diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,249632 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 10000,
+  "global_step": 207865,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999999998857893e-05,
+      "loss": 1.0258,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999999971447314e-05,
+      "loss": 1.3607,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999999885789244e-05,
+      "loss": 1.5179,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.99999997430258e-05,
+      "loss": 1.5094,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999999543156976e-05,
+      "loss": 1.8494,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999999928618278e-05,
+      "loss": 1.7089,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999998972103206e-05,
+      "loss": 1.387,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999999860091826e-05,
+      "loss": 1.5223,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999998172627943e-05,
+      "loss": 1.3672,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999999768723226e-05,
+      "loss": 1.5087,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999997144731212e-05,
+      "loss": 1.4472,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999996545124798e-05,
+      "loss": 1.3551,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999999588841303e-05,
+      "loss": 1.5519,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999995174595904e-05,
+      "loss": 1.444,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999994403673428e-05,
+      "loss": 1.6448,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999993575645603e-05,
+      "loss": 1.6252,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999999269051244e-05,
+      "loss": 1.5915,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999991748273934e-05,
+      "loss": 1.6129,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999990748930104e-05,
+      "loss": 1.4517,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999989692480946e-05,
+      "loss": 1.1957,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999988578926466e-05,
+      "loss": 1.4978,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999987408266675e-05,
+      "loss": 1.4419,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999986180501576e-05,
+      "loss": 1.4301,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999984895631176e-05,
+      "loss": 1.3633,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999983553655488e-05,
+      "loss": 1.7502,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999982154574513e-05,
+      "loss": 1.1596,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999998069838826e-05,
+      "loss": 1.6852,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999997918509674e-05,
+      "loss": 1.5627,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999997761469996e-05,
+      "loss": 1.3476,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999997598719793e-05,
+      "loss": 1.1338,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999997430259066e-05,
+      "loss": 1.2655,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999972560878155e-05,
+      "loss": 1.2277,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999997076206043e-05,
+      "loss": 1.2342,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999968906137494e-05,
+      "loss": 1.6791,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999966993109358e-05,
+      "loss": 1.2291,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999965022976032e-05,
+      "loss": 1.2927,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999962995737524e-05,
+      "loss": 1.57,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999996091139385e-05,
+      "loss": 1.3672,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999995876994502e-05,
+      "loss": 1.1193,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999995657139105e-05,
+      "loss": 1.0744,
+      "step": 195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999954315731947e-05,
+      "loss": 1.2387,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999952002967723e-05,
+      "loss": 1.6411,
+      "step": 205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.99999496330984e-05,
+      "loss": 1.4896,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999947206123984e-05,
+      "loss": 1.2735,
+      "step": 215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999944722044493e-05,
+      "loss": 1.4901,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999942180859938e-05,
+      "loss": 1.3097,
+      "step": 225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999939582570333e-05,
+      "loss": 1.1777,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999936927175694e-05,
+      "loss": 1.4237,
+      "step": 235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999993421467604e-05,
+      "loss": 1.2831,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999931445071385e-05,
+      "loss": 1.3956,
+      "step": 245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999928618361737e-05,
+      "loss": 1.4323,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999992573454712e-05,
+      "loss": 1.5805,
+      "step": 255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999922793627552e-05,
+      "loss": 1.3611,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999991979560304e-05,
+      "loss": 1.3747,
+      "step": 265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999916740473612e-05,
+      "loss": 1.6031,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999991362823928e-05,
+      "loss": 1.3129,
+      "step": 275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999910458900064e-05,
+      "loss": 1.3746,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999990723245598e-05,
+      "loss": 1.3728,
+      "step": 285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999903948907046e-05,
+      "loss": 1.3943,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999990060825328e-05,
+      "loss": 1.2526,
+      "step": 295
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999897210494707e-05,
+      "loss": 1.6433,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999893755631342e-05,
+      "loss": 1.1434,
+      "step": 305
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999890243663204e-05,
+      "loss": 1.2214,
+      "step": 310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999886674590314e-05,
+      "loss": 1.1086,
+      "step": 315
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999883048412695e-05,
+      "loss": 1.8266,
+      "step": 320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999987936513036e-05,
+      "loss": 1.3873,
+      "step": 325
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999875624743343e-05,
+      "loss": 1.0673,
+      "step": 330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999987182725165e-05,
+      "loss": 1.432,
+      "step": 335
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999867972655317e-05,
+      "loss": 1.2861,
+      "step": 340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999864060954354e-05,
+      "loss": 1.2748,
+      "step": 345
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999986009214879e-05,
+      "loss": 1.0814,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999985606623865e-05,
+      "loss": 1.2774,
+      "step": 355
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999851983223954e-05,
+      "loss": 1.5285,
+      "step": 360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999984784310472e-05,
+      "loss": 1.2669,
+      "step": 365
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999843645880978e-05,
+      "loss": 1.125,
+      "step": 370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999839391552748e-05,
+      "loss": 1.1198,
+      "step": 375
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999983508012006e-05,
+      "loss": 1.5111,
+      "step": 380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999830711582936e-05,
+      "loss": 1.378,
+      "step": 385
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999826285941398e-05,
+      "loss": 1.1921,
+      "step": 390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999821803195474e-05,
+      "loss": 1.2373,
+      "step": 395
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999981726334519e-05,
+      "loss": 1.2725,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999812666390568e-05,
+      "loss": 1.3351,
+      "step": 405
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999980801233164e-05,
+      "loss": 1.0018,
+      "step": 410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999980330116843e-05,
+      "loss": 1.1963,
+      "step": 415
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999798532900962e-05,
+      "loss": 1.4243,
+      "step": 420
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999979370752927e-05,
+      "loss": 1.2531,
+      "step": 425
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999788825053376e-05,
+      "loss": 1.1288,
+      "step": 430
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999978388547331e-05,
+      "loss": 1.5797,
+      "step": 435
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999778888789098e-05,
+      "loss": 1.6823,
+      "step": 440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999977383500077e-05,
+      "loss": 1.139,
+      "step": 445
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999768724108355e-05,
+      "loss": 1.1833,
+      "step": 450
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999763556111885e-05,
+      "loss": 1.3974,
+      "step": 455
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999975833101138e-05,
+      "loss": 1.2195,
+      "step": 460
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999753048806885e-05,
+      "loss": 1.0334,
+      "step": 465
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999747709498418e-05,
+      "loss": 1.248,
+      "step": 470
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999742313086013e-05,
+      "loss": 1.2612,
+      "step": 475
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.99997368595697e-05,
+      "loss": 1.378,
+      "step": 480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999731348949513e-05,
+      "loss": 1.3394,
+      "step": 485
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999972578122548e-05,
+      "loss": 1.0196,
+      "step": 490
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999720156397636e-05,
+      "loss": 1.6677,
+      "step": 495
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999714474466006e-05,
+      "loss": 1.1126,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999708735430637e-05,
+      "loss": 1.2853,
+      "step": 505
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999970293929155e-05,
+      "loss": 1.1659,
+      "step": 510
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999697086048775e-05,
+      "loss": 0.857,
+      "step": 515
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999691175702355e-05,
+      "loss": 1.233,
+      "step": 520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999685208252323e-05,
+      "loss": 1.343,
+      "step": 525
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999967918369871e-05,
+      "loss": 1.5479,
+      "step": 530
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999673102041545e-05,
+      "loss": 1.7049,
+      "step": 535
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999666963280872e-05,
+      "loss": 1.3518,
+      "step": 540
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999966076741672e-05,
+      "loss": 1.4052,
+      "step": 545
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999965451444913e-05,
+      "loss": 1.2536,
+      "step": 550
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999648204378132e-05,
+      "loss": 1.349,
+      "step": 555
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999641837203766e-05,
+      "loss": 1.3349,
+      "step": 560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999635412926065e-05,
+      "loss": 1.3917,
+      "step": 565
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999962893154507e-05,
+      "loss": 1.3191,
+      "step": 570
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999622393060812e-05,
+      "loss": 1.2765,
+      "step": 575
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999615797473336e-05,
+      "loss": 1.0695,
+      "step": 580
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999609144782673e-05,
+      "loss": 1.5153,
+      "step": 585
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999602434988864e-05,
+      "loss": 1.2533,
+      "step": 590
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999595668091945e-05,
+      "loss": 1.0707,
+      "step": 595
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999958884409196e-05,
+      "loss": 1.3526,
+      "step": 600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999958196298894e-05,
+      "loss": 1.3378,
+      "step": 605
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999575024782933e-05,
+      "loss": 1.2319,
+      "step": 610
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999956802947397e-05,
+      "loss": 1.0114,
+      "step": 615
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.99995609770621e-05,
+      "loss": 1.5617,
+      "step": 620
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999553867547356e-05,
+      "loss": 1.2009,
+      "step": 625
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999954670092978e-05,
+      "loss": 1.2912,
+      "step": 630
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999539477209415e-05,
+      "loss": 1.1615,
+      "step": 635
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.99995321963863e-05,
+      "loss": 1.1566,
+      "step": 640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999524858460482e-05,
+      "loss": 1.4267,
+      "step": 645
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999517463431997e-05,
+      "loss": 1.1853,
+      "step": 650
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999510011300888e-05,
+      "loss": 1.2694,
+      "step": 655
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999502502067198e-05,
+      "loss": 0.9305,
+      "step": 660
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999494935730972e-05,
+      "loss": 1.3301,
+      "step": 665
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999948731229225e-05,
+      "loss": 1.4041,
+      "step": 670
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999947963175108e-05,
+      "loss": 1.3485,
+      "step": 675
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.99994718941075e-05,
+      "loss": 1.0717,
+      "step": 680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999464099361558e-05,
+      "loss": 1.2562,
+      "step": 685
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999456247513297e-05,
+      "loss": 1.1009,
+      "step": 690
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999448338562767e-05,
+      "loss": 1.548,
+      "step": 695
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999440372510003e-05,
+      "loss": 1.5739,
+      "step": 700
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999943234935506e-05,
+      "loss": 1.5238,
+      "step": 705
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999942426909798e-05,
+      "loss": 0.9669,
+      "step": 710
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999416131738808e-05,
+      "loss": 1.2373,
+      "step": 715
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999407937277592e-05,
+      "loss": 1.6953,
+      "step": 720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999399685714378e-05,
+      "loss": 1.2403,
+      "step": 725
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999391377049214e-05,
+      "loss": 1.4917,
+      "step": 730
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999383011282147e-05,
+      "loss": 1.4203,
+      "step": 735
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999374588413225e-05,
+      "loss": 1.3485,
+      "step": 740
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999366108442495e-05,
+      "loss": 1.0583,
+      "step": 745
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999357571370008e-05,
+      "loss": 1.3338,
+      "step": 750
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999934897719581e-05,
+      "loss": 1.1792,
+      "step": 755
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999934032591995e-05,
+      "loss": 1.0468,
+      "step": 760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999933161754248e-05,
+      "loss": 1.2354,
+      "step": 765
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999932285206345e-05,
+      "loss": 1.1809,
+      "step": 770
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999314029482905e-05,
+      "loss": 1.0946,
+      "step": 775
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999305149800904e-05,
+      "loss": 1.2354,
+      "step": 780
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999296213017486e-05,
+      "loss": 1.1118,
+      "step": 785
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999287219132714e-05,
+      "loss": 1.0526,
+      "step": 790
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999278168146628e-05,
+      "loss": 1.2359,
+      "step": 795
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999926906005929e-05,
+      "loss": 1.2173,
+      "step": 800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999259894870744e-05,
+      "loss": 1.3357,
+      "step": 805
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999250672581047e-05,
+      "loss": 1.327,
+      "step": 810
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999241393190253e-05,
+      "loss": 1.5078,
+      "step": 815
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999923205669841e-05,
+      "loss": 1.2591,
+      "step": 820
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999222663105574e-05,
+      "loss": 1.422,
+      "step": 825
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.99992132124118e-05,
+      "loss": 1.2746,
+      "step": 830
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999203704617142e-05,
+      "loss": 1.0635,
+      "step": 835
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999919413972165e-05,
+      "loss": 1.4255,
+      "step": 840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999918451772538e-05,
+      "loss": 1.0915,
+      "step": 845
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999174838628394e-05,
+      "loss": 1.0714,
+      "step": 850
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999165102430736e-05,
+      "loss": 1.4302,
+      "step": 855
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999915530913247e-05,
+      "loss": 1.2158,
+      "step": 860
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999914545873365e-05,
+      "loss": 1.3395,
+      "step": 865
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999135551234326e-05,
+      "loss": 1.3719,
+      "step": 870
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999125586634563e-05,
+      "loss": 1.5028,
+      "step": 875
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999911556493442e-05,
+      "loss": 1.1432,
+      "step": 880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999105486133943e-05,
+      "loss": 1.0405,
+      "step": 885
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999095350233195e-05,
+      "loss": 1.1998,
+      "step": 890
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999085157232238e-05,
+      "loss": 1.4593,
+      "step": 895
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999074907131123e-05,
+      "loss": 1.2772,
+      "step": 900
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999064599929917e-05,
+      "loss": 1.5939,
+      "step": 905
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999905423562867e-05,
+      "loss": 1.2797,
+      "step": 910
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999043814227446e-05,
+      "loss": 1.4185,
+      "step": 915
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999033335726306e-05,
+      "loss": 1.0566,
+      "step": 920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999022800125306e-05,
+      "loss": 1.4083,
+      "step": 925
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999012207424507e-05,
+      "loss": 1.1619,
+      "step": 930
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999001557623974e-05,
+      "loss": 1.2815,
+      "step": 935
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999899085072376e-05,
+      "loss": 0.934,
+      "step": 940
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999898008672393e-05,
+      "loss": 1.1476,
+      "step": 945
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999896926562455e-05,
+      "loss": 1.4932,
+      "step": 950
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998958387425675e-05,
+      "loss": 1.1182,
+      "step": 955
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999894745212737e-05,
+      "loss": 1.1174,
+      "step": 960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998936459729698e-05,
+      "loss": 1.1109,
+      "step": 965
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998925410232718e-05,
+      "loss": 1.308,
+      "step": 970
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998914303636498e-05,
+      "loss": 1.1317,
+      "step": 975
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998903139941102e-05,
+      "loss": 1.264,
+      "step": 980
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998891919146586e-05,
+      "loss": 1.3861,
+      "step": 985
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998880641253023e-05,
+      "loss": 1.247,
+      "step": 990
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999886930626047e-05,
+      "loss": 1.6799,
+      "step": 995
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998857914168997e-05,
+      "loss": 1.3082,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999884646497867e-05,
+      "loss": 1.392,
+      "step": 1005
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999883495868955e-05,
+      "loss": 1.4562,
+      "step": 1010
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998823395301703e-05,
+      "loss": 1.4506,
+      "step": 1015
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998811774815195e-05,
+      "loss": 1.0513,
+      "step": 1020
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998800097230098e-05,
+      "loss": 1.2709,
+      "step": 1025
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998788362546472e-05,
+      "loss": 1.4961,
+      "step": 1030
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999877657076439e-05,
+      "loss": 1.1848,
+      "step": 1035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999876472188391e-05,
+      "loss": 1.285,
+      "step": 1040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998752815905114e-05,
+      "loss": 1.4173,
+      "step": 1045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998740852828057e-05,
+      "loss": 1.3801,
+      "step": 1050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999872883265281e-05,
+      "loss": 1.3254,
+      "step": 1055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998716755379445e-05,
+      "loss": 1.2398,
+      "step": 1060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999870462100803e-05,
+      "loss": 1.6193,
+      "step": 1065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998692429538634e-05,
+      "loss": 1.0443,
+      "step": 1070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998680180971328e-05,
+      "loss": 1.4149,
+      "step": 1075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998667875306174e-05,
+      "loss": 1.4814,
+      "step": 1080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998655512543254e-05,
+      "loss": 1.5378,
+      "step": 1085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998643092682633e-05,
+      "loss": 1.115,
+      "step": 1090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999863061572438e-05,
+      "loss": 1.2249,
+      "step": 1095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998618081668567e-05,
+      "loss": 1.1182,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998605490515274e-05,
+      "loss": 1.3632,
+      "step": 1105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999859284226456e-05,
+      "loss": 1.3083,
+      "step": 1110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998580136916503e-05,
+      "loss": 1.4238,
+      "step": 1115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998567374471177e-05,
+      "loss": 1.3233,
+      "step": 1120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998554554928654e-05,
+      "loss": 1.3187,
+      "step": 1125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998541678289006e-05,
+      "loss": 1.1243,
+      "step": 1130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999852874455231e-05,
+      "loss": 1.3218,
+      "step": 1135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998515753718634e-05,
+      "loss": 1.5325,
+      "step": 1140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998502705788056e-05,
+      "loss": 1.1336,
+      "step": 1145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999848960076065e-05,
+      "loss": 1.3542,
+      "step": 1150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998476438636492e-05,
+      "loss": 1.2375,
+      "step": 1155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998463219415652e-05,
+      "loss": 1.6395,
+      "step": 1160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998449943098212e-05,
+      "loss": 1.3271,
+      "step": 1165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998436609684246e-05,
+      "loss": 1.5619,
+      "step": 1170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999842321917383e-05,
+      "loss": 1.2337,
+      "step": 1175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998409771567034e-05,
+      "loss": 1.3469,
+      "step": 1180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998396266863948e-05,
+      "loss": 1.3802,
+      "step": 1185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998382705064637e-05,
+      "loss": 1.2715,
+      "step": 1190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998369086169184e-05,
+      "loss": 1.3735,
+      "step": 1195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998355410177668e-05,
+      "loss": 1.3237,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998341677090162e-05,
+      "loss": 1.2922,
+      "step": 1205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998327886906747e-05,
+      "loss": 1.1743,
+      "step": 1210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998314039627504e-05,
+      "loss": 1.0238,
+      "step": 1215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999830013525251e-05,
+      "loss": 1.4597,
+      "step": 1220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998286173781846e-05,
+      "loss": 1.3244,
+      "step": 1225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998272155215587e-05,
+      "loss": 0.9476,
+      "step": 1230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998258079553822e-05,
+      "loss": 1.5128,
+      "step": 1235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998243946796622e-05,
+      "loss": 1.3274,
+      "step": 1240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999822975694407e-05,
+      "loss": 1.7814,
+      "step": 1245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999821550999625e-05,
+      "loss": 1.8605,
+      "step": 1250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998201205953243e-05,
+      "loss": 1.2322,
+      "step": 1255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999818684481513e-05,
+      "loss": 1.2237,
+      "step": 1260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999817242658199e-05,
+      "loss": 1.292,
+      "step": 1265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999815795125391e-05,
+      "loss": 1.2144,
+      "step": 1270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998143418830967e-05,
+      "loss": 1.2193,
+      "step": 1275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999812882931325e-05,
+      "loss": 1.3432,
+      "step": 1280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998114182700842e-05,
+      "loss": 1.4002,
+      "step": 1285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998099478993823e-05,
+      "loss": 1.4279,
+      "step": 1290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998084718192277e-05,
+      "loss": 1.2951,
+      "step": 1295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999806990029629e-05,
+      "loss": 1.2204,
+      "step": 1300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998055025305947e-05,
+      "loss": 1.781,
+      "step": 1305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998040093221333e-05,
+      "loss": 0.9658,
+      "step": 1310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999802510404253e-05,
+      "loss": 1.3282,
+      "step": 1315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998010057769627e-05,
+      "loss": 1.1637,
+      "step": 1320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997994954402712e-05,
+      "loss": 1.3131,
+      "step": 1325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997979793941867e-05,
+      "loss": 1.4213,
+      "step": 1330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999796457638718e-05,
+      "loss": 1.1923,
+      "step": 1335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997949301738738e-05,
+      "loss": 1.2597,
+      "step": 1340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997933969996626e-05,
+      "loss": 1.23,
+      "step": 1345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997918581160936e-05,
+      "loss": 1.1332,
+      "step": 1350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997903135231753e-05,
+      "loss": 1.1103,
+      "step": 1355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997887632209164e-05,
+      "loss": 1.5728,
+      "step": 1360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999787207209326e-05,
+      "loss": 1.141,
+      "step": 1365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997856454884133e-05,
+      "loss": 1.471,
+      "step": 1370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997840780581864e-05,
+      "loss": 1.5945,
+      "step": 1375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997825049186548e-05,
+      "loss": 1.7122,
+      "step": 1380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999780926069827e-05,
+      "loss": 1.3147,
+      "step": 1385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999779341511713e-05,
+      "loss": 1.3256,
+      "step": 1390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999777751244321e-05,
+      "loss": 1.5589,
+      "step": 1395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99977615526766e-05,
+      "loss": 1.2491,
+      "step": 1400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997745535817396e-05,
+      "loss": 1.3827,
+      "step": 1405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999772946186569e-05,
+      "loss": 1.2154,
+      "step": 1410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999771333082157e-05,
+      "loss": 1.824,
+      "step": 1415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999769714268513e-05,
+      "loss": 1.5411,
+      "step": 1420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999768089745646e-05,
+      "loss": 1.3042,
+      "step": 1425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997664595135657e-05,
+      "loss": 1.3566,
+      "step": 1430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999764823572281e-05,
+      "loss": 1.7622,
+      "step": 1435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997631819218016e-05,
+      "loss": 1.4417,
+      "step": 1440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999761534562137e-05,
+      "loss": 1.3613,
+      "step": 1445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997598814932958e-05,
+      "loss": 1.3433,
+      "step": 1450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999758222715288e-05,
+      "loss": 1.0363,
+      "step": 1455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999756558228123e-05,
+      "loss": 1.369,
+      "step": 1460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997548880318107e-05,
+      "loss": 1.2401,
+      "step": 1465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997532121263602e-05,
+      "loss": 1.36,
+      "step": 1470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997515305117806e-05,
+      "loss": 1.4553,
+      "step": 1475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997498431880826e-05,
+      "loss": 1.5803,
+      "step": 1480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997481501552753e-05,
+      "loss": 1.5408,
+      "step": 1485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999746451413368e-05,
+      "loss": 1.2001,
+      "step": 1490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999744746962371e-05,
+      "loss": 1.2517,
+      "step": 1495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997430368022935e-05,
+      "loss": 1.0747,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999741320933146e-05,
+      "loss": 1.1227,
+      "step": 1505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997395993549375e-05,
+      "loss": 1.5405,
+      "step": 1510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997378720676783e-05,
+      "loss": 1.3432,
+      "step": 1515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999736139071378e-05,
+      "loss": 1.4938,
+      "step": 1520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997344003660466e-05,
+      "loss": 1.1219,
+      "step": 1525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997326559516944e-05,
+      "loss": 1.232,
+      "step": 1530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997309058283308e-05,
+      "loss": 1.2782,
+      "step": 1535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999729149995966e-05,
+      "loss": 1.2907,
+      "step": 1540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99972738845461e-05,
+      "loss": 1.5008,
+      "step": 1545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997256212042727e-05,
+      "loss": 1.2855,
+      "step": 1550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999723848244965e-05,
+      "loss": 1.3956,
+      "step": 1555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997220695766958e-05,
+      "loss": 1.2186,
+      "step": 1560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997202851994762e-05,
+      "loss": 1.2894,
+      "step": 1565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997184951133163e-05,
+      "loss": 1.1352,
+      "step": 1570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997166993182258e-05,
+      "loss": 1.0568,
+      "step": 1575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999714897814215e-05,
+      "loss": 1.1716,
+      "step": 1580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999713090601295e-05,
+      "loss": 1.3856,
+      "step": 1585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999711277679475e-05,
+      "loss": 1.0136,
+      "step": 1590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997094590487666e-05,
+      "loss": 1.351,
+      "step": 1595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999707634709179e-05,
+      "loss": 1.1206,
+      "step": 1600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997058046607233e-05,
+      "loss": 1.1116,
+      "step": 1605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997039689034094e-05,
+      "loss": 1.3035,
+      "step": 1610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997021274372487e-05,
+      "loss": 1.3049,
+      "step": 1615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997002802622506e-05,
+      "loss": 1.4062,
+      "step": 1620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996984273784265e-05,
+      "loss": 1.3591,
+      "step": 1625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996965687857868e-05,
+      "loss": 1.2571,
+      "step": 1630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999694704484342e-05,
+      "loss": 1.2189,
+      "step": 1635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996928344741025e-05,
+      "loss": 1.008,
+      "step": 1640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996909587550793e-05,
+      "loss": 0.9261,
+      "step": 1645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999689077327283e-05,
+      "loss": 1.1806,
+      "step": 1650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996871901907244e-05,
+      "loss": 1.4428,
+      "step": 1655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996852973454143e-05,
+      "loss": 1.868,
+      "step": 1660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996833987913634e-05,
+      "loss": 1.4634,
+      "step": 1665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996814945285828e-05,
+      "loss": 1.4856,
+      "step": 1670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999679584557083e-05,
+      "loss": 1.4027,
+      "step": 1675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996776688768753e-05,
+      "loss": 1.4062,
+      "step": 1680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99967574748797e-05,
+      "loss": 1.405,
+      "step": 1685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999673820390379e-05,
+      "loss": 1.1749,
+      "step": 1690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996718875841128e-05,
+      "loss": 1.335,
+      "step": 1695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996699490691823e-05,
+      "loss": 1.0966,
+      "step": 1700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996680048455987e-05,
+      "loss": 1.3,
+      "step": 1705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999666054913373e-05,
+      "loss": 1.1251,
+      "step": 1710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996640992725166e-05,
+      "loss": 1.2012,
+      "step": 1715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996621379230405e-05,
+      "loss": 1.2105,
+      "step": 1720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999660170864956e-05,
+      "loss": 1.4711,
+      "step": 1725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999658198098274e-05,
+      "loss": 1.3942,
+      "step": 1730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999656219623006e-05,
+      "loss": 1.2188,
+      "step": 1735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999654235439164e-05,
+      "loss": 1.2358,
+      "step": 1740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999652245546758e-05,
+      "loss": 1.3475,
+      "step": 1745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996502499458e-05,
+      "loss": 1.3608,
+      "step": 1750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996482486363016e-05,
+      "loss": 1.144,
+      "step": 1755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996462416182737e-05,
+      "loss": 1.1565,
+      "step": 1760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996442288917286e-05,
+      "loss": 1.2571,
+      "step": 1765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996422104566774e-05,
+      "loss": 1.1051,
+      "step": 1770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999640186313131e-05,
+      "loss": 1.2703,
+      "step": 1775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996381564611016e-05,
+      "loss": 1.4952,
+      "step": 1780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996361209006007e-05,
+      "loss": 1.1405,
+      "step": 1785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99963407963164e-05,
+      "loss": 1.4621,
+      "step": 1790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996320326542313e-05,
+      "loss": 1.3307,
+      "step": 1795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996299799683854e-05,
+      "loss": 1.4663,
+      "step": 1800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999627921574115e-05,
+      "loss": 1.3405,
+      "step": 1805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996258574714317e-05,
+      "loss": 1.4019,
+      "step": 1810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999623787660347e-05,
+      "loss": 1.0,
+      "step": 1815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996217121408725e-05,
+      "loss": 1.3007,
+      "step": 1820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996196309130206e-05,
+      "loss": 1.218,
+      "step": 1825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996175439768032e-05,
+      "loss": 1.2565,
+      "step": 1830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996154513322316e-05,
+      "loss": 1.1329,
+      "step": 1835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996133529793183e-05,
+      "loss": 1.2561,
+      "step": 1840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999611248918075e-05,
+      "loss": 1.3088,
+      "step": 1845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999609139148514e-05,
+      "loss": 1.1363,
+      "step": 1850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996070236706472e-05,
+      "loss": 1.132,
+      "step": 1855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996049024844866e-05,
+      "loss": 1.4022,
+      "step": 1860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996027755900442e-05,
+      "loss": 1.4587,
+      "step": 1865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9996006429873323e-05,
+      "loss": 1.5194,
+      "step": 1870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995985046763634e-05,
+      "loss": 1.0816,
+      "step": 1875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999596360657149e-05,
+      "loss": 1.1517,
+      "step": 1880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999594210929702e-05,
+      "loss": 1.3819,
+      "step": 1885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995920554940345e-05,
+      "loss": 1.0353,
+      "step": 1890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995898943501584e-05,
+      "loss": 1.7289,
+      "step": 1895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995877274980867e-05,
+      "loss": 1.1838,
+      "step": 1900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995855549378312e-05,
+      "loss": 1.4967,
+      "step": 1905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995833766694047e-05,
+      "loss": 1.2268,
+      "step": 1910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995811926928198e-05,
+      "loss": 1.1279,
+      "step": 1915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995790030080883e-05,
+      "loss": 1.2026,
+      "step": 1920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999576807615223e-05,
+      "loss": 1.0404,
+      "step": 1925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995746065142368e-05,
+      "loss": 1.2559,
+      "step": 1930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999572399705142e-05,
+      "loss": 0.9347,
+      "step": 1935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995701871879505e-05,
+      "loss": 1.1465,
+      "step": 1940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999567968962676e-05,
+      "loss": 1.5898,
+      "step": 1945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995657450293307e-05,
+      "loss": 1.4813,
+      "step": 1950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995635153879277e-05,
+      "loss": 1.526,
+      "step": 1955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999561280038479e-05,
+      "loss": 1.2717,
+      "step": 1960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999559038980998e-05,
+      "loss": 1.1622,
+      "step": 1965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995567922154976e-05,
+      "loss": 1.4986,
+      "step": 1970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99955453974199e-05,
+      "loss": 1.2453,
+      "step": 1975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995522815604883e-05,
+      "loss": 1.0614,
+      "step": 1980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995500176710055e-05,
+      "loss": 1.3614,
+      "step": 1985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995477480735547e-05,
+      "loss": 1.7098,
+      "step": 1990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999545472768148e-05,
+      "loss": 1.4479,
+      "step": 1995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995431917548e-05,
+      "loss": 1.0912,
+      "step": 2000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999540905033522e-05,
+      "loss": 1.3132,
+      "step": 2005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999538612604328e-05,
+      "loss": 1.0464,
+      "step": 2010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999536314467231e-05,
+      "loss": 1.155,
+      "step": 2015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995340106222442e-05,
+      "loss": 1.3222,
+      "step": 2020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995317010693802e-05,
+      "loss": 1.4382,
+      "step": 2025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999529385808653e-05,
+      "loss": 1.1369,
+      "step": 2030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995270648400753e-05,
+      "loss": 1.4491,
+      "step": 2035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995247381636603e-05,
+      "loss": 1.0443,
+      "step": 2040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995224057794217e-05,
+      "loss": 1.2192,
+      "step": 2045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995200676873723e-05,
+      "loss": 1.1347,
+      "step": 2050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999517723887526e-05,
+      "loss": 1.4007,
+      "step": 2055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995153743798953e-05,
+      "loss": 1.1644,
+      "step": 2060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995130191644948e-05,
+      "loss": 1.4453,
+      "step": 2065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995106582413374e-05,
+      "loss": 1.2323,
+      "step": 2070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995082916104363e-05,
+      "loss": 1.4546,
+      "step": 2075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995059192718053e-05,
+      "loss": 1.2772,
+      "step": 2080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999503541225458e-05,
+      "loss": 1.3487,
+      "step": 2085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995011574714078e-05,
+      "loss": 1.4051,
+      "step": 2090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994987680096684e-05,
+      "loss": 1.1992,
+      "step": 2095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994963728402537e-05,
+      "loss": 1.2443,
+      "step": 2100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999493971963177e-05,
+      "loss": 1.2023,
+      "step": 2105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999491565378452e-05,
+      "loss": 1.5508,
+      "step": 2110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994891530860924e-05,
+      "loss": 1.4321,
+      "step": 2115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994867350861125e-05,
+      "loss": 1.1997,
+      "step": 2120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994843113785258e-05,
+      "loss": 1.3156,
+      "step": 2125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999481881963346e-05,
+      "loss": 1.2119,
+      "step": 2130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994794468405873e-05,
+      "loss": 1.2588,
+      "step": 2135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994770060102627e-05,
+      "loss": 1.3705,
+      "step": 2140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994745594723875e-05,
+      "loss": 1.405,
+      "step": 2145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999472107226975e-05,
+      "loss": 1.4671,
+      "step": 2150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999469649274039e-05,
+      "loss": 1.2521,
+      "step": 2155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994671856135937e-05,
+      "loss": 1.2989,
+      "step": 2160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999464716245653e-05,
+      "loss": 1.0464,
+      "step": 2165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994622411702312e-05,
+      "loss": 1.4591,
+      "step": 2170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999459760387343e-05,
+      "loss": 1.205,
+      "step": 2175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999457273897002e-05,
+      "loss": 1.0434,
+      "step": 2180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994547816992217e-05,
+      "loss": 1.1634,
+      "step": 2185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994522837940177e-05,
+      "loss": 1.3809,
+      "step": 2190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999449780181403e-05,
+      "loss": 1.319,
+      "step": 2195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999447270861393e-05,
+      "loss": 1.4415,
+      "step": 2200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994447558340014e-05,
+      "loss": 1.7574,
+      "step": 2205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994422350992425e-05,
+      "loss": 1.3178,
+      "step": 2210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994397086571312e-05,
+      "loss": 1.2316,
+      "step": 2215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999437176507681e-05,
+      "loss": 1.5254,
+      "step": 2220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994346386509077e-05,
+      "loss": 1.2644,
+      "step": 2225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999432095086825e-05,
+      "loss": 1.5037,
+      "step": 2230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999429545815447e-05,
+      "loss": 0.9937,
+      "step": 2235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994269908367888e-05,
+      "loss": 1.2979,
+      "step": 2240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994244301508655e-05,
+      "loss": 1.3459,
+      "step": 2245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994218637576907e-05,
+      "loss": 1.3687,
+      "step": 2250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994192916572798e-05,
+      "loss": 1.1613,
+      "step": 2255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999416713849647e-05,
+      "loss": 1.1939,
+      "step": 2260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999414130334807e-05,
+      "loss": 1.1573,
+      "step": 2265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999411541112775e-05,
+      "loss": 0.9257,
+      "step": 2270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994089461835657e-05,
+      "loss": 1.6172,
+      "step": 2275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994063455471938e-05,
+      "loss": 1.4508,
+      "step": 2280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999403739203674e-05,
+      "loss": 1.1226,
+      "step": 2285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994011271530213e-05,
+      "loss": 1.2616,
+      "step": 2290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993985093952506e-05,
+      "loss": 1.1506,
+      "step": 2295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993958859303772e-05,
+      "loss": 1.1799,
+      "step": 2300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993932567584153e-05,
+      "loss": 1.318,
+      "step": 2305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993906218793806e-05,
+      "loss": 1.6039,
+      "step": 2310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999387981293288e-05,
+      "loss": 1.1135,
+      "step": 2315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993853350001525e-05,
+      "loss": 1.0927,
+      "step": 2320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999382682999989e-05,
+      "loss": 1.4625,
+      "step": 2325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999380025292813e-05,
+      "loss": 1.0676,
+      "step": 2330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993773618786395e-05,
+      "loss": 1.3484,
+      "step": 2335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999374692757484e-05,
+      "loss": 1.1844,
+      "step": 2340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999372017929361e-05,
+      "loss": 1.4724,
+      "step": 2345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993693373942866e-05,
+      "loss": 0.9435,
+      "step": 2350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993666511522755e-05,
+      "loss": 1.3911,
+      "step": 2355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999363959203344e-05,
+      "loss": 1.366,
+      "step": 2360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999361261547506e-05,
+      "loss": 1.2664,
+      "step": 2365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993585581847782e-05,
+      "loss": 1.1167,
+      "step": 2370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993558491151752e-05,
+      "loss": 1.3277,
+      "step": 2375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999353134338713e-05,
+      "loss": 1.4107,
+      "step": 2380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993504138554067e-05,
+      "loss": 1.5268,
+      "step": 2385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993476876652722e-05,
+      "loss": 1.4592,
+      "step": 2390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993449557683248e-05,
+      "loss": 1.1203,
+      "step": 2395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993422181645803e-05,
+      "loss": 1.0469,
+      "step": 2400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999339474854054e-05,
+      "loss": 1.3235,
+      "step": 2405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999336725836762e-05,
+      "loss": 1.1442,
+      "step": 2410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99933397111272e-05,
+      "loss": 1.2534,
+      "step": 2415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999331210681943e-05,
+      "loss": 1.6231,
+      "step": 2420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999328444544448e-05,
+      "loss": 1.1638,
+      "step": 2425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993256727002494e-05,
+      "loss": 1.0055,
+      "step": 2430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993228951493638e-05,
+      "loss": 1.2429,
+      "step": 2435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993201118918076e-05,
+      "loss": 1.2716,
+      "step": 2440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993173229275955e-05,
+      "loss": 1.0403,
+      "step": 2445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999314528256744e-05,
+      "loss": 1.2959,
+      "step": 2450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993117278792695e-05,
+      "loss": 1.0377,
+      "step": 2455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999308921795187e-05,
+      "loss": 1.2522,
+      "step": 2460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993061100045135e-05,
+      "loss": 1.4304,
+      "step": 2465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993032925072645e-05,
+      "loss": 1.1826,
+      "step": 2470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9993004693034562e-05,
+      "loss": 1.3453,
+      "step": 2475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992976403931047e-05,
+      "loss": 1.1502,
+      "step": 2480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992948057762264e-05,
+      "loss": 1.5489,
+      "step": 2485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999291965452837e-05,
+      "loss": 1.4626,
+      "step": 2490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992891194229532e-05,
+      "loss": 1.4757,
+      "step": 2495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999286267686591e-05,
+      "loss": 1.0684,
+      "step": 2500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992834102437667e-05,
+      "loss": 1.2266,
+      "step": 2505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999280547094497e-05,
+      "loss": 1.0737,
+      "step": 2510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992776782387976e-05,
+      "loss": 1.321,
+      "step": 2515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999274803676685e-05,
+      "loss": 1.2547,
+      "step": 2520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999271923408176e-05,
+      "loss": 1.3321,
+      "step": 2525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992690374332867e-05,
+      "loss": 1.0981,
+      "step": 2530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992661457520343e-05,
+      "loss": 1.2187,
+      "step": 2535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999263248364434e-05,
+      "loss": 1.5274,
+      "step": 2540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992603452705036e-05,
+      "loss": 1.4583,
+      "step": 2545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992574364702593e-05,
+      "loss": 1.4763,
+      "step": 2550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992545219637173e-05,
+      "loss": 1.1884,
+      "step": 2555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992516017508943e-05,
+      "loss": 1.1495,
+      "step": 2560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992486758318078e-05,
+      "loss": 1.3172,
+      "step": 2565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999245744206473e-05,
+      "loss": 1.2909,
+      "step": 2570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992428068749085e-05,
+      "loss": 1.2032,
+      "step": 2575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992398638371297e-05,
+      "loss": 1.4464,
+      "step": 2580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992369150931538e-05,
+      "loss": 1.0335,
+      "step": 2585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992339606429978e-05,
+      "loss": 1.1004,
+      "step": 2590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992310004866783e-05,
+      "loss": 1.3034,
+      "step": 2595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992280346242126e-05,
+      "loss": 1.2016,
+      "step": 2600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992250630556173e-05,
+      "loss": 1.1691,
+      "step": 2605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992220857809093e-05,
+      "loss": 1.1937,
+      "step": 2610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999219102800106e-05,
+      "loss": 1.0652,
+      "step": 2615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999216114113224e-05,
+      "loss": 1.5977,
+      "step": 2620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992131197202805e-05,
+      "loss": 1.1948,
+      "step": 2625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992101196212926e-05,
+      "loss": 0.8635,
+      "step": 2630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992071138162775e-05,
+      "loss": 1.1359,
+      "step": 2635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9992041023052527e-05,
+      "loss": 1.2671,
+      "step": 2640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999201085088235e-05,
+      "loss": 1.1429,
+      "step": 2645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991980621652415e-05,
+      "loss": 1.3101,
+      "step": 2650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991950335362896e-05,
+      "loss": 1.2979,
+      "step": 2655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991919992013966e-05,
+      "loss": 1.2994,
+      "step": 2660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991889591605802e-05,
+      "loss": 1.1883,
+      "step": 2665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991859134138572e-05,
+      "loss": 1.3228,
+      "step": 2670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999182861961245e-05,
+      "loss": 1.0386,
+      "step": 2675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991798048027617e-05,
+      "loss": 1.0111,
+      "step": 2680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999176741938424e-05,
+      "loss": 1.0779,
+      "step": 2685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991736733682496e-05,
+      "loss": 1.3726,
+      "step": 2690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999170599092256e-05,
+      "loss": 1.3075,
+      "step": 2695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991675191104614e-05,
+      "loss": 1.0187,
+      "step": 2700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991644334228827e-05,
+      "loss": 1.251,
+      "step": 2705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991613420295376e-05,
+      "loss": 0.917,
+      "step": 2710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999158244930444e-05,
+      "loss": 1.4163,
+      "step": 2715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991551421256193e-05,
+      "loss": 1.7127,
+      "step": 2720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999152033615081e-05,
+      "loss": 1.2469,
+      "step": 2725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991489193988478e-05,
+      "loss": 0.9603,
+      "step": 2730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991457994769363e-05,
+      "loss": 1.2009,
+      "step": 2735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991426738493655e-05,
+      "loss": 1.1193,
+      "step": 2740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991395425161523e-05,
+      "loss": 1.1736,
+      "step": 2745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999136405477315e-05,
+      "loss": 1.3103,
+      "step": 2750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999133262732871e-05,
+      "loss": 1.134,
+      "step": 2755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991301142828392e-05,
+      "loss": 1.1826,
+      "step": 2760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999126960127237e-05,
+      "loss": 1.1032,
+      "step": 2765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991238002660822e-05,
+      "loss": 1.2721,
+      "step": 2770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999120634699393e-05,
+      "loss": 1.1308,
+      "step": 2775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991174634271883e-05,
+      "loss": 1.2107,
+      "step": 2780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999114286449485e-05,
+      "loss": 1.0983,
+      "step": 2785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999111103766302e-05,
+      "loss": 1.1577,
+      "step": 2790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999107915377657e-05,
+      "loss": 1.053,
+      "step": 2795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991047212835687e-05,
+      "loss": 1.7503,
+      "step": 2800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999101521484055e-05,
+      "loss": 1.2732,
+      "step": 2805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990983159791338e-05,
+      "loss": 1.5427,
+      "step": 2810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999095104768824e-05,
+      "loss": 1.0419,
+      "step": 2815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999091887853144e-05,
+      "loss": 0.9169,
+      "step": 2820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990886652321123e-05,
+      "loss": 1.1803,
+      "step": 2825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990854369057466e-05,
+      "loss": 1.5184,
+      "step": 2830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990822028740653e-05,
+      "loss": 1.5375,
+      "step": 2835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990789631370876e-05,
+      "loss": 1.3887,
+      "step": 2840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990757176948318e-05,
+      "loss": 1.7262,
+      "step": 2845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990724665473163e-05,
+      "loss": 1.1992,
+      "step": 2850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990692096945594e-05,
+      "loss": 1.2063,
+      "step": 2855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990659471365805e-05,
+      "loss": 1.2594,
+      "step": 2860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990626788733973e-05,
+      "loss": 1.1143,
+      "step": 2865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990594049050284e-05,
+      "loss": 1.3394,
+      "step": 2870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990561252314935e-05,
+      "loss": 1.533,
+      "step": 2875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990528398528108e-05,
+      "loss": 1.0778,
+      "step": 2880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990495487689988e-05,
+      "loss": 1.2407,
+      "step": 2885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999046251980077e-05,
+      "loss": 1.4199,
+      "step": 2890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999042949486063e-05,
+      "loss": 1.1856,
+      "step": 2895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999039641286977e-05,
+      "loss": 1.1637,
+      "step": 2900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990363273828374e-05,
+      "loss": 1.2002,
+      "step": 2905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990330077736628e-05,
+      "loss": 1.2125,
+      "step": 2910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990296824594722e-05,
+      "loss": 1.1623,
+      "step": 2915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999026351440285e-05,
+      "loss": 1.1779,
+      "step": 2920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99902301471612e-05,
+      "loss": 1.6169,
+      "step": 2925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999019672286996e-05,
+      "loss": 1.5845,
+      "step": 2930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990163241529327e-05,
+      "loss": 1.07,
+      "step": 2935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990129703139488e-05,
+      "loss": 1.5606,
+      "step": 2940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990096107700635e-05,
+      "loss": 1.1276,
+      "step": 2945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999006245521296e-05,
+      "loss": 1.1697,
+      "step": 2950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9990028745676656e-05,
+      "loss": 1.4139,
+      "step": 2955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989994979091913e-05,
+      "loss": 1.56,
+      "step": 2960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989961155458925e-05,
+      "loss": 1.34,
+      "step": 2965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989927274777888e-05,
+      "loss": 1.0135,
+      "step": 2970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998989333704899e-05,
+      "loss": 1.4891,
+      "step": 2975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989859342272432e-05,
+      "loss": 1.543,
+      "step": 2980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989825290448402e-05,
+      "loss": 1.1451,
+      "step": 2985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989791181577096e-05,
+      "loss": 1.4817,
+      "step": 2990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998975701565871e-05,
+      "loss": 1.2326,
+      "step": 2995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998972279269344e-05,
+      "loss": 1.2476,
+      "step": 3000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989688512681477e-05,
+      "loss": 1.0286,
+      "step": 3005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998965417562302e-05,
+      "loss": 1.3189,
+      "step": 3010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989619781518267e-05,
+      "loss": 1.0645,
+      "step": 3015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989585330367408e-05,
+      "loss": 1.1543,
+      "step": 3020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998955082217065e-05,
+      "loss": 1.4037,
+      "step": 3025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998951625692818e-05,
+      "loss": 1.0412,
+      "step": 3030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99894816346402e-05,
+      "loss": 1.1774,
+      "step": 3035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989446955306904e-05,
+      "loss": 1.4903,
+      "step": 3040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989412218928498e-05,
+      "loss": 1.1771,
+      "step": 3045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998937742550517e-05,
+      "loss": 1.2034,
+      "step": 3050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998934257503713e-05,
+      "loss": 1.0356,
+      "step": 3055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989307667524566e-05,
+      "loss": 1.3203,
+      "step": 3060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989272702967685e-05,
+      "loss": 1.0247,
+      "step": 3065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989237681366685e-05,
+      "loss": 1.2092,
+      "step": 3070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989202602721764e-05,
+      "loss": 1.0271,
+      "step": 3075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989167467033125e-05,
+      "loss": 1.0622,
+      "step": 3080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989132274300963e-05,
+      "loss": 1.1748,
+      "step": 3085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989097024525487e-05,
+      "loss": 1.1957,
+      "step": 3090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998906171770689e-05,
+      "loss": 1.3377,
+      "step": 3095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989026353845384e-05,
+      "loss": 1.3324,
+      "step": 3100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998899093294116e-05,
+      "loss": 1.5039,
+      "step": 3105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.998895545499443e-05,
+      "loss": 1.5469,
+      "step": 3110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9988919920005388e-05,
+      "loss": 1.0656,
+      "step": 3115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988884327974238e-05,
+      "loss": 1.7284,
+      "step": 3120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998884867890119e-05,
+      "loss": 1.2493,
+      "step": 3125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988812972786444e-05,
+      "loss": 1.3123,
+      "step": 3130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99887772096302e-05,
+      "loss": 1.3696,
+      "step": 3135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998874138943267e-05,
+      "loss": 1.2591,
+      "step": 3140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998870551219405e-05,
+      "loss": 0.9435,
+      "step": 3145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988669577914553e-05,
+      "loss": 1.1076,
+      "step": 3150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998863358659438e-05,
+      "loss": 1.2175,
+      "step": 3155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988597538233732e-05,
+      "loss": 1.3075,
+      "step": 3160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988561432832823e-05,
+      "loss": 1.4398,
+      "step": 3165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988525270391855e-05,
+      "loss": 1.1588,
+      "step": 3170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988489050911037e-05,
+      "loss": 1.4456,
+      "step": 3175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988452774390574e-05,
+      "loss": 1.1487,
+      "step": 3180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998841644083067e-05,
+      "loss": 1.2789,
+      "step": 3185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988380050231543e-05,
+      "loss": 1.0727,
+      "step": 3190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988343602593388e-05,
+      "loss": 1.081,
+      "step": 3195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988307097916418e-05,
+      "loss": 1.1536,
+      "step": 3200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988270536200845e-05,
+      "loss": 0.9586,
+      "step": 3205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988233917446874e-05,
+      "loss": 1.2468,
+      "step": 3210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988197241654713e-05,
+      "loss": 1.4239,
+      "step": 3215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998816050882458e-05,
+      "loss": 1.3969,
+      "step": 3220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988123718956672e-05,
+      "loss": 1.2195,
+      "step": 3225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988086872051208e-05,
+      "loss": 1.3121,
+      "step": 3230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988049968108396e-05,
+      "loss": 1.1703,
+      "step": 3235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9988013007128448e-05,
+      "loss": 1.4183,
+      "step": 3240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998797598911157e-05,
+      "loss": 0.9785,
+      "step": 3245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998793891405798e-05,
+      "loss": 1.0995,
+      "step": 3250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987901781967883e-05,
+      "loss": 1.1184,
+      "step": 3255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99878645928415e-05,
+      "loss": 1.0878,
+      "step": 3260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987827346679034e-05,
+      "loss": 1.4086,
+      "step": 3265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987790043480703e-05,
+      "loss": 1.2398,
+      "step": 3270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998775268324672e-05,
+      "loss": 1.122,
+      "step": 3275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987715265977296e-05,
+      "loss": 0.9607,
+      "step": 3280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987677791672645e-05,
+      "loss": 1.0706,
+      "step": 3285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987640260332982e-05,
+      "loss": 1.2926,
+      "step": 3290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987602671958525e-05,
+      "loss": 1.3086,
+      "step": 3295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998756502654948e-05,
+      "loss": 1.0877,
+      "step": 3300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998752732410607e-05,
+      "loss": 1.6327,
+      "step": 3305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987489564628508e-05,
+      "loss": 1.4853,
+      "step": 3310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998745174811701e-05,
+      "loss": 1.2108,
+      "step": 3315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987413874571787e-05,
+      "loss": 1.2671,
+      "step": 3320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987375943993055e-05,
+      "loss": 1.2902,
+      "step": 3325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987337956381044e-05,
+      "loss": 1.1788,
+      "step": 3330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998729991173596e-05,
+      "loss": 1.3284,
+      "step": 3335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987261810058015e-05,
+      "loss": 1.1191,
+      "step": 3340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998722365134744e-05,
+      "loss": 0.9398,
+      "step": 3345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998718543560444e-05,
+      "loss": 1.0621,
+      "step": 3350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987147162829242e-05,
+      "loss": 1.2517,
+      "step": 3355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987108833022066e-05,
+      "loss": 1.1904,
+      "step": 3360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9987070446183124e-05,
+      "loss": 1.2257,
+      "step": 3365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998703200231264e-05,
+      "loss": 1.1555,
+      "step": 3370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998699350141083e-05,
+      "loss": 1.2172,
+      "step": 3375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986954943477915e-05,
+      "loss": 1.3997,
+      "step": 3380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986916328514116e-05,
+      "loss": 1.2609,
+      "step": 3385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986877656519656e-05,
+      "loss": 1.2832,
+      "step": 3390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998683892749475e-05,
+      "loss": 1.6585,
+      "step": 3395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986800141439625e-05,
+      "loss": 1.0182,
+      "step": 3400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986761298354494e-05,
+      "loss": 1.2534,
+      "step": 3405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998672239823959e-05,
+      "loss": 1.2704,
+      "step": 3410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986683441095128e-05,
+      "loss": 1.2391,
+      "step": 3415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986644426921332e-05,
+      "loss": 1.326,
+      "step": 3420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986605355718428e-05,
+      "loss": 1.0966,
+      "step": 3425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986566227486632e-05,
+      "loss": 1.0565,
+      "step": 3430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986527042226174e-05,
+      "loss": 1.0275,
+      "step": 3435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986487799937272e-05,
+      "loss": 1.3414,
+      "step": 3440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998644850062016e-05,
+      "loss": 1.2849,
+      "step": 3445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998640914427505e-05,
+      "loss": 1.3067,
+      "step": 3450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986369730902173e-05,
+      "loss": 1.0379,
+      "step": 3455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986330260501755e-05,
+      "loss": 1.2661,
+      "step": 3460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998629073307402e-05,
+      "loss": 1.2381,
+      "step": 3465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986251148619193e-05,
+      "loss": 1.5477,
+      "step": 3470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986211507137503e-05,
+      "loss": 1.0285,
+      "step": 3475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986171808629174e-05,
+      "loss": 1.1698,
+      "step": 3480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998613205309443e-05,
+      "loss": 1.477,
+      "step": 3485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986092240533504e-05,
+      "loss": 0.9896,
+      "step": 3490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986052370946618e-05,
+      "loss": 1.4433,
+      "step": 3495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9986012444334003e-05,
+      "loss": 1.1927,
+      "step": 3500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985972460695885e-05,
+      "loss": 1.3256,
+      "step": 3505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985932420032494e-05,
+      "loss": 1.3751,
+      "step": 3510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998589232234406e-05,
+      "loss": 1.3471,
+      "step": 3515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985852167630805e-05,
+      "loss": 1.151,
+      "step": 3520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985811955892965e-05,
+      "loss": 1.2945,
+      "step": 3525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998577168713077e-05,
+      "loss": 1.2169,
+      "step": 3530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985731361344447e-05,
+      "loss": 1.7054,
+      "step": 3535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985690978534226e-05,
+      "loss": 0.9485,
+      "step": 3540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998565053870034e-05,
+      "loss": 1.3839,
+      "step": 3545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998561004184302e-05,
+      "loss": 1.3258,
+      "step": 3550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985569487962493e-05,
+      "loss": 1.4144,
+      "step": 3555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998552887705899e-05,
+      "loss": 1.3854,
+      "step": 3560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985488209132754e-05,
+      "loss": 1.4312,
+      "step": 3565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985447484184008e-05,
+      "loss": 1.1864,
+      "step": 3570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985406702212983e-05,
+      "loss": 1.3584,
+      "step": 3575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985365863219917e-05,
+      "loss": 1.3261,
+      "step": 3580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998532496720504e-05,
+      "loss": 1.1455,
+      "step": 3585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985284014168587e-05,
+      "loss": 1.2045,
+      "step": 3590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985243004110794e-05,
+      "loss": 1.3005,
+      "step": 3595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998520193703189e-05,
+      "loss": 1.1107,
+      "step": 3600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985160812932113e-05,
+      "loss": 0.9846,
+      "step": 3605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99851196318117e-05,
+      "loss": 1.3503,
+      "step": 3610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998507839367088e-05,
+      "loss": 1.349,
+      "step": 3615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998503709850989e-05,
+      "loss": 1.2254,
+      "step": 3620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984995746328972e-05,
+      "loss": 1.1365,
+      "step": 3625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984954337128356e-05,
+      "loss": 1.0888,
+      "step": 3630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998491287090828e-05,
+      "loss": 1.4065,
+      "step": 3635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984871347668986e-05,
+      "loss": 1.1721,
+      "step": 3640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99848297674107e-05,
+      "loss": 1.3823,
+      "step": 3645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998478813013367e-05,
+      "loss": 1.0268,
+      "step": 3650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984746435838124e-05,
+      "loss": 1.6879,
+      "step": 3655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998470468452431e-05,
+      "loss": 1.2871,
+      "step": 3660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998466287619246e-05,
+      "loss": 1.3944,
+      "step": 3665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984621010842817e-05,
+      "loss": 1.2081,
+      "step": 3670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984579088475615e-05,
+      "loss": 1.1398,
+      "step": 3675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99845371090911e-05,
+      "loss": 0.9496,
+      "step": 3680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984495072689506e-05,
+      "loss": 1.5794,
+      "step": 3685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984452979271074e-05,
+      "loss": 1.2617,
+      "step": 3690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984410828836042e-05,
+      "loss": 1.058,
+      "step": 3695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998436862138466e-05,
+      "loss": 1.0278,
+      "step": 3700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984326356917162e-05,
+      "loss": 1.0021,
+      "step": 3705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984284035433788e-05,
+      "loss": 1.089,
+      "step": 3710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984241656934783e-05,
+      "loss": 1.1257,
+      "step": 3715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984199221420387e-05,
+      "loss": 1.4381,
+      "step": 3720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984156728890845e-05,
+      "loss": 1.3058,
+      "step": 3725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9984114179346398e-05,
+      "loss": 1.1287,
+      "step": 3730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998407157278729e-05,
+      "loss": 0.954,
+      "step": 3735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998402890921376e-05,
+      "loss": 1.5134,
+      "step": 3740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998398618862606e-05,
+      "loss": 1.1486,
+      "step": 3745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983943411024426e-05,
+      "loss": 1.1327,
+      "step": 3750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983900576409106e-05,
+      "loss": 1.4486,
+      "step": 3755
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998385768478034e-05,
+      "loss": 1.2974,
+      "step": 3760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983814736138382e-05,
+      "loss": 1.1361,
+      "step": 3765
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998377173048347e-05,
+      "loss": 1.2916,
+      "step": 3770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983728667815853e-05,
+      "loss": 1.105,
+      "step": 3775
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998368554813577e-05,
+      "loss": 1.3921,
+      "step": 3780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983642371443478e-05,
+      "loss": 1.1146,
+      "step": 3785
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998359913773922e-05,
+      "loss": 1.0138,
+      "step": 3790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983555847023237e-05,
+      "loss": 1.5803,
+      "step": 3795
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998351249929578e-05,
+      "loss": 1.4203,
+      "step": 3800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99834690945571e-05,
+      "loss": 1.6779,
+      "step": 3805
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998342563280744e-05,
+      "loss": 1.6436,
+      "step": 3810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983382114047054e-05,
+      "loss": 1.2483,
+      "step": 3815
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983338538276183e-05,
+      "loss": 0.9466,
+      "step": 3820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998329490549508e-05,
+      "loss": 1.1959,
+      "step": 3825
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983251215703992e-05,
+      "loss": 1.1924,
+      "step": 3830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998320746890317e-05,
+      "loss": 1.2454,
+      "step": 3835
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983163665092866e-05,
+      "loss": 1.3082,
+      "step": 3840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998311980427333e-05,
+      "loss": 1.3338,
+      "step": 3845
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983075886444804e-05,
+      "loss": 1.008,
+      "step": 3850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983031911607547e-05,
+      "loss": 1.0311,
+      "step": 3855
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9982987879761815e-05,
+      "loss": 1.501,
+      "step": 3860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998294379090785e-05,
+      "loss": 1.1989,
+      "step": 3865
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99828996450459e-05,
+      "loss": 1.4316,
+      "step": 3870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998285544217623e-05,
+      "loss": 1.1935,
+      "step": 3875
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9982811182299086e-05,
+      "loss": 1.3396,
+      "step": 3880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998276686541472e-05,
+      "loss": 1.2888,
+      "step": 3885
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9982722491523385e-05,
+      "loss": 1.1156,
+      "step": 3890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9982678060625337e-05,
+      "loss": 1.2704,
+      "step": 3895
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9982633572720823e-05,
+      "loss": 1.1804,
+      "step": 3900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9982589027810105e-05,
+      "loss": 1.1653,
+      "step": 3905
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998254442589344e-05,
+      "loss": 1.0423,
+      "step": 3910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998249976697107e-05,
+      "loss": 1.2121,
+      "step": 3915
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998245505104326e-05,
+      "loss": 1.1042,
+      "step": 3920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9982410278110262e-05,
+      "loss": 1.271,
+      "step": 3925
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998236544817233e-05,
+      "loss": 1.1148,
+      "step": 3930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9982320561229728e-05,
+      "loss": 1.3313,
+      "step": 3935
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9982275617282702e-05,
+      "loss": 1.1635,
+      "step": 3940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9982230616331515e-05,
+      "loss": 1.1945,
+      "step": 3945
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9982185558376423e-05,
+      "loss": 1.5509,
+      "step": 3950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998214044341768e-05,
+      "loss": 1.4676,
+      "step": 3955
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998209527145555e-05,
+      "loss": 1.0713,
+      "step": 3960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9982050042490282e-05,
+      "loss": 1.2814,
+      "step": 3965
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9982004756522144e-05,
+      "loss": 1.35,
+      "step": 3970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998195941355139e-05,
+      "loss": 1.567,
+      "step": 3975
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9981914013578274e-05,
+      "loss": 1.2339,
+      "step": 3980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9981868556603065e-05,
+      "loss": 1.2337,
+      "step": 3985
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9981823042626013e-05,
+      "loss": 1.116,
+      "step": 3990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998177747164739e-05,
+      "loss": 1.0853,
+      "step": 3995
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998173184366744e-05,
+      "loss": 1.0055,
+      "step": 4000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998168615868644e-05,
+      "loss": 1.4382,
+      "step": 4005
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998164041670464e-05,
+      "loss": 1.4883,
+      "step": 4010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9981594617722306e-05,
+      "loss": 1.3952,
+      "step": 4015
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9981548761739694e-05,
+      "loss": 1.415,
+      "step": 4020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9981502848757077e-05,
+      "loss": 1.2542,
+      "step": 4025
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9981456878774702e-05,
+      "loss": 1.5518,
+      "step": 4030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9981410851792848e-05,
+      "loss": 1.1454,
+      "step": 4035
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9981364767811765e-05,
+      "loss": 1.2567,
+      "step": 4040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998131862683172e-05,
+      "loss": 1.5474,
+      "step": 4045
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998127242885298e-05,
+      "loss": 1.3412,
+      "step": 4050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9981226173875802e-05,
+      "loss": 1.4475,
+      "step": 4055
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9981179861900456e-05,
+      "loss": 1.3041,
+      "step": 4060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9981133492927206e-05,
+      "loss": 1.086,
+      "step": 4065
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9981087066956313e-05,
+      "loss": 1.117,
+      "step": 4070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9981040583988048e-05,
+      "loss": 0.9535,
+      "step": 4075
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9980994044022668e-05,
+      "loss": 1.0812,
+      "step": 4080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9980947447060448e-05,
+      "loss": 1.3068,
+      "step": 4085
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998090079310165e-05,
+      "loss": 1.3654,
+      "step": 4090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9980854082146537e-05,
+      "loss": 1.2032,
+      "step": 4095
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9980807314195382e-05,
+      "loss": 1.3033,
+      "step": 4100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9980760489248444e-05,
+      "loss": 1.3664,
+      "step": 4105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9980713607306e-05,
+      "loss": 1.151,
+      "step": 4110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9980666668368315e-05,
+      "loss": 1.4509,
+      "step": 4115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998061967243565e-05,
+      "loss": 0.9909,
+      "step": 4120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998057261950828e-05,
+      "loss": 1.2108,
+      "step": 4125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998052550958647e-05,
+      "loss": 1.1467,
+      "step": 4130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99804783426705e-05,
+      "loss": 1.173,
+      "step": 4135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9980431118760623e-05,
+      "loss": 1.16,
+      "step": 4140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998038383785712e-05,
+      "loss": 1.1147,
+      "step": 4145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9980336499960256e-05,
+      "loss": 1.2486,
+      "step": 4150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99802891050703e-05,
+      "loss": 0.9499,
+      "step": 4155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9980241653187527e-05,
+      "loss": 1.2259,
+      "step": 4160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9980194144312208e-05,
+      "loss": 1.2696,
+      "step": 4165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998014657844461e-05,
+      "loss": 1.2724,
+      "step": 4170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998009895558501e-05,
+      "loss": 1.1557,
+      "step": 4175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9980051275733678e-05,
+      "loss": 1.4106,
+      "step": 4180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9980003538890882e-05,
+      "loss": 1.4529,
+      "step": 4185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9979955745056898e-05,
+      "loss": 1.3072,
+      "step": 4190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9979907894232e-05,
+      "loss": 1.2709,
+      "step": 4195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997985998641646e-05,
+      "loss": 1.0486,
+      "step": 4200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9979812021610552e-05,
+      "loss": 1.4813,
+      "step": 4205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9979763999814554e-05,
+      "loss": 1.3347,
+      "step": 4210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9979715921028728e-05,
+      "loss": 1.0769,
+      "step": 4215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9979667785253363e-05,
+      "loss": 1.4072,
+      "step": 4220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997961959248872e-05,
+      "loss": 1.0356,
+      "step": 4225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997957134273509e-05,
+      "loss": 1.2314,
+      "step": 4230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9979523035992738e-05,
+      "loss": 1.1203,
+      "step": 4235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997947467226194e-05,
+      "loss": 1.0559,
+      "step": 4240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997942625154297e-05,
+      "loss": 1.1271,
+      "step": 4245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9979377773836115e-05,
+      "loss": 1.0846,
+      "step": 4250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9979329239141646e-05,
+      "loss": 1.2246,
+      "step": 4255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9979280647459834e-05,
+      "loss": 1.3035,
+      "step": 4260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9979231998790966e-05,
+      "loss": 1.1589,
+      "step": 4265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9979183293135315e-05,
+      "loss": 1.4335,
+      "step": 4270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997913453049316e-05,
+      "loss": 1.3865,
+      "step": 4275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997908571086478e-05,
+      "loss": 1.0439,
+      "step": 4280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9979036834250452e-05,
+      "loss": 1.2113,
+      "step": 4285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997898790065046e-05,
+      "loss": 1.1291,
+      "step": 4290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9978938910065077e-05,
+      "loss": 1.3361,
+      "step": 4295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9978889862494584e-05,
+      "loss": 1.4122,
+      "step": 4300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9978840757939266e-05,
+      "loss": 1.2408,
+      "step": 4305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99787915963994e-05,
+      "loss": 1.0518,
+      "step": 4310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9978742377875266e-05,
+      "loss": 1.4533,
+      "step": 4315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9978693102367143e-05,
+      "loss": 1.5114,
+      "step": 4320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997864376987532e-05,
+      "loss": 1.4088,
+      "step": 4325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9978594380400073e-05,
+      "loss": 1.3162,
+      "step": 4330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997854493394168e-05,
+      "loss": 1.5228,
+      "step": 4335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9978495430500435e-05,
+      "loss": 1.7127,
+      "step": 4340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9978445870076612e-05,
+      "loss": 1.2099,
+      "step": 4345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9978396252670498e-05,
+      "loss": 1.1237,
+      "step": 4350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997834657828237e-05,
+      "loss": 1.6053,
+      "step": 4355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997829684691252e-05,
+      "loss": 1.0953,
+      "step": 4360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9978247058561225e-05,
+      "loss": 1.1762,
+      "step": 4365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9978197213228774e-05,
+      "loss": 1.1322,
+      "step": 4370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9978147310915454e-05,
+      "loss": 0.9611,
+      "step": 4375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997809735162154e-05,
+      "loss": 1.4599,
+      "step": 4380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9978047335347326e-05,
+      "loss": 1.5014,
+      "step": 4385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9977997262093094e-05,
+      "loss": 1.1496,
+      "step": 4390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997794713185913e-05,
+      "loss": 1.1935,
+      "step": 4395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9977896944645725e-05,
+      "loss": 1.0851,
+      "step": 4400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997784670045316e-05,
+      "loss": 1.2865,
+      "step": 4405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9977796399281722e-05,
+      "loss": 1.3831,
+      "step": 4410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9977746041131702e-05,
+      "loss": 1.1635,
+      "step": 4415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9977695626003382e-05,
+      "loss": 1.0858,
+      "step": 4420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9977645153897057e-05,
+      "loss": 1.1829,
+      "step": 4425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9977594624813013e-05,
+      "loss": 1.1756,
+      "step": 4430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9977544038751534e-05,
+      "loss": 1.2263,
+      "step": 4435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9977493395712916e-05,
+      "loss": 1.3753,
+      "step": 4440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997744269569744e-05,
+      "loss": 1.4979,
+      "step": 4445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9977391938705404e-05,
+      "loss": 1.0983,
+      "step": 4450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9977341124737092e-05,
+      "loss": 1.1451,
+      "step": 4455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9977290253792794e-05,
+      "loss": 1.298,
+      "step": 4460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99772393258728e-05,
+      "loss": 1.3433,
+      "step": 4465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997718834097741e-05,
+      "loss": 1.5244,
+      "step": 4470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9977137299106904e-05,
+      "loss": 1.1198,
+      "step": 4475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997708620026158e-05,
+      "loss": 1.3835,
+      "step": 4480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9977035044441723e-05,
+      "loss": 1.3272,
+      "step": 4485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9976983831647633e-05,
+      "loss": 1.2499,
+      "step": 4490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99769325618796e-05,
+      "loss": 1.3696,
+      "step": 4495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9976881235137915e-05,
+      "loss": 1.34,
+      "step": 4500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9976829851422873e-05,
+      "loss": 1.335,
+      "step": 4505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9976778410734765e-05,
+      "loss": 1.1674,
+      "step": 4510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9976726913073886e-05,
+      "loss": 1.0679,
+      "step": 4515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997667535844053e-05,
+      "loss": 1.2386,
+      "step": 4520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9976623746834995e-05,
+      "loss": 1.1336,
+      "step": 4525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9976572078257572e-05,
+      "loss": 1.2589,
+      "step": 4530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9976520352708553e-05,
+      "loss": 1.2,
+      "step": 4535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997646857018824e-05,
+      "loss": 1.5099,
+      "step": 4540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9976416730696924e-05,
+      "loss": 1.274,
+      "step": 4545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9976364834234902e-05,
+      "loss": 0.9577,
+      "step": 4550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9976312880802474e-05,
+      "loss": 1.2224,
+      "step": 4555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997626087039993e-05,
+      "loss": 1.3162,
+      "step": 4560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9976208803027575e-05,
+      "loss": 0.9278,
+      "step": 4565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9976156678685698e-05,
+      "loss": 1.4359,
+      "step": 4570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9976104497374603e-05,
+      "loss": 0.9904,
+      "step": 4575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9976052259094584e-05,
+      "loss": 1.1493,
+      "step": 4580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9975999963845943e-05,
+      "loss": 1.0518,
+      "step": 4585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997594761162897e-05,
+      "loss": 1.3466,
+      "step": 4590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997589520244398e-05,
+      "loss": 1.0213,
+      "step": 4595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9975842736291258e-05,
+      "loss": 1.1967,
+      "step": 4600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997579021317111e-05,
+      "loss": 1.51,
+      "step": 4605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9975737633083833e-05,
+      "loss": 1.1989,
+      "step": 4610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9975684996029728e-05,
+      "loss": 1.4391,
+      "step": 4615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9975632302009096e-05,
+      "loss": 1.4779,
+      "step": 4620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9975579551022242e-05,
+      "loss": 1.1719,
+      "step": 4625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997552674306946e-05,
+      "loss": 1.2789,
+      "step": 4630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9975473878151056e-05,
+      "loss": 1.353,
+      "step": 4635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997542095626733e-05,
+      "loss": 1.3703,
+      "step": 4640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9975367977418587e-05,
+      "loss": 1.3891,
+      "step": 4645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9975314941605128e-05,
+      "loss": 1.1404,
+      "step": 4650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997526184882725e-05,
+      "loss": 1.3729,
+      "step": 4655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9975208699085263e-05,
+      "loss": 1.3944,
+      "step": 4660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9975155492379475e-05,
+      "loss": 1.3773,
+      "step": 4665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9975102228710178e-05,
+      "loss": 1.1609,
+      "step": 4670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997504890807769e-05,
+      "loss": 1.3005,
+      "step": 4675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99749955304823e-05,
+      "loss": 1.3502,
+      "step": 4680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9974942095924322e-05,
+      "loss": 1.6135,
+      "step": 4685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997488860440406e-05,
+      "loss": 1.2209,
+      "step": 4690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997483505592182e-05,
+      "loss": 1.1922,
+      "step": 4695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9974781450477904e-05,
+      "loss": 1.0312,
+      "step": 4700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9974727788072625e-05,
+      "loss": 1.5122,
+      "step": 4705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9974674068706284e-05,
+      "loss": 1.1508,
+      "step": 4710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9974620292379187e-05,
+      "loss": 1.2445,
+      "step": 4715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9974566459091644e-05,
+      "loss": 1.2697,
+      "step": 4720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9974512568843966e-05,
+      "loss": 1.1233,
+      "step": 4725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997445862163645e-05,
+      "loss": 1.2156,
+      "step": 4730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9974404617469414e-05,
+      "loss": 1.0042,
+      "step": 4735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9974350556343162e-05,
+      "loss": 1.5705,
+      "step": 4740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9974296438258003e-05,
+      "loss": 1.0592,
+      "step": 4745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997424226321425e-05,
+      "loss": 1.2995,
+      "step": 4750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9974188031212205e-05,
+      "loss": 1.1714,
+      "step": 4755
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9974133742252183e-05,
+      "loss": 1.1072,
+      "step": 4760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9974079396334493e-05,
+      "loss": 1.1184,
+      "step": 4765
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9974024993459446e-05,
+      "loss": 1.4242,
+      "step": 4770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9973970533627352e-05,
+      "loss": 1.6112,
+      "step": 4775
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9973916016838516e-05,
+      "loss": 1.0159,
+      "step": 4780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997386144309326e-05,
+      "loss": 1.1834,
+      "step": 4785
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9973806812391893e-05,
+      "loss": 1.0418,
+      "step": 4790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9973752124734723e-05,
+      "loss": 1.131,
+      "step": 4795
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997369738012206e-05,
+      "loss": 1.2118,
+      "step": 4800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9973642578554223e-05,
+      "loss": 1.1031,
+      "step": 4805
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9973587720031523e-05,
+      "loss": 0.9868,
+      "step": 4810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9973532804554274e-05,
+      "loss": 1.5991,
+      "step": 4815
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9973477832122787e-05,
+      "loss": 1.201,
+      "step": 4820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9973422802737377e-05,
+      "loss": 1.2568,
+      "step": 4825
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997336771639836e-05,
+      "loss": 1.5864,
+      "step": 4830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997331257310605e-05,
+      "loss": 1.3474,
+      "step": 4835
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997325737286076e-05,
+      "loss": 1.4705,
+      "step": 4840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9973202115662806e-05,
+      "loss": 1.1543,
+      "step": 4845
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9973146801512498e-05,
+      "loss": 1.245,
+      "step": 4850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9973091430410165e-05,
+      "loss": 1.0904,
+      "step": 4855
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9973036002356115e-05,
+      "loss": 1.2851,
+      "step": 4860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9972980517350664e-05,
+      "loss": 1.0242,
+      "step": 4865
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997292497539413e-05,
+      "loss": 1.6198,
+      "step": 4870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997286937648683e-05,
+      "loss": 1.0606,
+      "step": 4875
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9972813720629084e-05,
+      "loss": 1.5042,
+      "step": 4880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9972758007821205e-05,
+      "loss": 1.6481,
+      "step": 4885
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9972702238063517e-05,
+      "loss": 1.3054,
+      "step": 4890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9972646411356335e-05,
+      "loss": 1.5038,
+      "step": 4895
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9972590527699976e-05,
+      "loss": 1.188,
+      "step": 4900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9972534587094763e-05,
+      "loss": 1.1569,
+      "step": 4905
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9972478589541012e-05,
+      "loss": 1.088,
+      "step": 4910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9972422535039046e-05,
+      "loss": 1.2164,
+      "step": 4915
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9972366423589185e-05,
+      "loss": 1.2008,
+      "step": 4920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9972310255191748e-05,
+      "loss": 1.6633,
+      "step": 4925
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9972254029847052e-05,
+      "loss": 1.1073,
+      "step": 4930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9972197747555423e-05,
+      "loss": 1.5043,
+      "step": 4935
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9972141408317186e-05,
+      "loss": 1.3543,
+      "step": 4940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9972085012132653e-05,
+      "loss": 1.4145,
+      "step": 4945
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9972028559002153e-05,
+      "loss": 1.0818,
+      "step": 4950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9971972048926003e-05,
+      "loss": 1.5431,
+      "step": 4955
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9971915481904537e-05,
+      "loss": 1.1728,
+      "step": 4960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9971858857938062e-05,
+      "loss": 1.116,
+      "step": 4965
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9971802177026914e-05,
+      "loss": 1.031,
+      "step": 4970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997174543917141e-05,
+      "loss": 1.29,
+      "step": 4975
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997168864437188e-05,
+      "loss": 1.1452,
+      "step": 4980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997163179262864e-05,
+      "loss": 1.0471,
+      "step": 4985
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997157488394202e-05,
+      "loss": 1.1059,
+      "step": 4990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9971517918312345e-05,
+      "loss": 1.6864,
+      "step": 4995
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997146089573994e-05,
+      "loss": 1.3927,
+      "step": 5000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997140381622513e-05,
+      "loss": 1.0544,
+      "step": 5005
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9971346679768242e-05,
+      "loss": 1.1512,
+      "step": 5010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9971289486369598e-05,
+      "loss": 1.3838,
+      "step": 5015
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9971232236029533e-05,
+      "loss": 1.4578,
+      "step": 5020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9971174928748365e-05,
+      "loss": 1.3975,
+      "step": 5025
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9971117564526425e-05,
+      "loss": 1.4513,
+      "step": 5030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9971060143364043e-05,
+      "loss": 1.412,
+      "step": 5035
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997100266526154e-05,
+      "loss": 1.2362,
+      "step": 5040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9970945130219256e-05,
+      "loss": 1.2341,
+      "step": 5045
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9970887538237508e-05,
+      "loss": 1.2322,
+      "step": 5050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9970829889316633e-05,
+      "loss": 0.9462,
+      "step": 5055
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997077218345695e-05,
+      "loss": 1.0217,
+      "step": 5060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99707144206588e-05,
+      "loss": 1.0786,
+      "step": 5065
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9970656600922508e-05,
+      "loss": 1.3958,
+      "step": 5070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9970598724248402e-05,
+      "loss": 1.4809,
+      "step": 5075
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9970540790636813e-05,
+      "loss": 1.3428,
+      "step": 5080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997048280008808e-05,
+      "loss": 1.6386,
+      "step": 5085
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997042475260252e-05,
+      "loss": 1.0525,
+      "step": 5090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9970366648180475e-05,
+      "loss": 1.1677,
+      "step": 5095
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9970308486822276e-05,
+      "loss": 1.3317,
+      "step": 5100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997025026852825e-05,
+      "loss": 1.2356,
+      "step": 5105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997019199329873e-05,
+      "loss": 1.4494,
+      "step": 5110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9970133661134058e-05,
+      "loss": 1.6876,
+      "step": 5115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9970075272034557e-05,
+      "loss": 1.2673,
+      "step": 5120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997001682600056e-05,
+      "loss": 0.8063,
+      "step": 5125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.996995832303241e-05,
+      "loss": 1.0395,
+      "step": 5130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9969899763130435e-05,
+      "loss": 1.4742,
+      "step": 5135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.996984114629497e-05,
+      "loss": 1.2982,
+      "step": 5140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.996978247252635e-05,
+      "loss": 1.3976,
+      "step": 5145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.996972374182491e-05,
+      "loss": 1.2362,
+      "step": 5150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9969664954190978e-05,
+      "loss": 1.529,
+      "step": 5155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9969606109624907e-05,
+      "loss": 1.4069,
+      "step": 5160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9969547208127014e-05,
+      "loss": 0.9794,
+      "step": 5165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9969488249697652e-05,
+      "loss": 1.4116,
+      "step": 5170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.996942923433715e-05,
+      "loss": 1.1823,
+      "step": 5175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.996937016204584e-05,
+      "loss": 1.3332,
+      "step": 5180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9969311032824064e-05,
+      "loss": 1.2227,
+      "step": 5185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9969251846672163e-05,
+      "loss": 1.1508,
+      "step": 5190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.996919260359047e-05,
+      "loss": 1.5042,
+      "step": 5195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9969133303579327e-05,
+      "loss": 1.376,
+      "step": 5200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996907394663907e-05,
+      "loss": 1.2476,
+      "step": 5205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996901453277004e-05,
+      "loss": 1.2189,
+      "step": 5210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968955061972574e-05,
+      "loss": 1.0215,
+      "step": 5215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996889553424701e-05,
+      "loss": 1.2127,
+      "step": 5220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968835949593696e-05,
+      "loss": 1.0127,
+      "step": 5225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968776308012966e-05,
+      "loss": 1.2391,
+      "step": 5230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968716609505157e-05,
+      "loss": 1.5827,
+      "step": 5235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996865685407062e-05,
+      "loss": 1.2071,
+      "step": 5240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996859704170969e-05,
+      "loss": 1.373,
+      "step": 5245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968537172422706e-05,
+      "loss": 1.2419,
+      "step": 5250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968477246210015e-05,
+      "loss": 1.0524,
+      "step": 5255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968417263071957e-05,
+      "loss": 1.2297,
+      "step": 5260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968357223008877e-05,
+      "loss": 1.2133,
+      "step": 5265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968297126021113e-05,
+      "loss": 1.2877,
+      "step": 5270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968236972109012e-05,
+      "loss": 1.048,
+      "step": 5275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996817676127292e-05,
+      "loss": 1.1711,
+      "step": 5280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996811649351317e-05,
+      "loss": 1.358,
+      "step": 5285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968056168830118e-05,
+      "loss": 1.2344,
+      "step": 5290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967995787224102e-05,
+      "loss": 1.2336,
+      "step": 5295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996793534869547e-05,
+      "loss": 1.5991,
+      "step": 5300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967874853244563e-05,
+      "loss": 0.9858,
+      "step": 5305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967814300871734e-05,
+      "loss": 1.1042,
+      "step": 5310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996775369157732e-05,
+      "loss": 1.1883,
+      "step": 5315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967693025361674e-05,
+      "loss": 1.0804,
+      "step": 5320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996763230222514e-05,
+      "loss": 1.672,
+      "step": 5325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967571522168062e-05,
+      "loss": 1.2918,
+      "step": 5330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996751068519079e-05,
+      "loss": 1.306,
+      "step": 5335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967449791293675e-05,
+      "loss": 1.3036,
+      "step": 5340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967388840477056e-05,
+      "loss": 0.9101,
+      "step": 5345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967327832741287e-05,
+      "loss": 1.3005,
+      "step": 5350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967266768086717e-05,
+      "loss": 1.1275,
+      "step": 5355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967205646513694e-05,
+      "loss": 1.2017,
+      "step": 5360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996714446802256e-05,
+      "loss": 1.2756,
+      "step": 5365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9967083232613673e-05,
+      "loss": 1.1872,
+      "step": 5370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996702194028738e-05,
+      "loss": 1.7011,
+      "step": 5375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966960591044033e-05,
+      "loss": 1.1364,
+      "step": 5380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996689918488398e-05,
+      "loss": 1.2308,
+      "step": 5385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966837721807567e-05,
+      "loss": 1.1224,
+      "step": 5390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966776201815155e-05,
+      "loss": 1.1025,
+      "step": 5395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966714624907086e-05,
+      "loss": 1.4667,
+      "step": 5400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966652991083722e-05,
+      "loss": 1.4393,
+      "step": 5405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966591300345403e-05,
+      "loss": 1.423,
+      "step": 5410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996652955269249e-05,
+      "loss": 1.0338,
+      "step": 5415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966467748125333e-05,
+      "loss": 1.1355,
+      "step": 5420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966405886644283e-05,
+      "loss": 1.3711,
+      "step": 5425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966343968249698e-05,
+      "loss": 1.055,
+      "step": 5430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996628199294192e-05,
+      "loss": 1.3201,
+      "step": 5435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966219960721324e-05,
+      "loss": 1.3422,
+      "step": 5440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966157871588244e-05,
+      "loss": 1.2691,
+      "step": 5445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966095725543044e-05,
+      "loss": 1.1956,
+      "step": 5450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966033522586076e-05,
+      "loss": 1.2469,
+      "step": 5455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9965971262717696e-05,
+      "loss": 1.2403,
+      "step": 5460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996590894593826e-05,
+      "loss": 1.0554,
+      "step": 5465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9965846572248128e-05,
+      "loss": 1.25,
+      "step": 5470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9965784141647647e-05,
+      "loss": 1.6047,
+      "step": 5475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996572165413718e-05,
+      "loss": 1.2735,
+      "step": 5480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996565910971708e-05,
+      "loss": 1.3217,
+      "step": 5485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996559650838771e-05,
+      "loss": 1.1467,
+      "step": 5490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9965533850149423e-05,
+      "loss": 1.3463,
+      "step": 5495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996547113500258e-05,
+      "loss": 1.4209,
+      "step": 5500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996540836294753e-05,
+      "loss": 1.2232,
+      "step": 5505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9965345533984647e-05,
+      "loss": 1.0734,
+      "step": 5510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9965282648114274e-05,
+      "loss": 1.1066,
+      "step": 5515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996521970533678e-05,
+      "loss": 1.1829,
+      "step": 5520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9965156705652523e-05,
+      "loss": 1.496,
+      "step": 5525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996509364906186e-05,
+      "loss": 1.4234,
+      "step": 5530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996503053556515e-05,
+      "loss": 1.2742,
+      "step": 5535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996496736516276e-05,
+      "loss": 1.3211,
+      "step": 5540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964904137855043e-05,
+      "loss": 1.158,
+      "step": 5545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964840853642367e-05,
+      "loss": 1.3634,
+      "step": 5550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964777512525087e-05,
+      "loss": 1.1811,
+      "step": 5555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996471411450357e-05,
+      "loss": 1.2917,
+      "step": 5560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964650659578178e-05,
+      "loss": 1.3127,
+      "step": 5565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996458714774927e-05,
+      "loss": 1.2224,
+      "step": 5570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996452357901721e-05,
+      "loss": 1.1626,
+      "step": 5575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964459953382355e-05,
+      "loss": 1.4575,
+      "step": 5580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996439627084508e-05,
+      "loss": 1.1691,
+      "step": 5585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964332531405742e-05,
+      "loss": 1.4571,
+      "step": 5590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964268735064706e-05,
+      "loss": 1.0365,
+      "step": 5595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964204881822334e-05,
+      "loss": 1.4499,
+      "step": 5600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964140971679e-05,
+      "loss": 1.1818,
+      "step": 5605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964077004635054e-05,
+      "loss": 1.5356,
+      "step": 5610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9964012980690874e-05,
+      "loss": 1.2299,
+      "step": 5615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963948899846815e-05,
+      "loss": 0.9508,
+      "step": 5620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963884762103254e-05,
+      "loss": 0.8958,
+      "step": 5625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996382056746055e-05,
+      "loss": 1.2729,
+      "step": 5630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996375631591907e-05,
+      "loss": 1.1457,
+      "step": 5635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963692007479184e-05,
+      "loss": 1.3386,
+      "step": 5640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963627642141258e-05,
+      "loss": 1.1793,
+      "step": 5645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996356321990566e-05,
+      "loss": 1.3562,
+      "step": 5650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963498740772756e-05,
+      "loss": 1.0335,
+      "step": 5655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963434204742916e-05,
+      "loss": 1.0703,
+      "step": 5660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963369611816507e-05,
+      "loss": 1.166,
+      "step": 5665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963304961993895e-05,
+      "loss": 1.1446,
+      "step": 5670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996324025527546e-05,
+      "loss": 1.4216,
+      "step": 5675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963175491661562e-05,
+      "loss": 1.2974,
+      "step": 5680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996311067115257e-05,
+      "loss": 1.1062,
+      "step": 5685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9963045793748857e-05,
+      "loss": 1.204,
+      "step": 5690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99629808594508e-05,
+      "loss": 1.1925,
+      "step": 5695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996291586825876e-05,
+      "loss": 1.4824,
+      "step": 5700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996285082017311e-05,
+      "loss": 1.4834,
+      "step": 5705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962785715194228e-05,
+      "loss": 1.3063,
+      "step": 5710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962720553322476e-05,
+      "loss": 1.1229,
+      "step": 5715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962655334558236e-05,
+      "loss": 1.1361,
+      "step": 5720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962590058901872e-05,
+      "loss": 1.3414,
+      "step": 5725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996252472635376e-05,
+      "loss": 1.1151,
+      "step": 5730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962459336914275e-05,
+      "loss": 1.1797,
+      "step": 5735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996239389058379e-05,
+      "loss": 0.9871,
+      "step": 5740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996232838736268e-05,
+      "loss": 1.0204,
+      "step": 5745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996226282725131e-05,
+      "loss": 1.6194,
+      "step": 5750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962197210250063e-05,
+      "loss": 1.4696,
+      "step": 5755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962131536359312e-05,
+      "loss": 1.1322,
+      "step": 5760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962065805579433e-05,
+      "loss": 1.3365,
+      "step": 5765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99620000179108e-05,
+      "loss": 1.4797,
+      "step": 5770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961934173353786e-05,
+      "loss": 1.6696,
+      "step": 5775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961868271908776e-05,
+      "loss": 1.0315,
+      "step": 5780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961802313576135e-05,
+      "loss": 1.2439,
+      "step": 5785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961736298356247e-05,
+      "loss": 1.1405,
+      "step": 5790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961670226249487e-05,
+      "loss": 1.2531,
+      "step": 5795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996160409725623e-05,
+      "loss": 1.2925,
+      "step": 5800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996153791137686e-05,
+      "loss": 1.3767,
+      "step": 5805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961471668611745e-05,
+      "loss": 1.4075,
+      "step": 5810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961405368961272e-05,
+      "loss": 0.9549,
+      "step": 5815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961339012425816e-05,
+      "loss": 1.1612,
+      "step": 5820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961272599005754e-05,
+      "loss": 1.1761,
+      "step": 5825
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996120612870147e-05,
+      "loss": 1.3504,
+      "step": 5830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961139601513344e-05,
+      "loss": 1.222,
+      "step": 5835
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996107301744175e-05,
+      "loss": 1.3197,
+      "step": 5840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9961006376487074e-05,
+      "loss": 1.4285,
+      "step": 5845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960939678649693e-05,
+      "loss": 1.1754,
+      "step": 5850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960872923929987e-05,
+      "loss": 1.5409,
+      "step": 5855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996080611232834e-05,
+      "loss": 1.2224,
+      "step": 5860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960739243845136e-05,
+      "loss": 1.2878,
+      "step": 5865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960672318480746e-05,
+      "loss": 1.3791,
+      "step": 5870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960605336235566e-05,
+      "loss": 1.4147,
+      "step": 5875
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960538297109973e-05,
+      "loss": 1.1994,
+      "step": 5880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960471201104345e-05,
+      "loss": 1.1172,
+      "step": 5885
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960404048219072e-05,
+      "loss": 1.1292,
+      "step": 5890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960336838454534e-05,
+      "loss": 1.1318,
+      "step": 5895
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960269571811114e-05,
+      "loss": 1.6204,
+      "step": 5900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960202248289197e-05,
+      "loss": 1.5496,
+      "step": 5905
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996013486788917e-05,
+      "loss": 1.0939,
+      "step": 5910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960067430611417e-05,
+      "loss": 1.1681,
+      "step": 5915
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995999993645632e-05,
+      "loss": 1.2109,
+      "step": 5920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9959932385424266e-05,
+      "loss": 1.4554,
+      "step": 5925
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9959864777515642e-05,
+      "loss": 1.2293,
+      "step": 5930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9959797112730832e-05,
+      "loss": 1.5043,
+      "step": 5935
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9959729391070227e-05,
+      "loss": 1.3896,
+      "step": 5940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9959661612534208e-05,
+      "loss": 1.1835,
+      "step": 5945
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9959593777123163e-05,
+      "loss": 1.2247,
+      "step": 5950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995952588483748e-05,
+      "loss": 1.0806,
+      "step": 5955
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995945793567755e-05,
+      "loss": 1.2232,
+      "step": 5960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995938992964376e-05,
+      "loss": 1.2787,
+      "step": 5965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9959321866736493e-05,
+      "loss": 1.399,
+      "step": 5970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9959253746956145e-05,
+      "loss": 1.2473,
+      "step": 5975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9959185570303098e-05,
+      "loss": 1.1659,
+      "step": 5980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995911733677775e-05,
+      "loss": 1.278,
+      "step": 5985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995904904638048e-05,
+      "loss": 1.3304,
+      "step": 5990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9958980699111686e-05,
+      "loss": 1.2583,
+      "step": 5995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9958912294971752e-05,
+      "loss": 1.1896,
+      "step": 6000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9958843833961078e-05,
+      "loss": 1.1833,
+      "step": 6005
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9958775316080046e-05,
+      "loss": 1.1538,
+      "step": 6010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9958706741329052e-05,
+      "loss": 1.2472,
+      "step": 6015
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9958638109708483e-05,
+      "loss": 1.2662,
+      "step": 6020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995856942121874e-05,
+      "loss": 1.417,
+      "step": 6025
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9958500675860205e-05,
+      "loss": 1.2011,
+      "step": 6030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9958431873633275e-05,
+      "loss": 1.1943,
+      "step": 6035
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9958363014538345e-05,
+      "loss": 1.1989,
+      "step": 6040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9958294098575806e-05,
+      "loss": 1.3175,
+      "step": 6045
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9958225125746052e-05,
+      "loss": 1.1853,
+      "step": 6050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995815609604947e-05,
+      "loss": 1.4385,
+      "step": 6055
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995808700948647e-05,
+      "loss": 1.192,
+      "step": 6060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995801786605743e-05,
+      "loss": 1.1352,
+      "step": 6065
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9957948665762755e-05,
+      "loss": 1.6038,
+      "step": 6070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995787940860284e-05,
+      "loss": 1.1829,
+      "step": 6075
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9957810094578074e-05,
+      "loss": 1.2658,
+      "step": 6080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9957740723688856e-05,
+      "loss": 1.5859,
+      "step": 6085
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9957671295935586e-05,
+      "loss": 1.1984,
+      "step": 6090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9957601811318653e-05,
+      "loss": 1.3058,
+      "step": 6095
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9957532269838463e-05,
+      "loss": 1.1099,
+      "step": 6100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9957462671495404e-05,
+      "loss": 1.5892,
+      "step": 6105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9957393016289877e-05,
+      "loss": 1.0532,
+      "step": 6110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9957323304222284e-05,
+      "loss": 1.0907,
+      "step": 6115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9957253535293016e-05,
+      "loss": 1.2916,
+      "step": 6120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9957183709502477e-05,
+      "loss": 1.3049,
+      "step": 6125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9957113826851064e-05,
+      "loss": 1.1776,
+      "step": 6130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9957043887339172e-05,
+      "loss": 1.341,
+      "step": 6135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9956973890967206e-05,
+      "loss": 1.1144,
+      "step": 6140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9956903837735565e-05,
+      "loss": 1.2114,
+      "step": 6145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9956833727644648e-05,
+      "loss": 1.1048,
+      "step": 6150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9956763560694853e-05,
+      "loss": 1.0749,
+      "step": 6155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9956693336886585e-05,
+      "loss": 1.1051,
+      "step": 6160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995662305622024e-05,
+      "loss": 1.1401,
+      "step": 6165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9956552718696224e-05,
+      "loss": 1.5849,
+      "step": 6170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9956482324314938e-05,
+      "loss": 1.3859,
+      "step": 6175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995641187307678e-05,
+      "loss": 1.4893,
+      "step": 6180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9956341364982156e-05,
+      "loss": 1.0593,
+      "step": 6185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9956270800031467e-05,
+      "loss": 1.3655,
+      "step": 6190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9956200178225118e-05,
+      "loss": 1.1461,
+      "step": 6195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995612949956351e-05,
+      "loss": 1.297,
+      "step": 6200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995605876404705e-05,
+      "loss": 1.2962,
+      "step": 6205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9955987971676137e-05,
+      "loss": 1.0012,
+      "step": 6210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9955917122451182e-05,
+      "loss": 1.0342,
+      "step": 6215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995584621637258e-05,
+      "loss": 1.563,
+      "step": 6220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9955775253440746e-05,
+      "loss": 1.3017,
+      "step": 6225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995570423365608e-05,
+      "loss": 1.0231,
+      "step": 6230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9955633157018983e-05,
+      "loss": 1.3194,
+      "step": 6235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9955562023529872e-05,
+      "loss": 1.1934,
+      "step": 6240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995549083318914e-05,
+      "loss": 1.0737,
+      "step": 6245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9955419585997207e-05,
+      "loss": 1.2502,
+      "step": 6250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9955348281954473e-05,
+      "loss": 1.3754,
+      "step": 6255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9955276921061343e-05,
+      "loss": 1.108,
+      "step": 6260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9955205503318228e-05,
+      "loss": 1.1237,
+      "step": 6265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9955134028725538e-05,
+      "loss": 1.1293,
+      "step": 6270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9955062497283676e-05,
+      "loss": 1.4053,
+      "step": 6275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9954990908993052e-05,
+      "loss": 1.0375,
+      "step": 6280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995491926385408e-05,
+      "loss": 1.5491,
+      "step": 6285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9954847561867162e-05,
+      "loss": 1.5426,
+      "step": 6290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995477580303271e-05,
+      "loss": 1.2495,
+      "step": 6295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9954703987351133e-05,
+      "loss": 1.1676,
+      "step": 6300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9954632114822842e-05,
+      "loss": 1.225,
+      "step": 6305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995456018544825e-05,
+      "loss": 1.3512,
+      "step": 6310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995448819922776e-05,
+      "loss": 1.2532,
+      "step": 6315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9954416156161796e-05,
+      "loss": 1.3618,
+      "step": 6320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9954344056250757e-05,
+      "loss": 1.315,
+      "step": 6325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9954271899495058e-05,
+      "loss": 1.0796,
+      "step": 6330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9954199685895114e-05,
+      "loss": 1.3333,
+      "step": 6335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9954127415451337e-05,
+      "loss": 1.4634,
+      "step": 6340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995405508816414e-05,
+      "loss": 1.1555,
+      "step": 6345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995398270403393e-05,
+      "loss": 1.1373,
+      "step": 6350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995391026306113e-05,
+      "loss": 1.1367,
+      "step": 6355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9953837765246146e-05,
+      "loss": 1.2147,
+      "step": 6360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9953765210589395e-05,
+      "loss": 1.1398,
+      "step": 6365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9953692599091292e-05,
+      "loss": 1.0799,
+      "step": 6370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9953619930752247e-05,
+      "loss": 1.1428,
+      "step": 6375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9953547205572684e-05,
+      "loss": 1.2635,
+      "step": 6380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995347442355301e-05,
+      "loss": 1.3943,
+      "step": 6385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9953401584693647e-05,
+      "loss": 1.1137,
+      "step": 6390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9953328688995e-05,
+      "loss": 1.2733,
+      "step": 6395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99532557364575e-05,
+      "loss": 1.3856,
+      "step": 6400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9953182727081554e-05,
+      "loss": 1.169,
+      "step": 6405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995310966086758e-05,
+      "loss": 0.9639,
+      "step": 6410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9953036537815996e-05,
+      "loss": 1.3882,
+      "step": 6415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9952963357927222e-05,
+      "loss": 1.3106,
+      "step": 6420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9952890121201673e-05,
+      "loss": 1.187,
+      "step": 6425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995281682763977e-05,
+      "loss": 1.0942,
+      "step": 6430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9952743477241927e-05,
+      "loss": 1.075,
+      "step": 6435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9952670070008565e-05,
+      "loss": 1.1893,
+      "step": 6440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9952596605940108e-05,
+      "loss": 1.2152,
+      "step": 6445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995252308503697e-05,
+      "loss": 1.1213,
+      "step": 6450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9952449507299572e-05,
+      "loss": 1.5182,
+      "step": 6455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9952375872728333e-05,
+      "loss": 1.6794,
+      "step": 6460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9952302181323676e-05,
+      "loss": 1.3221,
+      "step": 6465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995222843308602e-05,
+      "loss": 1.0024,
+      "step": 6470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9952154628015786e-05,
+      "loss": 1.0936,
+      "step": 6475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99520807661134e-05,
+      "loss": 1.4082,
+      "step": 6480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9952006847379276e-05,
+      "loss": 1.781,
+      "step": 6485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995193287181384e-05,
+      "loss": 1.4318,
+      "step": 6490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9951858839417518e-05,
+      "loss": 1.1513,
+      "step": 6495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995178475019073e-05,
+      "loss": 1.5595,
+      "step": 6500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9951710604133895e-05,
+      "loss": 1.3409,
+      "step": 6505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995163640124744e-05,
+      "loss": 1.0906,
+      "step": 6510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9951562141531792e-05,
+      "loss": 1.2434,
+      "step": 6515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9951487824987368e-05,
+      "loss": 1.4434,
+      "step": 6520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99514134516146e-05,
+      "loss": 1.2569,
+      "step": 6525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99513390214139e-05,
+      "loss": 1.0329,
+      "step": 6530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9951264534385712e-05,
+      "loss": 1.363,
+      "step": 6535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9951189990530445e-05,
+      "loss": 1.1402,
+      "step": 6540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9951115389848534e-05,
+      "loss": 1.2105,
+      "step": 6545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9951040732340402e-05,
+      "loss": 1.4815,
+      "step": 6550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9950966018006473e-05,
+      "loss": 1.2417,
+      "step": 6555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995089124684718e-05,
+      "loss": 0.954,
+      "step": 6560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9950816418862945e-05,
+      "loss": 1.4464,
+      "step": 6565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9950741534054193e-05,
+      "loss": 1.3972,
+      "step": 6570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9950666592421354e-05,
+      "loss": 1.6351,
+      "step": 6575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995059159396486e-05,
+      "loss": 1.2027,
+      "step": 6580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9950516538685133e-05,
+      "loss": 1.2695,
+      "step": 6585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995044142658261e-05,
+      "loss": 0.9714,
+      "step": 6590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995036625765771e-05,
+      "loss": 1.5456,
+      "step": 6595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9950291031910868e-05,
+      "loss": 1.1639,
+      "step": 6600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995021574934251e-05,
+      "loss": 1.4013,
+      "step": 6605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995014040995307e-05,
+      "loss": 1.1702,
+      "step": 6610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9950065013742982e-05,
+      "loss": 1.1352,
+      "step": 6615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9949989560712667e-05,
+      "loss": 1.1473,
+      "step": 6620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994991405086256e-05,
+      "loss": 1.1384,
+      "step": 6625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994983848419309e-05,
+      "loss": 1.4301,
+      "step": 6630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9949762860704692e-05,
+      "loss": 1.1077,
+      "step": 6635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9949687180397795e-05,
+      "loss": 1.1929,
+      "step": 6640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9949611443272836e-05,
+      "loss": 1.371,
+      "step": 6645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994953564933024e-05,
+      "loss": 1.4427,
+      "step": 6650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994945979857045e-05,
+      "loss": 1.2238,
+      "step": 6655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994938389099389e-05,
+      "loss": 1.295,
+      "step": 6660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9949307926600994e-05,
+      "loss": 1.0197,
+      "step": 6665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9949231905392206e-05,
+      "loss": 1.0934,
+      "step": 6670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9949155827367948e-05,
+      "loss": 1.2674,
+      "step": 6675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9949079692528657e-05,
+      "loss": 1.4884,
+      "step": 6680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9949003500874773e-05,
+      "loss": 1.1694,
+      "step": 6685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9948927252406726e-05,
+      "loss": 1.2099,
+      "step": 6690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9948850947124955e-05,
+      "loss": 1.3106,
+      "step": 6695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9948774585029892e-05,
+      "loss": 1.1807,
+      "step": 6700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9948698166121977e-05,
+      "loss": 1.2125,
+      "step": 6705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9948621690401642e-05,
+      "loss": 1.1667,
+      "step": 6710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994854515786933e-05,
+      "loss": 1.3208,
+      "step": 6715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994846856852547e-05,
+      "loss": 1.5172,
+      "step": 6720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9948391922370506e-05,
+      "loss": 1.2449,
+      "step": 6725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9948315219404875e-05,
+      "loss": 1.0898,
+      "step": 6730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994823845962901e-05,
+      "loss": 0.9265,
+      "step": 6735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9948161643043354e-05,
+      "loss": 1.3229,
+      "step": 6740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994808476964835e-05,
+      "loss": 1.2611,
+      "step": 6745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994800783944442e-05,
+      "loss": 1.1523,
+      "step": 6750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9947930852432025e-05,
+      "loss": 1.2563,
+      "step": 6755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994785380861159e-05,
+      "loss": 1.1468,
+      "step": 6760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9947776707983558e-05,
+      "loss": 1.1008,
+      "step": 6765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9947699550548372e-05,
+      "loss": 1.7342,
+      "step": 6770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9947622336306472e-05,
+      "loss": 1.1187,
+      "step": 6775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99475450652583e-05,
+      "loss": 1.2341,
+      "step": 6780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994746773740429e-05,
+      "loss": 1.242,
+      "step": 6785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9947390352744893e-05,
+      "loss": 1.5798,
+      "step": 6790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9947312911280546e-05,
+      "loss": 1.2003,
+      "step": 6795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994723541301169e-05,
+      "loss": 1.1904,
+      "step": 6800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994715785793877e-05,
+      "loss": 1.1499,
+      "step": 6805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994708024606223e-05,
+      "loss": 1.2379,
+      "step": 6810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9947002577382512e-05,
+      "loss": 1.076,
+      "step": 6815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994692485190006e-05,
+      "loss": 1.2376,
+      "step": 6820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9946847069615316e-05,
+      "loss": 1.1311,
+      "step": 6825
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9946769230528723e-05,
+      "loss": 1.4454,
+      "step": 6830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994669133464073e-05,
+      "loss": 1.0395,
+      "step": 6835
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994661338195178e-05,
+      "loss": 1.1515,
+      "step": 6840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994653537246232e-05,
+      "loss": 1.7929,
+      "step": 6845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994645730617279e-05,
+      "loss": 1.2698,
+      "step": 6850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9946379183083636e-05,
+      "loss": 1.3858,
+      "step": 6855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9946301003195315e-05,
+      "loss": 1.4902,
+      "step": 6860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994622276650826e-05,
+      "loss": 1.2016,
+      "step": 6865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9946144473022926e-05,
+      "loss": 1.5883,
+      "step": 6870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9946066122739755e-05,
+      "loss": 1.0461,
+      "step": 6875
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99459877156592e-05,
+      "loss": 1.4086,
+      "step": 6880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9945909251781702e-05,
+      "loss": 1.3186,
+      "step": 6885
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9945830731107717e-05,
+      "loss": 1.1327,
+      "step": 6890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9945752153637685e-05,
+      "loss": 1.1081,
+      "step": 6895
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9945673519372062e-05,
+      "loss": 1.4327,
+      "step": 6900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994559482831129e-05,
+      "loss": 1.3934,
+      "step": 6905
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994551608045583e-05,
+      "loss": 1.2407,
+      "step": 6910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9945437275806116e-05,
+      "loss": 1.1061,
+      "step": 6915
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9945358414362608e-05,
+      "loss": 1.324,
+      "step": 6920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9945279496125757e-05,
+      "loss": 1.0626,
+      "step": 6925
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9945200521096008e-05,
+      "loss": 1.2368,
+      "step": 6930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9945121489273814e-05,
+      "loss": 1.378,
+      "step": 6935
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9945042400659633e-05,
+      "loss": 1.2554,
+      "step": 6940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9944963255253906e-05,
+      "loss": 1.2732,
+      "step": 6945
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9944884053057092e-05,
+      "loss": 1.1604,
+      "step": 6950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994480479406964e-05,
+      "loss": 1.1137,
+      "step": 6955
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9944725478292005e-05,
+      "loss": 1.7043,
+      "step": 6960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9944646105724638e-05,
+      "loss": 1.1166,
+      "step": 6965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9944566676367995e-05,
+      "loss": 1.1517,
+      "step": 6970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9944487190222522e-05,
+      "loss": 1.2208,
+      "step": 6975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9944407647288684e-05,
+      "loss": 1.3677,
+      "step": 6980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994432804756693e-05,
+      "loss": 1.6058,
+      "step": 6985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994424839105771e-05,
+      "loss": 1.4158,
+      "step": 6990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9944168677761484e-05,
+      "loss": 1.5373,
+      "step": 6995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9944088907678705e-05,
+      "loss": 1.1946,
+      "step": 7000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9944009080809835e-05,
+      "loss": 1.2238,
+      "step": 7005
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9943929197155324e-05,
+      "loss": 1.2021,
+      "step": 7010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9943849256715623e-05,
+      "loss": 1.1734,
+      "step": 7015
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9943769259491202e-05,
+      "loss": 1.1662,
+      "step": 7020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9943689205482505e-05,
+      "loss": 1.6865,
+      "step": 7025
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9943609094689997e-05,
+      "loss": 1.2436,
+      "step": 7030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9943528927114132e-05,
+      "loss": 1.1883,
+      "step": 7035
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9943448702755366e-05,
+      "loss": 1.2086,
+      "step": 7040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9943368421614165e-05,
+      "loss": 1.0342,
+      "step": 7045
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994328808369098e-05,
+      "loss": 1.3536,
+      "step": 7050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9943207688986274e-05,
+      "loss": 1.3724,
+      "step": 7055
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99431272375005e-05,
+      "loss": 1.2551,
+      "step": 7060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9943046729234125e-05,
+      "loss": 1.1496,
+      "step": 7065
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99429661641876e-05,
+      "loss": 1.3707,
+      "step": 7070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9942885542361394e-05,
+      "loss": 1.383,
+      "step": 7075
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9942804863755965e-05,
+      "loss": 1.3712,
+      "step": 7080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994272412837177e-05,
+      "loss": 1.3645,
+      "step": 7085
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9942643336209276e-05,
+      "loss": 1.238,
+      "step": 7090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994256248726894e-05,
+      "loss": 1.0374,
+      "step": 7095
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994248158155122e-05,
+      "loss": 1.1958,
+      "step": 7100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9942400619056585e-05,
+      "loss": 1.3863,
+      "step": 7105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99423195997855e-05,
+      "loss": 1.2911,
+      "step": 7110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9942238523738416e-05,
+      "loss": 1.5038,
+      "step": 7115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9942157390915808e-05,
+      "loss": 1.48,
+      "step": 7120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9942076201318127e-05,
+      "loss": 1.2132,
+      "step": 7125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9941994954945846e-05,
+      "loss": 1.194,
+      "step": 7130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994191365179943e-05,
+      "loss": 1.1422,
+      "step": 7135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9941832291879338e-05,
+      "loss": 1.6606,
+      "step": 7140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9941750875186035e-05,
+      "loss": 1.186,
+      "step": 7145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994166940171999e-05,
+      "loss": 1.417,
+      "step": 7150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9941587871481662e-05,
+      "loss": 1.2468,
+      "step": 7155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994150628447152e-05,
+      "loss": 1.1609,
+      "step": 7160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9941424640690033e-05,
+      "loss": 1.4513,
+      "step": 7165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9941342940137663e-05,
+      "loss": 1.1223,
+      "step": 7170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994126118281488e-05,
+      "loss": 1.2192,
+      "step": 7175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9941179368722147e-05,
+      "loss": 1.3183,
+      "step": 7180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9941097497859935e-05,
+      "loss": 1.0591,
+      "step": 7185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9941015570228705e-05,
+      "loss": 1.1651,
+      "step": 7190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9940933585828933e-05,
+      "loss": 1.0378,
+      "step": 7195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994085154466108e-05,
+      "loss": 1.2768,
+      "step": 7200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994076944672562e-05,
+      "loss": 1.1598,
+      "step": 7205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994068729202302e-05,
+      "loss": 1.0858,
+      "step": 7210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994060508055375e-05,
+      "loss": 1.2792,
+      "step": 7215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9940522812318274e-05,
+      "loss": 1.4857,
+      "step": 7220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994044048731707e-05,
+      "loss": 1.1366,
+      "step": 7225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9940358105550603e-05,
+      "loss": 1.1612,
+      "step": 7230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9940275667019345e-05,
+      "loss": 1.0718,
+      "step": 7235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9940193171723767e-05,
+      "loss": 1.3822,
+      "step": 7240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994011061966434e-05,
+      "loss": 1.4561,
+      "step": 7245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9940028010841536e-05,
+      "loss": 0.8226,
+      "step": 7250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9939945345255825e-05,
+      "loss": 1.9772,
+      "step": 7255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.993986262290768e-05,
+      "loss": 1.1992,
+      "step": 7260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.993977984379757e-05,
+      "loss": 1.2101,
+      "step": 7265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9939697007925972e-05,
+      "loss": 1.2377,
+      "step": 7270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.993961411529336e-05,
+      "loss": 1.5278,
+      "step": 7275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9939531165900208e-05,
+      "loss": 1.1686,
+      "step": 7280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993944815974698e-05,
+      "loss": 1.6076,
+      "step": 7285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9939365096834164e-05,
+      "loss": 1.075,
+      "step": 7290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9939281977162225e-05,
+      "loss": 1.1738,
+      "step": 7295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9939198800731638e-05,
+      "loss": 1.0693,
+      "step": 7300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993911556754288e-05,
+      "loss": 1.1188,
+      "step": 7305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9939032277596428e-05,
+      "loss": 1.1881,
+      "step": 7310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9938948930892758e-05,
+      "loss": 1.2835,
+      "step": 7315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993886552743234e-05,
+      "loss": 1.1937,
+      "step": 7320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9938782067215656e-05,
+      "loss": 1.0639,
+      "step": 7325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993869855024318e-05,
+      "loss": 1.0295,
+      "step": 7330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9938614976515392e-05,
+      "loss": 1.5339,
+      "step": 7335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9938531346032763e-05,
+      "loss": 1.2064,
+      "step": 7340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993844765879578e-05,
+      "loss": 1.3001,
+      "step": 7345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993836391480491e-05,
+      "loss": 1.3142,
+      "step": 7350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993828011406064e-05,
+      "loss": 1.2035,
+      "step": 7355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9938196256563445e-05,
+      "loss": 1.2606,
+      "step": 7360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99381123423138e-05,
+      "loss": 1.1925,
+      "step": 7365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993802837131219e-05,
+      "loss": 1.3397,
+      "step": 7370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99379443435591e-05,
+      "loss": 1.1514,
+      "step": 7375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937860259054996e-05,
+      "loss": 1.2315,
+      "step": 7380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937776117800364e-05,
+      "loss": 2.2284,
+      "step": 7385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993769191979569e-05,
+      "loss": 1.2659,
+      "step": 7390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937607665041447e-05,
+      "loss": 1.25,
+      "step": 7395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937523353538118e-05,
+      "loss": 1.2841,
+      "step": 7400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937438985286187e-05,
+      "loss": 1.1903,
+      "step": 7405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937354560286136e-05,
+      "loss": 1.4402,
+      "step": 7410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993727007853844e-05,
+      "loss": 1.4766,
+      "step": 7415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937185540043594e-05,
+      "loss": 1.2243,
+      "step": 7420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937100944802073e-05,
+      "loss": 1.3953,
+      "step": 7425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9937016292814356e-05,
+      "loss": 1.16,
+      "step": 7430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9936931584080937e-05,
+      "loss": 1.1636,
+      "step": 7435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9936846818602294e-05,
+      "loss": 1.2163,
+      "step": 7440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9936761996378905e-05,
+      "loss": 1.3479,
+      "step": 7445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9936677117411264e-05,
+      "loss": 1.2049,
+      "step": 7450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9936592181699853e-05,
+      "loss": 1.1304,
+      "step": 7455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993650718924516e-05,
+      "loss": 1.0837,
+      "step": 7460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993642214004766e-05,
+      "loss": 0.9178,
+      "step": 7465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9936337034107847e-05,
+      "loss": 1.1802,
+      "step": 7470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9936251871426205e-05,
+      "loss": 1.143,
+      "step": 7475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9936166652003223e-05,
+      "loss": 1.218,
+      "step": 7480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9936081375839383e-05,
+      "loss": 1.5877,
+      "step": 7485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935996042935174e-05,
+      "loss": 1.1198,
+      "step": 7490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935910653291085e-05,
+      "loss": 1.283,
+      "step": 7495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99358252069076e-05,
+      "loss": 1.369,
+      "step": 7500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993573970378521e-05,
+      "loss": 1.2556,
+      "step": 7505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935654143924403e-05,
+      "loss": 1.1592,
+      "step": 7510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993556852732567e-05,
+      "loss": 0.9577,
+      "step": 7515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935482853989493e-05,
+      "loss": 1.1952,
+      "step": 7520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935397123916364e-05,
+      "loss": 1.2271,
+      "step": 7525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935311337106776e-05,
+      "loss": 1.715,
+      "step": 7530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935225493561213e-05,
+      "loss": 1.036,
+      "step": 7535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935139593280174e-05,
+      "loss": 1.2308,
+      "step": 7540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993505363626414e-05,
+      "loss": 1.0756,
+      "step": 7545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9934967622513608e-05,
+      "loss": 1.3914,
+      "step": 7550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9934881552029067e-05,
+      "loss": 1.3492,
+      "step": 7555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993479542481101e-05,
+      "loss": 1.3293,
+      "step": 7560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9934709240859926e-05,
+      "loss": 1.0542,
+      "step": 7565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993462300017631e-05,
+      "loss": 1.3996,
+      "step": 7570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9934536702760653e-05,
+      "loss": 1.1559,
+      "step": 7575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993445034861345e-05,
+      "loss": 1.4729,
+      "step": 7580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993436393773519e-05,
+      "loss": 1.0419,
+      "step": 7585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993427747012637e-05,
+      "loss": 1.2199,
+      "step": 7590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993419094578748e-05,
+      "loss": 1.1439,
+      "step": 7595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9934104364719018e-05,
+      "loss": 0.9356,
+      "step": 7600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993401772692148e-05,
+      "loss": 1.2065,
+      "step": 7605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993393103239535e-05,
+      "loss": 1.1444,
+      "step": 7610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933844281141137e-05,
+      "loss": 1.3841,
+      "step": 7615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933757473159328e-05,
+      "loss": 1.1841,
+      "step": 7620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993367060845042e-05,
+      "loss": 1.2637,
+      "step": 7625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933583687014914e-05,
+      "loss": 1.0156,
+      "step": 7630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99334967088533e-05,
+      "loss": 1.5503,
+      "step": 7635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933409673966077e-05,
+      "loss": 1.2593,
+      "step": 7640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933322582353743e-05,
+      "loss": 1.3801,
+      "step": 7645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933235434016792e-05,
+      "loss": 1.1655,
+      "step": 7650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933148228955725e-05,
+      "loss": 1.1876,
+      "step": 7655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9933060967171043e-05,
+      "loss": 0.9031,
+      "step": 7660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9932973648663235e-05,
+      "loss": 1.1522,
+      "step": 7665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9932886273432808e-05,
+      "loss": 1.0644,
+      "step": 7670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9932798841480258e-05,
+      "loss": 1.3547,
+      "step": 7675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9932711352806082e-05,
+      "loss": 1.3009,
+      "step": 7680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993262380741078e-05,
+      "loss": 1.5365,
+      "step": 7685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9932536205294858e-05,
+      "loss": 1.2529,
+      "step": 7690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993244854645881e-05,
+      "loss": 1.0895,
+      "step": 7695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9932360830903142e-05,
+      "loss": 1.1147,
+      "step": 7700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993227305862835e-05,
+      "loss": 1.2183,
+      "step": 7705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9932185229634936e-05,
+      "loss": 1.087,
+      "step": 7710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9932097343923403e-05,
+      "loss": 1.3348,
+      "step": 7715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993200940149425e-05,
+      "loss": 1.3041,
+      "step": 7720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993192140234798e-05,
+      "loss": 1.3177,
+      "step": 7725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99318333464851e-05,
+      "loss": 1.2218,
+      "step": 7730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993174523390611e-05,
+      "loss": 1.0634,
+      "step": 7735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9931657064611514e-05,
+      "loss": 1.1771,
+      "step": 7740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993156883860181e-05,
+      "loss": 1.4081,
+      "step": 7745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993148055587751e-05,
+      "loss": 1.4065,
+      "step": 7750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9931392216439112e-05,
+      "loss": 1.3188,
+      "step": 7755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9931303820287125e-05,
+      "loss": 1.0134,
+      "step": 7760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993121536742205e-05,
+      "loss": 1.1995,
+      "step": 7765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9931126857844395e-05,
+      "loss": 1.2619,
+      "step": 7770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9931038291554663e-05,
+      "loss": 1.0921,
+      "step": 7775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993094966855336e-05,
+      "loss": 1.3671,
+      "step": 7780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930860988840994e-05,
+      "loss": 1.1639,
+      "step": 7785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993077225241807e-05,
+      "loss": 1.3143,
+      "step": 7790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930683459285095e-05,
+      "loss": 1.4544,
+      "step": 7795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993059460944258e-05,
+      "loss": 1.0795,
+      "step": 7800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930505702891025e-05,
+      "loss": 1.201,
+      "step": 7805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930416739630943e-05,
+      "loss": 0.9917,
+      "step": 7810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993032771966284e-05,
+      "loss": 1.4431,
+      "step": 7815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993023864298722e-05,
+      "loss": 1.4298,
+      "step": 7820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930149509604604e-05,
+      "loss": 1.2084,
+      "step": 7825
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9930060319515486e-05,
+      "loss": 1.1887,
+      "step": 7830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992997107272039e-05,
+      "loss": 1.3208,
+      "step": 7835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9929881769219813e-05,
+      "loss": 1.3702,
+      "step": 7840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9929792409014272e-05,
+      "loss": 1.4483,
+      "step": 7845
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9929702992104275e-05,
+      "loss": 1.4166,
+      "step": 7850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9929613518490333e-05,
+      "loss": 1.1367,
+      "step": 7855
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992952398817296e-05,
+      "loss": 1.5203,
+      "step": 7860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992943440115266e-05,
+      "loss": 1.5564,
+      "step": 7865
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9929344757429953e-05,
+      "loss": 1.135,
+      "step": 7870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9929255057005347e-05,
+      "loss": 0.9559,
+      "step": 7875
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992916529987935e-05,
+      "loss": 1.2023,
+      "step": 7880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992907548605248e-05,
+      "loss": 1.026,
+      "step": 7885
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992898561552525e-05,
+      "loss": 1.3372,
+      "step": 7890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992889568829817e-05,
+      "loss": 1.379,
+      "step": 7895
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992880570437176e-05,
+      "loss": 0.9671,
+      "step": 7900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9928715663746522e-05,
+      "loss": 1.2505,
+      "step": 7905
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9928625566422983e-05,
+      "loss": 1.2379,
+      "step": 7910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992853541240165e-05,
+      "loss": 1.2694,
+      "step": 7915
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992844520168304e-05,
+      "loss": 1.4525,
+      "step": 7920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992835493426767e-05,
+      "loss": 1.219,
+      "step": 7925
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992826461015605e-05,
+      "loss": 1.261,
+      "step": 7930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99281742293487e-05,
+      "loss": 1.3133,
+      "step": 7935
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9928083791846135e-05,
+      "loss": 1.2385,
+      "step": 7940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9927993297648872e-05,
+      "loss": 1.2258,
+      "step": 7945
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9927902746757428e-05,
+      "loss": 1.1477,
+      "step": 7950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992781213917232e-05,
+      "loss": 1.0622,
+      "step": 7955
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9927721474894063e-05,
+      "loss": 1.3702,
+      "step": 7960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992763075392318e-05,
+      "loss": 1.119,
+      "step": 7965
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9927539976260185e-05,
+      "loss": 1.4414,
+      "step": 7970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9927449141905594e-05,
+      "loss": 1.1502,
+      "step": 7975
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992735825085993e-05,
+      "loss": 1.1139,
+      "step": 7980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992726730312371e-05,
+      "loss": 1.1507,
+      "step": 7985
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9927176298697457e-05,
+      "loss": 1.1326,
+      "step": 7990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992708523758169e-05,
+      "loss": 1.4337,
+      "step": 7995
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992699411977692e-05,
+      "loss": 1.3623,
+      "step": 8000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9926902945283683e-05,
+      "loss": 1.3186,
+      "step": 8005
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9926811714102484e-05,
+      "loss": 1.161,
+      "step": 8010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9926720426233852e-05,
+      "loss": 0.9462,
+      "step": 8015
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992662908167831e-05,
+      "loss": 1.8678,
+      "step": 8020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9926537680436374e-05,
+      "loss": 1.3106,
+      "step": 8025
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992644622250857e-05,
+      "loss": 1.1872,
+      "step": 8030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9926354707895423e-05,
+      "loss": 1.0467,
+      "step": 8035
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9926263136597447e-05,
+      "loss": 1.1398,
+      "step": 8040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992617150861517e-05,
+      "loss": 1.3976,
+      "step": 8045
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9926079823949116e-05,
+      "loss": 1.0335,
+      "step": 8050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9925988082599806e-05,
+      "loss": 1.1376,
+      "step": 8055
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9925896284567767e-05,
+      "loss": 1.4181,
+      "step": 8060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9925804429853523e-05,
+      "loss": 1.129,
+      "step": 8065
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9925712518457596e-05,
+      "loss": 1.1242,
+      "step": 8070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9925620550380507e-05,
+      "loss": 1.4343,
+      "step": 8075
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992552852562279e-05,
+      "loss": 1.2419,
+      "step": 8080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9925436444184968e-05,
+      "loss": 1.3173,
+      "step": 8085
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9925344306067564e-05,
+      "loss": 1.5075,
+      "step": 8090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9925252111271106e-05,
+      "loss": 1.6082,
+      "step": 8095
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992515985979612e-05,
+      "loss": 1.0558,
+      "step": 8100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9925067551643132e-05,
+      "loss": 1.2614,
+      "step": 8105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992497518681267e-05,
+      "loss": 1.2244,
+      "step": 8110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9924882765305266e-05,
+      "loss": 1.0856,
+      "step": 8115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9924790287121437e-05,
+      "loss": 1.3258,
+      "step": 8120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992469775226172e-05,
+      "loss": 1.1668,
+      "step": 8125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9924605160726642e-05,
+      "loss": 1.3219,
+      "step": 8130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992451251251673e-05,
+      "loss": 1.3516,
+      "step": 8135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9924419807632512e-05,
+      "loss": 1.1059,
+      "step": 8140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9924327046074522e-05,
+      "loss": 1.0697,
+      "step": 8145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9924234227843285e-05,
+      "loss": 1.0761,
+      "step": 8150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9924141352939334e-05,
+      "loss": 1.3289,
+      "step": 8155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9924048421363196e-05,
+      "loss": 1.1804,
+      "step": 8160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9923955433115405e-05,
+      "loss": 1.2253,
+      "step": 8165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9923862388196487e-05,
+      "loss": 1.3292,
+      "step": 8170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9923769286606983e-05,
+      "loss": 1.3393,
+      "step": 8175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9923676128347414e-05,
+      "loss": 1.0623,
+      "step": 8180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9923582913418322e-05,
+      "loss": 1.4196,
+      "step": 8185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9923489641820232e-05,
+      "loss": 1.1605,
+      "step": 8190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992339631355368e-05,
+      "loss": 1.0015,
+      "step": 8195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9923302928619196e-05,
+      "loss": 1.1405,
+      "step": 8200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9923209487017315e-05,
+      "loss": 1.4699,
+      "step": 8205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992311598874857e-05,
+      "loss": 1.1213,
+      "step": 8210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9923022433813496e-05,
+      "loss": 1.049,
+      "step": 8215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9922928822212627e-05,
+      "loss": 1.5005,
+      "step": 8220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9922835153946497e-05,
+      "loss": 1.4258,
+      "step": 8225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9922741429015642e-05,
+      "loss": 1.3227,
+      "step": 8230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9922647647420597e-05,
+      "loss": 1.3136,
+      "step": 8235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9922553809161897e-05,
+      "loss": 1.4117,
+      "step": 8240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9922459914240074e-05,
+      "loss": 1.2011,
+      "step": 8245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992236596265567e-05,
+      "loss": 1.4748,
+      "step": 8250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9922271954409225e-05,
+      "loss": 1.5292,
+      "step": 8255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9922177889501262e-05,
+      "loss": 1.2536,
+      "step": 8260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9922083767932335e-05,
+      "loss": 1.0452,
+      "step": 8265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992198958970297e-05,
+      "loss": 1.3372,
+      "step": 8270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9921895354813702e-05,
+      "loss": 1.2187,
+      "step": 8275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992180106326508e-05,
+      "loss": 1.0195,
+      "step": 8280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9921706715057637e-05,
+      "loss": 1.1309,
+      "step": 8285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9921612310191914e-05,
+      "loss": 1.286,
+      "step": 8290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9921517848668445e-05,
+      "loss": 1.4131,
+      "step": 8295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9921423330487775e-05,
+      "loss": 1.5204,
+      "step": 8300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992132875565044e-05,
+      "loss": 1.1149,
+      "step": 8305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9921234124156986e-05,
+      "loss": 1.2438,
+      "step": 8310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9921139436007944e-05,
+      "loss": 1.2451,
+      "step": 8315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992104469120386e-05,
+      "loss": 1.1362,
+      "step": 8320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9920949889745277e-05,
+      "loss": 1.2107,
+      "step": 8325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9920855031632735e-05,
+      "loss": 1.4883,
+      "step": 8330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992076011686677e-05,
+      "loss": 1.2439,
+      "step": 8335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9920665145447935e-05,
+      "loss": 1.0478,
+      "step": 8340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9920570117376762e-05,
+      "loss": 1.0548,
+      "step": 8345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99204750326538e-05,
+      "loss": 1.2046,
+      "step": 8350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9920379891279586e-05,
+      "loss": 1.3496,
+      "step": 8355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992028469325467e-05,
+      "loss": 1.3479,
+      "step": 8360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9920189438579592e-05,
+      "loss": 1.1536,
+      "step": 8365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99200941272549e-05,
+      "loss": 1.297,
+      "step": 8370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991999875928113e-05,
+      "loss": 1.2017,
+      "step": 8375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9919903334658835e-05,
+      "loss": 1.275,
+      "step": 8380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9919807853388555e-05,
+      "loss": 1.2353,
+      "step": 8385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991971231547084e-05,
+      "loss": 1.2574,
+      "step": 8390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9919616720906226e-05,
+      "loss": 1.1946,
+      "step": 8395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991952106969527e-05,
+      "loss": 1.3791,
+      "step": 8400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9919425361838512e-05,
+      "loss": 1.1961,
+      "step": 8405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9919329597336498e-05,
+      "loss": 1.028,
+      "step": 8410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991923377618978e-05,
+      "loss": 1.4459,
+      "step": 8415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99191378983989e-05,
+      "loss": 1.4668,
+      "step": 8420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9919041963964408e-05,
+      "loss": 1.1262,
+      "step": 8425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9918945972886854e-05,
+      "loss": 1.105,
+      "step": 8430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991884992516678e-05,
+      "loss": 1.4007,
+      "step": 8435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9918753820804743e-05,
+      "loss": 1.3218,
+      "step": 8440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991865765980128e-05,
+      "loss": 1.1041,
+      "step": 8445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991856144215695e-05,
+      "loss": 1.4142,
+      "step": 8450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9918465167872305e-05,
+      "loss": 1.6902,
+      "step": 8455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991836883694788e-05,
+      "loss": 1.409,
+      "step": 8460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991827244938424e-05,
+      "loss": 1.2541,
+      "step": 8465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9918176005181927e-05,
+      "loss": 1.3187,
+      "step": 8470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9918079504341497e-05,
+      "loss": 1.2607,
+      "step": 8475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9917982946863497e-05,
+      "loss": 1.2298,
+      "step": 8480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991788633274848e-05,
+      "loss": 1.4388,
+      "step": 8485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9917789661996998e-05,
+      "loss": 1.1381,
+      "step": 8490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9917692934609604e-05,
+      "loss": 1.2648,
+      "step": 8495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9917596150586845e-05,
+      "loss": 1.0814,
+      "step": 8500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9917499309929282e-05,
+      "loss": 1.335,
+      "step": 8505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9917402412637463e-05,
+      "loss": 1.1412,
+      "step": 8510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9917305458711943e-05,
+      "loss": 1.1389,
+      "step": 8515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9917208448153272e-05,
+      "loss": 1.3332,
+      "step": 8520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991711138096201e-05,
+      "loss": 1.1668,
+      "step": 8525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9917014257138702e-05,
+      "loss": 1.072,
+      "step": 8530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9916917076683914e-05,
+      "loss": 1.123,
+      "step": 8535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9916819839598194e-05,
+      "loss": 1.5237,
+      "step": 8540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99167225458821e-05,
+      "loss": 1.2122,
+      "step": 8545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9916625195536187e-05,
+      "loss": 1.2521,
+      "step": 8550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9916527788561008e-05,
+      "loss": 1.2081,
+      "step": 8555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9916430324957123e-05,
+      "loss": 1.2293,
+      "step": 8560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9916332804725087e-05,
+      "loss": 1.4103,
+      "step": 8565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991623522786546e-05,
+      "loss": 1.1434,
+      "step": 8570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991613759437879e-05,
+      "loss": 1.1555,
+      "step": 8575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9916039904265646e-05,
+      "loss": 0.9668,
+      "step": 8580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991594215752658e-05,
+      "loss": 1.0305,
+      "step": 8585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991584435416215e-05,
+      "loss": 1.1556,
+      "step": 8590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9915746494172915e-05,
+      "loss": 1.3302,
+      "step": 8595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9915648577559438e-05,
+      "loss": 1.5979,
+      "step": 8600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991555060432227e-05,
+      "loss": 1.0153,
+      "step": 8605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9915452574461978e-05,
+      "loss": 1.3899,
+      "step": 8610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9915354487979116e-05,
+      "loss": 1.1404,
+      "step": 8615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991525634487425e-05,
+      "loss": 1.3342,
+      "step": 8620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9915158145147936e-05,
+      "loss": 1.1493,
+      "step": 8625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9915059888800738e-05,
+      "loss": 1.2553,
+      "step": 8630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991496157583321e-05,
+      "loss": 1.2416,
+      "step": 8635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991486320624592e-05,
+      "loss": 0.9251,
+      "step": 8640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9914764780039434e-05,
+      "loss": 1.4921,
+      "step": 8645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99146662972143e-05,
+      "loss": 1.329,
+      "step": 8650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9914567757771094e-05,
+      "loss": 1.1772,
+      "step": 8655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9914469161710373e-05,
+      "loss": 1.6707,
+      "step": 8660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99143705090327e-05,
+      "loss": 1.5301,
+      "step": 8665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991427179973864e-05,
+      "loss": 1.4601,
+      "step": 8670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9914173033828757e-05,
+      "loss": 1.5826,
+      "step": 8675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991407421130361e-05,
+      "loss": 1.2054,
+      "step": 8680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9913975332163765e-05,
+      "loss": 0.9933,
+      "step": 8685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9913876396409793e-05,
+      "loss": 1.4174,
+      "step": 8690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9913777404042257e-05,
+      "loss": 1.4527,
+      "step": 8695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9913678355061716e-05,
+      "loss": 0.9797,
+      "step": 8700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9913579249468738e-05,
+      "loss": 1.1187,
+      "step": 8705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9913480087263896e-05,
+      "loss": 1.1622,
+      "step": 8710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9913380868447746e-05,
+      "loss": 1.372,
+      "step": 8715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9913281593020857e-05,
+      "loss": 1.1458,
+      "step": 8720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9913182260983802e-05,
+      "loss": 0.9477,
+      "step": 8725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9913082872337147e-05,
+      "loss": 1.1992,
+      "step": 8730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9912983427081453e-05,
+      "loss": 1.109,
+      "step": 8735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991288392521729e-05,
+      "loss": 1.19,
+      "step": 8740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9912784366745232e-05,
+      "loss": 0.9737,
+      "step": 8745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9912684751665846e-05,
+      "loss": 1.2151,
+      "step": 8750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9912585079979692e-05,
+      "loss": 1.1372,
+      "step": 8755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991248535168735e-05,
+      "loss": 1.229,
+      "step": 8760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9912385566789384e-05,
+      "loss": 1.1491,
+      "step": 8765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9912285725286365e-05,
+      "loss": 1.215,
+      "step": 8770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9912185827178865e-05,
+      "loss": 1.2611,
+      "step": 8775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991208587246745e-05,
+      "loss": 1.2349,
+      "step": 8780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9911985861152696e-05,
+      "loss": 0.9255,
+      "step": 8785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991188579323517e-05,
+      "loss": 1.1975,
+      "step": 8790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9911785668715443e-05,
+      "loss": 1.1261,
+      "step": 8795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9911685487594093e-05,
+      "loss": 1.5615,
+      "step": 8800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9911585249871687e-05,
+      "loss": 1.3382,
+      "step": 8805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9911484955548795e-05,
+      "loss": 1.1484,
+      "step": 8810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9911384604625996e-05,
+      "loss": 1.2801,
+      "step": 8815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991128419710386e-05,
+      "loss": 1.2171,
+      "step": 8820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9911183732982956e-05,
+      "loss": 1.2343,
+      "step": 8825
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9911083212263868e-05,
+      "loss": 1.2113,
+      "step": 8830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991098263494716e-05,
+      "loss": 1.5881,
+      "step": 8835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991088200103341e-05,
+      "loss": 1.1136,
+      "step": 8840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9910781310523197e-05,
+      "loss": 1.297,
+      "step": 8845
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991068056341709e-05,
+      "loss": 1.2344,
+      "step": 8850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9910579759715666e-05,
+      "loss": 1.3502,
+      "step": 8855
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9910478899419504e-05,
+      "loss": 1.2061,
+      "step": 8860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9910377982529174e-05,
+      "loss": 1.0848,
+      "step": 8865
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9910277009045257e-05,
+      "loss": 1.2082,
+      "step": 8870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991017597896833e-05,
+      "loss": 1.1069,
+      "step": 8875
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9910074892298964e-05,
+      "loss": 1.3378,
+      "step": 8880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990997374903774e-05,
+      "loss": 1.0772,
+      "step": 8885
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9909872549185237e-05,
+      "loss": 1.5293,
+      "step": 8890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9909771292742035e-05,
+      "loss": 1.1302,
+      "step": 8895
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9909669979708707e-05,
+      "loss": 1.1142,
+      "step": 8900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990956861008583e-05,
+      "loss": 1.3414,
+      "step": 8905
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990946718387399e-05,
+      "loss": 1.24,
+      "step": 8910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990936570107376e-05,
+      "loss": 1.0873,
+      "step": 8915
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9909264161685726e-05,
+      "loss": 1.3349,
+      "step": 8920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990916256571046e-05,
+      "loss": 1.1564,
+      "step": 8925
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9909060913148547e-05,
+      "loss": 1.2547,
+      "step": 8930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990895920400057e-05,
+      "loss": 1.1854,
+      "step": 8935
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9908857438267103e-05,
+      "loss": 1.2156,
+      "step": 8940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990875561594873e-05,
+      "loss": 1.4054,
+      "step": 8945
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9908653737046034e-05,
+      "loss": 0.9711,
+      "step": 8950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9908551801559594e-05,
+      "loss": 1.6443,
+      "step": 8955
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990844980949e-05,
+      "loss": 1.1181,
+      "step": 8960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9908347760837823e-05,
+      "loss": 1.2563,
+      "step": 8965
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9908245655603654e-05,
+      "loss": 1.1232,
+      "step": 8970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990814349378807e-05,
+      "loss": 1.4911,
+      "step": 8975
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990804127539166e-05,
+      "loss": 1.1076,
+      "step": 8980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9907939000415002e-05,
+      "loss": 1.1196,
+      "step": 8985
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9907836668858684e-05,
+      "loss": 1.0957,
+      "step": 8990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9907734280723292e-05,
+      "loss": 1.2863,
+      "step": 8995
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9907631836009407e-05,
+      "loss": 1.2681,
+      "step": 9000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9907529334717613e-05,
+      "loss": 1.1027,
+      "step": 9005
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99074267768485e-05,
+      "loss": 1.3326,
+      "step": 9010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9907324162402654e-05,
+      "loss": 1.4712,
+      "step": 9015
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9907221491380657e-05,
+      "loss": 1.2738,
+      "step": 9020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9907118763783097e-05,
+      "loss": 1.3904,
+      "step": 9025
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990701597961056e-05,
+      "loss": 1.1101,
+      "step": 9030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990691313886363e-05,
+      "loss": 1.4674,
+      "step": 9035
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99068102415429e-05,
+      "loss": 1.4758,
+      "step": 9040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9906707287648953e-05,
+      "loss": 1.0981,
+      "step": 9045
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9906604277182385e-05,
+      "loss": 1.1525,
+      "step": 9050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9906501210143774e-05,
+      "loss": 1.4813,
+      "step": 9055
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9906398086533713e-05,
+      "loss": 1.2374,
+      "step": 9060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990629490635279e-05,
+      "loss": 1.3107,
+      "step": 9065
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9906191669601596e-05,
+      "loss": 1.0175,
+      "step": 9070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990608837628072e-05,
+      "loss": 1.5528,
+      "step": 9075
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990598502639075e-05,
+      "loss": 1.1764,
+      "step": 9080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990588161993228e-05,
+      "loss": 1.1661,
+      "step": 9085
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9905778156905896e-05,
+      "loss": 1.4063,
+      "step": 9090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990567463731219e-05,
+      "loss": 1.2087,
+      "step": 9095
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9905571061151756e-05,
+      "loss": 1.6237,
+      "step": 9100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9905467428425183e-05,
+      "loss": 1.2059,
+      "step": 9105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9905363739133062e-05,
+      "loss": 1.1064,
+      "step": 9110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990525999327599e-05,
+      "loss": 1.1228,
+      "step": 9115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9905156190854555e-05,
+      "loss": 1.5409,
+      "step": 9120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9905052331869344e-05,
+      "loss": 1.2802,
+      "step": 9125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9904948416320964e-05,
+      "loss": 1.1927,
+      "step": 9130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990484444421e-05,
+      "loss": 1.2997,
+      "step": 9135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9904740415537046e-05,
+      "loss": 1.464,
+      "step": 9140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9904636330302694e-05,
+      "loss": 1.2711,
+      "step": 9145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9904532188507545e-05,
+      "loss": 1.1944,
+      "step": 9150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990442799015219e-05,
+      "loss": 1.0469,
+      "step": 9155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990432373523722e-05,
+      "loss": 1.0793,
+      "step": 9160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990421942376324e-05,
+      "loss": 1.2959,
+      "step": 9165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9904115055730837e-05,
+      "loss": 1.2139,
+      "step": 9170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9904010631140612e-05,
+      "loss": 1.4216,
+      "step": 9175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9903906149993157e-05,
+      "loss": 0.9439,
+      "step": 9180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9903801612289072e-05,
+      "loss": 1.02,
+      "step": 9185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9903697018028953e-05,
+      "loss": 0.9271,
+      "step": 9190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99035923672134e-05,
+      "loss": 1.226,
+      "step": 9195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9903487659843004e-05,
+      "loss": 1.0381,
+      "step": 9200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9903382895918367e-05,
+      "loss": 1.104,
+      "step": 9205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9903278075440088e-05,
+      "loss": 1.4177,
+      "step": 9210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9903173198408766e-05,
+      "loss": 1.4909,
+      "step": 9215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9903068264825e-05,
+      "loss": 1.4805,
+      "step": 9220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990296327468938e-05,
+      "loss": 1.0773,
+      "step": 9225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990285822800252e-05,
+      "loss": 1.3432,
+      "step": 9230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9902753124765012e-05,
+      "loss": 1.288,
+      "step": 9235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990264796497746e-05,
+      "loss": 1.2599,
+      "step": 9240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9902542748640458e-05,
+      "loss": 1.3901,
+      "step": 9245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9902437475754614e-05,
+      "loss": 1.3821,
+      "step": 9250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9902332146320522e-05,
+      "loss": 1.0754,
+      "step": 9255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990222676033879e-05,
+      "loss": 1.3713,
+      "step": 9260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990212131781002e-05,
+      "loss": 1.2233,
+      "step": 9265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9902015818734804e-05,
+      "loss": 1.5097,
+      "step": 9270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9901910263113757e-05,
+      "loss": 1.3623,
+      "step": 9275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9901804650947477e-05,
+      "loss": 1.1031,
+      "step": 9280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9901698982236567e-05,
+      "loss": 1.2577,
+      "step": 9285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9901593256981628e-05,
+      "loss": 1.4975,
+      "step": 9290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9901487475183264e-05,
+      "loss": 0.9557,
+      "step": 9295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9901381636842084e-05,
+      "loss": 1.0203,
+      "step": 9300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990127574195869e-05,
+      "loss": 1.1608,
+      "step": 9305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990116979053369e-05,
+      "loss": 1.0864,
+      "step": 9310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990106378256768e-05,
+      "loss": 0.9863,
+      "step": 9315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9900957718061268e-05,
+      "loss": 0.9859,
+      "step": 9320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9900851597015066e-05,
+      "loss": 1.1729,
+      "step": 9325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9900745419429676e-05,
+      "loss": 1.2107,
+      "step": 9330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99006391853057e-05,
+      "loss": 1.4476,
+      "step": 9335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9900532894643752e-05,
+      "loss": 1.2523,
+      "step": 9340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990042654744444e-05,
+      "loss": 1.3041,
+      "step": 9345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9900320143708363e-05,
+      "loss": 1.205,
+      "step": 9350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9900213683436135e-05,
+      "loss": 1.2128,
+      "step": 9355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990010716662836e-05,
+      "loss": 1.7896,
+      "step": 9360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.990000059328565e-05,
+      "loss": 1.1702,
+      "step": 9365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9899893963408613e-05,
+      "loss": 1.1623,
+      "step": 9370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9899787276997853e-05,
+      "loss": 1.2504,
+      "step": 9375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9899680534053986e-05,
+      "loss": 1.2068,
+      "step": 9380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989957373457762e-05,
+      "loss": 1.5042,
+      "step": 9385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989946687856936e-05,
+      "loss": 1.3911,
+      "step": 9390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989935996602982e-05,
+      "loss": 1.1025,
+      "step": 9395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989925299695961e-05,
+      "loss": 1.0655,
+      "step": 9400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9899145971359342e-05,
+      "loss": 1.6679,
+      "step": 9405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989903888922963e-05,
+      "loss": 1.1893,
+      "step": 9410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989893175057108e-05,
+      "loss": 1.412,
+      "step": 9415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9898824555384303e-05,
+      "loss": 1.149,
+      "step": 9420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9898717303669914e-05,
+      "loss": 1.1915,
+      "step": 9425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9898609995428523e-05,
+      "loss": 1.1518,
+      "step": 9430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989850263066075e-05,
+      "loss": 1.215,
+      "step": 9435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98983952093672e-05,
+      "loss": 1.4757,
+      "step": 9440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989828773154849e-05,
+      "loss": 1.2867,
+      "step": 9445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9898180197205232e-05,
+      "loss": 1.3449,
+      "step": 9450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9898072606338044e-05,
+      "loss": 0.9218,
+      "step": 9455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9897964958947534e-05,
+      "loss": 1.4454,
+      "step": 9460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9897857255034326e-05,
+      "loss": 1.2284,
+      "step": 9465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9897749494599026e-05,
+      "loss": 1.1128,
+      "step": 9470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9897641677642254e-05,
+      "loss": 1.0175,
+      "step": 9475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9897533804164624e-05,
+      "loss": 1.1218,
+      "step": 9480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9897425874166752e-05,
+      "loss": 0.9255,
+      "step": 9485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9897317887649254e-05,
+      "loss": 1.3486,
+      "step": 9490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989720984461275e-05,
+      "loss": 1.4,
+      "step": 9495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9897101745057854e-05,
+      "loss": 1.4699,
+      "step": 9500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989699358898518e-05,
+      "loss": 1.2206,
+      "step": 9505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9896885376395355e-05,
+      "loss": 1.2862,
+      "step": 9510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989677710728899e-05,
+      "loss": 1.6564,
+      "step": 9515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9896668781666704e-05,
+      "loss": 1.1663,
+      "step": 9520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9896560399529113e-05,
+      "loss": 1.3425,
+      "step": 9525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9896451960876843e-05,
+      "loss": 1.3767,
+      "step": 9530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989634346571051e-05,
+      "loss": 1.2799,
+      "step": 9535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9896234914030733e-05,
+      "loss": 1.6881,
+      "step": 9540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989612630583813e-05,
+      "loss": 1.2209,
+      "step": 9545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9896017641133323e-05,
+      "loss": 1.237,
+      "step": 9550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9895908919916933e-05,
+      "loss": 1.4765,
+      "step": 9555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9895800142189585e-05,
+      "loss": 1.7088,
+      "step": 9560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989569130795189e-05,
+      "loss": 1.2308,
+      "step": 9565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9895582417204476e-05,
+      "loss": 1.3563,
+      "step": 9570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9895473469947966e-05,
+      "loss": 1.2526,
+      "step": 9575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9895364466182978e-05,
+      "loss": 1.3359,
+      "step": 9580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9895255405910138e-05,
+      "loss": 1.0288,
+      "step": 9585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989514628913007e-05,
+      "loss": 1.1636,
+      "step": 9590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989503711584339e-05,
+      "loss": 1.1244,
+      "step": 9595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989492788605073e-05,
+      "loss": 1.1791,
+      "step": 9600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9894818599752704e-05,
+      "loss": 1.3037,
+      "step": 9605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9894709256949947e-05,
+      "loss": 1.0738,
+      "step": 9610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9894599857643077e-05,
+      "loss": 1.2277,
+      "step": 9615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989449040183272e-05,
+      "loss": 1.1937,
+      "step": 9620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9894380889519502e-05,
+      "loss": 1.1917,
+      "step": 9625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9894271320704046e-05,
+      "loss": 1.3834,
+      "step": 9630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989416169538698e-05,
+      "loss": 1.1577,
+      "step": 9635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9894052013568928e-05,
+      "loss": 0.9428,
+      "step": 9640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9893942275250516e-05,
+      "loss": 1.1291,
+      "step": 9645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9893832480432375e-05,
+      "loss": 1.1835,
+      "step": 9650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989372262911513e-05,
+      "loss": 1.3922,
+      "step": 9655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9893612721299402e-05,
+      "loss": 1.1248,
+      "step": 9660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989350275698583e-05,
+      "loss": 1.213,
+      "step": 9665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9893392736175033e-05,
+      "loss": 1.3818,
+      "step": 9670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9893282658867643e-05,
+      "loss": 0.9963,
+      "step": 9675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9893172525064287e-05,
+      "loss": 1.2942,
+      "step": 9680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9893062334765595e-05,
+      "loss": 0.9452,
+      "step": 9685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98929520879722e-05,
+      "loss": 1.1897,
+      "step": 9690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989284178468472e-05,
+      "loss": 1.0259,
+      "step": 9695
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98927314249038e-05,
+      "loss": 1.3206,
+      "step": 9700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989262100863006e-05,
+      "loss": 1.6516,
+      "step": 9705
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9892510535864132e-05,
+      "loss": 1.4037,
+      "step": 9710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989240000660665e-05,
+      "loss": 1.0368,
+      "step": 9715
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9892289420858243e-05,
+      "loss": 1.5531,
+      "step": 9720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989217877861954e-05,
+      "loss": 1.3799,
+      "step": 9725
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9892068079891177e-05,
+      "loss": 1.3424,
+      "step": 9730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9891957324673788e-05,
+      "loss": 1.4246,
+      "step": 9735
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9891846512968002e-05,
+      "loss": 1.2243,
+      "step": 9740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9891735644774446e-05,
+      "loss": 1.0598,
+      "step": 9745
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9891624720093764e-05,
+      "loss": 1.0711,
+      "step": 9750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9891513738926586e-05,
+      "loss": 1.2114,
+      "step": 9755
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989140270127354e-05,
+      "loss": 1.5318,
+      "step": 9760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989129160713527e-05,
+      "loss": 1.1534,
+      "step": 9765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9891180456512397e-05,
+      "loss": 1.1667,
+      "step": 9770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9891069249405567e-05,
+      "loss": 1.271,
+      "step": 9775
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989095798581541e-05,
+      "loss": 1.5872,
+      "step": 9780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989084666574257e-05,
+      "loss": 1.2265,
+      "step": 9785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989073528918767e-05,
+      "loss": 1.0201,
+      "step": 9790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9890623856151357e-05,
+      "loss": 1.0457,
+      "step": 9795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9890512366634255e-05,
+      "loss": 1.0311,
+      "step": 9800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9890400820637014e-05,
+      "loss": 1.2503,
+      "step": 9805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9890289218160262e-05,
+      "loss": 1.4068,
+      "step": 9810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9890177559204638e-05,
+      "loss": 1.3821,
+      "step": 9815
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989006584377078e-05,
+      "loss": 1.4525,
+      "step": 9820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9889954071859326e-05,
+      "loss": 0.9699,
+      "step": 9825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988984224347092e-05,
+      "loss": 1.3898,
+      "step": 9830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9889730358606193e-05,
+      "loss": 1.2761,
+      "step": 9835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9889618417265787e-05,
+      "loss": 1.1942,
+      "step": 9840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9889506419450336e-05,
+      "loss": 1.1532,
+      "step": 9845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988939436516049e-05,
+      "loss": 1.2619,
+      "step": 9850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988928225439688e-05,
+      "loss": 1.5465,
+      "step": 9855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988917008716015e-05,
+      "loss": 1.263,
+      "step": 9860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988905786345094e-05,
+      "loss": 1.53,
+      "step": 9865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9888945583269894e-05,
+      "loss": 1.0575,
+      "step": 9870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9888833246617646e-05,
+      "loss": 1.0476,
+      "step": 9875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9888720853494843e-05,
+      "loss": 1.2806,
+      "step": 9880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9888608403902125e-05,
+      "loss": 1.0644,
+      "step": 9885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9888495897840132e-05,
+      "loss": 1.4084,
+      "step": 9890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9888383335309513e-05,
+      "loss": 1.545,
+      "step": 9895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9888270716310903e-05,
+      "loss": 1.1981,
+      "step": 9900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988815804084495e-05,
+      "loss": 1.1743,
+      "step": 9905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98880453089123e-05,
+      "loss": 1.059,
+      "step": 9910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9887932520513585e-05,
+      "loss": 1.2824,
+      "step": 9915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9887819675649466e-05,
+      "loss": 1.4248,
+      "step": 9920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9887706774320572e-05,
+      "loss": 1.2857,
+      "step": 9925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9887593816527557e-05,
+      "loss": 1.2505,
+      "step": 9930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9887480802271062e-05,
+      "loss": 1.0554,
+      "step": 9935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9887367731551733e-05,
+      "loss": 1.1372,
+      "step": 9940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9887254604370218e-05,
+      "loss": 1.6018,
+      "step": 9945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988714142072716e-05,
+      "loss": 1.23,
+      "step": 9950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9887028180623208e-05,
+      "loss": 1.1246,
+      "step": 9955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9886914884059006e-05,
+      "loss": 1.2114,
+      "step": 9960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9886801531035204e-05,
+      "loss": 1.6327,
+      "step": 9965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9886688121552448e-05,
+      "loss": 1.1585,
+      "step": 9970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988657465561138e-05,
+      "loss": 1.3291,
+      "step": 9975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988646113321266e-05,
+      "loss": 1.0843,
+      "step": 9980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9886347554356926e-05,
+      "loss": 1.2927,
+      "step": 9985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9886233919044832e-05,
+      "loss": 0.9409,
+      "step": 9990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9886120227277023e-05,
+      "loss": 1.5901,
+      "step": 9995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9886006479054153e-05,
+      "loss": 1.1971,
+      "step": 10000
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 1.2409100532531738,
+      "eval_runtime": 6477.7475,
+      "eval_samples_per_second": 3.568,
+      "eval_steps_per_second": 1.784,
+      "step": 10000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9885892674376865e-05,
+      "loss": 1.2963,
+      "step": 10005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9885778813245814e-05,
+      "loss": 1.1636,
+      "step": 10010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988566489566165e-05,
+      "loss": 0.9621,
+      "step": 10015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988555092162502e-05,
+      "loss": 1.1449,
+      "step": 10020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988543689113658e-05,
+      "loss": 1.0954,
+      "step": 10025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9885322804196978e-05,
+      "loss": 1.1218,
+      "step": 10030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9885208660806864e-05,
+      "loss": 1.2043,
+      "step": 10035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9885094460966894e-05,
+      "loss": 1.0266,
+      "step": 10040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9884980204677718e-05,
+      "loss": 1.1203,
+      "step": 10045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9884865891939986e-05,
+      "loss": 1.2609,
+      "step": 10050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9884751522754355e-05,
+      "loss": 1.3692,
+      "step": 10055
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9884637097121476e-05,
+      "loss": 1.2597,
+      "step": 10060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9884522615042003e-05,
+      "loss": 1.1406,
+      "step": 10065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9884408076516587e-05,
+      "loss": 1.4196,
+      "step": 10070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9884293481545887e-05,
+      "loss": 1.6112,
+      "step": 10075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9884178830130556e-05,
+      "loss": 1.1824,
+      "step": 10080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9884064122271245e-05,
+      "loss": 1.3091,
+      "step": 10085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9883949357968614e-05,
+      "loss": 1.276,
+      "step": 10090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9883834537223315e-05,
+      "loss": 1.6573,
+      "step": 10095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9883719660036004e-05,
+      "loss": 1.1668,
+      "step": 10100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988360472640734e-05,
+      "loss": 1.2526,
+      "step": 10105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9883489736337976e-05,
+      "loss": 1.4807,
+      "step": 10110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9883374689828568e-05,
+      "loss": 1.0886,
+      "step": 10115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9883259586879775e-05,
+      "loss": 1.1517,
+      "step": 10120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9883144427492256e-05,
+      "loss": 1.1211,
+      "step": 10125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9883029211666664e-05,
+      "loss": 1.4874,
+      "step": 10130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988291393940366e-05,
+      "loss": 1.5091,
+      "step": 10135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9882798610703908e-05,
+      "loss": 1.2541,
+      "step": 10140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988268322556805e-05,
+      "loss": 1.1447,
+      "step": 10145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9882567783996764e-05,
+      "loss": 1.314,
+      "step": 10150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9882452285990697e-05,
+      "loss": 0.9884,
+      "step": 10155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988233673155051e-05,
+      "loss": 1.4214,
+      "step": 10160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988222112067687e-05,
+      "loss": 1.3074,
+      "step": 10165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9882105453370432e-05,
+      "loss": 1.161,
+      "step": 10170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9881989729631853e-05,
+      "loss": 1.0744,
+      "step": 10175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98818739494618e-05,
+      "loss": 1.2697,
+      "step": 10180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988175811286093e-05,
+      "loss": 1.5353,
+      "step": 10185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9881642219829906e-05,
+      "loss": 1.1183,
+      "step": 10190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9881526270369394e-05,
+      "loss": 1.0318,
+      "step": 10195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9881410264480047e-05,
+      "loss": 1.3856,
+      "step": 10200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988129420216254e-05,
+      "loss": 1.15,
+      "step": 10205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9881178083417517e-05,
+      "loss": 0.8726,
+      "step": 10210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9881061908245662e-05,
+      "loss": 1.2054,
+      "step": 10215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9880945676647626e-05,
+      "loss": 0.9944,
+      "step": 10220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9880829388624073e-05,
+      "loss": 1.358,
+      "step": 10225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9880713044175675e-05,
+      "loss": 1.2794,
+      "step": 10230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988059664330309e-05,
+      "loss": 0.973,
+      "step": 10235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988048018600698e-05,
+      "loss": 1.2693,
+      "step": 10240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988036367228802e-05,
+      "loss": 1.1531,
+      "step": 10245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9880247102146863e-05,
+      "loss": 1.2489,
+      "step": 10250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9880130475584184e-05,
+      "loss": 1.3648,
+      "step": 10255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9880013792600647e-05,
+      "loss": 1.3597,
+      "step": 10260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9879897053196916e-05,
+      "loss": 1.7223,
+      "step": 10265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987978025737366e-05,
+      "loss": 1.3152,
+      "step": 10270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987966340513154e-05,
+      "loss": 1.3869,
+      "step": 10275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9879546496471232e-05,
+      "loss": 1.1015,
+      "step": 10280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9879429531393398e-05,
+      "loss": 1.5379,
+      "step": 10285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987931250989871e-05,
+      "loss": 1.1558,
+      "step": 10290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9879195431987834e-05,
+      "loss": 1.3919,
+      "step": 10295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987907829766144e-05,
+      "loss": 1.0623,
+      "step": 10300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987896110692019e-05,
+      "loss": 1.2201,
+      "step": 10305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987884385976476e-05,
+      "loss": 1.8566,
+      "step": 10310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987872655619582e-05,
+      "loss": 1.3311,
+      "step": 10315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987860919621404e-05,
+      "loss": 1.0822,
+      "step": 10320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9878491779820085e-05,
+      "loss": 1.1824,
+      "step": 10325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9878374307014635e-05,
+      "loss": 1.2878,
+      "step": 10330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9878256777798346e-05,
+      "loss": 1.4227,
+      "step": 10335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9878139192171903e-05,
+      "loss": 1.272,
+      "step": 10340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9878021550135972e-05,
+      "loss": 0.9541,
+      "step": 10345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987790385169122e-05,
+      "loss": 1.3114,
+      "step": 10350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987778609683833e-05,
+      "loss": 1.5113,
+      "step": 10355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9877668285577965e-05,
+      "loss": 1.0491,
+      "step": 10360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9877550417910804e-05,
+      "loss": 1.4027,
+      "step": 10365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9877432493837516e-05,
+      "loss": 1.2956,
+      "step": 10370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9877314513358777e-05,
+      "loss": 1.1308,
+      "step": 10375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9877196476475256e-05,
+      "loss": 1.0753,
+      "step": 10380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9877078383187635e-05,
+      "loss": 1.3609,
+      "step": 10385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9876960233496582e-05,
+      "loss": 1.9996,
+      "step": 10390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9876842027402774e-05,
+      "loss": 1.2402,
+      "step": 10395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9876723764906888e-05,
+      "loss": 1.0923,
+      "step": 10400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9876605446009595e-05,
+      "loss": 1.1492,
+      "step": 10405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987648707071157e-05,
+      "loss": 1.1209,
+      "step": 10410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9876368639013494e-05,
+      "loss": 1.1345,
+      "step": 10415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9876250150916045e-05,
+      "loss": 1.5224,
+      "step": 10420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9876131606419892e-05,
+      "loss": 1.127,
+      "step": 10425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9876013005525714e-05,
+      "loss": 1.2273,
+      "step": 10430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9875894348234192e-05,
+      "loss": 1.5453,
+      "step": 10435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9875775634546e-05,
+      "loss": 1.2936,
+      "step": 10440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987565686446182e-05,
+      "loss": 1.207,
+      "step": 10445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9875538037982326e-05,
+      "loss": 1.0604,
+      "step": 10450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98754191551082e-05,
+      "loss": 1.0935,
+      "step": 10455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9875300215840115e-05,
+      "loss": 1.1386,
+      "step": 10460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9875181220178758e-05,
+      "loss": 0.972,
+      "step": 10465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9875062168124803e-05,
+      "loss": 1.1606,
+      "step": 10470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9874943059678932e-05,
+      "loss": 1.1025,
+      "step": 10475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9874823894841826e-05,
+      "loss": 1.2453,
+      "step": 10480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9874704673614165e-05,
+      "loss": 0.9414,
+      "step": 10485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9874585395996627e-05,
+      "loss": 1.392,
+      "step": 10490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9874466061989897e-05,
+      "loss": 1.2415,
+      "step": 10495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9874346671594653e-05,
+      "loss": 1.3719,
+      "step": 10500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987422722481158e-05,
+      "loss": 1.4879,
+      "step": 10505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9874107721641357e-05,
+      "loss": 1.295,
+      "step": 10510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987398816208467e-05,
+      "loss": 1.2331,
+      "step": 10515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9873868546142198e-05,
+      "loss": 1.08,
+      "step": 10520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9873748873814625e-05,
+      "loss": 1.3973,
+      "step": 10525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987362914510264e-05,
+      "loss": 1.3698,
+      "step": 10530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987350936000692e-05,
+      "loss": 1.1052,
+      "step": 10535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987338951852815e-05,
+      "loss": 1.1828,
+      "step": 10540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9873269620667014e-05,
+      "loss": 1.1824,
+      "step": 10545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98731496664242e-05,
+      "loss": 1.4254,
+      "step": 10550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9873029655800393e-05,
+      "loss": 1.0557,
+      "step": 10555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9872909588796273e-05,
+      "loss": 1.6545,
+      "step": 10560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9872789465412533e-05,
+      "loss": 1.4705,
+      "step": 10565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987266928564985e-05,
+      "loss": 1.4124,
+      "step": 10570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987254904950892e-05,
+      "loss": 1.0706,
+      "step": 10575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9872428756990424e-05,
+      "loss": 1.1204,
+      "step": 10580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9872308408095048e-05,
+      "loss": 1.4118,
+      "step": 10585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987218800282348e-05,
+      "loss": 1.1413,
+      "step": 10590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9872067541176413e-05,
+      "loss": 1.2695,
+      "step": 10595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9871947023154527e-05,
+      "loss": 1.3243,
+      "step": 10600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9871826448758516e-05,
+      "loss": 1.2357,
+      "step": 10605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9871705817989064e-05,
+      "loss": 1.0471,
+      "step": 10610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9871585130846865e-05,
+      "loss": 1.6457,
+      "step": 10615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9871464387332608e-05,
+      "loss": 1.185,
+      "step": 10620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9871343587446973e-05,
+      "loss": 1.136,
+      "step": 10625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987122273119066e-05,
+      "loss": 1.412,
+      "step": 10630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9871101818564356e-05,
+      "loss": 1.1109,
+      "step": 10635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987098084956875e-05,
+      "loss": 1.2231,
+      "step": 10640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9870859824204536e-05,
+      "loss": 1.5621,
+      "step": 10645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9870738742472405e-05,
+      "loss": 0.981,
+      "step": 10650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9870617604373043e-05,
+      "loss": 1.2283,
+      "step": 10655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9870496409907146e-05,
+      "loss": 1.1398,
+      "step": 10660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9870375159075407e-05,
+      "loss": 1.2184,
+      "step": 10665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987025385187852e-05,
+      "loss": 1.1094,
+      "step": 10670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987013248831717e-05,
+      "loss": 1.3011,
+      "step": 10675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9870011068392057e-05,
+      "loss": 1.0852,
+      "step": 10680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986988959210387e-05,
+      "loss": 1.2552,
+      "step": 10685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986976805945331e-05,
+      "loss": 1.1412,
+      "step": 10690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9869646470441058e-05,
+      "loss": 0.9488,
+      "step": 10695
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9869524825067823e-05,
+      "loss": 0.9769,
+      "step": 10700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9869403123334294e-05,
+      "loss": 1.1181,
+      "step": 10705
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986928136524116e-05,
+      "loss": 1.0681,
+      "step": 10710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9869159550789124e-05,
+      "loss": 1.0882,
+      "step": 10715
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9869037679978877e-05,
+      "loss": 1.3535,
+      "step": 10720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9868915752811116e-05,
+      "loss": 1.1684,
+      "step": 10725
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9868793769286537e-05,
+      "loss": 1.2424,
+      "step": 10730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9868671729405843e-05,
+      "loss": 0.9705,
+      "step": 10735
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9868549633169724e-05,
+      "loss": 0.9345,
+      "step": 10740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9868427480578876e-05,
+      "loss": 1.1214,
+      "step": 10745
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9868305271634e-05,
+      "loss": 1.3764,
+      "step": 10750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9868183006335794e-05,
+      "loss": 1.1509,
+      "step": 10755
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9868060684684955e-05,
+      "loss": 1.2361,
+      "step": 10760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986793830668218e-05,
+      "loss": 1.1931,
+      "step": 10765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9867815872328174e-05,
+      "loss": 1.0359,
+      "step": 10770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986769338162363e-05,
+      "loss": 1.5928,
+      "step": 10775
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986757083456925e-05,
+      "loss": 1.1128,
+      "step": 10780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986744823116573e-05,
+      "loss": 1.1367,
+      "step": 10785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9867325571413776e-05,
+      "loss": 0.8944,
+      "step": 10790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9867202855314087e-05,
+      "loss": 1.1478,
+      "step": 10795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986708008286736e-05,
+      "loss": 1.6205,
+      "step": 10800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9866957254074302e-05,
+      "loss": 1.2214,
+      "step": 10805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986683436893561e-05,
+      "loss": 1.1702,
+      "step": 10810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9866711427451983e-05,
+      "loss": 1.3409,
+      "step": 10815
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986658842962413e-05,
+      "loss": 1.232,
+      "step": 10820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986646537545275e-05,
+      "loss": 1.0151,
+      "step": 10825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9866342264938548e-05,
+      "loss": 1.2355,
+      "step": 10830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9866219098082222e-05,
+      "loss": 1.4583,
+      "step": 10835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986609587488448e-05,
+      "loss": 1.8161,
+      "step": 10840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9865972595346027e-05,
+      "loss": 1.0783,
+      "step": 10845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9865849259467557e-05,
+      "loss": 1.4209,
+      "step": 10850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9865725867249788e-05,
+      "loss": 1.3795,
+      "step": 10855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9865602418693414e-05,
+      "loss": 0.9859,
+      "step": 10860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9865478913799145e-05,
+      "loss": 1.1487,
+      "step": 10865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9865355352567684e-05,
+      "loss": 0.9843,
+      "step": 10870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986523173499974e-05,
+      "loss": 1.2784,
+      "step": 10875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9865108061096015e-05,
+      "loss": 1.185,
+      "step": 10880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986498433085722e-05,
+      "loss": 1.4853,
+      "step": 10885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9864860544284054e-05,
+      "loss": 1.2828,
+      "step": 10890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986473670137723e-05,
+      "loss": 1.3309,
+      "step": 10895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9864612802137456e-05,
+      "loss": 1.4696,
+      "step": 10900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9864488846565436e-05,
+      "loss": 1.3497,
+      "step": 10905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986436483466188e-05,
+      "loss": 1.3165,
+      "step": 10910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986424076642749e-05,
+      "loss": 1.0141,
+      "step": 10915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9864116641862986e-05,
+      "loss": 1.1502,
+      "step": 10920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9863992460969072e-05,
+      "loss": 1.648,
+      "step": 10925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986386822374645e-05,
+      "loss": 0.9742,
+      "step": 10930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9863743930195837e-05,
+      "loss": 1.1449,
+      "step": 10935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986361958031794e-05,
+      "loss": 1.114,
+      "step": 10940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986349517411347e-05,
+      "loss": 1.5397,
+      "step": 10945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9863370711583138e-05,
+      "loss": 1.6364,
+      "step": 10950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9863246192727655e-05,
+      "loss": 1.2116,
+      "step": 10955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986312161754773e-05,
+      "loss": 1.2281,
+      "step": 10960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986299698604408e-05,
+      "loss": 1.222,
+      "step": 10965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986287229821741e-05,
+      "loss": 1.1629,
+      "step": 10970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9862747554068433e-05,
+      "loss": 1.3408,
+      "step": 10975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9862622753597863e-05,
+      "loss": 1.065,
+      "step": 10980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986249789680641e-05,
+      "loss": 0.9804,
+      "step": 10985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9862372983694793e-05,
+      "loss": 1.2469,
+      "step": 10990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9862248014263722e-05,
+      "loss": 1.3296,
+      "step": 10995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986212298851391e-05,
+      "loss": 1.1393,
+      "step": 11000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986199790644607e-05,
+      "loss": 1.1336,
+      "step": 11005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986187276806092e-05,
+      "loss": 1.4058,
+      "step": 11010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986174757335917e-05,
+      "loss": 1.2721,
+      "step": 11015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9861622322341536e-05,
+      "loss": 1.3598,
+      "step": 11020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9861497015008734e-05,
+      "loss": 1.217,
+      "step": 11025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9861371651361482e-05,
+      "loss": 1.3356,
+      "step": 11030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98612462314005e-05,
+      "loss": 1.2336,
+      "step": 11035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9861120755126488e-05,
+      "loss": 1.221,
+      "step": 11040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9860995222540178e-05,
+      "loss": 1.1293,
+      "step": 11045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9860869633642278e-05,
+      "loss": 1.082,
+      "step": 11050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9860743988433512e-05,
+      "loss": 1.2635,
+      "step": 11055
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9860618286914596e-05,
+      "loss": 1.3784,
+      "step": 11060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986049252908624e-05,
+      "loss": 1.2206,
+      "step": 11065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.986036671494917e-05,
+      "loss": 1.3358,
+      "step": 11070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9860240844504106e-05,
+      "loss": 1.4454,
+      "step": 11075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9860114917751763e-05,
+      "loss": 1.1348,
+      "step": 11080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9859988934692858e-05,
+      "loss": 1.2675,
+      "step": 11085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9859862895328112e-05,
+      "loss": 1.2524,
+      "step": 11090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9859736799658253e-05,
+      "loss": 1.2737,
+      "step": 11095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9859610647683983e-05,
+      "loss": 1.0733,
+      "step": 11100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985948443940604e-05,
+      "loss": 1.1103,
+      "step": 11105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9859358174825137e-05,
+      "loss": 1.0363,
+      "step": 11110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9859231853941994e-05,
+      "loss": 1.5036,
+      "step": 11115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9859105476757337e-05,
+      "loss": 1.1192,
+      "step": 11120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9858979043271884e-05,
+      "loss": 1.1963,
+      "step": 11125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9858852553486357e-05,
+      "loss": 1.2365,
+      "step": 11130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985872600740148e-05,
+      "loss": 1.3814,
+      "step": 11135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9858599405017976e-05,
+      "loss": 1.0812,
+      "step": 11140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9858472746336566e-05,
+      "loss": 1.2029,
+      "step": 11145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9858346031357975e-05,
+      "loss": 1.1158,
+      "step": 11150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9858219260082924e-05,
+      "loss": 1.3801,
+      "step": 11155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985809243251214e-05,
+      "loss": 1.3947,
+      "step": 11160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9857965548646347e-05,
+      "loss": 1.0431,
+      "step": 11165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985783860848627e-05,
+      "loss": 1.7886,
+      "step": 11170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9857711612032626e-05,
+      "loss": 1.1055,
+      "step": 11175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9857584559286152e-05,
+      "loss": 1.2393,
+      "step": 11180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985745745024757e-05,
+      "loss": 1.03,
+      "step": 11185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98573302849176e-05,
+      "loss": 1.2805,
+      "step": 11190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985720306329697e-05,
+      "loss": 1.1462,
+      "step": 11195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9857075785386413e-05,
+      "loss": 1.3456,
+      "step": 11200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9856948451186652e-05,
+      "loss": 1.02,
+      "step": 11205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985682106069841e-05,
+      "loss": 1.3376,
+      "step": 11210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9856693613922423e-05,
+      "loss": 1.2378,
+      "step": 11215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985656611085941e-05,
+      "loss": 1.416,
+      "step": 11220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9856438551510107e-05,
+      "loss": 1.3436,
+      "step": 11225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9856310935875237e-05,
+      "loss": 1.0874,
+      "step": 11230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985618326395553e-05,
+      "loss": 1.1638,
+      "step": 11235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9856055535751712e-05,
+      "loss": 1.3187,
+      "step": 11240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985592775126452e-05,
+      "loss": 1.0937,
+      "step": 11245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985579991049468e-05,
+      "loss": 1.4024,
+      "step": 11250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9855672013442918e-05,
+      "loss": 1.1657,
+      "step": 11255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985554406010997e-05,
+      "loss": 1.4126,
+      "step": 11260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9855416050496564e-05,
+      "loss": 1.3912,
+      "step": 11265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9855287984603435e-05,
+      "loss": 1.2154,
+      "step": 11270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9855159862431305e-05,
+      "loss": 1.2804,
+      "step": 11275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9855031683980918e-05,
+      "loss": 1.5467,
+      "step": 11280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9854903449252994e-05,
+      "loss": 1.024,
+      "step": 11285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9854775158248275e-05,
+      "loss": 1.0998,
+      "step": 11290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9854646810967484e-05,
+      "loss": 1.3391,
+      "step": 11295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9854518407411362e-05,
+      "loss": 1.0636,
+      "step": 11300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985438994758064e-05,
+      "loss": 1.226,
+      "step": 11305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9854261431476053e-05,
+      "loss": 1.3227,
+      "step": 11310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9854132859098333e-05,
+      "loss": 1.274,
+      "step": 11315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9854004230448213e-05,
+      "loss": 1.2498,
+      "step": 11320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9853875545526428e-05,
+      "loss": 0.83,
+      "step": 11325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9853746804333713e-05,
+      "loss": 1.1327,
+      "step": 11330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9853618006870804e-05,
+      "loss": 1.1999,
+      "step": 11335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985348915313844e-05,
+      "loss": 1.0357,
+      "step": 11340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985336024313735e-05,
+      "loss": 1.2454,
+      "step": 11345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9853231276868277e-05,
+      "loss": 1.1345,
+      "step": 11350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985310225433195e-05,
+      "loss": 1.1031,
+      "step": 11355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985297317552911e-05,
+      "loss": 1.1486,
+      "step": 11360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985284404046049e-05,
+      "loss": 1.3602,
+      "step": 11365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985271484912684e-05,
+      "loss": 1.1613,
+      "step": 11370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9852585601528883e-05,
+      "loss": 1.1652,
+      "step": 11375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9852456297667367e-05,
+      "loss": 1.4337,
+      "step": 11380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9852326937543023e-05,
+      "loss": 1.1928,
+      "step": 11385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9852197521156595e-05,
+      "loss": 1.3482,
+      "step": 11390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985206804850882e-05,
+      "loss": 1.1555,
+      "step": 11395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9851938519600435e-05,
+      "loss": 1.5325,
+      "step": 11400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9851808934432186e-05,
+      "loss": 1.3917,
+      "step": 11405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9851679293004805e-05,
+      "loss": 1.2634,
+      "step": 11410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9851549595319037e-05,
+      "loss": 1.021,
+      "step": 11415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9851419841375624e-05,
+      "loss": 1.577,
+      "step": 11420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9851290031175305e-05,
+      "loss": 0.9013,
+      "step": 11425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.985116016471882e-05,
+      "loss": 1.0437,
+      "step": 11430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9851030242006914e-05,
+      "loss": 1.2488,
+      "step": 11435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9850900263040325e-05,
+      "loss": 1.371,
+      "step": 11440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98507702278198e-05,
+      "loss": 0.8482,
+      "step": 11445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9850640136346076e-05,
+      "loss": 1.3319,
+      "step": 11450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98505099886199e-05,
+      "loss": 1.5613,
+      "step": 11455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985037978464201e-05,
+      "loss": 1.2808,
+      "step": 11460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.985024952441316e-05,
+      "loss": 1.2154,
+      "step": 11465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9850119207934083e-05,
+      "loss": 1.4716,
+      "step": 11470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984998883520553e-05,
+      "loss": 1.1423,
+      "step": 11475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984985840622824e-05,
+      "loss": 1.1175,
+      "step": 11480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984972792100296e-05,
+      "loss": 1.1448,
+      "step": 11485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9849597379530438e-05,
+      "loss": 1.4043,
+      "step": 11490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984946678181142e-05,
+      "loss": 1.0753,
+      "step": 11495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9849336127846642e-05,
+      "loss": 1.2699,
+      "step": 11500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9849205417636862e-05,
+      "loss": 1.2057,
+      "step": 11505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9849074651182824e-05,
+      "loss": 1.0275,
+      "step": 11510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984894382848527e-05,
+      "loss": 1.2348,
+      "step": 11515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984881294954495e-05,
+      "loss": 1.2694,
+      "step": 11520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984868201436261e-05,
+      "loss": 1.1881,
+      "step": 11525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9848551022938998e-05,
+      "loss": 1.1068,
+      "step": 11530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984841997527486e-05,
+      "loss": 1.0816,
+      "step": 11535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984828887137095e-05,
+      "loss": 1.1879,
+      "step": 11540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9848157711228016e-05,
+      "loss": 1.3706,
+      "step": 11545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98480264948468e-05,
+      "loss": 1.1095,
+      "step": 11550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984789522222806e-05,
+      "loss": 1.0058,
+      "step": 11555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984776389337254e-05,
+      "loss": 1.1267,
+      "step": 11560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984763250828099e-05,
+      "loss": 1.0871,
+      "step": 11565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9847501066954162e-05,
+      "loss": 1.3626,
+      "step": 11570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9847369569392807e-05,
+      "loss": 0.9958,
+      "step": 11575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9847238015597677e-05,
+      "loss": 1.1831,
+      "step": 11580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984710640556952e-05,
+      "loss": 1.2024,
+      "step": 11585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984697473930909e-05,
+      "loss": 1.1863,
+      "step": 11590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9846843016817137e-05,
+      "loss": 1.5405,
+      "step": 11595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9846711238094417e-05,
+      "loss": 1.4188,
+      "step": 11600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9846579403141675e-05,
+      "loss": 1.2943,
+      "step": 11605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9846447511959672e-05,
+      "loss": 1.4569,
+      "step": 11610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9846315564549158e-05,
+      "loss": 1.2068,
+      "step": 11615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9846183560910886e-05,
+      "loss": 1.4474,
+      "step": 11620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9846051501045608e-05,
+      "loss": 1.3533,
+      "step": 11625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9845919384954084e-05,
+      "loss": 1.0758,
+      "step": 11630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9845787212637057e-05,
+      "loss": 1.1105,
+      "step": 11635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98456549840953e-05,
+      "loss": 1.3132,
+      "step": 11640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9845522699329546e-05,
+      "loss": 1.3511,
+      "step": 11645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984539035834057e-05,
+      "loss": 1.1389,
+      "step": 11650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984525796112912e-05,
+      "loss": 1.0299,
+      "step": 11655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9845125507695945e-05,
+      "loss": 1.2856,
+      "step": 11660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9844992998041814e-05,
+      "loss": 1.4832,
+      "step": 11665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9844860432167476e-05,
+      "loss": 1.1634,
+      "step": 11670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984472781007369e-05,
+      "loss": 1.0581,
+      "step": 11675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9844595131761216e-05,
+      "loss": 1.4322,
+      "step": 11680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9844462397230804e-05,
+      "loss": 1.0609,
+      "step": 11685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984432960648322e-05,
+      "loss": 1.184,
+      "step": 11690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984419675951922e-05,
+      "loss": 1.0865,
+      "step": 11695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984406385633956e-05,
+      "loss": 1.1915,
+      "step": 11700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9843930896944998e-05,
+      "loss": 1.0505,
+      "step": 11705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98437978813363e-05,
+      "loss": 1.2819,
+      "step": 11710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9843664809514222e-05,
+      "loss": 1.1441,
+      "step": 11715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9843531681479523e-05,
+      "loss": 1.1093,
+      "step": 11720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984339849723296e-05,
+      "loss": 1.3902,
+      "step": 11725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9843265256775298e-05,
+      "loss": 1.1287,
+      "step": 11730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9843131960107304e-05,
+      "loss": 0.9871,
+      "step": 11735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9842998607229725e-05,
+      "loss": 0.9815,
+      "step": 11740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984286519814333e-05,
+      "loss": 1.2145,
+      "step": 11745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9842731732848885e-05,
+      "loss": 1.2663,
+      "step": 11750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9842598211347145e-05,
+      "loss": 1.3638,
+      "step": 11755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9842464633638876e-05,
+      "loss": 1.2382,
+      "step": 11760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9842330999724843e-05,
+      "loss": 1.0557,
+      "step": 11765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9842197309605802e-05,
+      "loss": 1.1982,
+      "step": 11770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9842063563282524e-05,
+      "loss": 1.0403,
+      "step": 11775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9841929760755767e-05,
+      "loss": 1.1481,
+      "step": 11780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9841795902026297e-05,
+      "loss": 0.8844,
+      "step": 11785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9841661987094883e-05,
+      "loss": 1.2422,
+      "step": 11790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984152801596228e-05,
+      "loss": 1.8789,
+      "step": 11795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9841393988629263e-05,
+      "loss": 1.1147,
+      "step": 11800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9841259905096592e-05,
+      "loss": 1.3944,
+      "step": 11805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9841125765365033e-05,
+      "loss": 1.2658,
+      "step": 11810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9840991569435352e-05,
+      "loss": 1.0941,
+      "step": 11815
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984085731730832e-05,
+      "loss": 1.0669,
+      "step": 11820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9840723008984696e-05,
+      "loss": 1.2891,
+      "step": 11825
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9840588644465253e-05,
+      "loss": 0.9689,
+      "step": 11830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9840454223750755e-05,
+      "loss": 1.094,
+      "step": 11835
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9840319746841973e-05,
+      "loss": 1.1399,
+      "step": 11840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.984018521373967e-05,
+      "loss": 1.0234,
+      "step": 11845
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9840050624444615e-05,
+      "loss": 1.1422,
+      "step": 11850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9839915978957584e-05,
+      "loss": 1.0784,
+      "step": 11855
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9839781277279336e-05,
+      "loss": 1.3009,
+      "step": 11860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9839646519410644e-05,
+      "loss": 1.0545,
+      "step": 11865
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9839511705352283e-05,
+      "loss": 1.1935,
+      "step": 11870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9839376835105012e-05,
+      "loss": 1.0989,
+      "step": 11875
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983924190866961e-05,
+      "loss": 1.4702,
+      "step": 11880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9839106926046845e-05,
+      "loss": 0.9871,
+      "step": 11885
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9838971887237486e-05,
+      "loss": 1.0145,
+      "step": 11890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983883679224231e-05,
+      "loss": 1.5668,
+      "step": 11895
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983870164106208e-05,
+      "loss": 1.3032,
+      "step": 11900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983856643369757e-05,
+      "loss": 0.9709,
+      "step": 11905
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983843117014956e-05,
+      "loss": 1.5784,
+      "step": 11910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9838295850418808e-05,
+      "loss": 1.3816,
+      "step": 11915
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9838160474506102e-05,
+      "loss": 1.8916,
+      "step": 11920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9838025042412207e-05,
+      "loss": 1.2558,
+      "step": 11925
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9837889554137897e-05,
+      "loss": 1.3334,
+      "step": 11930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9837754009683947e-05,
+      "loss": 1.0767,
+      "step": 11935
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983761840905113e-05,
+      "loss": 1.3358,
+      "step": 11940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9837482752240218e-05,
+      "loss": 1.207,
+      "step": 11945
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9837347039251994e-05,
+      "loss": 1.0272,
+      "step": 11950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9837211270087223e-05,
+      "loss": 1.1798,
+      "step": 11955
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9837075444746685e-05,
+      "loss": 1.4565,
+      "step": 11960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9836939563231156e-05,
+      "loss": 1.2572,
+      "step": 11965
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9836803625541412e-05,
+      "loss": 0.9008,
+      "step": 11970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9836667631678228e-05,
+      "loss": 1.2131,
+      "step": 11975
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983653158164238e-05,
+      "loss": 1.2139,
+      "step": 11980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9836395475434645e-05,
+      "loss": 1.6618,
+      "step": 11985
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9836259313055804e-05,
+      "loss": 1.3725,
+      "step": 11990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983612309450663e-05,
+      "loss": 1.2832,
+      "step": 11995
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983598681978791e-05,
+      "loss": 1.1091,
+      "step": 12000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9835850488900405e-05,
+      "loss": 1.2554,
+      "step": 12005
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9835714101844906e-05,
+      "loss": 1.416,
+      "step": 12010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9835577658622195e-05,
+      "loss": 1.4751,
+      "step": 12015
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9835441159233038e-05,
+      "loss": 1.4148,
+      "step": 12020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9835304603678227e-05,
+      "loss": 1.653,
+      "step": 12025
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9835167991958535e-05,
+      "loss": 1.0489,
+      "step": 12030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9835031324074742e-05,
+      "loss": 1.3145,
+      "step": 12035
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9834894600027632e-05,
+      "loss": 1.2313,
+      "step": 12040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9834757819817986e-05,
+      "loss": 1.2314,
+      "step": 12045
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9834620983446583e-05,
+      "loss": 1.1954,
+      "step": 12050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9834484090914202e-05,
+      "loss": 1.201,
+      "step": 12055
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983434714222163e-05,
+      "loss": 1.1694,
+      "step": 12060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983421013736964e-05,
+      "loss": 1.0563,
+      "step": 12065
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9834073076359027e-05,
+      "loss": 1.2925,
+      "step": 12070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9833935959190568e-05,
+      "loss": 1.2233,
+      "step": 12075
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983379878586504e-05,
+      "loss": 1.2939,
+      "step": 12080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983366155638324e-05,
+      "loss": 1.3202,
+      "step": 12085
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9833524270745938e-05,
+      "loss": 1.492,
+      "step": 12090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9833386928953922e-05,
+      "loss": 1.382,
+      "step": 12095
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983324953100798e-05,
+      "loss": 1.2655,
+      "step": 12100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9833112076908895e-05,
+      "loss": 1.0472,
+      "step": 12105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983297456665745e-05,
+      "loss": 1.4484,
+      "step": 12110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9832837000254434e-05,
+      "loss": 1.011,
+      "step": 12115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9832699377700625e-05,
+      "loss": 1.1165,
+      "step": 12120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983256169899682e-05,
+      "loss": 1.0847,
+      "step": 12125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9832423964143793e-05,
+      "loss": 1.4724,
+      "step": 12130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9832286173142342e-05,
+      "loss": 1.5162,
+      "step": 12135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9832148325993248e-05,
+      "loss": 1.2651,
+      "step": 12140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9832010422697298e-05,
+      "loss": 1.1916,
+      "step": 12145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983187246325528e-05,
+      "loss": 1.2478,
+      "step": 12150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983173444766798e-05,
+      "loss": 1.0205,
+      "step": 12155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983159637593619e-05,
+      "loss": 1.4385,
+      "step": 12160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9831458248060697e-05,
+      "loss": 1.1417,
+      "step": 12165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9831320064042293e-05,
+      "loss": 1.2051,
+      "step": 12170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983118182388176e-05,
+      "loss": 1.3845,
+      "step": 12175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983104352757989e-05,
+      "loss": 1.1213,
+      "step": 12180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9830905175137476e-05,
+      "loss": 1.335,
+      "step": 12185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9830766766555304e-05,
+      "loss": 1.1424,
+      "step": 12190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9830628301834166e-05,
+      "loss": 1.4117,
+      "step": 12195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9830489780974856e-05,
+      "loss": 1.1377,
+      "step": 12200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983035120397816e-05,
+      "loss": 0.9247,
+      "step": 12205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.983021257084487e-05,
+      "loss": 1.0081,
+      "step": 12210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9830073881575786e-05,
+      "loss": 1.4535,
+      "step": 12215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9829935136171687e-05,
+      "loss": 1.2864,
+      "step": 12220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982979633463337e-05,
+      "loss": 1.2905,
+      "step": 12225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982965747696163e-05,
+      "loss": 1.3591,
+      "step": 12230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982951856315726e-05,
+      "loss": 1.2482,
+      "step": 12235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9829379593221052e-05,
+      "loss": 1.698,
+      "step": 12240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98292405671538e-05,
+      "loss": 0.8737,
+      "step": 12245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9829101484956295e-05,
+      "loss": 1.4299,
+      "step": 12250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982896234662934e-05,
+      "loss": 1.6366,
+      "step": 12255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982882315217372e-05,
+      "loss": 1.0695,
+      "step": 12260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9828683901590233e-05,
+      "loss": 0.9892,
+      "step": 12265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9828544594879674e-05,
+      "loss": 1.547,
+      "step": 12270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982840523204284e-05,
+      "loss": 1.0457,
+      "step": 12275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9828265813080528e-05,
+      "loss": 1.2599,
+      "step": 12280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9828126337993527e-05,
+      "loss": 0.9685,
+      "step": 12285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9827986806782642e-05,
+      "loss": 1.4196,
+      "step": 12290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9827847219448664e-05,
+      "loss": 1.1446,
+      "step": 12295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9827707575992395e-05,
+      "loss": 1.1565,
+      "step": 12300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982756787641463e-05,
+      "loss": 1.2019,
+      "step": 12305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9827428120716167e-05,
+      "loss": 1.3872,
+      "step": 12310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98272883088978e-05,
+      "loss": 1.4094,
+      "step": 12315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9827148440960334e-05,
+      "loss": 1.1682,
+      "step": 12320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9827008516904563e-05,
+      "loss": 1.3532,
+      "step": 12325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982686853673129e-05,
+      "loss": 1.0762,
+      "step": 12330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9826728500441313e-05,
+      "loss": 1.4171,
+      "step": 12335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9826588408035427e-05,
+      "loss": 1.2856,
+      "step": 12340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982644825951444e-05,
+      "loss": 1.5872,
+      "step": 12345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9826308054879145e-05,
+      "loss": 1.5496,
+      "step": 12350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982616779413035e-05,
+      "loss": 1.0796,
+      "step": 12355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982602747726885e-05,
+      "loss": 1.6539,
+      "step": 12360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9825887104295445e-05,
+      "loss": 1.0982,
+      "step": 12365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9825746675210943e-05,
+      "loss": 1.1168,
+      "step": 12370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9825606190016143e-05,
+      "loss": 1.5149,
+      "step": 12375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9825465648711844e-05,
+      "loss": 1.0984,
+      "step": 12380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9825325051298853e-05,
+      "loss": 1.2117,
+      "step": 12385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9825184397777972e-05,
+      "loss": 1.4432,
+      "step": 12390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9825043688150005e-05,
+      "loss": 1.2849,
+      "step": 12395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982490292241575e-05,
+      "loss": 1.1643,
+      "step": 12400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9824762100576018e-05,
+      "loss": 1.94,
+      "step": 12405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982462122263161e-05,
+      "loss": 1.4157,
+      "step": 12410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982448028858333e-05,
+      "loss": 1.1675,
+      "step": 12415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9824339298431984e-05,
+      "loss": 1.3197,
+      "step": 12420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9824198252178378e-05,
+      "loss": 1.5649,
+      "step": 12425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982405714982331e-05,
+      "loss": 1.4017,
+      "step": 12430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9823915991367598e-05,
+      "loss": 1.3608,
+      "step": 12435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9823774776812043e-05,
+      "loss": 1.4557,
+      "step": 12440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9823633506157444e-05,
+      "loss": 1.4622,
+      "step": 12445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982349217940462e-05,
+      "loss": 1.0071,
+      "step": 12450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9823350796554365e-05,
+      "loss": 1.2094,
+      "step": 12455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98232093576075e-05,
+      "loss": 1.3572,
+      "step": 12460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9823067862564824e-05,
+      "loss": 1.5611,
+      "step": 12465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9822926311427148e-05,
+      "loss": 1.1178,
+      "step": 12470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9822784704195276e-05,
+      "loss": 1.032,
+      "step": 12475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9822643040870024e-05,
+      "loss": 1.2735,
+      "step": 12480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9822501321452193e-05,
+      "loss": 1.1037,
+      "step": 12485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98223595459426e-05,
+      "loss": 1.316,
+      "step": 12490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9822217714342052e-05,
+      "loss": 1.0665,
+      "step": 12495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9822075826651355e-05,
+      "loss": 1.0764,
+      "step": 12500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9821933882871326e-05,
+      "loss": 1.0882,
+      "step": 12505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9821791883002766e-05,
+      "loss": 1.0108,
+      "step": 12510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9821649827046494e-05,
+      "loss": 1.3368,
+      "step": 12515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9821507715003322e-05,
+      "loss": 1.2657,
+      "step": 12520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9821365546874057e-05,
+      "loss": 1.5289,
+      "step": 12525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9821223322659507e-05,
+      "loss": 1.2158,
+      "step": 12530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9821081042360494e-05,
+      "loss": 1.1193,
+      "step": 12535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9820938705977827e-05,
+      "loss": 1.2099,
+      "step": 12540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9820796313512316e-05,
+      "loss": 1.1153,
+      "step": 12545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982065386496478e-05,
+      "loss": 1.1224,
+      "step": 12550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982051136033602e-05,
+      "loss": 1.1085,
+      "step": 12555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9820368799626865e-05,
+      "loss": 1.0213,
+      "step": 12560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982022618283812e-05,
+      "loss": 1.0192,
+      "step": 12565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9820083509970603e-05,
+      "loss": 1.335,
+      "step": 12570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9819940781025126e-05,
+      "loss": 1.1113,
+      "step": 12575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9819797996002507e-05,
+      "loss": 0.9923,
+      "step": 12580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9819655154903555e-05,
+      "loss": 1.5835,
+      "step": 12585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9819512257729095e-05,
+      "loss": 1.2485,
+      "step": 12590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.981936930447994e-05,
+      "loss": 1.1115,
+      "step": 12595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98192262951569e-05,
+      "loss": 0.885,
+      "step": 12600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98190832297608e-05,
+      "loss": 1.6136,
+      "step": 12605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.981894010829245e-05,
+      "loss": 0.9519,
+      "step": 12610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9818796930752677e-05,
+      "loss": 1.4413,
+      "step": 12615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9818653697142287e-05,
+      "loss": 1.3583,
+      "step": 12620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9818510407462107e-05,
+      "loss": 1.3023,
+      "step": 12625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.981836706171295e-05,
+      "loss": 1.2899,
+      "step": 12630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9818223659895633e-05,
+      "loss": 1.3739,
+      "step": 12635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9818080202010983e-05,
+      "loss": 1.4409,
+      "step": 12640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9817936688059815e-05,
+      "loss": 1.1489,
+      "step": 12645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9817793118042945e-05,
+      "loss": 1.2375,
+      "step": 12650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9817649491961197e-05,
+      "loss": 1.3428,
+      "step": 12655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.981750580981539e-05,
+      "loss": 1.1715,
+      "step": 12660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9817362071606343e-05,
+      "loss": 1.3356,
+      "step": 12665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.981721827733488e-05,
+      "loss": 1.2144,
+      "step": 12670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.981707442700182e-05,
+      "loss": 1.3718,
+      "step": 12675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9816930520607983e-05,
+      "loss": 1.1699,
+      "step": 12680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9816786558154195e-05,
+      "loss": 1.3054,
+      "step": 12685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9816642539641276e-05,
+      "loss": 1.39,
+      "step": 12690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9816498465070047e-05,
+      "loss": 1.2764,
+      "step": 12695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.981635433444133e-05,
+      "loss": 1.2402,
+      "step": 12700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9816210147755952e-05,
+      "loss": 1.1481,
+      "step": 12705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9816065905014733e-05,
+      "loss": 1.1827,
+      "step": 12710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98159216062185e-05,
+      "loss": 1.2206,
+      "step": 12715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9815777251368077e-05,
+      "loss": 1.3813,
+      "step": 12720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9815632840464283e-05,
+      "loss": 1.3659,
+      "step": 12725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9815488373507947e-05,
+      "loss": 1.1218,
+      "step": 12730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9815343850499895e-05,
+      "loss": 1.081,
+      "step": 12735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9815199271440944e-05,
+      "loss": 1.208,
+      "step": 12740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9815054636331933e-05,
+      "loss": 1.1264,
+      "step": 12745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9814909945173678e-05,
+      "loss": 1.262,
+      "step": 12750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9814765197967012e-05,
+      "loss": 1.3516,
+      "step": 12755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9814620394712752e-05,
+      "loss": 1.2605,
+      "step": 12760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9814475535411735e-05,
+      "loss": 1.1463,
+      "step": 12765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9814330620064783e-05,
+      "loss": 1.138,
+      "step": 12770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9814185648672725e-05,
+      "loss": 1.4094,
+      "step": 12775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9814040621236386e-05,
+      "loss": 0.9737,
+      "step": 12780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9813895537756596e-05,
+      "loss": 1.1834,
+      "step": 12785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.981375039823419e-05,
+      "loss": 1.2366,
+      "step": 12790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9813605202669986e-05,
+      "loss": 1.4874,
+      "step": 12795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9813459951064816e-05,
+      "loss": 1.1515,
+      "step": 12800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9813314643419515e-05,
+      "loss": 1.105,
+      "step": 12805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9813169279734908e-05,
+      "loss": 1.4955,
+      "step": 12810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9813023860011828e-05,
+      "loss": 1.1759,
+      "step": 12815
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9812878384251104e-05,
+      "loss": 1.4031,
+      "step": 12820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9812732852453563e-05,
+      "loss": 0.9884,
+      "step": 12825
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9812587264620044e-05,
+      "loss": 1.0804,
+      "step": 12830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9812441620751374e-05,
+      "loss": 1.089,
+      "step": 12835
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9812295920848382e-05,
+      "loss": 0.9567,
+      "step": 12840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98121501649119e-05,
+      "loss": 1.0628,
+      "step": 12845
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9812004352942766e-05,
+      "loss": 1.0254,
+      "step": 12850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9811858484941813e-05,
+      "loss": 1.0329,
+      "step": 12855
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9811712560909864e-05,
+      "loss": 1.1616,
+      "step": 12860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9811566580847767e-05,
+      "loss": 1.004,
+      "step": 12865
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.981142054475634e-05,
+      "loss": 1.2512,
+      "step": 12870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.981127445263643e-05,
+      "loss": 1.5121,
+      "step": 12875
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9811128304488863e-05,
+      "loss": 1.4529,
+      "step": 12880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.981098210031448e-05,
+      "loss": 1.2054,
+      "step": 12885
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.981083584011411e-05,
+      "loss": 1.239,
+      "step": 12890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.981068952388859e-05,
+      "loss": 1.2724,
+      "step": 12895
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9810543151638756e-05,
+      "loss": 1.2916,
+      "step": 12900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9810396723365444e-05,
+      "loss": 1.2566,
+      "step": 12905
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9810250239069492e-05,
+      "loss": 1.3209,
+      "step": 12910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9810103698751732e-05,
+      "loss": 1.2187,
+      "step": 12915
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9809957102413003e-05,
+      "loss": 1.1138,
+      "step": 12920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9809810450054147e-05,
+      "loss": 1.227,
+      "step": 12925
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9809663741675994e-05,
+      "loss": 1.2681,
+      "step": 12930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9809516977279385e-05,
+      "loss": 1.4083,
+      "step": 12935
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9809370156865157e-05,
+      "loss": 1.609,
+      "step": 12940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9809223280434153e-05,
+      "loss": 1.0964,
+      "step": 12945
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9809076347987207e-05,
+      "loss": 1.1075,
+      "step": 12950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9808929359525156e-05,
+      "loss": 1.1122,
+      "step": 12955
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9808782315048846e-05,
+      "loss": 1.4097,
+      "step": 12960
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980863521455911e-05,
+      "loss": 1.1811,
+      "step": 12965
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9808488058056796e-05,
+      "loss": 1.4019,
+      "step": 12970
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9808340845542736e-05,
+      "loss": 1.3456,
+      "step": 12975
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9808193577017775e-05,
+      "loss": 1.2942,
+      "step": 12980
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9808046252482754e-05,
+      "loss": 1.6423,
+      "step": 12985
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9807898871938513e-05,
+      "loss": 1.3189,
+      "step": 12990
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9807751435385895e-05,
+      "loss": 1.2166,
+      "step": 12995
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980760394282574e-05,
+      "loss": 0.9624,
+      "step": 13000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9807456394258894e-05,
+      "loss": 1.1326,
+      "step": 13005
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9807308789686195e-05,
+      "loss": 1.2649,
+      "step": 13010
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980716112910849e-05,
+      "loss": 1.1396,
+      "step": 13015
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980701341252662e-05,
+      "loss": 1.2654,
+      "step": 13020
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9806865639941424e-05,
+      "loss": 1.141,
+      "step": 13025
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9806717811353752e-05,
+      "loss": 1.1248,
+      "step": 13030
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980656992676445e-05,
+      "loss": 1.2906,
+      "step": 13035
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9806421986174356e-05,
+      "loss": 1.0084,
+      "step": 13040
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980627398958432e-05,
+      "loss": 1.1454,
+      "step": 13045
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9806125936995183e-05,
+      "loss": 1.1668,
+      "step": 13050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9805977828407797e-05,
+      "loss": 1.0509,
+      "step": 13055
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9805829663823003e-05,
+      "loss": 1.1743,
+      "step": 13060
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9805681443241642e-05,
+      "loss": 1.0374,
+      "step": 13065
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9805533166664573e-05,
+      "loss": 1.3899,
+      "step": 13070
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980538483409263e-05,
+      "loss": 1.1572,
+      "step": 13075
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980523644552667e-05,
+      "loss": 1.2711,
+      "step": 13080
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980508800096753e-05,
+      "loss": 1.2779,
+      "step": 13085
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9804939500416068e-05,
+      "loss": 1.3,
+      "step": 13090
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980479094387313e-05,
+      "loss": 1.2211,
+      "step": 13095
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9804642331339558e-05,
+      "loss": 1.2138,
+      "step": 13100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9804493662816208e-05,
+      "loss": 1.1673,
+      "step": 13105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9804344938303925e-05,
+      "loss": 1.224,
+      "step": 13110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9804196157803558e-05,
+      "loss": 1.2478,
+      "step": 13115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980404732131596e-05,
+      "loss": 1.4682,
+      "step": 13120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9803898428841973e-05,
+      "loss": 1.3131,
+      "step": 13125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9803749480382456e-05,
+      "loss": 1.0456,
+      "step": 13130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980360047593826e-05,
+      "loss": 1.1452,
+      "step": 13135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9803451415510227e-05,
+      "loss": 1.159,
+      "step": 13140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980330229909922e-05,
+      "loss": 1.2044,
+      "step": 13145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9803153126706078e-05,
+      "loss": 1.3093,
+      "step": 13150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9803003898331663e-05,
+      "loss": 0.9624,
+      "step": 13155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9802854613976823e-05,
+      "loss": 1.3404,
+      "step": 13160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980270527364241e-05,
+      "loss": 1.5372,
+      "step": 13165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9802555877329275e-05,
+      "loss": 1.3631,
+      "step": 13170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9802406425038275e-05,
+      "loss": 1.1539,
+      "step": 13175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9802256916770262e-05,
+      "loss": 1.3435,
+      "step": 13180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980210735252609e-05,
+      "loss": 1.0688,
+      "step": 13185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9801957732306617e-05,
+      "loss": 1.107,
+      "step": 13190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980180805611269e-05,
+      "loss": 1.2365,
+      "step": 13195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9801658323945164e-05,
+      "loss": 1.2264,
+      "step": 13200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.98015085358049e-05,
+      "loss": 1.3357,
+      "step": 13205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980135869169275e-05,
+      "loss": 1.3152,
+      "step": 13210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980120879160957e-05,
+      "loss": 0.9576,
+      "step": 13215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9801058835556218e-05,
+      "loss": 1.6253,
+      "step": 13220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9800908823533547e-05,
+      "loss": 1.1546,
+      "step": 13225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9800758755542415e-05,
+      "loss": 1.1934,
+      "step": 13230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980060863158368e-05,
+      "loss": 1.4024,
+      "step": 13235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9800458451658198e-05,
+      "loss": 1.0227,
+      "step": 13240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9800308215766826e-05,
+      "loss": 1.1489,
+      "step": 13245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9800157923910426e-05,
+      "loss": 1.0213,
+      "step": 13250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980000757608985e-05,
+      "loss": 1.3584,
+      "step": 13255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9799857172305962e-05,
+      "loss": 1.4833,
+      "step": 13260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9799706712559615e-05,
+      "loss": 1.2193,
+      "step": 13265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9799556196851676e-05,
+      "loss": 1.0727,
+      "step": 13270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9799405625183e-05,
+      "loss": 1.1131,
+      "step": 13275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9799254997554445e-05,
+      "loss": 1.3156,
+      "step": 13280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9799104313966876e-05,
+      "loss": 1.4516,
+      "step": 13285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9798953574421146e-05,
+      "loss": 1.4778,
+      "step": 13290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.979880277891813e-05,
+      "loss": 1.1571,
+      "step": 13295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.979865192745867e-05,
+      "loss": 1.2202,
+      "step": 13300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9798501020043642e-05,
+      "loss": 1.5797,
+      "step": 13305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9798350056673904e-05,
+      "loss": 1.2637,
+      "step": 13310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9798199037350314e-05,
+      "loss": 1.5373,
+      "step": 13315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9798047962073738e-05,
+      "loss": 1.3314,
+      "step": 13320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.979789683084504e-05,
+      "loss": 1.236,
+      "step": 13325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.979774564366508e-05,
+      "loss": 0.9962,
+      "step": 13330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9797594400534723e-05,
+      "loss": 1.1441,
+      "step": 13335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9797443101454834e-05,
+      "loss": 1.0728,
+      "step": 13340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9797291746426273e-05,
+      "loss": 1.1021,
+      "step": 13345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9797140335449906e-05,
+      "loss": 0.9794,
+      "step": 13350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9796988868526602e-05,
+      "loss": 1.5348,
+      "step": 13355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.979683734565722e-05,
+      "loss": 1.3052,
+      "step": 13360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9796685766842626e-05,
+      "loss": 1.4501,
+      "step": 13365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9796534132083688e-05,
+      "loss": 1.5883,
+      "step": 13370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9796382441381272e-05,
+      "loss": 1.4381,
+      "step": 13375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.979623069473624e-05,
+      "loss": 1.0612,
+      "step": 13380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9796078892149468e-05,
+      "loss": 1.3092,
+      "step": 13385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9795927033621813e-05,
+      "loss": 1.0874,
+      "step": 13390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9795775119154147e-05,
+      "loss": 1.2635,
+      "step": 13395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9795623148747334e-05,
+      "loss": 1.1457,
+      "step": 13400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9795471122402245e-05,
+      "loss": 1.0093,
+      "step": 13405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9795319040119748e-05,
+      "loss": 1.1817,
+      "step": 13410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9795166901900713e-05,
+      "loss": 1.3612,
+      "step": 13415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9795014707746003e-05,
+      "loss": 1.2705,
+      "step": 13420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9794862457656494e-05,
+      "loss": 1.2155,
+      "step": 13425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.979471015163305e-05,
+      "loss": 1.2998,
+      "step": 13430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9794557789676542e-05,
+      "loss": 1.4683,
+      "step": 13435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9794405371787845e-05,
+      "loss": 1.1735,
+      "step": 13440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.979425289796782e-05,
+      "loss": 1.1184,
+      "step": 13445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.979410036821735e-05,
+      "loss": 1.2239,
+      "step": 13450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9793947782537293e-05,
+      "loss": 1.0781,
+      "step": 13455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.979379514092853e-05,
+      "loss": 1.0708,
+      "step": 13460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9793642443391923e-05,
+      "loss": 1.2127,
+      "step": 13465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.979348968992836e-05,
+      "loss": 1.0594,
+      "step": 13470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9793336880538695e-05,
+      "loss": 1.1793,
+      "step": 13475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9793184015223813e-05,
+      "loss": 1.3467,
+      "step": 13480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9793031093984584e-05,
+      "loss": 1.2772,
+      "step": 13485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9792878116821877e-05,
+      "loss": 1.0613,
+      "step": 13490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.979272508373657e-05,
+      "loss": 0.9781,
+      "step": 13495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9792571994729535e-05,
+      "loss": 1.2955,
+      "step": 13500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.979241884980165e-05,
+      "loss": 1.3582,
+      "step": 13505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9792265648953782e-05,
+      "loss": 1.2472,
+      "step": 13510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9792112392186815e-05,
+      "loss": 0.992,
+      "step": 13515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9791959079501615e-05,
+      "loss": 1.1981,
+      "step": 13520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9791805710899065e-05,
+      "loss": 1.1977,
+      "step": 13525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9791652286380038e-05,
+      "loss": 1.5918,
+      "step": 13530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9791498805945408e-05,
+      "loss": 1.0761,
+      "step": 13535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9791345269596057e-05,
+      "loss": 1.5577,
+      "step": 13540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9791191677332854e-05,
+      "loss": 1.3756,
+      "step": 13545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979103802915668e-05,
+      "loss": 1.2849,
+      "step": 13550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9790884325068416e-05,
+      "loss": 1.2252,
+      "step": 13555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9790730565068936e-05,
+      "loss": 0.9845,
+      "step": 13560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9790576749159118e-05,
+      "loss": 1.0925,
+      "step": 13565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.979042287733984e-05,
+      "loss": 1.3067,
+      "step": 13570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9790268949611983e-05,
+      "loss": 0.983,
+      "step": 13575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9790114965976424e-05,
+      "loss": 1.5784,
+      "step": 13580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9789960926434042e-05,
+      "loss": 1.2614,
+      "step": 13585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978980683098572e-05,
+      "loss": 1.3811,
+      "step": 13590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978965267963233e-05,
+      "loss": 1.0908,
+      "step": 13595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9789498472374762e-05,
+      "loss": 1.1376,
+      "step": 13600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9789344209213893e-05,
+      "loss": 1.3131,
+      "step": 13605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9789189890150604e-05,
+      "loss": 1.3668,
+      "step": 13610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9789035515185776e-05,
+      "loss": 1.2728,
+      "step": 13615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9788881084320288e-05,
+      "loss": 1.0451,
+      "step": 13620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9788726597555022e-05,
+      "loss": 1.5374,
+      "step": 13625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9788572054890865e-05,
+      "loss": 1.512,
+      "step": 13630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9788417456328695e-05,
+      "loss": 1.5262,
+      "step": 13635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97882628018694e-05,
+      "loss": 0.9908,
+      "step": 13640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9788108091513856e-05,
+      "loss": 1.2806,
+      "step": 13645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978795332526295e-05,
+      "loss": 1.394,
+      "step": 13650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978779850311757e-05,
+      "loss": 0.9972,
+      "step": 13655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9787643625078593e-05,
+      "loss": 1.2407,
+      "step": 13660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978748869114691e-05,
+      "loss": 1.2202,
+      "step": 13665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97873337013234e-05,
+      "loss": 1.5004,
+      "step": 13670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978717865560895e-05,
+      "loss": 1.2309,
+      "step": 13675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9787023554004445e-05,
+      "loss": 1.2065,
+      "step": 13680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978686839651077e-05,
+      "loss": 1.4406,
+      "step": 13685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978671318312882e-05,
+      "loss": 1.0469,
+      "step": 13690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9786557913859466e-05,
+      "loss": 1.1471,
+      "step": 13695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978640258870361e-05,
+      "loss": 1.3044,
+      "step": 13700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9786247207662123e-05,
+      "loss": 1.0696,
+      "step": 13705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9786091770735908e-05,
+      "loss": 0.9818,
+      "step": 13710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978593627792584e-05,
+      "loss": 1.5496,
+      "step": 13715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9785780729232814e-05,
+      "loss": 1.073,
+      "step": 13720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9785625124657718e-05,
+      "loss": 1.57,
+      "step": 13725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9785469464201442e-05,
+      "loss": 1.0414,
+      "step": 13730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9785313747864868e-05,
+      "loss": 0.9996,
+      "step": 13735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9785157975648894e-05,
+      "loss": 1.7727,
+      "step": 13740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97850021475544e-05,
+      "loss": 1.2098,
+      "step": 13745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9784846263582284e-05,
+      "loss": 1.0548,
+      "step": 13750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978469032373343e-05,
+      "loss": 1.4077,
+      "step": 13755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9784534328008735e-05,
+      "loss": 1.2925,
+      "step": 13760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9784378276409083e-05,
+      "loss": 1.4908,
+      "step": 13765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978422216893537e-05,
+      "loss": 1.1222,
+      "step": 13770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9784066005588486e-05,
+      "loss": 1.2738,
+      "step": 13775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9783909786369325e-05,
+      "loss": 1.159,
+      "step": 13780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9783753511278774e-05,
+      "loss": 1.2516,
+      "step": 13785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978359718031773e-05,
+      "loss": 1.1927,
+      "step": 13790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9783440793487082e-05,
+      "loss": 1.1177,
+      "step": 13795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9783284350787725e-05,
+      "loss": 1.0095,
+      "step": 13800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9783127852220554e-05,
+      "loss": 1.3209,
+      "step": 13805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978297129778646e-05,
+      "loss": 1.4752,
+      "step": 13810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9782814687486338e-05,
+      "loss": 1.2949,
+      "step": 13815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9782658021321084e-05,
+      "loss": 1.5053,
+      "step": 13820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978250129929159e-05,
+      "loss": 1.1577,
+      "step": 13825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978234452139875e-05,
+      "loss": 1.4129,
+      "step": 13830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9782187687643467e-05,
+      "loss": 1.2844,
+      "step": 13835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9782030798026626e-05,
+      "loss": 1.115,
+      "step": 13840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9781873852549128e-05,
+      "loss": 1.0547,
+      "step": 13845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978171685121187e-05,
+      "loss": 1.2912,
+      "step": 13850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9781559794015747e-05,
+      "loss": 1.4544,
+      "step": 13855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978140268096166e-05,
+      "loss": 1.3112,
+      "step": 13860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97812455120505e-05,
+      "loss": 1.2465,
+      "step": 13865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9781088287283167e-05,
+      "loss": 1.6675,
+      "step": 13870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978093100666056e-05,
+      "loss": 1.3917,
+      "step": 13875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9780773670183576e-05,
+      "loss": 1.2308,
+      "step": 13880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9780616277853114e-05,
+      "loss": 1.2988,
+      "step": 13885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978045882967007e-05,
+      "loss": 1.0993,
+      "step": 13890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978030132563535e-05,
+      "loss": 1.0392,
+      "step": 13895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9780143765749848e-05,
+      "loss": 1.165,
+      "step": 13900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9779986150014463e-05,
+      "loss": 1.1052,
+      "step": 13905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9779828478430097e-05,
+      "loss": 1.0892,
+      "step": 13910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9779670750997647e-05,
+      "loss": 1.1647,
+      "step": 13915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9779512967718024e-05,
+      "loss": 1.2299,
+      "step": 13920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9779355128592116e-05,
+      "loss": 1.315,
+      "step": 13925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977919723362084e-05,
+      "loss": 1.3905,
+      "step": 13930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977903928280508e-05,
+      "loss": 1.0777,
+      "step": 13935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9778881276145744e-05,
+      "loss": 1.6219,
+      "step": 13940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977872321364374e-05,
+      "loss": 1.2348,
+      "step": 13945
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9778565095299966e-05,
+      "loss": 1.1503,
+      "step": 13950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9778406921115326e-05,
+      "loss": 1.0771,
+      "step": 13955
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9778248691090724e-05,
+      "loss": 1.1504,
+      "step": 13960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9778090405227063e-05,
+      "loss": 1.0526,
+      "step": 13965
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9777932063525245e-05,
+      "loss": 1.2079,
+      "step": 13970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9777773665986175e-05,
+      "loss": 1.0065,
+      "step": 13975
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9777615212610765e-05,
+      "loss": 1.7257,
+      "step": 13980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9777456703399905e-05,
+      "loss": 1.4592,
+      "step": 13985
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9777298138354513e-05,
+      "loss": 1.2991,
+      "step": 13990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9777139517475488e-05,
+      "loss": 1.2264,
+      "step": 13995
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9776980840763737e-05,
+      "loss": 1.17,
+      "step": 14000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977682210822017e-05,
+      "loss": 1.2731,
+      "step": 14005
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9776663319845687e-05,
+      "loss": 1.2458,
+      "step": 14010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9776504475641198e-05,
+      "loss": 1.0652,
+      "step": 14015
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9776345575607615e-05,
+      "loss": 1.1585,
+      "step": 14020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9776186619745836e-05,
+      "loss": 1.2945,
+      "step": 14025
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9776027608056773e-05,
+      "loss": 1.1888,
+      "step": 14030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9775868540541337e-05,
+      "loss": 1.1624,
+      "step": 14035
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977570941720043e-05,
+      "loss": 1.6292,
+      "step": 14040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9775550238034968e-05,
+      "loss": 1.198,
+      "step": 14045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9775391003045855e-05,
+      "loss": 1.144,
+      "step": 14050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9775231712234e-05,
+      "loss": 1.1303,
+      "step": 14055
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9775072365600318e-05,
+      "loss": 1.5683,
+      "step": 14060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9774912963145713e-05,
+      "loss": 1.0936,
+      "step": 14065
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97747535048711e-05,
+      "loss": 1.1034,
+      "step": 14070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9774593990777382e-05,
+      "loss": 1.741,
+      "step": 14075
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977443442086548e-05,
+      "loss": 1.2064,
+      "step": 14080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9774274795136296e-05,
+      "loss": 1.1082,
+      "step": 14085
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977411511359075e-05,
+      "loss": 1.189,
+      "step": 14090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9773955376229746e-05,
+      "loss": 1.0726,
+      "step": 14095
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9773795583054204e-05,
+      "loss": 1.4682,
+      "step": 14100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977363573406503e-05,
+      "loss": 1.179,
+      "step": 14105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9773475829263136e-05,
+      "loss": 1.5863,
+      "step": 14110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9773315868649442e-05,
+      "loss": 1.3353,
+      "step": 14115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9773155852224855e-05,
+      "loss": 1.1899,
+      "step": 14120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9772995779990294e-05,
+      "loss": 1.2714,
+      "step": 14125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977283565194667e-05,
+      "loss": 1.2612,
+      "step": 14130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97726754680949e-05,
+      "loss": 1.1677,
+      "step": 14135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9772515228435894e-05,
+      "loss": 1.5376,
+      "step": 14140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9772354932970574e-05,
+      "loss": 1.0941,
+      "step": 14145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9772194581699846e-05,
+      "loss": 1.0866,
+      "step": 14150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9772034174624634e-05,
+      "loss": 1.471,
+      "step": 14155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977187371174585e-05,
+      "loss": 1.0921,
+      "step": 14160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977171319306441e-05,
+      "loss": 1.5361,
+      "step": 14165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9771552618581237e-05,
+      "loss": 1.1156,
+      "step": 14170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9771391988297237e-05,
+      "loss": 1.1451,
+      "step": 14175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9771231302213335e-05,
+      "loss": 1.1993,
+      "step": 14180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9771070560330447e-05,
+      "loss": 1.431,
+      "step": 14185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9770909762649494e-05,
+      "loss": 1.2167,
+      "step": 14190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9770748909171388e-05,
+      "loss": 1.544,
+      "step": 14195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.977058799989705e-05,
+      "loss": 1.2978,
+      "step": 14200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9770427034827396e-05,
+      "loss": 1.4193,
+      "step": 14205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9770266013963355e-05,
+      "loss": 1.176,
+      "step": 14210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9770104937305836e-05,
+      "loss": 1.4377,
+      "step": 14215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9769943804855768e-05,
+      "loss": 1.1283,
+      "step": 14220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976978261661406e-05,
+      "loss": 1.1493,
+      "step": 14225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9769621372581644e-05,
+      "loss": 1.2261,
+      "step": 14230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976946007275943e-05,
+      "loss": 1.1999,
+      "step": 14235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9769298717148348e-05,
+      "loss": 1.268,
+      "step": 14240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9769137305749314e-05,
+      "loss": 1.1362,
+      "step": 14245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9768975838563253e-05,
+      "loss": 1.6004,
+      "step": 14250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9768814315591084e-05,
+      "loss": 1.4566,
+      "step": 14255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9768652736833732e-05,
+      "loss": 1.0395,
+      "step": 14260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976849110229212e-05,
+      "loss": 1.3457,
+      "step": 14265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976832941196717e-05,
+      "loss": 1.3503,
+      "step": 14270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9768167665859804e-05,
+      "loss": 1.0128,
+      "step": 14275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9768005863970945e-05,
+      "loss": 1.3222,
+      "step": 14280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9767844006301523e-05,
+      "loss": 1.0553,
+      "step": 14285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9767682092852454e-05,
+      "loss": 1.2453,
+      "step": 14290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9767520123624667e-05,
+      "loss": 1.4413,
+      "step": 14295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976735809861909e-05,
+      "loss": 1.3606,
+      "step": 14300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9767196017836642e-05,
+      "loss": 1.3839,
+      "step": 14305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976703388127825e-05,
+      "loss": 0.9436,
+      "step": 14310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9766871688944845e-05,
+      "loss": 1.3605,
+      "step": 14315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976670944083735e-05,
+      "loss": 1.1908,
+      "step": 14320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976654713695669e-05,
+      "loss": 1.714,
+      "step": 14325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976638477730379e-05,
+      "loss": 1.557,
+      "step": 14330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976622236187958e-05,
+      "loss": 1.1348,
+      "step": 14335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9766059890684988e-05,
+      "loss": 1.0021,
+      "step": 14340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9765897363720947e-05,
+      "loss": 1.2492,
+      "step": 14345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9765734780988375e-05,
+      "loss": 1.1582,
+      "step": 14350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9765572142488204e-05,
+      "loss": 1.424,
+      "step": 14355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9765409448221365e-05,
+      "loss": 1.5962,
+      "step": 14360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9765246698188788e-05,
+      "loss": 1.6204,
+      "step": 14365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9765083892391397e-05,
+      "loss": 1.2852,
+      "step": 14370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9764921030830124e-05,
+      "loss": 0.926,
+      "step": 14375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97647581135059e-05,
+      "loss": 1.3003,
+      "step": 14380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976459514041966e-05,
+      "loss": 1.3443,
+      "step": 14385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9764432111572326e-05,
+      "loss": 1.2654,
+      "step": 14390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976426902696483e-05,
+      "loss": 0.957,
+      "step": 14395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976410588659811e-05,
+      "loss": 1.1698,
+      "step": 14400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9763942690473094e-05,
+      "loss": 1.6755,
+      "step": 14405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976377943859071e-05,
+      "loss": 1.4141,
+      "step": 14410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9763616130951895e-05,
+      "loss": 1.4968,
+      "step": 14415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9763452767557584e-05,
+      "loss": 1.4074,
+      "step": 14420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9763289348408704e-05,
+      "loss": 1.4433,
+      "step": 14425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9763125873506192e-05,
+      "loss": 1.395,
+      "step": 14430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976296234285098e-05,
+      "loss": 1.6042,
+      "step": 14435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9762798756444e-05,
+      "loss": 1.0158,
+      "step": 14440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976263511428619e-05,
+      "loss": 1.5804,
+      "step": 14445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9762471416378483e-05,
+      "loss": 1.1277,
+      "step": 14450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976230766272181e-05,
+      "loss": 0.9173,
+      "step": 14455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976214385331711e-05,
+      "loss": 1.3267,
+      "step": 14460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9761979988165324e-05,
+      "loss": 1.238,
+      "step": 14465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9761816067267377e-05,
+      "loss": 1.2528,
+      "step": 14470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9761652090624214e-05,
+      "loss": 1.0022,
+      "step": 14475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9761488058236762e-05,
+      "loss": 1.3529,
+      "step": 14480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9761323970105963e-05,
+      "loss": 1.5728,
+      "step": 14485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9761159826232757e-05,
+      "loss": 1.0034,
+      "step": 14490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9760995626618077e-05,
+      "loss": 1.6662,
+      "step": 14495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9760831371262862e-05,
+      "loss": 1.2596,
+      "step": 14500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9760667060168052e-05,
+      "loss": 1.2383,
+      "step": 14505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.976050269333458e-05,
+      "loss": 1.381,
+      "step": 14510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9760338270763387e-05,
+      "loss": 1.2813,
+      "step": 14515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9760173792455417e-05,
+      "loss": 1.0319,
+      "step": 14520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9760009258411604e-05,
+      "loss": 1.5068,
+      "step": 14525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9759844668632884e-05,
+      "loss": 1.251,
+      "step": 14530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975968002312021e-05,
+      "loss": 1.6257,
+      "step": 14535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975951532187451e-05,
+      "loss": 1.1887,
+      "step": 14540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9759350564896727e-05,
+      "loss": 1.0677,
+      "step": 14545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9759185752187803e-05,
+      "loss": 1.1367,
+      "step": 14550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975902088374868e-05,
+      "loss": 1.141,
+      "step": 14555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97588559595803e-05,
+      "loss": 1.0912,
+      "step": 14560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9758690979683603e-05,
+      "loss": 1.0714,
+      "step": 14565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975852594405953e-05,
+      "loss": 1.3236,
+      "step": 14570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9758360852709028e-05,
+      "loss": 1.0357,
+      "step": 14575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9758195705633034e-05,
+      "loss": 1.2008,
+      "step": 14580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9758030502832498e-05,
+      "loss": 1.062,
+      "step": 14585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9757865244308355e-05,
+      "loss": 1.2549,
+      "step": 14590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9757699930061556e-05,
+      "loss": 1.0625,
+      "step": 14595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975753456009304e-05,
+      "loss": 1.2575,
+      "step": 14600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9757369134403756e-05,
+      "loss": 1.4566,
+      "step": 14605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9757203652994646e-05,
+      "loss": 1.0361,
+      "step": 14610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9757038115866655e-05,
+      "loss": 1.0932,
+      "step": 14615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9756872523020728e-05,
+      "loss": 1.137,
+      "step": 14620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975670687445781e-05,
+      "loss": 1.3862,
+      "step": 14625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9756541170178848e-05,
+      "loss": 1.2399,
+      "step": 14630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9756375410184793e-05,
+      "loss": 1.0176,
+      "step": 14635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975620959447658e-05,
+      "loss": 1.4383,
+      "step": 14640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975604372305517e-05,
+      "loss": 1.3198,
+      "step": 14645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9755877795921497e-05,
+      "loss": 1.4935,
+      "step": 14650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9755711813076516e-05,
+      "loss": 1.2217,
+      "step": 14655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9755545774521178e-05,
+      "loss": 1.4302,
+      "step": 14660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9755379680256422e-05,
+      "loss": 1.2526,
+      "step": 14665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97552135302832e-05,
+      "loss": 1.3066,
+      "step": 14670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9755047324602465e-05,
+      "loss": 1.3903,
+      "step": 14675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975488106321516e-05,
+      "loss": 1.0557,
+      "step": 14680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975471474612224e-05,
+      "loss": 1.4287,
+      "step": 14685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9754548373324652e-05,
+      "loss": 1.2901,
+      "step": 14690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9754381944823348e-05,
+      "loss": 1.2356,
+      "step": 14695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9754215460619274e-05,
+      "loss": 1.5005,
+      "step": 14700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975404892071338e-05,
+      "loss": 1.2276,
+      "step": 14705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9753882325106625e-05,
+      "loss": 1.1326,
+      "step": 14710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9753715673799956e-05,
+      "loss": 1.0573,
+      "step": 14715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9753548966794324e-05,
+      "loss": 1.074,
+      "step": 14720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975338220409068e-05,
+      "loss": 1.1805,
+      "step": 14725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975321538568998e-05,
+      "loss": 1.3621,
+      "step": 14730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9753048511593172e-05,
+      "loss": 1.271,
+      "step": 14735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9752881581801215e-05,
+      "loss": 1.1573,
+      "step": 14740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9752714596315054e-05,
+      "loss": 1.1924,
+      "step": 14745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975254755513565e-05,
+      "loss": 1.1987,
+      "step": 14750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9752380458263953e-05,
+      "loss": 1.1371,
+      "step": 14755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975221330570092e-05,
+      "loss": 1.2059,
+      "step": 14760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97520460974475e-05,
+      "loss": 1.1735,
+      "step": 14765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9751878833504655e-05,
+      "loss": 1.362,
+      "step": 14770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9751711513873337e-05,
+      "loss": 1.3088,
+      "step": 14775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9751544138554498e-05,
+      "loss": 1.1643,
+      "step": 14780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9751376707549097e-05,
+      "loss": 1.5467,
+      "step": 14785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975120922085809e-05,
+      "loss": 1.5015,
+      "step": 14790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975104167848244e-05,
+      "loss": 1.2636,
+      "step": 14795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975087408042309e-05,
+      "loss": 1.2623,
+      "step": 14800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975070642668101e-05,
+      "loss": 1.4984,
+      "step": 14805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975053871725715e-05,
+      "loss": 1.3931,
+      "step": 14810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9750370952152468e-05,
+      "loss": 1.3762,
+      "step": 14815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9750203131367923e-05,
+      "loss": 1.215,
+      "step": 14820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9750035254904477e-05,
+      "loss": 1.86,
+      "step": 14825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9749867322763085e-05,
+      "loss": 1.4761,
+      "step": 14830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9749699334944704e-05,
+      "loss": 1.2262,
+      "step": 14835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9749531291450298e-05,
+      "loss": 1.2792,
+      "step": 14840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9749363192280822e-05,
+      "loss": 1.3588,
+      "step": 14845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9749195037437244e-05,
+      "loss": 1.0806,
+      "step": 14850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9749026826920517e-05,
+      "loss": 1.187,
+      "step": 14855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97488585607316e-05,
+      "loss": 1.2269,
+      "step": 14860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974869023887146e-05,
+      "loss": 1.4904,
+      "step": 14865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974852186134105e-05,
+      "loss": 1.2003,
+      "step": 14870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974835342814134e-05,
+      "loss": 1.3796,
+      "step": 14875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9748184939273295e-05,
+      "loss": 1.1028,
+      "step": 14880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9748016394737864e-05,
+      "loss": 1.3552,
+      "step": 14885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974784779453602e-05,
+      "loss": 1.277,
+      "step": 14890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974767913866872e-05,
+      "loss": 1.1546,
+      "step": 14895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974751042713693e-05,
+      "loss": 1.1861,
+      "step": 14900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974734165994161e-05,
+      "loss": 1.0032,
+      "step": 14905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9747172837083732e-05,
+      "loss": 1.2061,
+      "step": 14910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974700395856425e-05,
+      "loss": 1.3137,
+      "step": 14915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9746835024384135e-05,
+      "loss": 1.4322,
+      "step": 14920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9746666034544347e-05,
+      "loss": 1.6461,
+      "step": 14925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9746496989045858e-05,
+      "loss": 1.268,
+      "step": 14930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9746327887889624e-05,
+      "loss": 1.3156,
+      "step": 14935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974615873107662e-05,
+      "loss": 1.4782,
+      "step": 14940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9745989518607804e-05,
+      "loss": 1.0874,
+      "step": 14945
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9745820250484147e-05,
+      "loss": 1.4867,
+      "step": 14950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9745650926706618e-05,
+      "loss": 1.1763,
+      "step": 14955
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974548154727618e-05,
+      "loss": 1.1794,
+      "step": 14960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9745312112193798e-05,
+      "loss": 1.1295,
+      "step": 14965
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9745142621460443e-05,
+      "loss": 1.4586,
+      "step": 14970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974497307507708e-05,
+      "loss": 1.3261,
+      "step": 14975
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974480347304468e-05,
+      "loss": 1.0368,
+      "step": 14980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9744633815364216e-05,
+      "loss": 1.3124,
+      "step": 14985
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974446410203665e-05,
+      "loss": 1.4999,
+      "step": 14990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9744294333062947e-05,
+      "loss": 1.2344,
+      "step": 14995
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974412450844409e-05,
+      "loss": 1.2119,
+      "step": 15000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9743954628181037e-05,
+      "loss": 0.9401,
+      "step": 15005
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9743784692274764e-05,
+      "loss": 1.5097,
+      "step": 15010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9743614700726237e-05,
+      "loss": 1.0179,
+      "step": 15015
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974344465353643e-05,
+      "loss": 1.6061,
+      "step": 15020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9743274550706318e-05,
+      "loss": 1.4228,
+      "step": 15025
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9743104392236863e-05,
+      "loss": 1.5211,
+      "step": 15030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9742934178129043e-05,
+      "loss": 1.3783,
+      "step": 15035
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9742763908383827e-05,
+      "loss": 1.4347,
+      "step": 15040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9742593583002194e-05,
+      "loss": 1.1683,
+      "step": 15045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9742423201985105e-05,
+      "loss": 1.0761,
+      "step": 15050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9742252765333544e-05,
+      "loss": 1.3955,
+      "step": 15055
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9742082273048476e-05,
+      "loss": 1.3349,
+      "step": 15060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974191172513088e-05,
+      "loss": 1.3317,
+      "step": 15065
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9741741121581734e-05,
+      "loss": 1.3879,
+      "step": 15070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9741570462402e-05,
+      "loss": 1.4654,
+      "step": 15075
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974139974759266e-05,
+      "loss": 1.0903,
+      "step": 15080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974122897715469e-05,
+      "loss": 1.2379,
+      "step": 15085
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974105815108906e-05,
+      "loss": 1.3298,
+      "step": 15090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9740887269396752e-05,
+      "loss": 1.4374,
+      "step": 15095
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9740716332078738e-05,
+      "loss": 1.2921,
+      "step": 15100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9740545339135995e-05,
+      "loss": 1.1166,
+      "step": 15105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9740374290569496e-05,
+      "loss": 1.3216,
+      "step": 15110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9740203186380224e-05,
+      "loss": 1.4808,
+      "step": 15115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9740032026569152e-05,
+      "loss": 1.3682,
+      "step": 15120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973986081113726e-05,
+      "loss": 1.3726,
+      "step": 15125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973968954008552e-05,
+      "loss": 1.0998,
+      "step": 15130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973951821341492e-05,
+      "loss": 1.1127,
+      "step": 15135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973934683112643e-05,
+      "loss": 1.1208,
+      "step": 15140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973917539322103e-05,
+      "loss": 1.5446,
+      "step": 15145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97390038996997e-05,
+      "loss": 1.1838,
+      "step": 15150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973883235056342e-05,
+      "loss": 1.2682,
+      "step": 15155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973866074581317e-05,
+      "loss": 1.3902,
+      "step": 15160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973848908544993e-05,
+      "loss": 1.3935,
+      "step": 15165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9738317369474677e-05,
+      "loss": 1.5397,
+      "step": 15170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9738145597888397e-05,
+      "loss": 1.1733,
+      "step": 15175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9737973770692066e-05,
+      "loss": 1.2874,
+      "step": 15180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9737801887886666e-05,
+      "loss": 1.304,
+      "step": 15185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973762994947318e-05,
+      "loss": 1.1904,
+      "step": 15190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9737457955452596e-05,
+      "loss": 1.3416,
+      "step": 15195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9737285905825884e-05,
+      "loss": 1.773,
+      "step": 15200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973711380059403e-05,
+      "loss": 1.1408,
+      "step": 15205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9736941639758025e-05,
+      "loss": 1.3796,
+      "step": 15210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9736769423318843e-05,
+      "loss": 1.2046,
+      "step": 15215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973659715127747e-05,
+      "loss": 1.2157,
+      "step": 15220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973642482363489e-05,
+      "loss": 1.4964,
+      "step": 15225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9736252440392092e-05,
+      "loss": 1.262,
+      "step": 15230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973608000155005e-05,
+      "loss": 1.1416,
+      "step": 15235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973590750710976e-05,
+      "loss": 1.161,
+      "step": 15240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9735734957072198e-05,
+      "loss": 1.1395,
+      "step": 15245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9735562351438355e-05,
+      "loss": 0.9982,
+      "step": 15250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9735389690209213e-05,
+      "loss": 1.48,
+      "step": 15255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9735216973385757e-05,
+      "loss": 1.3588,
+      "step": 15260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973504420096898e-05,
+      "loss": 1.04,
+      "step": 15265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9734871372959866e-05,
+      "loss": 1.1045,
+      "step": 15270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9734698489359398e-05,
+      "loss": 1.0717,
+      "step": 15275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9734525550168567e-05,
+      "loss": 1.102,
+      "step": 15280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9734352555388357e-05,
+      "loss": 1.3167,
+      "step": 15285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973417950501976e-05,
+      "loss": 1.2849,
+      "step": 15290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973400639906376e-05,
+      "loss": 1.4344,
+      "step": 15295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9733833237521355e-05,
+      "loss": 1.2402,
+      "step": 15300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9733660020393522e-05,
+      "loss": 1.2636,
+      "step": 15305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9733486747681253e-05,
+      "loss": 1.4483,
+      "step": 15310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9733313419385543e-05,
+      "loss": 1.3242,
+      "step": 15315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9733140035507376e-05,
+      "loss": 1.3449,
+      "step": 15320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9732966596047747e-05,
+      "loss": 1.0811,
+      "step": 15325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973279310100764e-05,
+      "loss": 1.1809,
+      "step": 15330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9732619550388053e-05,
+      "loss": 1.2181,
+      "step": 15335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973244594418997e-05,
+      "loss": 1.3495,
+      "step": 15340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9732272282414392e-05,
+      "loss": 1.195,
+      "step": 15345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97320985650623e-05,
+      "loss": 1.0447,
+      "step": 15350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9731924792134694e-05,
+      "loss": 1.216,
+      "step": 15355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973175096363256e-05,
+      "loss": 0.9649,
+      "step": 15360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9731577079556895e-05,
+      "loss": 1.2329,
+      "step": 15365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9731403139908688e-05,
+      "loss": 1.1758,
+      "step": 15370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9731229144688943e-05,
+      "loss": 1.1434,
+      "step": 15375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9731055093898635e-05,
+      "loss": 1.2327,
+      "step": 15380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9730880987538776e-05,
+      "loss": 1.1501,
+      "step": 15385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973070682561035e-05,
+      "loss": 1.3594,
+      "step": 15390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9730532608114353e-05,
+      "loss": 1.2418,
+      "step": 15395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9730358335051785e-05,
+      "loss": 1.5328,
+      "step": 15400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9730184006423636e-05,
+      "loss": 1.4477,
+      "step": 15405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9730009622230905e-05,
+      "loss": 1.2124,
+      "step": 15410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.972983518247458e-05,
+      "loss": 1.2199,
+      "step": 15415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9729660687155667e-05,
+      "loss": 1.1686,
+      "step": 15420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9729486136275156e-05,
+      "loss": 1.6925,
+      "step": 15425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.972931152983405e-05,
+      "loss": 1.0877,
+      "step": 15430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.972913686783334e-05,
+      "loss": 1.2439,
+      "step": 15435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9728962150274026e-05,
+      "loss": 1.0404,
+      "step": 15440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9728787377157104e-05,
+      "loss": 1.4994,
+      "step": 15445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9728612548483573e-05,
+      "loss": 0.9652,
+      "step": 15450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9728437664254435e-05,
+      "loss": 1.5395,
+      "step": 15455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9728262724470683e-05,
+      "loss": 0.9795,
+      "step": 15460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.972808772913332e-05,
+      "loss": 0.9239,
+      "step": 15465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.972791267824334e-05,
+      "loss": 1.0425,
+      "step": 15470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.972773757180175e-05,
+      "loss": 1.0715,
+      "step": 15475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9727562409809546e-05,
+      "loss": 1.0822,
+      "step": 15480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.972738719226773e-05,
+      "loss": 0.8867,
+      "step": 15485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9727211919177298e-05,
+      "loss": 0.963,
+      "step": 15490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9727036590539256e-05,
+      "loss": 0.915,
+      "step": 15495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9726861206354603e-05,
+      "loss": 1.1776,
+      "step": 15500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.972668576662434e-05,
+      "loss": 1.2738,
+      "step": 15505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9726510271349474e-05,
+      "loss": 1.0912,
+      "step": 15510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9726334720530997e-05,
+      "loss": 1.2145,
+      "step": 15515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.972615911416992e-05,
+      "loss": 1.1369,
+      "step": 15520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9725983452267243e-05,
+      "loss": 0.971,
+      "step": 15525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.972580773482397e-05,
+      "loss": 1.3098,
+      "step": 15530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9725631961841102e-05,
+      "loss": 1.795,
+      "step": 15535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9725456133319648e-05,
+      "loss": 1.1502,
+      "step": 15540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9725280249260605e-05,
+      "loss": 1.1811,
+      "step": 15545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9725104309664986e-05,
+      "loss": 1.3481,
+      "step": 15550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9724928314533784e-05,
+      "loss": 1.536,
+      "step": 15555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9724752263868015e-05,
+      "loss": 1.4411,
+      "step": 15560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9724576157668676e-05,
+      "loss": 1.1218,
+      "step": 15565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9724399995936783e-05,
+      "loss": 0.9266,
+      "step": 15570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9724223778673332e-05,
+      "loss": 1.2832,
+      "step": 15575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9724047505879333e-05,
+      "loss": 2.1365,
+      "step": 15580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9723871177555795e-05,
+      "loss": 1.4994,
+      "step": 15585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972369479370372e-05,
+      "loss": 1.3057,
+      "step": 15590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972351835432412e-05,
+      "loss": 0.9284,
+      "step": 15595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9723341859417996e-05,
+      "loss": 1.5208,
+      "step": 15600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9723165308986366e-05,
+      "loss": 1.1645,
+      "step": 15605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972298870303023e-05,
+      "loss": 1.2676,
+      "step": 15610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9722812041550598e-05,
+      "loss": 1.3883,
+      "step": 15615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.972263532454848e-05,
+      "loss": 1.2109,
+      "step": 15620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9722458552024885e-05,
+      "loss": 1.4427,
+      "step": 15625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9722281723980823e-05,
+      "loss": 1.3138,
+      "step": 15630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9722104840417305e-05,
+      "loss": 1.2832,
+      "step": 15635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9721927901335337e-05,
+      "loss": 1.4345,
+      "step": 15640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9721750906735934e-05,
+      "loss": 1.3504,
+      "step": 15645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97215738566201e-05,
+      "loss": 1.1426,
+      "step": 15650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9721396750988853e-05,
+      "loss": 1.2222,
+      "step": 15655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9721219589843203e-05,
+      "loss": 0.9629,
+      "step": 15660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9721042373184158e-05,
+      "loss": 1.1027,
+      "step": 15665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9720865101012735e-05,
+      "loss": 1.2213,
+      "step": 15670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9720687773329944e-05,
+      "loss": 1.0986,
+      "step": 15675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9720510390136795e-05,
+      "loss": 1.5026,
+      "step": 15680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9720332951434304e-05,
+      "loss": 1.2317,
+      "step": 15685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9720155457223487e-05,
+      "loss": 1.6807,
+      "step": 15690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971997790750535e-05,
+      "loss": 1.5379,
+      "step": 15695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9719800302280913e-05,
+      "loss": 1.4157,
+      "step": 15700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9719622641551188e-05,
+      "loss": 1.4356,
+      "step": 15705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9719444925317193e-05,
+      "loss": 1.2542,
+      "step": 15710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9719267153579937e-05,
+      "loss": 1.4353,
+      "step": 15715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9719089326340438e-05,
+      "loss": 1.2577,
+      "step": 15720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971891144359971e-05,
+      "loss": 1.1306,
+      "step": 15725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9718733505358774e-05,
+      "loss": 1.3191,
+      "step": 15730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9718555511618642e-05,
+      "loss": 1.3243,
+      "step": 15735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971837746238033e-05,
+      "loss": 1.0789,
+      "step": 15740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9718199357644857e-05,
+      "loss": 1.453,
+      "step": 15745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9718021197413237e-05,
+      "loss": 1.1408,
+      "step": 15750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9717842981686487e-05,
+      "loss": 1.2565,
+      "step": 15755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971766471046563e-05,
+      "loss": 1.2158,
+      "step": 15760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971748638375168e-05,
+      "loss": 1.3727,
+      "step": 15765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9717308001545652e-05,
+      "loss": 1.287,
+      "step": 15770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9717129563848577e-05,
+      "loss": 1.1353,
+      "step": 15775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971695107066146e-05,
+      "loss": 1.189,
+      "step": 15780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971677252198533e-05,
+      "loss": 1.0284,
+      "step": 15785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97165939178212e-05,
+      "loss": 1.1468,
+      "step": 15790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9716415258170093e-05,
+      "loss": 1.1881,
+      "step": 15795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971623654303303e-05,
+      "loss": 1.2126,
+      "step": 15800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9716057772411027e-05,
+      "loss": 1.4225,
+      "step": 15805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9715878946305112e-05,
+      "loss": 1.1671,
+      "step": 15810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97157000647163e-05,
+      "loss": 1.4368,
+      "step": 15815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9715521127645617e-05,
+      "loss": 1.3437,
+      "step": 15820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9715342135094083e-05,
+      "loss": 1.2049,
+      "step": 15825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9715163087062718e-05,
+      "loss": 1.3388,
+      "step": 15830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9714983983552548e-05,
+      "loss": 1.2291,
+      "step": 15835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9714804824564596e-05,
+      "loss": 1.2878,
+      "step": 15840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971462561009988e-05,
+      "loss": 1.3497,
+      "step": 15845
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9714446340159424e-05,
+      "loss": 1.2354,
+      "step": 15850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971426701474426e-05,
+      "loss": 1.2489,
+      "step": 15855
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9714087633855404e-05,
+      "loss": 1.0576,
+      "step": 15860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9713908197493887e-05,
+      "loss": 1.4905,
+      "step": 15865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9713728705660723e-05,
+      "loss": 1.1654,
+      "step": 15870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9713549158356947e-05,
+      "loss": 1.5846,
+      "step": 15875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9713369555583583e-05,
+      "loss": 1.6721,
+      "step": 15880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9713189897341652e-05,
+      "loss": 1.1954,
+      "step": 15885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9713010183632182e-05,
+      "loss": 1.0921,
+      "step": 15890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97128304144562e-05,
+      "loss": 1.3838,
+      "step": 15895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9712650589814732e-05,
+      "loss": 1.2691,
+      "step": 15900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9712470709708807e-05,
+      "loss": 1.2294,
+      "step": 15905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971229077413945e-05,
+      "loss": 1.2644,
+      "step": 15910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9712110783107684e-05,
+      "loss": 1.0717,
+      "step": 15915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9711930736614548e-05,
+      "loss": 1.1924,
+      "step": 15920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9711750634661062e-05,
+      "loss": 1.1272,
+      "step": 15925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9711570477248257e-05,
+      "loss": 1.1987,
+      "step": 15930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971139026437716e-05,
+      "loss": 1.2453,
+      "step": 15935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97112099960488e-05,
+      "loss": 1.2214,
+      "step": 15940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971102967226421e-05,
+      "loss": 1.5473,
+      "step": 15945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9710849293024416e-05,
+      "loss": 1.5956,
+      "step": 15950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.971066885833045e-05,
+      "loss": 1.1999,
+      "step": 15955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9710488368183343e-05,
+      "loss": 1.2443,
+      "step": 15960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9710307822584123e-05,
+      "loss": 1.4771,
+      "step": 15965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9710127221533824e-05,
+      "loss": 1.236,
+      "step": 15970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9709946565033475e-05,
+      "loss": 1.3041,
+      "step": 15975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970976585308411e-05,
+      "loss": 1.0582,
+      "step": 15980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970958508568676e-05,
+      "loss": 0.9796,
+      "step": 15985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9709404262842454e-05,
+      "loss": 1.2531,
+      "step": 15990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9709223384552227e-05,
+      "loss": 1.05,
+      "step": 15995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9709042450817117e-05,
+      "loss": 1.3577,
+      "step": 16000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9708861461638147e-05,
+      "loss": 1.1551,
+      "step": 16005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9708680417016358e-05,
+      "loss": 1.5977,
+      "step": 16010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9708499316952783e-05,
+      "loss": 1.3012,
+      "step": 16015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9708318161448456e-05,
+      "loss": 1.1312,
+      "step": 16020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9708136950504412e-05,
+      "loss": 1.1627,
+      "step": 16025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970795568412168e-05,
+      "loss": 1.1391,
+      "step": 16030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97077743623013e-05,
+      "loss": 1.1811,
+      "step": 16035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9707592985044307e-05,
+      "loss": 1.2002,
+      "step": 16040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970741155235174e-05,
+      "loss": 0.957,
+      "step": 16045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9707230064224632e-05,
+      "loss": 1.4292,
+      "step": 16050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9707048520664015e-05,
+      "loss": 1.1699,
+      "step": 16055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9706866921670932e-05,
+      "loss": 1.4367,
+      "step": 16060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970668526724642e-05,
+      "loss": 1.0751,
+      "step": 16065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9706503557391515e-05,
+      "loss": 1.1302,
+      "step": 16070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9706321792107254e-05,
+      "loss": 1.0667,
+      "step": 16075
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9706139971394668e-05,
+      "loss": 1.232,
+      "step": 16080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970595809525481e-05,
+      "loss": 0.9992,
+      "step": 16085
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970577616368871e-05,
+      "loss": 1.0748,
+      "step": 16090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9705594176697405e-05,
+      "loss": 1.3796,
+      "step": 16095
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9705412134281938e-05,
+      "loss": 1.5954,
+      "step": 16100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9705230036443346e-05,
+      "loss": 1.3948,
+      "step": 16105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9705047883182674e-05,
+      "loss": 1.1013,
+      "step": 16110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9704865674500958e-05,
+      "loss": 1.4512,
+      "step": 16115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9704683410399235e-05,
+      "loss": 1.5279,
+      "step": 16120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9704501090878555e-05,
+      "loss": 1.5556,
+      "step": 16125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970431871593995e-05,
+      "loss": 1.1216,
+      "step": 16130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9704136285584468e-05,
+      "loss": 1.1752,
+      "step": 16135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970395379981315e-05,
+      "loss": 1.1828,
+      "step": 16140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9703771258627036e-05,
+      "loss": 1.3863,
+      "step": 16145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9703588662027168e-05,
+      "loss": 1.1617,
+      "step": 16150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9703406010014587e-05,
+      "loss": 1.2538,
+      "step": 16155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9703223302590344e-05,
+      "loss": 1.0715,
+      "step": 16160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9703040539755473e-05,
+      "loss": 1.1747,
+      "step": 16165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9702857721511023e-05,
+      "loss": 0.9884,
+      "step": 16170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9702674847858038e-05,
+      "loss": 1.0798,
+      "step": 16175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9702491918797558e-05,
+      "loss": 1.1964,
+      "step": 16180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970230893433063e-05,
+      "loss": 1.7193,
+      "step": 16185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9702125894458306e-05,
+      "loss": 1.1751,
+      "step": 16190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970194279918162e-05,
+      "loss": 1.1304,
+      "step": 16195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9701759648501624e-05,
+      "loss": 1.3635,
+      "step": 16200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9701576442419358e-05,
+      "loss": 1.2272,
+      "step": 16205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970139318093588e-05,
+      "loss": 1.0026,
+      "step": 16210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9701209864052223e-05,
+      "loss": 1.2087,
+      "step": 16215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970102649176944e-05,
+      "loss": 1.422,
+      "step": 16220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9700843064088583e-05,
+      "loss": 1.419,
+      "step": 16225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970065958101069e-05,
+      "loss": 1.3544,
+      "step": 16230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9700476042536815e-05,
+      "loss": 1.569,
+      "step": 16235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9700292448668002e-05,
+      "loss": 1.6226,
+      "step": 16240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9700108799405308e-05,
+      "loss": 1.3621,
+      "step": 16245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969992509474977e-05,
+      "loss": 1.1167,
+      "step": 16250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9699741334702443e-05,
+      "loss": 1.3079,
+      "step": 16255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9699557519264376e-05,
+      "loss": 1.3442,
+      "step": 16260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969937364843662e-05,
+      "loss": 1.5602,
+      "step": 16265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9699189722220226e-05,
+      "loss": 1.0262,
+      "step": 16270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9699005740616237e-05,
+      "loss": 1.1971,
+      "step": 16275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969882170362571e-05,
+      "loss": 1.1395,
+      "step": 16280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9698637611249694e-05,
+      "loss": 1.1808,
+      "step": 16285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9698453463489242e-05,
+      "loss": 1.1041,
+      "step": 16290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9698269260345405e-05,
+      "loss": 1.0628,
+      "step": 16295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9698085001819233e-05,
+      "loss": 1.0795,
+      "step": 16300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969790068791178e-05,
+      "loss": 1.0438,
+      "step": 16305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9697716318624098e-05,
+      "loss": 1.4543,
+      "step": 16310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969753189395724e-05,
+      "loss": 1.2067,
+      "step": 16315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969734741391226e-05,
+      "loss": 1.6576,
+      "step": 16320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969716287849021e-05,
+      "loss": 1.5822,
+      "step": 16325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9696978287692143e-05,
+      "loss": 1.3374,
+      "step": 16330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9696793641519117e-05,
+      "loss": 1.3953,
+      "step": 16335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9696608939972183e-05,
+      "loss": 1.7317,
+      "step": 16340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9696424183052394e-05,
+      "loss": 1.4833,
+      "step": 16345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969623937076081e-05,
+      "loss": 1.1917,
+      "step": 16350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9696054503098482e-05,
+      "loss": 1.1783,
+      "step": 16355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9695869580066473e-05,
+      "loss": 1.1407,
+      "step": 16360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969568460166583e-05,
+      "loss": 1.3314,
+      "step": 16365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9695499567897612e-05,
+      "loss": 1.5616,
+      "step": 16370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9695314478762875e-05,
+      "loss": 1.1892,
+      "step": 16375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9695129334262684e-05,
+      "loss": 1.3462,
+      "step": 16380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9694944134398085e-05,
+      "loss": 1.4403,
+      "step": 16385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9694758879170145e-05,
+      "loss": 1.3521,
+      "step": 16390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969457356857991e-05,
+      "loss": 1.2976,
+      "step": 16395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969438820262845e-05,
+      "loss": 1.141,
+      "step": 16400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9694202781316822e-05,
+      "loss": 1.2409,
+      "step": 16405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969401730464608e-05,
+      "loss": 1.2857,
+      "step": 16410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9693831772617284e-05,
+      "loss": 1.3107,
+      "step": 16415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9693646185231496e-05,
+      "loss": 1.0182,
+      "step": 16420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9693460542489774e-05,
+      "loss": 1.1952,
+      "step": 16425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9693274844393178e-05,
+      "loss": 1.0312,
+      "step": 16430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969308909094277e-05,
+      "loss": 1.059,
+      "step": 16435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969290328213961e-05,
+      "loss": 1.1075,
+      "step": 16440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9692717417984757e-05,
+      "loss": 1.0944,
+      "step": 16445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9692531498479275e-05,
+      "loss": 1.7006,
+      "step": 16450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969234552362423e-05,
+      "loss": 1.4882,
+      "step": 16455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9692159493420672e-05,
+      "loss": 1.6082,
+      "step": 16460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9691973407869674e-05,
+      "loss": 1.2546,
+      "step": 16465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9691787266972294e-05,
+      "loss": 1.3858,
+      "step": 16470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9691601070729595e-05,
+      "loss": 1.0803,
+      "step": 16475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9691414819142643e-05,
+      "loss": 1.3917,
+      "step": 16480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.96912285122125e-05,
+      "loss": 1.519,
+      "step": 16485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969104214994023e-05,
+      "loss": 1.1724,
+      "step": 16490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9690855732326893e-05,
+      "loss": 1.2432,
+      "step": 16495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969066925937356e-05,
+      "loss": 1.2441,
+      "step": 16500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9690482731081294e-05,
+      "loss": 0.9288,
+      "step": 16505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969029614745116e-05,
+      "loss": 1.3546,
+      "step": 16510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9690109508484223e-05,
+      "loss": 1.1733,
+      "step": 16515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9689922814181546e-05,
+      "loss": 1.2239,
+      "step": 16520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9689736064544203e-05,
+      "loss": 1.2904,
+      "step": 16525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9689549259573253e-05,
+      "loss": 1.3591,
+      "step": 16530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9689362399269763e-05,
+      "loss": 1.3596,
+      "step": 16535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9689175483634802e-05,
+      "loss": 1.2091,
+      "step": 16540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9688988512669444e-05,
+      "loss": 1.1228,
+      "step": 16545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9688801486374744e-05,
+      "loss": 1.3108,
+      "step": 16550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9688614404751778e-05,
+      "loss": 1.3683,
+      "step": 16555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9688427267801616e-05,
+      "loss": 1.3179,
+      "step": 16560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.968824007552532e-05,
+      "loss": 1.1016,
+      "step": 16565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.968805282792396e-05,
+      "loss": 1.2855,
+      "step": 16570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9687865524998613e-05,
+      "loss": 1.3004,
+      "step": 16575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.968767816675034e-05,
+      "loss": 1.085,
+      "step": 16580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9687490753180213e-05,
+      "loss": 1.0491,
+      "step": 16585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9687303284289307e-05,
+      "loss": 1.3447,
+      "step": 16590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9687115760078687e-05,
+      "loss": 1.2135,
+      "step": 16595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9686928180549426e-05,
+      "loss": 1.0776,
+      "step": 16600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9686740545702594e-05,
+      "loss": 1.2666,
+      "step": 16605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9686552855539266e-05,
+      "loss": 1.3178,
+      "step": 16610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9686365110060506e-05,
+      "loss": 1.3354,
+      "step": 16615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9686177309267394e-05,
+      "loss": 1.626,
+      "step": 16620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9685989453161e-05,
+      "loss": 1.4444,
+      "step": 16625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9685801541742398e-05,
+      "loss": 1.3238,
+      "step": 16630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9685613575012662e-05,
+      "loss": 1.1915,
+      "step": 16635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.968542555297286e-05,
+      "loss": 1.3811,
+      "step": 16640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9685237475624067e-05,
+      "loss": 1.2469,
+      "step": 16645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9685049342967363e-05,
+      "loss": 1.3938,
+      "step": 16650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9684861155003812e-05,
+      "loss": 1.2657,
+      "step": 16655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.96846729117345e-05,
+      "loss": 1.1091,
+      "step": 16660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9684484613160493e-05,
+      "loss": 1.4787,
+      "step": 16665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9684296259282877e-05,
+      "loss": 1.6051,
+      "step": 16670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9684107850102713e-05,
+      "loss": 1.2142,
+      "step": 16675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.968391938562109e-05,
+      "loss": 1.1524,
+      "step": 16680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9683730865839076e-05,
+      "loss": 1.1894,
+      "step": 16685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9683542290757748e-05,
+      "loss": 1.0015,
+      "step": 16690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9683353660378192e-05,
+      "loss": 1.0729,
+      "step": 16695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9683164974701473e-05,
+      "loss": 1.4305,
+      "step": 16700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9682976233728674e-05,
+      "loss": 1.4228,
+      "step": 16705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9682787437460874e-05,
+      "loss": 1.2981,
+      "step": 16710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.968259858589915e-05,
+      "loss": 1.2222,
+      "step": 16715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.968240967904458e-05,
+      "loss": 1.7814,
+      "step": 16720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.968222071689824e-05,
+      "loss": 1.1776,
+      "step": 16725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9682031699461218e-05,
+      "loss": 1.4018,
+      "step": 16730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9681842626734582e-05,
+      "loss": 1.2043,
+      "step": 16735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9681653498719423e-05,
+      "loss": 1.3252,
+      "step": 16740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.968146431541681e-05,
+      "loss": 1.5725,
+      "step": 16745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9681275076827832e-05,
+      "loss": 1.5229,
+      "step": 16750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9681085782953562e-05,
+      "loss": 1.1767,
+      "step": 16755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.968089643379509e-05,
+      "loss": 1.5265,
+      "step": 16760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.968070702935349e-05,
+      "loss": 1.1993,
+      "step": 16765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9680517569629846e-05,
+      "loss": 1.1337,
+      "step": 16770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9680328054625238e-05,
+      "loss": 1.1855,
+      "step": 16775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9680138484340753e-05,
+      "loss": 1.0837,
+      "step": 16780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967994885877747e-05,
+      "loss": 1.3729,
+      "step": 16785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9679759177936474e-05,
+      "loss": 1.2219,
+      "step": 16790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9679569441818845e-05,
+      "loss": 1.0532,
+      "step": 16795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967937965042567e-05,
+      "loss": 1.175,
+      "step": 16800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967918980375803e-05,
+      "loss": 1.216,
+      "step": 16805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9678999901817012e-05,
+      "loss": 1.5103,
+      "step": 16810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9678809944603696e-05,
+      "loss": 1.262,
+      "step": 16815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967861993211917e-05,
+      "loss": 1.1947,
+      "step": 16820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967842986436452e-05,
+      "loss": 1.1964,
+      "step": 16825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967823974134083e-05,
+      "loss": 1.1122,
+      "step": 16830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967804956304918e-05,
+      "loss": 1.0566,
+      "step": 16835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9677859329490668e-05,
+      "loss": 1.1955,
+      "step": 16840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9677669040666375e-05,
+      "loss": 1.0746,
+      "step": 16845
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967747869657738e-05,
+      "loss": 1.2542,
+      "step": 16850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9677288297224782e-05,
+      "loss": 1.1606,
+      "step": 16855
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9677097842609663e-05,
+      "loss": 1.4997,
+      "step": 16860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967690733273311e-05,
+      "loss": 1.2796,
+      "step": 16865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967671676759621e-05,
+      "loss": 1.1711,
+      "step": 16870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967652614720005e-05,
+      "loss": 1.1571,
+      "step": 16875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9676335471545728e-05,
+      "loss": 1.1058,
+      "step": 16880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967614474063432e-05,
+      "loss": 1.1656,
+      "step": 16885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967595395446692e-05,
+      "loss": 1.3414,
+      "step": 16890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9675763113044625e-05,
+      "loss": 1.5234,
+      "step": 16895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9675572216368514e-05,
+      "loss": 1.5064,
+      "step": 16900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9675381264439684e-05,
+      "loss": 1.3924,
+      "step": 16905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967519025725922e-05,
+      "loss": 1.3017,
+      "step": 16910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9674999194828218e-05,
+      "loss": 1.3676,
+      "step": 16915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9674808077147766e-05,
+      "loss": 0.9842,
+      "step": 16920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967461690421896e-05,
+      "loss": 1.7946,
+      "step": 16925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967442567604288e-05,
+      "loss": 1.4373,
+      "step": 16930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9674234392620634e-05,
+      "loss": 1.1253,
+      "step": 16935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9674043053953304e-05,
+      "loss": 1.5998,
+      "step": 16940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9673851660041985e-05,
+      "loss": 1.1292,
+      "step": 16945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967366021088777e-05,
+      "loss": 1.1815,
+      "step": 16950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967346870649175e-05,
+      "loss": 1.1721,
+      "step": 16955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9673277146855022e-05,
+      "loss": 1.3813,
+      "step": 16960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967308553197868e-05,
+      "loss": 1.3188,
+      "step": 16965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9672893861863813e-05,
+      "loss": 1.3172,
+      "step": 16970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9672702136511524e-05,
+      "loss": 1.0284,
+      "step": 16975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.96725103559229e-05,
+      "loss": 1.2891,
+      "step": 16980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967231852009904e-05,
+      "loss": 1.2344,
+      "step": 16985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967212662904104e-05,
+      "loss": 1.6287,
+      "step": 16990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967193468275e-05,
+      "loss": 1.4448,
+      "step": 16995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9671742681227e-05,
+      "loss": 1.0314,
+      "step": 17000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9671550624473156e-05,
+      "loss": 1.4117,
+      "step": 17005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9671358512489554e-05,
+      "loss": 1.4404,
+      "step": 17010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967116634527729e-05,
+      "loss": 1.3894,
+      "step": 17015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9670974122837468e-05,
+      "loss": 1.2615,
+      "step": 17020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9670781845171182e-05,
+      "loss": 1.2674,
+      "step": 17025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9670589512279527e-05,
+      "loss": 1.541,
+      "step": 17030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9670397124163608e-05,
+      "loss": 1.075,
+      "step": 17035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9670204680824518e-05,
+      "loss": 1.0121,
+      "step": 17040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967001218226336e-05,
+      "loss": 1.1302,
+      "step": 17045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966981962848123e-05,
+      "loss": 1.2366,
+      "step": 17050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9669627019479226e-05,
+      "loss": 1.2385,
+      "step": 17055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9669434355258454e-05,
+      "loss": 0.9909,
+      "step": 17060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9669241635820013e-05,
+      "loss": 1.3118,
+      "step": 17065
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9669048861165e-05,
+      "loss": 1.3822,
+      "step": 17070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966885603129452e-05,
+      "loss": 1.1487,
+      "step": 17075
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9668663146209666e-05,
+      "loss": 1.2319,
+      "step": 17080
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966847020591155e-05,
+      "loss": 1.4379,
+      "step": 17085
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9668277210401266e-05,
+      "loss": 1.2362,
+      "step": 17090
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9668084159679922e-05,
+      "loss": 1.245,
+      "step": 17095
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9667891053748612e-05,
+      "loss": 1.3071,
+      "step": 17100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966769789260845e-05,
+      "loss": 1.3697,
+      "step": 17105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966750467626053e-05,
+      "loss": 1.0055,
+      "step": 17110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966731140470596e-05,
+      "loss": 1.128,
+      "step": 17115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966711807794584e-05,
+      "loss": 1.5158,
+      "step": 17120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966692469598128e-05,
+      "loss": 1.3246,
+      "step": 17125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9666731258813378e-05,
+      "loss": 0.9594,
+      "step": 17130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966653776644324e-05,
+      "loss": 1.3351,
+      "step": 17135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9666344218871977e-05,
+      "loss": 1.3636,
+      "step": 17140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9666150616100688e-05,
+      "loss": 1.1742,
+      "step": 17145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9665956958130477e-05,
+      "loss": 0.8614,
+      "step": 17150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9665763244962456e-05,
+      "loss": 1.155,
+      "step": 17155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9665569476597727e-05,
+      "loss": 1.5406,
+      "step": 17160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9665375653037396e-05,
+      "loss": 1.4073,
+      "step": 17165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9665181774282572e-05,
+      "loss": 1.1953,
+      "step": 17170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9664987840334363e-05,
+      "loss": 1.0751,
+      "step": 17175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9664793851193874e-05,
+      "loss": 1.3131,
+      "step": 17180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9664599806862215e-05,
+      "loss": 1.0817,
+      "step": 17185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9664405707340492e-05,
+      "loss": 1.2305,
+      "step": 17190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9664211552629815e-05,
+      "loss": 1.1968,
+      "step": 17195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9664017342731292e-05,
+      "loss": 1.3648,
+      "step": 17200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9663823077646032e-05,
+      "loss": 1.6717,
+      "step": 17205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9663628757375143e-05,
+      "loss": 1.4212,
+      "step": 17210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9663434381919738e-05,
+      "loss": 1.439,
+      "step": 17215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9663239951280926e-05,
+      "loss": 1.4464,
+      "step": 17220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9663045465459815e-05,
+      "loss": 1.2315,
+      "step": 17225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9662850924457516e-05,
+      "loss": 1.3383,
+      "step": 17230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9662656328275144e-05,
+      "loss": 1.3412,
+      "step": 17235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9662461676913807e-05,
+      "loss": 1.1342,
+      "step": 17240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9662266970374616e-05,
+      "loss": 1.568,
+      "step": 17245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9662072208658682e-05,
+      "loss": 1.2057,
+      "step": 17250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966187739176712e-05,
+      "loss": 1.1043,
+      "step": 17255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966168251970104e-05,
+      "loss": 1.1802,
+      "step": 17260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966148759246156e-05,
+      "loss": 1.1935,
+      "step": 17265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9661292610049787e-05,
+      "loss": 1.3468,
+      "step": 17270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9661097572466837e-05,
+      "loss": 1.1299,
+      "step": 17275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9660902479713825e-05,
+      "loss": 1.2347,
+      "step": 17280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966070733179186e-05,
+      "loss": 1.1689,
+      "step": 17285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9660512128702068e-05,
+      "loss": 1.4279,
+      "step": 17290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9660316870445547e-05,
+      "loss": 1.2329,
+      "step": 17295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9660121557023424e-05,
+      "loss": 1.0158,
+      "step": 17300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9659926188436813e-05,
+      "loss": 1.3429,
+      "step": 17305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9659730764686826e-05,
+      "loss": 1.4485,
+      "step": 17310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9659535285774584e-05,
+      "loss": 1.1227,
+      "step": 17315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9659339751701195e-05,
+      "loss": 1.5283,
+      "step": 17320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9659144162467784e-05,
+      "loss": 1.272,
+      "step": 17325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.965894851807546e-05,
+      "loss": 1.3175,
+      "step": 17330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.965875281852535e-05,
+      "loss": 1.2138,
+      "step": 17335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9658557063818567e-05,
+      "loss": 1.188,
+      "step": 17340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9658361253956224e-05,
+      "loss": 1.3074,
+      "step": 17345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.965816538893945e-05,
+      "loss": 1.2518,
+      "step": 17350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.965796946876935e-05,
+      "loss": 1.1388,
+      "step": 17355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9657773493447053e-05,
+      "loss": 1.9191,
+      "step": 17360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9657577462973674e-05,
+      "loss": 1.2748,
+      "step": 17365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9657381377350333e-05,
+      "loss": 1.206,
+      "step": 17370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.965718523657815e-05,
+      "loss": 1.4354,
+      "step": 17375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9656989040658247e-05,
+      "loss": 1.319,
+      "step": 17380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9656792789591743e-05,
+      "loss": 1.5903,
+      "step": 17385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9656596483379755e-05,
+      "loss": 1.2405,
+      "step": 17390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9656400122023408e-05,
+      "loss": 1.2684,
+      "step": 17395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9656203705523827e-05,
+      "loss": 1.3053,
+      "step": 17400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9656007233882123e-05,
+      "loss": 1.3168,
+      "step": 17405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9655810707099428e-05,
+      "loss": 1.0663,
+      "step": 17410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.965561412517686e-05,
+      "loss": 1.3552,
+      "step": 17415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.965541748811554e-05,
+      "loss": 1.2498,
+      "step": 17420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9655220795916595e-05,
+      "loss": 1.3549,
+      "step": 17425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9655024048581147e-05,
+      "loss": 1.4129,
+      "step": 17430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9654827246110316e-05,
+      "loss": 1.4351,
+      "step": 17435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9654630388505233e-05,
+      "loss": 1.1856,
+      "step": 17440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9654433475767012e-05,
+      "loss": 1.5075,
+      "step": 17445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9654236507896785e-05,
+      "loss": 1.4001,
+      "step": 17450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9654039484895678e-05,
+      "loss": 1.5444,
+      "step": 17455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9653842406764808e-05,
+      "loss": 1.0461,
+      "step": 17460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.965364527350531e-05,
+      "loss": 1.1672,
+      "step": 17465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9653448085118306e-05,
+      "loss": 1.3897,
+      "step": 17470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9653250841604916e-05,
+      "loss": 1.0397,
+      "step": 17475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9653053542966278e-05,
+      "loss": 1.2832,
+      "step": 17480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9652856189203507e-05,
+      "loss": 1.2416,
+      "step": 17485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9652658780317736e-05,
+      "loss": 1.404,
+      "step": 17490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9652461316310094e-05,
+      "loss": 1.0443,
+      "step": 17495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9652263797181704e-05,
+      "loss": 0.9282,
+      "step": 17500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.96520662229337e-05,
+      "loss": 1.3292,
+      "step": 17505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9651868593567204e-05,
+      "loss": 1.623,
+      "step": 17510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9651670909083343e-05,
+      "loss": 1.4182,
+      "step": 17515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9651473169483255e-05,
+      "loss": 1.5154,
+      "step": 17520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9651275374768065e-05,
+      "loss": 1.4564,
+      "step": 17525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.96510775249389e-05,
+      "loss": 1.2034,
+      "step": 17530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.965087961999689e-05,
+      "loss": 1.5122,
+      "step": 17535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9650681659943165e-05,
+      "loss": 1.3251,
+      "step": 17540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9650483644778865e-05,
+      "loss": 1.1221,
+      "step": 17545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9650285574505107e-05,
+      "loss": 1.2171,
+      "step": 17550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9650087449123026e-05,
+      "loss": 1.1908,
+      "step": 17555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9649889268633758e-05,
+      "loss": 1.4131,
+      "step": 17560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9649691033038433e-05,
+      "loss": 1.1954,
+      "step": 17565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9649492742338184e-05,
+      "loss": 1.4825,
+      "step": 17570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9649294396534136e-05,
+      "loss": 1.1918,
+      "step": 17575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9649095995627433e-05,
+      "loss": 1.307,
+      "step": 17580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9648897539619198e-05,
+      "loss": 1.2402,
+      "step": 17585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.964869902851057e-05,
+      "loss": 1.0643,
+      "step": 17590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.964850046230268e-05,
+      "loss": 1.3911,
+      "step": 17595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.964830184099666e-05,
+      "loss": 1.0078,
+      "step": 17600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.964810316459365e-05,
+      "loss": 1.1645,
+      "step": 17605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9647904433094783e-05,
+      "loss": 1.7442,
+      "step": 17610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9647705646501196e-05,
+      "loss": 1.6043,
+      "step": 17615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.964750680481401e-05,
+      "loss": 1.4775,
+      "step": 17620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.964730790803438e-05,
+      "loss": 1.1698,
+      "step": 17625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.964710895616343e-05,
+      "loss": 1.0596,
+      "step": 17630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.96469099492023e-05,
+      "loss": 1.2753,
+      "step": 17635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9646710887152124e-05,
+      "loss": 1.3763,
+      "step": 17640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9646511770014043e-05,
+      "loss": 1.4781,
+      "step": 17645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9646312597789187e-05,
+      "loss": 1.5436,
+      "step": 17650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9646113370478703e-05,
+      "loss": 1.4076,
+      "step": 17655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.964591408808372e-05,
+      "loss": 1.4253,
+      "step": 17660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.964571475060538e-05,
+      "loss": 1.372,
+      "step": 17665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9645515358044823e-05,
+      "loss": 1.3324,
+      "step": 17670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9645315910403183e-05,
+      "loss": 1.3106,
+      "step": 17675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9645116407681602e-05,
+      "loss": 1.1925,
+      "step": 17680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964491684988122e-05,
+      "loss": 1.3196,
+      "step": 17685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964471723700317e-05,
+      "loss": 1.1903,
+      "step": 17690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9644517569048604e-05,
+      "loss": 1.2175,
+      "step": 17695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9644317846018653e-05,
+      "loss": 1.4262,
+      "step": 17700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964411806791446e-05,
+      "loss": 1.7274,
+      "step": 17705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9643918234737166e-05,
+      "loss": 1.1904,
+      "step": 17710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964371834648791e-05,
+      "loss": 1.0686,
+      "step": 17715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9643518403167837e-05,
+      "loss": 1.0728,
+      "step": 17720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9643318404778087e-05,
+      "loss": 0.9805,
+      "step": 17725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96431183513198e-05,
+      "loss": 1.4235,
+      "step": 17730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964291824279412e-05,
+      "loss": 1.1079,
+      "step": 17735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9642718079202195e-05,
+      "loss": 1.424,
+      "step": 17740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9642517860545163e-05,
+      "loss": 1.339,
+      "step": 17745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9642317586824165e-05,
+      "loss": 1.2586,
+      "step": 17750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964211725804035e-05,
+      "loss": 1.3066,
+      "step": 17755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9641916874194857e-05,
+      "loss": 1.4547,
+      "step": 17760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9641716435288835e-05,
+      "loss": 1.0772,
+      "step": 17765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9641515941323424e-05,
+      "loss": 1.0645,
+      "step": 17770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964131539229977e-05,
+      "loss": 1.4352,
+      "step": 17775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9641114788219024e-05,
+      "loss": 1.1453,
+      "step": 17780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9640914129082324e-05,
+      "loss": 1.1364,
+      "step": 17785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964071341489082e-05,
+      "loss": 1.0556,
+      "step": 17790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9640512645645655e-05,
+      "loss": 1.3024,
+      "step": 17795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964031182134798e-05,
+      "loss": 1.2679,
+      "step": 17800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964011094199894e-05,
+      "loss": 1.2705,
+      "step": 17805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963991000759968e-05,
+      "loss": 1.2929,
+      "step": 17810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963970901815135e-05,
+      "loss": 1.1169,
+      "step": 17815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9639507973655096e-05,
+      "loss": 1.2149,
+      "step": 17820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963930687411207e-05,
+      "loss": 1.3664,
+      "step": 17825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963910571952341e-05,
+      "loss": 1.1288,
+      "step": 17830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9638904509890277e-05,
+      "loss": 1.2321,
+      "step": 17835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9638703245213816e-05,
+      "loss": 1.3885,
+      "step": 17840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9638501925495172e-05,
+      "loss": 1.1868,
+      "step": 17845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96383005507355e-05,
+      "loss": 1.2862,
+      "step": 17850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9638099120935944e-05,
+      "loss": 1.1836,
+      "step": 17855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9637897636097666e-05,
+      "loss": 1.3478,
+      "step": 17860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96376960962218e-05,
+      "loss": 1.1504,
+      "step": 17865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963749450130951e-05,
+      "loss": 1.4211,
+      "step": 17870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9637292851361944e-05,
+      "loss": 1.3849,
+      "step": 17875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963709114638025e-05,
+      "loss": 1.4489,
+      "step": 17880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9636889386365583e-05,
+      "loss": 0.925,
+      "step": 17885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9636687571319094e-05,
+      "loss": 1.3645,
+      "step": 17890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9636485701241938e-05,
+      "loss": 1.4902,
+      "step": 17895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9636283776135263e-05,
+      "loss": 1.2086,
+      "step": 17900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9636081796000227e-05,
+      "loss": 1.2743,
+      "step": 17905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963587976083798e-05,
+      "loss": 1.5472,
+      "step": 17910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9635677670649676e-05,
+      "loss": 1.258,
+      "step": 17915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9635475525436472e-05,
+      "loss": 1.1936,
+      "step": 17920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9635273325199517e-05,
+      "loss": 1.2595,
+      "step": 17925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9635071069939973e-05,
+      "loss": 1.0751,
+      "step": 17930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9634868759658988e-05,
+      "loss": 1.1409,
+      "step": 17935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963466639435772e-05,
+      "loss": 1.0885,
+      "step": 17940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9634463974037327e-05,
+      "loss": 1.052,
+      "step": 17945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9634261498698965e-05,
+      "loss": 1.1254,
+      "step": 17950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9634058968343784e-05,
+      "loss": 1.1254,
+      "step": 17955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9633856382972945e-05,
+      "loss": 1.1378,
+      "step": 17960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96336537425876e-05,
+      "loss": 1.0938,
+      "step": 17965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9633451047188917e-05,
+      "loss": 1.2592,
+      "step": 17970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9633248296778046e-05,
+      "loss": 1.2397,
+      "step": 17975
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963304549135615e-05,
+      "loss": 1.0893,
+      "step": 17980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9632842630924377e-05,
+      "loss": 0.9628,
+      "step": 17985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963263971548389e-05,
+      "loss": 1.8998,
+      "step": 17990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9632436745035856e-05,
+      "loss": 1.111,
+      "step": 17995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9632233719581423e-05,
+      "loss": 1.2566,
+      "step": 18000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9632030639121754e-05,
+      "loss": 1.423,
+      "step": 18005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9631827503658013e-05,
+      "loss": 1.3239,
+      "step": 18010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9631624313191353e-05,
+      "loss": 1.1302,
+      "step": 18015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963142106772294e-05,
+      "loss": 1.679,
+      "step": 18020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9631217767253932e-05,
+      "loss": 1.5895,
+      "step": 18025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9631014411785488e-05,
+      "loss": 1.1342,
+      "step": 18030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9630811001318775e-05,
+      "loss": 1.3347,
+      "step": 18035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963060753585495e-05,
+      "loss": 1.1565,
+      "step": 18040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9630404015395178e-05,
+      "loss": 1.2945,
+      "step": 18045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9630200439940616e-05,
+      "loss": 1.4093,
+      "step": 18050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9629996809492433e-05,
+      "loss": 1.5865,
+      "step": 18055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9629793124051787e-05,
+      "loss": 1.1564,
+      "step": 18060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9629589383619842e-05,
+      "loss": 1.2062,
+      "step": 18065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9629385588197762e-05,
+      "loss": 1.3853,
+      "step": 18070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9629181737786713e-05,
+      "loss": 1.3138,
+      "step": 18075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9628977832387858e-05,
+      "loss": 1.2715,
+      "step": 18080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9628773872002357e-05,
+      "loss": 1.224,
+      "step": 18085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9628569856631385e-05,
+      "loss": 1.4536,
+      "step": 18090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9628365786276095e-05,
+      "loss": 1.4676,
+      "step": 18095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9628161660937657e-05,
+      "loss": 1.4689,
+      "step": 18100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9627957480617236e-05,
+      "loss": 1.0297,
+      "step": 18105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9627753245316005e-05,
+      "loss": 1.1346,
+      "step": 18110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9627548955035123e-05,
+      "loss": 1.062,
+      "step": 18115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9627344609775757e-05,
+      "loss": 1.6199,
+      "step": 18120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962714020953907e-05,
+      "loss": 0.9779,
+      "step": 18125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9626935754326244e-05,
+      "loss": 1.0783,
+      "step": 18130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9626731244138434e-05,
+      "loss": 1.3851,
+      "step": 18135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9626526678976806e-05,
+      "loss": 1.2243,
+      "step": 18140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962632205884254e-05,
+      "loss": 1.4797,
+      "step": 18145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9626117383736793e-05,
+      "loss": 0.9935,
+      "step": 18150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962591265366074e-05,
+      "loss": 1.5269,
+      "step": 18155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9625707868615547e-05,
+      "loss": 1.158,
+      "step": 18160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9625503028602387e-05,
+      "loss": 1.0771,
+      "step": 18165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9625298133622427e-05,
+      "loss": 1.1337,
+      "step": 18170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9625093183676835e-05,
+      "loss": 1.3726,
+      "step": 18175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9624888178766784e-05,
+      "loss": 1.1355,
+      "step": 18180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962468311889345e-05,
+      "loss": 1.2375,
+      "step": 18185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9624478004057997e-05,
+      "loss": 0.9911,
+      "step": 18190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9624272834261596e-05,
+      "loss": 1.396,
+      "step": 18195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962406760950542e-05,
+      "loss": 1.383,
+      "step": 18200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962386232979064e-05,
+      "loss": 1.2365,
+      "step": 18205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9623656995118436e-05,
+      "loss": 1.4467,
+      "step": 18210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962345160548997e-05,
+      "loss": 0.9874,
+      "step": 18215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9623246160906422e-05,
+      "loss": 0.8187,
+      "step": 18220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962304066136896e-05,
+      "loss": 1.241,
+      "step": 18225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9622835106878763e-05,
+      "loss": 1.1989,
+      "step": 18230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9622629497437e-05,
+      "loss": 1.3857,
+      "step": 18235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9622423833044845e-05,
+      "loss": 1.3318,
+      "step": 18240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9622218113703476e-05,
+      "loss": 1.1693,
+      "step": 18245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9622012339414066e-05,
+      "loss": 1.237,
+      "step": 18250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9621806510177796e-05,
+      "loss": 1.0272,
+      "step": 18255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962160062599583e-05,
+      "loss": 1.4909,
+      "step": 18260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962139468686935e-05,
+      "loss": 0.9996,
+      "step": 18265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9621188692799528e-05,
+      "loss": 1.6408,
+      "step": 18270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962098264378755e-05,
+      "loss": 1.3883,
+      "step": 18275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9620776539834587e-05,
+      "loss": 1.1351,
+      "step": 18280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962057038094181e-05,
+      "loss": 1.1478,
+      "step": 18285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9620364167110403e-05,
+      "loss": 1.2822,
+      "step": 18290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9620157898341542e-05,
+      "loss": 0.9322,
+      "step": 18295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961995157463641e-05,
+      "loss": 1.2165,
+      "step": 18300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9619745195996174e-05,
+      "loss": 1.1196,
+      "step": 18305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9619538762422025e-05,
+      "loss": 1.3522,
+      "step": 18310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9619332273915132e-05,
+      "loss": 1.0579,
+      "step": 18315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9619125730476675e-05,
+      "loss": 1.1229,
+      "step": 18320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961891913210784e-05,
+      "loss": 1.4344,
+      "step": 18325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9618712478809804e-05,
+      "loss": 1.0584,
+      "step": 18330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9618505770583747e-05,
+      "loss": 1.2467,
+      "step": 18335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9618299007430846e-05,
+      "loss": 1.1075,
+      "step": 18340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9618092189352283e-05,
+      "loss": 1.1022,
+      "step": 18345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9617885316349245e-05,
+      "loss": 1.1655,
+      "step": 18350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9617678388422906e-05,
+      "loss": 1.3677,
+      "step": 18355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9617471405574453e-05,
+      "loss": 1.0657,
+      "step": 18360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961726436780506e-05,
+      "loss": 1.1338,
+      "step": 18365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9617057275115923e-05,
+      "loss": 1.1479,
+      "step": 18370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961685012750821e-05,
+      "loss": 1.1378,
+      "step": 18375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9616642924983112e-05,
+      "loss": 1.2561,
+      "step": 18380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961643566754181e-05,
+      "loss": 1.2724,
+      "step": 18385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9616228355185493e-05,
+      "loss": 1.0002,
+      "step": 18390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9616020987915334e-05,
+      "loss": 1.3244,
+      "step": 18395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9615813565732528e-05,
+      "loss": 1.2703,
+      "step": 18400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961560608863825e-05,
+      "loss": 1.2657,
+      "step": 18405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961539855663369e-05,
+      "loss": 1.143,
+      "step": 18410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9615190969720036e-05,
+      "loss": 1.0203,
+      "step": 18415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9614983327898466e-05,
+      "loss": 1.069,
+      "step": 18420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9614775631170174e-05,
+      "loss": 1.5295,
+      "step": 18425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9614567879536336e-05,
+      "loss": 1.2501,
+      "step": 18430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961436007299815e-05,
+      "loss": 1.2928,
+      "step": 18435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9614152211556793e-05,
+      "loss": 1.1546,
+      "step": 18440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9613944295213458e-05,
+      "loss": 1.4547,
+      "step": 18445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961373632396933e-05,
+      "loss": 1.3776,
+      "step": 18450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9613528297825598e-05,
+      "loss": 1.0526,
+      "step": 18455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9613320216783448e-05,
+      "loss": 1.6828,
+      "step": 18460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9613112080844067e-05,
+      "loss": 0.9902,
+      "step": 18465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9612903890008646e-05,
+      "loss": 1.1039,
+      "step": 18470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9612695644278375e-05,
+      "loss": 1.3292,
+      "step": 18475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961248734365444e-05,
+      "loss": 1.1805,
+      "step": 18480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9612278988138033e-05,
+      "loss": 1.2511,
+      "step": 18485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9612070577730344e-05,
+      "loss": 1.2954,
+      "step": 18490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9611862112432562e-05,
+      "loss": 1.0487,
+      "step": 18495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9611653592245876e-05,
+      "loss": 1.1959,
+      "step": 18500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961144501717148e-05,
+      "loss": 1.3482,
+      "step": 18505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9611236387210563e-05,
+      "loss": 1.3083,
+      "step": 18510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9611027702364317e-05,
+      "loss": 1.6901,
+      "step": 18515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.961081896263393e-05,
+      "loss": 1.2769,
+      "step": 18520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9610610168020602e-05,
+      "loss": 1.4512,
+      "step": 18525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9610401318525517e-05,
+      "loss": 1.2445,
+      "step": 18530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9610192414149877e-05,
+      "loss": 1.1728,
+      "step": 18535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9609983454894864e-05,
+      "loss": 1.2017,
+      "step": 18540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9609774440761677e-05,
+      "loss": 1.4032,
+      "step": 18545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9609565371751512e-05,
+      "loss": 1.2565,
+      "step": 18550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960935624786556e-05,
+      "loss": 1.2135,
+      "step": 18555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9609147069105017e-05,
+      "loss": 1.0282,
+      "step": 18560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9608937835471072e-05,
+      "loss": 1.2969,
+      "step": 18565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9608728546964926e-05,
+      "loss": 1.0849,
+      "step": 18570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9608519203587773e-05,
+      "loss": 1.3869,
+      "step": 18575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9608309805340804e-05,
+      "loss": 1.1976,
+      "step": 18580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960810035222522e-05,
+      "loss": 1.5356,
+      "step": 18585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9607890844242216e-05,
+      "loss": 1.1835,
+      "step": 18590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9607681281392988e-05,
+      "loss": 1.3588,
+      "step": 18595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960747166367873e-05,
+      "loss": 1.1799,
+      "step": 18600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9607261991100642e-05,
+      "loss": 1.1525,
+      "step": 18605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9607052263659923e-05,
+      "loss": 1.221,
+      "step": 18610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9606842481357767e-05,
+      "loss": 1.0552,
+      "step": 18615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960663264419537e-05,
+      "loss": 1.2144,
+      "step": 18620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960642275217394e-05,
+      "loss": 1.2757,
+      "step": 18625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9606212805294665e-05,
+      "loss": 1.3319,
+      "step": 18630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960600280355875e-05,
+      "loss": 1.1363,
+      "step": 18635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9605792746967388e-05,
+      "loss": 1.1097,
+      "step": 18640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9605582635521788e-05,
+      "loss": 1.1352,
+      "step": 18645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9605372469223144e-05,
+      "loss": 1.2213,
+      "step": 18650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9605162248072656e-05,
+      "loss": 1.285,
+      "step": 18655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9604951972071526e-05,
+      "loss": 1.3642,
+      "step": 18660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9604741641220954e-05,
+      "loss": 1.0996,
+      "step": 18665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9604531255522142e-05,
+      "loss": 1.3749,
+      "step": 18670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9604320814976287e-05,
+      "loss": 1.3532,
+      "step": 18675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96041103195846e-05,
+      "loss": 1.3392,
+      "step": 18680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9603899769348274e-05,
+      "loss": 1.4013,
+      "step": 18685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9603689164268515e-05,
+      "loss": 1.2689,
+      "step": 18690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9603478504346527e-05,
+      "loss": 1.6012,
+      "step": 18695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9603267789583512e-05,
+      "loss": 1.3133,
+      "step": 18700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9603057019980668e-05,
+      "loss": 0.8357,
+      "step": 18705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960284619553921e-05,
+      "loss": 1.2259,
+      "step": 18710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960263531626033e-05,
+      "loss": 1.709,
+      "step": 18715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960242438214524e-05,
+      "loss": 1.0972,
+      "step": 18720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9602213393195143e-05,
+      "loss": 1.2269,
+      "step": 18725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960200234941124e-05,
+      "loss": 1.3106,
+      "step": 18730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960179125079474e-05,
+      "loss": 1.2125,
+      "step": 18735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960158009734685e-05,
+      "loss": 1.3378,
+      "step": 18740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960136888906877e-05,
+      "loss": 1.2094,
+      "step": 18745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9601157625961712e-05,
+      "loss": 1.4102,
+      "step": 18750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9600946308026882e-05,
+      "loss": 1.4133,
+      "step": 18755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960073493526548e-05,
+      "loss": 1.4131,
+      "step": 18760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960052350767872e-05,
+      "loss": 1.188,
+      "step": 18765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9600312025267808e-05,
+      "loss": 1.3687,
+      "step": 18770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.960010048803395e-05,
+      "loss": 1.1821,
+      "step": 18775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9599888895978352e-05,
+      "loss": 1.4072,
+      "step": 18780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.959967724910223e-05,
+      "loss": 1.2363,
+      "step": 18785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9599465547406784e-05,
+      "loss": 1.3319,
+      "step": 18790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.959925379089323e-05,
+      "loss": 1.294,
+      "step": 18795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9599041979562767e-05,
+      "loss": 1.2392,
+      "step": 18800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.959883011341662e-05,
+      "loss": 1.2129,
+      "step": 18805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9598618192455984e-05,
+      "loss": 1.4107,
+      "step": 18810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9598406216682078e-05,
+      "loss": 1.326,
+      "step": 18815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.959819418609611e-05,
+      "loss": 1.1295,
+      "step": 18820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.959798210069929e-05,
+      "loss": 1.5532,
+      "step": 18825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.959776996049283e-05,
+      "loss": 1.2834,
+      "step": 18830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9597557765477944e-05,
+      "loss": 1.2943,
+      "step": 18835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9597345515655836e-05,
+      "loss": 1.1995,
+      "step": 18840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9597133211027728e-05,
+      "loss": 1.4857,
+      "step": 18845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9596920851594826e-05,
+      "loss": 1.3517,
+      "step": 18850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9596708437358344e-05,
+      "loss": 1.4873,
+      "step": 18855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9596495968319494e-05,
+      "loss": 1.1123,
+      "step": 18860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9596283444479493e-05,
+      "loss": 1.0347,
+      "step": 18865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.959607086583955e-05,
+      "loss": 1.11,
+      "step": 18870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9595858232400882e-05,
+      "loss": 1.4415,
+      "step": 18875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.95956455441647e-05,
+      "loss": 1.106,
+      "step": 18880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9595432801132225e-05,
+      "loss": 1.2626,
+      "step": 18885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9595220003304668e-05,
+      "loss": 1.5396,
+      "step": 18890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.959500715068324e-05,
+      "loss": 1.3733,
+      "step": 18895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9594794243269165e-05,
+      "loss": 1.4317,
+      "step": 18900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.959458128106365e-05,
+      "loss": 1.3913,
+      "step": 18905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9594368264067918e-05,
+      "loss": 1.1547,
+      "step": 18910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.959415519228318e-05,
+      "loss": 1.4115,
+      "step": 18915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9593942065710658e-05,
+      "loss": 1.1678,
+      "step": 18920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9593728884351567e-05,
+      "loss": 1.1599,
+      "step": 18925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.959351564820712e-05,
+      "loss": 1.3726,
+      "step": 18930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9593302357278544e-05,
+      "loss": 1.0292,
+      "step": 18935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9593089011567052e-05,
+      "loss": 1.3269,
+      "step": 18940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.959287561107386e-05,
+      "loss": 1.1088,
+      "step": 18945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9592662155800187e-05,
+      "loss": 1.3096,
+      "step": 18950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9592448645747253e-05,
+      "loss": 1.2453,
+      "step": 18955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9592235080916283e-05,
+      "loss": 1.444,
+      "step": 18960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9592021461308487e-05,
+      "loss": 1.3073,
+      "step": 18965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.959180778692509e-05,
+      "loss": 1.3493,
+      "step": 18970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9591594057767312e-05,
+      "loss": 1.2555,
+      "step": 18975
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9591380273836376e-05,
+      "loss": 1.3741,
+      "step": 18980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9591166435133497e-05,
+      "loss": 1.4761,
+      "step": 18985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.95909525416599e-05,
+      "loss": 1.3149,
+      "step": 18990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.95907385934168e-05,
+      "loss": 1.2313,
+      "step": 18995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9590524590405434e-05,
+      "loss": 1.2219,
+      "step": 19000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.959031053262701e-05,
+      "loss": 1.1568,
+      "step": 19005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.959009642008275e-05,
+      "loss": 1.2076,
+      "step": 19010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9589882252773887e-05,
+      "loss": 1.2691,
+      "step": 19015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.958966803070164e-05,
+      "loss": 1.2085,
+      "step": 19020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.958945375386723e-05,
+      "loss": 1.6011,
+      "step": 19025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9589239422271876e-05,
+      "loss": 1.2505,
+      "step": 19030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9589025035916813e-05,
+      "loss": 1.274,
+      "step": 19035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9588810594803255e-05,
+      "loss": 0.9993,
+      "step": 19040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.958859609893243e-05,
+      "loss": 1.2716,
+      "step": 19045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.958838154830557e-05,
+      "loss": 1.2698,
+      "step": 19050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.958816694292389e-05,
+      "loss": 1.175,
+      "step": 19055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.958795228278862e-05,
+      "loss": 0.998,
+      "step": 19060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9587737567900988e-05,
+      "loss": 1.2541,
+      "step": 19065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9587522798262215e-05,
+      "loss": 0.9978,
+      "step": 19070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9587307973873533e-05,
+      "loss": 1.4224,
+      "step": 19075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9587093094736165e-05,
+      "loss": 1.0256,
+      "step": 19080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9586878160851337e-05,
+      "loss": 1.2128,
+      "step": 19085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.958666317222028e-05,
+      "loss": 1.2422,
+      "step": 19090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.958644812884422e-05,
+      "loss": 1.6284,
+      "step": 19095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9586233030724388e-05,
+      "loss": 1.1698,
+      "step": 19100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9586017877862004e-05,
+      "loss": 1.5047,
+      "step": 19105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9585802670258306e-05,
+      "loss": 1.0996,
+      "step": 19110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9585587407914515e-05,
+      "loss": 1.2596,
+      "step": 19115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9585372090831867e-05,
+      "loss": 1.502,
+      "step": 19120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9585156719011588e-05,
+      "loss": 1.4109,
+      "step": 19125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9584941292454908e-05,
+      "loss": 1.2712,
+      "step": 19130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.958472581116306e-05,
+      "loss": 1.3805,
+      "step": 19135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9584510275137273e-05,
+      "loss": 1.1229,
+      "step": 19140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9584294684378774e-05,
+      "loss": 1.2844,
+      "step": 19145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.95840790388888e-05,
+      "loss": 1.1443,
+      "step": 19150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.958386333866858e-05,
+      "loss": 1.5444,
+      "step": 19155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9583647583719344e-05,
+      "loss": 0.9843,
+      "step": 19160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9583431774042327e-05,
+      "loss": 1.3002,
+      "step": 19165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.958321590963876e-05,
+      "loss": 1.1871,
+      "step": 19170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9582999990509875e-05,
+      "loss": 1.201,
+      "step": 19175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9582784016656907e-05,
+      "loss": 1.1084,
+      "step": 19180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9582567988081088e-05,
+      "loss": 1.3072,
+      "step": 19185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.958235190478365e-05,
+      "loss": 1.2975,
+      "step": 19190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.958213576676583e-05,
+      "loss": 1.2584,
+      "step": 19195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9581919574028863e-05,
+      "loss": 1.0492,
+      "step": 19200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9581703326573977e-05,
+      "loss": 1.3557,
+      "step": 19205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9581487024402416e-05,
+      "loss": 1.313,
+      "step": 19210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9581270667515406e-05,
+      "loss": 1.124,
+      "step": 19215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9581054255914192e-05,
+      "loss": 1.2677,
+      "step": 19220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9580837789600002e-05,
+      "loss": 1.3828,
+      "step": 19225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9580621268574077e-05,
+      "loss": 1.423,
+      "step": 19230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.958040469283765e-05,
+      "loss": 1.4795,
+      "step": 19235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.958018806239196e-05,
+      "loss": 0.9941,
+      "step": 19240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9579971377238242e-05,
+      "loss": 1.0747,
+      "step": 19245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9579754637377737e-05,
+      "loss": 1.1445,
+      "step": 19250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.957953784281168e-05,
+      "loss": 1.388,
+      "step": 19255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9579320993541308e-05,
+      "loss": 1.0825,
+      "step": 19260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9579104089567864e-05,
+      "loss": 1.2353,
+      "step": 19265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.957888713089258e-05,
+      "loss": 1.0579,
+      "step": 19270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.95786701175167e-05,
+      "loss": 1.3387,
+      "step": 19275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9578453049441463e-05,
+      "loss": 1.3783,
+      "step": 19280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.957823592666811e-05,
+      "loss": 1.2233,
+      "step": 19285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9578018749197867e-05,
+      "loss": 1.2383,
+      "step": 19290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9577801517031994e-05,
+      "loss": 1.467,
+      "step": 19295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9577584230171723e-05,
+      "loss": 1.0892,
+      "step": 19300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9577366888618294e-05,
+      "loss": 1.1135,
+      "step": 19305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9577149492372947e-05,
+      "loss": 1.0808,
+      "step": 19310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9576932041436926e-05,
+      "loss": 0.9261,
+      "step": 19315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9576714535811474e-05,
+      "loss": 1.2907,
+      "step": 19320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.957649697549783e-05,
+      "loss": 1.2778,
+      "step": 19325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9576279360497236e-05,
+      "loss": 1.1479,
+      "step": 19330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9576061690810938e-05,
+      "loss": 1.1566,
+      "step": 19335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9575843966440175e-05,
+      "loss": 1.4633,
+      "step": 19340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9575626187386196e-05,
+      "loss": 1.5023,
+      "step": 19345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.957540835365024e-05,
+      "loss": 1.318,
+      "step": 19350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9575190465233552e-05,
+      "loss": 1.2216,
+      "step": 19355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9574972522137375e-05,
+      "loss": 1.2276,
+      "step": 19360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9574754524362958e-05,
+      "loss": 1.1673,
+      "step": 19365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9574536471911537e-05,
+      "loss": 1.2551,
+      "step": 19370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.957431836478437e-05,
+      "loss": 1.4828,
+      "step": 19375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.957410020298269e-05,
+      "loss": 1.1739,
+      "step": 19380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.957388198650775e-05,
+      "loss": 1.3869,
+      "step": 19385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.95736637153608e-05,
+      "loss": 1.1286,
+      "step": 19390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9573445389543074e-05,
+      "loss": 1.3397,
+      "step": 19395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9573227009055827e-05,
+      "loss": 1.5811,
+      "step": 19400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9573008573900308e-05,
+      "loss": 1.0566,
+      "step": 19405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9572790084077757e-05,
+      "loss": 1.2963,
+      "step": 19410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.957257153958943e-05,
+      "loss": 1.4267,
+      "step": 19415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9572352940436566e-05,
+      "loss": 1.3885,
+      "step": 19420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9572134286620423e-05,
+      "loss": 1.2881,
+      "step": 19425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9571915578142242e-05,
+      "loss": 1.3362,
+      "step": 19430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9571696815003277e-05,
+      "loss": 1.1007,
+      "step": 19435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9571477997204775e-05,
+      "loss": 1.1677,
+      "step": 19440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.957125912474798e-05,
+      "loss": 1.3134,
+      "step": 19445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.957104019763415e-05,
+      "loss": 1.2158,
+      "step": 19450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9570821215864537e-05,
+      "loss": 1.0406,
+      "step": 19455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9570602179440387e-05,
+      "loss": 1.5745,
+      "step": 19460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9570383088362946e-05,
+      "loss": 1.3114,
+      "step": 19465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9570163942633473e-05,
+      "loss": 1.1069,
+      "step": 19470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9569944742253217e-05,
+      "loss": 1.2953,
+      "step": 19475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956972548722343e-05,
+      "loss": 1.2958,
+      "step": 19480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9569506177545362e-05,
+      "loss": 1.0719,
+      "step": 19485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9569286813220264e-05,
+      "loss": 1.1821,
+      "step": 19490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9569067394249393e-05,
+      "loss": 1.1657,
+      "step": 19495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9568847920634005e-05,
+      "loss": 1.1647,
+      "step": 19500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9568628392375346e-05,
+      "loss": 1.141,
+      "step": 19505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956840880947467e-05,
+      "loss": 1.1299,
+      "step": 19510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9568189171933235e-05,
+      "loss": 1.028,
+      "step": 19515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9567969479752294e-05,
+      "loss": 1.143,
+      "step": 19520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.95677497329331e-05,
+      "loss": 1.3594,
+      "step": 19525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956752993147691e-05,
+      "loss": 1.2789,
+      "step": 19530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9567310075384976e-05,
+      "loss": 1.1729,
+      "step": 19535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9567090164658557e-05,
+      "loss": 1.1143,
+      "step": 19540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956687019929891e-05,
+      "loss": 0.8476,
+      "step": 19545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9566650179307283e-05,
+      "loss": 1.2142,
+      "step": 19550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9566430104684943e-05,
+      "loss": 1.6155,
+      "step": 19555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956620997543314e-05,
+      "loss": 1.2109,
+      "step": 19560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9565989791553135e-05,
+      "loss": 1.2769,
+      "step": 19565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956576955304618e-05,
+      "loss": 1.1862,
+      "step": 19570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9565549259913535e-05,
+      "loss": 1.1062,
+      "step": 19575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9565328912156463e-05,
+      "loss": 1.1252,
+      "step": 19580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9565108509776217e-05,
+      "loss": 1.1211,
+      "step": 19585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9564888052774056e-05,
+      "loss": 1.4708,
+      "step": 19590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956466754115124e-05,
+      "loss": 1.1317,
+      "step": 19595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9564446974909025e-05,
+      "loss": 1.3303,
+      "step": 19600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956422635404868e-05,
+      "loss": 1.5766,
+      "step": 19605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9564005678571455e-05,
+      "loss": 1.4407,
+      "step": 19610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956378494847861e-05,
+      "loss": 1.1801,
+      "step": 19615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9563564163771413e-05,
+      "loss": 1.0308,
+      "step": 19620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956334332445112e-05,
+      "loss": 1.3144,
+      "step": 19625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9563122430518993e-05,
+      "loss": 1.1228,
+      "step": 19630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9562901481976294e-05,
+      "loss": 1.0582,
+      "step": 19635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9562680478824283e-05,
+      "loss": 1.3884,
+      "step": 19640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9562459421064224e-05,
+      "loss": 0.9908,
+      "step": 19645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9562238308697378e-05,
+      "loss": 1.0906,
+      "step": 19650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9562017141725007e-05,
+      "loss": 1.1635,
+      "step": 19655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9561795920148377e-05,
+      "loss": 1.3929,
+      "step": 19660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956157464396875e-05,
+      "loss": 1.2659,
+      "step": 19665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9561353313187387e-05,
+      "loss": 1.0484,
+      "step": 19670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9561131927805555e-05,
+      "loss": 1.3067,
+      "step": 19675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9560910487824518e-05,
+      "loss": 1.3629,
+      "step": 19680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956068899324554e-05,
+      "loss": 1.0694,
+      "step": 19685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9560467444069884e-05,
+      "loss": 1.3483,
+      "step": 19690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9560245840298818e-05,
+      "loss": 1.1415,
+      "step": 19695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9560024181933605e-05,
+      "loss": 0.795,
+      "step": 19700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9559802468975513e-05,
+      "loss": 2.2786,
+      "step": 19705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9559580701425805e-05,
+      "loss": 1.3525,
+      "step": 19710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.955935887928575e-05,
+      "loss": 1.2167,
+      "step": 19715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9559137002556618e-05,
+      "loss": 1.3656,
+      "step": 19720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9558915071239666e-05,
+      "loss": 1.0577,
+      "step": 19725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9558693085336175e-05,
+      "loss": 1.2307,
+      "step": 19730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.95584710448474e-05,
+      "loss": 1.1478,
+      "step": 19735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9558248949774618e-05,
+      "loss": 1.4657,
+      "step": 19740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.955802680011909e-05,
+      "loss": 1.2699,
+      "step": 19745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9557804595882093e-05,
+      "loss": 1.2315,
+      "step": 19750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9557582337064887e-05,
+      "loss": 1.0548,
+      "step": 19755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9557360023668744e-05,
+      "loss": 1.3084,
+      "step": 19760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955713765569494e-05,
+      "loss": 1.2564,
+      "step": 19765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9556915233144735e-05,
+      "loss": 1.0853,
+      "step": 19770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9556692756019406e-05,
+      "loss": 1.0833,
+      "step": 19775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955647022432022e-05,
+      "loss": 1.1256,
+      "step": 19780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955624763804845e-05,
+      "loss": 1.3202,
+      "step": 19785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9556024997205366e-05,
+      "loss": 1.2163,
+      "step": 19790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955580230179224e-05,
+      "loss": 0.9657,
+      "step": 19795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9555579551810342e-05,
+      "loss": 0.9085,
+      "step": 19800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955535674726095e-05,
+      "loss": 1.1394,
+      "step": 19805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9555133888145327e-05,
+      "loss": 1.1234,
+      "step": 19810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955491097446475e-05,
+      "loss": 1.334,
+      "step": 19815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9554688006220493e-05,
+      "loss": 0.9892,
+      "step": 19820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955446498341383e-05,
+      "loss": 1.038,
+      "step": 19825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955424190604603e-05,
+      "loss": 1.2617,
+      "step": 19830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9554018774118375e-05,
+      "loss": 1.3094,
+      "step": 19835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9553795587632128e-05,
+      "loss": 1.1192,
+      "step": 19840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955357234658857e-05,
+      "loss": 1.5831,
+      "step": 19845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955334905098898e-05,
+      "loss": 1.1039,
+      "step": 19850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9553125700834626e-05,
+      "loss": 1.6144,
+      "step": 19855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9552902296126787e-05,
+      "loss": 1.1311,
+      "step": 19860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955267883686674e-05,
+      "loss": 1.2907,
+      "step": 19865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9552455323055754e-05,
+      "loss": 1.3108,
+      "step": 19870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955223175469511e-05,
+      "loss": 1.2175,
+      "step": 19875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9552008131786092e-05,
+      "loss": 1.1464,
+      "step": 19880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9551784454329965e-05,
+      "loss": 1.2399,
+      "step": 19885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9551560722328008e-05,
+      "loss": 1.0985,
+      "step": 19890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9551336935781507e-05,
+      "loss": 1.2998,
+      "step": 19895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9551113094691736e-05,
+      "loss": 1.3062,
+      "step": 19900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955088919905997e-05,
+      "loss": 1.1084,
+      "step": 19905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.955066524888749e-05,
+      "loss": 1.4717,
+      "step": 19910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9550441244175574e-05,
+      "loss": 1.213,
+      "step": 19915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9550217184925498e-05,
+      "loss": 1.3867,
+      "step": 19920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954999307113855e-05,
+      "loss": 1.1308,
+      "step": 19925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9549768902816006e-05,
+      "loss": 1.3318,
+      "step": 19930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9549544679959144e-05,
+      "loss": 1.4084,
+      "step": 19935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9549320402569245e-05,
+      "loss": 0.8377,
+      "step": 19940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954909607064759e-05,
+      "loss": 1.424,
+      "step": 19945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9548871684195463e-05,
+      "loss": 1.093,
+      "step": 19950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954864724321414e-05,
+      "loss": 1.2638,
+      "step": 19955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9548422747704908e-05,
+      "loss": 1.318,
+      "step": 19960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9548198197669045e-05,
+      "loss": 1.2048,
+      "step": 19965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954797359310783e-05,
+      "loss": 1.2328,
+      "step": 19970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954774893402256e-05,
+      "loss": 1.2446,
+      "step": 19975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95475242204145e-05,
+      "loss": 1.2095,
+      "step": 19980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9547299452284947e-05,
+      "loss": 1.3382,
+      "step": 19985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9547074629635175e-05,
+      "loss": 1.3013,
+      "step": 19990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9546849752466474e-05,
+      "loss": 1.0682,
+      "step": 19995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9546624820780126e-05,
+      "loss": 1.1911,
+      "step": 20000
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 1.234906554222107,
+      "eval_runtime": 6468.7619,
+      "eval_samples_per_second": 3.573,
+      "eval_steps_per_second": 1.786,
+      "step": 20000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9546399834577413e-05,
+      "loss": 1.1365,
+      "step": 20005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954617479385963e-05,
+      "loss": 1.91,
+      "step": 20010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9545949698628048e-05,
+      "loss": 1.042,
+      "step": 20015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954572454888396e-05,
+      "loss": 1.0984,
+      "step": 20020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954549934462865e-05,
+      "loss": 1.3367,
+      "step": 20025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9545274085863408e-05,
+      "loss": 1.2211,
+      "step": 20030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9545048772589513e-05,
+      "loss": 1.5207,
+      "step": 20035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9544823404808256e-05,
+      "loss": 0.868,
+      "step": 20040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9544597982520927e-05,
+      "loss": 1.3005,
+      "step": 20045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954437250572881e-05,
+      "loss": 1.0364,
+      "step": 20050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9544146974433188e-05,
+      "loss": 1.4052,
+      "step": 20055
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9543921388635354e-05,
+      "loss": 1.3261,
+      "step": 20060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9543695748336597e-05,
+      "loss": 1.1369,
+      "step": 20065
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9543470053538206e-05,
+      "loss": 1.5081,
+      "step": 20070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9543244304241468e-05,
+      "loss": 1.1226,
+      "step": 20075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954301850044767e-05,
+      "loss": 1.5,
+      "step": 20080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9542792642158103e-05,
+      "loss": 1.0992,
+      "step": 20085
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954256672937406e-05,
+      "loss": 1.0155,
+      "step": 20090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9542340762096827e-05,
+      "loss": 1.6493,
+      "step": 20095
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9542114740327696e-05,
+      "loss": 1.1595,
+      "step": 20100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954188866406796e-05,
+      "loss": 1.3741,
+      "step": 20105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9541662533318905e-05,
+      "loss": 1.1869,
+      "step": 20110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9541436348081824e-05,
+      "loss": 1.4119,
+      "step": 20115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954121010835801e-05,
+      "loss": 1.1902,
+      "step": 20120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9540983814148758e-05,
+      "loss": 1.5608,
+      "step": 20125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9540757465455352e-05,
+      "loss": 1.0776,
+      "step": 20130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9540531062279092e-05,
+      "loss": 1.4458,
+      "step": 20135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954030460462127e-05,
+      "loss": 1.5088,
+      "step": 20140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9540078092483173e-05,
+      "loss": 1.2351,
+      "step": 20145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9539851525866102e-05,
+      "loss": 1.182,
+      "step": 20150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953962490477135e-05,
+      "loss": 1.2184,
+      "step": 20155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9539398229200203e-05,
+      "loss": 1.2184,
+      "step": 20160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9539171499153963e-05,
+      "loss": 1.3171,
+      "step": 20165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9538944714633926e-05,
+      "loss": 1.4778,
+      "step": 20170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9538717875641383e-05,
+      "loss": 1.1174,
+      "step": 20175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953849098217763e-05,
+      "loss": 1.4237,
+      "step": 20180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953826403424396e-05,
+      "loss": 1.1861,
+      "step": 20185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9538037031841676e-05,
+      "loss": 1.1762,
+      "step": 20190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9537809974972072e-05,
+      "loss": 1.1716,
+      "step": 20195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953758286363644e-05,
+      "loss": 1.2758,
+      "step": 20200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953735569783608e-05,
+      "loss": 1.1379,
+      "step": 20205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953712847757229e-05,
+      "loss": 1.2554,
+      "step": 20210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953690120284637e-05,
+      "loss": 1.1431,
+      "step": 20215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953667387365961e-05,
+      "loss": 1.5045,
+      "step": 20220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9536446490013318e-05,
+      "loss": 1.0301,
+      "step": 20225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953621905190878e-05,
+      "loss": 1.202,
+      "step": 20230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9535991559347307e-05,
+      "loss": 1.3676,
+      "step": 20235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9535764012330193e-05,
+      "loss": 1.082,
+      "step": 20240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9535536410858737e-05,
+      "loss": 1.0808,
+      "step": 20245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953530875493424e-05,
+      "loss": 1.438,
+      "step": 20250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9535081044558e-05,
+      "loss": 1.3526,
+      "step": 20255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953485327973132e-05,
+      "loss": 1.5509,
+      "step": 20260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95346254604555e-05,
+      "loss": 1.5743,
+      "step": 20265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953439758673184e-05,
+      "loss": 1.5429,
+      "step": 20270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953416965856164e-05,
+      "loss": 1.4524,
+      "step": 20275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9533941675946205e-05,
+      "loss": 1.0215,
+      "step": 20280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9533713638886833e-05,
+      "loss": 1.2776,
+      "step": 20285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9533485547384828e-05,
+      "loss": 1.2736,
+      "step": 20290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9533257401441498e-05,
+      "loss": 1.2919,
+      "step": 20295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9533029201058137e-05,
+      "loss": 1.3498,
+      "step": 20300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953280094623605e-05,
+      "loss": 1.3225,
+      "step": 20305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953257263697655e-05,
+      "loss": 1.0711,
+      "step": 20310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9532344273280923e-05,
+      "loss": 1.1229,
+      "step": 20315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953211585515049e-05,
+      "loss": 1.2175,
+      "step": 20320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9531887382586547e-05,
+      "loss": 1.3552,
+      "step": 20325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95316588555904e-05,
+      "loss": 1.0415,
+      "step": 20330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9531430274163355e-05,
+      "loss": 1.4672,
+      "step": 20335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9531201638306715e-05,
+      "loss": 1.2954,
+      "step": 20340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953097294802179e-05,
+      "loss": 1.3668,
+      "step": 20345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953074420330988e-05,
+      "loss": 1.1704,
+      "step": 20350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9530515404172298e-05,
+      "loss": 1.0533,
+      "step": 20355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9530286550610342e-05,
+      "loss": 1.2865,
+      "step": 20360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9530057642625327e-05,
+      "loss": 1.4738,
+      "step": 20365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952982868021856e-05,
+      "loss": 1.4419,
+      "step": 20370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952959966339134e-05,
+      "loss": 1.0078,
+      "step": 20375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9529370592144983e-05,
+      "loss": 1.2895,
+      "step": 20380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9529141466480795e-05,
+      "loss": 1.2372,
+      "step": 20385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9528912286400083e-05,
+      "loss": 1.2952,
+      "step": 20390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9528683051904158e-05,
+      "loss": 1.2763,
+      "step": 20395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9528453762994325e-05,
+      "loss": 1.3692,
+      "step": 20400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95282244196719e-05,
+      "loss": 1.1668,
+      "step": 20405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9527995021938187e-05,
+      "loss": 1.5622,
+      "step": 20410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9527765569794497e-05,
+      "loss": 1.1785,
+      "step": 20415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952753606324214e-05,
+      "loss": 1.2082,
+      "step": 20420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952730650228243e-05,
+      "loss": 1.284,
+      "step": 20425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9527076886916675e-05,
+      "loss": 1.1096,
+      "step": 20430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952684721714619e-05,
+      "loss": 1.1604,
+      "step": 20435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9526617492972278e-05,
+      "loss": 1.5018,
+      "step": 20440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952638771439626e-05,
+      "loss": 1.2062,
+      "step": 20445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952615788141944e-05,
+      "loss": 1.1018,
+      "step": 20450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9525927994043138e-05,
+      "loss": 1.37,
+      "step": 20455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952569805226866e-05,
+      "loss": 1.5316,
+      "step": 20460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952546805609733e-05,
+      "loss": 1.1617,
+      "step": 20465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9525238005530445e-05,
+      "loss": 1.105,
+      "step": 20470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9525007900569335e-05,
+      "loss": 1.167,
+      "step": 20475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9524777741215307e-05,
+      "loss": 1.129,
+      "step": 20480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952454752746967e-05,
+      "loss": 1.124,
+      "step": 20485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952431725933375e-05,
+      "loss": 1.4562,
+      "step": 20490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952408693680885e-05,
+      "loss": 1.4212,
+      "step": 20495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9523856559896294e-05,
+      "loss": 1.3307,
+      "step": 20500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9523626128597394e-05,
+      "loss": 1.4834,
+      "step": 20505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9523395642913467e-05,
+      "loss": 1.0222,
+      "step": 20510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952316510284583e-05,
+      "loss": 1.2971,
+      "step": 20515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9522934508395794e-05,
+      "loss": 1.067,
+      "step": 20520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952270385956468e-05,
+      "loss": 1.3035,
+      "step": 20525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952247315635381e-05,
+      "loss": 1.3845,
+      "step": 20530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9522242398764493e-05,
+      "loss": 1.1912,
+      "step": 20535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952201158679805e-05,
+      "loss": 1.1949,
+      "step": 20540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95217807204558e-05,
+      "loss": 1.2668,
+      "step": 20545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9521549799739063e-05,
+      "loss": 1.6051,
+      "step": 20550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9521318824649154e-05,
+      "loss": 1.343,
+      "step": 20555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9521087795187395e-05,
+      "loss": 1.08,
+      "step": 20560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9520856711355102e-05,
+      "loss": 1.3592,
+      "step": 20565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9520625573153598e-05,
+      "loss": 1.0845,
+      "step": 20570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95203943805842e-05,
+      "loss": 1.3039,
+      "step": 20575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.952016313364823e-05,
+      "loss": 1.2295,
+      "step": 20580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9519931832347006e-05,
+      "loss": 0.9361,
+      "step": 20585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9519700476681854e-05,
+      "loss": 1.2576,
+      "step": 20590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951946906665409e-05,
+      "loss": 1.0002,
+      "step": 20595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951923760226504e-05,
+      "loss": 1.1003,
+      "step": 20600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951900608351602e-05,
+      "loss": 1.0635,
+      "step": 20605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9518774510408355e-05,
+      "loss": 1.2661,
+      "step": 20610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951854288294337e-05,
+      "loss": 1.4027,
+      "step": 20615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9518311201122386e-05,
+      "loss": 1.1593,
+      "step": 20620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9518079464946727e-05,
+      "loss": 1.015,
+      "step": 20625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9517847674417712e-05,
+      "loss": 1.1417,
+      "step": 20630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9517615829536667e-05,
+      "loss": 1.1995,
+      "step": 20635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951738393030492e-05,
+      "loss": 1.1247,
+      "step": 20640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951715197672379e-05,
+      "loss": 1.463,
+      "step": 20645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9516919968794602e-05,
+      "loss": 1.2811,
+      "step": 20650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9516687906518682e-05,
+      "loss": 1.2897,
+      "step": 20655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951645578989736e-05,
+      "loss": 1.2501,
+      "step": 20660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9516223618931953e-05,
+      "loss": 1.129,
+      "step": 20665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951599139362379e-05,
+      "loss": 1.0546,
+      "step": 20670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.95157591139742e-05,
+      "loss": 1.085,
+      "step": 20675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9515526779984507e-05,
+      "loss": 1.0366,
+      "step": 20680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951529439165604e-05,
+      "loss": 1.3334,
+      "step": 20685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9515061948990122e-05,
+      "loss": 0.986,
+      "step": 20690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951482945198808e-05,
+      "loss": 1.1388,
+      "step": 20695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951459690065125e-05,
+      "loss": 1.0923,
+      "step": 20700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951436429498095e-05,
+      "loss": 1.1566,
+      "step": 20705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9514131634978515e-05,
+      "loss": 1.1649,
+      "step": 20710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951389892064527e-05,
+      "loss": 1.7342,
+      "step": 20715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9513666151982545e-05,
+      "loss": 1.1062,
+      "step": 20720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951343332899167e-05,
+      "loss": 1.3059,
+      "step": 20725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9513200451673975e-05,
+      "loss": 1.1013,
+      "step": 20730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9512967520030786e-05,
+      "loss": 1.3805,
+      "step": 20735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9512734534063438e-05,
+      "loss": 1.3806,
+      "step": 20740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9512501493773258e-05,
+      "loss": 1.3599,
+      "step": 20745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9512268399161578e-05,
+      "loss": 1.1167,
+      "step": 20750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9512035250229733e-05,
+      "loss": 1.2871,
+      "step": 20755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9511802046979043e-05,
+      "loss": 1.1532,
+      "step": 20760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9511568789410855e-05,
+      "loss": 1.3594,
+      "step": 20765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951133547752649e-05,
+      "loss": 1.1273,
+      "step": 20770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951110211132728e-05,
+      "loss": 1.3185,
+      "step": 20775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9510868690814568e-05,
+      "loss": 1.5525,
+      "step": 20780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.951063521598968e-05,
+      "loss": 1.6163,
+      "step": 20785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9510401686853946e-05,
+      "loss": 1.2887,
+      "step": 20790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9510168103408707e-05,
+      "loss": 1.1877,
+      "step": 20795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.950993446565529e-05,
+      "loss": 1.2148,
+      "step": 20800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9509700773595032e-05,
+      "loss": 1.1552,
+      "step": 20805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9509467027229266e-05,
+      "loss": 1.2769,
+      "step": 20810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.950923322655933e-05,
+      "loss": 1.1329,
+      "step": 20815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9508999371586564e-05,
+      "loss": 0.9789,
+      "step": 20820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.950876546231229e-05,
+      "loss": 1.5222,
+      "step": 20825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9508531498737854e-05,
+      "loss": 1.2473,
+      "step": 20830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9508297480864588e-05,
+      "loss": 1.1691,
+      "step": 20835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.950806340869383e-05,
+      "loss": 1.4237,
+      "step": 20840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9507829282226915e-05,
+      "loss": 1.1463,
+      "step": 20845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.950759510146518e-05,
+      "loss": 1.2257,
+      "step": 20850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9507360866409964e-05,
+      "loss": 1.5895,
+      "step": 20855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9507126577062604e-05,
+      "loss": 1.0079,
+      "step": 20860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9506892233424438e-05,
+      "loss": 1.1674,
+      "step": 20865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9506657835496806e-05,
+      "loss": 1.3249,
+      "step": 20870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9506423383281043e-05,
+      "loss": 1.3823,
+      "step": 20875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9506188876778488e-05,
+      "loss": 1.2325,
+      "step": 20880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9505954315990486e-05,
+      "loss": 1.1113,
+      "step": 20885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.950571970091837e-05,
+      "loss": 1.141,
+      "step": 20890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9505485031563477e-05,
+      "loss": 1.3743,
+      "step": 20895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9505250307927157e-05,
+      "loss": 1.2925,
+      "step": 20900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9505015530010746e-05,
+      "loss": 1.8606,
+      "step": 20905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.950478069781558e-05,
+      "loss": 1.2439,
+      "step": 20910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9504545811343006e-05,
+      "loss": 1.042,
+      "step": 20915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9504310870594364e-05,
+      "loss": 1.2437,
+      "step": 20920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9504075875570995e-05,
+      "loss": 1.2416,
+      "step": 20925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9503840826274237e-05,
+      "loss": 1.2485,
+      "step": 20930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9503605722705442e-05,
+      "loss": 1.181,
+      "step": 20935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9503370564865945e-05,
+      "loss": 1.2364,
+      "step": 20940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.950313535275709e-05,
+      "loss": 1.5688,
+      "step": 20945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.950290008638022e-05,
+      "loss": 1.2241,
+      "step": 20950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.950266476573668e-05,
+      "loss": 1.3949,
+      "step": 20955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9502429390827813e-05,
+      "loss": 1.1569,
+      "step": 20960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9502193961654966e-05,
+      "loss": 1.0404,
+      "step": 20965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9501958478219477e-05,
+      "loss": 1.6026,
+      "step": 20970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9501722940522694e-05,
+      "loss": 1.3824,
+      "step": 20975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9501487348565968e-05,
+      "loss": 1.4914,
+      "step": 20980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9501251702350635e-05,
+      "loss": 1.1371,
+      "step": 20985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9501016001878044e-05,
+      "loss": 1.5717,
+      "step": 20990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9500780247149543e-05,
+      "loss": 1.2625,
+      "step": 20995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9500544438166475e-05,
+      "loss": 1.2052,
+      "step": 21000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9500308574930188e-05,
+      "loss": 1.3421,
+      "step": 21005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.950007265744203e-05,
+      "loss": 1.0295,
+      "step": 21010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.949983668570335e-05,
+      "loss": 1.3949,
+      "step": 21015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9499600659715493e-05,
+      "loss": 1.4757,
+      "step": 21020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9499364579479803e-05,
+      "loss": 1.2228,
+      "step": 21025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9499128444997633e-05,
+      "loss": 1.3506,
+      "step": 21030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9498892256270335e-05,
+      "loss": 1.2295,
+      "step": 21035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9498656013299247e-05,
+      "loss": 1.0046,
+      "step": 21040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9498419716085728e-05,
+      "loss": 1.631,
+      "step": 21045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9498183364631122e-05,
+      "loss": 1.3061,
+      "step": 21050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.949794695893678e-05,
+      "loss": 1.1836,
+      "step": 21055
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9497710499004055e-05,
+      "loss": 0.9622,
+      "step": 21060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9497473984834295e-05,
+      "loss": 1.3001,
+      "step": 21065
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9497237416428846e-05,
+      "loss": 1.4687,
+      "step": 21070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9497000793789066e-05,
+      "loss": 1.2813,
+      "step": 21075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9496764116916304e-05,
+      "loss": 1.4555,
+      "step": 21080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.949652738581191e-05,
+      "loss": 1.3164,
+      "step": 21085
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9496290600477236e-05,
+      "loss": 1.0562,
+      "step": 21090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9496053760913636e-05,
+      "loss": 1.1572,
+      "step": 21095
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9495816867122462e-05,
+      "loss": 1.0096,
+      "step": 21100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9495579919105066e-05,
+      "loss": 1.2155,
+      "step": 21105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9495342916862802e-05,
+      "loss": 1.3274,
+      "step": 21110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9495105860397023e-05,
+      "loss": 1.4024,
+      "step": 21115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9494868749709083e-05,
+      "loss": 1.4122,
+      "step": 21120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.949463158480033e-05,
+      "loss": 1.3278,
+      "step": 21125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.949439436567213e-05,
+      "loss": 1.3632,
+      "step": 21130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9494157092325832e-05,
+      "loss": 1.2072,
+      "step": 21135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9493919764762787e-05,
+      "loss": 1.1896,
+      "step": 21140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9493682382984355e-05,
+      "loss": 1.1927,
+      "step": 21145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.949344494699189e-05,
+      "loss": 1.39,
+      "step": 21150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9493207456786753e-05,
+      "loss": 1.1881,
+      "step": 21155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.949296991237029e-05,
+      "loss": 1.3397,
+      "step": 21160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.949273231374386e-05,
+      "loss": 1.246,
+      "step": 21165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9492494660908828e-05,
+      "loss": 1.1983,
+      "step": 21170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.949225695386655e-05,
+      "loss": 1.0058,
+      "step": 21175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9492019192618373e-05,
+      "loss": 1.2393,
+      "step": 21180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9491781377165662e-05,
+      "loss": 1.3645,
+      "step": 21185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9491543507509774e-05,
+      "loss": 1.0125,
+      "step": 21190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.949130558365207e-05,
+      "loss": 1.0506,
+      "step": 21195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9491067605593904e-05,
+      "loss": 0.9419,
+      "step": 21200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9490829573336638e-05,
+      "loss": 1.2056,
+      "step": 21205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.949059148688163e-05,
+      "loss": 1.2405,
+      "step": 21210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.949035334623024e-05,
+      "loss": 1.4116,
+      "step": 21215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9490115151383828e-05,
+      "loss": 1.4794,
+      "step": 21220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9489876902343754e-05,
+      "loss": 1.3057,
+      "step": 21225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.948963859911138e-05,
+      "loss": 1.3921,
+      "step": 21230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9489400241688065e-05,
+      "loss": 1.5612,
+      "step": 21235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.948916183007517e-05,
+      "loss": 1.1683,
+      "step": 21240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.948892336427406e-05,
+      "loss": 0.9979,
+      "step": 21245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.948868484428609e-05,
+      "loss": 1.2069,
+      "step": 21250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9488446270112627e-05,
+      "loss": 1.2905,
+      "step": 21255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9488207641755034e-05,
+      "loss": 1.1852,
+      "step": 21260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.948796895921467e-05,
+      "loss": 1.1212,
+      "step": 21265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.94877302224929e-05,
+      "loss": 1.0976,
+      "step": 21270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.948749143159109e-05,
+      "loss": 1.2059,
+      "step": 21275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.94872525865106e-05,
+      "loss": 1.3153,
+      "step": 21280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9487013687252796e-05,
+      "loss": 1.1769,
+      "step": 21285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9486774733819042e-05,
+      "loss": 1.4528,
+      "step": 21290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9486535726210697e-05,
+      "loss": 1.2189,
+      "step": 21295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9486296664429133e-05,
+      "loss": 1.3242,
+      "step": 21300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9486057548475713e-05,
+      "loss": 1.1951,
+      "step": 21305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9485818378351804e-05,
+      "loss": 1.2366,
+      "step": 21310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.948557915405877e-05,
+      "loss": 1.3803,
+      "step": 21315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9485339875597976e-05,
+      "loss": 1.1895,
+      "step": 21320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9485100542970787e-05,
+      "loss": 1.0884,
+      "step": 21325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9484861156178575e-05,
+      "loss": 1.0465,
+      "step": 21330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9484621715222705e-05,
+      "loss": 1.0924,
+      "step": 21335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9484382220104544e-05,
+      "loss": 1.3078,
+      "step": 21340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9484142670825456e-05,
+      "loss": 1.218,
+      "step": 21345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9483903067386817e-05,
+      "loss": 1.1683,
+      "step": 21350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9483663409789986e-05,
+      "loss": 1.4619,
+      "step": 21355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.948342369803634e-05,
+      "loss": 1.3713,
+      "step": 21360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.948318393212724e-05,
+      "loss": 1.271,
+      "step": 21365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.948294411206406e-05,
+      "loss": 1.0896,
+      "step": 21370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9482704237848173e-05,
+      "loss": 1.429,
+      "step": 21375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.948246430948094e-05,
+      "loss": 1.5466,
+      "step": 21380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.948222432696374e-05,
+      "loss": 0.9395,
+      "step": 21385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9481984290297934e-05,
+      "loss": 1.2523,
+      "step": 21390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.94817441994849e-05,
+      "loss": 1.3743,
+      "step": 21395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9481504054526005e-05,
+      "loss": 1.2761,
+      "step": 21400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9481263855422624e-05,
+      "loss": 1.1932,
+      "step": 21405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9481023602176127e-05,
+      "loss": 1.3557,
+      "step": 21410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9480783294787888e-05,
+      "loss": 1.0085,
+      "step": 21415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9480542933259274e-05,
+      "loss": 1.2288,
+      "step": 21420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9480302517591662e-05,
+      "loss": 1.4873,
+      "step": 21425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9480062047786424e-05,
+      "loss": 1.3206,
+      "step": 21430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.947982152384493e-05,
+      "loss": 1.2101,
+      "step": 21435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.947958094576856e-05,
+      "loss": 1.2272,
+      "step": 21440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9479340313558683e-05,
+      "loss": 1.2396,
+      "step": 21445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9479099627216674e-05,
+      "loss": 1.4699,
+      "step": 21450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9478858886743904e-05,
+      "loss": 1.0833,
+      "step": 21455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9478618092141758e-05,
+      "loss": 1.446,
+      "step": 21460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.94783772434116e-05,
+      "loss": 1.3701,
+      "step": 21465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.947813634055481e-05,
+      "loss": 1.1076,
+      "step": 21470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.947789538357277e-05,
+      "loss": 1.3093,
+      "step": 21475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9477654372466844e-05,
+      "loss": 1.2184,
+      "step": 21480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9477413307238416e-05,
+      "loss": 1.1229,
+      "step": 21485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.947717218788886e-05,
+      "loss": 1.2851,
+      "step": 21490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.947693101441955e-05,
+      "loss": 0.9413,
+      "step": 21495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9476689786831872e-05,
+      "loss": 1.5446,
+      "step": 21500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9476448505127194e-05,
+      "loss": 1.2487,
+      "step": 21505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.94762071693069e-05,
+      "loss": 1.2955,
+      "step": 21510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9475965779372368e-05,
+      "loss": 1.1391,
+      "step": 21515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.947572433532497e-05,
+      "loss": 1.0756,
+      "step": 21520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.94754828371661e-05,
+      "loss": 0.9991,
+      "step": 21525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9475241284897114e-05,
+      "loss": 1.2334,
+      "step": 21530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.947499967851941e-05,
+      "loss": 1.3908,
+      "step": 21535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9474758018034362e-05,
+      "loss": 1.1945,
+      "step": 21540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9474516303443348e-05,
+      "loss": 1.0043,
+      "step": 21545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.947427453474775e-05,
+      "loss": 1.265,
+      "step": 21550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.947403271194895e-05,
+      "loss": 1.0046,
+      "step": 21555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9473790835048328e-05,
+      "loss": 1.1239,
+      "step": 21560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9473548904047264e-05,
+      "loss": 1.2256,
+      "step": 21565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.947330691894714e-05,
+      "loss": 1.3524,
+      "step": 21570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9473064879749337e-05,
+      "loss": 1.4812,
+      "step": 21575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9472822786455243e-05,
+      "loss": 1.0977,
+      "step": 21580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.947258063906623e-05,
+      "loss": 1.0234,
+      "step": 21585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.947233843758369e-05,
+      "loss": 1.4395,
+      "step": 21590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9472096182009e-05,
+      "loss": 0.9967,
+      "step": 21595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9471853872343553e-05,
+      "loss": 1.4585,
+      "step": 21600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9471611508588717e-05,
+      "loss": 1.3621,
+      "step": 21605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9471369090745892e-05,
+      "loss": 1.2078,
+      "step": 21610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9471126618816452e-05,
+      "loss": 1.2985,
+      "step": 21615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9470884092801787e-05,
+      "loss": 1.1207,
+      "step": 21620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9470641512703278e-05,
+      "loss": 1.1371,
+      "step": 21625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9470398878522312e-05,
+      "loss": 1.1705,
+      "step": 21630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9470156190260275e-05,
+      "loss": 1.5217,
+      "step": 21635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.946991344791855e-05,
+      "loss": 1.1248,
+      "step": 21640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.946967065149853e-05,
+      "loss": 1.2426,
+      "step": 21645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9469427801001596e-05,
+      "loss": 1.1944,
+      "step": 21650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9469184896429137e-05,
+      "loss": 1.1127,
+      "step": 21655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.946894193778254e-05,
+      "loss": 1.2713,
+      "step": 21660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9468698925063188e-05,
+      "loss": 1.2832,
+      "step": 21665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9468455858272473e-05,
+      "loss": 1.4039,
+      "step": 21670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9468212737411787e-05,
+      "loss": 1.2054,
+      "step": 21675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.946796956248251e-05,
+      "loss": 1.2724,
+      "step": 21680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.946772633348603e-05,
+      "loss": 1.1305,
+      "step": 21685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.946748305042375e-05,
+      "loss": 1.3849,
+      "step": 21690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9467239713297045e-05,
+      "loss": 1.2358,
+      "step": 21695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9466996322107308e-05,
+      "loss": 1.02,
+      "step": 21700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9466752876855934e-05,
+      "loss": 1.1712,
+      "step": 21705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9466509377544305e-05,
+      "loss": 1.1662,
+      "step": 21710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9466265824173816e-05,
+      "loss": 1.4019,
+      "step": 21715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.946602221674586e-05,
+      "loss": 1.3503,
+      "step": 21720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9465778555261826e-05,
+      "loss": 1.3489,
+      "step": 21725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9465534839723105e-05,
+      "loss": 1.1445,
+      "step": 21730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9465291070131086e-05,
+      "loss": 1.1562,
+      "step": 21735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9465047246487166e-05,
+      "loss": 1.6095,
+      "step": 21740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9464803368792736e-05,
+      "loss": 1.1477,
+      "step": 21745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.946455943704919e-05,
+      "loss": 1.5733,
+      "step": 21750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9464315451257918e-05,
+      "loss": 1.0654,
+      "step": 21755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9464071411420314e-05,
+      "loss": 1.3559,
+      "step": 21760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.946382731753777e-05,
+      "loss": 1.706,
+      "step": 21765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9463583169611685e-05,
+      "loss": 1.3125,
+      "step": 21770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9463338967643447e-05,
+      "loss": 1.3912,
+      "step": 21775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9463094711634457e-05,
+      "loss": 1.1038,
+      "step": 21780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9462850401586106e-05,
+      "loss": 1.3033,
+      "step": 21785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9462606037499784e-05,
+      "loss": 1.3028,
+      "step": 21790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9462361619376897e-05,
+      "loss": 1.2219,
+      "step": 21795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9462117147218837e-05,
+      "loss": 1.3451,
+      "step": 21800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9461872621026994e-05,
+      "loss": 1.0235,
+      "step": 21805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9461628040802773e-05,
+      "loss": 1.2862,
+      "step": 21810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9461383406547564e-05,
+      "loss": 1.5116,
+      "step": 21815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.946113871826277e-05,
+      "loss": 1.3206,
+      "step": 21820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9460893975949786e-05,
+      "loss": 1.2941,
+      "step": 21825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9460649179610005e-05,
+      "loss": 1.0542,
+      "step": 21830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9460404329244832e-05,
+      "loss": 1.183,
+      "step": 21835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.946015942485566e-05,
+      "loss": 1.2307,
+      "step": 21840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945991446644389e-05,
+      "loss": 1.3723,
+      "step": 21845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945966945401092e-05,
+      "loss": 1.0544,
+      "step": 21850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945942438755815e-05,
+      "loss": 1.3227,
+      "step": 21855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9459179267086982e-05,
+      "loss": 1.3538,
+      "step": 21860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9458934092598808e-05,
+      "loss": 1.1348,
+      "step": 21865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9458688864095035e-05,
+      "loss": 1.5319,
+      "step": 21870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945844358157706e-05,
+      "loss": 1.4848,
+      "step": 21875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9458198245046283e-05,
+      "loss": 1.2367,
+      "step": 21880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9457952854504114e-05,
+      "loss": 1.0602,
+      "step": 21885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9457707409951943e-05,
+      "loss": 1.1587,
+      "step": 21890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9457461911391173e-05,
+      "loss": 1.0561,
+      "step": 21895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9457216358823215e-05,
+      "loss": 1.0791,
+      "step": 21900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945697075224946e-05,
+      "loss": 1.2056,
+      "step": 21905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9456725091671318e-05,
+      "loss": 1.1487,
+      "step": 21910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9456479377090186e-05,
+      "loss": 1.2828,
+      "step": 21915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9456233608507475e-05,
+      "loss": 1.1427,
+      "step": 21920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945598778592458e-05,
+      "loss": 1.4633,
+      "step": 21925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9455741909342913e-05,
+      "loss": 1.0986,
+      "step": 21930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945549597876387e-05,
+      "loss": 1.1366,
+      "step": 21935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9455249994188862e-05,
+      "loss": 1.0572,
+      "step": 21940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945500395561929e-05,
+      "loss": 1.0955,
+      "step": 21945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945475786305656e-05,
+      "loss": 1.0745,
+      "step": 21950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945451171650208e-05,
+      "loss": 1.3071,
+      "step": 21955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9454265515957247e-05,
+      "loss": 1.1558,
+      "step": 21960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9454019261423477e-05,
+      "loss": 1.2048,
+      "step": 21965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945377295290217e-05,
+      "loss": 1.4509,
+      "step": 21970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9453526590394738e-05,
+      "loss": 1.0324,
+      "step": 21975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9453280173902584e-05,
+      "loss": 1.1126,
+      "step": 21980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9453033703427112e-05,
+      "loss": 1.1122,
+      "step": 21985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9452787178969736e-05,
+      "loss": 1.4379,
+      "step": 21990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945254060053186e-05,
+      "loss": 1.5548,
+      "step": 21995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9452293968114895e-05,
+      "loss": 1.0634,
+      "step": 22000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9452047281720246e-05,
+      "loss": 1.2785,
+      "step": 22005
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9451800541349323e-05,
+      "loss": 1.205,
+      "step": 22010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9451553747003537e-05,
+      "loss": 1.2349,
+      "step": 22015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9451306898684296e-05,
+      "loss": 1.0759,
+      "step": 22020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9451059996393003e-05,
+      "loss": 1.1617,
+      "step": 22025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9450813040131082e-05,
+      "loss": 1.2603,
+      "step": 22030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.945056602989993e-05,
+      "loss": 1.5763,
+      "step": 22035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9450318965700967e-05,
+      "loss": 1.2728,
+      "step": 22040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9450071847535596e-05,
+      "loss": 1.2246,
+      "step": 22045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9449824675405234e-05,
+      "loss": 1.1524,
+      "step": 22050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944957744931129e-05,
+      "loss": 1.1936,
+      "step": 22055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944933016925518e-05,
+      "loss": 1.5238,
+      "step": 22060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9449082835238305e-05,
+      "loss": 1.0732,
+      "step": 22065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9448835447262092e-05,
+      "loss": 1.2784,
+      "step": 22070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944858800532794e-05,
+      "loss": 1.1272,
+      "step": 22075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9448340509437274e-05,
+      "loss": 1.1352,
+      "step": 22080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.94480929595915e-05,
+      "loss": 1.1104,
+      "step": 22085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9447845355792032e-05,
+      "loss": 1.1177,
+      "step": 22090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9447597698040287e-05,
+      "loss": 1.2513,
+      "step": 22095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9447349986337676e-05,
+      "loss": 1.2457,
+      "step": 22100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944710222068562e-05,
+      "loss": 1.2522,
+      "step": 22105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9446854401085523e-05,
+      "loss": 1.8722,
+      "step": 22110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9446606527538812e-05,
+      "loss": 1.2269,
+      "step": 22115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9446358600046894e-05,
+      "loss": 1.1605,
+      "step": 22120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9446110618611188e-05,
+      "loss": 1.7115,
+      "step": 22125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9445862583233107e-05,
+      "loss": 1.6098,
+      "step": 22130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9445614493914077e-05,
+      "loss": 1.2723,
+      "step": 22135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.94453663506555e-05,
+      "loss": 1.6489,
+      "step": 22140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9445118153458806e-05,
+      "loss": 1.1734,
+      "step": 22145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9444869902325404e-05,
+      "loss": 1.2122,
+      "step": 22150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944462159725672e-05,
+      "loss": 1.2254,
+      "step": 22155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9444373238254162e-05,
+      "loss": 1.0558,
+      "step": 22160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9444124825319154e-05,
+      "loss": 1.1729,
+      "step": 22165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9443876358453115e-05,
+      "loss": 1.1447,
+      "step": 22170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9443627837657463e-05,
+      "loss": 1.3877,
+      "step": 22175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9443379262933616e-05,
+      "loss": 1.2475,
+      "step": 22180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9443130634282993e-05,
+      "loss": 1.1733,
+      "step": 22185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9442881951707017e-05,
+      "loss": 0.8633,
+      "step": 22190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9442633215207104e-05,
+      "loss": 1.0494,
+      "step": 22195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9442384424784677e-05,
+      "loss": 0.9981,
+      "step": 22200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944213558044116e-05,
+      "loss": 1.4914,
+      "step": 22205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9441886682177967e-05,
+      "loss": 1.1877,
+      "step": 22210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944163772999652e-05,
+      "loss": 1.4974,
+      "step": 22215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944138872389825e-05,
+      "loss": 1.4646,
+      "step": 22220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944113966388457e-05,
+      "loss": 1.3023,
+      "step": 22225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9440890549956905e-05,
+      "loss": 1.212,
+      "step": 22230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.944064138211667e-05,
+      "loss": 1.3299,
+      "step": 22235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.94403921603653e-05,
+      "loss": 1.1331,
+      "step": 22240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9440142884704218e-05,
+      "loss": 1.1217,
+      "step": 22245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943989355513484e-05,
+      "loss": 1.1907,
+      "step": 22250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943964417165859e-05,
+      "loss": 1.2589,
+      "step": 22255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9439394734276896e-05,
+      "loss": 1.1785,
+      "step": 22260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943914524299118e-05,
+      "loss": 1.3029,
+      "step": 22265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943889569780287e-05,
+      "loss": 1.2267,
+      "step": 22270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9438646098713385e-05,
+      "loss": 1.0741,
+      "step": 22275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9438396445724155e-05,
+      "loss": 1.3013,
+      "step": 22280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943814673883661e-05,
+      "loss": 1.3211,
+      "step": 22285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9437896978052168e-05,
+      "loss": 1.5035,
+      "step": 22290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9437647163372256e-05,
+      "loss": 1.163,
+      "step": 22295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9437397294798304e-05,
+      "loss": 1.4436,
+      "step": 22300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943714737233174e-05,
+      "loss": 1.2124,
+      "step": 22305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9436897395973986e-05,
+      "loss": 1.2491,
+      "step": 22310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943664736572647e-05,
+      "loss": 1.1548,
+      "step": 22315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9436397281590626e-05,
+      "loss": 1.0489,
+      "step": 22320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9436147143567877e-05,
+      "loss": 1.525,
+      "step": 22325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9435896951659653e-05,
+      "loss": 1.1814,
+      "step": 22330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943564670586738e-05,
+      "loss": 1.0932,
+      "step": 22335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943539640619249e-05,
+      "loss": 0.9121,
+      "step": 22340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943514605263641e-05,
+      "loss": 1.1703,
+      "step": 22345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9434895645200577e-05,
+      "loss": 1.2804,
+      "step": 22350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9434645183886413e-05,
+      "loss": 1.0942,
+      "step": 22355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943439466869535e-05,
+      "loss": 1.4131,
+      "step": 22360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943414409962882e-05,
+      "loss": 1.9427,
+      "step": 22365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943389347668825e-05,
+      "loss": 1.0322,
+      "step": 22370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9433642799875074e-05,
+      "loss": 0.998,
+      "step": 22375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9433392069190727e-05,
+      "loss": 1.3392,
+      "step": 22380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9433141284636637e-05,
+      "loss": 1.1743,
+      "step": 22385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9432890446214237e-05,
+      "loss": 1.3484,
+      "step": 22390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9432639553924956e-05,
+      "loss": 1.0955,
+      "step": 22395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943238860777023e-05,
+      "loss": 1.1709,
+      "step": 22400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9432137607751494e-05,
+      "loss": 1.1515,
+      "step": 22405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943188655387018e-05,
+      "loss": 1.1732,
+      "step": 22410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943163544612772e-05,
+      "loss": 1.3189,
+      "step": 22415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9431384284525548e-05,
+      "loss": 1.2149,
+      "step": 22420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.94311330690651e-05,
+      "loss": 0.9475,
+      "step": 22425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943088179974781e-05,
+      "loss": 1.2115,
+      "step": 22430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943063047657511e-05,
+      "loss": 0.9449,
+      "step": 22435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943037909954844e-05,
+      "loss": 1.0697,
+      "step": 22440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.943012766866923e-05,
+      "loss": 1.0352,
+      "step": 22445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9429876183938923e-05,
+      "loss": 1.0481,
+      "step": 22450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942962464535895e-05,
+      "loss": 1.4529,
+      "step": 22455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9429373052930747e-05,
+      "loss": 1.2563,
+      "step": 22460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9429121406655752e-05,
+      "loss": 1.541,
+      "step": 22465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9428869706535404e-05,
+      "loss": 1.4964,
+      "step": 22470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9428617952571137e-05,
+      "loss": 1.2901,
+      "step": 22475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9428366144764394e-05,
+      "loss": 1.3591,
+      "step": 22480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9428114283116608e-05,
+      "loss": 1.1881,
+      "step": 22485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9427862367629216e-05,
+      "loss": 1.1868,
+      "step": 22490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942761039830366e-05,
+      "loss": 1.2589,
+      "step": 22495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942735837514138e-05,
+      "loss": 0.972,
+      "step": 22500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942710629814381e-05,
+      "loss": 1.4848,
+      "step": 22505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9426854167312395e-05,
+      "loss": 1.2532,
+      "step": 22510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942660198264857e-05,
+      "loss": 1.2609,
+      "step": 22515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9426349744153777e-05,
+      "loss": 1.2401,
+      "step": 22520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9426097451829463e-05,
+      "loss": 1.2376,
+      "step": 22525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9425845105677056e-05,
+      "loss": 1.1258,
+      "step": 22530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9425592705698006e-05,
+      "loss": 1.3667,
+      "step": 22535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9425340251893752e-05,
+      "loss": 1.1662,
+      "step": 22540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942508774426574e-05,
+      "loss": 1.229,
+      "step": 22545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9424835182815402e-05,
+      "loss": 1.1742,
+      "step": 22550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942458256754419e-05,
+      "loss": 1.1437,
+      "step": 22555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942432989845354e-05,
+      "loss": 1.284,
+      "step": 22560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9424077175544895e-05,
+      "loss": 1.1598,
+      "step": 22565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9423824398819704e-05,
+      "loss": 1.3264,
+      "step": 22570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9423571568279407e-05,
+      "loss": 1.2542,
+      "step": 22575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9423318683925447e-05,
+      "loss": 1.1217,
+      "step": 22580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9423065745759267e-05,
+      "loss": 1.055,
+      "step": 22585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942281275378232e-05,
+      "loss": 1.2569,
+      "step": 22590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9422559707996037e-05,
+      "loss": 1.0688,
+      "step": 22595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9422306608401874e-05,
+      "loss": 1.3868,
+      "step": 22600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942205345500127e-05,
+      "loss": 1.1784,
+      "step": 22605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9421800247795675e-05,
+      "loss": 1.1863,
+      "step": 22610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9421546986786532e-05,
+      "loss": 1.3337,
+      "step": 22615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.942129367197529e-05,
+      "loss": 1.3478,
+      "step": 22620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9421040303363392e-05,
+      "loss": 1.5642,
+      "step": 22625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9420786880952287e-05,
+      "loss": 1.0744,
+      "step": 22630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9420533404743423e-05,
+      "loss": 1.1082,
+      "step": 22635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9420279874738247e-05,
+      "loss": 0.9444,
+      "step": 22640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9420026290938204e-05,
+      "loss": 1.0549,
+      "step": 22645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9419772653344746e-05,
+      "loss": 1.132,
+      "step": 22650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9419518961959318e-05,
+      "loss": 1.0018,
+      "step": 22655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.941926521678337e-05,
+      "loss": 1.1135,
+      "step": 22660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9419011417818353e-05,
+      "loss": 1.2263,
+      "step": 22665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9418757565065715e-05,
+      "loss": 1.1077,
+      "step": 22670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9418503658526905e-05,
+      "loss": 1.1793,
+      "step": 22675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9418249698203374e-05,
+      "loss": 1.3765,
+      "step": 22680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.941799568409657e-05,
+      "loss": 1.2049,
+      "step": 22685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9417741616207946e-05,
+      "loss": 1.2539,
+      "step": 22690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9417487494538952e-05,
+      "loss": 1.1309,
+      "step": 22695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.941723331909104e-05,
+      "loss": 1.2565,
+      "step": 22700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.941697908986566e-05,
+      "loss": 1.0279,
+      "step": 22705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9416724806864265e-05,
+      "loss": 1.6109,
+      "step": 22710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9416470470088304e-05,
+      "loss": 1.2622,
+      "step": 22715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9416216079539236e-05,
+      "loss": 1.2087,
+      "step": 22720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9415961635218507e-05,
+      "loss": 1.6153,
+      "step": 22725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.941570713712757e-05,
+      "loss": 1.1841,
+      "step": 22730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9415452585267887e-05,
+      "loss": 1.2842,
+      "step": 22735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.94151979796409e-05,
+      "loss": 1.0919,
+      "step": 22740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.941494332024807e-05,
+      "loss": 1.2103,
+      "step": 22745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9414688607090854e-05,
+      "loss": 1.4943,
+      "step": 22750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9414433840170698e-05,
+      "loss": 1.0945,
+      "step": 22755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9414179019489063e-05,
+      "loss": 1.1161,
+      "step": 22760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.94139241450474e-05,
+      "loss": 1.5931,
+      "step": 22765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9413669216847172e-05,
+      "loss": 1.062,
+      "step": 22770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9413414234889823e-05,
+      "loss": 1.1568,
+      "step": 22775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.941315919917682e-05,
+      "loss": 1.0014,
+      "step": 22780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9412904109709614e-05,
+      "loss": 1.2613,
+      "step": 22785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9412648966489664e-05,
+      "loss": 1.288,
+      "step": 22790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9412393769518426e-05,
+      "loss": 1.2064,
+      "step": 22795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9412138518797357e-05,
+      "loss": 0.9328,
+      "step": 22800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.941188321432791e-05,
+      "loss": 1.1937,
+      "step": 22805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.941162785611155e-05,
+      "loss": 1.5265,
+      "step": 22810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9411372444149737e-05,
+      "loss": 1.18,
+      "step": 22815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9411116978443924e-05,
+      "loss": 1.784,
+      "step": 22820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9410861458995567e-05,
+      "loss": 1.3985,
+      "step": 22825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.941060588580613e-05,
+      "loss": 1.2797,
+      "step": 22830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9410350258877077e-05,
+      "loss": 1.003,
+      "step": 22835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.941009457820986e-05,
+      "loss": 1.2124,
+      "step": 22840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.940983884380594e-05,
+      "loss": 0.8415,
+      "step": 22845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.940958305566678e-05,
+      "loss": 1.14,
+      "step": 22850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.940932721379384e-05,
+      "loss": 1.1517,
+      "step": 22855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9409071318188585e-05,
+      "loss": 1.4547,
+      "step": 22860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9408815368852468e-05,
+      "loss": 1.5385,
+      "step": 22865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9408559365786954e-05,
+      "loss": 1.1991,
+      "step": 22870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9408303308993506e-05,
+      "loss": 1.147,
+      "step": 22875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.940804719847359e-05,
+      "loss": 1.4459,
+      "step": 22880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9407791034228655e-05,
+      "loss": 0.993,
+      "step": 22885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.940753481626018e-05,
+      "loss": 1.3206,
+      "step": 22890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.940727854456962e-05,
+      "loss": 1.4343,
+      "step": 22895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.940702221915844e-05,
+      "loss": 1.0362,
+      "step": 22900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9406765840028103e-05,
+      "loss": 1.1553,
+      "step": 22905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9406509407180074e-05,
+      "loss": 1.3572,
+      "step": 22910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9406252920615817e-05,
+      "loss": 1.5309,
+      "step": 22915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9405996380336797e-05,
+      "loss": 1.2172,
+      "step": 22920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.940573978634448e-05,
+      "loss": 1.3347,
+      "step": 22925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.940548313864033e-05,
+      "loss": 1.6079,
+      "step": 22930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9405226437225807e-05,
+      "loss": 1.2439,
+      "step": 22935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9404969682102388e-05,
+      "loss": 0.9821,
+      "step": 22940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.940471287327153e-05,
+      "loss": 1.5628,
+      "step": 22945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9404456010734706e-05,
+      "loss": 1.1141,
+      "step": 22950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.940419909449338e-05,
+      "loss": 1.0886,
+      "step": 22955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.940394212454902e-05,
+      "loss": 1.4075,
+      "step": 22960
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9403685100903085e-05,
+      "loss": 1.6367,
+      "step": 22965
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9403428023557058e-05,
+      "loss": 1.1035,
+      "step": 22970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9403170892512398e-05,
+      "loss": 1.2048,
+      "step": 22975
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9402913707770576e-05,
+      "loss": 1.1821,
+      "step": 22980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9402656469333054e-05,
+      "loss": 1.4509,
+      "step": 22985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.940239917720131e-05,
+      "loss": 1.0307,
+      "step": 22990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9402141831376808e-05,
+      "loss": 1.2749,
+      "step": 22995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9401884431861023e-05,
+      "loss": 1.2735,
+      "step": 23000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9401626978655418e-05,
+      "loss": 1.2525,
+      "step": 23005
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9401369471761467e-05,
+      "loss": 1.3884,
+      "step": 23010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.940111191118064e-05,
+      "loss": 1.3598,
+      "step": 23015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9400854296914406e-05,
+      "loss": 1.6677,
+      "step": 23020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9400596628964238e-05,
+      "loss": 1.3515,
+      "step": 23025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.940033890733161e-05,
+      "loss": 1.1369,
+      "step": 23030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9400081132017988e-05,
+      "loss": 1.3165,
+      "step": 23035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9399823303024847e-05,
+      "loss": 1.4017,
+      "step": 23040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.939956542035366e-05,
+      "loss": 1.212,
+      "step": 23045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9399307484005897e-05,
+      "loss": 1.2158,
+      "step": 23050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9399049493983034e-05,
+      "loss": 1.2159,
+      "step": 23055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9398791450286543e-05,
+      "loss": 1.0635,
+      "step": 23060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9398533352917898e-05,
+      "loss": 1.5731,
+      "step": 23065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9398275201878575e-05,
+      "loss": 1.3708,
+      "step": 23070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9398016997170042e-05,
+      "loss": 1.1946,
+      "step": 23075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.939775873879378e-05,
+      "loss": 0.9944,
+      "step": 23080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.939750042675126e-05,
+      "loss": 1.2036,
+      "step": 23085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9397242061043956e-05,
+      "loss": 1.0729,
+      "step": 23090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9396983641673347e-05,
+      "loss": 1.19,
+      "step": 23095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9396725168640907e-05,
+      "loss": 1.0715,
+      "step": 23100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9396466641948114e-05,
+      "loss": 1.0852,
+      "step": 23105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9396208061596444e-05,
+      "loss": 1.3876,
+      "step": 23110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9395949427587367e-05,
+      "loss": 1.1444,
+      "step": 23115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.939569073992237e-05,
+      "loss": 1.7886,
+      "step": 23120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.939543199860292e-05,
+      "loss": 1.2237,
+      "step": 23125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9395173203630505e-05,
+      "loss": 1.4639,
+      "step": 23130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9394914355006597e-05,
+      "loss": 1.2888,
+      "step": 23135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9394655452732672e-05,
+      "loss": 1.139,
+      "step": 23140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9394396496810212e-05,
+      "loss": 1.4183,
+      "step": 23145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9394137487240696e-05,
+      "loss": 1.4614,
+      "step": 23150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9393878424025602e-05,
+      "loss": 1.3189,
+      "step": 23155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.939361930716641e-05,
+      "loss": 1.2631,
+      "step": 23160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9393360136664602e-05,
+      "loss": 1.1993,
+      "step": 23165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.939310091252165e-05,
+      "loss": 1.1153,
+      "step": 23170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9392841634739047e-05,
+      "loss": 1.3688,
+      "step": 23175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.939258230331826e-05,
+      "loss": 1.1557,
+      "step": 23180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.939232291826078e-05,
+      "loss": 1.3947,
+      "step": 23185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9392063479568078e-05,
+      "loss": 1.1558,
+      "step": 23190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9391803987241648e-05,
+      "loss": 1.3,
+      "step": 23195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.939154444128297e-05,
+      "loss": 1.5055,
+      "step": 23200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.939128484169351e-05,
+      "loss": 1.3617,
+      "step": 23205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.939102518847477e-05,
+      "loss": 1.8183,
+      "step": 23210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9390765481628224e-05,
+      "loss": 1.1299,
+      "step": 23215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9390505721155358e-05,
+      "loss": 1.0453,
+      "step": 23220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.939024590705765e-05,
+      "loss": 1.147,
+      "step": 23225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.938998603933659e-05,
+      "loss": 0.9318,
+      "step": 23230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.938972611799366e-05,
+      "loss": 1.5095,
+      "step": 23235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.938946614303034e-05,
+      "loss": 1.2088,
+      "step": 23240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.938920611444812e-05,
+      "loss": 1.3884,
+      "step": 23245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9388946032248485e-05,
+      "loss": 1.247,
+      "step": 23250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.938868589643292e-05,
+      "loss": 1.0006,
+      "step": 23255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9388425707002904e-05,
+      "loss": 1.2167,
+      "step": 23260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9388165463959928e-05,
+      "loss": 1.2547,
+      "step": 23265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.938790516730548e-05,
+      "loss": 1.4911,
+      "step": 23270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9387644817041043e-05,
+      "loss": 1.4996,
+      "step": 23275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9387384413168106e-05,
+      "loss": 1.1912,
+      "step": 23280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9387123955688154e-05,
+      "loss": 1.2197,
+      "step": 23285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9386863444602672e-05,
+      "loss": 1.096,
+      "step": 23290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9386602879913154e-05,
+      "loss": 1.3381,
+      "step": 23295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9386342261621087e-05,
+      "loss": 1.5635,
+      "step": 23300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9386081589727955e-05,
+      "loss": 0.9918,
+      "step": 23305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.938582086423525e-05,
+      "loss": 1.2372,
+      "step": 23310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.938556008514446e-05,
+      "loss": 1.4364,
+      "step": 23315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.938529925245707e-05,
+      "loss": 1.3534,
+      "step": 23320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9385038366174578e-05,
+      "loss": 1.5288,
+      "step": 23325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.938477742629847e-05,
+      "loss": 1.4925,
+      "step": 23330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9384516432830232e-05,
+      "loss": 1.5703,
+      "step": 23335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9384255385771354e-05,
+      "loss": 1.2279,
+      "step": 23340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9383994285123338e-05,
+      "loss": 1.2662,
+      "step": 23345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9383733130887664e-05,
+      "loss": 1.1637,
+      "step": 23350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9383471923065825e-05,
+      "loss": 1.0137,
+      "step": 23355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9383210661659318e-05,
+      "loss": 1.2088,
+      "step": 23360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.938294934666963e-05,
+      "loss": 1.7693,
+      "step": 23365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9382687978098253e-05,
+      "loss": 1.2877,
+      "step": 23370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9382426555946684e-05,
+      "loss": 1.2557,
+      "step": 23375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9382165080216412e-05,
+      "loss": 1.2589,
+      "step": 23380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9381903550908928e-05,
+      "loss": 1.2946,
+      "step": 23385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9381641968025732e-05,
+      "loss": 1.3758,
+      "step": 23390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9381380331568312e-05,
+      "loss": 1.208,
+      "step": 23395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.938111864153817e-05,
+      "loss": 1.3979,
+      "step": 23400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9380856897936787e-05,
+      "loss": 1.3734,
+      "step": 23405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.938059510076567e-05,
+      "loss": 1.3109,
+      "step": 23410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.938033325002631e-05,
+      "loss": 1.1561,
+      "step": 23415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9380071345720202e-05,
+      "loss": 1.3659,
+      "step": 23420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.937980938784884e-05,
+      "loss": 1.0417,
+      "step": 23425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9379547376413722e-05,
+      "loss": 1.3312,
+      "step": 23430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9379285311416344e-05,
+      "loss": 1.398,
+      "step": 23435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9379023192858202e-05,
+      "loss": 1.1565,
+      "step": 23440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9378761020740792e-05,
+      "loss": 1.4932,
+      "step": 23445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9378498795065615e-05,
+      "loss": 1.3898,
+      "step": 23450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9378236515834163e-05,
+      "loss": 1.06,
+      "step": 23455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.937797418304794e-05,
+      "loss": 1.191,
+      "step": 23460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9377711796708435e-05,
+      "loss": 1.0535,
+      "step": 23465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9377449356817155e-05,
+      "loss": 1.0287,
+      "step": 23470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9377186863375597e-05,
+      "loss": 1.1876,
+      "step": 23475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9376924316385254e-05,
+      "loss": 1.2637,
+      "step": 23480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9376661715847632e-05,
+      "loss": 1.1147,
+      "step": 23485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.937639906176423e-05,
+      "loss": 1.2442,
+      "step": 23490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9376136354136544e-05,
+      "loss": 1.5262,
+      "step": 23495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9375873592966077e-05,
+      "loss": 1.1975,
+      "step": 23500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.937561077825433e-05,
+      "loss": 1.0931,
+      "step": 23505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9375347910002802e-05,
+      "loss": 1.1458,
+      "step": 23510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9375084988212994e-05,
+      "loss": 1.0044,
+      "step": 23515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.937482201288641e-05,
+      "loss": 1.0843,
+      "step": 23520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.937455898402455e-05,
+      "loss": 1.1512,
+      "step": 23525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9374295901628915e-05,
+      "loss": 1.2668,
+      "step": 23530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.937403276570101e-05,
+      "loss": 1.4733,
+      "step": 23535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.937376957624233e-05,
+      "loss": 1.2654,
+      "step": 23540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.937350633325439e-05,
+      "loss": 1.275,
+      "step": 23545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9373243036738686e-05,
+      "loss": 1.2842,
+      "step": 23550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9372979686696724e-05,
+      "loss": 1.2627,
+      "step": 23555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9372716283130004e-05,
+      "loss": 1.3272,
+      "step": 23560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9372452826040035e-05,
+      "loss": 1.2549,
+      "step": 23565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.937218931542832e-05,
+      "loss": 0.8817,
+      "step": 23570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9371925751296362e-05,
+      "loss": 1.1216,
+      "step": 23575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9371662133645667e-05,
+      "loss": 1.3323,
+      "step": 23580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.937139846247774e-05,
+      "loss": 1.0545,
+      "step": 23585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9371134737794087e-05,
+      "loss": 1.6156,
+      "step": 23590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.937087095959622e-05,
+      "loss": 1.4077,
+      "step": 23595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9370607127885634e-05,
+      "loss": 1.6875,
+      "step": 23600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9370343242663844e-05,
+      "loss": 1.3132,
+      "step": 23605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.937007930393235e-05,
+      "loss": 1.3534,
+      "step": 23610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9369815311692668e-05,
+      "loss": 0.9544,
+      "step": 23615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9369551265946296e-05,
+      "loss": 1.0792,
+      "step": 23620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.936928716669475e-05,
+      "loss": 1.0451,
+      "step": 23625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.936902301393954e-05,
+      "loss": 1.6582,
+      "step": 23630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9368758807682163e-05,
+      "loss": 0.9948,
+      "step": 23635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9368494547924132e-05,
+      "loss": 1.5516,
+      "step": 23640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9368230234666962e-05,
+      "loss": 1.1216,
+      "step": 23645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.936796586791216e-05,
+      "loss": 1.2792,
+      "step": 23650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9367701447661233e-05,
+      "loss": 1.2686,
+      "step": 23655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9367436973915693e-05,
+      "loss": 1.1263,
+      "step": 23660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9367172446677048e-05,
+      "loss": 1.4669,
+      "step": 23665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9366907865946812e-05,
+      "loss": 1.0912,
+      "step": 23670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9366643231726493e-05,
+      "loss": 1.4567,
+      "step": 23675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9366378544017605e-05,
+      "loss": 1.4358,
+      "step": 23680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9366113802821656e-05,
+      "loss": 1.1822,
+      "step": 23685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9365849008140157e-05,
+      "loss": 1.1638,
+      "step": 23690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9365584159974626e-05,
+      "loss": 1.6097,
+      "step": 23695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.936531925832657e-05,
+      "loss": 1.311,
+      "step": 23700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9365054303197506e-05,
+      "loss": 1.2069,
+      "step": 23705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9364789294588946e-05,
+      "loss": 1.0243,
+      "step": 23710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.93645242325024e-05,
+      "loss": 1.4521,
+      "step": 23715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9364259116939385e-05,
+      "loss": 1.3956,
+      "step": 23720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9363993947901412e-05,
+      "loss": 1.1763,
+      "step": 23725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.936372872539e-05,
+      "loss": 1.005,
+      "step": 23730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.936346344940666e-05,
+      "loss": 1.4882,
+      "step": 23735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9363198119952905e-05,
+      "loss": 1.3265,
+      "step": 23740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9362932737030254e-05,
+      "loss": 1.2219,
+      "step": 23745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.936266730064022e-05,
+      "loss": 1.1714,
+      "step": 23750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9362401810784324e-05,
+      "loss": 1.0734,
+      "step": 23755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9362136267464074e-05,
+      "loss": 1.441,
+      "step": 23760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9361870670680992e-05,
+      "loss": 1.0699,
+      "step": 23765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9361605020436592e-05,
+      "loss": 1.0278,
+      "step": 23770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9361339316732396e-05,
+      "loss": 1.5958,
+      "step": 23775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9361073559569914e-05,
+      "loss": 1.7805,
+      "step": 23780
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.936080774895067e-05,
+      "loss": 1.2694,
+      "step": 23785
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9360541884876175e-05,
+      "loss": 0.9881,
+      "step": 23790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9360275967347953e-05,
+      "loss": 1.1638,
+      "step": 23795
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.936000999636752e-05,
+      "loss": 1.105,
+      "step": 23800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.93597439719364e-05,
+      "loss": 1.1051,
+      "step": 23805
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9359477894056104e-05,
+      "loss": 1.0097,
+      "step": 23810
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9359211762728157e-05,
+      "loss": 1.1155,
+      "step": 23815
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9358945577954075e-05,
+      "loss": 1.1522,
+      "step": 23820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.935867933973538e-05,
+      "loss": 1.2637,
+      "step": 23825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9358413048073595e-05,
+      "loss": 1.1347,
+      "step": 23830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9358146702970237e-05,
+      "loss": 1.4754,
+      "step": 23835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9357880304426824e-05,
+      "loss": 1.1548,
+      "step": 23840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9357613852444886e-05,
+      "loss": 1.0725,
+      "step": 23845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9357347347025936e-05,
+      "loss": 1.2466,
+      "step": 23850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9357080788171503e-05,
+      "loss": 1.0809,
+      "step": 23855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9356814175883104e-05,
+      "loss": 1.0645,
+      "step": 23860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9356547510162268e-05,
+      "loss": 1.09,
+      "step": 23865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9356280791010508e-05,
+      "loss": 1.2308,
+      "step": 23870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9356014018429354e-05,
+      "loss": 1.2118,
+      "step": 23875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9355747192420326e-05,
+      "loss": 1.2838,
+      "step": 23880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9355480312984952e-05,
+      "loss": 1.277,
+      "step": 23885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9355213380124754e-05,
+      "loss": 1.24,
+      "step": 23890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9354946393841256e-05,
+      "loss": 1.0606,
+      "step": 23895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.935467935413598e-05,
+      "loss": 1.1081,
+      "step": 23900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9354412261010457e-05,
+      "loss": 1.3965,
+      "step": 23905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9354145114466204e-05,
+      "loss": 1.1253,
+      "step": 23910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9353877914504754e-05,
+      "loss": 1.4783,
+      "step": 23915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9353610661127634e-05,
+      "loss": 0.9278,
+      "step": 23920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935334335433636e-05,
+      "loss": 1.0251,
+      "step": 23925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9353075994132468e-05,
+      "loss": 1.2916,
+      "step": 23930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935280858051748e-05,
+      "loss": 1.0031,
+      "step": 23935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9352541113492925e-05,
+      "loss": 1.3319,
+      "step": 23940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935227359306033e-05,
+      "loss": 1.3041,
+      "step": 23945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935200601922122e-05,
+      "loss": 1.1274,
+      "step": 23950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935173839197713e-05,
+      "loss": 1.4511,
+      "step": 23955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935147071132958e-05,
+      "loss": 1.2905,
+      "step": 23960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9351202977280107e-05,
+      "loss": 1.5288,
+      "step": 23965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9350935189830235e-05,
+      "loss": 1.5472,
+      "step": 23970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.935066734898149e-05,
+      "loss": 1.5304,
+      "step": 23975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9350399454735405e-05,
+      "loss": 1.4636,
+      "step": 23980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9350131507093514e-05,
+      "loss": 0.9939,
+      "step": 23985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9349863506057342e-05,
+      "loss": 1.2343,
+      "step": 23990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9349595451628418e-05,
+      "loss": 1.6983,
+      "step": 23995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9349327343808276e-05,
+      "loss": 1.3759,
+      "step": 24000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934905918259845e-05,
+      "loss": 1.1233,
+      "step": 24005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9348790968000463e-05,
+      "loss": 1.5619,
+      "step": 24010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9348522700015854e-05,
+      "loss": 1.4956,
+      "step": 24015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934825437864615e-05,
+      "loss": 0.7905,
+      "step": 24020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934798600389289e-05,
+      "loss": 1.3317,
+      "step": 24025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9347717575757596e-05,
+      "loss": 1.7883,
+      "step": 24030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9347449094241812e-05,
+      "loss": 1.282,
+      "step": 24035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9347180559347062e-05,
+      "loss": 1.1638,
+      "step": 24040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9346911971074888e-05,
+      "loss": 1.0671,
+      "step": 24045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9346643329426817e-05,
+      "loss": 1.333,
+      "step": 24050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9346374634404385e-05,
+      "loss": 1.4079,
+      "step": 24055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934610588600913e-05,
+      "loss": 1.4006,
+      "step": 24060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9345837084242575e-05,
+      "loss": 0.9993,
+      "step": 24065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934556822910627e-05,
+      "loss": 1.449,
+      "step": 24070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9345299320601743e-05,
+      "loss": 1.122,
+      "step": 24075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9345030358730536e-05,
+      "loss": 1.3024,
+      "step": 24080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9344761343494172e-05,
+      "loss": 1.066,
+      "step": 24085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9344492274894198e-05,
+      "loss": 1.4951,
+      "step": 24090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9344223152932144e-05,
+      "loss": 1.0083,
+      "step": 24095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9343953977609552e-05,
+      "loss": 1.3258,
+      "step": 24100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9343684748927957e-05,
+      "loss": 1.2041,
+      "step": 24105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9343415466888897e-05,
+      "loss": 1.1982,
+      "step": 24110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934314613149391e-05,
+      "loss": 1.5086,
+      "step": 24115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934287674274453e-05,
+      "loss": 1.4711,
+      "step": 24120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9342607300642302e-05,
+      "loss": 1.0485,
+      "step": 24125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934233780518876e-05,
+      "loss": 1.3562,
+      "step": 24130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9342068256385446e-05,
+      "loss": 1.1506,
+      "step": 24135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9341798654233896e-05,
+      "loss": 1.0297,
+      "step": 24140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934152899873565e-05,
+      "loss": 1.4607,
+      "step": 24145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934125928989225e-05,
+      "loss": 1.0168,
+      "step": 24150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934098952770524e-05,
+      "loss": 1.0479,
+      "step": 24155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.934071971217615e-05,
+      "loss": 1.2511,
+      "step": 24160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9340449843306527e-05,
+      "loss": 1.3768,
+      "step": 24165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9340179921097916e-05,
+      "loss": 2.0423,
+      "step": 24170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933990994555185e-05,
+      "loss": 1.5481,
+      "step": 24175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9339639916669874e-05,
+      "loss": 1.4792,
+      "step": 24180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9339369834453534e-05,
+      "loss": 1.5635,
+      "step": 24185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9339099698904367e-05,
+      "loss": 1.1147,
+      "step": 24190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933882951002392e-05,
+      "loss": 1.5713,
+      "step": 24195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933855926781373e-05,
+      "loss": 2.0757,
+      "step": 24200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9338288972275342e-05,
+      "loss": 1.2438,
+      "step": 24205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9338018623410306e-05,
+      "loss": 1.06,
+      "step": 24210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933774822122016e-05,
+      "loss": 1.4284,
+      "step": 24215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933747776570645e-05,
+      "loss": 1.2378,
+      "step": 24220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9337207256870718e-05,
+      "loss": 1.5966,
+      "step": 24225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9336936694714512e-05,
+      "loss": 1.238,
+      "step": 24230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9336666079239378e-05,
+      "loss": 1.4304,
+      "step": 24235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9336395410446853e-05,
+      "loss": 0.8846,
+      "step": 24240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9336124688338494e-05,
+      "loss": 1.2733,
+      "step": 24245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9335853912915838e-05,
+      "loss": 1.2915,
+      "step": 24250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9335583084180433e-05,
+      "loss": 1.4126,
+      "step": 24255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9335312202133832e-05,
+      "loss": 1.2875,
+      "step": 24260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9335041266777576e-05,
+      "loss": 1.1082,
+      "step": 24265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9334770278113213e-05,
+      "loss": 1.5585,
+      "step": 24270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9334499236142293e-05,
+      "loss": 1.3725,
+      "step": 24275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933422814086636e-05,
+      "loss": 1.1697,
+      "step": 24280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9333956992286963e-05,
+      "loss": 1.2221,
+      "step": 24285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933368579040565e-05,
+      "loss": 1.0758,
+      "step": 24290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9333414535223977e-05,
+      "loss": 1.2674,
+      "step": 24295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933314322674348e-05,
+      "loss": 1.3052,
+      "step": 24300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9332871864965722e-05,
+      "loss": 1.1225,
+      "step": 24305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933260044989224e-05,
+      "loss": 1.7253,
+      "step": 24310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9332328981524592e-05,
+      "loss": 1.3352,
+      "step": 24315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9332057459864328e-05,
+      "loss": 1.1988,
+      "step": 24320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933178588491299e-05,
+      "loss": 0.8643,
+      "step": 24325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9331514256672146e-05,
+      "loss": 1.1036,
+      "step": 24330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933124257514333e-05,
+      "loss": 1.2451,
+      "step": 24335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9330970840328102e-05,
+      "loss": 1.2164,
+      "step": 24340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.933069905222801e-05,
+      "loss": 1.8112,
+      "step": 24345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9330427210844608e-05,
+      "loss": 1.3401,
+      "step": 24350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9330155316179453e-05,
+      "loss": 1.0945,
+      "step": 24355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9329883368234085e-05,
+      "loss": 1.4154,
+      "step": 24360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932961136701007e-05,
+      "loss": 1.4046,
+      "step": 24365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9329339312508955e-05,
+      "loss": 1.2371,
+      "step": 24370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9329067204732296e-05,
+      "loss": 1.3248,
+      "step": 24375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932879504368164e-05,
+      "loss": 1.0585,
+      "step": 24380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9328522829358554e-05,
+      "loss": 1.4206,
+      "step": 24385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9328250561764583e-05,
+      "loss": 1.0721,
+      "step": 24390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932797824090128e-05,
+      "loss": 1.1719,
+      "step": 24395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932770586677021e-05,
+      "loss": 1.459,
+      "step": 24400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932743343937292e-05,
+      "loss": 1.0753,
+      "step": 24405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9327160958710973e-05,
+      "loss": 1.2782,
+      "step": 24410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9326888424785912e-05,
+      "loss": 1.1059,
+      "step": 24415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9326615837599307e-05,
+      "loss": 1.156,
+      "step": 24420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9326343197152705e-05,
+      "loss": 1.095,
+      "step": 24425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9326070503447672e-05,
+      "loss": 1.2023,
+      "step": 24430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9325797756485758e-05,
+      "loss": 1.3132,
+      "step": 24435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932552495626852e-05,
+      "loss": 1.2001,
+      "step": 24440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9325252102797526e-05,
+      "loss": 1.3569,
+      "step": 24445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9324979196074322e-05,
+      "loss": 1.3261,
+      "step": 24450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932470623610047e-05,
+      "loss": 1.2325,
+      "step": 24455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932443322287753e-05,
+      "loss": 1.3212,
+      "step": 24460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9324160156407067e-05,
+      "loss": 1.2679,
+      "step": 24465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9323887036690632e-05,
+      "loss": 1.1037,
+      "step": 24470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9323613863729784e-05,
+      "loss": 1.4973,
+      "step": 24475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9323340637526088e-05,
+      "loss": 1.333,
+      "step": 24480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9323067358081103e-05,
+      "loss": 1.2413,
+      "step": 24485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932279402539639e-05,
+      "loss": 1.0905,
+      "step": 24490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9322520639473505e-05,
+      "loss": 1.5985,
+      "step": 24495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9322247200314013e-05,
+      "loss": 1.2912,
+      "step": 24500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932197370791948e-05,
+      "loss": 1.0901,
+      "step": 24505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932170016229146e-05,
+      "loss": 1.398,
+      "step": 24510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932142656343152e-05,
+      "loss": 1.1879,
+      "step": 24515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.932115291134122e-05,
+      "loss": 1.2071,
+      "step": 24520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9320879206022125e-05,
+      "loss": 1.2062,
+      "step": 24525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9320605447475796e-05,
+      "loss": 1.1224,
+      "step": 24530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9320331635703794e-05,
+      "loss": 1.1631,
+      "step": 24535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9320057770707687e-05,
+      "loss": 1.2672,
+      "step": 24540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9319783852489037e-05,
+      "loss": 1.3981,
+      "step": 24545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931950988104941e-05,
+      "loss": 1.414,
+      "step": 24550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931923585639037e-05,
+      "loss": 1.0436,
+      "step": 24555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9318961778513477e-05,
+      "loss": 0.9879,
+      "step": 24560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9318687647420302e-05,
+      "loss": 1.2412,
+      "step": 24565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9318413463112408e-05,
+      "loss": 1.393,
+      "step": 24570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9318139225591363e-05,
+      "loss": 1.3552,
+      "step": 24575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931786493485873e-05,
+      "loss": 1.1193,
+      "step": 24580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9317590590916075e-05,
+      "loss": 1.212,
+      "step": 24585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931731619376497e-05,
+      "loss": 1.0333,
+      "step": 24590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9317041743406974e-05,
+      "loss": 1.035,
+      "step": 24595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931676723984366e-05,
+      "loss": 1.5412,
+      "step": 24600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9316492683076597e-05,
+      "loss": 1.2212,
+      "step": 24605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931621807310735e-05,
+      "loss": 1.0916,
+      "step": 24610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9315943409937482e-05,
+      "loss": 1.1718,
+      "step": 24615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9315668693568566e-05,
+      "loss": 1.2213,
+      "step": 24620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9315393924002176e-05,
+      "loss": 1.1098,
+      "step": 24625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9315119101239874e-05,
+      "loss": 1.3064,
+      "step": 24630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9314844225283233e-05,
+      "loss": 1.2211,
+      "step": 24635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931456929613382e-05,
+      "loss": 1.2624,
+      "step": 24640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9314294313793207e-05,
+      "loss": 1.1115,
+      "step": 24645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9314019278262965e-05,
+      "loss": 1.2489,
+      "step": 24650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931374418954466e-05,
+      "loss": 1.3404,
+      "step": 24655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931346904763987e-05,
+      "loss": 1.1589,
+      "step": 24660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931319385255016e-05,
+      "loss": 1.3941,
+      "step": 24665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9312918604277108e-05,
+      "loss": 1.4049,
+      "step": 24670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9312643302822274e-05,
+      "loss": 1.0756,
+      "step": 24675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9312367948187246e-05,
+      "loss": 1.147,
+      "step": 24680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9312092540373585e-05,
+      "loss": 1.3639,
+      "step": 24685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9311817079382867e-05,
+      "loss": 1.2057,
+      "step": 24690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9311541565216668e-05,
+      "loss": 1.2053,
+      "step": 24695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9311265997876552e-05,
+      "loss": 1.0777,
+      "step": 24700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9310990377364103e-05,
+      "loss": 1.2622,
+      "step": 24705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9310714703680894e-05,
+      "loss": 1.0155,
+      "step": 24710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931043897682849e-05,
+      "loss": 1.2991,
+      "step": 24715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9310163196808477e-05,
+      "loss": 0.9162,
+      "step": 24720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9309887363622422e-05,
+      "loss": 1.2715,
+      "step": 24725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93096114772719e-05,
+      "loss": 1.2537,
+      "step": 24730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9309335537758495e-05,
+      "loss": 1.3277,
+      "step": 24735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9309059545083776e-05,
+      "loss": 1.1536,
+      "step": 24740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9308783499249317e-05,
+      "loss": 1.0311,
+      "step": 24745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.93085074002567e-05,
+      "loss": 1.0089,
+      "step": 24750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9308231248107498e-05,
+      "loss": 1.339,
+      "step": 24755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.930795504280329e-05,
+      "loss": 1.0595,
+      "step": 24760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9307678784345653e-05,
+      "loss": 1.218,
+      "step": 24765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9307402472736162e-05,
+      "loss": 1.1179,
+      "step": 24770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9307126107976397e-05,
+      "loss": 1.2676,
+      "step": 24775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9306849690067937e-05,
+      "loss": 1.3955,
+      "step": 24780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9306573219012358e-05,
+      "loss": 1.4228,
+      "step": 24785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.930629669481124e-05,
+      "loss": 1.1889,
+      "step": 24790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9306020117466166e-05,
+      "loss": 1.2579,
+      "step": 24795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.930574348697871e-05,
+      "loss": 1.5908,
+      "step": 24800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9305466803350456e-05,
+      "loss": 1.121,
+      "step": 24805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.930519006658298e-05,
+      "loss": 1.0791,
+      "step": 24810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.930491327667786e-05,
+      "loss": 1.1441,
+      "step": 24815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9304636433636684e-05,
+      "loss": 1.0382,
+      "step": 24820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.930435953746103e-05,
+      "loss": 1.4582,
+      "step": 24825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9304082588152476e-05,
+      "loss": 1.2033,
+      "step": 24830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.930380558571261e-05,
+      "loss": 1.5857,
+      "step": 24835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9303528530143006e-05,
+      "loss": 1.5352,
+      "step": 24840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9303251421445252e-05,
+      "loss": 1.0033,
+      "step": 24845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.930297425962093e-05,
+      "loss": 1.0953,
+      "step": 24850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9302697044671622e-05,
+      "loss": 1.4777,
+      "step": 24855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9302419776598907e-05,
+      "loss": 1.5954,
+      "step": 24860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9302142455404375e-05,
+      "loss": 1.0838,
+      "step": 24865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9301865081089605e-05,
+      "loss": 1.2388,
+      "step": 24870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.930158765365618e-05,
+      "loss": 1.1704,
+      "step": 24875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9301310173105686e-05,
+      "loss": 1.2801,
+      "step": 24880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.930103263943971e-05,
+      "loss": 1.4502,
+      "step": 24885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.930075505265984e-05,
+      "loss": 1.2874,
+      "step": 24890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9300477412767652e-05,
+      "loss": 1.2808,
+      "step": 24895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9300199719764735e-05,
+      "loss": 1.1838,
+      "step": 24900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9299921973652676e-05,
+      "loss": 1.4047,
+      "step": 24905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.929964417443306e-05,
+      "loss": 1.03,
+      "step": 24910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9299366322107477e-05,
+      "loss": 1.292,
+      "step": 24915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9299088416677506e-05,
+      "loss": 1.2743,
+      "step": 24920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9298810458144742e-05,
+      "loss": 1.0449,
+      "step": 24925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9298532446510768e-05,
+      "loss": 1.1161,
+      "step": 24930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9298254381777172e-05,
+      "loss": 1.2096,
+      "step": 24935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.929797626394554e-05,
+      "loss": 0.7867,
+      "step": 24940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9297698093017466e-05,
+      "loss": 0.8586,
+      "step": 24945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9297419868994534e-05,
+      "loss": 1.4984,
+      "step": 24950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9297141591878333e-05,
+      "loss": 1.3165,
+      "step": 24955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9296863261670457e-05,
+      "loss": 1.289,
+      "step": 24960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.929658487837249e-05,
+      "loss": 1.3441,
+      "step": 24965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.929630644198602e-05,
+      "loss": 1.2099,
+      "step": 24970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9296027952512642e-05,
+      "loss": 1.2457,
+      "step": 24975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9295749409953944e-05,
+      "loss": 1.5441,
+      "step": 24980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.929547081431152e-05,
+      "loss": 1.3178,
+      "step": 24985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9295192165586955e-05,
+      "loss": 1.2312,
+      "step": 24990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9294913463781844e-05,
+      "loss": 1.1993,
+      "step": 24995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9294634708897777e-05,
+      "loss": 1.1798,
+      "step": 25000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.929435590093635e-05,
+      "loss": 1.1779,
+      "step": 25005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9294077039899147e-05,
+      "loss": 1.1561,
+      "step": 25010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9293798125787767e-05,
+      "loss": 1.2881,
+      "step": 25015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9293519158603802e-05,
+      "loss": 1.342,
+      "step": 25020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9293240138348842e-05,
+      "loss": 1.2444,
+      "step": 25025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9292961065024486e-05,
+      "loss": 1.1136,
+      "step": 25030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9292681938632324e-05,
+      "loss": 0.9378,
+      "step": 25035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9292402759173947e-05,
+      "loss": 0.9968,
+      "step": 25040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9292123526650952e-05,
+      "loss": 1.2587,
+      "step": 25045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9291844241064937e-05,
+      "loss": 1.1931,
+      "step": 25050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9291564902417492e-05,
+      "loss": 1.0008,
+      "step": 25055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9291285510710213e-05,
+      "loss": 1.4274,
+      "step": 25060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9291006065944698e-05,
+      "loss": 1.317,
+      "step": 25065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.929072656812254e-05,
+      "loss": 1.3654,
+      "step": 25070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.929044701724534e-05,
+      "loss": 1.5906,
+      "step": 25075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9290167413314687e-05,
+      "loss": 1.501,
+      "step": 25080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9289887756332182e-05,
+      "loss": 1.309,
+      "step": 25085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9289608046299418e-05,
+      "loss": 1.1493,
+      "step": 25090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9289328283218e-05,
+      "loss": 1.1237,
+      "step": 25095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.928904846708952e-05,
+      "loss": 0.9698,
+      "step": 25100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.928876859791558e-05,
+      "loss": 0.9155,
+      "step": 25105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.928848867569777e-05,
+      "loss": 1.1855,
+      "step": 25110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9288208700437695e-05,
+      "loss": 1.4062,
+      "step": 25115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9287928672136956e-05,
+      "loss": 1.1436,
+      "step": 25120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.928764859079715e-05,
+      "loss": 1.0317,
+      "step": 25125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.928736845641987e-05,
+      "loss": 1.6018,
+      "step": 25130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.928708826900672e-05,
+      "loss": 0.9204,
+      "step": 25135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9286808028559306e-05,
+      "loss": 1.796,
+      "step": 25140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.928652773507922e-05,
+      "loss": 0.9618,
+      "step": 25145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9286247388568067e-05,
+      "loss": 1.4395,
+      "step": 25150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9285966989027445e-05,
+      "loss": 1.5354,
+      "step": 25155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.928568653645896e-05,
+      "loss": 1.3158,
+      "step": 25160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.928540603086421e-05,
+      "loss": 1.4158,
+      "step": 25165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9285125472244796e-05,
+      "loss": 1.4114,
+      "step": 25170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.928484486060232e-05,
+      "loss": 1.007,
+      "step": 25175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.928456419593839e-05,
+      "loss": 1.2544,
+      "step": 25180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9284283478254603e-05,
+      "loss": 1.4816,
+      "step": 25185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9284002707552563e-05,
+      "loss": 1.63,
+      "step": 25190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9283721883833873e-05,
+      "loss": 0.9915,
+      "step": 25195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.928344100710014e-05,
+      "loss": 1.1612,
+      "step": 25200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9283160077352966e-05,
+      "loss": 1.2339,
+      "step": 25205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9282879094593953e-05,
+      "loss": 1.3871,
+      "step": 25210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.928259805882471e-05,
+      "loss": 1.2519,
+      "step": 25215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9282316970046838e-05,
+      "loss": 1.3702,
+      "step": 25220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9282035828261943e-05,
+      "loss": 1.1018,
+      "step": 25225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.928175463347163e-05,
+      "loss": 1.3701,
+      "step": 25230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.928147338567751e-05,
+      "loss": 1.2789,
+      "step": 25235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.928119208488118e-05,
+      "loss": 0.9869,
+      "step": 25240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9280910731084255e-05,
+      "loss": 1.2919,
+      "step": 25245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9280629324288335e-05,
+      "loss": 1.2336,
+      "step": 25250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9280347864495034e-05,
+      "loss": 1.0532,
+      "step": 25255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9280066351705954e-05,
+      "loss": 1.2199,
+      "step": 25260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.92797847859227e-05,
+      "loss": 1.4409,
+      "step": 25265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9279503167146887e-05,
+      "loss": 0.8491,
+      "step": 25270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9279221495380122e-05,
+      "loss": 1.3011,
+      "step": 25275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9278939770624005e-05,
+      "loss": 1.1521,
+      "step": 25280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9278657992880157e-05,
+      "loss": 1.1699,
+      "step": 25285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9278376162150176e-05,
+      "loss": 1.0813,
+      "step": 25290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9278094278435682e-05,
+      "loss": 1.3441,
+      "step": 25295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9277812341738275e-05,
+      "loss": 1.1349,
+      "step": 25300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9277530352059573e-05,
+      "loss": 1.0966,
+      "step": 25305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9277248309401183e-05,
+      "loss": 1.5228,
+      "step": 25310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9276966213764714e-05,
+      "loss": 1.1194,
+      "step": 25315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.927668406515178e-05,
+      "loss": 1.1763,
+      "step": 25320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9276401863563988e-05,
+      "loss": 1.4227,
+      "step": 25325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9276119609002952e-05,
+      "loss": 1.2193,
+      "step": 25330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9275837301470285e-05,
+      "loss": 1.1393,
+      "step": 25335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9275554940967596e-05,
+      "loss": 1.2296,
+      "step": 25340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9275272527496502e-05,
+      "loss": 1.177,
+      "step": 25345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.927499006105861e-05,
+      "loss": 0.9913,
+      "step": 25350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.927470754165554e-05,
+      "loss": 1.6427,
+      "step": 25355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9274424969288897e-05,
+      "loss": 1.1396,
+      "step": 25360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9274142343960304e-05,
+      "loss": 1.3343,
+      "step": 25365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.927385966567137e-05,
+      "loss": 1.3374,
+      "step": 25370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9273576934423707e-05,
+      "loss": 1.4179,
+      "step": 25375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9273294150218933e-05,
+      "loss": 1.2404,
+      "step": 25380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9273011313058658e-05,
+      "loss": 1.5644,
+      "step": 25385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9272728422944505e-05,
+      "loss": 1.9119,
+      "step": 25390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9272445479878085e-05,
+      "loss": 1.3735,
+      "step": 25395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9272162483861013e-05,
+      "loss": 1.2572,
+      "step": 25400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9271879434894908e-05,
+      "loss": 1.2307,
+      "step": 25405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.927159633298138e-05,
+      "loss": 1.6168,
+      "step": 25410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9271313178122056e-05,
+      "loss": 1.2368,
+      "step": 25415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9271029970318544e-05,
+      "loss": 1.47,
+      "step": 25420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9270746709572464e-05,
+      "loss": 1.7075,
+      "step": 25425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9270463395885436e-05,
+      "loss": 1.3035,
+      "step": 25430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9270180029259072e-05,
+      "loss": 1.1017,
+      "step": 25435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9269896609695e-05,
+      "loss": 1.3109,
+      "step": 25440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.926961313719483e-05,
+      "loss": 0.9848,
+      "step": 25445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.926932961176018e-05,
+      "loss": 1.2234,
+      "step": 25450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9269046033392675e-05,
+      "loss": 1.2691,
+      "step": 25455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9268762402093933e-05,
+      "loss": 1.0164,
+      "step": 25460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.926847871786557e-05,
+      "loss": 0.9494,
+      "step": 25465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.926819498070921e-05,
+      "loss": 1.0124,
+      "step": 25470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.926791119062647e-05,
+      "loss": 0.9233,
+      "step": 25475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9267627347618973e-05,
+      "loss": 1.1362,
+      "step": 25480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9267343451688344e-05,
+      "loss": 1.2403,
+      "step": 25485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.926705950283619e-05,
+      "loss": 1.5046,
+      "step": 25490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.926677550106415e-05,
+      "loss": 1.2927,
+      "step": 25495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9266491446373835e-05,
+      "loss": 1.3764,
+      "step": 25500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9266207338766868e-05,
+      "loss": 1.2053,
+      "step": 25505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9265923178244878e-05,
+      "loss": 1.2193,
+      "step": 25510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.926563896480948e-05,
+      "loss": 1.4636,
+      "step": 25515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.92653546984623e-05,
+      "loss": 1.2815,
+      "step": 25520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9265070379204962e-05,
+      "loss": 1.1203,
+      "step": 25525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.926478600703909e-05,
+      "loss": 1.5492,
+      "step": 25530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9264501581966303e-05,
+      "loss": 1.2689,
+      "step": 25535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.926421710398823e-05,
+      "loss": 1.0791,
+      "step": 25540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9263932573106495e-05,
+      "loss": 1.289,
+      "step": 25545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9263647989322724e-05,
+      "loss": 1.2656,
+      "step": 25550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9263363352638537e-05,
+      "loss": 1.2465,
+      "step": 25555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9263078663055567e-05,
+      "loss": 1.3608,
+      "step": 25560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9262793920575436e-05,
+      "loss": 1.3062,
+      "step": 25565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9262509125199766e-05,
+      "loss": 1.1685,
+      "step": 25570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.926222427693019e-05,
+      "loss": 1.0487,
+      "step": 25575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.926193937576833e-05,
+      "loss": 1.2023,
+      "step": 25580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9261654421715813e-05,
+      "loss": 1.4121,
+      "step": 25585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9261369414774274e-05,
+      "loss": 0.9903,
+      "step": 25590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.926108435494533e-05,
+      "loss": 1.5532,
+      "step": 25595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9260799242230614e-05,
+      "loss": 1.0608,
+      "step": 25600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.926051407663175e-05,
+      "loss": 1.0238,
+      "step": 25605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9260228858150372e-05,
+      "loss": 1.2753,
+      "step": 25610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.925994358678811e-05,
+      "loss": 1.1829,
+      "step": 25615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9259658262546586e-05,
+      "loss": 0.9419,
+      "step": 25620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9259372885427434e-05,
+      "loss": 1.139,
+      "step": 25625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9259087455432283e-05,
+      "loss": 1.211,
+      "step": 25630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.925880197256276e-05,
+      "loss": 1.408,
+      "step": 25635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.92585164368205e-05,
+      "loss": 1.1423,
+      "step": 25640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9258230848207135e-05,
+      "loss": 1.0203,
+      "step": 25645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9257945206724287e-05,
+      "loss": 1.1738,
+      "step": 25650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9257659512373596e-05,
+      "loss": 1.0206,
+      "step": 25655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.925737376515669e-05,
+      "loss": 1.1175,
+      "step": 25660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9257087965075195e-05,
+      "loss": 0.9767,
+      "step": 25665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9256802112130754e-05,
+      "loss": 1.1078,
+      "step": 25670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9256516206324993e-05,
+      "loss": 1.3037,
+      "step": 25675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9256230247659548e-05,
+      "loss": 1.3261,
+      "step": 25680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9255944236136045e-05,
+      "loss": 1.14,
+      "step": 25685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9255658171756124e-05,
+      "loss": 1.3488,
+      "step": 25690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.925537205452142e-05,
+      "loss": 1.2439,
+      "step": 25695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.925508588443356e-05,
+      "loss": 1.1575,
+      "step": 25700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.925479966149418e-05,
+      "loss": 1.2656,
+      "step": 25705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9254513385704922e-05,
+      "loss": 1.2399,
+      "step": 25710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9254227057067408e-05,
+      "loss": 1.1095,
+      "step": 25715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9253940675583285e-05,
+      "loss": 1.1651,
+      "step": 25720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9253654241254183e-05,
+      "loss": 1.0402,
+      "step": 25725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9253367754081738e-05,
+      "loss": 1.105,
+      "step": 25730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9253081214067586e-05,
+      "loss": 1.2484,
+      "step": 25735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.925279462121336e-05,
+      "loss": 1.3393,
+      "step": 25740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9252507975520705e-05,
+      "loss": 1.1237,
+      "step": 25745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9252221276991252e-05,
+      "loss": 1.2784,
+      "step": 25750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.925193452562664e-05,
+      "loss": 0.987,
+      "step": 25755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9251647721428504e-05,
+      "loss": 1.3794,
+      "step": 25760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9251360864398482e-05,
+      "loss": 1.2227,
+      "step": 25765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9251073954538215e-05,
+      "loss": 1.2399,
+      "step": 25770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.925078699184934e-05,
+      "loss": 1.1729,
+      "step": 25775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9250499976333498e-05,
+      "loss": 1.5688,
+      "step": 25780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9250212907992323e-05,
+      "loss": 1.4074,
+      "step": 25785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9249925786827457e-05,
+      "loss": 1.265,
+      "step": 25790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.924963861284054e-05,
+      "loss": 1.2338,
+      "step": 25795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9249351386033214e-05,
+      "loss": 1.3394,
+      "step": 25800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9249064106407115e-05,
+      "loss": 1.2392,
+      "step": 25805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9248776773963883e-05,
+      "loss": 1.2764,
+      "step": 25810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9248489388705165e-05,
+      "loss": 1.1561,
+      "step": 25815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9248201950632594e-05,
+      "loss": 1.4154,
+      "step": 25820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.924791445974782e-05,
+      "loss": 1.0786,
+      "step": 25825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9247626916052478e-05,
+      "loss": 1.0868,
+      "step": 25830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9247339319548212e-05,
+      "loss": 1.1069,
+      "step": 25835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9247051670236666e-05,
+      "loss": 0.8755,
+      "step": 25840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.924676396811948e-05,
+      "loss": 1.0959,
+      "step": 25845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9246476213198302e-05,
+      "loss": 1.0545,
+      "step": 25850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.924618840547477e-05,
+      "loss": 1.4367,
+      "step": 25855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.924590054495053e-05,
+      "loss": 1.1485,
+      "step": 25860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.924561263162722e-05,
+      "loss": 1.373,
+      "step": 25865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9245324665506495e-05,
+      "loss": 1.5318,
+      "step": 25870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9245036646589988e-05,
+      "loss": 0.9539,
+      "step": 25875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9244748574879354e-05,
+      "loss": 1.326,
+      "step": 25880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9244460450376233e-05,
+      "loss": 1.18,
+      "step": 25885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9244172273082268e-05,
+      "loss": 1.088,
+      "step": 25890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9243884042999106e-05,
+      "loss": 1.3453,
+      "step": 25895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9243595760128396e-05,
+      "loss": 1.1299,
+      "step": 25900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9243307424471783e-05,
+      "loss": 1.3892,
+      "step": 25905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9243019036030915e-05,
+      "loss": 1.1961,
+      "step": 25910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9242730594807434e-05,
+      "loss": 0.9985,
+      "step": 25915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9242442100802994e-05,
+      "loss": 1.2108,
+      "step": 25920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9242153554019234e-05,
+      "loss": 1.0776,
+      "step": 25925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.924186495445781e-05,
+      "loss": 1.153,
+      "step": 25930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9241576302120365e-05,
+      "loss": 1.1433,
+      "step": 25935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.924128759700855e-05,
+      "loss": 1.1442,
+      "step": 25940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9240998839124006e-05,
+      "loss": 1.2397,
+      "step": 25945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9240710028468395e-05,
+      "loss": 1.3592,
+      "step": 25950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9240421165043357e-05,
+      "loss": 1.1172,
+      "step": 25955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9240132248850548e-05,
+      "loss": 1.2957,
+      "step": 25960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9239843279891608e-05,
+      "loss": 1.215,
+      "step": 25965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9239554258168197e-05,
+      "loss": 1.1313,
+      "step": 25970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.923926518368196e-05,
+      "loss": 1.2282,
+      "step": 25975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.923897605643455e-05,
+      "loss": 1.0629,
+      "step": 25980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9238686876427617e-05,
+      "loss": 1.5274,
+      "step": 25985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9238397643662817e-05,
+      "loss": 1.3011,
+      "step": 25990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9238108358141793e-05,
+      "loss": 1.5854,
+      "step": 25995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92378190198662e-05,
+      "loss": 1.1595,
+      "step": 26000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9237529628837696e-05,
+      "loss": 1.3295,
+      "step": 26005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9237240185057928e-05,
+      "loss": 1.0387,
+      "step": 26010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923695068852855e-05,
+      "loss": 1.2519,
+      "step": 26015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9236661139251215e-05,
+      "loss": 0.9544,
+      "step": 26020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9236371537227575e-05,
+      "loss": 1.1039,
+      "step": 26025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923608188245929e-05,
+      "loss": 1.5505,
+      "step": 26030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923579217494801e-05,
+      "loss": 1.2545,
+      "step": 26035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9235502414695386e-05,
+      "loss": 1.3145,
+      "step": 26040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9235212601703076e-05,
+      "loss": 1.1539,
+      "step": 26045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9234922735972736e-05,
+      "loss": 1.0013,
+      "step": 26050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923463281750602e-05,
+      "loss": 1.1935,
+      "step": 26055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9234342846304585e-05,
+      "loss": 1.0939,
+      "step": 26060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9234052822370083e-05,
+      "loss": 1.0792,
+      "step": 26065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9233762745704173e-05,
+      "loss": 1.3026,
+      "step": 26070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9233472616308516e-05,
+      "loss": 1.4397,
+      "step": 26075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9233182434184757e-05,
+      "loss": 1.3114,
+      "step": 26080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9232892199334566e-05,
+      "loss": 1.1694,
+      "step": 26085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9232601911759592e-05,
+      "loss": 1.0237,
+      "step": 26090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9232311571461497e-05,
+      "loss": 1.4062,
+      "step": 26095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9232021178441936e-05,
+      "loss": 1.0929,
+      "step": 26100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923173073270257e-05,
+      "loss": 1.358,
+      "step": 26105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9231440234245054e-05,
+      "loss": 0.9796,
+      "step": 26110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.923114968307105e-05,
+      "loss": 1.3844,
+      "step": 26115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9230859079182216e-05,
+      "loss": 1.0557,
+      "step": 26120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9230568422580212e-05,
+      "loss": 1.4167,
+      "step": 26125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9230277713266697e-05,
+      "loss": 1.2207,
+      "step": 26130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9229986951243335e-05,
+      "loss": 1.8705,
+      "step": 26135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922969613651178e-05,
+      "loss": 1.3265,
+      "step": 26140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9229405269073697e-05,
+      "loss": 1.3616,
+      "step": 26145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9229114348930747e-05,
+      "loss": 1.1625,
+      "step": 26150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9228823376084588e-05,
+      "loss": 1.2056,
+      "step": 26155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9228532350536882e-05,
+      "loss": 1.0659,
+      "step": 26160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9228241272289297e-05,
+      "loss": 1.2159,
+      "step": 26165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9227950141343486e-05,
+      "loss": 1.5378,
+      "step": 26170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922765895770112e-05,
+      "loss": 1.2394,
+      "step": 26175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9227367721363857e-05,
+      "loss": 1.1872,
+      "step": 26180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9227076432333364e-05,
+      "loss": 1.1722,
+      "step": 26185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9226785090611298e-05,
+      "loss": 1.4043,
+      "step": 26190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9226493696199328e-05,
+      "loss": 1.201,
+      "step": 26195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922620224909912e-05,
+      "loss": 1.0771,
+      "step": 26200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922591074931233e-05,
+      "loss": 1.1767,
+      "step": 26205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9225619196840628e-05,
+      "loss": 1.2618,
+      "step": 26210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922532759168568e-05,
+      "loss": 1.1561,
+      "step": 26215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922503593384915e-05,
+      "loss": 1.1657,
+      "step": 26220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92247442233327e-05,
+      "loss": 1.2185,
+      "step": 26225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9224452460138005e-05,
+      "loss": 1.4158,
+      "step": 26230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922416064426672e-05,
+      "loss": 1.6176,
+      "step": 26235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9223868775720515e-05,
+      "loss": 1.2349,
+      "step": 26240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922357685450106e-05,
+      "loss": 1.4246,
+      "step": 26245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9223284880610023e-05,
+      "loss": 1.1029,
+      "step": 26250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9222992854049067e-05,
+      "loss": 1.0779,
+      "step": 26255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922270077481986e-05,
+      "loss": 1.1779,
+      "step": 26260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922240864292407e-05,
+      "loss": 1.0733,
+      "step": 26265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922211645836337e-05,
+      "loss": 1.1251,
+      "step": 26270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9221824221139423e-05,
+      "loss": 1.0798,
+      "step": 26275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92215319312539e-05,
+      "loss": 1.3198,
+      "step": 26280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922123958870847e-05,
+      "loss": 1.4634,
+      "step": 26285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92209471935048e-05,
+      "loss": 1.2208,
+      "step": 26290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922065474564457e-05,
+      "loss": 1.1399,
+      "step": 26295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9220362245129433e-05,
+      "loss": 1.1787,
+      "step": 26300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9220069691961074e-05,
+      "loss": 1.4601,
+      "step": 26305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9219777086141152e-05,
+      "loss": 1.4722,
+      "step": 26310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9219484427671347e-05,
+      "loss": 1.372,
+      "step": 26315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9219191716553328e-05,
+      "loss": 1.0195,
+      "step": 26320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9218898952788766e-05,
+      "loss": 1.2306,
+      "step": 26325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921860613637933e-05,
+      "loss": 1.4222,
+      "step": 26330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92183132673267e-05,
+      "loss": 1.4297,
+      "step": 26335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921802034563254e-05,
+      "loss": 1.4616,
+      "step": 26340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9217727371298523e-05,
+      "loss": 1.4694,
+      "step": 26345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921743434432633e-05,
+      "loss": 1.1084,
+      "step": 26350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9217141264717625e-05,
+      "loss": 1.0878,
+      "step": 26355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9216848132474086e-05,
+      "loss": 1.1851,
+      "step": 26360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921655494759739e-05,
+      "loss": 1.191,
+      "step": 26365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9216261710089208e-05,
+      "loss": 1.3403,
+      "step": 26370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9215968419951213e-05,
+      "loss": 1.3298,
+      "step": 26375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9215675077185082e-05,
+      "loss": 1.2342,
+      "step": 26380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921538168179249e-05,
+      "loss": 1.4601,
+      "step": 26385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921508823377511e-05,
+      "loss": 1.3958,
+      "step": 26390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921479473313462e-05,
+      "loss": 1.0265,
+      "step": 26395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92145011798727e-05,
+      "loss": 1.4243,
+      "step": 26400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921420757399102e-05,
+      "loss": 1.5407,
+      "step": 26405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9213913915491256e-05,
+      "loss": 1.3605,
+      "step": 26410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921362020437509e-05,
+      "loss": 0.9413,
+      "step": 26415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9213326440644197e-05,
+      "loss": 1.0624,
+      "step": 26420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9213032624300258e-05,
+      "loss": 1.0692,
+      "step": 26425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921273875534494e-05,
+      "loss": 1.1011,
+      "step": 26430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9212444833779934e-05,
+      "loss": 1.0276,
+      "step": 26435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9212150859606913e-05,
+      "loss": 1.4206,
+      "step": 26440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9211856832827555e-05,
+      "loss": 1.7397,
+      "step": 26445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921156275344354e-05,
+      "loss": 1.3796,
+      "step": 26450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9211268621456546e-05,
+      "loss": 1.08,
+      "step": 26455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9210974436868256e-05,
+      "loss": 1.4774,
+      "step": 26460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9210680199680347e-05,
+      "loss": 1.3495,
+      "step": 26465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92103859098945e-05,
+      "loss": 1.1465,
+      "step": 26470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9210091567512393e-05,
+      "loss": 1.3838,
+      "step": 26475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9209797172535714e-05,
+      "loss": 1.1784,
+      "step": 26480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9209502724966137e-05,
+      "loss": 1.565,
+      "step": 26485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9209208224805347e-05,
+      "loss": 1.3474,
+      "step": 26490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9208913672055026e-05,
+      "loss": 1.0655,
+      "step": 26495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9208619066716855e-05,
+      "loss": 1.2443,
+      "step": 26500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9208324408792513e-05,
+      "loss": 1.3105,
+      "step": 26505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9208029698283685e-05,
+      "loss": 1.0315,
+      "step": 26510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9207734935192057e-05,
+      "loss": 0.9946,
+      "step": 26515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920744011951931e-05,
+      "loss": 1.4496,
+      "step": 26520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920714525126713e-05,
+      "loss": 1.0672,
+      "step": 26525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9206850330437196e-05,
+      "loss": 1.2646,
+      "step": 26530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9206555357031194e-05,
+      "loss": 1.3131,
+      "step": 26535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9206260331050813e-05,
+      "loss": 0.9826,
+      "step": 26540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9205965252497733e-05,
+      "loss": 1.3572,
+      "step": 26545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9205670121373637e-05,
+      "loss": 1.0744,
+      "step": 26550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920537493768021e-05,
+      "loss": 1.114,
+      "step": 26555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920507970141915e-05,
+      "loss": 1.0375,
+      "step": 26560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920478441259213e-05,
+      "loss": 1.2112,
+      "step": 26565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920448907120084e-05,
+      "loss": 1.3093,
+      "step": 26570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9204193677246964e-05,
+      "loss": 1.6452,
+      "step": 26575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9203898230732197e-05,
+      "loss": 1.4476,
+      "step": 26580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9203602731658216e-05,
+      "loss": 1.6756,
+      "step": 26585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9203307180026714e-05,
+      "loss": 1.0403,
+      "step": 26590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920301157583938e-05,
+      "loss": 1.1331,
+      "step": 26595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.92027159190979e-05,
+      "loss": 1.1452,
+      "step": 26600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920242020980396e-05,
+      "loss": 1.3306,
+      "step": 26605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9202124447959255e-05,
+      "loss": 1.4265,
+      "step": 26610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9201828633565465e-05,
+      "loss": 1.1837,
+      "step": 26615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9201532766624286e-05,
+      "loss": 1.3219,
+      "step": 26620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9201236847137407e-05,
+      "loss": 1.1109,
+      "step": 26625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920094087510652e-05,
+      "loss": 1.1124,
+      "step": 26630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920064485053331e-05,
+      "loss": 1.1187,
+      "step": 26635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9200348773419463e-05,
+      "loss": 1.0633,
+      "step": 26640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920005264376668e-05,
+      "loss": 1.6182,
+      "step": 26645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919975646157665e-05,
+      "loss": 1.2952,
+      "step": 26650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9199460226851065e-05,
+      "loss": 1.599,
+      "step": 26655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9199163939591608e-05,
+      "loss": 1.2129,
+      "step": 26660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919886759979998e-05,
+      "loss": 1.1564,
+      "step": 26665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919857120747787e-05,
+      "loss": 1.0774,
+      "step": 26670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9198274762626972e-05,
+      "loss": 1.418,
+      "step": 26675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9197978265248975e-05,
+      "loss": 1.3628,
+      "step": 26680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9197681715345577e-05,
+      "loss": 1.4121,
+      "step": 26685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919738511291847e-05,
+      "loss": 1.4873,
+      "step": 26690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9197088457969342e-05,
+      "loss": 1.0572,
+      "step": 26695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.91967917504999e-05,
+      "loss": 1.4979,
+      "step": 26700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9196494990511826e-05,
+      "loss": 1.1394,
+      "step": 26705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919619817800682e-05,
+      "loss": 0.9739,
+      "step": 26710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9195901312986578e-05,
+      "loss": 1.2612,
+      "step": 26715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9195604395452792e-05,
+      "loss": 1.2553,
+      "step": 26720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9195307425407158e-05,
+      "loss": 0.9671,
+      "step": 26725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9195010402851373e-05,
+      "loss": 1.384,
+      "step": 26730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9194713327787134e-05,
+      "loss": 1.0584,
+      "step": 26735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9194416200216136e-05,
+      "loss": 1.1871,
+      "step": 26740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9194119020140077e-05,
+      "loss": 0.9233,
+      "step": 26745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9193821787560653e-05,
+      "loss": 1.3347,
+      "step": 26750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9193524502479562e-05,
+      "loss": 1.3573,
+      "step": 26755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.91932271648985e-05,
+      "loss": 1.3186,
+      "step": 26760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919292977481917e-05,
+      "loss": 1.0102,
+      "step": 26765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919263233224326e-05,
+      "loss": 1.1894,
+      "step": 26770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919233483717248e-05,
+      "loss": 1.1545,
+      "step": 26775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9192037289608523e-05,
+      "loss": 1.5307,
+      "step": 26780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9191739689553088e-05,
+      "loss": 1.0871,
+      "step": 26785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9191442037007878e-05,
+      "loss": 1.8431,
+      "step": 26790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919114433197459e-05,
+      "loss": 1.1062,
+      "step": 26795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919084657445492e-05,
+      "loss": 1.3116,
+      "step": 26800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.919054876445058e-05,
+      "loss": 1.0408,
+      "step": 26805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9190250901963256e-05,
+      "loss": 1.0348,
+      "step": 26810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9189952986994662e-05,
+      "loss": 1.3047,
+      "step": 26815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9189655019546492e-05,
+      "loss": 1.3477,
+      "step": 26820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9189356999620447e-05,
+      "loss": 1.208,
+      "step": 26825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9189058927218233e-05,
+      "loss": 1.3859,
+      "step": 26830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.918876080234155e-05,
+      "loss": 0.8371,
+      "step": 26835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.91884626249921e-05,
+      "loss": 1.2172,
+      "step": 26840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9188164395171585e-05,
+      "loss": 1.2392,
+      "step": 26845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9187866112881713e-05,
+      "loss": 1.1293,
+      "step": 26850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9187567778124183e-05,
+      "loss": 1.2968,
+      "step": 26855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9187269390900698e-05,
+      "loss": 1.3172,
+      "step": 26860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.918697095121296e-05,
+      "loss": 1.2046,
+      "step": 26865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9186672459062686e-05,
+      "loss": 1.1448,
+      "step": 26870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9186373914451565e-05,
+      "loss": 1.1003,
+      "step": 26875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9186075317381308e-05,
+      "loss": 1.2221,
+      "step": 26880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.918577666785362e-05,
+      "loss": 1.205,
+      "step": 26885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9185477965870208e-05,
+      "loss": 0.9989,
+      "step": 26890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9185179211432773e-05,
+      "loss": 1.0023,
+      "step": 26895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.918488040454303e-05,
+      "loss": 1.0124,
+      "step": 26900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9184581545202678e-05,
+      "loss": 1.5165,
+      "step": 26905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9184282633413423e-05,
+      "loss": 1.4667,
+      "step": 26910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9183983669176975e-05,
+      "loss": 1.1325,
+      "step": 26915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9183684652495043e-05,
+      "loss": 1.3395,
+      "step": 26920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.918338558336933e-05,
+      "loss": 0.9897,
+      "step": 26925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9183086461801544e-05,
+      "loss": 1.3041,
+      "step": 26930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.91827872877934e-05,
+      "loss": 1.1326,
+      "step": 26935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9182488061346597e-05,
+      "loss": 1.3086,
+      "step": 26940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9182188782462854e-05,
+      "loss": 0.9832,
+      "step": 26945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9181889451143873e-05,
+      "loss": 1.6528,
+      "step": 26950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.918159006739136e-05,
+      "loss": 1.1032,
+      "step": 26955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9181290631207033e-05,
+      "loss": 1.5496,
+      "step": 26960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9180991142592596e-05,
+      "loss": 1.0719,
+      "step": 26965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9180691601549765e-05,
+      "loss": 1.1666,
+      "step": 26970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9180392008080245e-05,
+      "loss": 1.2393,
+      "step": 26975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9180092362185746e-05,
+      "loss": 1.172,
+      "step": 26980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.917979266386799e-05,
+      "loss": 1.1034,
+      "step": 26985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9179492913128676e-05,
+      "loss": 1.3546,
+      "step": 26990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9179193109969517e-05,
+      "loss": 1.0893,
+      "step": 26995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9178893254392232e-05,
+      "loss": 1.3703,
+      "step": 27000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.917859334639853e-05,
+      "loss": 1.1376,
+      "step": 27005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9178293385990122e-05,
+      "loss": 1.2821,
+      "step": 27010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9177993373168723e-05,
+      "loss": 0.9424,
+      "step": 27015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9177693307936045e-05,
+      "loss": 0.7951,
+      "step": 27020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.91773931902938e-05,
+      "loss": 0.9791,
+      "step": 27025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9177093020243705e-05,
+      "loss": 1.2906,
+      "step": 27030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9176792797787475e-05,
+      "loss": 1.1342,
+      "step": 27035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9176492522926818e-05,
+      "loss": 1.4607,
+      "step": 27040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9176192195663458e-05,
+      "loss": 1.364,
+      "step": 27045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9175891815999103e-05,
+      "loss": 1.0889,
+      "step": 27050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9175591383935473e-05,
+      "loss": 1.5875,
+      "step": 27055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9175290899474277e-05,
+      "loss": 1.37,
+      "step": 27060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9174990362617235e-05,
+      "loss": 1.3061,
+      "step": 27065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9174689773366066e-05,
+      "loss": 1.0725,
+      "step": 27070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9174389131722484e-05,
+      "loss": 1.1114,
+      "step": 27075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9174088437688203e-05,
+      "loss": 1.2069,
+      "step": 27080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9173787691264944e-05,
+      "loss": 0.9883,
+      "step": 27085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9173486892454422e-05,
+      "loss": 1.3708,
+      "step": 27090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9173186041258358e-05,
+      "loss": 1.5986,
+      "step": 27095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9172885137678466e-05,
+      "loss": 1.7158,
+      "step": 27100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9172584181716468e-05,
+      "loss": 1.2234,
+      "step": 27105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.917228317337408e-05,
+      "loss": 1.3913,
+      "step": 27110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9171982112653023e-05,
+      "loss": 1.2757,
+      "step": 27115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9171680999555012e-05,
+      "loss": 1.5049,
+      "step": 27120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9171379834081773e-05,
+      "loss": 1.3304,
+      "step": 27125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.917107861623502e-05,
+      "loss": 1.0207,
+      "step": 27130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9170777346016476e-05,
+      "loss": 1.0284,
+      "step": 27135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.917047602342786e-05,
+      "loss": 1.117,
+      "step": 27140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9170174648470895e-05,
+      "loss": 1.4404,
+      "step": 27145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.91698732211473e-05,
+      "loss": 1.3052,
+      "step": 27150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9169571741458796e-05,
+      "loss": 1.1516,
+      "step": 27155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9169270209407105e-05,
+      "loss": 1.0173,
+      "step": 27160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.916896862499395e-05,
+      "loss": 1.264,
+      "step": 27165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9168666988221053e-05,
+      "loss": 1.0813,
+      "step": 27170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9168365299090138e-05,
+      "loss": 0.9391,
+      "step": 27175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9168063557602923e-05,
+      "loss": 1.2099,
+      "step": 27180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9167761763761134e-05,
+      "loss": 1.0937,
+      "step": 27185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9167459917566496e-05,
+      "loss": 1.1763,
+      "step": 27190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.916715801902073e-05,
+      "loss": 1.2431,
+      "step": 27195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9166856068125563e-05,
+      "loss": 1.5276,
+      "step": 27200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9166554064882715e-05,
+      "loss": 1.1142,
+      "step": 27205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.916625200929392e-05,
+      "loss": 1.0712,
+      "step": 27210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9165949901360888e-05,
+      "loss": 1.0361,
+      "step": 27215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9165647741085355e-05,
+      "loss": 1.2279,
+      "step": 27220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.916534552846904e-05,
+      "loss": 1.2387,
+      "step": 27225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9165043263513676e-05,
+      "loss": 1.3095,
+      "step": 27230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9164740946220986e-05,
+      "loss": 1.1152,
+      "step": 27235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9164438576592694e-05,
+      "loss": 1.5168,
+      "step": 27240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.916413615463053e-05,
+      "loss": 1.7925,
+      "step": 27245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9163833680336217e-05,
+      "loss": 1.2746,
+      "step": 27250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9163531153711488e-05,
+      "loss": 1.001,
+      "step": 27255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9163228574758065e-05,
+      "loss": 1.2322,
+      "step": 27260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9162925943477685e-05,
+      "loss": 1.0187,
+      "step": 27265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.916262325987206e-05,
+      "loss": 1.1737,
+      "step": 27270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9162320523942935e-05,
+      "loss": 1.3938,
+      "step": 27275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.916201773569203e-05,
+      "loss": 1.162,
+      "step": 27280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9161714895121073e-05,
+      "loss": 1.7008,
+      "step": 27285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9161412002231802e-05,
+      "loss": 1.2559,
+      "step": 27290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9161109057025934e-05,
+      "loss": 1.2717,
+      "step": 27295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.916080605950521e-05,
+      "loss": 1.2473,
+      "step": 27300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9160503009671357e-05,
+      "loss": 1.6476,
+      "step": 27305
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9160199907526103e-05,
+      "loss": 1.1859,
+      "step": 27310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.915989675307118e-05,
+      "loss": 1.4891,
+      "step": 27315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9159593546308322e-05,
+      "loss": 0.972,
+      "step": 27320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9159290287239258e-05,
+      "loss": 1.3183,
+      "step": 27325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.915898697586572e-05,
+      "loss": 1.4363,
+      "step": 27330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.915868361218944e-05,
+      "loss": 1.0973,
+      "step": 27335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.915838019621215e-05,
+      "loss": 1.3592,
+      "step": 27340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9158076727935585e-05,
+      "loss": 1.3006,
+      "step": 27345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9157773207361473e-05,
+      "loss": 1.1301,
+      "step": 27350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9157469634491553e-05,
+      "loss": 1.5134,
+      "step": 27355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9157166009327554e-05,
+      "loss": 1.5289,
+      "step": 27360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9156862331871214e-05,
+      "loss": 1.3529,
+      "step": 27365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9156558602124264e-05,
+      "loss": 0.9688,
+      "step": 27370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.915625482008844e-05,
+      "loss": 0.9429,
+      "step": 27375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9155950985765476e-05,
+      "loss": 1.1739,
+      "step": 27380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9155647099157108e-05,
+      "loss": 1.2552,
+      "step": 27385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9155343160265068e-05,
+      "loss": 1.1213,
+      "step": 27390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9155039169091096e-05,
+      "loss": 1.3666,
+      "step": 27395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9154735125636925e-05,
+      "loss": 1.0611,
+      "step": 27400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9154431029904294e-05,
+      "loss": 1.2114,
+      "step": 27405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9154126881894937e-05,
+      "loss": 0.9672,
+      "step": 27410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9153822681610594e-05,
+      "loss": 1.1045,
+      "step": 27415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9153518429053e-05,
+      "loss": 1.4785,
+      "step": 27420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9153214124223894e-05,
+      "loss": 1.1186,
+      "step": 27425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.915290976712501e-05,
+      "loss": 1.1526,
+      "step": 27430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9152605357758088e-05,
+      "loss": 1.2699,
+      "step": 27435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9152300896124866e-05,
+      "loss": 1.0465,
+      "step": 27440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9151996382227085e-05,
+      "loss": 1.1178,
+      "step": 27445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.915169181606648e-05,
+      "loss": 1.325,
+      "step": 27450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9151387197644796e-05,
+      "loss": 1.2744,
+      "step": 27455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9151082526963765e-05,
+      "loss": 1.1742,
+      "step": 27460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9150777804025134e-05,
+      "loss": 0.9079,
+      "step": 27465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.915047302883064e-05,
+      "loss": 1.3098,
+      "step": 27470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.915016820138202e-05,
+      "loss": 1.1279,
+      "step": 27475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.914986332168102e-05,
+      "loss": 0.9424,
+      "step": 27480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.914955838972938e-05,
+      "loss": 1.1295,
+      "step": 27485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.914925340552884e-05,
+      "loss": 1.0523,
+      "step": 27490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9148948369081143e-05,
+      "loss": 1.2508,
+      "step": 27495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9148643280388027e-05,
+      "loss": 1.3717,
+      "step": 27500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.914833813945124e-05,
+      "loss": 0.9356,
+      "step": 27505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.914803294627252e-05,
+      "loss": 1.2196,
+      "step": 27510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9147727700853614e-05,
+      "loss": 1.1392,
+      "step": 27515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.914742240319626e-05,
+      "loss": 1.1451,
+      "step": 27520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9147117053302206e-05,
+      "loss": 1.0883,
+      "step": 27525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9146811651173195e-05,
+      "loss": 1.2466,
+      "step": 27530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.914650619681097e-05,
+      "loss": 1.1372,
+      "step": 27535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.914620069021727e-05,
+      "loss": 1.5923,
+      "step": 27540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9145895131393847e-05,
+      "loss": 1.1596,
+      "step": 27545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9145589520342448e-05,
+      "loss": 1.0108,
+      "step": 27550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.914528385706481e-05,
+      "loss": 1.1379,
+      "step": 27555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9144978141562682e-05,
+      "loss": 1.5038,
+      "step": 27560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.914467237383781e-05,
+      "loss": 1.1964,
+      "step": 27565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9144366553891944e-05,
+      "loss": 1.1233,
+      "step": 27570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9144060681726823e-05,
+      "loss": 1.6805,
+      "step": 27575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9143754757344198e-05,
+      "loss": 1.1298,
+      "step": 27580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9143448780745814e-05,
+      "loss": 1.4432,
+      "step": 27585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9143142751933422e-05,
+      "loss": 1.6813,
+      "step": 27590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9142836670908767e-05,
+      "loss": 1.0765,
+      "step": 27595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9142530537673598e-05,
+      "loss": 1.2679,
+      "step": 27600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.914222435222966e-05,
+      "loss": 1.111,
+      "step": 27605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9141918114578706e-05,
+      "loss": 1.0945,
+      "step": 27610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9141611824722482e-05,
+      "loss": 1.2193,
+      "step": 27615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9141305482662734e-05,
+      "loss": 1.2947,
+      "step": 27620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.914099908840122e-05,
+      "loss": 1.2853,
+      "step": 27625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9140692641939683e-05,
+      "loss": 0.883,
+      "step": 27630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9140386143279874e-05,
+      "loss": 1.4935,
+      "step": 27635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9140079592423543e-05,
+      "loss": 1.6289,
+      "step": 27640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9139772989372446e-05,
+      "loss": 0.9004,
+      "step": 27645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9139466334128323e-05,
+      "loss": 1.3303,
+      "step": 27650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9139159626692934e-05,
+      "loss": 1.2664,
+      "step": 27655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.913885286706803e-05,
+      "loss": 1.1508,
+      "step": 27660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.913854605525536e-05,
+      "loss": 1.0689,
+      "step": 27665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9138239191256674e-05,
+      "loss": 1.0721,
+      "step": 27670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9137932275073732e-05,
+      "loss": 1.3005,
+      "step": 27675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9137625306708282e-05,
+      "loss": 1.2307,
+      "step": 27680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.913731828616207e-05,
+      "loss": 1.2414,
+      "step": 27685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9137011213436863e-05,
+      "loss": 1.3572,
+      "step": 27690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9136704088534405e-05,
+      "loss": 1.1424,
+      "step": 27695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.913639691145645e-05,
+      "loss": 1.1815,
+      "step": 27700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.913608968220476e-05,
+      "loss": 1.4211,
+      "step": 27705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.913578240078108e-05,
+      "loss": 1.0722,
+      "step": 27710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.913547506718717e-05,
+      "loss": 1.1096,
+      "step": 27715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.913516768142478e-05,
+      "loss": 1.1057,
+      "step": 27720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9134860243495676e-05,
+      "loss": 1.258,
+      "step": 27725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9134552753401603e-05,
+      "loss": 1.367,
+      "step": 27730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.913424521114432e-05,
+      "loss": 1.1435,
+      "step": 27735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9133937616725588e-05,
+      "loss": 1.27,
+      "step": 27740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9133629970147155e-05,
+      "loss": 1.4523,
+      "step": 27745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9133322271410785e-05,
+      "loss": 1.0846,
+      "step": 27750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9133014520518232e-05,
+      "loss": 1.0215,
+      "step": 27755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9132706717471256e-05,
+      "loss": 1.0318,
+      "step": 27760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.913239886227161e-05,
+      "loss": 0.999,
+      "step": 27765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9132090954921054e-05,
+      "loss": 1.2128,
+      "step": 27770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9131782995421345e-05,
+      "loss": 0.9755,
+      "step": 27775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9131474983774247e-05,
+      "loss": 1.5296,
+      "step": 27780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9131166919981513e-05,
+      "loss": 1.4599,
+      "step": 27785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.913085880404491e-05,
+      "loss": 1.1265,
+      "step": 27790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9130550635966187e-05,
+      "loss": 1.2508,
+      "step": 27795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.913024241574711e-05,
+      "loss": 1.4249,
+      "step": 27800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9129934143389438e-05,
+      "loss": 1.5168,
+      "step": 27805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9129625818894935e-05,
+      "loss": 1.2352,
+      "step": 27810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9129317442265355e-05,
+      "loss": 1.0183,
+      "step": 27815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.912900901350246e-05,
+      "loss": 1.1658,
+      "step": 27820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9128700532608014e-05,
+      "loss": 1.1559,
+      "step": 27825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9128391999583784e-05,
+      "loss": 1.2364,
+      "step": 27830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.912808341443152e-05,
+      "loss": 1.147,
+      "step": 27835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9127774777152992e-05,
+      "loss": 1.2977,
+      "step": 27840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.912746608774996e-05,
+      "loss": 1.7345,
+      "step": 27845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.912715734622419e-05,
+      "loss": 1.1136,
+      "step": 27850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.912684855257744e-05,
+      "loss": 1.1237,
+      "step": 27855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9126539706811475e-05,
+      "loss": 1.2568,
+      "step": 27860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9126230808928064e-05,
+      "loss": 1.0021,
+      "step": 27865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9125921858928963e-05,
+      "loss": 1.2226,
+      "step": 27870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.912561285681594e-05,
+      "loss": 1.4319,
+      "step": 27875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9125303802590758e-05,
+      "loss": 1.1177,
+      "step": 27880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9124994696255185e-05,
+      "loss": 1.158,
+      "step": 27885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9124685537810985e-05,
+      "loss": 1.2531,
+      "step": 27890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.912437632725992e-05,
+      "loss": 1.2319,
+      "step": 27895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9124067064603764e-05,
+      "loss": 1.3632,
+      "step": 27900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9123757749844276e-05,
+      "loss": 1.0658,
+      "step": 27905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.912344838298322e-05,
+      "loss": 1.0305,
+      "step": 27910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9123138964022367e-05,
+      "loss": 1.0325,
+      "step": 27915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9122829492963487e-05,
+      "loss": 1.3337,
+      "step": 27920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.912251996980834e-05,
+      "loss": 1.2713,
+      "step": 27925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9122210394558703e-05,
+      "loss": 1.3091,
+      "step": 27930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9121900767216335e-05,
+      "loss": 1.5587,
+      "step": 27935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9121591087783007e-05,
+      "loss": 1.2459,
+      "step": 27940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.912128135626049e-05,
+      "loss": 1.1256,
+      "step": 27945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9120971572650545e-05,
+      "loss": 1.2521,
+      "step": 27950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9120661736954952e-05,
+      "loss": 1.1222,
+      "step": 27955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9120351849175476e-05,
+      "loss": 1.0522,
+      "step": 27960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9120041909313883e-05,
+      "loss": 1.0244,
+      "step": 27965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9119731917371943e-05,
+      "loss": 1.1821,
+      "step": 27970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9119421873351433e-05,
+      "loss": 1.009,
+      "step": 27975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9119111777254117e-05,
+      "loss": 1.1119,
+      "step": 27980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9118801629081768e-05,
+      "loss": 0.9925,
+      "step": 27985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9118491428836158e-05,
+      "loss": 1.0746,
+      "step": 27990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9118181176519055e-05,
+      "loss": 1.0678,
+      "step": 27995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9117870872132234e-05,
+      "loss": 1.3514,
+      "step": 28000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9117560515677472e-05,
+      "loss": 1.1314,
+      "step": 28005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.911725010715653e-05,
+      "loss": 1.2887,
+      "step": 28010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9116939646571186e-05,
+      "loss": 1.1322,
+      "step": 28015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9116629133923216e-05,
+      "loss": 1.158,
+      "step": 28020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.911631856921439e-05,
+      "loss": 1.2892,
+      "step": 28025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9116007952446476e-05,
+      "loss": 1.2517,
+      "step": 28030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.911569728362126e-05,
+      "loss": 1.185,
+      "step": 28035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9115386562740506e-05,
+      "loss": 1.5896,
+      "step": 28040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9115075789805996e-05,
+      "loss": 1.1943,
+      "step": 28045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9114764964819496e-05,
+      "loss": 1.5103,
+      "step": 28050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9114454087782787e-05,
+      "loss": 1.2214,
+      "step": 28055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9114143158697645e-05,
+      "loss": 1.2675,
+      "step": 28060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9113832177565842e-05,
+      "loss": 1.1787,
+      "step": 28065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9113521144389153e-05,
+      "loss": 0.8999,
+      "step": 28070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911321005916936e-05,
+      "loss": 1.3014,
+      "step": 28075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9112898921908236e-05,
+      "loss": 1.1257,
+      "step": 28080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9112587732607556e-05,
+      "loss": 1.3497,
+      "step": 28085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91122764912691e-05,
+      "loss": 1.1314,
+      "step": 28090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9111965197894646e-05,
+      "loss": 1.1333,
+      "step": 28095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9111653852485966e-05,
+      "loss": 1.2812,
+      "step": 28100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9111342455044842e-05,
+      "loss": 1.5117,
+      "step": 28105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9111031005573055e-05,
+      "loss": 1.0941,
+      "step": 28110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911071950407238e-05,
+      "loss": 1.3459,
+      "step": 28115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9110407950544593e-05,
+      "loss": 1.2066,
+      "step": 28120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.911009634499148e-05,
+      "loss": 1.1454,
+      "step": 28125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9109784687414814e-05,
+      "loss": 1.1079,
+      "step": 28130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910947297781638e-05,
+      "loss": 1.2344,
+      "step": 28135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9109161216197957e-05,
+      "loss": 1.1051,
+      "step": 28140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9108849402561322e-05,
+      "loss": 1.1901,
+      "step": 28145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910853753690826e-05,
+      "loss": 1.6487,
+      "step": 28150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9108225619240546e-05,
+      "loss": 1.2379,
+      "step": 28155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9107913649559968e-05,
+      "loss": 1.2063,
+      "step": 28160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.91076016278683e-05,
+      "loss": 1.3094,
+      "step": 28165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910728955416733e-05,
+      "loss": 1.1217,
+      "step": 28170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9106977428458845e-05,
+      "loss": 1.1434,
+      "step": 28175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9106665250744614e-05,
+      "loss": 1.044,
+      "step": 28180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9106353021026426e-05,
+      "loss": 1.0849,
+      "step": 28185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9106040739306063e-05,
+      "loss": 1.1046,
+      "step": 28190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9105728405585317e-05,
+      "loss": 0.9196,
+      "step": 28195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9105416019865957e-05,
+      "loss": 1.1093,
+      "step": 28200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910510358214978e-05,
+      "loss": 1.0749,
+      "step": 28205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9104791092438558e-05,
+      "loss": 1.5976,
+      "step": 28210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9104478550734086e-05,
+      "loss": 1.3288,
+      "step": 28215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9104165957038146e-05,
+      "loss": 1.2898,
+      "step": 28220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910385331135252e-05,
+      "loss": 1.3425,
+      "step": 28225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9103540613678993e-05,
+      "loss": 1.5056,
+      "step": 28230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9103227864019354e-05,
+      "loss": 1.1516,
+      "step": 28235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910291506237539e-05,
+      "loss": 1.1717,
+      "step": 28240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910260220874888e-05,
+      "loss": 1.1708,
+      "step": 28245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9102289303141618e-05,
+      "loss": 1.2705,
+      "step": 28250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910197634555539e-05,
+      "loss": 1.2934,
+      "step": 28255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910166333599198e-05,
+      "loss": 1.2341,
+      "step": 28260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9101350274453176e-05,
+      "loss": 1.1472,
+      "step": 28265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910103716094077e-05,
+      "loss": 1.3289,
+      "step": 28270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9100723995456546e-05,
+      "loss": 1.2776,
+      "step": 28275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910041077800229e-05,
+      "loss": 1.0524,
+      "step": 28280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9100097508579797e-05,
+      "loss": 1.5149,
+      "step": 28285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9099784187190852e-05,
+      "loss": 1.0513,
+      "step": 28290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9099470813837242e-05,
+      "loss": 1.3271,
+      "step": 28295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9099157388520763e-05,
+      "loss": 1.2615,
+      "step": 28300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.90988439112432e-05,
+      "loss": 1.7584,
+      "step": 28305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9098530382006342e-05,
+      "loss": 1.1647,
+      "step": 28310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9098216800811986e-05,
+      "loss": 1.1148,
+      "step": 28315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909790316766192e-05,
+      "loss": 1.2166,
+      "step": 28320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909758948255793e-05,
+      "loss": 1.3608,
+      "step": 28325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909727574550181e-05,
+      "loss": 1.2816,
+      "step": 28330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9096961956495354e-05,
+      "loss": 1.1907,
+      "step": 28335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9096648115540353e-05,
+      "loss": 1.1044,
+      "step": 28340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.90963342226386e-05,
+      "loss": 1.2815,
+      "step": 28345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9096020277791885e-05,
+      "loss": 1.0207,
+      "step": 28350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9095706281002e-05,
+      "loss": 1.245,
+      "step": 28355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909539223227074e-05,
+      "loss": 1.0943,
+      "step": 28360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.90950781315999e-05,
+      "loss": 1.3467,
+      "step": 28365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9094763978991273e-05,
+      "loss": 1.5583,
+      "step": 28370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909444977444665e-05,
+      "loss": 1.4623,
+      "step": 28375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9094135517967826e-05,
+      "loss": 1.3021,
+      "step": 28380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.90938212095566e-05,
+      "loss": 1.1184,
+      "step": 28385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909350684921476e-05,
+      "loss": 1.0351,
+      "step": 28390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909319243694411e-05,
+      "loss": 1.3723,
+      "step": 28395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.909287797274644e-05,
+      "loss": 1.8685,
+      "step": 28400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9092563456623542e-05,
+      "loss": 1.2336,
+      "step": 28405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9092248888577218e-05,
+      "loss": 1.1011,
+      "step": 28410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9091934268609264e-05,
+      "loss": 1.3732,
+      "step": 28415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9091619596721475e-05,
+      "loss": 1.0958,
+      "step": 28420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9091304872915647e-05,
+      "loss": 1.4715,
+      "step": 28425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9090990097193578e-05,
+      "loss": 1.302,
+      "step": 28430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9090675269557066e-05,
+      "loss": 1.2242,
+      "step": 28435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9090360390007908e-05,
+      "loss": 1.4441,
+      "step": 28440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9090045458547905e-05,
+      "loss": 1.6247,
+      "step": 28445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9089730475178853e-05,
+      "loss": 0.9238,
+      "step": 28450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908941543990255e-05,
+      "loss": 1.1275,
+      "step": 28455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.90891003527208e-05,
+      "loss": 1.2882,
+      "step": 28460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9088785213635392e-05,
+      "loss": 1.3724,
+      "step": 28465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9088470022648136e-05,
+      "loss": 1.0435,
+      "step": 28470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9088154779760827e-05,
+      "loss": 1.1797,
+      "step": 28475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9087839484975267e-05,
+      "loss": 1.1981,
+      "step": 28480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9087524138293252e-05,
+      "loss": 1.3481,
+      "step": 28485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908720873971659e-05,
+      "loss": 0.9123,
+      "step": 28490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9086893289247077e-05,
+      "loss": 1.1474,
+      "step": 28495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9086577786886518e-05,
+      "loss": 1.5713,
+      "step": 28500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908626223263671e-05,
+      "loss": 1.4337,
+      "step": 28505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9085946626499458e-05,
+      "loss": 1.3641,
+      "step": 28510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9085630968476565e-05,
+      "loss": 1.1458,
+      "step": 28515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9085315258569834e-05,
+      "loss": 1.2572,
+      "step": 28520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908499949678106e-05,
+      "loss": 1.2254,
+      "step": 28525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908468368311206e-05,
+      "loss": 1.3087,
+      "step": 28530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9084367817564626e-05,
+      "loss": 1.3519,
+      "step": 28535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9084051900140566e-05,
+      "loss": 1.1501,
+      "step": 28540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9083735930841684e-05,
+      "loss": 1.0625,
+      "step": 28545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9083419909669783e-05,
+      "loss": 1.1349,
+      "step": 28550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908310383662667e-05,
+      "loss": 1.2402,
+      "step": 28555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9082787711714153e-05,
+      "loss": 1.1792,
+      "step": 28560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908247153493403e-05,
+      "loss": 1.5399,
+      "step": 28565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9082155306288105e-05,
+      "loss": 1.347,
+      "step": 28570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908183902577819e-05,
+      "loss": 1.469,
+      "step": 28575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9081522693406094e-05,
+      "loss": 1.2685,
+      "step": 28580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9081206309173615e-05,
+      "loss": 1.6695,
+      "step": 28585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9080889873082567e-05,
+      "loss": 1.0447,
+      "step": 28590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9080573385134752e-05,
+      "loss": 1.1097,
+      "step": 28595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9080256845331977e-05,
+      "loss": 1.1259,
+      "step": 28600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9079940253676054e-05,
+      "loss": 1.1321,
+      "step": 28605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907962361016879e-05,
+      "loss": 1.3413,
+      "step": 28610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9079306914811987e-05,
+      "loss": 1.2684,
+      "step": 28615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907899016760746e-05,
+      "loss": 1.4974,
+      "step": 28620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9078673368557016e-05,
+      "loss": 1.0758,
+      "step": 28625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9078356517662464e-05,
+      "loss": 1.2529,
+      "step": 28630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907803961492561e-05,
+      "loss": 1.2555,
+      "step": 28635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9077722660348274e-05,
+      "loss": 1.2862,
+      "step": 28640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9077405653932256e-05,
+      "loss": 1.3266,
+      "step": 28645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907708859567937e-05,
+      "loss": 1.5012,
+      "step": 28650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9076771485591426e-05,
+      "loss": 1.0689,
+      "step": 28655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9076454323670232e-05,
+      "loss": 1.3173,
+      "step": 28660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907613710991761e-05,
+      "loss": 1.0317,
+      "step": 28665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9075819844335354e-05,
+      "loss": 0.9195,
+      "step": 28670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9075502526925288e-05,
+      "loss": 1.2946,
+      "step": 28675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9075185157689226e-05,
+      "loss": 1.1807,
+      "step": 28680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907486773662897e-05,
+      "loss": 1.1486,
+      "step": 28685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907455026374634e-05,
+      "loss": 0.9754,
+      "step": 28690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9074232739043147e-05,
+      "loss": 1.0283,
+      "step": 28695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9073915162521206e-05,
+      "loss": 1.1701,
+      "step": 28700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907359753418233e-05,
+      "loss": 1.755,
+      "step": 28705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907327985402833e-05,
+      "loss": 1.2511,
+      "step": 28710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907296212206102e-05,
+      "loss": 1.2255,
+      "step": 28715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907264433828222e-05,
+      "loss": 1.2553,
+      "step": 28720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9072326502693737e-05,
+      "loss": 1.2032,
+      "step": 28725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9072008615297393e-05,
+      "loss": 1.0915,
+      "step": 28730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9071690676095e-05,
+      "loss": 1.1607,
+      "step": 28735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907137268508837e-05,
+      "loss": 1.1025,
+      "step": 28740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907105464227933e-05,
+      "loss": 1.2646,
+      "step": 28745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9070736547669683e-05,
+      "loss": 1.3584,
+      "step": 28750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9070418401261258e-05,
+      "loss": 1.2873,
+      "step": 28755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.907010020305586e-05,
+      "loss": 1.3106,
+      "step": 28760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9069781953055312e-05,
+      "loss": 1.4959,
+      "step": 28765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9069463651261434e-05,
+      "loss": 0.9778,
+      "step": 28770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9069145297676038e-05,
+      "loss": 1.2838,
+      "step": 28775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9068826892300943e-05,
+      "loss": 1.0174,
+      "step": 28780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9068508435137973e-05,
+      "loss": 1.2336,
+      "step": 28785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906818992618894e-05,
+      "loss": 1.1761,
+      "step": 28790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9067871365455667e-05,
+      "loss": 1.1153,
+      "step": 28795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906755275293997e-05,
+      "loss": 0.956,
+      "step": 28800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906723408864367e-05,
+      "loss": 1.2129,
+      "step": 28805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9066915372568583e-05,
+      "loss": 1.3527,
+      "step": 28810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9066596604716537e-05,
+      "loss": 1.0811,
+      "step": 28815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9066277785089348e-05,
+      "loss": 1.386,
+      "step": 28820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906595891368883e-05,
+      "loss": 1.436,
+      "step": 28825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9065639990516818e-05,
+      "loss": 1.0685,
+      "step": 28830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9065321015575123e-05,
+      "loss": 1.3479,
+      "step": 28835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906500198886557e-05,
+      "loss": 1.3209,
+      "step": 28840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9064682910389976e-05,
+      "loss": 1.218,
+      "step": 28845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9064363780150174e-05,
+      "loss": 1.0982,
+      "step": 28850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9064044598147975e-05,
+      "loss": 1.1659,
+      "step": 28855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9063725364385205e-05,
+      "loss": 1.4131,
+      "step": 28860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906340607886369e-05,
+      "loss": 1.1183,
+      "step": 28865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9063086741585255e-05,
+      "loss": 1.0945,
+      "step": 28870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9062767352551717e-05,
+      "loss": 1.4709,
+      "step": 28875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9062447911764903e-05,
+      "loss": 1.2508,
+      "step": 28880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9062128419226636e-05,
+      "loss": 1.2392,
+      "step": 28885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9061808874938742e-05,
+      "loss": 1.4347,
+      "step": 28890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9061489278903047e-05,
+      "loss": 0.9075,
+      "step": 28895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9061169631121374e-05,
+      "loss": 1.557,
+      "step": 28900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.906084993159555e-05,
+      "loss": 1.2643,
+      "step": 28905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9060530180327396e-05,
+      "loss": 1.1511,
+      "step": 28910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9060210377318743e-05,
+      "loss": 1.5826,
+      "step": 28915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.905989052257142e-05,
+      "loss": 1.2473,
+      "step": 28920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9059570616087243e-05,
+      "loss": 1.508,
+      "step": 28925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9059250657868046e-05,
+      "loss": 1.2895,
+      "step": 28930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9058930647915658e-05,
+      "loss": 1.0148,
+      "step": 28935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9058610586231904e-05,
+      "loss": 1.2585,
+      "step": 28940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.905829047281861e-05,
+      "loss": 1.0236,
+      "step": 28945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9057970307677604e-05,
+      "loss": 1.0581,
+      "step": 28950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9057650090810716e-05,
+      "loss": 1.2621,
+      "step": 28955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9057329822219775e-05,
+      "loss": 1.2469,
+      "step": 28960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.905700950190661e-05,
+      "loss": 1.518,
+      "step": 28965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9056689129873048e-05,
+      "loss": 1.4158,
+      "step": 28970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.905636870612092e-05,
+      "loss": 1.4439,
+      "step": 28975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.905604823065206e-05,
+      "loss": 1.1524,
+      "step": 28980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9055727703468284e-05,
+      "loss": 1.2877,
+      "step": 28985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.905540712457144e-05,
+      "loss": 1.2999,
+      "step": 28990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9055086493963348e-05,
+      "loss": 1.1661,
+      "step": 28995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9054765811645842e-05,
+      "loss": 1.2255,
+      "step": 29000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.905444507762076e-05,
+      "loss": 1.1271,
+      "step": 29005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.905412429188992e-05,
+      "loss": 1.1289,
+      "step": 29010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.905380345445516e-05,
+      "loss": 1.1447,
+      "step": 29015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.905348256531831e-05,
+      "loss": 1.3022,
+      "step": 29020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9053161624481213e-05,
+      "loss": 1.0231,
+      "step": 29025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.905284063194569e-05,
+      "loss": 0.9313,
+      "step": 29030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9052519587713576e-05,
+      "loss": 1.0764,
+      "step": 29035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9052198491786707e-05,
+      "loss": 1.1893,
+      "step": 29040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.905187734416692e-05,
+      "loss": 1.3149,
+      "step": 29045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.905155614485604e-05,
+      "loss": 1.2664,
+      "step": 29050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9051234893855907e-05,
+      "loss": 1.119,
+      "step": 29055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9050913591168353e-05,
+      "loss": 1.113,
+      "step": 29060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9050592236795218e-05,
+      "loss": 1.1216,
+      "step": 29065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9050270830738335e-05,
+      "loss": 1.3797,
+      "step": 29070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9049949372999533e-05,
+      "loss": 1.0534,
+      "step": 29075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9049627863580653e-05,
+      "loss": 1.4433,
+      "step": 29080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904930630248353e-05,
+      "loss": 1.5894,
+      "step": 29085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9048984689710004e-05,
+      "loss": 1.1174,
+      "step": 29090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9048663025261905e-05,
+      "loss": 1.5647,
+      "step": 29095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904834130914108e-05,
+      "loss": 1.3983,
+      "step": 29100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904801954134935e-05,
+      "loss": 1.448,
+      "step": 29105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904769772188857e-05,
+      "loss": 1.1323,
+      "step": 29110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9047375850760565e-05,
+      "loss": 1.358,
+      "step": 29115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904705392796718e-05,
+      "loss": 1.0977,
+      "step": 29120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904673195351025e-05,
+      "loss": 0.9658,
+      "step": 29125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9046409927391613e-05,
+      "loss": 1.4357,
+      "step": 29130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9046087849613113e-05,
+      "loss": 1.3336,
+      "step": 29135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9045765720176584e-05,
+      "loss": 1.221,
+      "step": 29140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9045443539083867e-05,
+      "loss": 1.855,
+      "step": 29145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9045121306336803e-05,
+      "loss": 1.1616,
+      "step": 29150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9044799021937233e-05,
+      "loss": 1.1765,
+      "step": 29155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9044476685886994e-05,
+      "loss": 1.1902,
+      "step": 29160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9044154298187928e-05,
+      "loss": 1.6542,
+      "step": 29165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9043831858841878e-05,
+      "loss": 1.8596,
+      "step": 29170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9043509367850684e-05,
+      "loss": 1.4233,
+      "step": 29175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904318682521619e-05,
+      "loss": 1.4191,
+      "step": 29180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904286423094023e-05,
+      "loss": 1.8056,
+      "step": 29185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9042541585024655e-05,
+      "loss": 1.2924,
+      "step": 29190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.90422188874713e-05,
+      "loss": 1.2552,
+      "step": 29195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9041896138282015e-05,
+      "loss": 1.1455,
+      "step": 29200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904157333745864e-05,
+      "loss": 1.0743,
+      "step": 29205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904125048500302e-05,
+      "loss": 1.3098,
+      "step": 29210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904092758091699e-05,
+      "loss": 1.2617,
+      "step": 29215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9040604625202407e-05,
+      "loss": 1.0864,
+      "step": 29220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9040281617861108e-05,
+      "loss": 1.248,
+      "step": 29225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9039958558894937e-05,
+      "loss": 1.1535,
+      "step": 29230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9039635448305742e-05,
+      "loss": 1.1532,
+      "step": 29235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9039312286095368e-05,
+      "loss": 1.0037,
+      "step": 29240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9038989072265654e-05,
+      "loss": 1.2179,
+      "step": 29245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9038665806818458e-05,
+      "loss": 1.3731,
+      "step": 29250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.903834248975561e-05,
+      "loss": 1.2331,
+      "step": 29255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9038019121078974e-05,
+      "loss": 1.5114,
+      "step": 29260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9037695700790383e-05,
+      "loss": 1.3117,
+      "step": 29265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9037372228891686e-05,
+      "loss": 0.9646,
+      "step": 29270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9037048705384738e-05,
+      "loss": 1.3072,
+      "step": 29275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.903672513027138e-05,
+      "loss": 1.719,
+      "step": 29280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.903640150355346e-05,
+      "loss": 1.3459,
+      "step": 29285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9036077825232826e-05,
+      "loss": 1.3048,
+      "step": 29290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9035754095311328e-05,
+      "loss": 1.2615,
+      "step": 29295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9035430313790815e-05,
+      "loss": 1.0629,
+      "step": 29300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9035106480673135e-05,
+      "loss": 1.3329,
+      "step": 29305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.903478259596014e-05,
+      "loss": 1.6941,
+      "step": 29310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9034458659653674e-05,
+      "loss": 1.1875,
+      "step": 29315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9034134671755587e-05,
+      "loss": 1.1492,
+      "step": 29320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9033810632267738e-05,
+      "loss": 1.2367,
+      "step": 29325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.903348654119197e-05,
+      "loss": 1.045,
+      "step": 29330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9033162398530135e-05,
+      "loss": 1.4207,
+      "step": 29335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.903283820428408e-05,
+      "loss": 1.3673,
+      "step": 29340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9032513958455666e-05,
+      "loss": 1.1711,
+      "step": 29345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9032189661046735e-05,
+      "loss": 1.6219,
+      "step": 29350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.903186531205915e-05,
+      "loss": 1.1862,
+      "step": 29355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9031540911494748e-05,
+      "loss": 0.9567,
+      "step": 29360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9031216459355395e-05,
+      "loss": 1.22,
+      "step": 29365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9030891955642933e-05,
+      "loss": 1.2813,
+      "step": 29370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9030567400359227e-05,
+      "loss": 1.2102,
+      "step": 29375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.903024279350612e-05,
+      "loss": 1.1797,
+      "step": 29380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9029918135085473e-05,
+      "loss": 1.551,
+      "step": 29385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9029593425099133e-05,
+      "loss": 1.3084,
+      "step": 29390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9029268663548958e-05,
+      "loss": 1.2905,
+      "step": 29395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9028943850436804e-05,
+      "loss": 1.4542,
+      "step": 29400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9028618985764525e-05,
+      "loss": 1.162,
+      "step": 29405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9028294069533977e-05,
+      "loss": 1.4777,
+      "step": 29410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.902796910174701e-05,
+      "loss": 1.1321,
+      "step": 29415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.902764408240549e-05,
+      "loss": 1.027,
+      "step": 29420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.902731901151126e-05,
+      "loss": 1.1424,
+      "step": 29425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9026993889066188e-05,
+      "loss": 1.0406,
+      "step": 29430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.902666871507212e-05,
+      "loss": 1.3405,
+      "step": 29435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9026343489530924e-05,
+      "loss": 1.2157,
+      "step": 29440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.902601821244445e-05,
+      "loss": 1.317,
+      "step": 29445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9025692883814555e-05,
+      "loss": 0.9764,
+      "step": 29450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9025367503643103e-05,
+      "loss": 1.602,
+      "step": 29455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.902504207193195e-05,
+      "loss": 1.2816,
+      "step": 29460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9024716588682946e-05,
+      "loss": 1.3936,
+      "step": 29465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.902439105389796e-05,
+      "loss": 1.074,
+      "step": 29470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9024065467578844e-05,
+      "loss": 1.3123,
+      "step": 29475
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9023739829727463e-05,
+      "loss": 1.3152,
+      "step": 29480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9023414140345675e-05,
+      "loss": 1.24,
+      "step": 29485
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9023088399435342e-05,
+      "loss": 1.0841,
+      "step": 29490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9022762606998314e-05,
+      "loss": 1.1535,
+      "step": 29495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9022436763036462e-05,
+      "loss": 1.1715,
+      "step": 29500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9022110867551646e-05,
+      "loss": 1.5131,
+      "step": 29505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.902178492054572e-05,
+      "loss": 1.1947,
+      "step": 29510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9021458922020553e-05,
+      "loss": 1.6167,
+      "step": 29515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9021132871978e-05,
+      "loss": 1.7485,
+      "step": 29520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.902080677041993e-05,
+      "loss": 1.3512,
+      "step": 29525
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9020480617348197e-05,
+      "loss": 1.0499,
+      "step": 29530
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9020154412764673e-05,
+      "loss": 1.2853,
+      "step": 29535
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9019828156671216e-05,
+      "loss": 1.5193,
+      "step": 29540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9019501849069685e-05,
+      "loss": 1.2418,
+      "step": 29545
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9019175489961947e-05,
+      "loss": 1.7867,
+      "step": 29550
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.901884907934987e-05,
+      "loss": 1.0925,
+      "step": 29555
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.901852261723531e-05,
+      "loss": 1.3431,
+      "step": 29560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.901819610362014e-05,
+      "loss": 1.8567,
+      "step": 29565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9017869538506212e-05,
+      "loss": 1.3164,
+      "step": 29570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9017542921895403e-05,
+      "loss": 1.0731,
+      "step": 29575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9017216253789576e-05,
+      "loss": 1.2743,
+      "step": 29580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.901688953419059e-05,
+      "loss": 1.4995,
+      "step": 29585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.901656276310032e-05,
+      "loss": 1.4641,
+      "step": 29590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9016235940520623e-05,
+      "loss": 1.1931,
+      "step": 29595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.901590906645337e-05,
+      "loss": 1.2019,
+      "step": 29600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.901558214090043e-05,
+      "loss": 1.1998,
+      "step": 29605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9015255163863664e-05,
+      "loss": 1.2354,
+      "step": 29610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9014928135344945e-05,
+      "loss": 1.254,
+      "step": 29615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9014601055346134e-05,
+      "loss": 1.3007,
+      "step": 29620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9014273923869103e-05,
+      "loss": 1.059,
+      "step": 29625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.901394674091572e-05,
+      "loss": 0.9539,
+      "step": 29630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9013619506487853e-05,
+      "loss": 1.3142,
+      "step": 29635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.901329222058737e-05,
+      "loss": 1.4629,
+      "step": 29640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9012964883216142e-05,
+      "loss": 1.2012,
+      "step": 29645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9012637494376037e-05,
+      "loss": 1.1044,
+      "step": 29650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9012310054068927e-05,
+      "loss": 1.2407,
+      "step": 29655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9011982562296674e-05,
+      "loss": 1.0008,
+      "step": 29660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.901165501906116e-05,
+      "loss": 1.3175,
+      "step": 29665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.901132742436424e-05,
+      "loss": 1.0799,
+      "step": 29670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.90109997782078e-05,
+      "loss": 1.0535,
+      "step": 29675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9010672080593702e-05,
+      "loss": 1.5323,
+      "step": 29680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.901034433152382e-05,
+      "loss": 1.3276,
+      "step": 29685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9010016531000025e-05,
+      "loss": 1.3033,
+      "step": 29690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 1.628,
+      "step": 29695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.900936077559819e-05,
+      "loss": 1.1608,
+      "step": 29700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9009032820723892e-05,
+      "loss": 1.1506,
+      "step": 29705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9008704814403172e-05,
+      "loss": 1.0357,
+      "step": 29710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.90083767566379e-05,
+      "loss": 1.4189,
+      "step": 29715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9008048647429955e-05,
+      "loss": 1.4868,
+      "step": 29720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9007720486781204e-05,
+      "loss": 1.5269,
+      "step": 29725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9007392274693525e-05,
+      "loss": 1.4838,
+      "step": 29730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9007064011168795e-05,
+      "loss": 1.3571,
+      "step": 29735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.900673569620888e-05,
+      "loss": 1.5499,
+      "step": 29740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.900640732981566e-05,
+      "loss": 1.1894,
+      "step": 29745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9006078911991014e-05,
+      "loss": 1.1519,
+      "step": 29750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.900575044273681e-05,
+      "loss": 1.3985,
+      "step": 29755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9005421922054927e-05,
+      "loss": 1.2769,
+      "step": 29760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9005093349947244e-05,
+      "loss": 1.2333,
+      "step": 29765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.900476472641563e-05,
+      "loss": 1.2178,
+      "step": 29770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9004436051461974e-05,
+      "loss": 1.134,
+      "step": 29775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.900410732508814e-05,
+      "loss": 1.127,
+      "step": 29780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.900377854729601e-05,
+      "loss": 1.4157,
+      "step": 29785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.900344971808746e-05,
+      "loss": 1.2513,
+      "step": 29790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9003120837464374e-05,
+      "loss": 1.0541,
+      "step": 29795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9002791905428624e-05,
+      "loss": 0.9843,
+      "step": 29800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.900246292198209e-05,
+      "loss": 1.2657,
+      "step": 29805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.900213388712665e-05,
+      "loss": 1.1517,
+      "step": 29810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9001804800864186e-05,
+      "loss": 1.3719,
+      "step": 29815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.900147566319657e-05,
+      "loss": 1.0954,
+      "step": 29820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9001146474125692e-05,
+      "loss": 1.1746,
+      "step": 29825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9000817233653424e-05,
+      "loss": 1.4419,
+      "step": 29830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.900048794178165e-05,
+      "loss": 1.0713,
+      "step": 29835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9000158598512247e-05,
+      "loss": 1.1229,
+      "step": 29840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8999829203847095e-05,
+      "loss": 1.2162,
+      "step": 29845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8999499757788083e-05,
+      "loss": 0.9385,
+      "step": 29850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8999170260337085e-05,
+      "loss": 1.399,
+      "step": 29855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8998840711495984e-05,
+      "loss": 1.1833,
+      "step": 29860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8998511111266663e-05,
+      "loss": 1.1997,
+      "step": 29865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8998181459651003e-05,
+      "loss": 1.4536,
+      "step": 29870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.899785175665089e-05,
+      "loss": 1.6128,
+      "step": 29875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.89975220022682e-05,
+      "loss": 1.3688,
+      "step": 29880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.899719219650482e-05,
+      "loss": 1.3968,
+      "step": 29885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8996862339362636e-05,
+      "loss": 1.288,
+      "step": 29890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.899653243084353e-05,
+      "loss": 1.451,
+      "step": 29895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8996202470949382e-05,
+      "loss": 1.0675,
+      "step": 29900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8995872459682082e-05,
+      "loss": 1.1576,
+      "step": 29905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.899554239704351e-05,
+      "loss": 1.4755,
+      "step": 29910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.899521228303555e-05,
+      "loss": 1.3065,
+      "step": 29915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8994882117660094e-05,
+      "loss": 0.9417,
+      "step": 29920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.899455190091902e-05,
+      "loss": 1.4266,
+      "step": 29925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.899422163281422e-05,
+      "loss": 1.1555,
+      "step": 29930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8993891313347577e-05,
+      "loss": 1.2243,
+      "step": 29935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8993560942520975e-05,
+      "loss": 1.291,
+      "step": 29940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8993230520336302e-05,
+      "loss": 1.318,
+      "step": 29945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.899290004679545e-05,
+      "loss": 1.2159,
+      "step": 29950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.89925695219003e-05,
+      "loss": 1.7674,
+      "step": 29955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8992238945652737e-05,
+      "loss": 1.068,
+      "step": 29960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.899190831805466e-05,
+      "loss": 1.6585,
+      "step": 29965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8991577639107944e-05,
+      "loss": 1.3974,
+      "step": 29970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8991246908814488e-05,
+      "loss": 1.2482,
+      "step": 29975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8990916127176175e-05,
+      "loss": 1.2655,
+      "step": 29980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8990585294194895e-05,
+      "loss": 1.4713,
+      "step": 29985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8990254409872536e-05,
+      "loss": 1.0248,
+      "step": 29990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.898992347421099e-05,
+      "loss": 1.2048,
+      "step": 29995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8989592487212145e-05,
+      "loss": 1.4103,
+      "step": 30000
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 1.2316739559173584,
+      "eval_runtime": 6462.631,
+      "eval_samples_per_second": 3.576,
+      "eval_steps_per_second": 1.788,
+      "step": 30000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8989261448877895e-05,
+      "loss": 1.2053,
+      "step": 30005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.898893035921012e-05,
+      "loss": 0.837,
+      "step": 30010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8988599218210728e-05,
+      "loss": 1.0792,
+      "step": 30015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8988268025881595e-05,
+      "loss": 1.3862,
+      "step": 30020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8987936782224613e-05,
+      "loss": 1.173,
+      "step": 30025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8987605487241684e-05,
+      "loss": 1.1851,
+      "step": 30030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.898727414093469e-05,
+      "loss": 1.3907,
+      "step": 30035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.898694274330553e-05,
+      "loss": 1.348,
+      "step": 30040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.898661129435609e-05,
+      "loss": 1.0469,
+      "step": 30045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.898627979408827e-05,
+      "loss": 1.1818,
+      "step": 30050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8985948242503952e-05,
+      "loss": 1.4551,
+      "step": 30055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8985616639605045e-05,
+      "loss": 1.2521,
+      "step": 30060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.898528498539343e-05,
+      "loss": 1.0695,
+      "step": 30065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8984953279871002e-05,
+      "loss": 1.1091,
+      "step": 30070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8984621523039662e-05,
+      "loss": 1.2875,
+      "step": 30075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.89842897149013e-05,
+      "loss": 1.4828,
+      "step": 30080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8983957855457814e-05,
+      "loss": 1.1929,
+      "step": 30085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.898362594471109e-05,
+      "loss": 1.1033,
+      "step": 30090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8983293982663037e-05,
+      "loss": 1.3595,
+      "step": 30095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.898296196931554e-05,
+      "loss": 1.2816,
+      "step": 30100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8982629904670496e-05,
+      "loss": 1.3897,
+      "step": 30105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.898229778872981e-05,
+      "loss": 1.1114,
+      "step": 30110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8981965621495368e-05,
+      "loss": 0.9817,
+      "step": 30115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8981633402969075e-05,
+      "loss": 1.0541,
+      "step": 30120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8981301133152824e-05,
+      "loss": 1.2932,
+      "step": 30125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.898096881204851e-05,
+      "loss": 1.2442,
+      "step": 30130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8980636439658036e-05,
+      "loss": 1.6035,
+      "step": 30135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8980304015983294e-05,
+      "loss": 1.111,
+      "step": 30140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897997154102619e-05,
+      "loss": 1.1862,
+      "step": 30145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8979639014788616e-05,
+      "loss": 1.5339,
+      "step": 30150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8979306437272478e-05,
+      "loss": 1.2255,
+      "step": 30155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8978973808479665e-05,
+      "loss": 1.0102,
+      "step": 30160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8978641128412086e-05,
+      "loss": 1.323,
+      "step": 30165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8978308397071637e-05,
+      "loss": 1.3996,
+      "step": 30170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8977975614460216e-05,
+      "loss": 1.1338,
+      "step": 30175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8977642780579727e-05,
+      "loss": 1.4059,
+      "step": 30180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8977309895432068e-05,
+      "loss": 1.2135,
+      "step": 30185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897697695901914e-05,
+      "loss": 1.0809,
+      "step": 30190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8976643971342847e-05,
+      "loss": 1.0529,
+      "step": 30195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897631093240509e-05,
+      "loss": 1.4772,
+      "step": 30200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8975977842207766e-05,
+      "loss": 0.9424,
+      "step": 30205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8975644700752785e-05,
+      "loss": 1.3164,
+      "step": 30210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8975311508042042e-05,
+      "loss": 1.2583,
+      "step": 30215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8974978264077443e-05,
+      "loss": 1.3084,
+      "step": 30220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8974644968860893e-05,
+      "loss": 1.2671,
+      "step": 30225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8974311622394293e-05,
+      "loss": 1.163,
+      "step": 30230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8973978224679543e-05,
+      "loss": 1.2914,
+      "step": 30235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8973644775718554e-05,
+      "loss": 1.3231,
+      "step": 30240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8973311275513228e-05,
+      "loss": 1.6376,
+      "step": 30245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8972977724065464e-05,
+      "loss": 1.3018,
+      "step": 30250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8972644121377174e-05,
+      "loss": 1.1792,
+      "step": 30255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897231046745026e-05,
+      "loss": 1.2414,
+      "step": 30260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897197676228663e-05,
+      "loss": 0.9882,
+      "step": 30265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897164300588818e-05,
+      "loss": 1.1519,
+      "step": 30270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897130919825683e-05,
+      "loss": 1.1907,
+      "step": 30275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8970975339394474e-05,
+      "loss": 1.1386,
+      "step": 30280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8970641429303026e-05,
+      "loss": 1.4081,
+      "step": 30285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897030746798439e-05,
+      "loss": 1.2052,
+      "step": 30290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8969973455440477e-05,
+      "loss": 1.2976,
+      "step": 30295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896963939167319e-05,
+      "loss": 1.4727,
+      "step": 30300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8969305276684436e-05,
+      "loss": 1.4341,
+      "step": 30305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8968971110476124e-05,
+      "loss": 1.1792,
+      "step": 30310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8968636893050165e-05,
+      "loss": 1.1995,
+      "step": 30315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8968302624408465e-05,
+      "loss": 1.3281,
+      "step": 30320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8967968304552933e-05,
+      "loss": 1.418,
+      "step": 30325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8967633933485475e-05,
+      "loss": 1.3392,
+      "step": 30330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896729951120801e-05,
+      "loss": 1.1061,
+      "step": 30335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896696503772244e-05,
+      "loss": 1.3202,
+      "step": 30340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8966630513030676e-05,
+      "loss": 1.1982,
+      "step": 30345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8966295937134628e-05,
+      "loss": 1.2628,
+      "step": 30350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8965961310036208e-05,
+      "loss": 1.084,
+      "step": 30355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896562663173733e-05,
+      "loss": 1.0974,
+      "step": 30360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8965291902239897e-05,
+      "loss": 1.0273,
+      "step": 30365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8964957121545827e-05,
+      "loss": 1.3558,
+      "step": 30370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896462228965703e-05,
+      "loss": 1.0101,
+      "step": 30375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896428740657542e-05,
+      "loss": 1.1837,
+      "step": 30380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8963952472302904e-05,
+      "loss": 1.6073,
+      "step": 30385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8963617486841398e-05,
+      "loss": 1.2266,
+      "step": 30390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896328245019282e-05,
+      "loss": 1.4586,
+      "step": 30395
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896294736235907e-05,
+      "loss": 1.1049,
+      "step": 30400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8962612223342076e-05,
+      "loss": 1.5734,
+      "step": 30405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8962277033143745e-05,
+      "loss": 1.1715,
+      "step": 30410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8961941791765993e-05,
+      "loss": 1.0708,
+      "step": 30415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896160649921073e-05,
+      "loss": 1.4169,
+      "step": 30420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8961271155479875e-05,
+      "loss": 1.4287,
+      "step": 30425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896093576057534e-05,
+      "loss": 1.5781,
+      "step": 30430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.896060031449905e-05,
+      "loss": 1.2006,
+      "step": 30435
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8960264817252906e-05,
+      "loss": 1.1365,
+      "step": 30440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.895992926883883e-05,
+      "loss": 1.2403,
+      "step": 30445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.895959366925874e-05,
+      "loss": 1.2332,
+      "step": 30450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8959258018514553e-05,
+      "loss": 1.7017,
+      "step": 30455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8958922316608182e-05,
+      "loss": 1.2107,
+      "step": 30460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8958586563541547e-05,
+      "loss": 1.2669,
+      "step": 30465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8958250759316564e-05,
+      "loss": 1.1488,
+      "step": 30470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8957914903935152e-05,
+      "loss": 1.2311,
+      "step": 30475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8957578997399225e-05,
+      "loss": 1.2608,
+      "step": 30480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8957243039710707e-05,
+      "loss": 1.2675,
+      "step": 30485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8956907030871512e-05,
+      "loss": 1.3262,
+      "step": 30490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8956570970883562e-05,
+      "loss": 1.1247,
+      "step": 30495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.895623485974877e-05,
+      "loss": 1.0188,
+      "step": 30500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8955898697469064e-05,
+      "loss": 1.3409,
+      "step": 30505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.895556248404636e-05,
+      "loss": 1.2203,
+      "step": 30510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8955226219482576e-05,
+      "loss": 1.9928,
+      "step": 30515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8954889903779632e-05,
+      "loss": 1.0994,
+      "step": 30520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.895455353693945e-05,
+      "loss": 1.0996,
+      "step": 30525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8954217118963957e-05,
+      "loss": 1.5469,
+      "step": 30530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.895388064985506e-05,
+      "loss": 1.441,
+      "step": 30535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8953544129614697e-05,
+      "loss": 1.5696,
+      "step": 30540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8953207558244778e-05,
+      "loss": 1.1249,
+      "step": 30545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8952870935747225e-05,
+      "loss": 1.8314,
+      "step": 30550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8952534262123966e-05,
+      "loss": 1.0563,
+      "step": 30555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8952197537376922e-05,
+      "loss": 1.0796,
+      "step": 30560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8951860761508015e-05,
+      "loss": 1.272,
+      "step": 30565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8951523934519168e-05,
+      "loss": 1.4335,
+      "step": 30570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8951187056412303e-05,
+      "loss": 1.1418,
+      "step": 30575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8950850127189348e-05,
+      "loss": 1.704,
+      "step": 30580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8950513146852223e-05,
+      "loss": 1.1833,
+      "step": 30585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8950176115402853e-05,
+      "loss": 1.4834,
+      "step": 30590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8949839032843167e-05,
+      "loss": 1.084,
+      "step": 30595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.894950189917508e-05,
+      "loss": 1.3441,
+      "step": 30600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8949164714400533e-05,
+      "loss": 1.0976,
+      "step": 30605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8948827478521436e-05,
+      "loss": 1.1588,
+      "step": 30610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8948490191539722e-05,
+      "loss": 1.3533,
+      "step": 30615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8948152853457314e-05,
+      "loss": 0.9185,
+      "step": 30620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8947815464276146e-05,
+      "loss": 1.5347,
+      "step": 30625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8947478023998135e-05,
+      "loss": 1.3195,
+      "step": 30630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8947140532625213e-05,
+      "loss": 1.2101,
+      "step": 30635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8946802990159305e-05,
+      "loss": 1.2127,
+      "step": 30640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.894646539660234e-05,
+      "loss": 1.3175,
+      "step": 30645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8946127751956245e-05,
+      "loss": 1.1781,
+      "step": 30650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.894579005622295e-05,
+      "loss": 0.8339,
+      "step": 30655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8945452309404384e-05,
+      "loss": 1.0312,
+      "step": 30660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8945114511502473e-05,
+      "loss": 1.0567,
+      "step": 30665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8944776662519145e-05,
+      "loss": 1.1815,
+      "step": 30670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8944438762456333e-05,
+      "loss": 1.1453,
+      "step": 30675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8944100811315967e-05,
+      "loss": 1.112,
+      "step": 30680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8943762809099972e-05,
+      "loss": 1.3712,
+      "step": 30685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8943424755810282e-05,
+      "loss": 0.9962,
+      "step": 30690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8943086651448825e-05,
+      "loss": 1.398,
+      "step": 30695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8942748496017532e-05,
+      "loss": 1.3717,
+      "step": 30700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.894241028951834e-05,
+      "loss": 1.2563,
+      "step": 30705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8942072031953172e-05,
+      "loss": 1.2063,
+      "step": 30710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8941733723323964e-05,
+      "loss": 0.8603,
+      "step": 30715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8941395363632648e-05,
+      "loss": 1.2843,
+      "step": 30720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8941056952881153e-05,
+      "loss": 1.578,
+      "step": 30725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8940718491071416e-05,
+      "loss": 1.5571,
+      "step": 30730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8940379978205366e-05,
+      "loss": 1.5062,
+      "step": 30735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.894004141428494e-05,
+      "loss": 1.1257,
+      "step": 30740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8939702799312066e-05,
+      "loss": 1.1922,
+      "step": 30745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8939364133288686e-05,
+      "loss": 1.1082,
+      "step": 30750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8939025416216725e-05,
+      "loss": 1.4494,
+      "step": 30755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893868664809812e-05,
+      "loss": 1.1342,
+      "step": 30760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8938347828934807e-05,
+      "loss": 0.9668,
+      "step": 30765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893800895872872e-05,
+      "loss": 1.3681,
+      "step": 30770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8937670037481795e-05,
+      "loss": 1.3799,
+      "step": 30775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893733106519597e-05,
+      "loss": 0.8463,
+      "step": 30780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8936992041873175e-05,
+      "loss": 1.111,
+      "step": 30785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893665296751535e-05,
+      "loss": 1.1734,
+      "step": 30790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8936313842124428e-05,
+      "loss": 1.0423,
+      "step": 30795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8935974665702348e-05,
+      "loss": 1.2378,
+      "step": 30800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8935635438251047e-05,
+      "loss": 1.0086,
+      "step": 30805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8935296159772464e-05,
+      "loss": 1.2685,
+      "step": 30810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893495683026853e-05,
+      "loss": 1.1242,
+      "step": 30815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893461744974119e-05,
+      "loss": 1.4268,
+      "step": 30820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8934278018192373e-05,
+      "loss": 1.4275,
+      "step": 30825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8933938535624028e-05,
+      "loss": 1.5639,
+      "step": 30830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893359900203809e-05,
+      "loss": 1.0611,
+      "step": 30835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8933259417436493e-05,
+      "loss": 1.5689,
+      "step": 30840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8932919781821183e-05,
+      "loss": 1.2507,
+      "step": 30845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8932580095194097e-05,
+      "loss": 0.9685,
+      "step": 30850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8932240357557172e-05,
+      "loss": 1.378,
+      "step": 30855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8931900568912353e-05,
+      "loss": 1.2628,
+      "step": 30860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8931560729261572e-05,
+      "loss": 0.9333,
+      "step": 30865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893122083860678e-05,
+      "loss": 1.0053,
+      "step": 30870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8930880896949914e-05,
+      "loss": 1.3756,
+      "step": 30875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893054090429291e-05,
+      "loss": 1.3102,
+      "step": 30880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8930200860637716e-05,
+      "loss": 1.4218,
+      "step": 30885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8929860765986274e-05,
+      "loss": 0.9889,
+      "step": 30890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8929520620340524e-05,
+      "loss": 1.5326,
+      "step": 30895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8929180423702408e-05,
+      "loss": 1.5472,
+      "step": 30900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8928840176073867e-05,
+      "loss": 1.0711,
+      "step": 30905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.892849987745685e-05,
+      "loss": 1.29,
+      "step": 30910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8928159527853295e-05,
+      "loss": 0.9934,
+      "step": 30915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8927819127265146e-05,
+      "loss": 1.2213,
+      "step": 30920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8927478675694348e-05,
+      "loss": 1.5451,
+      "step": 30925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8927138173142846e-05,
+      "loss": 1.0976,
+      "step": 30930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.892679761961258e-05,
+      "loss": 0.9957,
+      "step": 30935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8926457015105503e-05,
+      "loss": 1.1465,
+      "step": 30940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8926116359623554e-05,
+      "loss": 1.2148,
+      "step": 30945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8925775653168677e-05,
+      "loss": 1.2904,
+      "step": 30950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8925434895742823e-05,
+      "loss": 1.4914,
+      "step": 30955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8925094087347938e-05,
+      "loss": 1.2713,
+      "step": 30960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8924753227985962e-05,
+      "loss": 1.3322,
+      "step": 30965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8924412317658844e-05,
+      "loss": 1.1996,
+      "step": 30970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8924071356368534e-05,
+      "loss": 1.2493,
+      "step": 30975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8923730344116976e-05,
+      "loss": 1.2963,
+      "step": 30980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8923389280906116e-05,
+      "loss": 1.236,
+      "step": 30985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8923048166737907e-05,
+      "loss": 1.3326,
+      "step": 30990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8922707001614292e-05,
+      "loss": 1.2464,
+      "step": 30995
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.892236578553722e-05,
+      "loss": 1.1215,
+      "step": 31000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8922024518508642e-05,
+      "loss": 1.2728,
+      "step": 31005
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8921683200530502e-05,
+      "loss": 1.0631,
+      "step": 31010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8921341831604757e-05,
+      "loss": 1.4044,
+      "step": 31015
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.892100041173335e-05,
+      "loss": 1.6322,
+      "step": 31020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.892065894091823e-05,
+      "loss": 0.966,
+      "step": 31025
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8920317419161354e-05,
+      "loss": 1.3304,
+      "step": 31030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8919975846464663e-05,
+      "loss": 1.5073,
+      "step": 31035
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8919634222830117e-05,
+      "loss": 1.1677,
+      "step": 31040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.891929254825966e-05,
+      "loss": 1.2411,
+      "step": 31045
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8918950822755244e-05,
+      "loss": 1.2682,
+      "step": 31050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8918609046318824e-05,
+      "loss": 1.3387,
+      "step": 31055
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.891826721895235e-05,
+      "loss": 0.9734,
+      "step": 31060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8917925340657768e-05,
+      "loss": 1.2052,
+      "step": 31065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.891758341143704e-05,
+      "loss": 1.1885,
+      "step": 31070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8917241431292115e-05,
+      "loss": 1.1813,
+      "step": 31075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8916899400224944e-05,
+      "loss": 1.3184,
+      "step": 31080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.891655731823748e-05,
+      "loss": 0.9753,
+      "step": 31085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.891621518533168e-05,
+      "loss": 1.5485,
+      "step": 31090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8915873001509496e-05,
+      "loss": 1.1839,
+      "step": 31095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8915530766772883e-05,
+      "loss": 1.1377,
+      "step": 31100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8915188481123793e-05,
+      "loss": 1.1291,
+      "step": 31105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.891484614456418e-05,
+      "loss": 1.1681,
+      "step": 31110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8914503757096e-05,
+      "loss": 1.5317,
+      "step": 31115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8914161318721208e-05,
+      "loss": 1.1167,
+      "step": 31120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8913818829441765e-05,
+      "loss": 1.0717,
+      "step": 31125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.891347628925962e-05,
+      "loss": 1.0664,
+      "step": 31130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.891313369817673e-05,
+      "loss": 1.2943,
+      "step": 31135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8912791056195054e-05,
+      "loss": 1.3325,
+      "step": 31140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8912448363316548e-05,
+      "loss": 1.072,
+      "step": 31145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8912105619543166e-05,
+      "loss": 1.0813,
+      "step": 31150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.891176282487687e-05,
+      "loss": 1.1633,
+      "step": 31155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8911419979319614e-05,
+      "loss": 0.989,
+      "step": 31160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8911077082873358e-05,
+      "loss": 1.0045,
+      "step": 31165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8910734135540055e-05,
+      "loss": 1.4691,
+      "step": 31170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8910391137321672e-05,
+      "loss": 1.2058,
+      "step": 31175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.891004808822016e-05,
+      "loss": 1.109,
+      "step": 31180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8909704988237484e-05,
+      "loss": 1.199,
+      "step": 31185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8909361837375598e-05,
+      "loss": 1.195,
+      "step": 31190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8909018635636465e-05,
+      "loss": 1.2701,
+      "step": 31195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8908675383022043e-05,
+      "loss": 1.1538,
+      "step": 31200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8908332079534294e-05,
+      "loss": 1.7645,
+      "step": 31205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8907988725175175e-05,
+      "loss": 1.1528,
+      "step": 31210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8907645319946653e-05,
+      "loss": 1.2307,
+      "step": 31215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8907301863850685e-05,
+      "loss": 1.1395,
+      "step": 31220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.890695835688923e-05,
+      "loss": 1.1505,
+      "step": 31225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8906614799064256e-05,
+      "loss": 1.3204,
+      "step": 31230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8906271190377715e-05,
+      "loss": 1.6013,
+      "step": 31235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.890592753083158e-05,
+      "loss": 1.0968,
+      "step": 31240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8905583820427806e-05,
+      "loss": 1.5917,
+      "step": 31245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.890524005916836e-05,
+      "loss": 1.3457,
+      "step": 31250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8904896247055206e-05,
+      "loss": 1.3544,
+      "step": 31255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8904552384090302e-05,
+      "loss": 1.367,
+      "step": 31260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8904208470275612e-05,
+      "loss": 1.397,
+      "step": 31265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8903864505613108e-05,
+      "loss": 1.224,
+      "step": 31270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8903520490104747e-05,
+      "loss": 1.3836,
+      "step": 31275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.890317642375249e-05,
+      "loss": 1.3527,
+      "step": 31280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8902832306558316e-05,
+      "loss": 1.5588,
+      "step": 31285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8902488138524173e-05,
+      "loss": 0.953,
+      "step": 31290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8902143919652037e-05,
+      "loss": 1.109,
+      "step": 31295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8901799649943874e-05,
+      "loss": 1.2806,
+      "step": 31300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8901455329401646e-05,
+      "loss": 1.5119,
+      "step": 31305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8901110958027316e-05,
+      "loss": 1.228,
+      "step": 31310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8900766535822862e-05,
+      "loss": 1.2912,
+      "step": 31315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.890042206279024e-05,
+      "loss": 1.2681,
+      "step": 31320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.890007753893142e-05,
+      "loss": 1.2694,
+      "step": 31325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.889973296424837e-05,
+      "loss": 1.3704,
+      "step": 31330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.889938833874306e-05,
+      "loss": 1.4338,
+      "step": 31335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.889904366241746e-05,
+      "loss": 1.1357,
+      "step": 31340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8898698935273526e-05,
+      "loss": 1.3555,
+      "step": 31345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8898354157313242e-05,
+      "loss": 1.2177,
+      "step": 31350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8898009328538567e-05,
+      "loss": 1.0978,
+      "step": 31355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8897664448951472e-05,
+      "loss": 1.2899,
+      "step": 31360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8897319518553928e-05,
+      "loss": 1.439,
+      "step": 31365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8896974537347904e-05,
+      "loss": 1.5226,
+      "step": 31370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8896629505335368e-05,
+      "loss": 1.3144,
+      "step": 31375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8896284422518296e-05,
+      "loss": 1.128,
+      "step": 31380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8895939288898652e-05,
+      "loss": 1.2396,
+      "step": 31385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8895594104478413e-05,
+      "loss": 1.2381,
+      "step": 31390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8895248869259545e-05,
+      "loss": 1.0755,
+      "step": 31395
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8894903583244024e-05,
+      "loss": 1.5036,
+      "step": 31400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8894558246433816e-05,
+      "loss": 1.4454,
+      "step": 31405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.88942128588309e-05,
+      "loss": 0.9908,
+      "step": 31410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8893867420437243e-05,
+      "loss": 1.1083,
+      "step": 31415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8893521931254818e-05,
+      "loss": 1.2519,
+      "step": 31420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.88931763912856e-05,
+      "loss": 1.303,
+      "step": 31425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8892830800531565e-05,
+      "loss": 1.215,
+      "step": 31430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.889248515899468e-05,
+      "loss": 1.3932,
+      "step": 31435
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.889213946667692e-05,
+      "loss": 1.0678,
+      "step": 31440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.889179372358027e-05,
+      "loss": 1.4707,
+      "step": 31445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8891447929706685e-05,
+      "loss": 1.08,
+      "step": 31450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8891102085058156e-05,
+      "loss": 1.2582,
+      "step": 31455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.889075618963665e-05,
+      "loss": 1.3403,
+      "step": 31460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8890410243444142e-05,
+      "loss": 1.1574,
+      "step": 31465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8890064246482612e-05,
+      "loss": 1.1332,
+      "step": 31470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8889718198754034e-05,
+      "loss": 1.2688,
+      "step": 31475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8889372100260382e-05,
+      "loss": 1.1589,
+      "step": 31480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8889025951003636e-05,
+      "loss": 1.0584,
+      "step": 31485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.888867975098577e-05,
+      "loss": 1.2514,
+      "step": 31490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.888833350020876e-05,
+      "loss": 1.2077,
+      "step": 31495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.888798719867459e-05,
+      "loss": 1.4054,
+      "step": 31500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.888764084638523e-05,
+      "loss": 1.5164,
+      "step": 31505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.888729444334266e-05,
+      "loss": 0.8417,
+      "step": 31510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.888694798954886e-05,
+      "loss": 1.2557,
+      "step": 31515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8886601485005803e-05,
+      "loss": 1.1777,
+      "step": 31520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8886254929715475e-05,
+      "loss": 1.1837,
+      "step": 31525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8885908323679855e-05,
+      "loss": 1.3929,
+      "step": 31530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8885561666900916e-05,
+      "loss": 1.2702,
+      "step": 31535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.888521495938064e-05,
+      "loss": 1.0289,
+      "step": 31540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8884868201121006e-05,
+      "loss": 1.4404,
+      "step": 31545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8884521392124e-05,
+      "loss": 1.2955,
+      "step": 31550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.88841745323916e-05,
+      "loss": 1.1803,
+      "step": 31555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8883827621925777e-05,
+      "loss": 1.0261,
+      "step": 31560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8883480660728523e-05,
+      "loss": 1.437,
+      "step": 31565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.888313364880182e-05,
+      "loss": 1.3071,
+      "step": 31570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8882786586147643e-05,
+      "loss": 1.0039,
+      "step": 31575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.888243947276798e-05,
+      "loss": 1.0724,
+      "step": 31580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8882092308664808e-05,
+      "loss": 1.5995,
+      "step": 31585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8881745093840113e-05,
+      "loss": 1.3015,
+      "step": 31590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8881397828295874e-05,
+      "loss": 1.1294,
+      "step": 31595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8881050512034077e-05,
+      "loss": 1.352,
+      "step": 31600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8880703145056708e-05,
+      "loss": 1.3777,
+      "step": 31605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8880355727365745e-05,
+      "loss": 1.1039,
+      "step": 31610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8880008258963174e-05,
+      "loss": 1.2044,
+      "step": 31615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8879660739850982e-05,
+      "loss": 1.4694,
+      "step": 31620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8879313170031147e-05,
+      "loss": 1.2452,
+      "step": 31625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8878965549505663e-05,
+      "loss": 1.5355,
+      "step": 31630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.887861787827651e-05,
+      "loss": 0.9386,
+      "step": 31635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.887827015634567e-05,
+      "loss": 1.4844,
+      "step": 31640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8877922383715132e-05,
+      "loss": 1.1335,
+      "step": 31645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8877574560386884e-05,
+      "loss": 1.0904,
+      "step": 31650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.887722668636291e-05,
+      "loss": 1.117,
+      "step": 31655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8876878761645198e-05,
+      "loss": 1.087,
+      "step": 31660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8876530786235734e-05,
+      "loss": 1.1839,
+      "step": 31665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8876182760136503e-05,
+      "loss": 1.4848,
+      "step": 31670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8875834683349495e-05,
+      "loss": 1.2662,
+      "step": 31675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.88754865558767e-05,
+      "loss": 1.1119,
+      "step": 31680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8875138377720098e-05,
+      "loss": 1.2716,
+      "step": 31685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8874790148881684e-05,
+      "loss": 1.1221,
+      "step": 31690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8874441869363445e-05,
+      "loss": 1.312,
+      "step": 31695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8874093539167374e-05,
+      "loss": 1.1045,
+      "step": 31700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.887374515829545e-05,
+      "loss": 1.0952,
+      "step": 31705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8873396726749675e-05,
+      "loss": 1.2723,
+      "step": 31710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8873048244532027e-05,
+      "loss": 1.3447,
+      "step": 31715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.88726997116445e-05,
+      "loss": 1.3781,
+      "step": 31720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.887235112808909e-05,
+      "loss": 1.2488,
+      "step": 31725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8872002493867778e-05,
+      "loss": 1.0374,
+      "step": 31730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8871653808982563e-05,
+      "loss": 1.323,
+      "step": 31735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8871305073435433e-05,
+      "loss": 1.2503,
+      "step": 31740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.887095628722838e-05,
+      "loss": 1.4176,
+      "step": 31745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8870607450363394e-05,
+      "loss": 0.9931,
+      "step": 31750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8870258562842466e-05,
+      "loss": 1.0782,
+      "step": 31755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8869909624667596e-05,
+      "loss": 1.2688,
+      "step": 31760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.886956063584077e-05,
+      "loss": 1.3032,
+      "step": 31765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.886921159636398e-05,
+      "loss": 1.4657,
+      "step": 31770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8868862506239223e-05,
+      "loss": 1.2017,
+      "step": 31775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8868513365468492e-05,
+      "loss": 1.1904,
+      "step": 31780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8868164174053776e-05,
+      "loss": 1.3472,
+      "step": 31785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8867814931997078e-05,
+      "loss": 0.9465,
+      "step": 31790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8867465639300383e-05,
+      "loss": 1.3381,
+      "step": 31795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8867116295965693e-05,
+      "loss": 0.9742,
+      "step": 31800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8866766901994995e-05,
+      "loss": 1.4267,
+      "step": 31805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8866417457390293e-05,
+      "loss": 1.1846,
+      "step": 31810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8866067962153577e-05,
+      "loss": 1.3138,
+      "step": 31815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8865718416286848e-05,
+      "loss": 1.4402,
+      "step": 31820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8865368819792093e-05,
+      "loss": 1.7915,
+      "step": 31825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8865019172671317e-05,
+      "loss": 1.3229,
+      "step": 31830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8864669474926513e-05,
+      "loss": 1.1315,
+      "step": 31835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8864319726559675e-05,
+      "loss": 1.0843,
+      "step": 31840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8863969927572807e-05,
+      "loss": 1.0972,
+      "step": 31845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8863620077967904e-05,
+      "loss": 1.34,
+      "step": 31850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.886327017774696e-05,
+      "loss": 0.9285,
+      "step": 31855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8862920226911978e-05,
+      "loss": 1.266,
+      "step": 31860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8862570225464957e-05,
+      "loss": 1.1459,
+      "step": 31865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.886222017340789e-05,
+      "loss": 1.3233,
+      "step": 31870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8861870070742777e-05,
+      "loss": 1.3358,
+      "step": 31875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8861519917471625e-05,
+      "loss": 1.1241,
+      "step": 31880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8861169713596425e-05,
+      "loss": 0.9524,
+      "step": 31885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8860819459119182e-05,
+      "loss": 1.3582,
+      "step": 31890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8860469154041888e-05,
+      "loss": 1.4105,
+      "step": 31895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8860118798366556e-05,
+      "loss": 1.474,
+      "step": 31900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8859768392095178e-05,
+      "loss": 1.1511,
+      "step": 31905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.885941793522976e-05,
+      "loss": 1.3362,
+      "step": 31910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8859067427772296e-05,
+      "loss": 1.4382,
+      "step": 31915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8858716869724794e-05,
+      "loss": 1.3228,
+      "step": 31920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8858366261089255e-05,
+      "loss": 1.1413,
+      "step": 31925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.885801560186768e-05,
+      "loss": 1.2064,
+      "step": 31930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.885766489206207e-05,
+      "loss": 1.1965,
+      "step": 31935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8857314131674428e-05,
+      "loss": 0.9583,
+      "step": 31940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8856963320706763e-05,
+      "loss": 1.1233,
+      "step": 31945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.885661245916107e-05,
+      "loss": 1.3322,
+      "step": 31950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8856261547039357e-05,
+      "loss": 1.5533,
+      "step": 31955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.885591058434363e-05,
+      "loss": 1.351,
+      "step": 31960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.885555957107589e-05,
+      "loss": 1.1442,
+      "step": 31965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.885520850723814e-05,
+      "loss": 1.0346,
+      "step": 31970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8854857392832384e-05,
+      "loss": 0.9811,
+      "step": 31975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8854506227860635e-05,
+      "loss": 1.508,
+      "step": 31980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.885415501232489e-05,
+      "loss": 1.5006,
+      "step": 31985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8853803746227157e-05,
+      "loss": 1.4892,
+      "step": 31990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8853452429569446e-05,
+      "loss": 1.2689,
+      "step": 31995
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8853101062353755e-05,
+      "loss": 1.7033,
+      "step": 32000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.88527496445821e-05,
+      "loss": 1.1096,
+      "step": 32005
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8852398176256478e-05,
+      "loss": 1.0646,
+      "step": 32010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8852046657378907e-05,
+      "loss": 1.2079,
+      "step": 32015
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8851695087951382e-05,
+      "loss": 1.2769,
+      "step": 32020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.885134346797592e-05,
+      "loss": 1.3646,
+      "step": 32025
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8850991797454528e-05,
+      "loss": 1.0732,
+      "step": 32030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.885064007638921e-05,
+      "loss": 1.2307,
+      "step": 32035
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.885028830478198e-05,
+      "loss": 0.9444,
+      "step": 32040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8849936482634842e-05,
+      "loss": 1.2425,
+      "step": 32045
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8849584609949804e-05,
+      "loss": 1.5381,
+      "step": 32050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8849232686728882e-05,
+      "loss": 1.6613,
+      "step": 32055
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8848880712974076e-05,
+      "loss": 1.5717,
+      "step": 32060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.884852868868741e-05,
+      "loss": 0.9801,
+      "step": 32065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.884817661387088e-05,
+      "loss": 1.1675,
+      "step": 32070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8847824488526505e-05,
+      "loss": 1.4403,
+      "step": 32075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8847472312656292e-05,
+      "loss": 0.9908,
+      "step": 32080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8847120086262255e-05,
+      "loss": 1.4026,
+      "step": 32085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8846767809346402e-05,
+      "loss": 1.0396,
+      "step": 32090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8846415481910746e-05,
+      "loss": 1.0205,
+      "step": 32095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8846063103957303e-05,
+      "loss": 1.0724,
+      "step": 32100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.884571067548808e-05,
+      "loss": 1.512,
+      "step": 32105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.884535819650509e-05,
+      "loss": 1.0334,
+      "step": 32110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.884500566701035e-05,
+      "loss": 1.3458,
+      "step": 32115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.884465308700587e-05,
+      "loss": 1.295,
+      "step": 32120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.884430045649366e-05,
+      "loss": 1.0858,
+      "step": 32125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.884394777547574e-05,
+      "loss": 1.2197,
+      "step": 32130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8843595043954122e-05,
+      "loss": 1.1731,
+      "step": 32135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8843242261930818e-05,
+      "loss": 1.2022,
+      "step": 32140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8842889429407844e-05,
+      "loss": 1.1859,
+      "step": 32145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.884253654638722e-05,
+      "loss": 0.9833,
+      "step": 32150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.884218361287095e-05,
+      "loss": 1.4764,
+      "step": 32155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.884183062886106e-05,
+      "loss": 1.1838,
+      "step": 32160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.884147759435956e-05,
+      "loss": 1.0737,
+      "step": 32165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8841124509368468e-05,
+      "loss": 1.4885,
+      "step": 32170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.88407713738898e-05,
+      "loss": 1.3362,
+      "step": 32175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.884041818792557e-05,
+      "loss": 1.4664,
+      "step": 32180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8840064951477803e-05,
+      "loss": 1.1534,
+      "step": 32185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8839711664548507e-05,
+      "loss": 1.2198,
+      "step": 32190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.88393583271397e-05,
+      "loss": 1.5163,
+      "step": 32195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.883900493925341e-05,
+      "loss": 1.1997,
+      "step": 32200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.883865150089164e-05,
+      "loss": 1.0418,
+      "step": 32205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.883829801205642e-05,
+      "loss": 1.3178,
+      "step": 32210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8837944472749764e-05,
+      "loss": 1.5232,
+      "step": 32215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883759088297369e-05,
+      "loss": 1.2053,
+      "step": 32220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883723724273022e-05,
+      "loss": 1.2047,
+      "step": 32225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8836883552021373e-05,
+      "loss": 1.0607,
+      "step": 32230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8836529810849168e-05,
+      "loss": 1.2602,
+      "step": 32235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8836176019215625e-05,
+      "loss": 1.2364,
+      "step": 32240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8835822177122765e-05,
+      "loss": 1.6478,
+      "step": 32245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8835468284572606e-05,
+      "loss": 1.0433,
+      "step": 32250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883511434156717e-05,
+      "loss": 1.3286,
+      "step": 32255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883476034810848e-05,
+      "loss": 1.4877,
+      "step": 32260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8834406304198558e-05,
+      "loss": 1.1573,
+      "step": 32265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8834052209839428e-05,
+      "loss": 1.2942,
+      "step": 32270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8833698065033103e-05,
+      "loss": 1.3734,
+      "step": 32275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8833343869781612e-05,
+      "loss": 1.1361,
+      "step": 32280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8832989624086978e-05,
+      "loss": 1.5331,
+      "step": 32285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8832635327951217e-05,
+      "loss": 1.1904,
+      "step": 32290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8832280981376362e-05,
+      "loss": 1.0253,
+      "step": 32295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883192658436443e-05,
+      "loss": 1.1977,
+      "step": 32300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.883157213691745e-05,
+      "loss": 1.2078,
+      "step": 32305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8831217639037444e-05,
+      "loss": 1.4585,
+      "step": 32310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8830863090726433e-05,
+      "loss": 1.4213,
+      "step": 32315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8830508491986444e-05,
+      "loss": 1.1015,
+      "step": 32320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8830153842819503e-05,
+      "loss": 1.1296,
+      "step": 32325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882979914322763e-05,
+      "loss": 1.2765,
+      "step": 32330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882944439321286e-05,
+      "loss": 1.5525,
+      "step": 32335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8829089592777213e-05,
+      "loss": 1.204,
+      "step": 32340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8828734741922712e-05,
+      "loss": 1.3652,
+      "step": 32345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882837984065139e-05,
+      "loss": 1.3118,
+      "step": 32350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882802488896527e-05,
+      "loss": 0.9876,
+      "step": 32355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882766988686638e-05,
+      "loss": 1.1367,
+      "step": 32360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882731483435675e-05,
+      "loss": 1.1062,
+      "step": 32365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88269597314384e-05,
+      "loss": 1.3545,
+      "step": 32370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8826604578113364e-05,
+      "loss": 0.9761,
+      "step": 32375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882624937438367e-05,
+      "loss": 1.2483,
+      "step": 32380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882589412025134e-05,
+      "loss": 1.653,
+      "step": 32385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8825538815718412e-05,
+      "loss": 1.5275,
+      "step": 32390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882518346078691e-05,
+      "loss": 1.247,
+      "step": 32395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8824828055458862e-05,
+      "loss": 1.2583,
+      "step": 32400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8824472599736297e-05,
+      "loss": 1.3265,
+      "step": 32405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8824117093621255e-05,
+      "loss": 1.4487,
+      "step": 32410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8823761537115753e-05,
+      "loss": 1.1961,
+      "step": 32415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882340593022183e-05,
+      "loss": 1.3205,
+      "step": 32420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8823050272941506e-05,
+      "loss": 0.9853,
+      "step": 32425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8822694565276828e-05,
+      "loss": 1.0458,
+      "step": 32430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882233880722981e-05,
+      "loss": 1.3118,
+      "step": 32435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88219829988025e-05,
+      "loss": 1.6132,
+      "step": 32440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882162713999692e-05,
+      "loss": 1.2281,
+      "step": 32445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8821271230815105e-05,
+      "loss": 1.0922,
+      "step": 32450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8820915271259085e-05,
+      "loss": 1.1141,
+      "step": 32455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8820559261330896e-05,
+      "loss": 1.0641,
+      "step": 32460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.882020320103257e-05,
+      "loss": 1.6522,
+      "step": 32465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8819847090366138e-05,
+      "loss": 1.0935,
+      "step": 32470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8819490929333634e-05,
+      "loss": 1.3375,
+      "step": 32475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8819134717937097e-05,
+      "loss": 1.3255,
+      "step": 32480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8818778456178552e-05,
+      "loss": 1.2895,
+      "step": 32485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8818422144060046e-05,
+      "loss": 0.9191,
+      "step": 32490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.88180657815836e-05,
+      "loss": 1.2454,
+      "step": 32495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881770936875126e-05,
+      "loss": 1.115,
+      "step": 32500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881735290556505e-05,
+      "loss": 1.6201,
+      "step": 32505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881699639202702e-05,
+      "loss": 1.2998,
+      "step": 32510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8816639828139195e-05,
+      "loss": 1.3474,
+      "step": 32515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8816283213903617e-05,
+      "loss": 1.1209,
+      "step": 32520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8815926549322315e-05,
+      "loss": 1.6302,
+      "step": 32525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8815569834397335e-05,
+      "loss": 1.4003,
+      "step": 32530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881521306913071e-05,
+      "loss": 1.7379,
+      "step": 32535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8814856253524477e-05,
+      "loss": 0.9436,
+      "step": 32540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8814499387580672e-05,
+      "loss": 1.5539,
+      "step": 32545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8814142471301337e-05,
+      "loss": 1.1359,
+      "step": 32550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8813785504688503e-05,
+      "loss": 1.1384,
+      "step": 32555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881342848774422e-05,
+      "loss": 1.1999,
+      "step": 32560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8813071420470515e-05,
+      "loss": 1.2457,
+      "step": 32565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8812714302869436e-05,
+      "loss": 1.1008,
+      "step": 32570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8812357134943013e-05,
+      "loss": 1.1645,
+      "step": 32575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8811999916693292e-05,
+      "loss": 1.1914,
+      "step": 32580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8811642648122316e-05,
+      "loss": 1.1416,
+      "step": 32585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8811285329232118e-05,
+      "loss": 1.0196,
+      "step": 32590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8810927960024744e-05,
+      "loss": 0.9974,
+      "step": 32595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8810570540502227e-05,
+      "loss": 1.3856,
+      "step": 32600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881021307066662e-05,
+      "loss": 1.2221,
+      "step": 32605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8809855550519955e-05,
+      "loss": 1.5059,
+      "step": 32610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8809497980064275e-05,
+      "loss": 1.5346,
+      "step": 32615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8809140359301627e-05,
+      "loss": 1.3307,
+      "step": 32620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8808782688234045e-05,
+      "loss": 1.1872,
+      "step": 32625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880842496686358e-05,
+      "loss": 1.3544,
+      "step": 32630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8808067195192268e-05,
+      "loss": 1.4741,
+      "step": 32635
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8807709373222155e-05,
+      "loss": 1.5405,
+      "step": 32640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8807351500955287e-05,
+      "loss": 1.2669,
+      "step": 32645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8806993578393702e-05,
+      "loss": 1.4021,
+      "step": 32650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880663560553945e-05,
+      "loss": 1.4396,
+      "step": 32655
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880627758239457e-05,
+      "loss": 1.4297,
+      "step": 32660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880591950896111e-05,
+      "loss": 1.0926,
+      "step": 32665
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880556138524111e-05,
+      "loss": 1.2709,
+      "step": 32670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880520321123662e-05,
+      "loss": 1.3962,
+      "step": 32675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8804844986949682e-05,
+      "loss": 1.2468,
+      "step": 32680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8804486712382347e-05,
+      "loss": 1.296,
+      "step": 32685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8804128387536657e-05,
+      "loss": 1.0503,
+      "step": 32690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8803770012414656e-05,
+      "loss": 1.2709,
+      "step": 32695
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8803411587018393e-05,
+      "loss": 1.2305,
+      "step": 32700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880305311134992e-05,
+      "loss": 1.3658,
+      "step": 32705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8802694585411272e-05,
+      "loss": 1.1939,
+      "step": 32710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880233600920451e-05,
+      "loss": 1.177,
+      "step": 32715
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8801977382731672e-05,
+      "loss": 1.146,
+      "step": 32720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8801618705994805e-05,
+      "loss": 1.1262,
+      "step": 32725
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8801259978995964e-05,
+      "loss": 1.0734,
+      "step": 32730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8800901201737198e-05,
+      "loss": 1.0895,
+      "step": 32735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.880054237422055e-05,
+      "loss": 1.3274,
+      "step": 32740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8800183496448074e-05,
+      "loss": 1.4252,
+      "step": 32745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8799824568421813e-05,
+      "loss": 1.1941,
+      "step": 32750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8799465590143822e-05,
+      "loss": 1.1958,
+      "step": 32755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879910656161615e-05,
+      "loss": 1.4034,
+      "step": 32760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8798747482840847e-05,
+      "loss": 1.1729,
+      "step": 32765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8798388353819967e-05,
+      "loss": 1.1753,
+      "step": 32770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879802917455555e-05,
+      "loss": 1.2101,
+      "step": 32775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879766994504966e-05,
+      "loss": 0.9137,
+      "step": 32780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879731066530434e-05,
+      "loss": 1.2363,
+      "step": 32785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8796951335321644e-05,
+      "loss": 1.2961,
+      "step": 32790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879659195510363e-05,
+      "loss": 1.3225,
+      "step": 32795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879623252465234e-05,
+      "loss": 1.2133,
+      "step": 32800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879587304396983e-05,
+      "loss": 1.4,
+      "step": 32805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8795513513058155e-05,
+      "loss": 1.7928,
+      "step": 32810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879515393191937e-05,
+      "loss": 1.1185,
+      "step": 32815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879479430055552e-05,
+      "loss": 1.1885,
+      "step": 32820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879443461896867e-05,
+      "loss": 1.157,
+      "step": 32825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8794074887160865e-05,
+      "loss": 1.1197,
+      "step": 32830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8793715105134163e-05,
+      "loss": 1.2616,
+      "step": 32835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879335527289062e-05,
+      "loss": 1.1355,
+      "step": 32840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8792995390432286e-05,
+      "loss": 1.0963,
+      "step": 32845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8792635457761222e-05,
+      "loss": 1.3632,
+      "step": 32850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879227547487948e-05,
+      "loss": 1.6196,
+      "step": 32855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8791915441789115e-05,
+      "loss": 1.3821,
+      "step": 32860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8791555358492182e-05,
+      "loss": 1.0561,
+      "step": 32865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879119522499074e-05,
+      "loss": 1.4736,
+      "step": 32870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8790835041286848e-05,
+      "loss": 1.4119,
+      "step": 32875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879047480738256e-05,
+      "loss": 1.4543,
+      "step": 32880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8790114523279926e-05,
+      "loss": 1.018,
+      "step": 32885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8789754188981018e-05,
+      "loss": 1.5367,
+      "step": 32890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8789393804487883e-05,
+      "loss": 1.1437,
+      "step": 32895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8789033369802583e-05,
+      "loss": 1.146,
+      "step": 32900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8788672884927174e-05,
+      "loss": 1.1816,
+      "step": 32905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8788312349863717e-05,
+      "loss": 1.2771,
+      "step": 32910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8787951764614266e-05,
+      "loss": 0.8585,
+      "step": 32915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8787591129180888e-05,
+      "loss": 1.388,
+      "step": 32920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8787230443565636e-05,
+      "loss": 1.3202,
+      "step": 32925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8786869707770573e-05,
+      "loss": 1.1879,
+      "step": 32930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.878650892179776e-05,
+      "loss": 1.388,
+      "step": 32935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8786148085649252e-05,
+      "loss": 1.1492,
+      "step": 32940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8785787199327114e-05,
+      "loss": 0.855,
+      "step": 32945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8785426262833405e-05,
+      "loss": 1.1777,
+      "step": 32950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8785065276170185e-05,
+      "loss": 1.1802,
+      "step": 32955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.878470423933952e-05,
+      "loss": 1.3087,
+      "step": 32960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8784343152343466e-05,
+      "loss": 1.0677,
+      "step": 32965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.878398201518409e-05,
+      "loss": 1.4291,
+      "step": 32970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8783620827863456e-05,
+      "loss": 1.2126,
+      "step": 32975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.878325959038362e-05,
+      "loss": 1.2583,
+      "step": 32980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8782898302746645e-05,
+      "loss": 0.9062,
+      "step": 32985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8782536964954598e-05,
+      "loss": 1.0278,
+      "step": 32990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8782175577009543e-05,
+      "loss": 1.4376,
+      "step": 32995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8781814138913537e-05,
+      "loss": 1.3111,
+      "step": 33000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8781452650668655e-05,
+      "loss": 1.086,
+      "step": 33005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8781091112276956e-05,
+      "loss": 1.2699,
+      "step": 33010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8780729523740496e-05,
+      "loss": 1.0952,
+      "step": 33015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8780367885061356e-05,
+      "loss": 1.0874,
+      "step": 33020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.878000619624159e-05,
+      "loss": 1.3071,
+      "step": 33025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8779644457283265e-05,
+      "loss": 1.3812,
+      "step": 33030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8779282668188448e-05,
+      "loss": 1.4624,
+      "step": 33035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8778920828959204e-05,
+      "loss": 0.9258,
+      "step": 33040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.87785589395976e-05,
+      "loss": 1.3082,
+      "step": 33045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8778197000105707e-05,
+      "loss": 1.2666,
+      "step": 33050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8777835010485584e-05,
+      "loss": 1.1284,
+      "step": 33055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8777472970739305e-05,
+      "loss": 1.5902,
+      "step": 33060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8777110880868933e-05,
+      "loss": 1.102,
+      "step": 33065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8776748740876538e-05,
+      "loss": 1.1108,
+      "step": 33070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8776386550764184e-05,
+      "loss": 1.1746,
+      "step": 33075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8776024310533944e-05,
+      "loss": 1.3563,
+      "step": 33080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8775662020187885e-05,
+      "loss": 1.1755,
+      "step": 33085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8775299679728077e-05,
+      "loss": 1.1295,
+      "step": 33090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8774937289156587e-05,
+      "loss": 1.1454,
+      "step": 33095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8774574848475485e-05,
+      "loss": 1.5705,
+      "step": 33100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.877421235768684e-05,
+      "loss": 1.3773,
+      "step": 33105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.877384981679273e-05,
+      "loss": 0.8551,
+      "step": 33110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.877348722579521e-05,
+      "loss": 1.2714,
+      "step": 33115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8773124584696365e-05,
+      "loss": 0.9983,
+      "step": 33120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.877276189349826e-05,
+      "loss": 1.1962,
+      "step": 33125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.877239915220296e-05,
+      "loss": 1.1285,
+      "step": 33130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8772036360812547e-05,
+      "loss": 1.1158,
+      "step": 33135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8771673519329088e-05,
+      "loss": 1.403,
+      "step": 33140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8771310627754652e-05,
+      "loss": 1.2364,
+      "step": 33145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8770947686091323e-05,
+      "loss": 1.6717,
+      "step": 33150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8770584694341156e-05,
+      "loss": 1.486,
+      "step": 33155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8770221652506237e-05,
+      "loss": 1.4626,
+      "step": 33160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8769858560588637e-05,
+      "loss": 1.0804,
+      "step": 33165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8769495418590425e-05,
+      "loss": 1.4414,
+      "step": 33170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8769132226513677e-05,
+      "loss": 1.266,
+      "step": 33175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.876876898436047e-05,
+      "loss": 1.2743,
+      "step": 33180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8768405692132873e-05,
+      "loss": 1.5728,
+      "step": 33185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8768042349832968e-05,
+      "loss": 1.3185,
+      "step": 33190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.876767895746282e-05,
+      "loss": 1.1854,
+      "step": 33195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8767315515024513e-05,
+      "loss": 1.1958,
+      "step": 33200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.876695202252012e-05,
+      "loss": 1.33,
+      "step": 33205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8766588479951712e-05,
+      "loss": 1.4909,
+      "step": 33210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.876622488732137e-05,
+      "loss": 1.2105,
+      "step": 33215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8765861244631168e-05,
+      "loss": 1.2752,
+      "step": 33220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8765497551883185e-05,
+      "loss": 1.2062,
+      "step": 33225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8765133809079496e-05,
+      "loss": 0.9942,
+      "step": 33230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.876477001622218e-05,
+      "loss": 1.3056,
+      "step": 33235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.876440617331331e-05,
+      "loss": 1.398,
+      "step": 33240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.876404228035497e-05,
+      "loss": 1.0888,
+      "step": 33245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8763678337349233e-05,
+      "loss": 1.6666,
+      "step": 33250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8763314344298177e-05,
+      "loss": 1.3396,
+      "step": 33255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8762950301203886e-05,
+      "loss": 1.2603,
+      "step": 33260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8762586208068433e-05,
+      "loss": 1.4264,
+      "step": 33265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8762222064893906e-05,
+      "loss": 1.0829,
+      "step": 33270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.876185787168237e-05,
+      "loss": 1.1973,
+      "step": 33275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8761493628435916e-05,
+      "loss": 1.0872,
+      "step": 33280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8761129335156624e-05,
+      "loss": 1.191,
+      "step": 33285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.876076499184657e-05,
+      "loss": 1.1669,
+      "step": 33290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8760400598507833e-05,
+      "loss": 1.2544,
+      "step": 33295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.87600361551425e-05,
+      "loss": 1.3278,
+      "step": 33300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8759671661752645e-05,
+      "loss": 0.9875,
+      "step": 33305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8759307118340356e-05,
+      "loss": 1.5577,
+      "step": 33310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8758942524907712e-05,
+      "loss": 1.3718,
+      "step": 33315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8758577881456797e-05,
+      "loss": 1.1593,
+      "step": 33320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8758213187989688e-05,
+      "loss": 1.2983,
+      "step": 33325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8757848444508474e-05,
+      "loss": 1.5165,
+      "step": 33330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8757483651015237e-05,
+      "loss": 1.2186,
+      "step": 33335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8757118807512053e-05,
+      "loss": 1.1013,
+      "step": 33340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.875675391400101e-05,
+      "loss": 1.4452,
+      "step": 33345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8756388970484197e-05,
+      "loss": 1.2591,
+      "step": 33350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8756023976963692e-05,
+      "loss": 1.3346,
+      "step": 33355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.875565893344158e-05,
+      "loss": 1.3299,
+      "step": 33360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8755293839919945e-05,
+      "loss": 1.1073,
+      "step": 33365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8754928696400872e-05,
+      "loss": 1.3957,
+      "step": 33370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8754563502886453e-05,
+      "loss": 1.03,
+      "step": 33375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8754198259378763e-05,
+      "loss": 1.2654,
+      "step": 33380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8753832965879894e-05,
+      "loss": 1.3612,
+      "step": 33385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8753467622391928e-05,
+      "loss": 1.4394,
+      "step": 33390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8753102228916956e-05,
+      "loss": 1.3171,
+      "step": 33395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8752736785457063e-05,
+      "loss": 1.2866,
+      "step": 33400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8752371292014335e-05,
+      "loss": 1.4428,
+      "step": 33405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.875200574859086e-05,
+      "loss": 1.1635,
+      "step": 33410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8751640155188723e-05,
+      "loss": 1.1902,
+      "step": 33415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8751274511810016e-05,
+      "loss": 1.5265,
+      "step": 33420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8750908818456823e-05,
+      "loss": 1.0404,
+      "step": 33425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8750543075131233e-05,
+      "loss": 1.0621,
+      "step": 33430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.875017728183534e-05,
+      "loss": 0.9614,
+      "step": 33435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8749811438571225e-05,
+      "loss": 1.2881,
+      "step": 33440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8749445545340977e-05,
+      "loss": 1.3353,
+      "step": 33445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8749079602146695e-05,
+      "loss": 1.254,
+      "step": 33450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.874871360899046e-05,
+      "loss": 1.042,
+      "step": 33455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8748347565874364e-05,
+      "loss": 0.9707,
+      "step": 33460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8747981472800498e-05,
+      "loss": 1.4809,
+      "step": 33465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8747615329770954e-05,
+      "loss": 1.1524,
+      "step": 33470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.874724913678782e-05,
+      "loss": 1.4386,
+      "step": 33475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8746882893853194e-05,
+      "loss": 1.1557,
+      "step": 33480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.874651660096916e-05,
+      "loss": 1.2064,
+      "step": 33485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8746150258137807e-05,
+      "loss": 1.2201,
+      "step": 33490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.874578386536124e-05,
+      "loss": 1.3743,
+      "step": 33495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8745417422641538e-05,
+      "loss": 1.1253,
+      "step": 33500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8745050929980798e-05,
+      "loss": 1.3681,
+      "step": 33505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8744684387381113e-05,
+      "loss": 1.0922,
+      "step": 33510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8744317794844577e-05,
+      "loss": 1.0956,
+      "step": 33515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8743951152373287e-05,
+      "loss": 1.6371,
+      "step": 33520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8743584459969327e-05,
+      "loss": 1.0268,
+      "step": 33525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8743217717634804e-05,
+      "loss": 1.3316,
+      "step": 33530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8742850925371802e-05,
+      "loss": 1.556,
+      "step": 33535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.874248408318242e-05,
+      "loss": 1.3755,
+      "step": 33540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.874211719106875e-05,
+      "loss": 0.8986,
+      "step": 33545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8741750249032888e-05,
+      "loss": 1.844,
+      "step": 33550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8741383257076932e-05,
+      "loss": 1.2493,
+      "step": 33555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8741016215202977e-05,
+      "loss": 1.2149,
+      "step": 33560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8740649123413116e-05,
+      "loss": 1.1022,
+      "step": 33565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8740281981709447e-05,
+      "loss": 1.2049,
+      "step": 33570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.873991479009407e-05,
+      "loss": 1.5214,
+      "step": 33575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8739547548569078e-05,
+      "loss": 1.3531,
+      "step": 33580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8739180257136564e-05,
+      "loss": 1.1868,
+      "step": 33585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8738812915798633e-05,
+      "loss": 1.2879,
+      "step": 33590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8738445524557384e-05,
+      "loss": 1.0389,
+      "step": 33595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.873807808341491e-05,
+      "loss": 1.3338,
+      "step": 33600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8737710592373307e-05,
+      "loss": 0.9585,
+      "step": 33605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.873734305143468e-05,
+      "loss": 1.1068,
+      "step": 33610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.873697546060112e-05,
+      "loss": 1.4905,
+      "step": 33615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.873660781987474e-05,
+      "loss": 1.5291,
+      "step": 33620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.873624012925762e-05,
+      "loss": 1.2856,
+      "step": 33625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8735872388751875e-05,
+      "loss": 1.4125,
+      "step": 33630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.87355045983596e-05,
+      "loss": 1.1323,
+      "step": 33635
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8735136758082895e-05,
+      "loss": 1.1667,
+      "step": 33640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.873476886792386e-05,
+      "loss": 1.1076,
+      "step": 33645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8734400927884598e-05,
+      "loss": 1.2081,
+      "step": 33650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8734032937967208e-05,
+      "loss": 1.3319,
+      "step": 33655
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8733664898173793e-05,
+      "loss": 1.1159,
+      "step": 33660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8733296808506455e-05,
+      "loss": 1.5209,
+      "step": 33665
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8732928668967294e-05,
+      "loss": 1.1542,
+      "step": 33670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.873256047955841e-05,
+      "loss": 0.953,
+      "step": 33675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8732192240281914e-05,
+      "loss": 1.2435,
+      "step": 33680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.87318239511399e-05,
+      "loss": 1.0902,
+      "step": 33685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.873145561213448e-05,
+      "loss": 1.1276,
+      "step": 33690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8731087223267746e-05,
+      "loss": 1.2992,
+      "step": 33695
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.873071878454181e-05,
+      "loss": 1.4748,
+      "step": 33700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8730350295958776e-05,
+      "loss": 1.1548,
+      "step": 33705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8729981757520745e-05,
+      "loss": 1.2349,
+      "step": 33710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.872961316922982e-05,
+      "loss": 1.5161,
+      "step": 33715
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8729244531088113e-05,
+      "loss": 1.0062,
+      "step": 33720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.872887584309772e-05,
+      "loss": 1.2376,
+      "step": 33725
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8728507105260755e-05,
+      "loss": 1.4262,
+      "step": 33730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.872813831757932e-05,
+      "loss": 1.5282,
+      "step": 33735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.872776948005552e-05,
+      "loss": 1.4226,
+      "step": 33740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.872740059269146e-05,
+      "loss": 1.0556,
+      "step": 33745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.872703165548925e-05,
+      "loss": 1.1428,
+      "step": 33750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8726662668450993e-05,
+      "loss": 1.1511,
+      "step": 33755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8726293631578802e-05,
+      "loss": 1.2344,
+      "step": 33760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.872592454487478e-05,
+      "loss": 1.151,
+      "step": 33765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8725555408341037e-05,
+      "loss": 1.1932,
+      "step": 33770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8725186221979678e-05,
+      "loss": 1.0068,
+      "step": 33775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8724816985792812e-05,
+      "loss": 1.3853,
+      "step": 33780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.872444769978255e-05,
+      "loss": 1.056,
+      "step": 33785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8724078363950995e-05,
+      "loss": 1.1009,
+      "step": 33790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8723708978300266e-05,
+      "loss": 1.2072,
+      "step": 33795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8723339542832463e-05,
+      "loss": 1.3125,
+      "step": 33800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.87229700575497e-05,
+      "loss": 1.3274,
+      "step": 33805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8722600522454085e-05,
+      "loss": 1.5841,
+      "step": 33810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8722230937547736e-05,
+      "loss": 1.4492,
+      "step": 33815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.872186130283275e-05,
+      "loss": 1.0086,
+      "step": 33820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8721491618311245e-05,
+      "loss": 1.1468,
+      "step": 33825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8721121883985332e-05,
+      "loss": 1.228,
+      "step": 33830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8720752099857128e-05,
+      "loss": 1.3323,
+      "step": 33835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8720382265928733e-05,
+      "loss": 1.1847,
+      "step": 33840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.872001238220227e-05,
+      "loss": 1.1681,
+      "step": 33845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8719642448679845e-05,
+      "loss": 1.101,
+      "step": 33850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.871927246536357e-05,
+      "loss": 1.204,
+      "step": 33855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.871890243225556e-05,
+      "loss": 1.651,
+      "step": 33860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8718532349357927e-05,
+      "loss": 1.2513,
+      "step": 33865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8718162216672786e-05,
+      "loss": 1.3574,
+      "step": 33870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.871779203420225e-05,
+      "loss": 1.0605,
+      "step": 33875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.871742180194843e-05,
+      "loss": 1.3523,
+      "step": 33880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8717051519913444e-05,
+      "loss": 1.3791,
+      "step": 33885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.871668118809941e-05,
+      "loss": 1.4074,
+      "step": 33890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8716310806508434e-05,
+      "loss": 1.3377,
+      "step": 33895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8715940375142633e-05,
+      "loss": 1.227,
+      "step": 33900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.871556989400413e-05,
+      "loss": 1.4065,
+      "step": 33905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8715199363095032e-05,
+      "loss": 1.194,
+      "step": 33910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8714828782417455e-05,
+      "loss": 1.2001,
+      "step": 33915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8714458151973523e-05,
+      "loss": 0.9536,
+      "step": 33920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8714087471765346e-05,
+      "loss": 1.0468,
+      "step": 33925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8713716741795043e-05,
+      "loss": 1.2602,
+      "step": 33930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8713345962064733e-05,
+      "loss": 1.239,
+      "step": 33935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8712975132576526e-05,
+      "loss": 1.0794,
+      "step": 33940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.871260425333255e-05,
+      "loss": 0.8765,
+      "step": 33945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8712233324334913e-05,
+      "loss": 1.4591,
+      "step": 33950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.871186234558574e-05,
+      "loss": 1.0116,
+      "step": 33955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.871149131708715e-05,
+      "loss": 1.2628,
+      "step": 33960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8711120238841257e-05,
+      "loss": 0.968,
+      "step": 33965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.871074911085018e-05,
+      "loss": 1.4602,
+      "step": 33970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8710377933116045e-05,
+      "loss": 1.3551,
+      "step": 33975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.871000670564096e-05,
+      "loss": 1.1013,
+      "step": 33980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8709635428427056e-05,
+      "loss": 1.4379,
+      "step": 33985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8709264101476453e-05,
+      "loss": 1.0506,
+      "step": 33990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8708892724791264e-05,
+      "loss": 1.19,
+      "step": 33995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8708521298373613e-05,
+      "loss": 1.0492,
+      "step": 34000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8708149822225624e-05,
+      "loss": 0.9973,
+      "step": 34005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8707778296349412e-05,
+      "loss": 1.3177,
+      "step": 34010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8707406720747104e-05,
+      "loss": 0.8646,
+      "step": 34015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.870703509542082e-05,
+      "loss": 1.4617,
+      "step": 34020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8706663420372684e-05,
+      "loss": 1.2013,
+      "step": 34025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8706291695604815e-05,
+      "loss": 1.2308,
+      "step": 34030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.870591992111934e-05,
+      "loss": 1.1748,
+      "step": 34035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.870554809691838e-05,
+      "loss": 1.297,
+      "step": 34040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8705176223004055e-05,
+      "loss": 1.1108,
+      "step": 34045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8704804299378495e-05,
+      "loss": 1.4435,
+      "step": 34050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.870443232604382e-05,
+      "loss": 1.3071,
+      "step": 34055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.870406030300215e-05,
+      "loss": 1.3797,
+      "step": 34060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.870368823025562e-05,
+      "loss": 1.3759,
+      "step": 34065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8703316107806347e-05,
+      "loss": 1.2495,
+      "step": 34070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8702943935656457e-05,
+      "loss": 1.495,
+      "step": 34075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8702571713808075e-05,
+      "loss": 1.1922,
+      "step": 34080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.870219944226333e-05,
+      "loss": 1.3765,
+      "step": 34085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8701827121024347e-05,
+      "loss": 1.1872,
+      "step": 34090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.870145475009325e-05,
+      "loss": 1.1181,
+      "step": 34095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8701082329472164e-05,
+      "loss": 1.4883,
+      "step": 34100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.870070985916322e-05,
+      "loss": 1.3689,
+      "step": 34105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8700337339168546e-05,
+      "loss": 1.3917,
+      "step": 34110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8699964769490258e-05,
+      "loss": 1.0942,
+      "step": 34115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.86995921501305e-05,
+      "loss": 1.1997,
+      "step": 34120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8699219481091388e-05,
+      "loss": 1.1613,
+      "step": 34125
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.869884676237505e-05,
+      "loss": 1.2041,
+      "step": 34130
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8698473993983626e-05,
+      "loss": 1.3801,
+      "step": 34135
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8698101175919232e-05,
+      "loss": 1.2959,
+      "step": 34140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8697728308184006e-05,
+      "loss": 1.0529,
+      "step": 34145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8697355390780068e-05,
+      "loss": 1.3307,
+      "step": 34150
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8696982423709555e-05,
+      "loss": 1.3271,
+      "step": 34155
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.86966094069746e-05,
+      "loss": 1.1174,
+      "step": 34160
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.869623634057732e-05,
+      "loss": 1.4578,
+      "step": 34165
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8695863224519856e-05,
+      "loss": 1.5529,
+      "step": 34170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8695490058804338e-05,
+      "loss": 1.3526,
+      "step": 34175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.869511684343289e-05,
+      "loss": 1.2998,
+      "step": 34180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8694743578407656e-05,
+      "loss": 1.3039,
+      "step": 34185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.869437026373075e-05,
+      "loss": 1.0906,
+      "step": 34190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.869399689940432e-05,
+      "loss": 1.1739,
+      "step": 34195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.869362348543049e-05,
+      "loss": 1.6473,
+      "step": 34200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8693250021811392e-05,
+      "loss": 1.2744,
+      "step": 34205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8692876508549165e-05,
+      "loss": 1.5486,
+      "step": 34210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8692502945645932e-05,
+      "loss": 1.1398,
+      "step": 34215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8692129333103837e-05,
+      "loss": 0.9503,
+      "step": 34220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.869175567092501e-05,
+      "loss": 1.3486,
+      "step": 34225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8691381959111578e-05,
+      "loss": 1.2898,
+      "step": 34230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8691008197665682e-05,
+      "loss": 1.0798,
+      "step": 34235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8690634386589456e-05,
+      "loss": 1.3544,
+      "step": 34240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8690260525885034e-05,
+      "loss": 1.5095,
+      "step": 34245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8689886615554547e-05,
+      "loss": 1.4555,
+      "step": 34250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8689512655600134e-05,
+      "loss": 1.3923,
+      "step": 34255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8689138646023932e-05,
+      "loss": 1.0512,
+      "step": 34260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8688764586828073e-05,
+      "loss": 1.3044,
+      "step": 34265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8688390478014697e-05,
+      "loss": 1.4456,
+      "step": 34270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.868801631958594e-05,
+      "loss": 0.9239,
+      "step": 34275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8687642111543934e-05,
+      "loss": 1.1908,
+      "step": 34280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8687267853890817e-05,
+      "loss": 1.2663,
+      "step": 34285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.868689354662873e-05,
+      "loss": 1.095,
+      "step": 34290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8686519189759812e-05,
+      "loss": 1.0528,
+      "step": 34295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868614478328619e-05,
+      "loss": 1.2927,
+      "step": 34300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8685770327210015e-05,
+      "loss": 1.0844,
+      "step": 34305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868539582153342e-05,
+      "loss": 1.4194,
+      "step": 34310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868502126625854e-05,
+      "loss": 1.2382,
+      "step": 34315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868464666138752e-05,
+      "loss": 1.4315,
+      "step": 34320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8684272006922495e-05,
+      "loss": 1.1372,
+      "step": 34325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8683897302865606e-05,
+      "loss": 1.4553,
+      "step": 34330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8683522549218994e-05,
+      "loss": 1.2283,
+      "step": 34335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86831477459848e-05,
+      "loss": 0.9524,
+      "step": 34340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8682772893165157e-05,
+      "loss": 1.3024,
+      "step": 34345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868239799076221e-05,
+      "loss": 1.2465,
+      "step": 34350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8682023038778108e-05,
+      "loss": 1.3735,
+      "step": 34355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868164803721498e-05,
+      "loss": 1.167,
+      "step": 34360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868127298607497e-05,
+      "loss": 1.2842,
+      "step": 34365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8680897885360227e-05,
+      "loss": 1.189,
+      "step": 34370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8680522735072886e-05,
+      "loss": 1.2256,
+      "step": 34375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.868014753521509e-05,
+      "loss": 1.0089,
+      "step": 34380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8679772285788985e-05,
+      "loss": 1.0455,
+      "step": 34385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8679396986796713e-05,
+      "loss": 1.1368,
+      "step": 34390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8679021638240413e-05,
+      "loss": 1.3222,
+      "step": 34395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867864624012223e-05,
+      "loss": 1.0811,
+      "step": 34400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867827079244431e-05,
+      "loss": 1.1515,
+      "step": 34405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86778952952088e-05,
+      "loss": 1.1978,
+      "step": 34410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867751974841784e-05,
+      "loss": 1.3424,
+      "step": 34415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867714415207357e-05,
+      "loss": 1.4138,
+      "step": 34420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8676768506178144e-05,
+      "loss": 1.5867,
+      "step": 34425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86763928107337e-05,
+      "loss": 1.0305,
+      "step": 34430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867601706574239e-05,
+      "loss": 1.3847,
+      "step": 34435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8675641271206357e-05,
+      "loss": 1.1201,
+      "step": 34440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8675265427127743e-05,
+      "loss": 1.1899,
+      "step": 34445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86748895335087e-05,
+      "loss": 1.173,
+      "step": 34450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8674513590351367e-05,
+      "loss": 1.355,
+      "step": 34455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86741375976579e-05,
+      "loss": 1.2924,
+      "step": 34460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8673761555430438e-05,
+      "loss": 1.2793,
+      "step": 34465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8673385463671137e-05,
+      "loss": 1.3218,
+      "step": 34470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8673009322382136e-05,
+      "loss": 1.2062,
+      "step": 34475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867263313156559e-05,
+      "loss": 1.1268,
+      "step": 34480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8672256891223642e-05,
+      "loss": 1.6389,
+      "step": 34485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8671880601358446e-05,
+      "loss": 1.0731,
+      "step": 34490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8671504261972146e-05,
+      "loss": 1.1973,
+      "step": 34495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867112787306689e-05,
+      "loss": 1.3505,
+      "step": 34500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867075143464483e-05,
+      "loss": 1.1525,
+      "step": 34505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867037494670812e-05,
+      "loss": 1.6973,
+      "step": 34510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8669998409258904e-05,
+      "loss": 1.1359,
+      "step": 34515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8669621822299337e-05,
+      "loss": 1.2624,
+      "step": 34520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866924518583156e-05,
+      "loss": 1.281,
+      "step": 34525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8668868499857733e-05,
+      "loss": 1.2959,
+      "step": 34530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8668491764380002e-05,
+      "loss": 1.5509,
+      "step": 34535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8668114979400527e-05,
+      "loss": 1.4329,
+      "step": 34540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8667738144921447e-05,
+      "loss": 1.5262,
+      "step": 34545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8667361260944922e-05,
+      "loss": 1.1588,
+      "step": 34550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8666984327473104e-05,
+      "loss": 1.4481,
+      "step": 34555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866660734450814e-05,
+      "loss": 1.5539,
+      "step": 34560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8666230312052192e-05,
+      "loss": 1.3155,
+      "step": 34565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8665853230107408e-05,
+      "loss": 1.8022,
+      "step": 34570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8665476098675937e-05,
+      "loss": 1.1146,
+      "step": 34575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8665098917759937e-05,
+      "loss": 1.4963,
+      "step": 34580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8664721687361563e-05,
+      "loss": 1.0544,
+      "step": 34585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866434440748297e-05,
+      "loss": 1.2647,
+      "step": 34590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8663967078126307e-05,
+      "loss": 1.2092,
+      "step": 34595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8663589699293734e-05,
+      "loss": 1.2607,
+      "step": 34600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8663212270987403e-05,
+      "loss": 1.4857,
+      "step": 34605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866283479320947e-05,
+      "loss": 1.0402,
+      "step": 34610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8662457265962093e-05,
+      "loss": 1.0794,
+      "step": 34615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8662079689247424e-05,
+      "loss": 1.5192,
+      "step": 34620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866170206306762e-05,
+      "loss": 1.234,
+      "step": 34625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8661324387424843e-05,
+      "loss": 1.2319,
+      "step": 34630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866094666232124e-05,
+      "loss": 1.2376,
+      "step": 34635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8660568887758974e-05,
+      "loss": 1.2269,
+      "step": 34640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8660191063740205e-05,
+      "loss": 1.52,
+      "step": 34645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8659813190267086e-05,
+      "loss": 1.4544,
+      "step": 34650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8659435267341774e-05,
+      "loss": 1.0622,
+      "step": 34655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8659057294966428e-05,
+      "loss": 1.2774,
+      "step": 34660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8658679273143213e-05,
+      "loss": 1.1809,
+      "step": 34665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8658301201874277e-05,
+      "loss": 1.4256,
+      "step": 34670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8657923081161788e-05,
+      "loss": 1.0008,
+      "step": 34675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8657544911007898e-05,
+      "loss": 1.2291,
+      "step": 34680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8657166691414775e-05,
+      "loss": 1.2041,
+      "step": 34685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8656788422384568e-05,
+      "loss": 1.5161,
+      "step": 34690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865641010391945e-05,
+      "loss": 1.2934,
+      "step": 34695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865603173602157e-05,
+      "loss": 1.2324,
+      "step": 34700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8655653318693093e-05,
+      "loss": 1.2505,
+      "step": 34705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865527485193618e-05,
+      "loss": 1.1727,
+      "step": 34710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8654896335752993e-05,
+      "loss": 1.3639,
+      "step": 34715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8654517770145692e-05,
+      "loss": 1.1419,
+      "step": 34720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865413915511644e-05,
+      "loss": 1.1359,
+      "step": 34725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86537604906674e-05,
+      "loss": 1.2236,
+      "step": 34730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865338177680073e-05,
+      "loss": 1.2598,
+      "step": 34735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86530030135186e-05,
+      "loss": 1.1333,
+      "step": 34740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8652624200823167e-05,
+      "loss": 1.6573,
+      "step": 34745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8652245338716594e-05,
+      "loss": 1.2247,
+      "step": 34750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.865186642720105e-05,
+      "loss": 1.1248,
+      "step": 34755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8651487466278695e-05,
+      "loss": 1.4179,
+      "step": 34760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8651108455951693e-05,
+      "loss": 1.1448,
+      "step": 34765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8650729396222206e-05,
+      "loss": 1.1849,
+      "step": 34770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8650350287092402e-05,
+      "loss": 1.0223,
+      "step": 34775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.864997112856445e-05,
+      "loss": 1.0706,
+      "step": 34780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8649591920640504e-05,
+      "loss": 1.5487,
+      "step": 34785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.864921266332274e-05,
+      "loss": 1.294,
+      "step": 34790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8648833356613316e-05,
+      "loss": 1.32,
+      "step": 34795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8648454000514403e-05,
+      "loss": 1.3698,
+      "step": 34800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.864807459502817e-05,
+      "loss": 1.0437,
+      "step": 34805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8647695140156777e-05,
+      "loss": 1.3676,
+      "step": 34810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8647315635902394e-05,
+      "loss": 1.2121,
+      "step": 34815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8646936082267185e-05,
+      "loss": 1.1979,
+      "step": 34820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8646556479253324e-05,
+      "loss": 1.4068,
+      "step": 34825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8646176826862974e-05,
+      "loss": 1.3338,
+      "step": 34830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8645797125098305e-05,
+      "loss": 1.6872,
+      "step": 34835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8645417373961478e-05,
+      "loss": 1.0131,
+      "step": 34840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8645037573454675e-05,
+      "loss": 1.3204,
+      "step": 34845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8644657723580054e-05,
+      "loss": 1.5051,
+      "step": 34850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.864427782433979e-05,
+      "loss": 1.1561,
+      "step": 34855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.864389787573605e-05,
+      "loss": 0.9592,
+      "step": 34860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8643517877771002e-05,
+      "loss": 1.0296,
+      "step": 34865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8643137830446818e-05,
+      "loss": 1.2508,
+      "step": 34870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.864275773376567e-05,
+      "loss": 1.4071,
+      "step": 34875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8642377587729728e-05,
+      "loss": 1.0508,
+      "step": 34880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.864199739234116e-05,
+      "loss": 1.1499,
+      "step": 34885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8641617147602136e-05,
+      "loss": 1.187,
+      "step": 34890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8641236853514835e-05,
+      "loss": 0.9721,
+      "step": 34895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.864085651008142e-05,
+      "loss": 1.2704,
+      "step": 34900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8640476117304068e-05,
+      "loss": 1.029,
+      "step": 34905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8640095675184952e-05,
+      "loss": 1.209,
+      "step": 34910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8639715183726242e-05,
+      "loss": 1.5133,
+      "step": 34915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8639334642930112e-05,
+      "loss": 1.2307,
+      "step": 34920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863895405279873e-05,
+      "loss": 1.3754,
+      "step": 34925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8638573413334278e-05,
+      "loss": 1.1935,
+      "step": 34930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8638192724538924e-05,
+      "loss": 1.2565,
+      "step": 34935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8637811986414844e-05,
+      "loss": 1.3252,
+      "step": 34940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8637431198964214e-05,
+      "loss": 1.1767,
+      "step": 34945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8637050362189203e-05,
+      "loss": 1.2954,
+      "step": 34950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8636669476091987e-05,
+      "loss": 1.0723,
+      "step": 34955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8636288540674747e-05,
+      "loss": 1.1965,
+      "step": 34960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8635907555939654e-05,
+      "loss": 1.0718,
+      "step": 34965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8635526521888884e-05,
+      "loss": 0.9595,
+      "step": 34970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863514543852461e-05,
+      "loss": 1.2953,
+      "step": 34975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8634764305849016e-05,
+      "loss": 1.2046,
+      "step": 34980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863438312386427e-05,
+      "loss": 1.3053,
+      "step": 34985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8634001892572553e-05,
+      "loss": 1.1476,
+      "step": 34990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863362061197604e-05,
+      "loss": 0.9745,
+      "step": 34995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8633239282076913e-05,
+      "loss": 1.1615,
+      "step": 35000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8632857902877343e-05,
+      "loss": 1.0728,
+      "step": 35005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8632476474379514e-05,
+      "loss": 1.007,
+      "step": 35010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86320949965856e-05,
+      "loss": 1.2111,
+      "step": 35015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863171346949778e-05,
+      "loss": 1.18,
+      "step": 35020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8631331893118235e-05,
+      "loss": 1.5978,
+      "step": 35025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8630950267449144e-05,
+      "loss": 1.1009,
+      "step": 35030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8630568592492682e-05,
+      "loss": 0.9293,
+      "step": 35035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863018686825103e-05,
+      "loss": 1.4354,
+      "step": 35040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862980509472637e-05,
+      "loss": 1.0185,
+      "step": 35045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8629423271920884e-05,
+      "loss": 1.0882,
+      "step": 35050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862904139983675e-05,
+      "loss": 1.5625,
+      "step": 35055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8628659478476147e-05,
+      "loss": 1.3144,
+      "step": 35060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8628277507841256e-05,
+      "loss": 1.103,
+      "step": 35065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8627895487934262e-05,
+      "loss": 1.3721,
+      "step": 35070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8627513418757344e-05,
+      "loss": 1.1207,
+      "step": 35075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8627131300312683e-05,
+      "loss": 1.1478,
+      "step": 35080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8626749132602462e-05,
+      "loss": 1.2715,
+      "step": 35085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8626366915628866e-05,
+      "loss": 1.1415,
+      "step": 35090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8625984649394073e-05,
+      "loss": 1.1927,
+      "step": 35095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8625602333900265e-05,
+      "loss": 0.9525,
+      "step": 35100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8625219969149634e-05,
+      "loss": 1.2689,
+      "step": 35105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8624837555144353e-05,
+      "loss": 1.2822,
+      "step": 35110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8624455091886613e-05,
+      "loss": 1.5097,
+      "step": 35115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8624072579378596e-05,
+      "loss": 1.0778,
+      "step": 35120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862369001762249e-05,
+      "loss": 1.1811,
+      "step": 35125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862330740662047e-05,
+      "loss": 1.5006,
+      "step": 35130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8622924746374726e-05,
+      "loss": 1.5405,
+      "step": 35135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8622542036887446e-05,
+      "loss": 1.2704,
+      "step": 35140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8622159278160813e-05,
+      "loss": 1.3107,
+      "step": 35145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8621776470197015e-05,
+      "loss": 1.4961,
+      "step": 35150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8621393612998234e-05,
+      "loss": 1.3139,
+      "step": 35155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862101070656666e-05,
+      "loss": 1.1835,
+      "step": 35160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.862062775090448e-05,
+      "loss": 1.0045,
+      "step": 35165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8620244746013876e-05,
+      "loss": 1.2807,
+      "step": 35170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8619861691897038e-05,
+      "loss": 1.1288,
+      "step": 35175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.861947858855615e-05,
+      "loss": 1.2933,
+      "step": 35180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.861909543599341e-05,
+      "loss": 1.6077,
+      "step": 35185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8618712234210994e-05,
+      "loss": 1.2205,
+      "step": 35190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86183289832111e-05,
+      "loss": 1.1315,
+      "step": 35195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.861794568299591e-05,
+      "loss": 0.9486,
+      "step": 35200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8617562333567614e-05,
+      "loss": 1.1408,
+      "step": 35205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8617178934928403e-05,
+      "loss": 1.0948,
+      "step": 35210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8616795487080466e-05,
+      "loss": 1.2036,
+      "step": 35215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.861641199002599e-05,
+      "loss": 1.1494,
+      "step": 35220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.861602844376717e-05,
+      "loss": 1.1035,
+      "step": 35225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8615644848306192e-05,
+      "loss": 1.2109,
+      "step": 35230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8615261203645248e-05,
+      "loss": 1.0375,
+      "step": 35235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.861487750978653e-05,
+      "loss": 1.2785,
+      "step": 35240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8614493766732225e-05,
+      "loss": 1.4851,
+      "step": 35245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.861410997448453e-05,
+      "loss": 1.0959,
+      "step": 35250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8613726133045636e-05,
+      "loss": 1.6286,
+      "step": 35255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8613342242417728e-05,
+      "loss": 1.4259,
+      "step": 35260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8612958302603006e-05,
+      "loss": 1.2978,
+      "step": 35265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.861257431360366e-05,
+      "loss": 1.3628,
+      "step": 35270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.861219027542188e-05,
+      "loss": 1.0244,
+      "step": 35275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.861180618805986e-05,
+      "loss": 1.3776,
+      "step": 35280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86114220515198e-05,
+      "loss": 1.3513,
+      "step": 35285
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8611037865803885e-05,
+      "loss": 1.2657,
+      "step": 35290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8610653630914313e-05,
+      "loss": 1.208,
+      "step": 35295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.861026934685328e-05,
+      "loss": 1.0329,
+      "step": 35300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8609885013622975e-05,
+      "loss": 1.1475,
+      "step": 35305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8609500631225597e-05,
+      "loss": 0.836,
+      "step": 35310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8609116199663336e-05,
+      "loss": 1.1854,
+      "step": 35315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8608731718938396e-05,
+      "loss": 1.0718,
+      "step": 35320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8608347189052964e-05,
+      "loss": 1.2188,
+      "step": 35325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8607962610009244e-05,
+      "loss": 1.2413,
+      "step": 35330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8607577981809423e-05,
+      "loss": 1.4866,
+      "step": 35335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8607193304455704e-05,
+      "loss": 1.3904,
+      "step": 35340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8606808577950283e-05,
+      "loss": 1.5742,
+      "step": 35345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8606423802295355e-05,
+      "loss": 1.137,
+      "step": 35350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8606038977493116e-05,
+      "loss": 1.1269,
+      "step": 35355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.860565410354577e-05,
+      "loss": 1.3265,
+      "step": 35360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.860526918045551e-05,
+      "loss": 1.0157,
+      "step": 35365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.860488420822453e-05,
+      "loss": 1.0769,
+      "step": 35370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8604499186855033e-05,
+      "loss": 1.0145,
+      "step": 35375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8604114116349223e-05,
+      "loss": 1.063,
+      "step": 35380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.860372899670929e-05,
+      "loss": 1.3636,
+      "step": 35385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.860334382793744e-05,
+      "loss": 1.1323,
+      "step": 35390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8602958610035866e-05,
+      "loss": 1.0992,
+      "step": 35395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8602573343006775e-05,
+      "loss": 1.3218,
+      "step": 35400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.860218802685236e-05,
+      "loss": 1.247,
+      "step": 35405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.860180266157482e-05,
+      "loss": 1.1305,
+      "step": 35410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8601417247176368e-05,
+      "loss": 0.9523,
+      "step": 35415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8601031783659194e-05,
+      "loss": 0.9526,
+      "step": 35420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8600646271025507e-05,
+      "loss": 1.1344,
+      "step": 35425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86002607092775e-05,
+      "loss": 1.2282,
+      "step": 35430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8599875098417378e-05,
+      "loss": 1.1813,
+      "step": 35435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8599489438447344e-05,
+      "loss": 1.4681,
+      "step": 35440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8599103729369603e-05,
+      "loss": 1.2269,
+      "step": 35445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.859871797118635e-05,
+      "loss": 1.2554,
+      "step": 35450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8598332163899797e-05,
+      "loss": 1.1795,
+      "step": 35455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.859794630751214e-05,
+      "loss": 1.2013,
+      "step": 35460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8597560402025588e-05,
+      "loss": 1.0919,
+      "step": 35465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.859717444744234e-05,
+      "loss": 1.2841,
+      "step": 35470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8596788443764598e-05,
+      "loss": 1.2107,
+      "step": 35475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8596402390994575e-05,
+      "loss": 1.4652,
+      "step": 35480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.859601628913447e-05,
+      "loss": 1.095,
+      "step": 35485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8595630138186487e-05,
+      "loss": 1.1107,
+      "step": 35490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8595243938152837e-05,
+      "loss": 1.3451,
+      "step": 35495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8594857689035715e-05,
+      "loss": 1.0917,
+      "step": 35500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8594471390837336e-05,
+      "loss": 0.8884,
+      "step": 35505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8594085043559904e-05,
+      "loss": 1.0684,
+      "step": 35510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8593698647205623e-05,
+      "loss": 1.1739,
+      "step": 35515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8593312201776702e-05,
+      "loss": 1.2649,
+      "step": 35520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.859292570727534e-05,
+      "loss": 1.2764,
+      "step": 35525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.859253916370376e-05,
+      "loss": 1.5408,
+      "step": 35530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.859215257106415e-05,
+      "loss": 1.2224,
+      "step": 35535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8591765929358737e-05,
+      "loss": 1.0601,
+      "step": 35540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8591379238589712e-05,
+      "loss": 1.0368,
+      "step": 35545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8590992498759295e-05,
+      "loss": 1.0394,
+      "step": 35550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.859060570986969e-05,
+      "loss": 1.2432,
+      "step": 35555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.85902188719231e-05,
+      "loss": 1.2513,
+      "step": 35560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.858983198492175e-05,
+      "loss": 1.1447,
+      "step": 35565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8589445048867833e-05,
+      "loss": 1.0494,
+      "step": 35570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8589058063763566e-05,
+      "loss": 1.4098,
+      "step": 35575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8588671029611158e-05,
+      "loss": 0.9926,
+      "step": 35580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.858828394641282e-05,
+      "loss": 1.1582,
+      "step": 35585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8587896814170757e-05,
+      "loss": 1.1384,
+      "step": 35590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8587509632887188e-05,
+      "loss": 1.168,
+      "step": 35595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8587122402564318e-05,
+      "loss": 1.4812,
+      "step": 35600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8586735123204366e-05,
+      "loss": 1.0785,
+      "step": 35605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.858634779480953e-05,
+      "loss": 1.4455,
+      "step": 35610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8585960417382036e-05,
+      "loss": 1.2809,
+      "step": 35615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.858557299092409e-05,
+      "loss": 1.2126,
+      "step": 35620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.85851855154379e-05,
+      "loss": 1.2143,
+      "step": 35625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.858479799092569e-05,
+      "loss": 1.2993,
+      "step": 35630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.858441041738966e-05,
+      "loss": 1.0559,
+      "step": 35635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.858402279483203e-05,
+      "loss": 1.413,
+      "step": 35640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8583635123255015e-05,
+      "loss": 1.6645,
+      "step": 35645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8583247402660825e-05,
+      "loss": 1.0796,
+      "step": 35650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8582859633051675e-05,
+      "loss": 0.9852,
+      "step": 35655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.858247181442978e-05,
+      "loss": 0.9883,
+      "step": 35660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.858208394679736e-05,
+      "loss": 0.9892,
+      "step": 35665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.858169603015662e-05,
+      "loss": 1.2472,
+      "step": 35670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.858130806450978e-05,
+      "loss": 0.92,
+      "step": 35675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8580920049859058e-05,
+      "loss": 1.1558,
+      "step": 35680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.858053198620667e-05,
+      "loss": 1.3182,
+      "step": 35685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8580143873554822e-05,
+      "loss": 1.2471,
+      "step": 35690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8579755711905742e-05,
+      "loss": 1.0048,
+      "step": 35695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.857936750126164e-05,
+      "loss": 1.1938,
+      "step": 35700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8578979241624736e-05,
+      "loss": 1.6116,
+      "step": 35705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.857859093299725e-05,
+      "loss": 1.3498,
+      "step": 35710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.857820257538139e-05,
+      "loss": 0.957,
+      "step": 35715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8577814168779384e-05,
+      "loss": 1.3562,
+      "step": 35720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.857742571319344e-05,
+      "loss": 0.9637,
+      "step": 35725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8577037208625787e-05,
+      "loss": 1.1449,
+      "step": 35730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8576648655078634e-05,
+      "loss": 1.2255,
+      "step": 35735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8576260052554208e-05,
+      "loss": 1.4783,
+      "step": 35740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8575871401054725e-05,
+      "loss": 1.4328,
+      "step": 35745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.85754827005824e-05,
+      "loss": 1.2042,
+      "step": 35750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.857509395113946e-05,
+      "loss": 1.0255,
+      "step": 35755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8574705152728118e-05,
+      "loss": 1.1691,
+      "step": 35760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.85743163053506e-05,
+      "loss": 1.0552,
+      "step": 35765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8573927409009123e-05,
+      "loss": 1.3469,
+      "step": 35770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.857353846370591e-05,
+      "loss": 1.5234,
+      "step": 35775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.857314946944318e-05,
+      "loss": 1.1057,
+      "step": 35780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8572760426223157e-05,
+      "loss": 1.3528,
+      "step": 35785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.857237133404806e-05,
+      "loss": 1.416,
+      "step": 35790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8571982192920114e-05,
+      "loss": 1.0276,
+      "step": 35795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8571593002841536e-05,
+      "loss": 1.2735,
+      "step": 35800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8571203763814553e-05,
+      "loss": 1.1955,
+      "step": 35805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8570814475841385e-05,
+      "loss": 1.2567,
+      "step": 35810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8570425138924258e-05,
+      "loss": 1.2573,
+      "step": 35815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8570035753065395e-05,
+      "loss": 1.1503,
+      "step": 35820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8569646318267018e-05,
+      "loss": 1.4339,
+      "step": 35825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.856925683453135e-05,
+      "loss": 1.1916,
+      "step": 35830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.856886730186062e-05,
+      "loss": 1.3173,
+      "step": 35835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8568477720257042e-05,
+      "loss": 1.3595,
+      "step": 35840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8568088089722852e-05,
+      "loss": 1.031,
+      "step": 35845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8567698410260272e-05,
+      "loss": 1.205,
+      "step": 35850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8567308681871524e-05,
+      "loss": 1.0674,
+      "step": 35855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8566918904558833e-05,
+      "loss": 1.4338,
+      "step": 35860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.856652907832443e-05,
+      "loss": 1.4692,
+      "step": 35865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.856613920317054e-05,
+      "loss": 1.712,
+      "step": 35870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8565749279099384e-05,
+      "loss": 1.1303,
+      "step": 35875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.85653593061132e-05,
+      "loss": 1.0217,
+      "step": 35880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8564969284214202e-05,
+      "loss": 1.0874,
+      "step": 35885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8564579213404623e-05,
+      "loss": 1.4236,
+      "step": 35890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8564189093686686e-05,
+      "loss": 1.3714,
+      "step": 35895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.856379892506263e-05,
+      "loss": 1.1842,
+      "step": 35900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8563408707534672e-05,
+      "loss": 1.5817,
+      "step": 35905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8563018441105046e-05,
+      "loss": 1.1858,
+      "step": 35910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8562628125775977e-05,
+      "loss": 1.2552,
+      "step": 35915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.85622377615497e-05,
+      "loss": 1.3803,
+      "step": 35920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.856184734842844e-05,
+      "loss": 1.1615,
+      "step": 35925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.856145688641442e-05,
+      "loss": 1.4285,
+      "step": 35930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8561066375509883e-05,
+      "loss": 1.5763,
+      "step": 35935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.856067581571705e-05,
+      "loss": 1.2109,
+      "step": 35940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8560285207038154e-05,
+      "loss": 1.3982,
+      "step": 35945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8559894549475425e-05,
+      "loss": 1.257,
+      "step": 35950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8559503843031095e-05,
+      "loss": 1.1478,
+      "step": 35955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8559113087707394e-05,
+      "loss": 1.1126,
+      "step": 35960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8558722283506555e-05,
+      "loss": 1.1493,
+      "step": 35965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8558331430430807e-05,
+      "loss": 1.4049,
+      "step": 35970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8557940528482382e-05,
+      "loss": 1.1678,
+      "step": 35975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8557549577663514e-05,
+      "loss": 1.417,
+      "step": 35980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8557158577976436e-05,
+      "loss": 1.0065,
+      "step": 35985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.855676752942338e-05,
+      "loss": 1.4382,
+      "step": 35990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.855637643200658e-05,
+      "loss": 1.6697,
+      "step": 35995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.855598528572827e-05,
+      "loss": 1.0402,
+      "step": 36000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.855559409059068e-05,
+      "loss": 1.0289,
+      "step": 36005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8555202846596045e-05,
+      "loss": 1.3089,
+      "step": 36010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.85548115537466e-05,
+      "loss": 1.2674,
+      "step": 36015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8554420212044583e-05,
+      "loss": 1.1061,
+      "step": 36020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8554028821492226e-05,
+      "loss": 1.0944,
+      "step": 36025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8553637382091757e-05,
+      "loss": 1.4445,
+      "step": 36030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8553245893845426e-05,
+      "loss": 1.1439,
+      "step": 36035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8552854356755455e-05,
+      "loss": 1.4835,
+      "step": 36040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.855246277082409e-05,
+      "loss": 1.0924,
+      "step": 36045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.855207113605356e-05,
+      "loss": 1.1045,
+      "step": 36050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.85516794524461e-05,
+      "loss": 1.1989,
+      "step": 36055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.855128772000396e-05,
+      "loss": 1.2808,
+      "step": 36060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.855089593872936e-05,
+      "loss": 1.1043,
+      "step": 36065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8550504108624548e-05,
+      "loss": 1.1185,
+      "step": 36070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.855011222969176e-05,
+      "loss": 1.2423,
+      "step": 36075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.854972030193323e-05,
+      "loss": 1.2093,
+      "step": 36080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8549328325351198e-05,
+      "loss": 1.1268,
+      "step": 36085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8548936299947905e-05,
+      "loss": 1.2888,
+      "step": 36090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8548544225725587e-05,
+      "loss": 1.3984,
+      "step": 36095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8548152102686484e-05,
+      "loss": 1.2821,
+      "step": 36100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8547759930832834e-05,
+      "loss": 1.1065,
+      "step": 36105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8547367710166878e-05,
+      "loss": 1.05,
+      "step": 36110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8546975440690858e-05,
+      "loss": 1.117,
+      "step": 36115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8546583122407008e-05,
+      "loss": 1.1431,
+      "step": 36120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.854619075531757e-05,
+      "loss": 1.4067,
+      "step": 36125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.854579833942479e-05,
+      "loss": 1.1905,
+      "step": 36130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.85454058747309e-05,
+      "loss": 1.0612,
+      "step": 36135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.854501336123815e-05,
+      "loss": 1.1636,
+      "step": 36140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8544620798948777e-05,
+      "loss": 1.4995,
+      "step": 36145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8544228187865024e-05,
+      "loss": 1.4808,
+      "step": 36150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8543835527989134e-05,
+      "loss": 1.107,
+      "step": 36155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8543442819323346e-05,
+      "loss": 1.0859,
+      "step": 36160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8543050061869904e-05,
+      "loss": 1.1538,
+      "step": 36165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.854265725563105e-05,
+      "loss": 1.115,
+      "step": 36170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8542264400609032e-05,
+      "loss": 1.3212,
+      "step": 36175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8541871496806088e-05,
+      "loss": 1.2634,
+      "step": 36180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8541478544224464e-05,
+      "loss": 1.4852,
+      "step": 36185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8541085542866404e-05,
+      "loss": 1.3748,
+      "step": 36190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8540692492734148e-05,
+      "loss": 1.0767,
+      "step": 36195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8540299393829947e-05,
+      "loss": 1.1867,
+      "step": 36200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8539906246156043e-05,
+      "loss": 1.1826,
+      "step": 36205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.853951304971468e-05,
+      "loss": 1.2367,
+      "step": 36210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8539119804508107e-05,
+      "loss": 1.2812,
+      "step": 36215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8538726510538565e-05,
+      "loss": 1.2472,
+      "step": 36220
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8538333167808305e-05,
+      "loss": 1.2471,
+      "step": 36225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8537939776319565e-05,
+      "loss": 1.4636,
+      "step": 36230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8537546336074602e-05,
+      "loss": 1.3679,
+      "step": 36235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8537152847075656e-05,
+      "loss": 0.9964,
+      "step": 36240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8536759309324973e-05,
+      "loss": 1.3362,
+      "step": 36245
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8536365722824803e-05,
+      "loss": 1.089,
+      "step": 36250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8535972087577396e-05,
+      "loss": 1.2321,
+      "step": 36255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8535578403584998e-05,
+      "loss": 0.8964,
+      "step": 36260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.853518467084985e-05,
+      "loss": 1.2559,
+      "step": 36265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8534790889374214e-05,
+      "loss": 1.0454,
+      "step": 36270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.853439705916033e-05,
+      "loss": 1.3314,
+      "step": 36275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8534003180210448e-05,
+      "loss": 1.4227,
+      "step": 36280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8533609252526816e-05,
+      "loss": 1.1698,
+      "step": 36285
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8533215276111685e-05,
+      "loss": 0.9204,
+      "step": 36290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.853282125096731e-05,
+      "loss": 1.482,
+      "step": 36295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.853242717709593e-05,
+      "loss": 1.2791,
+      "step": 36300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8532033054499806e-05,
+      "loss": 1.3835,
+      "step": 36305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.853163888318118e-05,
+      "loss": 1.2787,
+      "step": 36310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8531244663142315e-05,
+      "loss": 1.2837,
+      "step": 36315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8530850394385448e-05,
+      "loss": 1.0083,
+      "step": 36320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.853045607691284e-05,
+      "loss": 1.02,
+      "step": 36325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8530061710726736e-05,
+      "loss": 1.2185,
+      "step": 36330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8529667295829394e-05,
+      "loss": 1.0803,
+      "step": 36335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8529272832223066e-05,
+      "loss": 1.1273,
+      "step": 36340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.852887831991e-05,
+      "loss": 1.2616,
+      "step": 36345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.852848375889245e-05,
+      "loss": 1.2384,
+      "step": 36350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8528089149172673e-05,
+      "loss": 1.4421,
+      "step": 36355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.852769449075292e-05,
+      "loss": 1.2479,
+      "step": 36360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8527299783635447e-05,
+      "loss": 0.8005,
+      "step": 36365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8526905027822502e-05,
+      "loss": 1.2043,
+      "step": 36370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8526510223316347e-05,
+      "loss": 1.1966,
+      "step": 36375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8526115370119232e-05,
+      "loss": 0.9826,
+      "step": 36380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852572046823341e-05,
+      "loss": 1.2402,
+      "step": 36385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852532551766114e-05,
+      "loss": 1.9125,
+      "step": 36390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8524930518404674e-05,
+      "loss": 1.2368,
+      "step": 36395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852453547046627e-05,
+      "loss": 1.2477,
+      "step": 36400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8524140373848187e-05,
+      "loss": 1.0688,
+      "step": 36405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852374522855267e-05,
+      "loss": 1.4407,
+      "step": 36410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852335003458199e-05,
+      "loss": 1.1301,
+      "step": 36415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8522954791938397e-05,
+      "loss": 1.0789,
+      "step": 36420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8522559500624145e-05,
+      "loss": 1.2903,
+      "step": 36425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8522164160641494e-05,
+      "loss": 1.272,
+      "step": 36430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8521768771992705e-05,
+      "loss": 1.2499,
+      "step": 36435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8521373334680032e-05,
+      "loss": 1.2786,
+      "step": 36440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8520977848705734e-05,
+      "loss": 1.1487,
+      "step": 36445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.852058231407207e-05,
+      "loss": 1.3421,
+      "step": 36450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8520186730781295e-05,
+      "loss": 1.0674,
+      "step": 36455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8519791098835674e-05,
+      "loss": 1.0807,
+      "step": 36460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8519395418237463e-05,
+      "loss": 1.3588,
+      "step": 36465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8518999688988923e-05,
+      "loss": 1.0675,
+      "step": 36470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851860391109231e-05,
+      "loss": 1.5154,
+      "step": 36475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851820808454989e-05,
+      "loss": 1.3548,
+      "step": 36480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851781220936392e-05,
+      "loss": 1.1999,
+      "step": 36485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8517416285536657e-05,
+      "loss": 1.1937,
+      "step": 36490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851702031307037e-05,
+      "loss": 1.0691,
+      "step": 36495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8516624291967314e-05,
+      "loss": 1.4354,
+      "step": 36500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8516228222229757e-05,
+      "loss": 1.105,
+      "step": 36505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851583210385995e-05,
+      "loss": 1.272,
+      "step": 36510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8515435936860166e-05,
+      "loss": 1.4853,
+      "step": 36515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8515039721232656e-05,
+      "loss": 1.3703,
+      "step": 36520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8514643456979698e-05,
+      "loss": 1.2665,
+      "step": 36525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8514247144103542e-05,
+      "loss": 1.4496,
+      "step": 36530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8513850782606455e-05,
+      "loss": 1.2377,
+      "step": 36535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8513454372490703e-05,
+      "loss": 1.1312,
+      "step": 36540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8513057913758547e-05,
+      "loss": 1.1994,
+      "step": 36545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851266140641225e-05,
+      "loss": 1.1735,
+      "step": 36550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8512264850454077e-05,
+      "loss": 1.4172,
+      "step": 36555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8511868245886296e-05,
+      "loss": 1.3234,
+      "step": 36560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851147159271117e-05,
+      "loss": 1.0409,
+      "step": 36565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851107489093096e-05,
+      "loss": 1.1411,
+      "step": 36570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8510678140547935e-05,
+      "loss": 1.1026,
+      "step": 36575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.851028134156436e-05,
+      "loss": 1.0715,
+      "step": 36580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8509884493982507e-05,
+      "loss": 1.2806,
+      "step": 36585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850948759780463e-05,
+      "loss": 1.2361,
+      "step": 36590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8509090653033005e-05,
+      "loss": 1.0086,
+      "step": 36595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8508693659669895e-05,
+      "loss": 1.2545,
+      "step": 36600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8508296617717567e-05,
+      "loss": 1.9443,
+      "step": 36605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850789952717829e-05,
+      "loss": 1.1327,
+      "step": 36610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850750238805433e-05,
+      "loss": 1.1386,
+      "step": 36615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8507105200347956e-05,
+      "loss": 1.6971,
+      "step": 36620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8506707964061438e-05,
+      "loss": 0.8636,
+      "step": 36625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850631067919704e-05,
+      "loss": 1.3484,
+      "step": 36630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8505913345757034e-05,
+      "loss": 1.2982,
+      "step": 36635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8505515963743684e-05,
+      "loss": 1.4144,
+      "step": 36640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8505118533159267e-05,
+      "loss": 1.3099,
+      "step": 36645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8504721054006048e-05,
+      "loss": 1.7368,
+      "step": 36650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8504323526286298e-05,
+      "loss": 1.134,
+      "step": 36655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8503925950002286e-05,
+      "loss": 1.1988,
+      "step": 36660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850352832515628e-05,
+      "loss": 1.1943,
+      "step": 36665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850313065175056e-05,
+      "loss": 1.1683,
+      "step": 36670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8502732929787384e-05,
+      "loss": 1.3468,
+      "step": 36675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8502335159269035e-05,
+      "loss": 1.3393,
+      "step": 36680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8501937340197774e-05,
+      "loss": 1.2662,
+      "step": 36685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8501539472575882e-05,
+      "loss": 1.2779,
+      "step": 36690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8501141556405626e-05,
+      "loss": 1.1949,
+      "step": 36695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850074359168928e-05,
+      "loss": 0.9427,
+      "step": 36700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8500345578429116e-05,
+      "loss": 1.018,
+      "step": 36705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8499947516627405e-05,
+      "loss": 1.316,
+      "step": 36710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8499549406286423e-05,
+      "loss": 1.0843,
+      "step": 36715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849915124740844e-05,
+      "loss": 1.3535,
+      "step": 36720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8498753039995735e-05,
+      "loss": 1.2857,
+      "step": 36725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849835478405058e-05,
+      "loss": 1.423,
+      "step": 36730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8497956479575246e-05,
+      "loss": 1.6102,
+      "step": 36735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8497558126572014e-05,
+      "loss": 0.9793,
+      "step": 36740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849715972504315e-05,
+      "loss": 1.0031,
+      "step": 36745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8496761274990936e-05,
+      "loss": 1.1573,
+      "step": 36750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849636277641764e-05,
+      "loss": 1.1689,
+      "step": 36755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849596422932555e-05,
+      "loss": 1.2615,
+      "step": 36760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8495565633716933e-05,
+      "loss": 0.9388,
+      "step": 36765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8495166989594063e-05,
+      "loss": 1.2987,
+      "step": 36770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8494768296959225e-05,
+      "loss": 1.2349,
+      "step": 36775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8494369555814687e-05,
+      "loss": 1.4315,
+      "step": 36780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849397076616273e-05,
+      "loss": 1.2053,
+      "step": 36785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8493571928005637e-05,
+      "loss": 1.1626,
+      "step": 36790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8493173041345674e-05,
+      "loss": 1.3288,
+      "step": 36795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849277410618513e-05,
+      "loss": 1.4319,
+      "step": 36800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8492375122526275e-05,
+      "loss": 1.2802,
+      "step": 36805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849197609037139e-05,
+      "loss": 1.8781,
+      "step": 36810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8491577009722757e-05,
+      "loss": 1.5353,
+      "step": 36815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849117788058265e-05,
+      "loss": 1.3304,
+      "step": 36820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849077870295335e-05,
+      "loss": 1.1764,
+      "step": 36825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8490379476837138e-05,
+      "loss": 1.1342,
+      "step": 36830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848998020223629e-05,
+      "loss": 1.4264,
+      "step": 36835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848958087915309e-05,
+      "loss": 1.2656,
+      "step": 36840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8489181507589817e-05,
+      "loss": 0.9472,
+      "step": 36845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8488782087548752e-05,
+      "loss": 1.3061,
+      "step": 36850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8488382619032178e-05,
+      "loss": 1.4348,
+      "step": 36855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848798310204237e-05,
+      "loss": 1.3273,
+      "step": 36860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8487583536581615e-05,
+      "loss": 1.3002,
+      "step": 36865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8487183922652195e-05,
+      "loss": 1.1194,
+      "step": 36870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8486784260256386e-05,
+      "loss": 1.2807,
+      "step": 36875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8486384549396475e-05,
+      "loss": 1.0341,
+      "step": 36880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8485984790074744e-05,
+      "loss": 0.9139,
+      "step": 36885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8485584982293478e-05,
+      "loss": 1.1917,
+      "step": 36890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8485185126054955e-05,
+      "loss": 1.6361,
+      "step": 36895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848478522136146e-05,
+      "loss": 1.1254,
+      "step": 36900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848438526821528e-05,
+      "loss": 1.4186,
+      "step": 36905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8483985266618698e-05,
+      "loss": 0.933,
+      "step": 36910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8483585216573998e-05,
+      "loss": 1.1943,
+      "step": 36915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848318511808346e-05,
+      "loss": 1.2715,
+      "step": 36920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8482784971149374e-05,
+      "loss": 1.0379,
+      "step": 36925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848238477577402e-05,
+      "loss": 1.0411,
+      "step": 36930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8481984531959693e-05,
+      "loss": 1.2173,
+      "step": 36935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8481584239708666e-05,
+      "loss": 1.1471,
+      "step": 36940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8481183899023235e-05,
+      "loss": 1.4335,
+      "step": 36945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8480783509905684e-05,
+      "loss": 1.3241,
+      "step": 36950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848038307235829e-05,
+      "loss": 1.2789,
+      "step": 36955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8479982586383355e-05,
+      "loss": 1.2868,
+      "step": 36960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8479582051983157e-05,
+      "loss": 1.2585,
+      "step": 36965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847918146915998e-05,
+      "loss": 1.3397,
+      "step": 36970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847878083791612e-05,
+      "loss": 1.0219,
+      "step": 36975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8478380158253864e-05,
+      "loss": 1.4311,
+      "step": 36980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8477979430175495e-05,
+      "loss": 1.0973,
+      "step": 36985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8477578653683302e-05,
+      "loss": 1.2503,
+      "step": 36990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8477177828779574e-05,
+      "loss": 0.9258,
+      "step": 36995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8476776955466605e-05,
+      "loss": 1.3536,
+      "step": 37000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8476376033746676e-05,
+      "loss": 1.1817,
+      "step": 37005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8475975063622084e-05,
+      "loss": 1.2064,
+      "step": 37010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8475574045095115e-05,
+      "loss": 1.3882,
+      "step": 37015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847517297816806e-05,
+      "loss": 1.2951,
+      "step": 37020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847477186284321e-05,
+      "loss": 1.2205,
+      "step": 37025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8474370699122853e-05,
+      "loss": 0.9089,
+      "step": 37030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847396948700928e-05,
+      "loss": 1.4698,
+      "step": 37035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8473568226504784e-05,
+      "loss": 1.359,
+      "step": 37040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8473166917611654e-05,
+      "loss": 1.1643,
+      "step": 37045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8472765560332188e-05,
+      "loss": 0.9747,
+      "step": 37050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847236415466867e-05,
+      "loss": 1.136,
+      "step": 37055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8471962700623402e-05,
+      "loss": 1.1528,
+      "step": 37060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8471561198198663e-05,
+      "loss": 1.3264,
+      "step": 37065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8471159647396756e-05,
+      "loss": 1.4882,
+      "step": 37070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.847075804821997e-05,
+      "loss": 1.1715,
+      "step": 37075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.84703564006706e-05,
+      "loss": 1.2238,
+      "step": 37080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.846995470475094e-05,
+      "loss": 1.0851,
+      "step": 37085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8469552960463285e-05,
+      "loss": 1.3879,
+      "step": 37090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.846915116780992e-05,
+      "loss": 1.1058,
+      "step": 37095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8468749326793153e-05,
+      "loss": 1.3243,
+      "step": 37100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8468347437415274e-05,
+      "loss": 1.175,
+      "step": 37105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8467945499678572e-05,
+      "loss": 1.1914,
+      "step": 37110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.846754351358535e-05,
+      "loss": 1.0387,
+      "step": 37115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.84671414791379e-05,
+      "loss": 1.102,
+      "step": 37120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8466739396338514e-05,
+      "loss": 1.1543,
+      "step": 37125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8466337265189493e-05,
+      "loss": 1.2534,
+      "step": 37130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8465935085693135e-05,
+      "loss": 1.2895,
+      "step": 37135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8465532857851732e-05,
+      "loss": 1.5398,
+      "step": 37140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8465130581667587e-05,
+      "loss": 1.2883,
+      "step": 37145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8464728257142993e-05,
+      "loss": 1.2347,
+      "step": 37150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8464325884280242e-05,
+      "loss": 1.5812,
+      "step": 37155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8463923463081644e-05,
+      "loss": 1.4144,
+      "step": 37160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.846352099354949e-05,
+      "loss": 1.1203,
+      "step": 37165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8463118475686075e-05,
+      "loss": 1.1741,
+      "step": 37170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8462715909493702e-05,
+      "loss": 1.0888,
+      "step": 37175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8462313294974674e-05,
+      "loss": 0.887,
+      "step": 37180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.846191063213128e-05,
+      "loss": 1.1628,
+      "step": 37185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.846150792096583e-05,
+      "loss": 1.1971,
+      "step": 37190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8461105161480616e-05,
+      "loss": 1.2759,
+      "step": 37195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8460702353677943e-05,
+      "loss": 1.3239,
+      "step": 37200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8460299497560107e-05,
+      "loss": 1.6539,
+      "step": 37205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.845989659312941e-05,
+      "loss": 1.1164,
+      "step": 37210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8459493640388153e-05,
+      "loss": 1.2819,
+      "step": 37215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.845909063933864e-05,
+      "loss": 1.13,
+      "step": 37220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.845868758998317e-05,
+      "loss": 1.4149,
+      "step": 37225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8458284492324042e-05,
+      "loss": 1.5818,
+      "step": 37230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.845788134636356e-05,
+      "loss": 1.0913,
+      "step": 37235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8457478152104028e-05,
+      "loss": 1.3057,
+      "step": 37240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.845707490954774e-05,
+      "loss": 1.18,
+      "step": 37245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8456671618697015e-05,
+      "loss": 1.2854,
+      "step": 37250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8456268279554145e-05,
+      "loss": 1.0218,
+      "step": 37255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8455864892121435e-05,
+      "loss": 0.9241,
+      "step": 37260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8455461456401186e-05,
+      "loss": 1.2844,
+      "step": 37265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8455057972395702e-05,
+      "loss": 1.3772,
+      "step": 37270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8454654440107293e-05,
+      "loss": 0.995,
+      "step": 37275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.845425085953826e-05,
+      "loss": 1.1323,
+      "step": 37280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8453847230690905e-05,
+      "loss": 1.3169,
+      "step": 37285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8453443553567537e-05,
+      "loss": 1.27,
+      "step": 37290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.845303982817046e-05,
+      "loss": 1.367,
+      "step": 37295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8452636054501978e-05,
+      "loss": 1.3432,
+      "step": 37300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.84522322325644e-05,
+      "loss": 1.1565,
+      "step": 37305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8451828362360026e-05,
+      "loss": 1.207,
+      "step": 37310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.845142444389117e-05,
+      "loss": 1.2079,
+      "step": 37315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8451020477160132e-05,
+      "loss": 1.1401,
+      "step": 37320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8450616462169226e-05,
+      "loss": 1.0735,
+      "step": 37325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8450212398920752e-05,
+      "loss": 1.2679,
+      "step": 37330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8449808287417018e-05,
+      "loss": 1.2784,
+      "step": 37335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8449404127660338e-05,
+      "loss": 1.3236,
+      "step": 37340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8448999919653015e-05,
+      "loss": 0.9827,
+      "step": 37345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8448595663397357e-05,
+      "loss": 1.1666,
+      "step": 37350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8448191358895672e-05,
+      "loss": 1.4962,
+      "step": 37355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8447787006150273e-05,
+      "loss": 1.3779,
+      "step": 37360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8447382605163465e-05,
+      "loss": 1.1253,
+      "step": 37365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.844697815593756e-05,
+      "loss": 1.0198,
+      "step": 37370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.844657365847487e-05,
+      "loss": 1.244,
+      "step": 37375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8446169112777695e-05,
+      "loss": 1.3309,
+      "step": 37380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8445764518848355e-05,
+      "loss": 1.0591,
+      "step": 37385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8445359876689156e-05,
+      "loss": 1.2977,
+      "step": 37390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.844495518630241e-05,
+      "loss": 1.0791,
+      "step": 37395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8444550447690427e-05,
+      "loss": 1.1789,
+      "step": 37400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.844414566085552e-05,
+      "loss": 1.4075,
+      "step": 37405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.84437408258e-05,
+      "loss": 1.2438,
+      "step": 37410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8443335942526177e-05,
+      "loss": 1.0606,
+      "step": 37415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8442931011036366e-05,
+      "loss": 1.1573,
+      "step": 37420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8442526031332873e-05,
+      "loss": 1.1099,
+      "step": 37425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8442121003418024e-05,
+      "loss": 1.1493,
+      "step": 37430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8441715927294118e-05,
+      "loss": 1.2432,
+      "step": 37435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8441310802963476e-05,
+      "loss": 1.0621,
+      "step": 37440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8440905630428407e-05,
+      "loss": 1.3265,
+      "step": 37445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8440500409691226e-05,
+      "loss": 1.2185,
+      "step": 37450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.844009514075425e-05,
+      "loss": 1.478,
+      "step": 37455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.843968982361979e-05,
+      "loss": 0.9912,
+      "step": 37460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8439284458290165e-05,
+      "loss": 0.9215,
+      "step": 37465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8438879044767683e-05,
+      "loss": 1.1468,
+      "step": 37470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8438473583054664e-05,
+      "loss": 1.3118,
+      "step": 37475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8438068073153422e-05,
+      "loss": 1.0774,
+      "step": 37480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8437662515066272e-05,
+      "loss": 1.4171,
+      "step": 37485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8437256908795534e-05,
+      "loss": 1.3203,
+      "step": 37490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.843685125434352e-05,
+      "loss": 1.3125,
+      "step": 37495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8436445551712544e-05,
+      "loss": 1.4376,
+      "step": 37500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.843603980090493e-05,
+      "loss": 1.2459,
+      "step": 37505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.843563400192299e-05,
+      "loss": 0.9972,
+      "step": 37510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8435228154769046e-05,
+      "loss": 1.2257,
+      "step": 37515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8434822259445408e-05,
+      "loss": 1.2496,
+      "step": 37520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.84344163159544e-05,
+      "loss": 1.5233,
+      "step": 37525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.843401032429834e-05,
+      "loss": 1.5473,
+      "step": 37530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.843360428447954e-05,
+      "loss": 1.0596,
+      "step": 37535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8433198196500327e-05,
+      "loss": 1.0339,
+      "step": 37540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8432792060363012e-05,
+      "loss": 1.2684,
+      "step": 37545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8432385876069923e-05,
+      "loss": 1.1089,
+      "step": 37550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8431979643623377e-05,
+      "loss": 0.9859,
+      "step": 37555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.843157336302569e-05,
+      "loss": 1.2551,
+      "step": 37560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8431167034279183e-05,
+      "loss": 1.7098,
+      "step": 37565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.843076065738618e-05,
+      "loss": 1.1489,
+      "step": 37570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8430354232348998e-05,
+      "loss": 1.446,
+      "step": 37575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.842994775916996e-05,
+      "loss": 1.2159,
+      "step": 37580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8429541237851384e-05,
+      "loss": 1.0514,
+      "step": 37585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8429134668395595e-05,
+      "loss": 1.3433,
+      "step": 37590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8428728050804913e-05,
+      "loss": 1.4046,
+      "step": 37595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.842832138508166e-05,
+      "loss": 1.505,
+      "step": 37600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8427914671228157e-05,
+      "loss": 1.5187,
+      "step": 37605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.842750790924673e-05,
+      "loss": 1.2967,
+      "step": 37610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8427101099139702e-05,
+      "loss": 1.2745,
+      "step": 37615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8426694240909392e-05,
+      "loss": 1.0019,
+      "step": 37620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8426287334558127e-05,
+      "loss": 1.2055,
+      "step": 37625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8425880380088228e-05,
+      "loss": 1.6131,
+      "step": 37630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.842547337750202e-05,
+      "loss": 0.8863,
+      "step": 37635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.842506632680183e-05,
+      "loss": 1.2419,
+      "step": 37640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8424659227989974e-05,
+      "loss": 1.0902,
+      "step": 37645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.842425208106879e-05,
+      "loss": 1.197,
+      "step": 37650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.842384488604059e-05,
+      "loss": 1.0196,
+      "step": 37655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.842343764290771e-05,
+      "loss": 1.545,
+      "step": 37660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8423030351672466e-05,
+      "loss": 1.0519,
+      "step": 37665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.842262301233719e-05,
+      "loss": 1.2123,
+      "step": 37670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8422215624904204e-05,
+      "loss": 1.3093,
+      "step": 37675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.842180818937584e-05,
+      "loss": 1.1934,
+      "step": 37680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.842140070575442e-05,
+      "loss": 1.1848,
+      "step": 37685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8420993174042275e-05,
+      "loss": 1.2884,
+      "step": 37690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8420585594241727e-05,
+      "loss": 1.331,
+      "step": 37695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.842017796635511e-05,
+      "loss": 1.0594,
+      "step": 37700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8419770290384745e-05,
+      "loss": 1.2785,
+      "step": 37705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8419362566332962e-05,
+      "loss": 1.0758,
+      "step": 37710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.841895479420209e-05,
+      "loss": 1.1198,
+      "step": 37715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.841854697399446e-05,
+      "loss": 1.2512,
+      "step": 37720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8418139105712398e-05,
+      "loss": 1.0841,
+      "step": 37725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8417731189358233e-05,
+      "loss": 1.2003,
+      "step": 37730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.84173232249343e-05,
+      "loss": 1.4534,
+      "step": 37735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8416915212442918e-05,
+      "loss": 1.3312,
+      "step": 37740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8416507151886426e-05,
+      "loss": 1.2918,
+      "step": 37745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.841609904326715e-05,
+      "loss": 1.1929,
+      "step": 37750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8415690886587424e-05,
+      "loss": 1.2749,
+      "step": 37755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8415282681849576e-05,
+      "loss": 1.0308,
+      "step": 37760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.841487442905594e-05,
+      "loss": 1.1365,
+      "step": 37765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.841446612820884e-05,
+      "loss": 1.0995,
+      "step": 37770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8414057779310616e-05,
+      "loss": 0.9114,
+      "step": 37775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8413649382363595e-05,
+      "loss": 1.5215,
+      "step": 37780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8413240937370113e-05,
+      "loss": 1.8661,
+      "step": 37785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8412832444332497e-05,
+      "loss": 1.1258,
+      "step": 37790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8412423903253085e-05,
+      "loss": 1.4126,
+      "step": 37795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8412015314134207e-05,
+      "loss": 1.4335,
+      "step": 37800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8411606676978196e-05,
+      "loss": 1.1145,
+      "step": 37805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.841119799178739e-05,
+      "loss": 1.2787,
+      "step": 37810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8410789258564115e-05,
+      "loss": 1.2056,
+      "step": 37815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8410380477310714e-05,
+      "loss": 1.4706,
+      "step": 37820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8409971648029518e-05,
+      "loss": 1.3645,
+      "step": 37825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8409562770722855e-05,
+      "loss": 1.3823,
+      "step": 37830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.840915384539307e-05,
+      "loss": 1.3484,
+      "step": 37835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.840874487204249e-05,
+      "loss": 1.9245,
+      "step": 37840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8408335850673458e-05,
+      "loss": 1.475,
+      "step": 37845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8407926781288305e-05,
+      "loss": 1.0501,
+      "step": 37850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8407517663889367e-05,
+      "loss": 1.1769,
+      "step": 37855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.840710849847898e-05,
+      "loss": 1.3032,
+      "step": 37860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.840669928505948e-05,
+      "loss": 1.2079,
+      "step": 37865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.840629002363321e-05,
+      "loss": 1.5025,
+      "step": 37870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.84058807142025e-05,
+      "loss": 1.4909,
+      "step": 37875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.840547135676969e-05,
+      "loss": 1.2495,
+      "step": 37880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.840506195133712e-05,
+      "loss": 1.1286,
+      "step": 37885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8404652497907124e-05,
+      "loss": 1.2662,
+      "step": 37890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8404242996482042e-05,
+      "loss": 1.0183,
+      "step": 37895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8403833447064208e-05,
+      "loss": 1.2641,
+      "step": 37900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.840342384965597e-05,
+      "loss": 1.4537,
+      "step": 37905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8403014204259662e-05,
+      "loss": 1.4505,
+      "step": 37910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8402604510877622e-05,
+      "loss": 1.0805,
+      "step": 37915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8402194769512192e-05,
+      "loss": 1.244,
+      "step": 37920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.840178498016571e-05,
+      "loss": 1.2043,
+      "step": 37925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8401375142840517e-05,
+      "loss": 1.419,
+      "step": 37930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.840096525753895e-05,
+      "loss": 1.4516,
+      "step": 37935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8400555324263357e-05,
+      "loss": 1.256,
+      "step": 37940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8400145343016075e-05,
+      "loss": 1.1699,
+      "step": 37945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8399735313799444e-05,
+      "loss": 1.2722,
+      "step": 37950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8399325236615805e-05,
+      "loss": 1.1875,
+      "step": 37955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8398915111467505e-05,
+      "loss": 1.2358,
+      "step": 37960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.839850493835688e-05,
+      "loss": 1.0997,
+      "step": 37965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8398094717286273e-05,
+      "loss": 0.995,
+      "step": 37970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.839768444825803e-05,
+      "loss": 1.1676,
+      "step": 37975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.839727413127449e-05,
+      "loss": 1.058,
+      "step": 37980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8396863766338e-05,
+      "loss": 1.3793,
+      "step": 37985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8396453353450905e-05,
+      "loss": 1.2003,
+      "step": 37990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.839604289261554e-05,
+      "loss": 1.2801,
+      "step": 37995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8395632383834255e-05,
+      "loss": 1.2062,
+      "step": 38000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8395221827109394e-05,
+      "loss": 1.0563,
+      "step": 38005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8394811222443306e-05,
+      "loss": 1.3427,
+      "step": 38010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8394400569838327e-05,
+      "loss": 1.3892,
+      "step": 38015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8393989869296805e-05,
+      "loss": 1.2622,
+      "step": 38020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8393579120821083e-05,
+      "loss": 1.6501,
+      "step": 38025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.839316832441351e-05,
+      "loss": 1.4278,
+      "step": 38030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8392757480076438e-05,
+      "loss": 1.1998,
+      "step": 38035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8392346587812204e-05,
+      "loss": 1.5629,
+      "step": 38040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8391935647623155e-05,
+      "loss": 0.9965,
+      "step": 38045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8391524659511644e-05,
+      "loss": 1.5124,
+      "step": 38050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.839111362348001e-05,
+      "loss": 1.2876,
+      "step": 38055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8390702539530604e-05,
+      "loss": 1.3031,
+      "step": 38060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8390291407665778e-05,
+      "loss": 1.3049,
+      "step": 38065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8389880227887872e-05,
+      "loss": 1.6738,
+      "step": 38070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8389469000199235e-05,
+      "loss": 1.4492,
+      "step": 38075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.838905772460222e-05,
+      "loss": 1.3394,
+      "step": 38080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8388646401099175e-05,
+      "loss": 1.4191,
+      "step": 38085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.838823502969245e-05,
+      "loss": 1.4167,
+      "step": 38090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8387823610384383e-05,
+      "loss": 1.0598,
+      "step": 38095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8387412143177336e-05,
+      "loss": 0.9421,
+      "step": 38100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8387000628073655e-05,
+      "loss": 1.0739,
+      "step": 38105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.838658906507569e-05,
+      "loss": 1.3638,
+      "step": 38110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.838617745418579e-05,
+      "loss": 1.0476,
+      "step": 38115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8385765795406303e-05,
+      "loss": 1.093,
+      "step": 38120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8385354088739586e-05,
+      "loss": 1.3128,
+      "step": 38125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8384942334187985e-05,
+      "loss": 0.9443,
+      "step": 38130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8384530531753856e-05,
+      "loss": 1.4016,
+      "step": 38135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.838411868143955e-05,
+      "loss": 1.2249,
+      "step": 38140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8383706783247414e-05,
+      "loss": 1.1158,
+      "step": 38145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.83832948371798e-05,
+      "loss": 1.247,
+      "step": 38150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8382882843239067e-05,
+      "loss": 1.2077,
+      "step": 38155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8382470801427565e-05,
+      "loss": 1.0406,
+      "step": 38160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8382058711747643e-05,
+      "loss": 1.1395,
+      "step": 38165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8381646574201663e-05,
+      "loss": 1.3982,
+      "step": 38170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8381234388791967e-05,
+      "loss": 1.2857,
+      "step": 38175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.838082215552092e-05,
+      "loss": 1.3374,
+      "step": 38180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8380409874390867e-05,
+      "loss": 1.2263,
+      "step": 38185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8379997545404164e-05,
+      "loss": 1.1759,
+      "step": 38190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8379585168563173e-05,
+      "loss": 1.2925,
+      "step": 38195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8379172743870242e-05,
+      "loss": 1.1084,
+      "step": 38200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8378760271327728e-05,
+      "loss": 0.9624,
+      "step": 38205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8378347750937986e-05,
+      "loss": 1.3576,
+      "step": 38210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.837793518270337e-05,
+      "loss": 1.4137,
+      "step": 38215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8377522566626243e-05,
+      "loss": 1.2179,
+      "step": 38220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8377109902708953e-05,
+      "loss": 1.7433,
+      "step": 38225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.837669719095386e-05,
+      "loss": 1.1515,
+      "step": 38230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.837628443136332e-05,
+      "loss": 1.341,
+      "step": 38235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.837587162393969e-05,
+      "loss": 1.0769,
+      "step": 38240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.837545876868533e-05,
+      "loss": 1.2044,
+      "step": 38245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8375045865602595e-05,
+      "loss": 1.1385,
+      "step": 38250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8374632914693844e-05,
+      "loss": 0.9987,
+      "step": 38255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8374219915961435e-05,
+      "loss": 1.3832,
+      "step": 38260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8373806869407726e-05,
+      "loss": 1.196,
+      "step": 38265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8373393775035076e-05,
+      "loss": 0.9648,
+      "step": 38270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8372980632845844e-05,
+      "loss": 1.0575,
+      "step": 38275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8372567442842388e-05,
+      "loss": 1.2554,
+      "step": 38280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.837215420502707e-05,
+      "loss": 1.115,
+      "step": 38285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8371740919402245e-05,
+      "loss": 1.1906,
+      "step": 38290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.837132758597028e-05,
+      "loss": 1.1792,
+      "step": 38295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.837091420473353e-05,
+      "loss": 1.6142,
+      "step": 38300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.837050077569436e-05,
+      "loss": 1.8804,
+      "step": 38305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8370087298855128e-05,
+      "loss": 1.1597,
+      "step": 38310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8369673774218194e-05,
+      "loss": 1.635,
+      "step": 38315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8369260201785922e-05,
+      "loss": 1.1114,
+      "step": 38320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8368846581560677e-05,
+      "loss": 1.1521,
+      "step": 38325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8368432913544813e-05,
+      "loss": 1.6186,
+      "step": 38330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8368019197740695e-05,
+      "loss": 1.0906,
+      "step": 38335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8367605434150685e-05,
+      "loss": 1.163,
+      "step": 38340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.836719162277715e-05,
+      "loss": 1.0408,
+      "step": 38345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.836677776362245e-05,
+      "loss": 1.0956,
+      "step": 38350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.836636385668895e-05,
+      "loss": 1.1947,
+      "step": 38355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8365949901979013e-05,
+      "loss": 1.5094,
+      "step": 38360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8365535899494998e-05,
+      "loss": 1.1508,
+      "step": 38365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8365121849239277e-05,
+      "loss": 1.2347,
+      "step": 38370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.836470775121421e-05,
+      "loss": 1.1928,
+      "step": 38375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8364293605422166e-05,
+      "loss": 1.2067,
+      "step": 38380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8363879411865506e-05,
+      "loss": 1.4027,
+      "step": 38385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8363465170546596e-05,
+      "loss": 1.0417,
+      "step": 38390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8363050881467798e-05,
+      "loss": 1.3875,
+      "step": 38395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8362636544631483e-05,
+      "loss": 1.0415,
+      "step": 38400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8362222160040016e-05,
+      "loss": 1.5011,
+      "step": 38405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8361807727695763e-05,
+      "loss": 1.3853,
+      "step": 38410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.836139324760109e-05,
+      "loss": 1.0419,
+      "step": 38415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8360978719758365e-05,
+      "loss": 1.5305,
+      "step": 38420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.836056414416995e-05,
+      "loss": 1.3245,
+      "step": 38425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8360149520838226e-05,
+      "loss": 1.2519,
+      "step": 38430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8359734849765546e-05,
+      "loss": 1.218,
+      "step": 38435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8359320130954286e-05,
+      "loss": 1.0661,
+      "step": 38440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.835890536440681e-05,
+      "loss": 1.4007,
+      "step": 38445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8358490550125487e-05,
+      "loss": 1.2205,
+      "step": 38450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8358075688112688e-05,
+      "loss": 1.1676,
+      "step": 38455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8357660778370782e-05,
+      "loss": 1.1384,
+      "step": 38460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.835724582090214e-05,
+      "loss": 1.1206,
+      "step": 38465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.835683081570913e-05,
+      "loss": 1.036,
+      "step": 38470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8356415762794118e-05,
+      "loss": 1.0289,
+      "step": 38475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.835600066215948e-05,
+      "loss": 1.1803,
+      "step": 38480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.835558551380758e-05,
+      "loss": 1.4789,
+      "step": 38485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8355170317740796e-05,
+      "loss": 1.2289,
+      "step": 38490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8354755073961493e-05,
+      "loss": 1.3852,
+      "step": 38495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8354339782472047e-05,
+      "loss": 1.2348,
+      "step": 38500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8353924443274828e-05,
+      "loss": 1.5565,
+      "step": 38505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.835350905637221e-05,
+      "loss": 1.273,
+      "step": 38510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8353093621766554e-05,
+      "loss": 1.2757,
+      "step": 38515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8352678139460246e-05,
+      "loss": 1.2655,
+      "step": 38520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8352262609455652e-05,
+      "loss": 1.1781,
+      "step": 38525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8351847031755148e-05,
+      "loss": 1.1348,
+      "step": 38530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8351431406361105e-05,
+      "loss": 1.4082,
+      "step": 38535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8351015733275896e-05,
+      "loss": 1.3776,
+      "step": 38540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8350600012501898e-05,
+      "loss": 1.1079,
+      "step": 38545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8350184244041477e-05,
+      "loss": 1.2894,
+      "step": 38550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834976842789702e-05,
+      "loss": 0.9623,
+      "step": 38555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8349352564070888e-05,
+      "loss": 1.2007,
+      "step": 38560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8348936652565467e-05,
+      "loss": 0.9826,
+      "step": 38565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8348520693383122e-05,
+      "loss": 1.1076,
+      "step": 38570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8348104686526237e-05,
+      "loss": 1.2313,
+      "step": 38575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8347688631997185e-05,
+      "loss": 1.5512,
+      "step": 38580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8347272529798336e-05,
+      "loss": 1.2894,
+      "step": 38585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8346856379932077e-05,
+      "loss": 1.0444,
+      "step": 38590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8346440182400777e-05,
+      "loss": 1.1783,
+      "step": 38595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8346023937206814e-05,
+      "loss": 1.2748,
+      "step": 38600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8345607644352564e-05,
+      "loss": 1.1212,
+      "step": 38605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834519130384041e-05,
+      "loss": 1.2851,
+      "step": 38610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834477491567272e-05,
+      "loss": 1.4136,
+      "step": 38615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834435847985188e-05,
+      "loss": 0.9397,
+      "step": 38620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8343941996380263e-05,
+      "loss": 1.2638,
+      "step": 38625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834352546526025e-05,
+      "loss": 1.303,
+      "step": 38630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834310888649422e-05,
+      "loss": 1.0741,
+      "step": 38635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834269226008455e-05,
+      "loss": 1.0263,
+      "step": 38640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8342275586033623e-05,
+      "loss": 1.1812,
+      "step": 38645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8341858864343814e-05,
+      "loss": 1.124,
+      "step": 38650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83414420950175e-05,
+      "loss": 1.0775,
+      "step": 38655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.834102527805707e-05,
+      "loss": 1.167,
+      "step": 38660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8340608413464897e-05,
+      "loss": 1.129,
+      "step": 38665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8340191501243364e-05,
+      "loss": 1.4888,
+      "step": 38670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8339774541394856e-05,
+      "loss": 1.0916,
+      "step": 38675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8339357533921745e-05,
+      "loss": 1.4274,
+      "step": 38680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8338940478826418e-05,
+      "loss": 1.5082,
+      "step": 38685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833852337611126e-05,
+      "loss": 1.0209,
+      "step": 38690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833810622577864e-05,
+      "loss": 1.3451,
+      "step": 38695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8337689027830954e-05,
+      "loss": 1.2817,
+      "step": 38700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833727178227058e-05,
+      "loss": 1.3947,
+      "step": 38705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83368544890999e-05,
+      "loss": 1.4614,
+      "step": 38710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8336437148321294e-05,
+      "loss": 1.1082,
+      "step": 38715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8336019759937148e-05,
+      "loss": 1.1731,
+      "step": 38720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833560232394985e-05,
+      "loss": 1.0529,
+      "step": 38725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8335184840361777e-05,
+      "loss": 1.2416,
+      "step": 38730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8334767309175316e-05,
+      "loss": 1.0812,
+      "step": 38735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833434973039285e-05,
+      "loss": 1.837,
+      "step": 38740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8333932104016762e-05,
+      "loss": 1.4225,
+      "step": 38745
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8333514430049444e-05,
+      "loss": 1.4165,
+      "step": 38750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8333096708493273e-05,
+      "loss": 1.156,
+      "step": 38755
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833267893935064e-05,
+      "loss": 1.5589,
+      "step": 38760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8332261122623927e-05,
+      "loss": 1.5314,
+      "step": 38765
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8331843258315522e-05,
+      "loss": 1.1482,
+      "step": 38770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833142534642781e-05,
+      "loss": 0.9408,
+      "step": 38775
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8331007386963177e-05,
+      "loss": 1.0634,
+      "step": 38780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8330589379924015e-05,
+      "loss": 1.2108,
+      "step": 38785
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8330171325312704e-05,
+      "loss": 1.0796,
+      "step": 38790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8329753223131633e-05,
+      "loss": 1.0953,
+      "step": 38795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8329335073383194e-05,
+      "loss": 1.0347,
+      "step": 38800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8328916876069772e-05,
+      "loss": 1.6899,
+      "step": 38805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8328498631193753e-05,
+      "loss": 1.3493,
+      "step": 38810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832808033875753e-05,
+      "loss": 1.0782,
+      "step": 38815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8327661998763484e-05,
+      "loss": 1.1077,
+      "step": 38820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8327243611214012e-05,
+      "loss": 1.6376,
+      "step": 38825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83268251761115e-05,
+      "loss": 1.1686,
+      "step": 38830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832640669345834e-05,
+      "loss": 1.1653,
+      "step": 38835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8325988163256914e-05,
+      "loss": 1.483,
+      "step": 38840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8325569585509622e-05,
+      "loss": 1.2276,
+      "step": 38845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8325150960218848e-05,
+      "loss": 1.3144,
+      "step": 38850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8324732287386985e-05,
+      "loss": 1.3555,
+      "step": 38855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8324313567016424e-05,
+      "loss": 1.127,
+      "step": 38860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832389479910955e-05,
+      "loss": 1.2769,
+      "step": 38865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8323475983668767e-05,
+      "loss": 1.2265,
+      "step": 38870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8323057120696455e-05,
+      "loss": 1.5295,
+      "step": 38875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832263821019501e-05,
+      "loss": 1.709,
+      "step": 38880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832221925216683e-05,
+      "loss": 1.1352,
+      "step": 38885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8321800246614298e-05,
+      "loss": 1.2454,
+      "step": 38890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8321381193539807e-05,
+      "loss": 1.0937,
+      "step": 38895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832096209294576e-05,
+      "loss": 1.0662,
+      "step": 38900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.832054294483454e-05,
+      "loss": 1.3731,
+      "step": 38905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8320123749208544e-05,
+      "loss": 1.1555,
+      "step": 38910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831970450607017e-05,
+      "loss": 1.1436,
+      "step": 38915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831928521542181e-05,
+      "loss": 1.2268,
+      "step": 38920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831886587726585e-05,
+      "loss": 1.0778,
+      "step": 38925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8318446491604692e-05,
+      "loss": 0.9621,
+      "step": 38930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8318027058440737e-05,
+      "loss": 1.1762,
+      "step": 38935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8317607577776367e-05,
+      "loss": 1.5594,
+      "step": 38940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831718804961399e-05,
+      "loss": 1.22,
+      "step": 38945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831676847395599e-05,
+      "loss": 1.0973,
+      "step": 38950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8316348850804772e-05,
+      "loss": 1.0961,
+      "step": 38955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831592918016273e-05,
+      "loss": 1.0475,
+      "step": 38960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831550946203226e-05,
+      "loss": 1.6446,
+      "step": 38965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8315089696415754e-05,
+      "loss": 1.2914,
+      "step": 38970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831466988331562e-05,
+      "loss": 1.335,
+      "step": 38975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8314250022734244e-05,
+      "loss": 1.1877,
+      "step": 38980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831383011467403e-05,
+      "loss": 1.3939,
+      "step": 38985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8313410159137376e-05,
+      "loss": 0.9904,
+      "step": 38990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8312990156126676e-05,
+      "loss": 1.0645,
+      "step": 38995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831257010564433e-05,
+      "loss": 1.405,
+      "step": 39000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831215000769274e-05,
+      "loss": 1.1476,
+      "step": 39005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8311729862274304e-05,
+      "loss": 1.2442,
+      "step": 39010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.831130966939142e-05,
+      "loss": 1.1685,
+      "step": 39015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8310889429046485e-05,
+      "loss": 1.6795,
+      "step": 39020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8310469141241902e-05,
+      "loss": 1.218,
+      "step": 39025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8310048805980073e-05,
+      "loss": 1.1393,
+      "step": 39030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8309628423263392e-05,
+      "loss": 1.3305,
+      "step": 39035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8309207993094262e-05,
+      "loss": 1.3524,
+      "step": 39040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.830878751547509e-05,
+      "loss": 0.9713,
+      "step": 39045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8308366990408272e-05,
+      "loss": 1.2896,
+      "step": 39050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8307946417896207e-05,
+      "loss": 1.523,
+      "step": 39055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8307525797941302e-05,
+      "loss": 1.1586,
+      "step": 39060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8307105130545954e-05,
+      "loss": 1.0611,
+      "step": 39065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8306684415712568e-05,
+      "loss": 1.1347,
+      "step": 39070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8306263653443547e-05,
+      "loss": 1.4621,
+      "step": 39075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8305842843741294e-05,
+      "loss": 1.0152,
+      "step": 39080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8305421986608212e-05,
+      "loss": 1.5858,
+      "step": 39085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83050010820467e-05,
+      "loss": 1.2238,
+      "step": 39090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8304580130059168e-05,
+      "loss": 1.0524,
+      "step": 39095
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8304159130648016e-05,
+      "loss": 1.1338,
+      "step": 39100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8303738083815648e-05,
+      "loss": 1.1376,
+      "step": 39105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8303316989564467e-05,
+      "loss": 1.4935,
+      "step": 39110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8302895847896884e-05,
+      "loss": 1.3816,
+      "step": 39115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83024746588153e-05,
+      "loss": 1.1481,
+      "step": 39120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.830205342232212e-05,
+      "loss": 1.1863,
+      "step": 39125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8301632138419748e-05,
+      "loss": 1.1008,
+      "step": 39130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.830121080711059e-05,
+      "loss": 1.5859,
+      "step": 39135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8300789428397056e-05,
+      "loss": 1.0534,
+      "step": 39140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.830036800228155e-05,
+      "loss": 1.1849,
+      "step": 39145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8299946528766477e-05,
+      "loss": 1.0374,
+      "step": 39150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8299525007854247e-05,
+      "loss": 1.2789,
+      "step": 39155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8299103439547264e-05,
+      "loss": 1.5736,
+      "step": 39160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8298681823847934e-05,
+      "loss": 1.1513,
+      "step": 39165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8298260160758672e-05,
+      "loss": 1.0447,
+      "step": 39170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8297838450281878e-05,
+      "loss": 1.2065,
+      "step": 39175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8297416692419964e-05,
+      "loss": 1.4627,
+      "step": 39180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829699488717534e-05,
+      "loss": 1.1899,
+      "step": 39185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8296573034550413e-05,
+      "loss": 1.5934,
+      "step": 39190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829615113454759e-05,
+      "loss": 1.2758,
+      "step": 39195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829572918716928e-05,
+      "loss": 1.6454,
+      "step": 39200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8295307192417896e-05,
+      "loss": 1.1159,
+      "step": 39205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8294885150295848e-05,
+      "loss": 1.389,
+      "step": 39210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8294463060805542e-05,
+      "loss": 1.2331,
+      "step": 39215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829404092394939e-05,
+      "loss": 1.3915,
+      "step": 39220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8293618739729807e-05,
+      "loss": 1.3327,
+      "step": 39225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.82931965081492e-05,
+      "loss": 1.3524,
+      "step": 39230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8292774229209976e-05,
+      "loss": 1.2681,
+      "step": 39235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8292351902914554e-05,
+      "loss": 1.42,
+      "step": 39240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.829192952926534e-05,
+      "loss": 1.1843,
+      "step": 39245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8291507108264755e-05,
+      "loss": 0.966,
+      "step": 39250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.82910846399152e-05,
+      "loss": 1.0759,
+      "step": 39255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8290662124219093e-05,
+      "loss": 1.2412,
+      "step": 39260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8290239561178843e-05,
+      "loss": 1.3906,
+      "step": 39265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828981695079687e-05,
+      "loss": 1.3726,
+      "step": 39270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8289394293075583e-05,
+      "loss": 1.3224,
+      "step": 39275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.82889715880174e-05,
+      "loss": 0.9905,
+      "step": 39280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8288548835624725e-05,
+      "loss": 1.3248,
+      "step": 39285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8288126035899977e-05,
+      "loss": 1.3922,
+      "step": 39290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8287703188845575e-05,
+      "loss": 1.0514,
+      "step": 39295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828728029446393e-05,
+      "loss": 1.6882,
+      "step": 39300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828685735275746e-05,
+      "loss": 1.3843,
+      "step": 39305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828643436372857e-05,
+      "loss": 1.1495,
+      "step": 39310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828601132737969e-05,
+      "loss": 1.2037,
+      "step": 39315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8285588243713226e-05,
+      "loss": 1.1305,
+      "step": 39320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8285165112731595e-05,
+      "loss": 1.2418,
+      "step": 39325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8284741934437217e-05,
+      "loss": 1.4751,
+      "step": 39330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8284318708832508e-05,
+      "loss": 1.2816,
+      "step": 39335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828389543591988e-05,
+      "loss": 1.2647,
+      "step": 39340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8283472115701755e-05,
+      "loss": 1.5952,
+      "step": 39345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828304874818055e-05,
+      "loss": 1.0941,
+      "step": 39350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828262533335868e-05,
+      "loss": 1.3868,
+      "step": 39355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8282201871238566e-05,
+      "loss": 1.3898,
+      "step": 39360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8281778361822624e-05,
+      "loss": 1.3924,
+      "step": 39365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828135480511327e-05,
+      "loss": 1.1064,
+      "step": 39370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828093120111293e-05,
+      "loss": 1.3766,
+      "step": 39375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828050754982402e-05,
+      "loss": 1.2957,
+      "step": 39380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8280083851248954e-05,
+      "loss": 1.437,
+      "step": 39385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.827966010539016e-05,
+      "loss": 0.9524,
+      "step": 39390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.827923631225005e-05,
+      "loss": 1.3896,
+      "step": 39395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8278812471831048e-05,
+      "loss": 1.2655,
+      "step": 39400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8278388584135576e-05,
+      "loss": 1.0097,
+      "step": 39405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8277964649166052e-05,
+      "loss": 1.3638,
+      "step": 39410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8277540666924898e-05,
+      "loss": 1.2832,
+      "step": 39415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8277116637414535e-05,
+      "loss": 1.1003,
+      "step": 39420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.827669256063738e-05,
+      "loss": 1.4736,
+      "step": 39425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8276268436595864e-05,
+      "loss": 1.6919,
+      "step": 39430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8275844265292402e-05,
+      "loss": 1.342,
+      "step": 39435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8275420046729418e-05,
+      "loss": 1.0545,
+      "step": 39440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8274995780909334e-05,
+      "loss": 1.2777,
+      "step": 39445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8274571467834576e-05,
+      "loss": 1.6017,
+      "step": 39450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.827414710750756e-05,
+      "loss": 1.3157,
+      "step": 39455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8273722699930715e-05,
+      "loss": 1.3032,
+      "step": 39460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8273298245106464e-05,
+      "loss": 1.1847,
+      "step": 39465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.827287374303723e-05,
+      "loss": 1.2718,
+      "step": 39470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.827244919372544e-05,
+      "loss": 1.1484,
+      "step": 39475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.827202459717351e-05,
+      "loss": 1.202,
+      "step": 39480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8271599953383877e-05,
+      "loss": 1.2542,
+      "step": 39485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8271175262358956e-05,
+      "loss": 1.2454,
+      "step": 39490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8270750524101177e-05,
+      "loss": 1.4044,
+      "step": 39495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8270325738612962e-05,
+      "loss": 0.9972,
+      "step": 39500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.826990090589674e-05,
+      "loss": 1.3812,
+      "step": 39505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8269476025954936e-05,
+      "loss": 1.0215,
+      "step": 39510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8269051098789975e-05,
+      "loss": 1.0971,
+      "step": 39515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8268626124404286e-05,
+      "loss": 1.1156,
+      "step": 39520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8268201102800294e-05,
+      "loss": 1.2707,
+      "step": 39525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8267776033980427e-05,
+      "loss": 1.7269,
+      "step": 39530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8267350917947112e-05,
+      "loss": 1.458,
+      "step": 39535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8266925754702776e-05,
+      "loss": 1.0368,
+      "step": 39540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8266500544249845e-05,
+      "loss": 1.0566,
+      "step": 39545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8266075286590754e-05,
+      "loss": 1.2918,
+      "step": 39550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8265649981727927e-05,
+      "loss": 1.4061,
+      "step": 39555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8265224629663793e-05,
+      "loss": 1.1747,
+      "step": 39560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8264799230400777e-05,
+      "loss": 1.0553,
+      "step": 39565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8264373783941316e-05,
+      "loss": 1.0714,
+      "step": 39570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8263948290287835e-05,
+      "loss": 1.2655,
+      "step": 39575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8263522749442764e-05,
+      "loss": 1.2441,
+      "step": 39580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.826309716140853e-05,
+      "loss": 1.2751,
+      "step": 39585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8262671526187575e-05,
+      "loss": 1.2216,
+      "step": 39590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8262245843782315e-05,
+      "loss": 1.3704,
+      "step": 39595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.826182011419519e-05,
+      "loss": 1.188,
+      "step": 39600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8261394337428627e-05,
+      "loss": 0.9681,
+      "step": 39605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8260968513485057e-05,
+      "loss": 1.1342,
+      "step": 39610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8260542642366918e-05,
+      "loss": 1.3378,
+      "step": 39615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8260116724076637e-05,
+      "loss": 1.2303,
+      "step": 39620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8259690758616643e-05,
+      "loss": 1.1703,
+      "step": 39625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8259264745989375e-05,
+      "loss": 1.3226,
+      "step": 39630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8258838686197263e-05,
+      "loss": 0.9389,
+      "step": 39635
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.825841257924274e-05,
+      "loss": 1.3688,
+      "step": 39640
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.825798642512824e-05,
+      "loss": 1.0885,
+      "step": 39645
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8257560223856195e-05,
+      "loss": 1.1772,
+      "step": 39650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.825713397542904e-05,
+      "loss": 1.2508,
+      "step": 39655
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.825670767984921e-05,
+      "loss": 1.4975,
+      "step": 39660
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8256281337119138e-05,
+      "loss": 1.7599,
+      "step": 39665
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8255854947241255e-05,
+      "loss": 1.3032,
+      "step": 39670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8255428510218005e-05,
+      "loss": 1.3905,
+      "step": 39675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8255002026051816e-05,
+      "loss": 0.9752,
+      "step": 39680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8254575494745122e-05,
+      "loss": 1.1102,
+      "step": 39685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.825414891630037e-05,
+      "loss": 1.3545,
+      "step": 39690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.825372229071998e-05,
+      "loss": 1.2474,
+      "step": 39695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8253295618006402e-05,
+      "loss": 1.2917,
+      "step": 39700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8252868898162067e-05,
+      "loss": 1.1388,
+      "step": 39705
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8252442131189408e-05,
+      "loss": 1.5992,
+      "step": 39710
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.825201531709087e-05,
+      "loss": 1.2784,
+      "step": 39715
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8251588455868884e-05,
+      "loss": 1.1349,
+      "step": 39720
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8251161547525887e-05,
+      "loss": 1.3038,
+      "step": 39725
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8250734592064322e-05,
+      "loss": 1.0776,
+      "step": 39730
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8250307589486626e-05,
+      "loss": 0.9789,
+      "step": 39735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8249880539795234e-05,
+      "loss": 1.456,
+      "step": 39740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.824945344299259e-05,
+      "loss": 0.9567,
+      "step": 39745
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.824902629908113e-05,
+      "loss": 0.9676,
+      "step": 39750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.824859910806329e-05,
+      "loss": 1.2776,
+      "step": 39755
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8248171869941512e-05,
+      "loss": 1.1944,
+      "step": 39760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.824774458471824e-05,
+      "loss": 1.5479,
+      "step": 39765
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8247317252395907e-05,
+      "loss": 1.2819,
+      "step": 39770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.824688987297696e-05,
+      "loss": 1.016,
+      "step": 39775
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8246462446463834e-05,
+      "loss": 1.1681,
+      "step": 39780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8246034972858974e-05,
+      "loss": 1.0746,
+      "step": 39785
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8245607452164815e-05,
+      "loss": 1.295,
+      "step": 39790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8245179884383806e-05,
+      "loss": 1.4568,
+      "step": 39795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8244752269518383e-05,
+      "loss": 1.1898,
+      "step": 39800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.824432460757099e-05,
+      "loss": 1.1616,
+      "step": 39805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8243896898544072e-05,
+      "loss": 1.5553,
+      "step": 39810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8243469142440067e-05,
+      "loss": 1.3757,
+      "step": 39815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.824304133926142e-05,
+      "loss": 1.2723,
+      "step": 39820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8242613489010575e-05,
+      "loss": 1.279,
+      "step": 39825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.824218559168997e-05,
+      "loss": 1.3725,
+      "step": 39830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8241757647302054e-05,
+      "loss": 1.3998,
+      "step": 39835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.824132965584927e-05,
+      "loss": 1.0184,
+      "step": 39840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.824090161733406e-05,
+      "loss": 1.1475,
+      "step": 39845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8240473531758867e-05,
+      "loss": 1.0427,
+      "step": 39850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8240045399126144e-05,
+      "loss": 1.3731,
+      "step": 39855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8239617219438322e-05,
+      "loss": 0.9761,
+      "step": 39860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8239188992697857e-05,
+      "loss": 1.1204,
+      "step": 39865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8238760718907196e-05,
+      "loss": 1.0459,
+      "step": 39870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8238332398068777e-05,
+      "loss": 1.323,
+      "step": 39875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8237904030185047e-05,
+      "loss": 1.1721,
+      "step": 39880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8237475615258456e-05,
+      "loss": 1.4178,
+      "step": 39885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8237047153291446e-05,
+      "loss": 1.2068,
+      "step": 39890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8236618644286474e-05,
+      "loss": 1.1582,
+      "step": 39895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8236190088245972e-05,
+      "loss": 1.3115,
+      "step": 39900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8235761485172397e-05,
+      "loss": 1.5228,
+      "step": 39905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8235332835068192e-05,
+      "loss": 1.0333,
+      "step": 39910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8234904137935814e-05,
+      "loss": 1.4875,
+      "step": 39915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8234475393777697e-05,
+      "loss": 1.2458,
+      "step": 39920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.82340466025963e-05,
+      "loss": 1.2125,
+      "step": 39925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.823361776439407e-05,
+      "loss": 1.169,
+      "step": 39930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.823318887917345e-05,
+      "loss": 1.424,
+      "step": 39935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8232759946936896e-05,
+      "loss": 1.1153,
+      "step": 39940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.823233096768685e-05,
+      "loss": 1.2335,
+      "step": 39945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.823190194142577e-05,
+      "loss": 1.2254,
+      "step": 39950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8231472868156106e-05,
+      "loss": 1.225,
+      "step": 39955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.82310437478803e-05,
+      "loss": 1.2362,
+      "step": 39960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8230614580600807e-05,
+      "loss": 1.2154,
+      "step": 39965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8230185366320082e-05,
+      "loss": 0.9613,
+      "step": 39970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.822975610504057e-05,
+      "loss": 1.4025,
+      "step": 39975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.822932679676472e-05,
+      "loss": 1.1723,
+      "step": 39980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.822889744149499e-05,
+      "loss": 1.3406,
+      "step": 39985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.822846803923383e-05,
+      "loss": 1.228,
+      "step": 39990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8228038589983696e-05,
+      "loss": 1.1655,
+      "step": 39995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.822760909374703e-05,
+      "loss": 1.192,
+      "step": 40000
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 1.229492425918579,
+      "eval_runtime": 6462.7155,
+      "eval_samples_per_second": 3.576,
+      "eval_steps_per_second": 1.788,
+      "step": 40000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8227179550526296e-05,
+      "loss": 1.3472,
+      "step": 40005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8226749960323936e-05,
+      "loss": 1.0328,
+      "step": 40010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8226320323142412e-05,
+      "loss": 1.1965,
+      "step": 40015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8225890638984176e-05,
+      "loss": 1.006,
+      "step": 40020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.822546090785168e-05,
+      "loss": 1.3798,
+      "step": 40025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8225031129747376e-05,
+      "loss": 1.3298,
+      "step": 40030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.822460130467372e-05,
+      "loss": 1.1299,
+      "step": 40035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8224171432633172e-05,
+      "loss": 0.9483,
+      "step": 40040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8223741513628174e-05,
+      "loss": 1.0565,
+      "step": 40045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8223311547661192e-05,
+      "loss": 1.3453,
+      "step": 40050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.822288153473468e-05,
+      "loss": 1.4991,
+      "step": 40055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.822245147485109e-05,
+      "loss": 1.2573,
+      "step": 40060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.822202136801288e-05,
+      "loss": 1.2575,
+      "step": 40065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8221591214222507e-05,
+      "loss": 1.345,
+      "step": 40070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8221161013482428e-05,
+      "loss": 1.1898,
+      "step": 40075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.822073076579509e-05,
+      "loss": 1.3764,
+      "step": 40080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8220300471162967e-05,
+      "loss": 1.1086,
+      "step": 40085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8219870129588502e-05,
+      "loss": 0.9336,
+      "step": 40090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8219439741074157e-05,
+      "loss": 1.108,
+      "step": 40095
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8219009305622394e-05,
+      "loss": 1.3353,
+      "step": 40100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8218578823235664e-05,
+      "loss": 1.1365,
+      "step": 40105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.821814829391643e-05,
+      "loss": 1.05,
+      "step": 40110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8217717717667147e-05,
+      "loss": 1.2587,
+      "step": 40115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8217287094490276e-05,
+      "loss": 1.3282,
+      "step": 40120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8216856424388275e-05,
+      "loss": 1.0306,
+      "step": 40125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.821642570736361e-05,
+      "loss": 1.4754,
+      "step": 40130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.821599494341873e-05,
+      "loss": 1.2384,
+      "step": 40135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8215564132556103e-05,
+      "loss": 1.3703,
+      "step": 40140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8215133274778184e-05,
+      "loss": 1.5814,
+      "step": 40145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8214702370087437e-05,
+      "loss": 1.2592,
+      "step": 40150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.821427141848632e-05,
+      "loss": 1.5711,
+      "step": 40155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8213840419977297e-05,
+      "loss": 1.2969,
+      "step": 40160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8213409374562823e-05,
+      "loss": 1.1904,
+      "step": 40165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8212978282245367e-05,
+      "loss": 1.1752,
+      "step": 40170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8212547143027386e-05,
+      "loss": 0.9621,
+      "step": 40175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8212115956911348e-05,
+      "loss": 1.0851,
+      "step": 40180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8211684723899707e-05,
+      "loss": 1.4125,
+      "step": 40185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.821125344399493e-05,
+      "loss": 0.9238,
+      "step": 40190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.821082211719948e-05,
+      "loss": 1.1221,
+      "step": 40195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8210390743515818e-05,
+      "loss": 1.3105,
+      "step": 40200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.820995932294641e-05,
+      "loss": 1.1621,
+      "step": 40205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8209527855493722e-05,
+      "loss": 1.177,
+      "step": 40210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.820909634116021e-05,
+      "loss": 1.2381,
+      "step": 40215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8208664779948343e-05,
+      "loss": 1.2109,
+      "step": 40220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8208233171860586e-05,
+      "loss": 1.1758,
+      "step": 40225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.82078015168994e-05,
+      "loss": 1.1285,
+      "step": 40230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.820736981506726e-05,
+      "loss": 1.3091,
+      "step": 40235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8206938066366614e-05,
+      "loss": 1.2297,
+      "step": 40240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8206506270799946e-05,
+      "loss": 1.3173,
+      "step": 40245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8206074428369712e-05,
+      "loss": 1.1277,
+      "step": 40250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8205642539078377e-05,
+      "loss": 1.3671,
+      "step": 40255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.820521060292841e-05,
+      "loss": 1.1553,
+      "step": 40260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8204778619922278e-05,
+      "loss": 1.1876,
+      "step": 40265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8204346590062443e-05,
+      "loss": 1.1458,
+      "step": 40270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.820391451335138e-05,
+      "loss": 1.2891,
+      "step": 40275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8203482389791552e-05,
+      "loss": 1.2064,
+      "step": 40280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8203050219385428e-05,
+      "loss": 1.2399,
+      "step": 40285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.820261800213548e-05,
+      "loss": 1.1487,
+      "step": 40290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8202185738044163e-05,
+      "loss": 0.9307,
+      "step": 40295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.820175342711396e-05,
+      "loss": 1.0708,
+      "step": 40300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8201321069347328e-05,
+      "loss": 1.2444,
+      "step": 40305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8200888664746745e-05,
+      "loss": 1.2843,
+      "step": 40310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8200456213314676e-05,
+      "loss": 1.5803,
+      "step": 40315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.820002371505359e-05,
+      "loss": 1.0799,
+      "step": 40320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.819959116996596e-05,
+      "loss": 1.0639,
+      "step": 40325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8199158578054255e-05,
+      "loss": 1.3577,
+      "step": 40330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8198725939320945e-05,
+      "loss": 1.6355,
+      "step": 40335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.81982932537685e-05,
+      "loss": 1.2796,
+      "step": 40340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.819786052139939e-05,
+      "loss": 1.2443,
+      "step": 40345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8197427742216086e-05,
+      "loss": 1.0176,
+      "step": 40350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8196994916221064e-05,
+      "loss": 1.0529,
+      "step": 40355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8196562043416785e-05,
+      "loss": 0.9211,
+      "step": 40360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8196129123805737e-05,
+      "loss": 1.0686,
+      "step": 40365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.819569615739038e-05,
+      "loss": 1.2867,
+      "step": 40370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8195263144173187e-05,
+      "loss": 1.0514,
+      "step": 40375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8194830084156636e-05,
+      "loss": 1.5814,
+      "step": 40380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.81943969773432e-05,
+      "loss": 1.1222,
+      "step": 40385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8193963823735344e-05,
+      "loss": 1.1086,
+      "step": 40390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8193530623335554e-05,
+      "loss": 1.2637,
+      "step": 40395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.819309737614629e-05,
+      "loss": 1.1487,
+      "step": 40400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.819266408217004e-05,
+      "loss": 1.2851,
+      "step": 40405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.819223074140927e-05,
+      "loss": 0.9804,
+      "step": 40410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8191797353866455e-05,
+      "loss": 1.2225,
+      "step": 40415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8191363919544072e-05,
+      "loss": 1.102,
+      "step": 40420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8190930438444593e-05,
+      "loss": 1.2358,
+      "step": 40425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.81904969105705e-05,
+      "loss": 0.9742,
+      "step": 40430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.819006333592426e-05,
+      "loss": 1.0079,
+      "step": 40435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8189629714508353e-05,
+      "loss": 1.3438,
+      "step": 40440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8189196046325263e-05,
+      "loss": 1.321,
+      "step": 40445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8188762331377453e-05,
+      "loss": 1.2169,
+      "step": 40450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8188328569667404e-05,
+      "loss": 1.697,
+      "step": 40455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.81878947611976e-05,
+      "loss": 1.3758,
+      "step": 40460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8187460905970508e-05,
+      "loss": 0.987,
+      "step": 40465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8187027003988615e-05,
+      "loss": 1.2143,
+      "step": 40470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8186593055254392e-05,
+      "loss": 1.2265,
+      "step": 40475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8186159059770317e-05,
+      "loss": 0.9863,
+      "step": 40480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8185725017538877e-05,
+      "loss": 1.1516,
+      "step": 40485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8185290928562538e-05,
+      "loss": 1.1248,
+      "step": 40490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.818485679284379e-05,
+      "loss": 1.1331,
+      "step": 40495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8184422610385106e-05,
+      "loss": 1.1857,
+      "step": 40500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8183988381188965e-05,
+      "loss": 1.2795,
+      "step": 40505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.818355410525785e-05,
+      "loss": 1.367,
+      "step": 40510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8183119782594242e-05,
+      "loss": 1.131,
+      "step": 40515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8182685413200612e-05,
+      "loss": 1.4438,
+      "step": 40520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8182250997079454e-05,
+      "loss": 1.2515,
+      "step": 40525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.818181653423324e-05,
+      "loss": 1.3008,
+      "step": 40530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8181382024664453e-05,
+      "loss": 1.1909,
+      "step": 40535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8180947468375575e-05,
+      "loss": 1.2872,
+      "step": 40540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8180512865369086e-05,
+      "loss": 1.0059,
+      "step": 40545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.818007821564747e-05,
+      "loss": 1.3112,
+      "step": 40550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.81796435192132e-05,
+      "loss": 1.3034,
+      "step": 40555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8179208776068777e-05,
+      "loss": 1.2508,
+      "step": 40560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8178773986216665e-05,
+      "loss": 1.3585,
+      "step": 40565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817833914965936e-05,
+      "loss": 1.5657,
+      "step": 40570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8177904266399335e-05,
+      "loss": 1.0067,
+      "step": 40575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817746933643908e-05,
+      "loss": 1.2156,
+      "step": 40580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8177034359781075e-05,
+      "loss": 1.1906,
+      "step": 40585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817659933642781e-05,
+      "loss": 1.5523,
+      "step": 40590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817616426638176e-05,
+      "loss": 0.923,
+      "step": 40595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8175729149645418e-05,
+      "loss": 1.1596,
+      "step": 40600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8175293986221266e-05,
+      "loss": 1.2064,
+      "step": 40605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8174858776111786e-05,
+      "loss": 1.3357,
+      "step": 40610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8174423519319467e-05,
+      "loss": 1.0972,
+      "step": 40615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8173988215846793e-05,
+      "loss": 1.4944,
+      "step": 40620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8173552865696248e-05,
+      "loss": 1.2031,
+      "step": 40625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817311746887032e-05,
+      "loss": 1.243,
+      "step": 40630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8172682025371497e-05,
+      "loss": 1.4009,
+      "step": 40635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8172246535202265e-05,
+      "loss": 1.1496,
+      "step": 40640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817181099836511e-05,
+      "loss": 1.0765,
+      "step": 40645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8171375414862516e-05,
+      "loss": 1.0324,
+      "step": 40650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8170939784696973e-05,
+      "loss": 1.4854,
+      "step": 40655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.817050410787097e-05,
+      "loss": 1.0353,
+      "step": 40660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8170068384386997e-05,
+      "loss": 1.4321,
+      "step": 40665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8169632614247538e-05,
+      "loss": 1.3264,
+      "step": 40670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816919679745508e-05,
+      "loss": 1.3598,
+      "step": 40675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8168760934012116e-05,
+      "loss": 1.1562,
+      "step": 40680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8168325023921134e-05,
+      "loss": 1.1003,
+      "step": 40685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816788906718462e-05,
+      "loss": 1.2654,
+      "step": 40690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816745306380507e-05,
+      "loss": 1.2692,
+      "step": 40695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816701701378497e-05,
+      "loss": 1.4203,
+      "step": 40700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816658091712681e-05,
+      "loss": 1.7028,
+      "step": 40705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8166144773833077e-05,
+      "loss": 1.1555,
+      "step": 40710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816570858390627e-05,
+      "loss": 1.1779,
+      "step": 40715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816527234734887e-05,
+      "loss": 1.2437,
+      "step": 40720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8164836064163377e-05,
+      "loss": 1.4101,
+      "step": 40725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8164399734352275e-05,
+      "loss": 1.2022,
+      "step": 40730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8163963357918064e-05,
+      "loss": 1.1498,
+      "step": 40735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8163526934863228e-05,
+      "loss": 1.3326,
+      "step": 40740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8163090465190266e-05,
+      "loss": 1.1992,
+      "step": 40745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8162653948901662e-05,
+      "loss": 1.0467,
+      "step": 40750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8162217385999918e-05,
+      "loss": 1.2543,
+      "step": 40755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816178077648752e-05,
+      "loss": 1.2371,
+      "step": 40760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8161344120366965e-05,
+      "loss": 1.1818,
+      "step": 40765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.816090741764074e-05,
+      "loss": 1.2233,
+      "step": 40770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8160470668311355e-05,
+      "loss": 1.1844,
+      "step": 40775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8160033872381285e-05,
+      "loss": 1.2623,
+      "step": 40780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815959702985304e-05,
+      "loss": 1.0317,
+      "step": 40785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.81591601407291e-05,
+      "loss": 1.0494,
+      "step": 40790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815872320501197e-05,
+      "loss": 1.1765,
+      "step": 40795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8158286222704142e-05,
+      "loss": 1.2087,
+      "step": 40800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815784919380811e-05,
+      "loss": 1.0075,
+      "step": 40805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8157412118326375e-05,
+      "loss": 1.1626,
+      "step": 40810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8156974996261428e-05,
+      "loss": 1.2649,
+      "step": 40815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8156537827615767e-05,
+      "loss": 0.9536,
+      "step": 40820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8156100612391886e-05,
+      "loss": 1.2425,
+      "step": 40825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8155663350592285e-05,
+      "loss": 1.251,
+      "step": 40830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815522604221946e-05,
+      "loss": 1.053,
+      "step": 40835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8154788687275904e-05,
+      "loss": 1.254,
+      "step": 40840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8154351285764125e-05,
+      "loss": 1.3727,
+      "step": 40845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815391383768661e-05,
+      "loss": 1.0569,
+      "step": 40850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8153476343045863e-05,
+      "loss": 1.5372,
+      "step": 40855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815303880184438e-05,
+      "loss": 1.2835,
+      "step": 40860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8152601214084663e-05,
+      "loss": 1.2593,
+      "step": 40865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8152163579769205e-05,
+      "loss": 1.4661,
+      "step": 40870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815172589890051e-05,
+      "loss": 1.4586,
+      "step": 40875
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8151288171481075e-05,
+      "loss": 1.2938,
+      "step": 40880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8150850397513398e-05,
+      "loss": 1.2989,
+      "step": 40885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8150412576999987e-05,
+      "loss": 1.1728,
+      "step": 40890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814997470994333e-05,
+      "loss": 1.1978,
+      "step": 40895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8149536796345938e-05,
+      "loss": 1.2415,
+      "step": 40900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814909883621031e-05,
+      "loss": 1.1874,
+      "step": 40905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814866082953894e-05,
+      "loss": 1.1431,
+      "step": 40910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8148222776334333e-05,
+      "loss": 1.2161,
+      "step": 40915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8147784676598994e-05,
+      "loss": 0.9067,
+      "step": 40920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814734653033542e-05,
+      "loss": 1.211,
+      "step": 40925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814690833754612e-05,
+      "loss": 1.2552,
+      "step": 40930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8146470098233587e-05,
+      "loss": 1.0693,
+      "step": 40935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814603181240033e-05,
+      "loss": 1.0163,
+      "step": 40940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814559348004885e-05,
+      "loss": 1.1925,
+      "step": 40945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814515510118165e-05,
+      "loss": 1.2604,
+      "step": 40950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8144716675801232e-05,
+      "loss": 1.1432,
+      "step": 40955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8144278203910105e-05,
+      "loss": 1.294,
+      "step": 40960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8143839685510768e-05,
+      "loss": 1.1527,
+      "step": 40965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814340112060572e-05,
+      "loss": 1.1345,
+      "step": 40970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814296250919748e-05,
+      "loss": 1.6167,
+      "step": 40975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814252385128854e-05,
+      "loss": 1.0555,
+      "step": 40980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814208514688141e-05,
+      "loss": 1.2231,
+      "step": 40985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8141646395978594e-05,
+      "loss": 1.0389,
+      "step": 40990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.81412075985826e-05,
+      "loss": 1.1325,
+      "step": 40995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814076875469593e-05,
+      "loss": 1.0165,
+      "step": 41000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8140329864321092e-05,
+      "loss": 1.0834,
+      "step": 41005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8139890927460594e-05,
+      "loss": 1.4097,
+      "step": 41010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.813945194411694e-05,
+      "loss": 1.1307,
+      "step": 41015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8139012914292635e-05,
+      "loss": 1.357,
+      "step": 41020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.813857383799019e-05,
+      "loss": 1.2103,
+      "step": 41025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.813813471521211e-05,
+      "loss": 1.0823,
+      "step": 41030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8137695545960907e-05,
+      "loss": 1.1223,
+      "step": 41035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8137256330239084e-05,
+      "loss": 1.2949,
+      "step": 41040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8136817068049152e-05,
+      "loss": 1.2505,
+      "step": 41045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8136377759393615e-05,
+      "loss": 1.3701,
+      "step": 41050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8135938404274986e-05,
+      "loss": 1.254,
+      "step": 41055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8135499002695774e-05,
+      "loss": 1.2564,
+      "step": 41060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8135059554658485e-05,
+      "loss": 1.0993,
+      "step": 41065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8134620060165633e-05,
+      "loss": 1.2442,
+      "step": 41070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8134180519219723e-05,
+      "loss": 1.1401,
+      "step": 41075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8133740931823268e-05,
+      "loss": 1.1885,
+      "step": 41080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.813330129797878e-05,
+      "loss": 1.2206,
+      "step": 41085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8132861617688765e-05,
+      "loss": 1.0821,
+      "step": 41090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8132421890955732e-05,
+      "loss": 1.0739,
+      "step": 41095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.81319821177822e-05,
+      "loss": 1.231,
+      "step": 41100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8131542298170677e-05,
+      "loss": 1.3278,
+      "step": 41105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8131102432123672e-05,
+      "loss": 1.2284,
+      "step": 41110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.81306625196437e-05,
+      "loss": 1.1077,
+      "step": 41115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.813022256073327e-05,
+      "loss": 1.0822,
+      "step": 41120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.81297825553949e-05,
+      "loss": 1.2876,
+      "step": 41125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8129342503631097e-05,
+      "loss": 1.1083,
+      "step": 41130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8128902405444376e-05,
+      "loss": 1.3237,
+      "step": 41135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.812846226083725e-05,
+      "loss": 1.1749,
+      "step": 41140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8128022069812235e-05,
+      "loss": 1.238,
+      "step": 41145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8127581832371837e-05,
+      "loss": 1.1653,
+      "step": 41150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8127141548518577e-05,
+      "loss": 1.1782,
+      "step": 41155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8126701218254973e-05,
+      "loss": 1.0497,
+      "step": 41160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8126260841583527e-05,
+      "loss": 1.3359,
+      "step": 41165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8125820418506765e-05,
+      "loss": 1.056,
+      "step": 41170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8125379949027197e-05,
+      "loss": 1.1939,
+      "step": 41175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8124939433147336e-05,
+      "loss": 1.2956,
+      "step": 41180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8124498870869705e-05,
+      "loss": 1.3755,
+      "step": 41185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8124058262196813e-05,
+      "loss": 1.0885,
+      "step": 41190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.812361760713118e-05,
+      "loss": 1.2145,
+      "step": 41195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8123176905675322e-05,
+      "loss": 1.3931,
+      "step": 41200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8122736157831756e-05,
+      "loss": 0.951,
+      "step": 41205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8122295363602995e-05,
+      "loss": 1.053,
+      "step": 41210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.812185452299156e-05,
+      "loss": 1.0116,
+      "step": 41215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8121413635999965e-05,
+      "loss": 1.1986,
+      "step": 41220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.812097270263073e-05,
+      "loss": 0.9704,
+      "step": 41225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.812053172288638e-05,
+      "loss": 0.9495,
+      "step": 41230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.812009069676942e-05,
+      "loss": 0.9967,
+      "step": 41235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8119649624282373e-05,
+      "loss": 1.1822,
+      "step": 41240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8119208505427764e-05,
+      "loss": 1.0541,
+      "step": 41245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8118767340208106e-05,
+      "loss": 1.3507,
+      "step": 41250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8118326128625917e-05,
+      "loss": 1.3531,
+      "step": 41255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8117884870683724e-05,
+      "loss": 1.285,
+      "step": 41260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8117443566384038e-05,
+      "loss": 0.9615,
+      "step": 41265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8117002215729386e-05,
+      "loss": 1.2728,
+      "step": 41270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8116560818722286e-05,
+      "loss": 1.3139,
+      "step": 41275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.811611937536526e-05,
+      "loss": 1.4378,
+      "step": 41280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.811567788566082e-05,
+      "loss": 1.2172,
+      "step": 41285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.81152363496115e-05,
+      "loss": 1.4305,
+      "step": 41290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8114794767219816e-05,
+      "loss": 1.0335,
+      "step": 41295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8114353138488286e-05,
+      "loss": 1.3011,
+      "step": 41300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8113911463419436e-05,
+      "loss": 1.1773,
+      "step": 41305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8113469742015787e-05,
+      "loss": 1.3368,
+      "step": 41310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8113027974279866e-05,
+      "loss": 1.2851,
+      "step": 41315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.811258616021419e-05,
+      "loss": 0.9809,
+      "step": 41320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.811214429982128e-05,
+      "loss": 1.0346,
+      "step": 41325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8111702393103664e-05,
+      "loss": 1.0992,
+      "step": 41330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8111260440063866e-05,
+      "loss": 1.213,
+      "step": 41335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.811081844070441e-05,
+      "loss": 1.1203,
+      "step": 41340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8110376395027814e-05,
+      "loss": 1.188,
+      "step": 41345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8109934303036612e-05,
+      "loss": 1.2162,
+      "step": 41350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8109492164733317e-05,
+      "loss": 1.3354,
+      "step": 41355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8109049980120464e-05,
+      "loss": 1.0172,
+      "step": 41360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8108607749200575e-05,
+      "loss": 1.135,
+      "step": 41365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8108165471976172e-05,
+      "loss": 1.2733,
+      "step": 41370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8107723148449784e-05,
+      "loss": 1.3395,
+      "step": 41375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8107280778623934e-05,
+      "loss": 1.2129,
+      "step": 41380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8106838362501155e-05,
+      "loss": 1.3766,
+      "step": 41385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8106395900083965e-05,
+      "loss": 1.1676,
+      "step": 41390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8105953391374894e-05,
+      "loss": 1.3565,
+      "step": 41395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.810551083637647e-05,
+      "loss": 1.1636,
+      "step": 41400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8105068235091223e-05,
+      "loss": 1.285,
+      "step": 41405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.810462558752167e-05,
+      "loss": 1.07,
+      "step": 41410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8104182893670354e-05,
+      "loss": 1.0951,
+      "step": 41415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.810374015353979e-05,
+      "loss": 1.2387,
+      "step": 41420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8103297367132514e-05,
+      "loss": 1.1231,
+      "step": 41425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8102854534451047e-05,
+      "loss": 1.2905,
+      "step": 41430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8102411655497925e-05,
+      "loss": 1.3362,
+      "step": 41435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8101968730275676e-05,
+      "loss": 1.2774,
+      "step": 41440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8101525758786823e-05,
+      "loss": 1.1784,
+      "step": 41445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8101082741033905e-05,
+      "loss": 1.4395,
+      "step": 41450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8100639677019446e-05,
+      "loss": 1.0625,
+      "step": 41455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.810019656674598e-05,
+      "loss": 1.1716,
+      "step": 41460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.809975341021603e-05,
+      "loss": 1.3596,
+      "step": 41465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8099310207432137e-05,
+      "loss": 1.0167,
+      "step": 41470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.809886695839682e-05,
+      "loss": 0.9593,
+      "step": 41475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8098423663112623e-05,
+      "loss": 1.3745,
+      "step": 41480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.809798032158207e-05,
+      "loss": 1.3379,
+      "step": 41485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.809753693380769e-05,
+      "loss": 1.244,
+      "step": 41490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8097093499792022e-05,
+      "loss": 1.4686,
+      "step": 41495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8096650019537593e-05,
+      "loss": 0.8793,
+      "step": 41500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.809620649304694e-05,
+      "loss": 1.1108,
+      "step": 41505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.809576292032259e-05,
+      "loss": 1.4737,
+      "step": 41510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8095319301367084e-05,
+      "loss": 1.2209,
+      "step": 41515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.809487563618295e-05,
+      "loss": 1.3526,
+      "step": 41520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.809443192477272e-05,
+      "loss": 0.9614,
+      "step": 41525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.809398816713893e-05,
+      "loss": 1.0822,
+      "step": 41530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8093544363284116e-05,
+      "loss": 0.9351,
+      "step": 41535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.809310051321081e-05,
+      "loss": 1.2912,
+      "step": 41540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8092656616921546e-05,
+      "loss": 1.5891,
+      "step": 41545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.809221267441886e-05,
+      "loss": 1.0846,
+      "step": 41550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8091768685705288e-05,
+      "loss": 1.2887,
+      "step": 41555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8091324650783365e-05,
+      "loss": 1.1791,
+      "step": 41560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8090880569655626e-05,
+      "loss": 1.585,
+      "step": 41565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8090436442324607e-05,
+      "loss": 1.4012,
+      "step": 41570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8089992268792844e-05,
+      "loss": 1.5196,
+      "step": 41575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8089548049062874e-05,
+      "loss": 1.2774,
+      "step": 41580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8089103783137235e-05,
+      "loss": 1.38,
+      "step": 41585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.808865947101846e-05,
+      "loss": 1.0393,
+      "step": 41590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8088215112709092e-05,
+      "loss": 1.3098,
+      "step": 41595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8087770708211665e-05,
+      "loss": 1.1858,
+      "step": 41600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8087326257528716e-05,
+      "loss": 1.2321,
+      "step": 41605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8086881760662786e-05,
+      "loss": 1.271,
+      "step": 41610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.808643721761641e-05,
+      "loss": 1.0278,
+      "step": 41615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8085992628392128e-05,
+      "loss": 1.2862,
+      "step": 41620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.808554799299248e-05,
+      "loss": 1.3424,
+      "step": 41625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8085103311420004e-05,
+      "loss": 1.3187,
+      "step": 41630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8084658583677236e-05,
+      "loss": 1.1841,
+      "step": 41635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.808421380976672e-05,
+      "loss": 1.418,
+      "step": 41640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8083768989691002e-05,
+      "loss": 1.2266,
+      "step": 41645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.808332412345261e-05,
+      "loss": 1.3502,
+      "step": 41650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.808287921105409e-05,
+      "loss": 1.0459,
+      "step": 41655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8082434252497982e-05,
+      "loss": 1.3088,
+      "step": 41660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8081989247786828e-05,
+      "loss": 1.3023,
+      "step": 41665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8081544196923166e-05,
+      "loss": 1.1001,
+      "step": 41670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.808109909990954e-05,
+      "loss": 1.2379,
+      "step": 41675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8080653956748497e-05,
+      "loss": 1.2224,
+      "step": 41680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8080208767442572e-05,
+      "loss": 1.3414,
+      "step": 41685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8079763531994307e-05,
+      "loss": 1.3213,
+      "step": 41690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8079318250406247e-05,
+      "loss": 1.4048,
+      "step": 41695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8078872922680935e-05,
+      "loss": 1.0936,
+      "step": 41700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.807842754882091e-05,
+      "loss": 1.1279,
+      "step": 41705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8077982128828723e-05,
+      "loss": 1.3002,
+      "step": 41710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8077536662706912e-05,
+      "loss": 1.1937,
+      "step": 41715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.807709115045802e-05,
+      "loss": 1.3375,
+      "step": 41720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8076645592084597e-05,
+      "loss": 1.356,
+      "step": 41725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8076199987589183e-05,
+      "loss": 1.2929,
+      "step": 41730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.807575433697432e-05,
+      "loss": 1.4132,
+      "step": 41735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8075308640242557e-05,
+      "loss": 1.2954,
+      "step": 41740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.807486289739644e-05,
+      "loss": 1.2087,
+      "step": 41745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8074417108438514e-05,
+      "loss": 1.5627,
+      "step": 41750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.807397127337132e-05,
+      "loss": 1.0395,
+      "step": 41755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.807352539219741e-05,
+      "loss": 1.1164,
+      "step": 41760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8073079464919326e-05,
+      "loss": 1.28,
+      "step": 41765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8072633491539617e-05,
+      "loss": 1.3348,
+      "step": 41770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.807218747206083e-05,
+      "loss": 1.1876,
+      "step": 41775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.807174140648551e-05,
+      "loss": 1.2907,
+      "step": 41780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8071295294816205e-05,
+      "loss": 1.431,
+      "step": 41785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.807084913705546e-05,
+      "loss": 1.2133,
+      "step": 41790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.807040293320583e-05,
+      "loss": 1.3124,
+      "step": 41795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8069956683269855e-05,
+      "loss": 1.2064,
+      "step": 41800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8069510387250088e-05,
+      "loss": 1.3266,
+      "step": 41805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.806906404514908e-05,
+      "loss": 1.1553,
+      "step": 41810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.806861765696937e-05,
+      "loss": 1.2535,
+      "step": 41815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.806817122271352e-05,
+      "loss": 1.0329,
+      "step": 41820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.806772474238407e-05,
+      "loss": 1.4288,
+      "step": 41825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.806727821598357e-05,
+      "loss": 1.2406,
+      "step": 41830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8066831643514575e-05,
+      "loss": 1.197,
+      "step": 41835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8066385024979635e-05,
+      "loss": 1.2931,
+      "step": 41840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.806593836038129e-05,
+      "loss": 1.6443,
+      "step": 41845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8065491649722108e-05,
+      "loss": 1.5111,
+      "step": 41850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8065044893004625e-05,
+      "loss": 1.0752,
+      "step": 41855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.80645980902314e-05,
+      "loss": 1.1678,
+      "step": 41860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8064151241404978e-05,
+      "loss": 1.523,
+      "step": 41865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.806370434652792e-05,
+      "loss": 1.15,
+      "step": 41870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8063257405602773e-05,
+      "loss": 1.4288,
+      "step": 41875
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.806281041863209e-05,
+      "loss": 1.1572,
+      "step": 41880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8062363385618418e-05,
+      "loss": 1.0062,
+      "step": 41885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8061916306564318e-05,
+      "loss": 1.4942,
+      "step": 41890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8061469181472338e-05,
+      "loss": 1.6124,
+      "step": 41895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8061022010345036e-05,
+      "loss": 1.1163,
+      "step": 41900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8060574793184962e-05,
+      "loss": 1.155,
+      "step": 41905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.806012752999467e-05,
+      "loss": 1.1826,
+      "step": 41910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8059680220776713e-05,
+      "loss": 1.0939,
+      "step": 41915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.805923286553365e-05,
+      "loss": 1.3792,
+      "step": 41920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8058785464268028e-05,
+      "loss": 1.2475,
+      "step": 41925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.805833801698241e-05,
+      "loss": 1.062,
+      "step": 41930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8057890523679347e-05,
+      "loss": 1.3392,
+      "step": 41935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.80574429843614e-05,
+      "loss": 1.243,
+      "step": 41940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8056995399031116e-05,
+      "loss": 1.2747,
+      "step": 41945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.805654776769105e-05,
+      "loss": 1.3705,
+      "step": 41950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.805610009034377e-05,
+      "loss": 1.0123,
+      "step": 41955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8055652366991823e-05,
+      "loss": 1.3518,
+      "step": 41960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8055204597637764e-05,
+      "loss": 1.2315,
+      "step": 41965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8054756782284158e-05,
+      "loss": 1.2583,
+      "step": 41970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.805430892093356e-05,
+      "loss": 1.1688,
+      "step": 41975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8053861013588526e-05,
+      "loss": 1.1159,
+      "step": 41980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8053413060251613e-05,
+      "loss": 1.3638,
+      "step": 41985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.805296506092538e-05,
+      "loss": 1.5526,
+      "step": 41990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8052517015612383e-05,
+      "loss": 1.0856,
+      "step": 41995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8052068924315185e-05,
+      "loss": 1.6693,
+      "step": 42000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8051620787036343e-05,
+      "loss": 1.1705,
+      "step": 42005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.805117260377841e-05,
+      "loss": 1.2052,
+      "step": 42010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8050724374543958e-05,
+      "loss": 1.1336,
+      "step": 42015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8050276099335538e-05,
+      "loss": 1.1428,
+      "step": 42020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8049827778155713e-05,
+      "loss": 1.333,
+      "step": 42025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.804937941100704e-05,
+      "loss": 1.2364,
+      "step": 42030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.804893099789208e-05,
+      "loss": 1.4052,
+      "step": 42035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8048482538813395e-05,
+      "loss": 1.2801,
+      "step": 42040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.804803403377355e-05,
+      "loss": 1.6844,
+      "step": 42045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8047585482775097e-05,
+      "loss": 1.777,
+      "step": 42050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8047136885820606e-05,
+      "loss": 1.1492,
+      "step": 42055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8046688242912632e-05,
+      "loss": 1.0596,
+      "step": 42060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.804623955405374e-05,
+      "loss": 1.1831,
+      "step": 42065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8045790819246498e-05,
+      "loss": 0.996,
+      "step": 42070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8045342038493456e-05,
+      "loss": 1.1618,
+      "step": 42075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8044893211797188e-05,
+      "loss": 1.3146,
+      "step": 42080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8044444339160252e-05,
+      "loss": 1.3102,
+      "step": 42085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.804399542058521e-05,
+      "loss": 1.1325,
+      "step": 42090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.804354645607463e-05,
+      "loss": 1.2926,
+      "step": 42095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8043097445631073e-05,
+      "loss": 1.057,
+      "step": 42100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.80426483892571e-05,
+      "loss": 1.6987,
+      "step": 42105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8042199286955286e-05,
+      "loss": 1.1229,
+      "step": 42110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8041750138728185e-05,
+      "loss": 1.2454,
+      "step": 42115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8041300944578366e-05,
+      "loss": 1.3855,
+      "step": 42120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.804085170450839e-05,
+      "loss": 1.6193,
+      "step": 42125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8040402418520832e-05,
+      "loss": 1.1792,
+      "step": 42130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8039953086618248e-05,
+      "loss": 1.4872,
+      "step": 42135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.803950370880321e-05,
+      "loss": 1.2421,
+      "step": 42140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.803905428507828e-05,
+      "loss": 1.4762,
+      "step": 42145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8038604815446023e-05,
+      "loss": 1.4742,
+      "step": 42150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8038155299909013e-05,
+      "loss": 1.5595,
+      "step": 42155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8037705738469814e-05,
+      "loss": 1.1672,
+      "step": 42160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.803725613113099e-05,
+      "loss": 1.1885,
+      "step": 42165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.803680647789511e-05,
+      "loss": 1.0925,
+      "step": 42170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8036356778764743e-05,
+      "loss": 1.0924,
+      "step": 42175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8035907033742456e-05,
+      "loss": 1.256,
+      "step": 42180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.803545724283082e-05,
+      "loss": 1.5008,
+      "step": 42185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8035007406032397e-05,
+      "loss": 1.5843,
+      "step": 42190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8034557523349763e-05,
+      "loss": 1.1851,
+      "step": 42195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8034107594785485e-05,
+      "loss": 1.4225,
+      "step": 42200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.803365762034213e-05,
+      "loss": 1.2378,
+      "step": 42205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8033207600022268e-05,
+      "loss": 1.2282,
+      "step": 42210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8032757533828475e-05,
+      "loss": 1.1621,
+      "step": 42215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8032307421763308e-05,
+      "loss": 1.2684,
+      "step": 42220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8031857263829352e-05,
+      "loss": 1.1964,
+      "step": 42225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8031407060029168e-05,
+      "loss": 0.9978,
+      "step": 42230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8030956810365332e-05,
+      "loss": 1.2437,
+      "step": 42235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.803050651484041e-05,
+      "loss": 1.4622,
+      "step": 42240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.803005617345698e-05,
+      "loss": 1.2036,
+      "step": 42245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.802960578621761e-05,
+      "loss": 1.3648,
+      "step": 42250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8029155353124868e-05,
+      "loss": 1.2183,
+      "step": 42255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8028704874181336e-05,
+      "loss": 1.3224,
+      "step": 42260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8028254349389577e-05,
+      "loss": 1.4117,
+      "step": 42265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.802780377875217e-05,
+      "loss": 1.1846,
+      "step": 42270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8027353162271683e-05,
+      "loss": 1.0404,
+      "step": 42275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8026902499950695e-05,
+      "loss": 1.1886,
+      "step": 42280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8026451791791776e-05,
+      "loss": 1.0775,
+      "step": 42285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.80260010377975e-05,
+      "loss": 1.4222,
+      "step": 42290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8025550237970436e-05,
+      "loss": 1.1228,
+      "step": 42295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.802509939231317e-05,
+      "loss": 1.2341,
+      "step": 42300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.802464850082827e-05,
+      "loss": 1.0376,
+      "step": 42305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8024197563518305e-05,
+      "loss": 1.4878,
+      "step": 42310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8023746580385865e-05,
+      "loss": 1.6031,
+      "step": 42315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.802329555143351e-05,
+      "loss": 1.4175,
+      "step": 42320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.802284447666382e-05,
+      "loss": 1.2543,
+      "step": 42325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.802239335607938e-05,
+      "loss": 1.3617,
+      "step": 42330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8021942189682755e-05,
+      "loss": 1.1519,
+      "step": 42335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.802149097747653e-05,
+      "loss": 1.3712,
+      "step": 42340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.802103971946327e-05,
+      "loss": 1.3364,
+      "step": 42345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.802058841564556e-05,
+      "loss": 1.1454,
+      "step": 42350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.802013706602598e-05,
+      "loss": 1.3127,
+      "step": 42355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8019685670607103e-05,
+      "loss": 1.6236,
+      "step": 42360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8019234229391506e-05,
+      "loss": 1.0833,
+      "step": 42365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.801878274238177e-05,
+      "loss": 1.2689,
+      "step": 42370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.801833120958047e-05,
+      "loss": 1.0709,
+      "step": 42375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.801787963099019e-05,
+      "loss": 1.2007,
+      "step": 42380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.80174280066135e-05,
+      "loss": 1.2141,
+      "step": 42385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8016976336452984e-05,
+      "loss": 1.0093,
+      "step": 42390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8016524620511223e-05,
+      "loss": 1.0664,
+      "step": 42395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8016072858790797e-05,
+      "loss": 1.1985,
+      "step": 42400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.801562105129428e-05,
+      "loss": 1.4754,
+      "step": 42405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8015169198024257e-05,
+      "loss": 1.0382,
+      "step": 42410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8014717298983307e-05,
+      "loss": 1.1986,
+      "step": 42415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.801426535417401e-05,
+      "loss": 1.0397,
+      "step": 42420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8013813363598944e-05,
+      "loss": 1.2326,
+      "step": 42425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.80133613272607e-05,
+      "loss": 1.3174,
+      "step": 42430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8012909245161848e-05,
+      "loss": 1.4642,
+      "step": 42435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8012457117304976e-05,
+      "loss": 1.2231,
+      "step": 42440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8012004943692665e-05,
+      "loss": 1.6013,
+      "step": 42445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8011552724327493e-05,
+      "loss": 1.1371,
+      "step": 42450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8011100459212052e-05,
+      "loss": 1.5445,
+      "step": 42455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8010648148348915e-05,
+      "loss": 1.2074,
+      "step": 42460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8010195791740668e-05,
+      "loss": 1.3573,
+      "step": 42465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8009743389389895e-05,
+      "loss": 1.2952,
+      "step": 42470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8009290941299183e-05,
+      "loss": 1.1427,
+      "step": 42475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8008838447471108e-05,
+      "loss": 1.1006,
+      "step": 42480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8008385907908257e-05,
+      "loss": 0.8603,
+      "step": 42485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8007933322613218e-05,
+      "loss": 1.4201,
+      "step": 42490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.800748069158857e-05,
+      "loss": 1.7968,
+      "step": 42495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8007028014836906e-05,
+      "loss": 1.2484,
+      "step": 42500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.80065752923608e-05,
+      "loss": 1.348,
+      "step": 42505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.800612252416284e-05,
+      "loss": 1.3686,
+      "step": 42510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.800566971024562e-05,
+      "loss": 1.1525,
+      "step": 42515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.800521685061172e-05,
+      "loss": 1.2587,
+      "step": 42520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8004763945263724e-05,
+      "loss": 1.5963,
+      "step": 42525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.800431099420422e-05,
+      "loss": 1.4022,
+      "step": 42530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8003857997435798e-05,
+      "loss": 1.4355,
+      "step": 42535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.800340495496104e-05,
+      "loss": 1.3729,
+      "step": 42540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8002951866782533e-05,
+      "loss": 1.2139,
+      "step": 42545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8002498732902868e-05,
+      "loss": 1.3373,
+      "step": 42550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8002045553324633e-05,
+      "loss": 0.9801,
+      "step": 42555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.800159232805041e-05,
+      "loss": 1.4597,
+      "step": 42560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8001139057082797e-05,
+      "loss": 1.4504,
+      "step": 42565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8000685740424372e-05,
+      "loss": 1.5235,
+      "step": 42570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8000232378077732e-05,
+      "loss": 1.3166,
+      "step": 42575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.799977897004546e-05,
+      "loss": 1.1695,
+      "step": 42580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7999325516330147e-05,
+      "loss": 1.0721,
+      "step": 42585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7998872016934382e-05,
+      "loss": 1.1978,
+      "step": 42590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7998418471860753e-05,
+      "loss": 1.3277,
+      "step": 42595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.799796488111186e-05,
+      "loss": 1.2868,
+      "step": 42600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.7997511244690283e-05,
+      "loss": 1.4073,
+      "step": 42605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.799705756259861e-05,
+      "loss": 1.602,
+      "step": 42610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7996603834839442e-05,
+      "loss": 1.2376,
+      "step": 42615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7996150061415366e-05,
+      "loss": 1.1209,
+      "step": 42620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.799569624232897e-05,
+      "loss": 1.3409,
+      "step": 42625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7995242377582848e-05,
+      "loss": 1.2636,
+      "step": 42630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7994788467179592e-05,
+      "loss": 1.3122,
+      "step": 42635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7994334511121796e-05,
+      "loss": 0.9617,
+      "step": 42640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7993880509412048e-05,
+      "loss": 1.2904,
+      "step": 42645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7993426462052946e-05,
+      "loss": 1.1033,
+      "step": 42650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7992972369047073e-05,
+      "loss": 1.1701,
+      "step": 42655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7992518230397034e-05,
+      "loss": 1.2667,
+      "step": 42660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7992064046105414e-05,
+      "loss": 1.3089,
+      "step": 42665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7991609816174812e-05,
+      "loss": 1.5568,
+      "step": 42670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.799115554060782e-05,
+      "loss": 1.0466,
+      "step": 42675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.799070121940703e-05,
+      "loss": 1.2031,
+      "step": 42680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7990246852575038e-05,
+      "loss": 1.2122,
+      "step": 42685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7989792440114443e-05,
+      "loss": 1.5251,
+      "step": 42690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.798933798202783e-05,
+      "loss": 1.0291,
+      "step": 42695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7988883478317808e-05,
+      "loss": 1.3439,
+      "step": 42700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7988428928986955e-05,
+      "loss": 1.163,
+      "step": 42705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.798797433403788e-05,
+      "loss": 1.1051,
+      "step": 42710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7987519693473178e-05,
+      "loss": 1.4592,
+      "step": 42715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7987065007295437e-05,
+      "loss": 1.4761,
+      "step": 42720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7986610275507267e-05,
+      "loss": 1.1481,
+      "step": 42725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.798615549811125e-05,
+      "loss": 1.4799,
+      "step": 42730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7985700675109992e-05,
+      "loss": 1.0207,
+      "step": 42735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7985245806506085e-05,
+      "loss": 0.9928,
+      "step": 42740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7984790892302136e-05,
+      "loss": 1.2506,
+      "step": 42745
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7984335932500732e-05,
+      "loss": 1.2434,
+      "step": 42750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7983880927104477e-05,
+      "loss": 1.5187,
+      "step": 42755
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7983425876115965e-05,
+      "loss": 1.1901,
+      "step": 42760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79829707795378e-05,
+      "loss": 1.4642,
+      "step": 42765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7982515637372578e-05,
+      "loss": 1.3359,
+      "step": 42770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7982060449622895e-05,
+      "loss": 1.1546,
+      "step": 42775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.798160521629136e-05,
+      "loss": 1.2308,
+      "step": 42780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7981149937380558e-05,
+      "loss": 1.1027,
+      "step": 42785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7980694612893104e-05,
+      "loss": 1.0939,
+      "step": 42790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.798023924283159e-05,
+      "loss": 1.1331,
+      "step": 42795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7979783827198617e-05,
+      "loss": 1.187,
+      "step": 42800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7979328365996783e-05,
+      "loss": 1.3203,
+      "step": 42805
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79788728592287e-05,
+      "loss": 1.3037,
+      "step": 42810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7978417306896957e-05,
+      "loss": 1.7088,
+      "step": 42815
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7977961709004158e-05,
+      "loss": 1.0536,
+      "step": 42820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7977506065552908e-05,
+      "loss": 1.275,
+      "step": 42825
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.797705037654581e-05,
+      "loss": 1.3452,
+      "step": 42830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.797659464198546e-05,
+      "loss": 1.3291,
+      "step": 42835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.797613886187447e-05,
+      "loss": 1.2236,
+      "step": 42840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.797568303621543e-05,
+      "loss": 1.4939,
+      "step": 42845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7975227165010957e-05,
+      "loss": 1.0292,
+      "step": 42850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7974771248263643e-05,
+      "loss": 1.3192,
+      "step": 42855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79743152859761e-05,
+      "loss": 1.7013,
+      "step": 42860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.797385927815092e-05,
+      "loss": 1.043,
+      "step": 42865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.797340322479072e-05,
+      "loss": 1.45,
+      "step": 42870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79729471258981e-05,
+      "loss": 1.0109,
+      "step": 42875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.797249098147566e-05,
+      "loss": 1.1881,
+      "step": 42880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7972034791526014e-05,
+      "loss": 1.0912,
+      "step": 42885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7971578556051755e-05,
+      "loss": 1.1264,
+      "step": 42890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7971122275055503e-05,
+      "loss": 1.5887,
+      "step": 42895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.797066594853985e-05,
+      "loss": 1.3601,
+      "step": 42900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7970209576507406e-05,
+      "loss": 1.37,
+      "step": 42905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796975315896078e-05,
+      "loss": 1.4557,
+      "step": 42910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796929669590258e-05,
+      "loss": 1.2403,
+      "step": 42915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7968840187335406e-05,
+      "loss": 1.132,
+      "step": 42920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796838363326187e-05,
+      "loss": 1.1794,
+      "step": 42925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7967927033684578e-05,
+      "loss": 1.3692,
+      "step": 42930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7967470388606138e-05,
+      "loss": 1.2241,
+      "step": 42935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7967013698029154e-05,
+      "loss": 1.0585,
+      "step": 42940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7966556961956238e-05,
+      "loss": 1.3317,
+      "step": 42945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7966100180389995e-05,
+      "loss": 1.326,
+      "step": 42950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796564335333304e-05,
+      "loss": 1.0364,
+      "step": 42955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7965186480787973e-05,
+      "loss": 1.5107,
+      "step": 42960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7964729562757408e-05,
+      "loss": 1.253,
+      "step": 42965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7964272599243955e-05,
+      "loss": 1.2012,
+      "step": 42970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796381559025022e-05,
+      "loss": 1.0362,
+      "step": 42975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7963358535778815e-05,
+      "loss": 1.3246,
+      "step": 42980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796290143583235e-05,
+      "loss": 1.0561,
+      "step": 42985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7962444290413435e-05,
+      "loss": 1.2895,
+      "step": 42990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796198709952468e-05,
+      "loss": 1.0836,
+      "step": 42995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7961529863168698e-05,
+      "loss": 1.2745,
+      "step": 43000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7961072581348094e-05,
+      "loss": 1.2993,
+      "step": 43005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.796061525406549e-05,
+      "loss": 1.4118,
+      "step": 43010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7960157881323484e-05,
+      "loss": 1.281,
+      "step": 43015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79597004631247e-05,
+      "loss": 1.0828,
+      "step": 43020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7959242999471743e-05,
+      "loss": 1.4531,
+      "step": 43025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7958785490367228e-05,
+      "loss": 1.2485,
+      "step": 43030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.795832793581377e-05,
+      "loss": 1.1334,
+      "step": 43035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7957870335813975e-05,
+      "loss": 1.2969,
+      "step": 43040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7957412690370465e-05,
+      "loss": 1.7718,
+      "step": 43045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7956954999485846e-05,
+      "loss": 1.0467,
+      "step": 43050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7956497263162737e-05,
+      "loss": 1.0449,
+      "step": 43055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.795603948140374e-05,
+      "loss": 1.4352,
+      "step": 43060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7955581654211487e-05,
+      "loss": 1.1292,
+      "step": 43065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7955123781588583e-05,
+      "loss": 1.0672,
+      "step": 43070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7954665863537642e-05,
+      "loss": 0.98,
+      "step": 43075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.795420790006128e-05,
+      "loss": 1.1912,
+      "step": 43080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7953749891162115e-05,
+      "loss": 1.3131,
+      "step": 43085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7953291836842756e-05,
+      "loss": 1.117,
+      "step": 43090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7952833737105827e-05,
+      "loss": 1.0205,
+      "step": 43095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7952375591953938e-05,
+      "loss": 1.6441,
+      "step": 43100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7951917401389706e-05,
+      "loss": 1.4803,
+      "step": 43105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.795145916541575e-05,
+      "loss": 1.4077,
+      "step": 43110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.795100088403469e-05,
+      "loss": 1.0851,
+      "step": 43115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.795054255724913e-05,
+      "loss": 0.9402,
+      "step": 43120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79500841850617e-05,
+      "loss": 1.3241,
+      "step": 43125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7949625767475015e-05,
+      "loss": 1.18,
+      "step": 43130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.794916730449169e-05,
+      "loss": 1.4997,
+      "step": 43135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7948708796114343e-05,
+      "loss": 1.2725,
+      "step": 43140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.794825024234559e-05,
+      "loss": 1.1594,
+      "step": 43145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7947791643188058e-05,
+      "loss": 1.2058,
+      "step": 43150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7947332998644365e-05,
+      "loss": 1.2532,
+      "step": 43155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.794687430871712e-05,
+      "loss": 1.3506,
+      "step": 43160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7946415573408948e-05,
+      "loss": 1.0859,
+      "step": 43165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7945956792722472e-05,
+      "loss": 1.5235,
+      "step": 43170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.794549796666031e-05,
+      "loss": 0.9583,
+      "step": 43175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.794503909522508e-05,
+      "loss": 1.2197,
+      "step": 43180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79445801784194e-05,
+      "loss": 1.2226,
+      "step": 43185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7944121216245896e-05,
+      "loss": 0.9951,
+      "step": 43190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.794366220870719e-05,
+      "loss": 1.0354,
+      "step": 43195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7943203155805896e-05,
+      "loss": 1.149,
+      "step": 43200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.794274405754464e-05,
+      "loss": 1.1014,
+      "step": 43205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7942284913926045e-05,
+      "loss": 1.2592,
+      "step": 43210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.794182572495273e-05,
+      "loss": 1.2138,
+      "step": 43215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7941366490627322e-05,
+      "loss": 1.1458,
+      "step": 43220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.794090721095244e-05,
+      "loss": 1.2606,
+      "step": 43225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7940447885930706e-05,
+      "loss": 1.5147,
+      "step": 43230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793998851556474e-05,
+      "loss": 1.3491,
+      "step": 43235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793952909985717e-05,
+      "loss": 1.1695,
+      "step": 43240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7939069638810623e-05,
+      "loss": 1.4831,
+      "step": 43245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7938610132427715e-05,
+      "loss": 1.0626,
+      "step": 43250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7938150580711074e-05,
+      "loss": 1.509,
+      "step": 43255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7937690983663324e-05,
+      "loss": 1.1156,
+      "step": 43260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7937231341287087e-05,
+      "loss": 1.5292,
+      "step": 43265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7936771653584993e-05,
+      "loss": 0.9323,
+      "step": 43270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7936311920559662e-05,
+      "loss": 1.3254,
+      "step": 43275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7935852142213722e-05,
+      "loss": 1.5461,
+      "step": 43280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79353923185498e-05,
+      "loss": 1.2995,
+      "step": 43285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7934932449570517e-05,
+      "loss": 1.6774,
+      "step": 43290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7934472535278503e-05,
+      "loss": 1.0126,
+      "step": 43295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7934012575676382e-05,
+      "loss": 1.3786,
+      "step": 43300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7933552570766784e-05,
+      "loss": 1.198,
+      "step": 43305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7933092520552332e-05,
+      "loss": 1.2657,
+      "step": 43310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7932632425035654e-05,
+      "loss": 1.2723,
+      "step": 43315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7932172284219382e-05,
+      "loss": 1.3227,
+      "step": 43320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7931712098106137e-05,
+      "loss": 1.0672,
+      "step": 43325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793125186669855e-05,
+      "loss": 1.1003,
+      "step": 43330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.793079158999925e-05,
+      "loss": 1.4795,
+      "step": 43335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7930331268010862e-05,
+      "loss": 1.235,
+      "step": 43340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7929870900736022e-05,
+      "loss": 1.6073,
+      "step": 43345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792941048817735e-05,
+      "loss": 1.302,
+      "step": 43350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7928950030337478e-05,
+      "loss": 1.1535,
+      "step": 43355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792848952721904e-05,
+      "loss": 1.2483,
+      "step": 43360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7928028978824662e-05,
+      "loss": 0.9994,
+      "step": 43365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7927568385156974e-05,
+      "loss": 1.1938,
+      "step": 43370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7927107746218607e-05,
+      "loss": 1.1521,
+      "step": 43375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792664706201219e-05,
+      "loss": 1.2368,
+      "step": 43380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7926186332540357e-05,
+      "loss": 1.0422,
+      "step": 43385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7925725557805736e-05,
+      "loss": 1.2043,
+      "step": 43390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792526473781096e-05,
+      "loss": 1.313,
+      "step": 43395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7924803872558658e-05,
+      "loss": 1.6021,
+      "step": 43400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7924342962051462e-05,
+      "loss": 1.3839,
+      "step": 43405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792388200629201e-05,
+      "loss": 1.5917,
+      "step": 43410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7923421005282927e-05,
+      "loss": 1.6386,
+      "step": 43415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7922959959026846e-05,
+      "loss": 1.5156,
+      "step": 43420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792249886752641e-05,
+      "loss": 1.2206,
+      "step": 43425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7922037730784235e-05,
+      "loss": 1.1225,
+      "step": 43430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792157654880297e-05,
+      "loss": 1.0129,
+      "step": 43435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7921115321585238e-05,
+      "loss": 1.0144,
+      "step": 43440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.792065404913368e-05,
+      "loss": 1.1287,
+      "step": 43445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7920192731450927e-05,
+      "loss": 1.299,
+      "step": 43450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7919731368539616e-05,
+      "loss": 1.128,
+      "step": 43455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7919269960402376e-05,
+      "loss": 1.1204,
+      "step": 43460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7918808507041844e-05,
+      "loss": 1.4074,
+      "step": 43465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791834700846066e-05,
+      "loss": 1.4107,
+      "step": 43470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7917885464661454e-05,
+      "loss": 1.5713,
+      "step": 43475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791742387564686e-05,
+      "loss": 1.2889,
+      "step": 43480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7916962241419522e-05,
+      "loss": 1.2982,
+      "step": 43485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7916500561982068e-05,
+      "loss": 1.1819,
+      "step": 43490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791603883733714e-05,
+      "loss": 1.0499,
+      "step": 43495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7915577067487373e-05,
+      "loss": 1.3399,
+      "step": 43500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7915115252435403e-05,
+      "loss": 1.3557,
+      "step": 43505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7914653392183866e-05,
+      "loss": 0.8939,
+      "step": 43510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7914191486735405e-05,
+      "loss": 1.1396,
+      "step": 43515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791372953609265e-05,
+      "loss": 1.4107,
+      "step": 43520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7913267540258247e-05,
+      "loss": 1.402,
+      "step": 43525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7912805499234825e-05,
+      "loss": 1.0476,
+      "step": 43530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791234341302503e-05,
+      "loss": 1.4533,
+      "step": 43535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79118812816315e-05,
+      "loss": 0.9871,
+      "step": 43540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791141910505687e-05,
+      "loss": 1.4642,
+      "step": 43545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7910956883303784e-05,
+      "loss": 1.1053,
+      "step": 43550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7910494616374882e-05,
+      "loss": 1.0257,
+      "step": 43555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7910032304272795e-05,
+      "loss": 1.2855,
+      "step": 43560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7909569947000173e-05,
+      "loss": 1.2561,
+      "step": 43565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.790910754455965e-05,
+      "loss": 1.1628,
+      "step": 43570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.790864509695387e-05,
+      "loss": 1.257,
+      "step": 43575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.790818260418547e-05,
+      "loss": 1.2708,
+      "step": 43580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79077200662571e-05,
+      "loss": 1.0366,
+      "step": 43585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.790725748317139e-05,
+      "loss": 1.1142,
+      "step": 43590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.790679485493099e-05,
+      "loss": 1.1561,
+      "step": 43595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7906332181538536e-05,
+      "loss": 1.5639,
+      "step": 43600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7905869462996675e-05,
+      "loss": 1.1969,
+      "step": 43605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.790540669930805e-05,
+      "loss": 1.3875,
+      "step": 43610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.790494389047529e-05,
+      "loss": 1.3005,
+      "step": 43615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.790448103650106e-05,
+      "loss": 1.2307,
+      "step": 43620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7904018137387988e-05,
+      "loss": 1.3883,
+      "step": 43625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.790355519313872e-05,
+      "loss": 1.3272,
+      "step": 43630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.79030922037559e-05,
+      "loss": 1.3453,
+      "step": 43635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7902629169242177e-05,
+      "loss": 1.3183,
+      "step": 43640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7902166089600192e-05,
+      "loss": 1.1629,
+      "step": 43645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7901702964832584e-05,
+      "loss": 1.4747,
+      "step": 43650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7901239794942e-05,
+      "loss": 1.4983,
+      "step": 43655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7900776579931093e-05,
+      "loss": 1.3445,
+      "step": 43660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7900313319802498e-05,
+      "loss": 1.3208,
+      "step": 43665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7899850014558864e-05,
+      "loss": 1.3203,
+      "step": 43670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7899386664202842e-05,
+      "loss": 1.0674,
+      "step": 43675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.789892326873707e-05,
+      "loss": 1.3819,
+      "step": 43680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.78984598281642e-05,
+      "loss": 1.5008,
+      "step": 43685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7897996342486875e-05,
+      "loss": 1.3183,
+      "step": 43690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7897532811707743e-05,
+      "loss": 1.2324,
+      "step": 43695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7897069235829453e-05,
+      "loss": 1.2973,
+      "step": 43700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7896605614854645e-05,
+      "loss": 1.2197,
+      "step": 43705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7896141948785977e-05,
+      "loss": 1.8911,
+      "step": 43710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.789567823762609e-05,
+      "loss": 1.272,
+      "step": 43715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7895214481377632e-05,
+      "loss": 1.285,
+      "step": 43720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7894750680043255e-05,
+      "loss": 1.2199,
+      "step": 43725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7894286833625602e-05,
+      "loss": 1.1304,
+      "step": 43730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.789382294212733e-05,
+      "loss": 0.8534,
+      "step": 43735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.789335900555108e-05,
+      "loss": 1.3358,
+      "step": 43740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7892895023899505e-05,
+      "loss": 1.2998,
+      "step": 43745
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7892430997175253e-05,
+      "loss": 1.3913,
+      "step": 43750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7891966925380975e-05,
+      "loss": 1.2013,
+      "step": 43755
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7891502808519327e-05,
+      "loss": 1.0472,
+      "step": 43760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.789103864659295e-05,
+      "loss": 1.3417,
+      "step": 43765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7890574439604496e-05,
+      "loss": 1.2686,
+      "step": 43770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.789011018755662e-05,
+      "loss": 1.0672,
+      "step": 43775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7889645890451967e-05,
+      "loss": 1.2672,
+      "step": 43780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7889181548293198e-05,
+      "loss": 1.4048,
+      "step": 43785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.788871716108296e-05,
+      "loss": 1.3544,
+      "step": 43790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7888252728823897e-05,
+      "loss": 1.2233,
+      "step": 43795
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7887788251518674e-05,
+      "loss": 1.5742,
+      "step": 43800
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7887323729169936e-05,
+      "loss": 1.1516,
+      "step": 43805
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7886859161780336e-05,
+      "loss": 1.1529,
+      "step": 43810
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.788639454935253e-05,
+      "loss": 1.2919,
+      "step": 43815
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.788592989188917e-05,
+      "loss": 1.0832,
+      "step": 43820
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.788546518939291e-05,
+      "loss": 1.4464,
+      "step": 43825
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7885000441866398e-05,
+      "loss": 1.5242,
+      "step": 43830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7884535649312293e-05,
+      "loss": 1.3619,
+      "step": 43835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.788407081173325e-05,
+      "loss": 1.471,
+      "step": 43840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7883605929131924e-05,
+      "loss": 1.2732,
+      "step": 43845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7883141001510965e-05,
+      "loss": 1.2693,
+      "step": 43850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7882676028873037e-05,
+      "loss": 1.408,
+      "step": 43855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7882211011220784e-05,
+      "loss": 1.2256,
+      "step": 43860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7881745948556865e-05,
+      "loss": 1.3367,
+      "step": 43865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7881280840883942e-05,
+      "loss": 1.2021,
+      "step": 43870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7880815688204662e-05,
+      "loss": 1.4749,
+      "step": 43875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7880350490521688e-05,
+      "loss": 1.1321,
+      "step": 43880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7879885247837677e-05,
+      "loss": 1.4774,
+      "step": 43885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.787941996015528e-05,
+      "loss": 1.1263,
+      "step": 43890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7878954627477156e-05,
+      "loss": 1.2713,
+      "step": 43895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7878489249805966e-05,
+      "loss": 1.3143,
+      "step": 43900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7878023827144363e-05,
+      "loss": 1.2629,
+      "step": 43905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7877558359495008e-05,
+      "loss": 1.1831,
+      "step": 43910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7877092846860556e-05,
+      "loss": 1.1209,
+      "step": 43915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7876627289243667e-05,
+      "loss": 1.2305,
+      "step": 43920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7876161686647e-05,
+      "loss": 1.5909,
+      "step": 43925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7875696039073214e-05,
+      "loss": 1.5559,
+      "step": 43930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.787523034652497e-05,
+      "loss": 1.0375,
+      "step": 43935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7874764609004924e-05,
+      "loss": 1.2339,
+      "step": 43940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7874298826515736e-05,
+      "loss": 1.5612,
+      "step": 43945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7873832999060066e-05,
+      "loss": 1.0921,
+      "step": 43950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7873367126640574e-05,
+      "loss": 1.1092,
+      "step": 43955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7872901209259922e-05,
+      "loss": 1.0359,
+      "step": 43960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.787243524692077e-05,
+      "loss": 1.0991,
+      "step": 43965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.787196923962578e-05,
+      "loss": 1.3044,
+      "step": 43970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7871503187377608e-05,
+      "loss": 1.2446,
+      "step": 43975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7871037090178922e-05,
+      "loss": 1.253,
+      "step": 43980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7870570948032376e-05,
+      "loss": 1.2599,
+      "step": 43985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.787010476094064e-05,
+      "loss": 1.5354,
+      "step": 43990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7869638528906376e-05,
+      "loss": 1.2185,
+      "step": 43995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7869172251932242e-05,
+      "loss": 1.0607,
+      "step": 44000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7868705930020896e-05,
+      "loss": 1.0063,
+      "step": 44005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7868239563175015e-05,
+      "loss": 0.8102,
+      "step": 44010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.786777315139725e-05,
+      "loss": 1.5273,
+      "step": 44015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.786730669469027e-05,
+      "loss": 1.1695,
+      "step": 44020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7866840193056734e-05,
+      "loss": 1.1691,
+      "step": 44025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.786637364649931e-05,
+      "loss": 1.2616,
+      "step": 44030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7865907055020665e-05,
+      "loss": 1.2329,
+      "step": 44035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.786544041862346e-05,
+      "loss": 1.3583,
+      "step": 44040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7864973737310358e-05,
+      "loss": 1.4087,
+      "step": 44045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7864507011084023e-05,
+      "loss": 1.1065,
+      "step": 44050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.786404023994713e-05,
+      "loss": 1.5013,
+      "step": 44055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7863573423902333e-05,
+      "loss": 1.2646,
+      "step": 44060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.78631065629523e-05,
+      "loss": 1.5926,
+      "step": 44065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7862639657099704e-05,
+      "loss": 1.1027,
+      "step": 44070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7862172706347205e-05,
+      "loss": 1.2091,
+      "step": 44075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.786170571069747e-05,
+      "loss": 1.1654,
+      "step": 44080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.786123867015317e-05,
+      "loss": 1.3104,
+      "step": 44085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7860771584716965e-05,
+      "loss": 1.163,
+      "step": 44090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.786030445439153e-05,
+      "loss": 1.4628,
+      "step": 44095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7859837279179528e-05,
+      "loss": 1.136,
+      "step": 44100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7859370059083626e-05,
+      "loss": 1.1413,
+      "step": 44105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.785890279410649e-05,
+      "loss": 1.2251,
+      "step": 44110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.78584354842508e-05,
+      "loss": 1.3204,
+      "step": 44115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.785796812951921e-05,
+      "loss": 1.3838,
+      "step": 44120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.78575007299144e-05,
+      "loss": 1.0895,
+      "step": 44125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7857033285439036e-05,
+      "loss": 1.4254,
+      "step": 44130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7856565796095783e-05,
+      "loss": 1.3426,
+      "step": 44135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7856098261887315e-05,
+      "loss": 1.0944,
+      "step": 44140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7855630682816296e-05,
+      "loss": 1.1302,
+      "step": 44145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.785516305888541e-05,
+      "loss": 1.4318,
+      "step": 44150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7854695390097307e-05,
+      "loss": 1.3605,
+      "step": 44155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7854227676454677e-05,
+      "loss": 1.0569,
+      "step": 44160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.785375991796018e-05,
+      "loss": 1.2517,
+      "step": 44165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7853292114616486e-05,
+      "loss": 1.3138,
+      "step": 44170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7852824266426277e-05,
+      "loss": 1.2005,
+      "step": 44175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7852356373392213e-05,
+      "loss": 1.3494,
+      "step": 44180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.785188843551697e-05,
+      "loss": 0.9952,
+      "step": 44185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7851420452803222e-05,
+      "loss": 1.1932,
+      "step": 44190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.785095242525364e-05,
+      "loss": 1.0175,
+      "step": 44195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.78504843528709e-05,
+      "loss": 1.3168,
+      "step": 44200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.785001623565767e-05,
+      "loss": 1.1246,
+      "step": 44205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7849548073616623e-05,
+      "loss": 1.0137,
+      "step": 44210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7849079866750442e-05,
+      "loss": 0.9404,
+      "step": 44215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7848611615061788e-05,
+      "loss": 1.2441,
+      "step": 44220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.784814331855334e-05,
+      "loss": 1.3079,
+      "step": 44225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.784767497722777e-05,
+      "loss": 1.2944,
+      "step": 44230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7847206591087757e-05,
+      "loss": 1.5572,
+      "step": 44235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7846738160135973e-05,
+      "loss": 1.2865,
+      "step": 44240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7846269684375097e-05,
+      "loss": 1.1641,
+      "step": 44245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.78458011638078e-05,
+      "loss": 1.1876,
+      "step": 44250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7845332598436757e-05,
+      "loss": 1.5259,
+      "step": 44255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7844863988264647e-05,
+      "loss": 1.328,
+      "step": 44260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7844395333294144e-05,
+      "loss": 1.1817,
+      "step": 44265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7843926633527923e-05,
+      "loss": 1.1959,
+      "step": 44270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7843457888968663e-05,
+      "loss": 1.2818,
+      "step": 44275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.784298909961904e-05,
+      "loss": 1.2307,
+      "step": 44280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.784252026548173e-05,
+      "loss": 1.3271,
+      "step": 44285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7842051386559412e-05,
+      "loss": 1.1874,
+      "step": 44290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7841582462854764e-05,
+      "loss": 1.0025,
+      "step": 44295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7841113494370458e-05,
+      "loss": 1.0005,
+      "step": 44300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.784064448110918e-05,
+      "loss": 1.1373,
+      "step": 44305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7840175423073605e-05,
+      "loss": 1.2337,
+      "step": 44310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.783970632026641e-05,
+      "loss": 1.3013,
+      "step": 44315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7839237172690277e-05,
+      "loss": 1.0902,
+      "step": 44320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7838767980347882e-05,
+      "loss": 1.2685,
+      "step": 44325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7838298743241907e-05,
+      "loss": 1.105,
+      "step": 44330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7837829461375028e-05,
+      "loss": 1.6271,
+      "step": 44335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7837360134749932e-05,
+      "loss": 1.3697,
+      "step": 44340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.783689076336929e-05,
+      "loss": 1.2875,
+      "step": 44345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7836421347235785e-05,
+      "loss": 1.0577,
+      "step": 44350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7835951886352104e-05,
+      "loss": 1.1574,
+      "step": 44355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.783548238072092e-05,
+      "loss": 1.0953,
+      "step": 44360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7835012830344918e-05,
+      "loss": 1.1984,
+      "step": 44365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.783454323522678e-05,
+      "loss": 1.0975,
+      "step": 44370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7834073595369184e-05,
+      "loss": 1.2931,
+      "step": 44375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7833603910774816e-05,
+      "loss": 1.4819,
+      "step": 44380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7833134181446356e-05,
+      "loss": 1.7084,
+      "step": 44385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7832664407386487e-05,
+      "loss": 1.1594,
+      "step": 44390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7832194588597895e-05,
+      "loss": 1.1502,
+      "step": 44395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7831724725083252e-05,
+      "loss": 1.2913,
+      "step": 44400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7831254816845253e-05,
+      "loss": 1.2926,
+      "step": 44405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7830784863886574e-05,
+      "loss": 1.1008,
+      "step": 44410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7830314866209905e-05,
+      "loss": 1.5533,
+      "step": 44415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7829844823817923e-05,
+      "loss": 1.5318,
+      "step": 44420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7829374736713318e-05,
+      "loss": 1.2525,
+      "step": 44425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7828904604898773e-05,
+      "loss": 1.1236,
+      "step": 44430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7828434428376972e-05,
+      "loss": 1.1916,
+      "step": 44435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7827964207150595e-05,
+      "loss": 1.3278,
+      "step": 44440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.782749394122234e-05,
+      "loss": 1.3283,
+      "step": 44445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.782702363059488e-05,
+      "loss": 1.1222,
+      "step": 44450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7826553275270903e-05,
+      "loss": 1.3175,
+      "step": 44455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7826082875253103e-05,
+      "loss": 1.3571,
+      "step": 44460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7825612430544157e-05,
+      "loss": 1.4342,
+      "step": 44465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7825141941146756e-05,
+      "loss": 1.4411,
+      "step": 44470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7824671407063583e-05,
+      "loss": 1.0929,
+      "step": 44475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7824200828297332e-05,
+      "loss": 1.2658,
+      "step": 44480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7823730204850682e-05,
+      "loss": 1.1574,
+      "step": 44485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7823259536726327e-05,
+      "loss": 1.5259,
+      "step": 44490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7822788823926954e-05,
+      "loss": 1.3793,
+      "step": 44495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7822318066455247e-05,
+      "loss": 1.1076,
+      "step": 44500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7821847264313895e-05,
+      "loss": 1.5231,
+      "step": 44505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7821376417505592e-05,
+      "loss": 1.3279,
+      "step": 44510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7820905526033018e-05,
+      "loss": 1.3898,
+      "step": 44515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.782043458989887e-05,
+      "loss": 1.4233,
+      "step": 44520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7819963609105832e-05,
+      "loss": 1.2327,
+      "step": 44525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.78194925836566e-05,
+      "loss": 1.315,
+      "step": 44530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7819021513553857e-05,
+      "loss": 1.4482,
+      "step": 44535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7818550398800295e-05,
+      "loss": 1.0179,
+      "step": 44540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.781807923939861e-05,
+      "loss": 1.1605,
+      "step": 44545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.781760803535148e-05,
+      "loss": 1.3997,
+      "step": 44550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7817136786661607e-05,
+      "loss": 1.4891,
+      "step": 44555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.781666549333168e-05,
+      "loss": 1.191,
+      "step": 44560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7816194155364387e-05,
+      "loss": 1.3733,
+      "step": 44565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.781572277276242e-05,
+      "loss": 1.5139,
+      "step": 44570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7815251345528474e-05,
+      "loss": 1.0653,
+      "step": 44575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.781477987366524e-05,
+      "loss": 1.1328,
+      "step": 44580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7814308357175408e-05,
+      "loss": 1.0035,
+      "step": 44585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7813836796061673e-05,
+      "loss": 1.3033,
+      "step": 44590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.781336519032673e-05,
+      "loss": 1.0768,
+      "step": 44595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.781289353997327e-05,
+      "loss": 1.3939,
+      "step": 44600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.781242184500398e-05,
+      "loss": 1.2261,
+      "step": 44605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.781195010542156e-05,
+      "loss": 1.4575,
+      "step": 44610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7811478321228704e-05,
+      "loss": 1.2004,
+      "step": 44615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.781100649242811e-05,
+      "loss": 1.2508,
+      "step": 44620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7810534619022464e-05,
+      "loss": 1.4264,
+      "step": 44625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7810062701014462e-05,
+      "loss": 1.0136,
+      "step": 44630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7809590738406805e-05,
+      "loss": 1.0102,
+      "step": 44635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7809118731202186e-05,
+      "loss": 1.1521,
+      "step": 44640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7808646679403297e-05,
+      "loss": 1.2837,
+      "step": 44645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7808174583012834e-05,
+      "loss": 1.2231,
+      "step": 44650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7807702442033496e-05,
+      "loss": 1.3074,
+      "step": 44655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7807230256467976e-05,
+      "loss": 1.7127,
+      "step": 44660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7806758026318976e-05,
+      "loss": 1.122,
+      "step": 44665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7806285751589185e-05,
+      "loss": 1.4111,
+      "step": 44670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7805813432281303e-05,
+      "loss": 1.588,
+      "step": 44675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.780534106839803e-05,
+      "loss": 0.9929,
+      "step": 44680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.780486865994206e-05,
+      "loss": 1.2255,
+      "step": 44685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7804396206916096e-05,
+      "loss": 1.3568,
+      "step": 44690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.780392370932283e-05,
+      "loss": 1.5643,
+      "step": 44695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7803451167164957e-05,
+      "loss": 1.0059,
+      "step": 44700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7802978580445186e-05,
+      "loss": 1.1333,
+      "step": 44705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.780250594916621e-05,
+      "loss": 0.9887,
+      "step": 44710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.780203327333073e-05,
+      "loss": 1.5605,
+      "step": 44715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.780156055294144e-05,
+      "loss": 1.1357,
+      "step": 44720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7801087788001045e-05,
+      "loss": 1.1415,
+      "step": 44725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7800614978512243e-05,
+      "loss": 1.1689,
+      "step": 44730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7800142124477737e-05,
+      "loss": 1.5519,
+      "step": 44735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7799669225900218e-05,
+      "loss": 1.3816,
+      "step": 44740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7799196282782393e-05,
+      "loss": 1.9012,
+      "step": 44745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7798723295126966e-05,
+      "loss": 1.1362,
+      "step": 44750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7798250262936637e-05,
+      "loss": 1.2306,
+      "step": 44755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.77977771862141e-05,
+      "loss": 1.2381,
+      "step": 44760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779730406496206e-05,
+      "loss": 1.3396,
+      "step": 44765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7796830899183225e-05,
+      "loss": 1.5893,
+      "step": 44770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779635768888029e-05,
+      "loss": 1.1738,
+      "step": 44775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7795884434055955e-05,
+      "loss": 1.0435,
+      "step": 44780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779541113471293e-05,
+      "loss": 1.1438,
+      "step": 44785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7794937790853917e-05,
+      "loss": 1.2557,
+      "step": 44790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7794464402481612e-05,
+      "loss": 0.9208,
+      "step": 44795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779399096959873e-05,
+      "loss": 1.4923,
+      "step": 44800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779351749220796e-05,
+      "loss": 1.035,
+      "step": 44805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7793043970312018e-05,
+      "loss": 1.0935,
+      "step": 44810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.77925704039136e-05,
+      "loss": 1.0425,
+      "step": 44815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7792096793015418e-05,
+      "loss": 1.211,
+      "step": 44820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7791623137620167e-05,
+      "loss": 1.1517,
+      "step": 44825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.779114943773056e-05,
+      "loss": 1.1368,
+      "step": 44830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7790675693349306e-05,
+      "loss": 1.347,
+      "step": 44835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7790201904479095e-05,
+      "loss": 1.3618,
+      "step": 44840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7789728071122643e-05,
+      "loss": 1.3193,
+      "step": 44845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7789254193282653e-05,
+      "loss": 1.3182,
+      "step": 44850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7788780270961835e-05,
+      "loss": 1.5269,
+      "step": 44855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7788306304162887e-05,
+      "loss": 1.4093,
+      "step": 44860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7787832292888527e-05,
+      "loss": 1.1648,
+      "step": 44865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7787358237141453e-05,
+      "loss": 1.7398,
+      "step": 44870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7786884136924377e-05,
+      "loss": 1.1641,
+      "step": 44875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7786409992240003e-05,
+      "loss": 1.3697,
+      "step": 44880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778593580309104e-05,
+      "loss": 1.2845,
+      "step": 44885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7785461569480197e-05,
+      "loss": 1.2323,
+      "step": 44890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7784987291410177e-05,
+      "loss": 1.4962,
+      "step": 44895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7784512968883697e-05,
+      "loss": 1.2374,
+      "step": 44900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778403860190346e-05,
+      "loss": 1.1244,
+      "step": 44905
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7783564190472175e-05,
+      "loss": 1.2553,
+      "step": 44910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778308973459255e-05,
+      "loss": 1.5793,
+      "step": 44915
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7782615234267303e-05,
+      "loss": 0.9869,
+      "step": 44920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7782140689499134e-05,
+      "loss": 1.3249,
+      "step": 44925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778166610029075e-05,
+      "loss": 1.4003,
+      "step": 44930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7781191466644874e-05,
+      "loss": 1.2559,
+      "step": 44935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.778071678856421e-05,
+      "loss": 1.0724,
+      "step": 44940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7780242066051464e-05,
+      "loss": 1.4348,
+      "step": 44945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.777976729910935e-05,
+      "loss": 1.2238,
+      "step": 44950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7779292487740588e-05,
+      "loss": 1.1938,
+      "step": 44955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7778817631947876e-05,
+      "loss": 1.3113,
+      "step": 44960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7778342731733932e-05,
+      "loss": 1.4241,
+      "step": 44965
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7777867787101468e-05,
+      "loss": 1.2287,
+      "step": 44970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7777392798053198e-05,
+      "loss": 1.2027,
+      "step": 44975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.777691776459183e-05,
+      "loss": 1.1846,
+      "step": 44980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7776442686720076e-05,
+      "loss": 1.4367,
+      "step": 44985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7775967564440656e-05,
+      "loss": 1.3671,
+      "step": 44990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7775492397756275e-05,
+      "loss": 1.3203,
+      "step": 44995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.777501718666965e-05,
+      "loss": 1.6157,
+      "step": 45000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7774541931183498e-05,
+      "loss": 1.21,
+      "step": 45005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7774066631300534e-05,
+      "loss": 1.2633,
+      "step": 45010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7773591287023464e-05,
+      "loss": 1.0386,
+      "step": 45015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7773115898355004e-05,
+      "loss": 1.7029,
+      "step": 45020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7772640465297874e-05,
+      "loss": 1.1188,
+      "step": 45025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7772164987854788e-05,
+      "loss": 1.1195,
+      "step": 45030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7771689466028455e-05,
+      "loss": 1.5022,
+      "step": 45035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.77712138998216e-05,
+      "loss": 1.2281,
+      "step": 45040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7770738289236933e-05,
+      "loss": 0.9799,
+      "step": 45045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7770262634277168e-05,
+      "loss": 1.0727,
+      "step": 45050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7769786934945025e-05,
+      "loss": 1.1458,
+      "step": 45055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7769311191243224e-05,
+      "loss": 1.7115,
+      "step": 45060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776883540317447e-05,
+      "loss": 1.4046,
+      "step": 45065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7768359570741495e-05,
+      "loss": 1.3879,
+      "step": 45070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7767883693947002e-05,
+      "loss": 1.2651,
+      "step": 45075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776740777279372e-05,
+      "loss": 1.0034,
+      "step": 45080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7766931807284357e-05,
+      "loss": 1.2291,
+      "step": 45085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776645579742164e-05,
+      "loss": 1.3015,
+      "step": 45090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7765979743208285e-05,
+      "loss": 1.0306,
+      "step": 45095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7765503644647e-05,
+      "loss": 1.4483,
+      "step": 45100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7765027501740523e-05,
+      "loss": 1.0229,
+      "step": 45105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7764551314491555e-05,
+      "loss": 1.4589,
+      "step": 45110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7764075082902828e-05,
+      "loss": 1.7408,
+      "step": 45115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7763598806977052e-05,
+      "loss": 1.1582,
+      "step": 45120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7763122486716952e-05,
+      "loss": 1.1305,
+      "step": 45125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7762646122125252e-05,
+      "loss": 1.3116,
+      "step": 45130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776216971320466e-05,
+      "loss": 1.4955,
+      "step": 45135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7761693259957912e-05,
+      "loss": 1.3676,
+      "step": 45140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7761216762387717e-05,
+      "loss": 1.2081,
+      "step": 45145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7760740220496798e-05,
+      "loss": 1.0808,
+      "step": 45150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.776026363428788e-05,
+      "loss": 1.2192,
+      "step": 45155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.775978700376368e-05,
+      "loss": 1.4003,
+      "step": 45160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7759310328926927e-05,
+      "loss": 1.0973,
+      "step": 45165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7758833609780337e-05,
+      "loss": 1.2415,
+      "step": 45170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7758356846326636e-05,
+      "loss": 1.0136,
+      "step": 45175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.775788003856854e-05,
+      "loss": 0.9625,
+      "step": 45180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7757403186508776e-05,
+      "loss": 1.3089,
+      "step": 45185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7756926290150073e-05,
+      "loss": 1.1264,
+      "step": 45190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7756449349495148e-05,
+      "loss": 1.3207,
+      "step": 45195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7755972364546718e-05,
+      "loss": 1.1959,
+      "step": 45200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7755495335307522e-05,
+      "loss": 1.2082,
+      "step": 45205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7755018261780274e-05,
+      "loss": 1.163,
+      "step": 45210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.77545411439677e-05,
+      "loss": 1.3329,
+      "step": 45215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7754063981872524e-05,
+      "loss": 1.4619,
+      "step": 45220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7753586775497475e-05,
+      "loss": 1.2107,
+      "step": 45225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7753109524845276e-05,
+      "loss": 1.2533,
+      "step": 45230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7752632229918648e-05,
+      "loss": 1.0565,
+      "step": 45235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.775215489072032e-05,
+      "loss": 0.8614,
+      "step": 45240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.775167750725302e-05,
+      "loss": 1.2897,
+      "step": 45245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7751200079519474e-05,
+      "loss": 1.1688,
+      "step": 45250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7750722607522407e-05,
+      "loss": 1.3665,
+      "step": 45255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7750245091264542e-05,
+      "loss": 1.2043,
+      "step": 45260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.774976753074861e-05,
+      "loss": 1.1871,
+      "step": 45265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7749289925977336e-05,
+      "loss": 1.2707,
+      "step": 45270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.774881227695345e-05,
+      "loss": 1.3473,
+      "step": 45275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7748334583679678e-05,
+      "loss": 1.274,
+      "step": 45280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7747856846158748e-05,
+      "loss": 1.0817,
+      "step": 45285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7747379064393387e-05,
+      "loss": 1.4862,
+      "step": 45290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7746901238386327e-05,
+      "loss": 1.0965,
+      "step": 45295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.774642336814029e-05,
+      "loss": 1.3033,
+      "step": 45300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7745945453658014e-05,
+      "loss": 1.2377,
+      "step": 45305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7745467494942223e-05,
+      "loss": 1.3864,
+      "step": 45310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7744989491995642e-05,
+      "loss": 0.9482,
+      "step": 45315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7744511444821012e-05,
+      "loss": 1.0983,
+      "step": 45320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.774403335342105e-05,
+      "loss": 1.165,
+      "step": 45325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7743555217798497e-05,
+      "loss": 1.2144,
+      "step": 45330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7743077037956075e-05,
+      "loss": 1.2232,
+      "step": 45335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.774259881389652e-05,
+      "loss": 1.4409,
+      "step": 45340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7742120545622564e-05,
+      "loss": 1.4885,
+      "step": 45345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7741642233136933e-05,
+      "loss": 1.3053,
+      "step": 45350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7741163876442362e-05,
+      "loss": 1.1887,
+      "step": 45355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.774068547554158e-05,
+      "loss": 1.1557,
+      "step": 45360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.774020703043732e-05,
+      "loss": 1.2354,
+      "step": 45365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7739728541132317e-05,
+      "loss": 1.0738,
+      "step": 45370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.77392500076293e-05,
+      "loss": 1.0621,
+      "step": 45375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7738771429931004e-05,
+      "loss": 1.1375,
+      "step": 45380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7738292808040162e-05,
+      "loss": 1.4908,
+      "step": 45385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7737814141959503e-05,
+      "loss": 1.1849,
+      "step": 45390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7737335431691764e-05,
+      "loss": 1.3664,
+      "step": 45395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.773685667723968e-05,
+      "loss": 1.3841,
+      "step": 45400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.773637787860598e-05,
+      "loss": 1.5261,
+      "step": 45405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7735899035793405e-05,
+      "loss": 1.3702,
+      "step": 45410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7735420148804686e-05,
+      "loss": 1.2465,
+      "step": 45415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7734941217642555e-05,
+      "loss": 1.1267,
+      "step": 45420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7734462242309747e-05,
+      "loss": 0.9478,
+      "step": 45425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7733983222809e-05,
+      "loss": 1.3486,
+      "step": 45430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7733504159143057e-05,
+      "loss": 1.2108,
+      "step": 45435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7733025051314638e-05,
+      "loss": 1.1812,
+      "step": 45440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7732545899326488e-05,
+      "loss": 1.4563,
+      "step": 45445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7732066703181345e-05,
+      "loss": 1.5998,
+      "step": 45450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.773158746288194e-05,
+      "loss": 1.1851,
+      "step": 45455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7731108178431013e-05,
+      "loss": 1.0466,
+      "step": 45460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7730628849831296e-05,
+      "loss": 1.5892,
+      "step": 45465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7730149477085533e-05,
+      "loss": 1.0562,
+      "step": 45470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7729670060196462e-05,
+      "loss": 1.3355,
+      "step": 45475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7729190599166813e-05,
+      "loss": 1.2408,
+      "step": 45480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7728711093999334e-05,
+      "loss": 1.145,
+      "step": 45485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7728231544696753e-05,
+      "loss": 1.0927,
+      "step": 45490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7727751951261814e-05,
+      "loss": 1.1168,
+      "step": 45495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772727231369726e-05,
+      "loss": 0.8929,
+      "step": 45500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772679263200582e-05,
+      "loss": 1.1997,
+      "step": 45505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7726312906190236e-05,
+      "loss": 1.161,
+      "step": 45510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7725833136253255e-05,
+      "loss": 1.3214,
+      "step": 45515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7725353322197607e-05,
+      "loss": 1.3849,
+      "step": 45520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772487346402604e-05,
+      "loss": 1.1702,
+      "step": 45525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7724393561741287e-05,
+      "loss": 1.2159,
+      "step": 45530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7723913615346092e-05,
+      "loss": 1.2291,
+      "step": 45535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.77234336248432e-05,
+      "loss": 1.3989,
+      "step": 45540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7722953590235347e-05,
+      "loss": 1.1106,
+      "step": 45545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772247351152527e-05,
+      "loss": 1.3052,
+      "step": 45550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7721993388715723e-05,
+      "loss": 1.2336,
+      "step": 45555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772151322180944e-05,
+      "loss": 1.6883,
+      "step": 45560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772103301080916e-05,
+      "loss": 1.3414,
+      "step": 45565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772055275571763e-05,
+      "loss": 1.3691,
+      "step": 45570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.772007245653759e-05,
+      "loss": 1.1853,
+      "step": 45575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7719592113271785e-05,
+      "loss": 1.1572,
+      "step": 45580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7719111725922958e-05,
+      "loss": 0.9959,
+      "step": 45585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.771863129449385e-05,
+      "loss": 0.9819,
+      "step": 45590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7718150818987204e-05,
+      "loss": 1.0976,
+      "step": 45595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.771767029940577e-05,
+      "loss": 1.2183,
+      "step": 45600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7717189735752286e-05,
+      "loss": 1.6275,
+      "step": 45605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7716709128029498e-05,
+      "loss": 1.1241,
+      "step": 45610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7716228476240146e-05,
+      "loss": 1.0532,
+      "step": 45615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.771574778038698e-05,
+      "loss": 1.2059,
+      "step": 45620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.771526704047275e-05,
+      "loss": 1.1273,
+      "step": 45625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7714786256500194e-05,
+      "loss": 1.1269,
+      "step": 45630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7714305428472055e-05,
+      "loss": 1.2862,
+      "step": 45635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.771382455639109e-05,
+      "loss": 1.311,
+      "step": 45640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.771334364026003e-05,
+      "loss": 1.1442,
+      "step": 45645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7712862680081632e-05,
+      "loss": 1.6376,
+      "step": 45650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7712381675858642e-05,
+      "loss": 1.3281,
+      "step": 45655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7711900627593804e-05,
+      "loss": 1.2421,
+      "step": 45660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7711419535289864e-05,
+      "loss": 1.1982,
+      "step": 45665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.771093839894957e-05,
+      "loss": 1.2187,
+      "step": 45670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.771045721857567e-05,
+      "loss": 1.3219,
+      "step": 45675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7709975994170915e-05,
+      "loss": 1.7756,
+      "step": 45680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.770949472573805e-05,
+      "loss": 1.1493,
+      "step": 45685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7709013413279825e-05,
+      "loss": 1.2583,
+      "step": 45690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7708532056798983e-05,
+      "loss": 1.115,
+      "step": 45695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7708050656298276e-05,
+      "loss": 1.1989,
+      "step": 45700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7707569211780458e-05,
+      "loss": 1.2444,
+      "step": 45705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7707087723248274e-05,
+      "loss": 1.1864,
+      "step": 45710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.770660619070447e-05,
+      "loss": 1.2824,
+      "step": 45715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.77061246141518e-05,
+      "loss": 1.3365,
+      "step": 45720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.770564299359302e-05,
+      "loss": 1.3577,
+      "step": 45725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.770516132903087e-05,
+      "loss": 1.3168,
+      "step": 45730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7704679620468103e-05,
+      "loss": 1.2682,
+      "step": 45735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7704197867907472e-05,
+      "loss": 1.1203,
+      "step": 45740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.770371607135173e-05,
+      "loss": 1.2929,
+      "step": 45745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7703234230803627e-05,
+      "loss": 1.2153,
+      "step": 45750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7702752346265908e-05,
+      "loss": 1.123,
+      "step": 45755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7702270417741335e-05,
+      "loss": 1.4198,
+      "step": 45760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.770178844523265e-05,
+      "loss": 1.2012,
+      "step": 45765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7701306428742615e-05,
+      "loss": 1.007,
+      "step": 45770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7700824368273975e-05,
+      "loss": 1.2785,
+      "step": 45775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7700342263829488e-05,
+      "loss": 1.6547,
+      "step": 45780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7699860115411905e-05,
+      "loss": 1.9492,
+      "step": 45785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.769937792302398e-05,
+      "loss": 1.4233,
+      "step": 45790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7698895686668466e-05,
+      "loss": 1.1493,
+      "step": 45795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.769841340634811e-05,
+      "loss": 0.9474,
+      "step": 45800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7697931082065682e-05,
+      "loss": 1.4258,
+      "step": 45805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7697448713823924e-05,
+      "loss": 0.9268,
+      "step": 45810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7696966301625594e-05,
+      "loss": 1.253,
+      "step": 45815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7696483845473444e-05,
+      "loss": 1.1555,
+      "step": 45820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7696001345370234e-05,
+      "loss": 1.1226,
+      "step": 45825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7695518801318717e-05,
+      "loss": 1.2443,
+      "step": 45830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7695036213321647e-05,
+      "loss": 1.094,
+      "step": 45835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7694553581381785e-05,
+      "loss": 1.1155,
+      "step": 45840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7694070905501882e-05,
+      "loss": 1.1615,
+      "step": 45845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7693588185684692e-05,
+      "loss": 1.0356,
+      "step": 45850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7693105421932978e-05,
+      "loss": 1.0249,
+      "step": 45855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7692622614249497e-05,
+      "loss": 0.9821,
+      "step": 45860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7692139762637e-05,
+      "loss": 1.072,
+      "step": 45865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7691656867098248e-05,
+      "loss": 1.2391,
+      "step": 45870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7691173927636e-05,
+      "loss": 1.257,
+      "step": 45875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.769069094425301e-05,
+      "loss": 1.3174,
+      "step": 45880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.769020791695204e-05,
+      "loss": 1.1245,
+      "step": 45885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.768972484573585e-05,
+      "loss": 1.1966,
+      "step": 45890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.768924173060719e-05,
+      "loss": 1.2508,
+      "step": 45895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7688758571568825e-05,
+      "loss": 1.4241,
+      "step": 45900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7688275368623514e-05,
+      "loss": 1.5751,
+      "step": 45905
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7687792121774015e-05,
+      "loss": 1.0102,
+      "step": 45910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7687308831023087e-05,
+      "loss": 1.2041,
+      "step": 45915
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.768682549637349e-05,
+      "loss": 1.2562,
+      "step": 45920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.768634211782799e-05,
+      "loss": 1.3226,
+      "step": 45925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7685858695389338e-05,
+      "loss": 1.15,
+      "step": 45930
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.76853752290603e-05,
+      "loss": 1.0791,
+      "step": 45935
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7684891718843638e-05,
+      "loss": 1.5457,
+      "step": 45940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.768440816474211e-05,
+      "loss": 1.3189,
+      "step": 45945
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7683924566758477e-05,
+      "loss": 1.3438,
+      "step": 45950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7683440924895502e-05,
+      "loss": 0.8733,
+      "step": 45955
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.768295723915595e-05,
+      "loss": 1.4554,
+      "step": 45960
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7682473509542575e-05,
+      "loss": 1.2582,
+      "step": 45965
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7681989736058147e-05,
+      "loss": 1.154,
+      "step": 45970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7681505918705423e-05,
+      "loss": 1.4768,
+      "step": 45975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7681022057487173e-05,
+      "loss": 1.0061,
+      "step": 45980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7680538152406152e-05,
+      "loss": 1.3518,
+      "step": 45985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7680054203465128e-05,
+      "loss": 1.5298,
+      "step": 45990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.767957021066687e-05,
+      "loss": 1.2084,
+      "step": 45995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7679086174014127e-05,
+      "loss": 1.1839,
+      "step": 46000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7678602093509675e-05,
+      "loss": 0.9537,
+      "step": 46005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7678117969156273e-05,
+      "loss": 1.1488,
+      "step": 46010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.767763380095669e-05,
+      "loss": 0.9164,
+      "step": 46015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7677149588913688e-05,
+      "loss": 1.244,
+      "step": 46020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7676665333030035e-05,
+      "loss": 1.4743,
+      "step": 46025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.767618103330849e-05,
+      "loss": 1.1321,
+      "step": 46030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7675696689751824e-05,
+      "loss": 1.3251,
+      "step": 46035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7675212302362803e-05,
+      "loss": 1.3398,
+      "step": 46040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7674727871144193e-05,
+      "loss": 0.9255,
+      "step": 46045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7674243396098754e-05,
+      "loss": 1.1522,
+      "step": 46050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7673758877229264e-05,
+      "loss": 1.2933,
+      "step": 46055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7673274314538476e-05,
+      "loss": 1.3464,
+      "step": 46060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.767278970802917e-05,
+      "loss": 0.9423,
+      "step": 46065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7672305057704107e-05,
+      "loss": 1.218,
+      "step": 46070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7671820363566054e-05,
+      "loss": 1.1466,
+      "step": 46075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.767133562561778e-05,
+      "loss": 1.1391,
+      "step": 46080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7670850843862056e-05,
+      "loss": 1.003,
+      "step": 46085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7670366018301648e-05,
+      "loss": 1.1482,
+      "step": 46090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7669881148939322e-05,
+      "loss": 1.0851,
+      "step": 46095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.766939623577785e-05,
+      "loss": 1.2067,
+      "step": 46100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7668911278820003e-05,
+      "loss": 1.2096,
+      "step": 46105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7668426278068546e-05,
+      "loss": 1.4046,
+      "step": 46110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7667941233526248e-05,
+      "loss": 1.4708,
+      "step": 46115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7667456145195883e-05,
+      "loss": 1.0184,
+      "step": 46120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7666971013080224e-05,
+      "loss": 1.0022,
+      "step": 46125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7666485837182032e-05,
+      "loss": 1.0481,
+      "step": 46130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7666000617504084e-05,
+      "loss": 1.4198,
+      "step": 46135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.766551535404915e-05,
+      "loss": 1.5038,
+      "step": 46140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7665030046819997e-05,
+      "loss": 1.2103,
+      "step": 46145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7664544695819403e-05,
+      "loss": 1.2836,
+      "step": 46150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7664059301050133e-05,
+      "loss": 1.142,
+      "step": 46155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7663573862514966e-05,
+      "loss": 1.4086,
+      "step": 46160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.766308838021667e-05,
+      "loss": 1.2472,
+      "step": 46165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7662602854158017e-05,
+      "loss": 1.3542,
+      "step": 46170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.766211728434178e-05,
+      "loss": 1.2021,
+      "step": 46175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7661631670770733e-05,
+      "loss": 1.0659,
+      "step": 46180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7661146013447646e-05,
+      "loss": 1.3224,
+      "step": 46185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7660660312375294e-05,
+      "loss": 1.3113,
+      "step": 46190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7660174567556455e-05,
+      "loss": 1.5163,
+      "step": 46195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.76596887789939e-05,
+      "loss": 1.2472,
+      "step": 46200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.76592029466904e-05,
+      "loss": 1.487,
+      "step": 46205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7658717070648732e-05,
+      "loss": 1.0559,
+      "step": 46210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7658231150871667e-05,
+      "loss": 1.1359,
+      "step": 46215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.765774518736199e-05,
+      "loss": 1.4726,
+      "step": 46220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7657259180122467e-05,
+      "loss": 1.2463,
+      "step": 46225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.765677312915587e-05,
+      "loss": 2.0234,
+      "step": 46230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7656287034464985e-05,
+      "loss": 1.3631,
+      "step": 46235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7655800896052585e-05,
+      "loss": 1.2215,
+      "step": 46240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.765531471392144e-05,
+      "loss": 1.2759,
+      "step": 46245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7654828488074332e-05,
+      "loss": 1.5118,
+      "step": 46250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7654342218514036e-05,
+      "loss": 1.0481,
+      "step": 46255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.765385590524333e-05,
+      "loss": 1.2483,
+      "step": 46260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.765336954826499e-05,
+      "loss": 0.9677,
+      "step": 46265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.765288314758179e-05,
+      "loss": 1.5283,
+      "step": 46270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7652396703196514e-05,
+      "loss": 1.4784,
+      "step": 46275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.765191021511194e-05,
+      "loss": 1.2179,
+      "step": 46280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7651423683330838e-05,
+      "loss": 1.3484,
+      "step": 46285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7650937107855994e-05,
+      "loss": 1.114,
+      "step": 46290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7650450488690183e-05,
+      "loss": 1.352,
+      "step": 46295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7649963825836186e-05,
+      "loss": 1.4588,
+      "step": 46300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.764947711929678e-05,
+      "loss": 1.094,
+      "step": 46305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7648990369074748e-05,
+      "loss": 1.0708,
+      "step": 46310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.764850357517286e-05,
+      "loss": 1.3576,
+      "step": 46315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.764801673759391e-05,
+      "loss": 1.3262,
+      "step": 46320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.764752985634067e-05,
+      "loss": 1.3217,
+      "step": 46325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7647042931415922e-05,
+      "loss": 1.1481,
+      "step": 46330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.764655596282244e-05,
+      "loss": 1.2955,
+      "step": 46335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7646068950563012e-05,
+      "loss": 1.1893,
+      "step": 46340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7645581894640425e-05,
+      "loss": 1.3117,
+      "step": 46345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7645094795057445e-05,
+      "loss": 1.1692,
+      "step": 46350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7644607651816867e-05,
+      "loss": 1.207,
+      "step": 46355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7644120464921468e-05,
+      "loss": 1.0675,
+      "step": 46360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7643633234374025e-05,
+      "loss": 1.022,
+      "step": 46365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.764314596017733e-05,
+      "loss": 1.0231,
+      "step": 46370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7642658642334156e-05,
+      "loss": 1.2526,
+      "step": 46375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7642171280847295e-05,
+      "loss": 1.3995,
+      "step": 46380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.764168387571952e-05,
+      "loss": 1.0743,
+      "step": 46385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7641196426953624e-05,
+      "loss": 1.6153,
+      "step": 46390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.764070893455239e-05,
+      "loss": 1.4329,
+      "step": 46395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7640221398518592e-05,
+      "loss": 1.4855,
+      "step": 46400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7639733818855024e-05,
+      "loss": 1.0895,
+      "step": 46405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7639246195564466e-05,
+      "loss": 1.1861,
+      "step": 46410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.76387585286497e-05,
+      "loss": 1.1972,
+      "step": 46415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.763827081811352e-05,
+      "loss": 1.1356,
+      "step": 46420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7637783063958705e-05,
+      "loss": 1.3409,
+      "step": 46425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7637295266188036e-05,
+      "loss": 1.438,
+      "step": 46430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7636807424804305e-05,
+      "loss": 1.2297,
+      "step": 46435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7636319539810297e-05,
+      "loss": 1.4586,
+      "step": 46440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7635831611208796e-05,
+      "loss": 1.0826,
+      "step": 46445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7635343639002592e-05,
+      "loss": 1.2956,
+      "step": 46450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.763485562319447e-05,
+      "loss": 1.0931,
+      "step": 46455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7634367563787215e-05,
+      "loss": 1.271,
+      "step": 46460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7633879460783613e-05,
+      "loss": 1.1737,
+      "step": 46465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7633391314186453e-05,
+      "loss": 1.0584,
+      "step": 46470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7632903123998525e-05,
+      "loss": 1.0101,
+      "step": 46475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7632414890222614e-05,
+      "loss": 1.223,
+      "step": 46480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7631926612861508e-05,
+      "loss": 1.0839,
+      "step": 46485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7631438291918e-05,
+      "loss": 1.6339,
+      "step": 46490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7630949927394868e-05,
+      "loss": 1.151,
+      "step": 46495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7630461519294914e-05,
+      "loss": 1.07,
+      "step": 46500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.762997306762092e-05,
+      "loss": 1.0159,
+      "step": 46505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.762948457237567e-05,
+      "loss": 1.5309,
+      "step": 46510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7628996033561965e-05,
+      "loss": 1.2238,
+      "step": 46515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7628507451182587e-05,
+      "loss": 1.0901,
+      "step": 46520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.762801882524033e-05,
+      "loss": 1.328,
+      "step": 46525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7627530155737984e-05,
+      "loss": 1.2236,
+      "step": 46530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7627041442678335e-05,
+      "loss": 1.0427,
+      "step": 46535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.762655268606418e-05,
+      "loss": 1.1907,
+      "step": 46540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7626063885898305e-05,
+      "loss": 1.1747,
+      "step": 46545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7625575042183506e-05,
+      "loss": 1.3612,
+      "step": 46550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7625086154922567e-05,
+      "loss": 1.1748,
+      "step": 46555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.762459722411829e-05,
+      "loss": 1.42,
+      "step": 46560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7624108249773463e-05,
+      "loss": 1.086,
+      "step": 46565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7623619231890875e-05,
+      "loss": 1.332,
+      "step": 46570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.762313017047332e-05,
+      "loss": 1.2774,
+      "step": 46575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.762264106552359e-05,
+      "loss": 1.1492,
+      "step": 46580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7622151917044484e-05,
+      "loss": 1.5453,
+      "step": 46585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7621662725038787e-05,
+      "loss": 1.3225,
+      "step": 46590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.76211734895093e-05,
+      "loss": 1.348,
+      "step": 46595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7620684210458812e-05,
+      "loss": 1.0475,
+      "step": 46600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7620194887890115e-05,
+      "loss": 1.3238,
+      "step": 46605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.761970552180601e-05,
+      "loss": 1.0348,
+      "step": 46610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7619216112209286e-05,
+      "loss": 1.2189,
+      "step": 46615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7618726659102742e-05,
+      "loss": 1.124,
+      "step": 46620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7618237162489174e-05,
+      "loss": 1.3225,
+      "step": 46625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.761774762237137e-05,
+      "loss": 1.0347,
+      "step": 46630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.761725803875213e-05,
+      "loss": 0.9609,
+      "step": 46635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.761676841163425e-05,
+      "loss": 1.1997,
+      "step": 46640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7616278741020525e-05,
+      "loss": 1.2725,
+      "step": 46645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7615789026913755e-05,
+      "loss": 1.2196,
+      "step": 46650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7615299269316732e-05,
+      "loss": 1.2827,
+      "step": 46655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7614809468232255e-05,
+      "loss": 1.3262,
+      "step": 46660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7614319623663116e-05,
+      "loss": 1.1068,
+      "step": 46665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.761382973561212e-05,
+      "loss": 1.0473,
+      "step": 46670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7613339804082064e-05,
+      "loss": 1.2211,
+      "step": 46675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7612849829075736e-05,
+      "loss": 1.2793,
+      "step": 46680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7612359810595947e-05,
+      "loss": 1.3349,
+      "step": 46685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7611869748645488e-05,
+      "loss": 1.3578,
+      "step": 46690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7611379643227155e-05,
+      "loss": 1.3868,
+      "step": 46695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7610889494343754e-05,
+      "loss": 1.4247,
+      "step": 46700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7610399301998078e-05,
+      "loss": 1.0474,
+      "step": 46705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.760990906619293e-05,
+      "loss": 1.1569,
+      "step": 46710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7609418786931108e-05,
+      "loss": 1.3582,
+      "step": 46715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7608928464215414e-05,
+      "loss": 1.2164,
+      "step": 46720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7608438098048644e-05,
+      "loss": 1.1984,
+      "step": 46725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.76079476884336e-05,
+      "loss": 1.1995,
+      "step": 46730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7607457235373085e-05,
+      "loss": 1.4801,
+      "step": 46735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7606966738869892e-05,
+      "loss": 1.0272,
+      "step": 46740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7606476198926835e-05,
+      "loss": 1.2409,
+      "step": 46745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7605985615546703e-05,
+      "loss": 1.0221,
+      "step": 46750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7605494988732303e-05,
+      "loss": 1.0139,
+      "step": 46755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7605004318486437e-05,
+      "loss": 1.1195,
+      "step": 46760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7604513604811905e-05,
+      "loss": 1.392,
+      "step": 46765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.760402284771151e-05,
+      "loss": 1.3223,
+      "step": 46770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7603532047188056e-05,
+      "loss": 1.3322,
+      "step": 46775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7603041203244342e-05,
+      "loss": 1.3832,
+      "step": 46780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7602550315883173e-05,
+      "loss": 1.2377,
+      "step": 46785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7602059385107355e-05,
+      "loss": 1.0309,
+      "step": 46790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7601568410919688e-05,
+      "loss": 1.1593,
+      "step": 46795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7601077393322978e-05,
+      "loss": 1.4631,
+      "step": 46800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7600586332320028e-05,
+      "loss": 1.1629,
+      "step": 46805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7600095227913643e-05,
+      "loss": 1.1516,
+      "step": 46810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7599604080106624e-05,
+      "loss": 1.2197,
+      "step": 46815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.759911288890178e-05,
+      "loss": 1.1153,
+      "step": 46820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7598621654301912e-05,
+      "loss": 1.2387,
+      "step": 46825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.759813037630983e-05,
+      "loss": 1.0542,
+      "step": 46830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7597639054928334e-05,
+      "loss": 1.1734,
+      "step": 46835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7597147690160236e-05,
+      "loss": 1.484,
+      "step": 46840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7596656282008337e-05,
+      "loss": 1.0395,
+      "step": 46845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7596164830475443e-05,
+      "loss": 1.109,
+      "step": 46850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7595673335564365e-05,
+      "loss": 1.0853,
+      "step": 46855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7595181797277905e-05,
+      "loss": 1.3393,
+      "step": 46860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7594690215618873e-05,
+      "loss": 1.1288,
+      "step": 46865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7594198590590076e-05,
+      "loss": 1.2234,
+      "step": 46870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7593706922194318e-05,
+      "loss": 1.222,
+      "step": 46875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.759321521043441e-05,
+      "loss": 1.3057,
+      "step": 46880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.759272345531316e-05,
+      "loss": 1.2532,
+      "step": 46885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7592231656833375e-05,
+      "loss": 1.466,
+      "step": 46890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7591739814997863e-05,
+      "loss": 1.5071,
+      "step": 46895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7591247929809432e-05,
+      "loss": 1.1309,
+      "step": 46900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7590756001270896e-05,
+      "loss": 1.3177,
+      "step": 46905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.759026402938506e-05,
+      "loss": 1.1684,
+      "step": 46910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7589772014154733e-05,
+      "loss": 1.3106,
+      "step": 46915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7589279955582724e-05,
+      "loss": 1.5725,
+      "step": 46920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7588787853671844e-05,
+      "loss": 1.5424,
+      "step": 46925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7588295708424908e-05,
+      "loss": 1.4737,
+      "step": 46930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7587803519844716e-05,
+      "loss": 1.2342,
+      "step": 46935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.758731128793409e-05,
+      "loss": 1.1105,
+      "step": 46940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7586819012695836e-05,
+      "loss": 1.2087,
+      "step": 46945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.758632669413276e-05,
+      "loss": 1.007,
+      "step": 46950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.758583433224768e-05,
+      "loss": 1.1617,
+      "step": 46955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.758534192704341e-05,
+      "loss": 1.266,
+      "step": 46960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7584849478522755e-05,
+      "loss": 1.1617,
+      "step": 46965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.758435698668853e-05,
+      "loss": 1.0278,
+      "step": 46970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7583864451543548e-05,
+      "loss": 1.2002,
+      "step": 46975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7583371873090617e-05,
+      "loss": 1.0676,
+      "step": 46980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.758287925133256e-05,
+      "loss": 1.2083,
+      "step": 46985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7582386586272182e-05,
+      "loss": 1.1342,
+      "step": 46990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.75818938779123e-05,
+      "loss": 1.2165,
+      "step": 46995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.758140112625572e-05,
+      "loss": 0.9808,
+      "step": 47000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7580908331305267e-05,
+      "loss": 1.0848,
+      "step": 47005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7580415493063746e-05,
+      "loss": 1.3377,
+      "step": 47010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.757992261153398e-05,
+      "loss": 1.4309,
+      "step": 47015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7579429686718774e-05,
+      "loss": 1.1763,
+      "step": 47020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.757893671862095e-05,
+      "loss": 1.3644,
+      "step": 47025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.757844370724332e-05,
+      "loss": 1.2805,
+      "step": 47030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7577950652588708e-05,
+      "loss": 0.9413,
+      "step": 47035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7577457554659913e-05,
+      "loss": 1.3387,
+      "step": 47040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7576964413459763e-05,
+      "loss": 1.0265,
+      "step": 47045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.757647122899107e-05,
+      "loss": 1.2027,
+      "step": 47050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7575978001256654e-05,
+      "loss": 1.2238,
+      "step": 47055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7575484730259326e-05,
+      "loss": 1.1091,
+      "step": 47060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7574991416001905e-05,
+      "loss": 1.2755,
+      "step": 47065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7574498058487212e-05,
+      "loss": 1.1933,
+      "step": 47070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.757400465771806e-05,
+      "loss": 1.2821,
+      "step": 47075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.757351121369727e-05,
+      "loss": 1.0547,
+      "step": 47080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.757301772642765e-05,
+      "loss": 0.9949,
+      "step": 47085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.757252419591203e-05,
+      "loss": 1.1856,
+      "step": 47090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7572030622153224e-05,
+      "loss": 1.264,
+      "step": 47095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.757153700515405e-05,
+      "loss": 1.0505,
+      "step": 47100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.757104334491733e-05,
+      "loss": 1.4336,
+      "step": 47105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.757054964144588e-05,
+      "loss": 1.2227,
+      "step": 47110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.757005589474252e-05,
+      "loss": 1.3984,
+      "step": 47115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7569562104810063e-05,
+      "loss": 1.0414,
+      "step": 47120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.756906827165134e-05,
+      "loss": 1.3213,
+      "step": 47125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7568574395269167e-05,
+      "loss": 1.1954,
+      "step": 47130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.756808047566636e-05,
+      "loss": 1.2083,
+      "step": 47135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.756758651284575e-05,
+      "loss": 1.065,
+      "step": 47140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.756709250681014e-05,
+      "loss": 0.9963,
+      "step": 47145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.756659845756237e-05,
+      "loss": 1.4202,
+      "step": 47150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.756610436510525e-05,
+      "loss": 1.1253,
+      "step": 47155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7565610229441605e-05,
+      "loss": 1.4732,
+      "step": 47160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7565116050574256e-05,
+      "loss": 1.3844,
+      "step": 47165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7564621828506028e-05,
+      "loss": 1.4676,
+      "step": 47170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.756412756323974e-05,
+      "loss": 1.0596,
+      "step": 47175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7563633254778213e-05,
+      "loss": 1.0572,
+      "step": 47180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7563138903124274e-05,
+      "loss": 1.0692,
+      "step": 47185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7562644508280745e-05,
+      "loss": 1.4162,
+      "step": 47190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7562150070250442e-05,
+      "loss": 1.2861,
+      "step": 47195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.75616555890362e-05,
+      "loss": 1.0785,
+      "step": 47200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.756116106464084e-05,
+      "loss": 1.1822,
+      "step": 47205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.756066649706718e-05,
+      "loss": 1.1246,
+      "step": 47210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7560171886318046e-05,
+      "loss": 1.2004,
+      "step": 47215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7559677232396267e-05,
+      "loss": 1.1607,
+      "step": 47220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7559182535304664e-05,
+      "loss": 1.0845,
+      "step": 47225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7558687795046063e-05,
+      "loss": 1.501,
+      "step": 47230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.755819301162329e-05,
+      "loss": 1.1022,
+      "step": 47235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7557698185039173e-05,
+      "loss": 1.1958,
+      "step": 47240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7557203315296526e-05,
+      "loss": 1.2093,
+      "step": 47245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.755670840239819e-05,
+      "loss": 1.1359,
+      "step": 47250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7556213446346985e-05,
+      "loss": 1.131,
+      "step": 47255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7555718447145733e-05,
+      "loss": 1.2126,
+      "step": 47260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7555223404797266e-05,
+      "loss": 1.3352,
+      "step": 47265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.755472831930441e-05,
+      "loss": 1.2248,
+      "step": 47270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7554233190669995e-05,
+      "loss": 1.3162,
+      "step": 47275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7553738018896843e-05,
+      "loss": 1.0656,
+      "step": 47280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.755324280398778e-05,
+      "loss": 1.1854,
+      "step": 47285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7552747545945644e-05,
+      "loss": 1.0999,
+      "step": 47290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7552252244773257e-05,
+      "loss": 1.1898,
+      "step": 47295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.755175690047344e-05,
+      "loss": 1.4656,
+      "step": 47300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.755126151304904e-05,
+      "loss": 1.2746,
+      "step": 47305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.755076608250287e-05,
+      "loss": 1.3164,
+      "step": 47310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7550270608837763e-05,
+      "loss": 1.1124,
+      "step": 47315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7549775092056555e-05,
+      "loss": 1.2088,
+      "step": 47320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7549279532162067e-05,
+      "loss": 1.3796,
+      "step": 47325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7548783929157132e-05,
+      "loss": 1.2335,
+      "step": 47330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7548288283044584e-05,
+      "loss": 1.2991,
+      "step": 47335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7547792593827247e-05,
+      "loss": 1.133,
+      "step": 47340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7547296861507954e-05,
+      "loss": 1.0841,
+      "step": 47345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7546801086089538e-05,
+      "loss": 1.3449,
+      "step": 47350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7546305267574827e-05,
+      "loss": 1.6275,
+      "step": 47355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7545809405966658e-05,
+      "loss": 1.1138,
+      "step": 47360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7545313501267856e-05,
+      "loss": 1.1875,
+      "step": 47365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7544817553481257e-05,
+      "loss": 1.3751,
+      "step": 47370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7544321562609693e-05,
+      "loss": 0.9762,
+      "step": 47375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.754382552865599e-05,
+      "loss": 1.475,
+      "step": 47380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.754332945162299e-05,
+      "loss": 0.9859,
+      "step": 47385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.754283333151352e-05,
+      "loss": 0.9711,
+      "step": 47390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7542337168330418e-05,
+      "loss": 1.0961,
+      "step": 47395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.754184096207651e-05,
+      "loss": 1.3288,
+      "step": 47400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7541344712754635e-05,
+      "loss": 1.3615,
+      "step": 47405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7540848420367623e-05,
+      "loss": 1.1662,
+      "step": 47410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7540352084918317e-05,
+      "loss": 1.2002,
+      "step": 47415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.753985570640954e-05,
+      "loss": 1.2714,
+      "step": 47420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7539359284844134e-05,
+      "loss": 1.3202,
+      "step": 47425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.753886282022493e-05,
+      "loss": 1.2811,
+      "step": 47430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7538366312554763e-05,
+      "loss": 1.4408,
+      "step": 47435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.753786976183647e-05,
+      "loss": 1.2479,
+      "step": 47440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.753737316807289e-05,
+      "loss": 1.5202,
+      "step": 47445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7536876531266856e-05,
+      "loss": 1.216,
+      "step": 47450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7536379851421198e-05,
+      "loss": 1.427,
+      "step": 47455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.753588312853876e-05,
+      "loss": 1.0814,
+      "step": 47460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7535386362622374e-05,
+      "loss": 1.2641,
+      "step": 47465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7534889553674884e-05,
+      "loss": 1.2816,
+      "step": 47470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.753439270169912e-05,
+      "loss": 1.1637,
+      "step": 47475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7533895806697917e-05,
+      "loss": 1.3141,
+      "step": 47480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7533398868674122e-05,
+      "loss": 1.4466,
+      "step": 47485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7532901887630568e-05,
+      "loss": 1.2694,
+      "step": 47490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.753240486357009e-05,
+      "loss": 1.7133,
+      "step": 47495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.753190779649553e-05,
+      "loss": 1.4215,
+      "step": 47500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7531410686409724e-05,
+      "loss": 1.1132,
+      "step": 47505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7530913533315512e-05,
+      "loss": 1.1595,
+      "step": 47510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7530416337215737e-05,
+      "loss": 0.9476,
+      "step": 47515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.752991909811323e-05,
+      "loss": 1.2645,
+      "step": 47520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7529421816010836e-05,
+      "loss": 1.0074,
+      "step": 47525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7528924490911395e-05,
+      "loss": 1.2972,
+      "step": 47530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7528427122817745e-05,
+      "loss": 1.3208,
+      "step": 47535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7527929711732727e-05,
+      "loss": 1.1244,
+      "step": 47540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.752743225765918e-05,
+      "loss": 1.1666,
+      "step": 47545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.752693476059995e-05,
+      "loss": 1.2659,
+      "step": 47550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7526437220557872e-05,
+      "loss": 1.3152,
+      "step": 47555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7525939637535788e-05,
+      "loss": 1.2829,
+      "step": 47560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7525442011536543e-05,
+      "loss": 1.0123,
+      "step": 47565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7524944342562976e-05,
+      "loss": 1.4902,
+      "step": 47570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7524446630617927e-05,
+      "loss": 1.0767,
+      "step": 47575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7523948875704245e-05,
+      "loss": 1.418,
+      "step": 47580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7523451077824766e-05,
+      "loss": 1.2049,
+      "step": 47585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7522953236982335e-05,
+      "loss": 0.9723,
+      "step": 47590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7522455353179793e-05,
+      "loss": 1.0225,
+      "step": 47595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7521957426419986e-05,
+      "loss": 1.0161,
+      "step": 47600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7521459456705758e-05,
+      "loss": 1.4816,
+      "step": 47605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.752096144403995e-05,
+      "loss": 1.0947,
+      "step": 47610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7520463388425408e-05,
+      "loss": 1.18,
+      "step": 47615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7519965289864974e-05,
+      "loss": 1.0853,
+      "step": 47620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7519467148361494e-05,
+      "loss": 1.4841,
+      "step": 47625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.751896896391781e-05,
+      "loss": 1.2108,
+      "step": 47630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7518470736536774e-05,
+      "loss": 1.3479,
+      "step": 47635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7517972466221223e-05,
+      "loss": 1.4222,
+      "step": 47640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7517474152974006e-05,
+      "loss": 1.3453,
+      "step": 47645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7516975796797964e-05,
+      "loss": 0.8601,
+      "step": 47650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7516477397695953e-05,
+      "loss": 1.0634,
+      "step": 47655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.751597895567081e-05,
+      "loss": 1.2082,
+      "step": 47660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7515480470725388e-05,
+      "loss": 1.3535,
+      "step": 47665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7514981942862527e-05,
+      "loss": 1.1144,
+      "step": 47670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.751448337208508e-05,
+      "loss": 1.0881,
+      "step": 47675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.751398475839589e-05,
+      "loss": 1.4641,
+      "step": 47680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7513486101797808e-05,
+      "loss": 1.5341,
+      "step": 47685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7512987402293674e-05,
+      "loss": 1.3362,
+      "step": 47690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7512488659886346e-05,
+      "loss": 1.1989,
+      "step": 47695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7511989874578663e-05,
+      "loss": 0.9669,
+      "step": 47700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.751149104637348e-05,
+      "loss": 1.2582,
+      "step": 47705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7510992175273645e-05,
+      "loss": 1.0938,
+      "step": 47710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7510493261282002e-05,
+      "loss": 1.1009,
+      "step": 47715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7509994304401402e-05,
+      "loss": 1.163,
+      "step": 47720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7509495304634697e-05,
+      "loss": 1.0672,
+      "step": 47725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7508996261984736e-05,
+      "loss": 1.094,
+      "step": 47730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7508497176454366e-05,
+      "loss": 1.2253,
+      "step": 47735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7507998048046436e-05,
+      "loss": 1.1963,
+      "step": 47740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7507498876763803e-05,
+      "loss": 1.3535,
+      "step": 47745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7506999662609315e-05,
+      "loss": 0.9462,
+      "step": 47750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.750650040558582e-05,
+      "loss": 1.4241,
+      "step": 47755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7506001105696167e-05,
+      "loss": 1.441,
+      "step": 47760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7505501762943212e-05,
+      "loss": 1.1795,
+      "step": 47765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.75050023773298e-05,
+      "loss": 1.4319,
+      "step": 47770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7504502948858797e-05,
+      "loss": 1.4683,
+      "step": 47775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7504003477533044e-05,
+      "loss": 1.3105,
+      "step": 47780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7503503963355393e-05,
+      "loss": 1.0176,
+      "step": 47785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7503004406328697e-05,
+      "loss": 1.1558,
+      "step": 47790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7502504806455812e-05,
+      "loss": 1.7752,
+      "step": 47795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.750200516373959e-05,
+      "loss": 1.1884,
+      "step": 47800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7501505478182884e-05,
+      "loss": 1.2061,
+      "step": 47805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7501005749788548e-05,
+      "loss": 1.2795,
+      "step": 47810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.750050597855943e-05,
+      "loss": 0.9707,
+      "step": 47815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7500006164498394e-05,
+      "loss": 1.1821,
+      "step": 47820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7499506307608282e-05,
+      "loss": 1.1667,
+      "step": 47825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7499006407891964e-05,
+      "loss": 1.3282,
+      "step": 47830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.749850646535228e-05,
+      "loss": 1.2484,
+      "step": 47835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.749800647999209e-05,
+      "loss": 1.0317,
+      "step": 47840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7497506451814255e-05,
+      "loss": 1.061,
+      "step": 47845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7497006380821627e-05,
+      "loss": 1.1869,
+      "step": 47850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7496506267017054e-05,
+      "loss": 1.301,
+      "step": 47855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7496006110403402e-05,
+      "loss": 0.9649,
+      "step": 47860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.749550591098352e-05,
+      "loss": 1.133,
+      "step": 47865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7495005668760272e-05,
+      "loss": 1.359,
+      "step": 47870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.749450538373651e-05,
+      "loss": 1.2154,
+      "step": 47875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.749400505591509e-05,
+      "loss": 1.181,
+      "step": 47880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.749350468529887e-05,
+      "loss": 1.4662,
+      "step": 47885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.749300427189071e-05,
+      "loss": 1.145,
+      "step": 47890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.749250381569346e-05,
+      "loss": 1.3428,
+      "step": 47895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.749200331670999e-05,
+      "loss": 1.128,
+      "step": 47900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7491502774943144e-05,
+      "loss": 1.0907,
+      "step": 47905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7491002190395793e-05,
+      "loss": 1.1974,
+      "step": 47910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.749050156307079e-05,
+      "loss": 1.49,
+      "step": 47915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7490000892970997e-05,
+      "loss": 1.1487,
+      "step": 47920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7489500180099268e-05,
+      "loss": 1.4826,
+      "step": 47925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.748899942445846e-05,
+      "loss": 1.2291,
+      "step": 47930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7488498626051445e-05,
+      "loss": 1.3595,
+      "step": 47935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.748799778488107e-05,
+      "loss": 1.1982,
+      "step": 47940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7487496900950203e-05,
+      "loss": 1.5265,
+      "step": 47945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.74869959742617e-05,
+      "loss": 1.3462,
+      "step": 47950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7486495004818424e-05,
+      "loss": 1.2741,
+      "step": 47955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7485993992623233e-05,
+      "loss": 1.3781,
+      "step": 47960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7485492937678993e-05,
+      "loss": 1.2151,
+      "step": 47965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7484991839988562e-05,
+      "loss": 1.2215,
+      "step": 47970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.74844906995548e-05,
+      "loss": 1.1496,
+      "step": 47975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.748398951638057e-05,
+      "loss": 1.0895,
+      "step": 47980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7483488290468736e-05,
+      "loss": 1.1944,
+      "step": 47985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7482987021822157e-05,
+      "loss": 1.8504,
+      "step": 47990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.74824857104437e-05,
+      "loss": 1.6749,
+      "step": 47995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7481984356336223e-05,
+      "loss": 1.1397,
+      "step": 48000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.748148295950259e-05,
+      "loss": 1.2347,
+      "step": 48005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.748098151994567e-05,
+      "loss": 1.4421,
+      "step": 48010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.748048003766832e-05,
+      "loss": 1.4515,
+      "step": 48015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7479978512673405e-05,
+      "loss": 1.1129,
+      "step": 48020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7479476944963788e-05,
+      "loss": 1.4538,
+      "step": 48025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7478975334542335e-05,
+      "loss": 1.1674,
+      "step": 48030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7478473681411912e-05,
+      "loss": 1.3324,
+      "step": 48035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7477971985575382e-05,
+      "loss": 1.2869,
+      "step": 48040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7477470247035612e-05,
+      "loss": 1.1499,
+      "step": 48045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.747696846579546e-05,
+      "loss": 1.0798,
+      "step": 48050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7476466641857798e-05,
+      "loss": 1.3434,
+      "step": 48055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.747596477522549e-05,
+      "loss": 1.2425,
+      "step": 48060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7475462865901404e-05,
+      "loss": 1.3095,
+      "step": 48065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7474960913888406e-05,
+      "loss": 1.2748,
+      "step": 48070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7474458919189355e-05,
+      "loss": 1.2777,
+      "step": 48075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.747395688180713e-05,
+      "loss": 1.5286,
+      "step": 48080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7473454801744588e-05,
+      "loss": 1.0319,
+      "step": 48085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.74729526790046e-05,
+      "loss": 1.0819,
+      "step": 48090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.747245051359003e-05,
+      "loss": 1.2463,
+      "step": 48095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7471948305503753e-05,
+      "loss": 1.2904,
+      "step": 48100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7471446054748628e-05,
+      "loss": 1.4753,
+      "step": 48105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.747094376132753e-05,
+      "loss": 1.0883,
+      "step": 48110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7470441425243327e-05,
+      "loss": 1.622,
+      "step": 48115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7469939046498884e-05,
+      "loss": 1.209,
+      "step": 48120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.746943662509707e-05,
+      "loss": 1.1935,
+      "step": 48125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7468934161040754e-05,
+      "loss": 1.279,
+      "step": 48130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7468431654332808e-05,
+      "loss": 1.0023,
+      "step": 48135
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7467929104976102e-05,
+      "loss": 1.1743,
+      "step": 48140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7467426512973503e-05,
+      "loss": 1.144,
+      "step": 48145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7466923878327884e-05,
+      "loss": 1.3243,
+      "step": 48150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.746642120104211e-05,
+      "loss": 1.4018,
+      "step": 48155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.746591848111906e-05,
+      "loss": 1.1069,
+      "step": 48160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7465415718561595e-05,
+      "loss": 1.0557,
+      "step": 48165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7464912913372594e-05,
+      "loss": 1.1603,
+      "step": 48170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.746441006555492e-05,
+      "loss": 1.1959,
+      "step": 48175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7463907175111458e-05,
+      "loss": 1.0446,
+      "step": 48180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.746340424204507e-05,
+      "loss": 1.2282,
+      "step": 48185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7462901266358626e-05,
+      "loss": 1.105,
+      "step": 48190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7462398248055e-05,
+      "loss": 1.3082,
+      "step": 48195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.746189518713707e-05,
+      "loss": 1.1362,
+      "step": 48200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7461392083607705e-05,
+      "loss": 1.1009,
+      "step": 48205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.746088893746978e-05,
+      "loss": 1.3191,
+      "step": 48210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7460385748726163e-05,
+      "loss": 1.1546,
+      "step": 48215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7459882517379728e-05,
+      "loss": 1.3226,
+      "step": 48220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.745937924343336e-05,
+      "loss": 1.239,
+      "step": 48225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7458875926889914e-05,
+      "loss": 1.092,
+      "step": 48230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7458372567752282e-05,
+      "loss": 1.025,
+      "step": 48235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7457869166023333e-05,
+      "loss": 0.9013,
+      "step": 48240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7457365721705933e-05,
+      "loss": 1.3045,
+      "step": 48245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7456862234802966e-05,
+      "loss": 1.1741,
+      "step": 48250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7456358705317308e-05,
+      "loss": 1.2486,
+      "step": 48255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.745585513325183e-05,
+      "loss": 1.0071,
+      "step": 48260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.745535151860941e-05,
+      "loss": 1.4547,
+      "step": 48265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.745484786139292e-05,
+      "loss": 1.0195,
+      "step": 48270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.745434416160524e-05,
+      "loss": 1.181,
+      "step": 48275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7453840419249245e-05,
+      "loss": 1.2467,
+      "step": 48280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7453336634327812e-05,
+      "loss": 1.3953,
+      "step": 48285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.745283280684382e-05,
+      "loss": 1.1726,
+      "step": 48290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7452328936800143e-05,
+      "loss": 1.4555,
+      "step": 48295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.745182502419966e-05,
+      "loss": 1.113,
+      "step": 48300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7451321069045248e-05,
+      "loss": 0.9555,
+      "step": 48305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7450817071339784e-05,
+      "loss": 1.5538,
+      "step": 48310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7450313031086147e-05,
+      "loss": 1.4145,
+      "step": 48315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7449808948287217e-05,
+      "loss": 1.2309,
+      "step": 48320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.744930482294587e-05,
+      "loss": 1.255,
+      "step": 48325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.744880065506499e-05,
+      "loss": 1.1691,
+      "step": 48330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7448296444647445e-05,
+      "loss": 1.0124,
+      "step": 48335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7447792191696123e-05,
+      "loss": 1.3211,
+      "step": 48340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7447287896213905e-05,
+      "loss": 1.1007,
+      "step": 48345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7446783558203668e-05,
+      "loss": 1.1346,
+      "step": 48350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.744627917766829e-05,
+      "loss": 1.1643,
+      "step": 48355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.744577475461065e-05,
+      "loss": 1.2717,
+      "step": 48360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.744527028903363e-05,
+      "loss": 0.969,
+      "step": 48365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7444765780940115e-05,
+      "loss": 1.31,
+      "step": 48370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7444261230332987e-05,
+      "loss": 1.3968,
+      "step": 48375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7443756637215117e-05,
+      "loss": 1.204,
+      "step": 48380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.74432520015894e-05,
+      "loss": 1.2623,
+      "step": 48385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7442747323458708e-05,
+      "loss": 1.2158,
+      "step": 48390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7442242602825924e-05,
+      "loss": 1.1738,
+      "step": 48395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7441737839693933e-05,
+      "loss": 1.4429,
+      "step": 48400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7441233034065614e-05,
+      "loss": 0.9547,
+      "step": 48405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7440728185943855e-05,
+      "loss": 1.1657,
+      "step": 48410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7440223295331534e-05,
+      "loss": 1.7829,
+      "step": 48415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.743971836223154e-05,
+      "loss": 1.625,
+      "step": 48420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.743921338664675e-05,
+      "loss": 1.2924,
+      "step": 48425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.743870836858005e-05,
+      "loss": 1.2599,
+      "step": 48430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7438203308034326e-05,
+      "loss": 1.1175,
+      "step": 48435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7437698205012458e-05,
+      "loss": 1.6229,
+      "step": 48440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7437193059517333e-05,
+      "loss": 1.2148,
+      "step": 48445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7436687871551837e-05,
+      "loss": 1.2304,
+      "step": 48450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7436182641118857e-05,
+      "loss": 1.4662,
+      "step": 48455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7435677368221265e-05,
+      "loss": 0.9643,
+      "step": 48460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7435172052861963e-05,
+      "loss": 1.4191,
+      "step": 48465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.743466669504383e-05,
+      "loss": 1.1611,
+      "step": 48470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.743416129476975e-05,
+      "loss": 1.0813,
+      "step": 48475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7433655852042607e-05,
+      "loss": 1.5205,
+      "step": 48480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7433150366865295e-05,
+      "loss": 1.0693,
+      "step": 48485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7432644839240695e-05,
+      "loss": 1.1148,
+      "step": 48490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7432139269171697e-05,
+      "loss": 1.2838,
+      "step": 48495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7431633656661186e-05,
+      "loss": 1.0484,
+      "step": 48500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7431128001712046e-05,
+      "loss": 1.414,
+      "step": 48505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7430622304327173e-05,
+      "loss": 1.1031,
+      "step": 48510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7430116564509446e-05,
+      "loss": 1.4278,
+      "step": 48515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.742961078226176e-05,
+      "loss": 1.3257,
+      "step": 48520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7429104957587e-05,
+      "loss": 1.5026,
+      "step": 48525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7428599090488054e-05,
+      "loss": 1.2545,
+      "step": 48530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7428093180967814e-05,
+      "loss": 1.6117,
+      "step": 48535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7427587229029164e-05,
+      "loss": 1.3019,
+      "step": 48540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7427081234674997e-05,
+      "loss": 1.3316,
+      "step": 48545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7426575197908203e-05,
+      "loss": 1.2253,
+      "step": 48550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.742606911873167e-05,
+      "loss": 1.3762,
+      "step": 48555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7425562997148284e-05,
+      "loss": 1.0784,
+      "step": 48560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7425056833160943e-05,
+      "loss": 1.2844,
+      "step": 48565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.742455062677253e-05,
+      "loss": 1.5345,
+      "step": 48570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7424044377985944e-05,
+      "loss": 1.4698,
+      "step": 48575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7423538086804073e-05,
+      "loss": 1.4698,
+      "step": 48580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7423031753229802e-05,
+      "loss": 1.3242,
+      "step": 48585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.742252537726603e-05,
+      "loss": 1.1616,
+      "step": 48590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7422018958915645e-05,
+      "loss": 1.2365,
+      "step": 48595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7421512498181542e-05,
+      "loss": 1.568,
+      "step": 48600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7421005995066607e-05,
+      "loss": 1.2219,
+      "step": 48605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.742049944957374e-05,
+      "loss": 1.3974,
+      "step": 48610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7419992861705825e-05,
+      "loss": 0.9685,
+      "step": 48615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7419486231465766e-05,
+      "loss": 1.5169,
+      "step": 48620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7418979558856445e-05,
+      "loss": 1.2689,
+      "step": 48625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.741847284388076e-05,
+      "loss": 1.1014,
+      "step": 48630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7417966086541612e-05,
+      "loss": 1.387,
+      "step": 48635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7417459286841882e-05,
+      "loss": 1.1814,
+      "step": 48640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7416952444784473e-05,
+      "loss": 1.7873,
+      "step": 48645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7416445560372276e-05,
+      "loss": 1.335,
+      "step": 48650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.741593863360818e-05,
+      "loss": 1.1765,
+      "step": 48655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.741543166449509e-05,
+      "loss": 1.1149,
+      "step": 48660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.74149246530359e-05,
+      "loss": 1.3511,
+      "step": 48665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.74144175992335e-05,
+      "loss": 1.1325,
+      "step": 48670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7413910503090784e-05,
+      "loss": 1.4434,
+      "step": 48675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7413403364610655e-05,
+      "loss": 1.2049,
+      "step": 48680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7412896183796005e-05,
+      "loss": 1.0906,
+      "step": 48685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7412388960649732e-05,
+      "loss": 1.047,
+      "step": 48690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.741188169517473e-05,
+      "loss": 1.3203,
+      "step": 48695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.74113743873739e-05,
+      "loss": 1.525,
+      "step": 48700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.741086703725013e-05,
+      "loss": 1.2322,
+      "step": 48705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7410359644806324e-05,
+      "loss": 1.3118,
+      "step": 48710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.740985221004538e-05,
+      "loss": 1.1713,
+      "step": 48715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7409344732970196e-05,
+      "loss": 1.0616,
+      "step": 48720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.740883721358367e-05,
+      "loss": 1.3434,
+      "step": 48725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7408329651888696e-05,
+      "loss": 1.2251,
+      "step": 48730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7407822047888176e-05,
+      "loss": 1.3037,
+      "step": 48735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.740731440158501e-05,
+      "loss": 0.9932,
+      "step": 48740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.740680671298209e-05,
+      "loss": 1.3728,
+      "step": 48745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7406298982082323e-05,
+      "loss": 1.2536,
+      "step": 48750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.74057912088886e-05,
+      "loss": 1.3928,
+      "step": 48755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7405283393403833e-05,
+      "loss": 1.1389,
+      "step": 48760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7404775535630915e-05,
+      "loss": 1.1543,
+      "step": 48765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7404267635572742e-05,
+      "loss": 1.1892,
+      "step": 48770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7403759693232217e-05,
+      "loss": 1.0735,
+      "step": 48775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7403251708612245e-05,
+      "loss": 1.116,
+      "step": 48780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7402743681715724e-05,
+      "loss": 0.9785,
+      "step": 48785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7402235612545554e-05,
+      "loss": 1.1428,
+      "step": 48790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7401727501104638e-05,
+      "loss": 1.144,
+      "step": 48795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.740121934739588e-05,
+      "loss": 1.0586,
+      "step": 48800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7400711151422174e-05,
+      "loss": 1.2567,
+      "step": 48805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.740020291318643e-05,
+      "loss": 1.1283,
+      "step": 48810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7399694632691544e-05,
+      "loss": 1.3733,
+      "step": 48815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7399186309940424e-05,
+      "loss": 1.0448,
+      "step": 48820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7398677944935967e-05,
+      "loss": 1.1781,
+      "step": 48825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7398169537681083e-05,
+      "loss": 1.3182,
+      "step": 48830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7397661088178672e-05,
+      "loss": 1.1042,
+      "step": 48835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7397152596431635e-05,
+      "loss": 1.208,
+      "step": 48840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7396644062442882e-05,
+      "loss": 1.7149,
+      "step": 48845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.739613548621531e-05,
+      "loss": 1.048,
+      "step": 48850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7395626867751825e-05,
+      "loss": 1.1547,
+      "step": 48855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.739511820705534e-05,
+      "loss": 1.3272,
+      "step": 48860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7394609504128744e-05,
+      "loss": 1.4416,
+      "step": 48865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7394100758974955e-05,
+      "loss": 1.2522,
+      "step": 48870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.739359197159687e-05,
+      "loss": 1.1828,
+      "step": 48875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.73930831419974e-05,
+      "loss": 1.1505,
+      "step": 48880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.739257427017945e-05,
+      "loss": 1.0582,
+      "step": 48885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7392065356145926e-05,
+      "loss": 1.1894,
+      "step": 48890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7391556399899732e-05,
+      "loss": 1.1839,
+      "step": 48895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7391047401443776e-05,
+      "loss": 1.0743,
+      "step": 48900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.739053836078096e-05,
+      "loss": 1.1486,
+      "step": 48905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.73900292779142e-05,
+      "loss": 1.7487,
+      "step": 48910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7389520152846396e-05,
+      "loss": 1.4956,
+      "step": 48915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7389010985580456e-05,
+      "loss": 1.4727,
+      "step": 48920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7388501776119292e-05,
+      "loss": 1.1995,
+      "step": 48925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7387992524465808e-05,
+      "loss": 1.1765,
+      "step": 48930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.738748323062291e-05,
+      "loss": 1.6442,
+      "step": 48935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7386973894593513e-05,
+      "loss": 1.2186,
+      "step": 48940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7386464516380523e-05,
+      "loss": 1.815,
+      "step": 48945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7385955095986844e-05,
+      "loss": 1.0651,
+      "step": 48950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.738544563341539e-05,
+      "loss": 1.168,
+      "step": 48955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7384936128669067e-05,
+      "loss": 1.2908,
+      "step": 48960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.738442658175079e-05,
+      "loss": 1.3358,
+      "step": 48965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7383916992663466e-05,
+      "loss": 1.2359,
+      "step": 48970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.738340736141e-05,
+      "loss": 1.3105,
+      "step": 48975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7382897687993314e-05,
+      "loss": 1.1102,
+      "step": 48980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7382387972416306e-05,
+      "loss": 1.1089,
+      "step": 48985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7381878214681893e-05,
+      "loss": 1.175,
+      "step": 48990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7381368414792988e-05,
+      "loss": 1.3367,
+      "step": 48995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7380858572752495e-05,
+      "loss": 1.0598,
+      "step": 49000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.738034868856333e-05,
+      "loss": 1.3129,
+      "step": 49005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7379838762228408e-05,
+      "loss": 1.1552,
+      "step": 49010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7379328793750632e-05,
+      "loss": 1.4653,
+      "step": 49015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7378818783132927e-05,
+      "loss": 1.1181,
+      "step": 49020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7378308730378192e-05,
+      "loss": 1.5155,
+      "step": 49025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7377798635489348e-05,
+      "loss": 1.4722,
+      "step": 49030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7377288498469305e-05,
+      "loss": 1.0584,
+      "step": 49035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7376778319320978e-05,
+      "loss": 1.2479,
+      "step": 49040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7376268098047276e-05,
+      "loss": 1.3756,
+      "step": 49045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7375757834651117e-05,
+      "loss": 1.4017,
+      "step": 49050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.737524752913542e-05,
+      "loss": 1.0634,
+      "step": 49055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7374737181503083e-05,
+      "loss": 1.183,
+      "step": 49060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7374226791757036e-05,
+      "loss": 1.1848,
+      "step": 49065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7373716359900186e-05,
+      "loss": 1.3025,
+      "step": 49070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.737320588593545e-05,
+      "loss": 1.1397,
+      "step": 49075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7372695369865746e-05,
+      "loss": 1.418,
+      "step": 49080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7372184811693983e-05,
+      "loss": 1.2267,
+      "step": 49085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7371674211423077e-05,
+      "loss": 1.3463,
+      "step": 49090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7371163569055947e-05,
+      "loss": 1.0206,
+      "step": 49095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7370652884595513e-05,
+      "loss": 1.1854,
+      "step": 49100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7370142158044682e-05,
+      "loss": 1.0121,
+      "step": 49105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7369631389406378e-05,
+      "loss": 1.2769,
+      "step": 49110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7369120578683515e-05,
+      "loss": 1.2734,
+      "step": 49115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7368609725879006e-05,
+      "loss": 2.0108,
+      "step": 49120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7368098830995776e-05,
+      "loss": 1.0362,
+      "step": 49125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.736758789403674e-05,
+      "loss": 1.3279,
+      "step": 49130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7367076915004813e-05,
+      "loss": 1.0603,
+      "step": 49135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7366565893902914e-05,
+      "loss": 1.5102,
+      "step": 49140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.736605483073396e-05,
+      "loss": 1.3217,
+      "step": 49145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7365543725500873e-05,
+      "loss": 1.1588,
+      "step": 49150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.736503257820657e-05,
+      "loss": 1.2664,
+      "step": 49155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.736452138885397e-05,
+      "loss": 1.29,
+      "step": 49160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.736401015744599e-05,
+      "loss": 1.3284,
+      "step": 49165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.736349888398555e-05,
+      "loss": 1.0913,
+      "step": 49170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7362987568475572e-05,
+      "loss": 1.3628,
+      "step": 49175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.736247621091898e-05,
+      "loss": 1.2833,
+      "step": 49180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7361964811318683e-05,
+      "loss": 1.4793,
+      "step": 49185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.736145336967761e-05,
+      "loss": 1.0659,
+      "step": 49190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7360941885998676e-05,
+      "loss": 1.2907,
+      "step": 49195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7360430360284806e-05,
+      "loss": 1.3863,
+      "step": 49200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7359918792538922e-05,
+      "loss": 1.2976,
+      "step": 49205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7359407182763944e-05,
+      "loss": 1.1958,
+      "step": 49210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.735889553096279e-05,
+      "loss": 1.6036,
+      "step": 49215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7358383837138383e-05,
+      "loss": 1.4274,
+      "step": 49220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.735787210129365e-05,
+      "loss": 1.0153,
+      "step": 49225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7357360323431506e-05,
+      "loss": 0.9809,
+      "step": 49230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.735684850355488e-05,
+      "loss": 1.0877,
+      "step": 49235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7356336641666692e-05,
+      "loss": 1.4037,
+      "step": 49240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7355824737769868e-05,
+      "loss": 1.3559,
+      "step": 49245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7355312791867324e-05,
+      "loss": 1.0837,
+      "step": 49250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7354800803961988e-05,
+      "loss": 1.4044,
+      "step": 49255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7354288774056784e-05,
+      "loss": 1.2681,
+      "step": 49260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.735377670215464e-05,
+      "loss": 1.2922,
+      "step": 49265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.735326458825847e-05,
+      "loss": 1.2709,
+      "step": 49270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7352752432371203e-05,
+      "loss": 1.6146,
+      "step": 49275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.735224023449577e-05,
+      "loss": 1.1138,
+      "step": 49280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.735172799463509e-05,
+      "loss": 1.1345,
+      "step": 49285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7351215712792088e-05,
+      "loss": 1.2815,
+      "step": 49290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.735070338896969e-05,
+      "loss": 1.113,
+      "step": 49295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7350191023170824e-05,
+      "loss": 1.1266,
+      "step": 49300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7349678615398412e-05,
+      "loss": 1.096,
+      "step": 49305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7349166165655386e-05,
+      "loss": 1.3603,
+      "step": 49310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.734865367394466e-05,
+      "loss": 1.1761,
+      "step": 49315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7348141140269177e-05,
+      "loss": 1.1641,
+      "step": 49320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7347628564631854e-05,
+      "loss": 1.4539,
+      "step": 49325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.734711594703562e-05,
+      "loss": 1.3044,
+      "step": 49330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7346603287483398e-05,
+      "loss": 1.216,
+      "step": 49335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7346090585978124e-05,
+      "loss": 1.0518,
+      "step": 49340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.734557784252272e-05,
+      "loss": 1.4173,
+      "step": 49345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7345065057120116e-05,
+      "loss": 1.3902,
+      "step": 49350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.734455222977324e-05,
+      "loss": 1.2001,
+      "step": 49355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7344039360485024e-05,
+      "loss": 1.0766,
+      "step": 49360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.734352644925839e-05,
+      "loss": 1.4974,
+      "step": 49365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.734301349609627e-05,
+      "loss": 1.1941,
+      "step": 49370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7342500501001593e-05,
+      "loss": 1.2951,
+      "step": 49375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.734198746397729e-05,
+      "loss": 1.3601,
+      "step": 49380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.734147438502629e-05,
+      "loss": 1.1547,
+      "step": 49385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7340961264151523e-05,
+      "loss": 1.7866,
+      "step": 49390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7340448101355917e-05,
+      "loss": 1.2684,
+      "step": 49395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7339934896642403e-05,
+      "loss": 1.5122,
+      "step": 49400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7339421650013917e-05,
+      "loss": 1.2427,
+      "step": 49405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7338908361473383e-05,
+      "loss": 1.1329,
+      "step": 49410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7338395031023737e-05,
+      "loss": 1.646,
+      "step": 49415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7337881658667907e-05,
+      "loss": 1.5042,
+      "step": 49420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7337368244408826e-05,
+      "loss": 0.9816,
+      "step": 49425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7336854788249427e-05,
+      "loss": 1.2021,
+      "step": 49430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7336341290192638e-05,
+      "loss": 0.8428,
+      "step": 49435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.73358277502414e-05,
+      "loss": 1.0663,
+      "step": 49440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7335314168398635e-05,
+      "loss": 1.0956,
+      "step": 49445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7334800544667283e-05,
+      "loss": 0.9145,
+      "step": 49450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7334286879050274e-05,
+      "loss": 1.0101,
+      "step": 49455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7333773171550544e-05,
+      "loss": 1.0659,
+      "step": 49460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.733325942217102e-05,
+      "loss": 1.0523,
+      "step": 49465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7332745630914646e-05,
+      "loss": 1.5429,
+      "step": 49470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7332231797784346e-05,
+      "loss": 1.0567,
+      "step": 49475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.733171792278306e-05,
+      "loss": 1.5505,
+      "step": 49480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7331204005913725e-05,
+      "loss": 1.2418,
+      "step": 49485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.733069004717927e-05,
+      "loss": 1.208,
+      "step": 49490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.733017604658263e-05,
+      "loss": 1.1382,
+      "step": 49495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7329662004126745e-05,
+      "loss": 1.1966,
+      "step": 49500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7329147919814543e-05,
+      "loss": 0.956,
+      "step": 49505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.732863379364897e-05,
+      "loss": 1.1957,
+      "step": 49510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7328119625632954e-05,
+      "loss": 1.2211,
+      "step": 49515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7327605415769433e-05,
+      "loss": 1.1687,
+      "step": 49520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7327091164061345e-05,
+      "loss": 1.3884,
+      "step": 49525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7326576870511625e-05,
+      "loss": 1.3872,
+      "step": 49530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7326062535123208e-05,
+      "loss": 1.3033,
+      "step": 49535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7325548157899037e-05,
+      "loss": 1.1167,
+      "step": 49540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7325033738842047e-05,
+      "loss": 1.1256,
+      "step": 49545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7324519277955168e-05,
+      "loss": 1.1409,
+      "step": 49550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.732400477524135e-05,
+      "loss": 1.0461,
+      "step": 49555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7323490230703525e-05,
+      "loss": 1.0462,
+      "step": 49560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.732297564434463e-05,
+      "loss": 1.0951,
+      "step": 49565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7322461016167608e-05,
+      "loss": 1.0962,
+      "step": 49570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7321946346175393e-05,
+      "loss": 1.1477,
+      "step": 49575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7321431634370928e-05,
+      "loss": 1.3709,
+      "step": 49580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7320916880757147e-05,
+      "loss": 1.2957,
+      "step": 49585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7320402085336997e-05,
+      "loss": 1.3194,
+      "step": 49590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.731988724811341e-05,
+      "loss": 1.0585,
+      "step": 49595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7319372369089334e-05,
+      "loss": 1.2374,
+      "step": 49600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.73188574482677e-05,
+      "loss": 1.0854,
+      "step": 49605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.731834248565146e-05,
+      "loss": 1.3098,
+      "step": 49610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7317827481243547e-05,
+      "loss": 1.1956,
+      "step": 49615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7317312435046896e-05,
+      "loss": 1.1046,
+      "step": 49620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7316797347064464e-05,
+      "loss": 1.1321,
+      "step": 49625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7316282217299177e-05,
+      "loss": 1.1117,
+      "step": 49630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.731576704575399e-05,
+      "loss": 1.2625,
+      "step": 49635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7315251832431833e-05,
+      "loss": 1.6459,
+      "step": 49640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.731473657733566e-05,
+      "loss": 1.0276,
+      "step": 49645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.73142212804684e-05,
+      "loss": 1.2949,
+      "step": 49650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7313705941833007e-05,
+      "loss": 1.0369,
+      "step": 49655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7313190561432412e-05,
+      "loss": 1.4221,
+      "step": 49660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7312675139269573e-05,
+      "loss": 1.4561,
+      "step": 49665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7312159675347424e-05,
+      "loss": 1.2658,
+      "step": 49670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.731164416966891e-05,
+      "loss": 1.5852,
+      "step": 49675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7311128622236974e-05,
+      "loss": 1.0637,
+      "step": 49680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7310613033054565e-05,
+      "loss": 1.2799,
+      "step": 49685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.731009740212462e-05,
+      "loss": 1.1202,
+      "step": 49690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.730958172945009e-05,
+      "loss": 0.9221,
+      "step": 49695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7309066015033914e-05,
+      "loss": 1.1946,
+      "step": 49700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.730855025887904e-05,
+      "loss": 1.3448,
+      "step": 49705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7308034460988416e-05,
+      "loss": 1.2876,
+      "step": 49710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7307518621364982e-05,
+      "loss": 1.3305,
+      "step": 49715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7307002740011684e-05,
+      "loss": 1.2117,
+      "step": 49720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7306486816931473e-05,
+      "loss": 1.0211,
+      "step": 49725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7305970852127293e-05,
+      "loss": 1.3073,
+      "step": 49730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7305454845602088e-05,
+      "loss": 1.1144,
+      "step": 49735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7304938797358807e-05,
+      "loss": 1.1062,
+      "step": 49740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.73044227074004e-05,
+      "loss": 1.1394,
+      "step": 49745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.730390657572981e-05,
+      "loss": 1.3763,
+      "step": 49750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.730339040234998e-05,
+      "loss": 1.1537,
+      "step": 49755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7302874187263866e-05,
+      "loss": 1.1944,
+      "step": 49760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7302357930474415e-05,
+      "loss": 1.2084,
+      "step": 49765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.730184163198457e-05,
+      "loss": 1.2088,
+      "step": 49770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.730132529179728e-05,
+      "loss": 1.4612,
+      "step": 49775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7300808909915498e-05,
+      "loss": 1.1976,
+      "step": 49780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.730029248634217e-05,
+      "loss": 1.1142,
+      "step": 49785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7299776021080242e-05,
+      "loss": 1.2605,
+      "step": 49790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.729925951413267e-05,
+      "loss": 1.1864,
+      "step": 49795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7298742965502403e-05,
+      "loss": 1.2077,
+      "step": 49800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7298226375192384e-05,
+      "loss": 1.2853,
+      "step": 49805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.729770974320557e-05,
+      "loss": 1.1402,
+      "step": 49810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7297193069544907e-05,
+      "loss": 1.2073,
+      "step": 49815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7296676354213343e-05,
+      "loss": 1.1786,
+      "step": 49820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7296159597213837e-05,
+      "loss": 1.1655,
+      "step": 49825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7295642798549336e-05,
+      "loss": 1.2166,
+      "step": 49830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7295125958222788e-05,
+      "loss": 1.461,
+      "step": 49835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7294609076237146e-05,
+      "loss": 1.119,
+      "step": 49840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7294092152595366e-05,
+      "loss": 0.9639,
+      "step": 49845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.72935751873004e-05,
+      "loss": 1.0497,
+      "step": 49850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.729305818035519e-05,
+      "loss": 1.2946,
+      "step": 49855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7292541131762693e-05,
+      "loss": 1.29,
+      "step": 49860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.729202404152587e-05,
+      "loss": 1.642,
+      "step": 49865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7291506909647668e-05,
+      "loss": 1.1668,
+      "step": 49870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7290989736131036e-05,
+      "loss": 1.7029,
+      "step": 49875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7290472520978934e-05,
+      "loss": 1.1034,
+      "step": 49880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7289955264194312e-05,
+      "loss": 1.3267,
+      "step": 49885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.728943796578012e-05,
+      "loss": 1.2312,
+      "step": 49890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7288920625739326e-05,
+      "loss": 1.5799,
+      "step": 49895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7288403244074868e-05,
+      "loss": 1.3745,
+      "step": 49900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7287885820789708e-05,
+      "loss": 1.2204,
+      "step": 49905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.72873683558868e-05,
+      "loss": 1.3267,
+      "step": 49910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.72868508493691e-05,
+      "loss": 1.2325,
+      "step": 49915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7286333301239568e-05,
+      "loss": 0.9492,
+      "step": 49920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7285815711501148e-05,
+      "loss": 1.1698,
+      "step": 49925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7285298080156798e-05,
+      "loss": 1.1764,
+      "step": 49930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7284780407209483e-05,
+      "loss": 1.341,
+      "step": 49935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.728426269266215e-05,
+      "loss": 1.2592,
+      "step": 49940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7283744936517766e-05,
+      "loss": 1.327,
+      "step": 49945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7283227138779275e-05,
+      "loss": 1.1681,
+      "step": 49950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.728270929944964e-05,
+      "loss": 0.9486,
+      "step": 49955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.728219141853182e-05,
+      "loss": 1.3442,
+      "step": 49960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.728167349602877e-05,
+      "loss": 1.2116,
+      "step": 49965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7281155531943447e-05,
+      "loss": 1.169,
+      "step": 49970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.728063752627881e-05,
+      "loss": 1.0806,
+      "step": 49975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7280119479037816e-05,
+      "loss": 0.9475,
+      "step": 49980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7279601390223426e-05,
+      "loss": 1.597,
+      "step": 49985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7279083259838594e-05,
+      "loss": 1.5552,
+      "step": 49990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7278565087886284e-05,
+      "loss": 1.3767,
+      "step": 49995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.727804687436945e-05,
+      "loss": 1.4831,
+      "step": 50000
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 1.2274746894836426,
+      "eval_runtime": 6470.275,
+      "eval_samples_per_second": 3.572,
+      "eval_steps_per_second": 1.786,
+      "step": 50000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7277528619291055e-05,
+      "loss": 1.067,
+      "step": 50005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.727701032265406e-05,
+      "loss": 1.133,
+      "step": 50010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.727649198446142e-05,
+      "loss": 1.3225,
+      "step": 50015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7275973604716094e-05,
+      "loss": 1.0013,
+      "step": 50020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.727545518342105e-05,
+      "loss": 1.3698,
+      "step": 50025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.727493672057924e-05,
+      "loss": 1.2925,
+      "step": 50030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.727441821619363e-05,
+      "loss": 1.1305,
+      "step": 50035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.727389967026718e-05,
+      "loss": 1.0998,
+      "step": 50040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7273381082802852e-05,
+      "loss": 1.3885,
+      "step": 50045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7272862453803605e-05,
+      "loss": 1.1185,
+      "step": 50050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.72723437832724e-05,
+      "loss": 1.4008,
+      "step": 50055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7271825071212202e-05,
+      "loss": 1.2491,
+      "step": 50060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7271306317625973e-05,
+      "loss": 1.1326,
+      "step": 50065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7270787522516673e-05,
+      "loss": 1.3644,
+      "step": 50070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7270268685887262e-05,
+      "loss": 1.1502,
+      "step": 50075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.726974980774071e-05,
+      "loss": 1.5105,
+      "step": 50080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7269230888079977e-05,
+      "loss": 1.1974,
+      "step": 50085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7268711926908025e-05,
+      "loss": 1.3136,
+      "step": 50090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.726819292422782e-05,
+      "loss": 1.1406,
+      "step": 50095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7267673880042323e-05,
+      "loss": 1.1288,
+      "step": 50100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7267154794354498e-05,
+      "loss": 1.4358,
+      "step": 50105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7266635667167314e-05,
+      "loss": 1.0893,
+      "step": 50110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.726611649848373e-05,
+      "loss": 1.5343,
+      "step": 50115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.726559728830671e-05,
+      "loss": 1.205,
+      "step": 50120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7265078036639225e-05,
+      "loss": 1.1289,
+      "step": 50125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7264558743484232e-05,
+      "loss": 1.6094,
+      "step": 50130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7264039408844707e-05,
+      "loss": 1.2519,
+      "step": 50135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7263520032723606e-05,
+      "loss": 1.1294,
+      "step": 50140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7263000615123902e-05,
+      "loss": 1.6399,
+      "step": 50145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7262481156048555e-05,
+      "loss": 1.2478,
+      "step": 50150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7261961655500538e-05,
+      "loss": 1.2802,
+      "step": 50155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7261442113482813e-05,
+      "loss": 1.0797,
+      "step": 50160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7260922529998344e-05,
+      "loss": 1.2816,
+      "step": 50165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7260402905050107e-05,
+      "loss": 1.1352,
+      "step": 50170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7259883238641064e-05,
+      "loss": 1.2269,
+      "step": 50175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.725936353077418e-05,
+      "loss": 1.2689,
+      "step": 50180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.725884378145243e-05,
+      "loss": 1.0777,
+      "step": 50185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7258323990678775e-05,
+      "loss": 1.3905,
+      "step": 50190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7257804158456185e-05,
+      "loss": 0.9481,
+      "step": 50195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7257284284787633e-05,
+      "loss": 0.9101,
+      "step": 50200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.725676436967608e-05,
+      "loss": 1.3976,
+      "step": 50205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.72562444131245e-05,
+      "loss": 1.0591,
+      "step": 50210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7255724415135866e-05,
+      "loss": 1.4046,
+      "step": 50215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.725520437571314e-05,
+      "loss": 1.1758,
+      "step": 50220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7254684294859296e-05,
+      "loss": 1.1104,
+      "step": 50225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7254164172577305e-05,
+      "loss": 1.1256,
+      "step": 50230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.725364400887013e-05,
+      "loss": 1.3559,
+      "step": 50235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7253123803740746e-05,
+      "loss": 1.0287,
+      "step": 50240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7252603557192126e-05,
+      "loss": 1.2747,
+      "step": 50245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.725208326922724e-05,
+      "loss": 1.2191,
+      "step": 50250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7251562939849058e-05,
+      "loss": 1.3348,
+      "step": 50255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.725104256906055e-05,
+      "loss": 1.1279,
+      "step": 50260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7250522156864688e-05,
+      "loss": 0.9784,
+      "step": 50265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7250001703264448e-05,
+      "loss": 1.133,
+      "step": 50270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7249481208262796e-05,
+      "loss": 1.4786,
+      "step": 50275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7248960671862706e-05,
+      "loss": 1.0107,
+      "step": 50280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7248440094067154e-05,
+      "loss": 1.3326,
+      "step": 50285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.724791947487911e-05,
+      "loss": 1.2156,
+      "step": 50290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.724739881430155e-05,
+      "loss": 1.1728,
+      "step": 50295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7246878112337442e-05,
+      "loss": 1.219,
+      "step": 50300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7246357368989765e-05,
+      "loss": 1.2924,
+      "step": 50305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7245836584261482e-05,
+      "loss": 1.0691,
+      "step": 50310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7245315758155583e-05,
+      "loss": 1.1631,
+      "step": 50315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7244794890675034e-05,
+      "loss": 1.1224,
+      "step": 50320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7244273981822804e-05,
+      "loss": 1.3026,
+      "step": 50325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7243753031601878e-05,
+      "loss": 1.3003,
+      "step": 50330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7243232040015224e-05,
+      "loss": 1.2623,
+      "step": 50335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7242711007065824e-05,
+      "loss": 1.2504,
+      "step": 50340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.724218993275664e-05,
+      "loss": 1.425,
+      "step": 50345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.724166881709066e-05,
+      "loss": 1.1525,
+      "step": 50350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.724114766007086e-05,
+      "loss": 1.4414,
+      "step": 50355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7240626461700207e-05,
+      "loss": 1.6218,
+      "step": 50360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7240105221981685e-05,
+      "loss": 1.0414,
+      "step": 50365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.723958394091827e-05,
+      "loss": 1.6485,
+      "step": 50370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7239062618512933e-05,
+      "loss": 2.0716,
+      "step": 50375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7238541254768655e-05,
+      "loss": 1.2966,
+      "step": 50380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7238019849688418e-05,
+      "loss": 0.9231,
+      "step": 50385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.723749840327519e-05,
+      "loss": 1.2971,
+      "step": 50390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.723697691553195e-05,
+      "loss": 1.2885,
+      "step": 50395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.723645538646169e-05,
+      "loss": 1.0595,
+      "step": 50400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.723593381606737e-05,
+      "loss": 1.2251,
+      "step": 50405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7235412204351978e-05,
+      "loss": 1.1537,
+      "step": 50410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7234890551318488e-05,
+      "loss": 1.1658,
+      "step": 50415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.723436885696989e-05,
+      "loss": 1.107,
+      "step": 50420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7233847121309148e-05,
+      "loss": 1.2508,
+      "step": 50425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7233325344339247e-05,
+      "loss": 1.409,
+      "step": 50430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7232803526063168e-05,
+      "loss": 1.2453,
+      "step": 50435
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7232281666483892e-05,
+      "loss": 1.3803,
+      "step": 50440
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.72317597656044e-05,
+      "loss": 1.3399,
+      "step": 50445
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.723123782342767e-05,
+      "loss": 1.0308,
+      "step": 50450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.723071583995668e-05,
+      "loss": 1.313,
+      "step": 50455
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7230193815194413e-05,
+      "loss": 1.2946,
+      "step": 50460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7229671749143856e-05,
+      "loss": 1.1949,
+      "step": 50465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.722914964180798e-05,
+      "loss": 1.073,
+      "step": 50470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.722862749318977e-05,
+      "loss": 1.0948,
+      "step": 50475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.722810530329221e-05,
+      "loss": 1.1528,
+      "step": 50480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7227583072118284e-05,
+      "loss": 1.5082,
+      "step": 50485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.722706079967097e-05,
+      "loss": 1.2282,
+      "step": 50490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.722653848595325e-05,
+      "loss": 1.1545,
+      "step": 50495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7226016130968114e-05,
+      "loss": 1.1262,
+      "step": 50500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7225493734718534e-05,
+      "loss": 1.1164,
+      "step": 50505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.72249712972075e-05,
+      "loss": 1.2101,
+      "step": 50510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7224448818437994e-05,
+      "loss": 1.122,
+      "step": 50515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7223926298412998e-05,
+      "loss": 1.0934,
+      "step": 50520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.72234037371355e-05,
+      "loss": 1.1889,
+      "step": 50525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.722288113460848e-05,
+      "loss": 1.1044,
+      "step": 50530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7222358490834923e-05,
+      "loss": 1.4435,
+      "step": 50535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7221835805817816e-05,
+      "loss": 1.1429,
+      "step": 50540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7221313079560144e-05,
+      "loss": 1.6474,
+      "step": 50545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7220790312064887e-05,
+      "loss": 0.995,
+      "step": 50550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7220267503335036e-05,
+      "loss": 1.1829,
+      "step": 50555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7219744653373573e-05,
+      "loss": 1.353,
+      "step": 50560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7219221762183482e-05,
+      "loss": 1.0184,
+      "step": 50565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7218698829767756e-05,
+      "loss": 1.0592,
+      "step": 50570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7218175856129375e-05,
+      "loss": 1.24,
+      "step": 50575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.721765284127133e-05,
+      "loss": 1.3093,
+      "step": 50580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7217129785196602e-05,
+      "loss": 1.0443,
+      "step": 50585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7216606687908183e-05,
+      "loss": 1.1266,
+      "step": 50590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.721608354940906e-05,
+      "loss": 1.3422,
+      "step": 50595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7215560369702213e-05,
+      "loss": 1.1025,
+      "step": 50600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7215037148790643e-05,
+      "loss": 1.2924,
+      "step": 50605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7214513886677323e-05,
+      "loss": 1.1629,
+      "step": 50610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7213990583365256e-05,
+      "loss": 1.1342,
+      "step": 50615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7213467238857416e-05,
+      "loss": 1.1811,
+      "step": 50620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7212943853156802e-05,
+      "loss": 0.9977,
+      "step": 50625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.72124204262664e-05,
+      "loss": 1.2666,
+      "step": 50630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7211896958189192e-05,
+      "loss": 1.278,
+      "step": 50635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7211373448928176e-05,
+      "loss": 1.1013,
+      "step": 50640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7210849898486342e-05,
+      "loss": 1.249,
+      "step": 50645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7210326306866675e-05,
+      "loss": 1.1689,
+      "step": 50650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7209802674072168e-05,
+      "loss": 1.3662,
+      "step": 50655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.720927900010581e-05,
+      "loss": 1.3071,
+      "step": 50660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7208755284970588e-05,
+      "loss": 1.4414,
+      "step": 50665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.72082315286695e-05,
+      "loss": 1.0799,
+      "step": 50670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.720770773120553e-05,
+      "loss": 1.1201,
+      "step": 50675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7207183892581673e-05,
+      "loss": 1.2414,
+      "step": 50680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.720666001280092e-05,
+      "loss": 1.3208,
+      "step": 50685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7206136091866262e-05,
+      "loss": 1.0509,
+      "step": 50690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7205612129780694e-05,
+      "loss": 1.043,
+      "step": 50695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.72050881265472e-05,
+      "loss": 1.1091,
+      "step": 50700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7204564082168782e-05,
+      "loss": 1.4091,
+      "step": 50705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7204039996648426e-05,
+      "loss": 1.243,
+      "step": 50710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7203515869989126e-05,
+      "loss": 1.1913,
+      "step": 50715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7202991702193877e-05,
+      "loss": 1.1995,
+      "step": 50720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7202467493265672e-05,
+      "loss": 1.1839,
+      "step": 50725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7201943243207503e-05,
+      "loss": 1.4085,
+      "step": 50730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7201418952022363e-05,
+      "loss": 1.1891,
+      "step": 50735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7200894619713252e-05,
+      "loss": 1.335,
+      "step": 50740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7200370246283154e-05,
+      "loss": 1.545,
+      "step": 50745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7199845831735072e-05,
+      "loss": 1.4182,
+      "step": 50750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7199321376071997e-05,
+      "loss": 1.4228,
+      "step": 50755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7198796879296925e-05,
+      "loss": 1.3039,
+      "step": 50760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7198272341412853e-05,
+      "loss": 1.3477,
+      "step": 50765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.719774776242277e-05,
+      "loss": 1.361,
+      "step": 50770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7197223142329682e-05,
+      "loss": 1.0636,
+      "step": 50775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7196698481136573e-05,
+      "loss": 0.8565,
+      "step": 50780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.719617377884645e-05,
+      "loss": 1.1326,
+      "step": 50785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.71956490354623e-05,
+      "loss": 1.369,
+      "step": 50790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7195124250987124e-05,
+      "loss": 1.1305,
+      "step": 50795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7194599425423924e-05,
+      "loss": 1.1033,
+      "step": 50800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7194074558775686e-05,
+      "loss": 1.5845,
+      "step": 50805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7193549651045415e-05,
+      "loss": 1.2213,
+      "step": 50810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7193024702236102e-05,
+      "loss": 1.1697,
+      "step": 50815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7192499712350757e-05,
+      "loss": 1.3318,
+      "step": 50820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7191974681392364e-05,
+      "loss": 0.9004,
+      "step": 50825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7191449609363932e-05,
+      "loss": 1.1951,
+      "step": 50830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.719092449626845e-05,
+      "loss": 1.5005,
+      "step": 50835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7190399342108924e-05,
+      "loss": 1.4916,
+      "step": 50840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7189874146888352e-05,
+      "loss": 1.0905,
+      "step": 50845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.718934891060973e-05,
+      "loss": 1.5615,
+      "step": 50850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.718882363327606e-05,
+      "loss": 1.3047,
+      "step": 50855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7188298314890337e-05,
+      "loss": 1.0381,
+      "step": 50860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.718777295545557e-05,
+      "loss": 1.6078,
+      "step": 50865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.718724755497475e-05,
+      "loss": 1.26,
+      "step": 50870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7186722113450878e-05,
+      "loss": 1.3696,
+      "step": 50875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.718619663088696e-05,
+      "loss": 1.5705,
+      "step": 50880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7185671107285994e-05,
+      "loss": 0.956,
+      "step": 50885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7185145542650982e-05,
+      "loss": 1.153,
+      "step": 50890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7184619936984926e-05,
+      "loss": 1.6245,
+      "step": 50895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7184094290290823e-05,
+      "loss": 1.1684,
+      "step": 50900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7183568602571678e-05,
+      "loss": 1.3529,
+      "step": 50905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7183042873830494e-05,
+      "loss": 1.1974,
+      "step": 50910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.718251710407027e-05,
+      "loss": 1.1976,
+      "step": 50915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.718199129329401e-05,
+      "loss": 1.1566,
+      "step": 50920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7181465441504718e-05,
+      "loss": 1.2334,
+      "step": 50925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7180939548705395e-05,
+      "loss": 1.6741,
+      "step": 50930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7180413614899046e-05,
+      "loss": 1.2994,
+      "step": 50935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7179887640088676e-05,
+      "loss": 1.0198,
+      "step": 50940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.717936162427728e-05,
+      "loss": 1.6488,
+      "step": 50945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.717883556746787e-05,
+      "loss": 1.1298,
+      "step": 50950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7178309469663448e-05,
+      "loss": 1.0731,
+      "step": 50955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.717778333086702e-05,
+      "loss": 1.2364,
+      "step": 50960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7177257151081585e-05,
+      "loss": 1.3326,
+      "step": 50965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7176730930310153e-05,
+      "loss": 1.279,
+      "step": 50970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7176204668555727e-05,
+      "loss": 1.4389,
+      "step": 50975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7175678365821314e-05,
+      "loss": 1.3355,
+      "step": 50980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7175152022109918e-05,
+      "loss": 1.1405,
+      "step": 50985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7174625637424543e-05,
+      "loss": 1.098,
+      "step": 50990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7174099211768197e-05,
+      "loss": 1.2705,
+      "step": 50995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7173572745143887e-05,
+      "loss": 1.1875,
+      "step": 51000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7173046237554617e-05,
+      "loss": 1.1282,
+      "step": 51005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7172519689003397e-05,
+      "loss": 1.1144,
+      "step": 51010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.717199309949323e-05,
+      "loss": 1.5569,
+      "step": 51015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7171466469027124e-05,
+      "loss": 1.3934,
+      "step": 51020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.717093979760809e-05,
+      "loss": 0.9175,
+      "step": 51025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7170413085239134e-05,
+      "loss": 1.2604,
+      "step": 51030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.716988633192326e-05,
+      "loss": 1.2954,
+      "step": 51035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7169359537663477e-05,
+      "loss": 1.138,
+      "step": 51040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.71688327024628e-05,
+      "loss": 0.9872,
+      "step": 51045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7168305826324228e-05,
+      "loss": 1.1966,
+      "step": 51050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7167778909250775e-05,
+      "loss": 1.3431,
+      "step": 51055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.716725195124545e-05,
+      "loss": 1.2908,
+      "step": 51060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.716672495231126e-05,
+      "loss": 1.3077,
+      "step": 51065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.716619791245122e-05,
+      "loss": 1.0679,
+      "step": 51070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.716567083166833e-05,
+      "loss": 1.1658,
+      "step": 51075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.716514370996561e-05,
+      "loss": 1.2557,
+      "step": 51080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7164616547346065e-05,
+      "loss": 1.212,
+      "step": 51085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7164089343812703e-05,
+      "loss": 1.0408,
+      "step": 51090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.716356209936854e-05,
+      "loss": 1.4215,
+      "step": 51095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7163034814016587e-05,
+      "loss": 1.0593,
+      "step": 51100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7162507487759846e-05,
+      "loss": 1.1792,
+      "step": 51105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.716198012060134e-05,
+      "loss": 1.469,
+      "step": 51110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.716145271254407e-05,
+      "loss": 1.4479,
+      "step": 51115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7160925263591058e-05,
+      "loss": 1.5281,
+      "step": 51120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7160397773745313e-05,
+      "loss": 1.2268,
+      "step": 51125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7159870243009842e-05,
+      "loss": 1.374,
+      "step": 51130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7159342671387664e-05,
+      "loss": 1.3293,
+      "step": 51135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7158815058881787e-05,
+      "loss": 1.0601,
+      "step": 51140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7158287405495223e-05,
+      "loss": 1.1351,
+      "step": 51145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.715775971123099e-05,
+      "loss": 1.1563,
+      "step": 51150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.71572319760921e-05,
+      "loss": 1.6299,
+      "step": 51155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.715670420008156e-05,
+      "loss": 1.6189,
+      "step": 51160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.71561763832024e-05,
+      "loss": 1.437,
+      "step": 51165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7155648525457617e-05,
+      "loss": 1.2804,
+      "step": 51170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7155120626850235e-05,
+      "loss": 1.3721,
+      "step": 51175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7154592687383264e-05,
+      "loss": 1.4338,
+      "step": 51180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.715406470705972e-05,
+      "loss": 1.595,
+      "step": 51185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.715353668588262e-05,
+      "loss": 1.179,
+      "step": 51190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.715300862385498e-05,
+      "loss": 0.9894,
+      "step": 51195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7152480520979812e-05,
+      "loss": 1.2927,
+      "step": 51200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7151952377260133e-05,
+      "loss": 1.2357,
+      "step": 51205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7151424192698958e-05,
+      "loss": 1.25,
+      "step": 51210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7150895967299305e-05,
+      "loss": 1.105,
+      "step": 51215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.715036770106419e-05,
+      "loss": 1.7062,
+      "step": 51220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.714983939399663e-05,
+      "loss": 1.3241,
+      "step": 51225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.714931104609964e-05,
+      "loss": 1.2752,
+      "step": 51230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.714878265737624e-05,
+      "loss": 1.4171,
+      "step": 51235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7148254227829448e-05,
+      "loss": 1.1308,
+      "step": 51240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7147725757462276e-05,
+      "loss": 1.5068,
+      "step": 51245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.714719724627775e-05,
+      "loss": 1.6335,
+      "step": 51250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7146668694278882e-05,
+      "loss": 1.8439,
+      "step": 51255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.714614010146869e-05,
+      "loss": 1.3034,
+      "step": 51260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7145611467850196e-05,
+      "loss": 1.2009,
+      "step": 51265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7145082793426414e-05,
+      "loss": 1.0859,
+      "step": 51270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.714455407820037e-05,
+      "loss": 1.5246,
+      "step": 51275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.714402532217508e-05,
+      "loss": 1.1245,
+      "step": 51280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.714349652535356e-05,
+      "loss": 1.1467,
+      "step": 51285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7142967687738835e-05,
+      "loss": 1.1813,
+      "step": 51290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7142438809333917e-05,
+      "loss": 1.1968,
+      "step": 51295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.714190989014184e-05,
+      "loss": 1.1273,
+      "step": 51300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7141380930165613e-05,
+      "loss": 1.0951,
+      "step": 51305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.714085192940826e-05,
+      "loss": 1.1169,
+      "step": 51310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.71403228878728e-05,
+      "loss": 1.268,
+      "step": 51315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7139793805562254e-05,
+      "loss": 1.1843,
+      "step": 51320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7139264682479652e-05,
+      "loss": 1.1845,
+      "step": 51325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7138735518628006e-05,
+      "loss": 1.5627,
+      "step": 51330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.713820631401034e-05,
+      "loss": 1.4597,
+      "step": 51335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7137677068629673e-05,
+      "loss": 1.1225,
+      "step": 51340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7137147782489038e-05,
+      "loss": 1.3347,
+      "step": 51345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7136618455591443e-05,
+      "loss": 1.3845,
+      "step": 51350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7136089087939923e-05,
+      "loss": 1.2376,
+      "step": 51355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7135559679537498e-05,
+      "loss": 1.1401,
+      "step": 51360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7135030230387185e-05,
+      "loss": 1.2342,
+      "step": 51365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.713450074049201e-05,
+      "loss": 1.2953,
+      "step": 51370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7133971209855004e-05,
+      "loss": 1.2458,
+      "step": 51375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.713344163847918e-05,
+      "loss": 1.3271,
+      "step": 51380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7132912026367573e-05,
+      "loss": 1.3903,
+      "step": 51385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.71323823735232e-05,
+      "loss": 1.105,
+      "step": 51390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7131852679949085e-05,
+      "loss": 1.2992,
+      "step": 51395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.713132294564826e-05,
+      "loss": 1.0379,
+      "step": 51400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.713079317062374e-05,
+      "loss": 1.3865,
+      "step": 51405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.713026335487856e-05,
+      "loss": 1.0834,
+      "step": 51410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7129733498415738e-05,
+      "loss": 1.2596,
+      "step": 51415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7129203601238304e-05,
+      "loss": 1.4678,
+      "step": 51420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7128673663349286e-05,
+      "loss": 1.4241,
+      "step": 51425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7128143684751705e-05,
+      "loss": 1.5086,
+      "step": 51430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.712761366544859e-05,
+      "loss": 1.9201,
+      "step": 51435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7127083605442963e-05,
+      "loss": 1.1854,
+      "step": 51440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.712655350473786e-05,
+      "loss": 1.5793,
+      "step": 51445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7126023363336305e-05,
+      "loss": 1.3792,
+      "step": 51450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.712549318124132e-05,
+      "loss": 1.679,
+      "step": 51455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7124962958455936e-05,
+      "loss": 1.0765,
+      "step": 51460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7124432694983183e-05,
+      "loss": 1.0444,
+      "step": 51465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.712390239082609e-05,
+      "loss": 1.1718,
+      "step": 51470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.712337204598768e-05,
+      "loss": 1.1355,
+      "step": 51475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7122841660470983e-05,
+      "loss": 1.662,
+      "step": 51480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.712231123427903e-05,
+      "loss": 1.1877,
+      "step": 51485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.712178076741485e-05,
+      "loss": 1.1147,
+      "step": 51490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7121250259881473e-05,
+      "loss": 1.6075,
+      "step": 51495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7120719711681923e-05,
+      "loss": 1.169,
+      "step": 51500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7120189122819238e-05,
+      "loss": 1.6182,
+      "step": 51505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7119658493296437e-05,
+      "loss": 1.9683,
+      "step": 51510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7119127823116563e-05,
+      "loss": 1.3574,
+      "step": 51515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7118597112282638e-05,
+      "loss": 1.2679,
+      "step": 51520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7118066360797697e-05,
+      "loss": 1.1321,
+      "step": 51525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7117535568664764e-05,
+      "loss": 0.9978,
+      "step": 51530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7117004735886876e-05,
+      "loss": 1.5586,
+      "step": 51535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7116473862467064e-05,
+      "loss": 1.2116,
+      "step": 51540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.711594294840836e-05,
+      "loss": 0.9523,
+      "step": 51545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.711541199371379e-05,
+      "loss": 1.3544,
+      "step": 51550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7114880998386395e-05,
+      "loss": 1.1168,
+      "step": 51555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.71143499624292e-05,
+      "loss": 1.2849,
+      "step": 51560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.711381888584524e-05,
+      "loss": 1.3145,
+      "step": 51565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7113287768637553e-05,
+      "loss": 1.6531,
+      "step": 51570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7112756610809163e-05,
+      "loss": 1.1381,
+      "step": 51575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7112225412363107e-05,
+      "loss": 1.472,
+      "step": 51580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.711169417330242e-05,
+      "loss": 1.2597,
+      "step": 51585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.711116289363013e-05,
+      "loss": 1.1851,
+      "step": 51590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.711063157334928e-05,
+      "loss": 1.2405,
+      "step": 51595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7110100212462898e-05,
+      "loss": 1.0731,
+      "step": 51600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7109568810974018e-05,
+      "loss": 1.4476,
+      "step": 51605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7109037368885677e-05,
+      "loss": 1.3614,
+      "step": 51610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7108505886200914e-05,
+      "loss": 1.5878,
+      "step": 51615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7107974362922754e-05,
+      "loss": 1.0765,
+      "step": 51620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7107442799054236e-05,
+      "loss": 1.4227,
+      "step": 51625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7106911194598397e-05,
+      "loss": 1.3375,
+      "step": 51630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7106379549558275e-05,
+      "loss": 1.1728,
+      "step": 51635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7105847863936903e-05,
+      "loss": 1.1845,
+      "step": 51640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.710531613773732e-05,
+      "loss": 1.1864,
+      "step": 51645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7104784370962554e-05,
+      "loss": 1.0905,
+      "step": 51650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7104252563615652e-05,
+      "loss": 1.2156,
+      "step": 51655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7103720715699647e-05,
+      "loss": 1.2085,
+      "step": 51660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7103188827217577e-05,
+      "loss": 1.6611,
+      "step": 51665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7102656898172474e-05,
+      "loss": 1.1622,
+      "step": 51670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7102124928567386e-05,
+      "loss": 1.5269,
+      "step": 51675
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.710159291840534e-05,
+      "loss": 1.2212,
+      "step": 51680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7101060867689382e-05,
+      "loss": 1.0388,
+      "step": 51685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7100528776422546e-05,
+      "loss": 1.1398,
+      "step": 51690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7099996644607873e-05,
+      "loss": 1.5294,
+      "step": 51695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7099464472248396e-05,
+      "loss": 1.1802,
+      "step": 51700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7098932259347164e-05,
+      "loss": 0.9433,
+      "step": 51705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7098400005907207e-05,
+      "loss": 1.028,
+      "step": 51710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.709786771193157e-05,
+      "loss": 1.2482,
+      "step": 51715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7097335377423287e-05,
+      "loss": 1.3289,
+      "step": 51720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7096803002385404e-05,
+      "loss": 1.1318,
+      "step": 51725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.709627058682096e-05,
+      "loss": 1.1843,
+      "step": 51730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7095738130732993e-05,
+      "loss": 1.2343,
+      "step": 51735
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7095205634124547e-05,
+      "loss": 1.1894,
+      "step": 51740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7094673096998656e-05,
+      "loss": 1.0834,
+      "step": 51745
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.709414051935837e-05,
+      "loss": 1.2649,
+      "step": 51750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7093607901206723e-05,
+      "loss": 1.4133,
+      "step": 51755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7093075242546762e-05,
+      "loss": 1.1638,
+      "step": 51760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7092542543381527e-05,
+      "loss": 1.33,
+      "step": 51765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7092009803714056e-05,
+      "loss": 1.133,
+      "step": 51770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7091477023547395e-05,
+      "loss": 1.4792,
+      "step": 51775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7090944202884587e-05,
+      "loss": 1.3433,
+      "step": 51780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7090411341728678e-05,
+      "loss": 0.9821,
+      "step": 51785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.70898784400827e-05,
+      "loss": 1.1596,
+      "step": 51790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7089345497949706e-05,
+      "loss": 1.6367,
+      "step": 51795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7088812515332732e-05,
+      "loss": 1.3371,
+      "step": 51800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7088279492234833e-05,
+      "loss": 1.2044,
+      "step": 51805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7087746428659043e-05,
+      "loss": 1.3571,
+      "step": 51810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7087213324608406e-05,
+      "loss": 1.3952,
+      "step": 51815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7086680180085973e-05,
+      "loss": 1.4795,
+      "step": 51820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7086146995094782e-05,
+      "loss": 1.4003,
+      "step": 51825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.708561376963788e-05,
+      "loss": 1.2172,
+      "step": 51830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7085080503718316e-05,
+      "loss": 0.8644,
+      "step": 51835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7084547197339128e-05,
+      "loss": 1.3224,
+      "step": 51840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7084013850503363e-05,
+      "loss": 1.1178,
+      "step": 51845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7083480463214074e-05,
+      "loss": 1.2396,
+      "step": 51850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7082947035474296e-05,
+      "loss": 1.1489,
+      "step": 51855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7082413567287086e-05,
+      "loss": 0.8941,
+      "step": 51860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7081880058655485e-05,
+      "loss": 1.1609,
+      "step": 51865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7081346509582537e-05,
+      "loss": 1.3116,
+      "step": 51870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.708081292007129e-05,
+      "loss": 1.114,
+      "step": 51875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.70802792901248e-05,
+      "loss": 1.0289,
+      "step": 51880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.70797456197461e-05,
+      "loss": 1.346,
+      "step": 51885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.707921190893825e-05,
+      "loss": 1.2083,
+      "step": 51890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.707867815770429e-05,
+      "loss": 1.193,
+      "step": 51895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.707814436604727e-05,
+      "loss": 1.4652,
+      "step": 51900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.707761053397024e-05,
+      "loss": 1.2007,
+      "step": 51905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7077076661476245e-05,
+      "loss": 1.165,
+      "step": 51910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7076542748568335e-05,
+      "loss": 1.2091,
+      "step": 51915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7076008795249563e-05,
+      "loss": 0.9223,
+      "step": 51920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7075474801522977e-05,
+      "loss": 1.1267,
+      "step": 51925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.707494076739162e-05,
+      "loss": 1.0345,
+      "step": 51930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7074406692858545e-05,
+      "loss": 1.1726,
+      "step": 51935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7073872577926805e-05,
+      "loss": 1.1405,
+      "step": 51940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7073338422599447e-05,
+      "loss": 1.0977,
+      "step": 51945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7072804226879524e-05,
+      "loss": 1.0712,
+      "step": 51950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7072269990770082e-05,
+      "loss": 1.1811,
+      "step": 51955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7071735714274177e-05,
+      "loss": 1.0806,
+      "step": 51960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7071201397394858e-05,
+      "loss": 1.1566,
+      "step": 51965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7070667040135174e-05,
+      "loss": 1.2664,
+      "step": 51970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.707013264249818e-05,
+      "loss": 1.29,
+      "step": 51975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7069598204486924e-05,
+      "loss": 1.0291,
+      "step": 51980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.706906372610446e-05,
+      "loss": 1.1419,
+      "step": 51985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7068529207353842e-05,
+      "loss": 1.1564,
+      "step": 51990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7067994648238116e-05,
+      "loss": 1.2423,
+      "step": 51995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7067460048760342e-05,
+      "loss": 1.117,
+      "step": 52000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7066925408923572e-05,
+      "loss": 1.5322,
+      "step": 52005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7066390728730856e-05,
+      "loss": 1.4372,
+      "step": 52010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7065856008185244e-05,
+      "loss": 1.0737,
+      "step": 52015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.70653212472898e-05,
+      "loss": 1.1537,
+      "step": 52020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7064786446047565e-05,
+      "loss": 1.4172,
+      "step": 52025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7064251604461605e-05,
+      "loss": 1.4009,
+      "step": 52030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7063716722534965e-05,
+      "loss": 1.081,
+      "step": 52035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.706318180027071e-05,
+      "loss": 1.2536,
+      "step": 52040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.706264683767188e-05,
+      "loss": 1.1403,
+      "step": 52045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.706211183474154e-05,
+      "loss": 1.2699,
+      "step": 52050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7061576791482745e-05,
+      "loss": 1.5211,
+      "step": 52055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7061041707898547e-05,
+      "loss": 0.8897,
+      "step": 52060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7060506583992002e-05,
+      "loss": 1.2829,
+      "step": 52065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7059971419766167e-05,
+      "loss": 1.1708,
+      "step": 52070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7059436215224102e-05,
+      "loss": 0.9973,
+      "step": 52075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7058900970368856e-05,
+      "loss": 1.4556,
+      "step": 52080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7058365685203488e-05,
+      "loss": 1.599,
+      "step": 52085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7057830359731057e-05,
+      "loss": 1.3518,
+      "step": 52090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.705729499395462e-05,
+      "loss": 1.0409,
+      "step": 52095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.705675958787723e-05,
+      "loss": 1.1852,
+      "step": 52100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.705622414150195e-05,
+      "loss": 1.1333,
+      "step": 52105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7055688654831834e-05,
+      "loss": 1.1822,
+      "step": 52110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7055153127869937e-05,
+      "loss": 1.077,
+      "step": 52115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7054617560619328e-05,
+      "loss": 1.3286,
+      "step": 52120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7054081953083056e-05,
+      "loss": 1.2207,
+      "step": 52125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7053546305264177e-05,
+      "loss": 1.4694,
+      "step": 52130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.705301061716576e-05,
+      "loss": 1.5078,
+      "step": 52135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7052474888790857e-05,
+      "loss": 1.0131,
+      "step": 52140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7051939120142532e-05,
+      "loss": 1.1619,
+      "step": 52145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.705140331122384e-05,
+      "loss": 1.2104,
+      "step": 52150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7050867462037837e-05,
+      "loss": 1.2111,
+      "step": 52155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7050331572587598e-05,
+      "loss": 1.0949,
+      "step": 52160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7049795642876167e-05,
+      "loss": 1.3207,
+      "step": 52165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7049259672906613e-05,
+      "loss": 1.3604,
+      "step": 52170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7048723662681997e-05,
+      "loss": 1.6818,
+      "step": 52175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7048187612205377e-05,
+      "loss": 1.0563,
+      "step": 52180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.704765152147981e-05,
+      "loss": 1.1057,
+      "step": 52185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7047115390508368e-05,
+      "loss": 1.3121,
+      "step": 52190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7046579219294102e-05,
+      "loss": 1.2292,
+      "step": 52195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7046043007840083e-05,
+      "loss": 1.1189,
+      "step": 52200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.704550675614937e-05,
+      "loss": 1.3386,
+      "step": 52205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7044970464225018e-05,
+      "loss": 1.4703,
+      "step": 52210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.70444341320701e-05,
+      "loss": 1.6642,
+      "step": 52215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7043897759687672e-05,
+      "loss": 1.3982,
+      "step": 52220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.70433613470808e-05,
+      "loss": 1.1187,
+      "step": 52225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7042824894252545e-05,
+      "loss": 0.9768,
+      "step": 52230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7042288401205973e-05,
+      "loss": 1.1378,
+      "step": 52235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7041751867944147e-05,
+      "loss": 1.4338,
+      "step": 52240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.704121529447013e-05,
+      "loss": 1.532,
+      "step": 52245
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7040678680786988e-05,
+      "loss": 1.3175,
+      "step": 52250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7040142026897783e-05,
+      "loss": 1.2531,
+      "step": 52255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.703960533280558e-05,
+      "loss": 1.2531,
+      "step": 52260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7039068598513443e-05,
+      "loss": 1.1306,
+      "step": 52265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.703853182402444e-05,
+      "loss": 1.3538,
+      "step": 52270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7037995009341633e-05,
+      "loss": 1.3106,
+      "step": 52275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7037458154468092e-05,
+      "loss": 1.2751,
+      "step": 52280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7036921259406876e-05,
+      "loss": 1.4825,
+      "step": 52285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.703638432416106e-05,
+      "loss": 1.2663,
+      "step": 52290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.70358473487337e-05,
+      "loss": 0.9864,
+      "step": 52295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7035310333127867e-05,
+      "loss": 0.8633,
+      "step": 52300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.703477327734663e-05,
+      "loss": 1.2664,
+      "step": 52305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7034236181393056e-05,
+      "loss": 1.1359,
+      "step": 52310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7033699045270208e-05,
+      "loss": 1.3592,
+      "step": 52315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7033161868981158e-05,
+      "loss": 1.2116,
+      "step": 52320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7032624652528967e-05,
+      "loss": 1.1163,
+      "step": 52325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7032087395916708e-05,
+      "loss": 1.1104,
+      "step": 52330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.703155009914745e-05,
+      "loss": 1.1888,
+      "step": 52335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7031012762224254e-05,
+      "loss": 1.0855,
+      "step": 52340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7030475385150198e-05,
+      "loss": 1.0566,
+      "step": 52345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7029937967928346e-05,
+      "loss": 1.1779,
+      "step": 52350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7029400510561764e-05,
+      "loss": 1.1213,
+      "step": 52355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7028863013053528e-05,
+      "loss": 0.9775,
+      "step": 52360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.70283254754067e-05,
+      "loss": 1.042,
+      "step": 52365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7027787897624356e-05,
+      "loss": 1.1456,
+      "step": 52370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7027250279709563e-05,
+      "loss": 0.9765,
+      "step": 52375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.702671262166539e-05,
+      "loss": 1.0823,
+      "step": 52380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7026174923494908e-05,
+      "loss": 1.4669,
+      "step": 52385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7025637185201192e-05,
+      "loss": 1.3323,
+      "step": 52390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7025099406787303e-05,
+      "loss": 1.1816,
+      "step": 52395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7024561588256324e-05,
+      "loss": 1.3552,
+      "step": 52400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7024023729611316e-05,
+      "loss": 1.2304,
+      "step": 52405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7023485830855356e-05,
+      "loss": 1.0577,
+      "step": 52410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7022947891991512e-05,
+      "loss": 0.9898,
+      "step": 52415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7022409913022865e-05,
+      "loss": 1.2543,
+      "step": 52420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7021871893952475e-05,
+      "loss": 1.3692,
+      "step": 52425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.702133383478342e-05,
+      "loss": 1.0265,
+      "step": 52430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.702079573551877e-05,
+      "loss": 1.1762,
+      "step": 52435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7020257596161608e-05,
+      "loss": 1.2925,
+      "step": 52440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.701971941671499e-05,
+      "loss": 1.3011,
+      "step": 52445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7019181197182005e-05,
+      "loss": 1.1535,
+      "step": 52450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7018642937565717e-05,
+      "loss": 1.4491,
+      "step": 52455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7018104637869206e-05,
+      "loss": 1.2546,
+      "step": 52460
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.701756629809554e-05,
+      "loss": 1.3889,
+      "step": 52465
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.70170279182478e-05,
+      "loss": 0.9326,
+      "step": 52470
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7016489498329048e-05,
+      "loss": 0.8448,
+      "step": 52475
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7015951038342376e-05,
+      "loss": 1.1159,
+      "step": 52480
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7015412538290847e-05,
+      "loss": 1.2782,
+      "step": 52485
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7014873998177535e-05,
+      "loss": 1.3607,
+      "step": 52490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7014335418005525e-05,
+      "loss": 1.1813,
+      "step": 52495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7013796797777885e-05,
+      "loss": 1.2622,
+      "step": 52500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7013258137497693e-05,
+      "loss": 1.8665,
+      "step": 52505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7012719437168025e-05,
+      "loss": 1.0881,
+      "step": 52510
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7012180696791957e-05,
+      "loss": 1.1418,
+      "step": 52515
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7011641916372566e-05,
+      "loss": 1.3221,
+      "step": 52520
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7011103095912927e-05,
+      "loss": 1.2561,
+      "step": 52525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7010564235416122e-05,
+      "loss": 1.1001,
+      "step": 52530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.701002533488522e-05,
+      "loss": 1.1337,
+      "step": 52535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7009486394323306e-05,
+      "loss": 1.3902,
+      "step": 52540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7008947413733454e-05,
+      "loss": 0.9587,
+      "step": 52545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.700840839311874e-05,
+      "loss": 1.38,
+      "step": 52550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.700786933248225e-05,
+      "loss": 1.1419,
+      "step": 52555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7007330231827054e-05,
+      "loss": 1.1865,
+      "step": 52560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7006791091156232e-05,
+      "loss": 1.1567,
+      "step": 52565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7006251910472865e-05,
+      "loss": 1.068,
+      "step": 52570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7005712689780032e-05,
+      "loss": 1.3983,
+      "step": 52575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7005173429080813e-05,
+      "loss": 1.3383,
+      "step": 52580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7004634128378285e-05,
+      "loss": 1.1991,
+      "step": 52585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7004094787675527e-05,
+      "loss": 1.5773,
+      "step": 52590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.700355540697562e-05,
+      "loss": 1.4803,
+      "step": 52595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7003015986281648e-05,
+      "loss": 1.1906,
+      "step": 52600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7002476525596686e-05,
+      "loss": 1.1673,
+      "step": 52605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7001937024923818e-05,
+      "loss": 1.4338,
+      "step": 52610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.700139748426612e-05,
+      "loss": 1.1431,
+      "step": 52615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.700085790362668e-05,
+      "loss": 1.4145,
+      "step": 52620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7000318283008578e-05,
+      "loss": 1.0819,
+      "step": 52625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.699977862241489e-05,
+      "loss": 1.3519,
+      "step": 52630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6999238921848703e-05,
+      "loss": 1.2648,
+      "step": 52635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6998699181313095e-05,
+      "loss": 1.4905,
+      "step": 52640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.699815940081115e-05,
+      "loss": 1.0823,
+      "step": 52645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6997619580345955e-05,
+      "loss": 1.1971,
+      "step": 52650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6997079719920586e-05,
+      "loss": 1.5097,
+      "step": 52655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6996539819538128e-05,
+      "loss": 1.2781,
+      "step": 52660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6995999879201665e-05,
+      "loss": 1.4684,
+      "step": 52665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.699545989891428e-05,
+      "loss": 1.025,
+      "step": 52670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6994919878679054e-05,
+      "loss": 1.0911,
+      "step": 52675
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6994379818499077e-05,
+      "loss": 1.4709,
+      "step": 52680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6993839718377428e-05,
+      "loss": 1.3712,
+      "step": 52685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6993299578317192e-05,
+      "loss": 1.0671,
+      "step": 52690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6992759398321454e-05,
+      "loss": 0.9263,
+      "step": 52695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6992219178393302e-05,
+      "loss": 1.0353,
+      "step": 52700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6991678918535815e-05,
+      "loss": 1.2044,
+      "step": 52705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.699113861875208e-05,
+      "loss": 1.1736,
+      "step": 52710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6990598279045183e-05,
+      "loss": 1.0952,
+      "step": 52715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6990057899418212e-05,
+      "loss": 1.1302,
+      "step": 52720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6989517479874248e-05,
+      "loss": 0.9344,
+      "step": 52725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6988977020416384e-05,
+      "loss": 1.2222,
+      "step": 52730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6988436521047696e-05,
+      "loss": 1.0829,
+      "step": 52735
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.698789598177128e-05,
+      "loss": 1.0984,
+      "step": 52740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6987355402590223e-05,
+      "loss": 1.3078,
+      "step": 52745
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6986814783507607e-05,
+      "loss": 1.3747,
+      "step": 52750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6986274124526517e-05,
+      "loss": 1.3143,
+      "step": 52755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6985733425650047e-05,
+      "loss": 1.349,
+      "step": 52760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.698519268688128e-05,
+      "loss": 1.1734,
+      "step": 52765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6984651908223308e-05,
+      "loss": 1.0675,
+      "step": 52770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6984111089679216e-05,
+      "loss": 1.0644,
+      "step": 52775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6983570231252093e-05,
+      "loss": 1.6153,
+      "step": 52780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6983029332945025e-05,
+      "loss": 1.4789,
+      "step": 52785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.698248839476111e-05,
+      "loss": 1.2556,
+      "step": 52790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6981947416703428e-05,
+      "loss": 1.0245,
+      "step": 52795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.698140639877507e-05,
+      "loss": 1.119,
+      "step": 52800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6980865340979126e-05,
+      "loss": 1.1352,
+      "step": 52805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6980324243318688e-05,
+      "loss": 1.2567,
+      "step": 52810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6979783105796844e-05,
+      "loss": 1.583,
+      "step": 52815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.697924192841668e-05,
+      "loss": 1.2772,
+      "step": 52820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6978700711181298e-05,
+      "loss": 1.0538,
+      "step": 52825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6978159454093778e-05,
+      "loss": 1.1503,
+      "step": 52830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6977618157157213e-05,
+      "loss": 1.1364,
+      "step": 52835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6977076820374698e-05,
+      "loss": 1.3591,
+      "step": 52840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6976535443749317e-05,
+      "loss": 1.5134,
+      "step": 52845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6975994027284173e-05,
+      "loss": 1.0526,
+      "step": 52850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6975452570982347e-05,
+      "loss": 1.1646,
+      "step": 52855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6974911074846937e-05,
+      "loss": 1.2429,
+      "step": 52860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.697436953888103e-05,
+      "loss": 1.0174,
+      "step": 52865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.697382796308772e-05,
+      "loss": 1.5147,
+      "step": 52870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6973286347470107e-05,
+      "loss": 1.159,
+      "step": 52875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.697274469203128e-05,
+      "loss": 1.3239,
+      "step": 52880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6972202996774324e-05,
+      "loss": 1.0518,
+      "step": 52885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.697166126170234e-05,
+      "loss": 1.018,
+      "step": 52890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6971119486818427e-05,
+      "loss": 1.0622,
+      "step": 52895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6970577672125665e-05,
+      "loss": 1.3382,
+      "step": 52900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.697003581762716e-05,
+      "loss": 1.2789,
+      "step": 52905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6969493923326e-05,
+      "loss": 1.142,
+      "step": 52910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.696895198922528e-05,
+      "loss": 1.3227,
+      "step": 52915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6968410015328097e-05,
+      "loss": 1.2408,
+      "step": 52920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6967868001637545e-05,
+      "loss": 1.2671,
+      "step": 52925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.696732594815672e-05,
+      "loss": 0.8963,
+      "step": 52930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6966783854888712e-05,
+      "loss": 1.3205,
+      "step": 52935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6966241721836625e-05,
+      "loss": 1.1902,
+      "step": 52940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6965699549003547e-05,
+      "loss": 1.1362,
+      "step": 52945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6965157336392583e-05,
+      "loss": 1.199,
+      "step": 52950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6964615084006825e-05,
+      "loss": 1.0985,
+      "step": 52955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6964072791849363e-05,
+      "loss": 0.989,
+      "step": 52960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.69635304599233e-05,
+      "loss": 1.2051,
+      "step": 52965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.696298808823174e-05,
+      "loss": 1.4931,
+      "step": 52970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6962445676777768e-05,
+      "loss": 1.1817,
+      "step": 52975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.696190322556449e-05,
+      "loss": 0.9555,
+      "step": 52980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6961360734594996e-05,
+      "loss": 1.3351,
+      "step": 52985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.696081820387239e-05,
+      "loss": 1.4239,
+      "step": 52990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6960275633399765e-05,
+      "loss": 1.2283,
+      "step": 52995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.695973302318023e-05,
+      "loss": 1.1917,
+      "step": 53000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6959190373216867e-05,
+      "loss": 1.128,
+      "step": 53005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.695864768351279e-05,
+      "loss": 1.4372,
+      "step": 53010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6958104954071085e-05,
+      "loss": 0.782,
+      "step": 53015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6957562184894866e-05,
+      "loss": 1.3317,
+      "step": 53020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.695701937598722e-05,
+      "loss": 1.4311,
+      "step": 53025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6956476527351255e-05,
+      "loss": 1.081,
+      "step": 53030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6955933638990066e-05,
+      "loss": 0.8586,
+      "step": 53035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6955390710906756e-05,
+      "loss": 0.9843,
+      "step": 53040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6954847743104425e-05,
+      "loss": 1.2144,
+      "step": 53045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.695430473558617e-05,
+      "loss": 1.4159,
+      "step": 53050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6953761688355094e-05,
+      "loss": 1.2621,
+      "step": 53055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6953218601414302e-05,
+      "loss": 1.1919,
+      "step": 53060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.695267547476689e-05,
+      "loss": 1.5048,
+      "step": 53065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.695213230841596e-05,
+      "loss": 0.8801,
+      "step": 53070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6951589102364615e-05,
+      "loss": 1.2971,
+      "step": 53075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.695104585661596e-05,
+      "loss": 1.2908,
+      "step": 53080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6950502571173092e-05,
+      "loss": 1.3428,
+      "step": 53085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6949959246039117e-05,
+      "loss": 1.5518,
+      "step": 53090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.694941588121714e-05,
+      "loss": 1.551,
+      "step": 53095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6948872476710258e-05,
+      "loss": 1.4432,
+      "step": 53100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6948329032521577e-05,
+      "loss": 1.2616,
+      "step": 53105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6947785548654193e-05,
+      "loss": 1.1927,
+      "step": 53110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6947242025111224e-05,
+      "loss": 1.323,
+      "step": 53115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6946698461895766e-05,
+      "loss": 1.0528,
+      "step": 53120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6946154859010926e-05,
+      "loss": 1.212,
+      "step": 53125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6945611216459804e-05,
+      "loss": 1.1617,
+      "step": 53130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.69450675342455e-05,
+      "loss": 1.284,
+      "step": 53135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6944523812371132e-05,
+      "loss": 1.4915,
+      "step": 53140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6943980050839796e-05,
+      "loss": 1.0217,
+      "step": 53145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.69434362496546e-05,
+      "loss": 1.3291,
+      "step": 53150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6942892408818648e-05,
+      "loss": 1.1383,
+      "step": 53155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6942348528335046e-05,
+      "loss": 1.0953,
+      "step": 53160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6941804608206897e-05,
+      "loss": 1.501,
+      "step": 53165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.694126064843731e-05,
+      "loss": 1.2119,
+      "step": 53170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6940716649029396e-05,
+      "loss": 1.1301,
+      "step": 53175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6940172609986255e-05,
+      "loss": 0.9846,
+      "step": 53180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6939628531310994e-05,
+      "loss": 1.0197,
+      "step": 53185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6939084413006724e-05,
+      "loss": 1.1706,
+      "step": 53190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.693854025507655e-05,
+      "loss": 1.1219,
+      "step": 53195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6937996057523575e-05,
+      "loss": 1.247,
+      "step": 53200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6937451820350916e-05,
+      "loss": 1.1572,
+      "step": 53205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6936907543561672e-05,
+      "loss": 1.3424,
+      "step": 53210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6936363227158957e-05,
+      "loss": 1.2321,
+      "step": 53215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.693581887114588e-05,
+      "loss": 1.2697,
+      "step": 53220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.693527447552554e-05,
+      "loss": 1.1209,
+      "step": 53225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6934730040301056e-05,
+      "loss": 1.1007,
+      "step": 53230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6934185565475533e-05,
+      "loss": 1.2081,
+      "step": 53235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.693364105105208e-05,
+      "loss": 1.4002,
+      "step": 53240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6933096497033812e-05,
+      "loss": 1.4719,
+      "step": 53245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.693255190342383e-05,
+      "loss": 1.3707,
+      "step": 53250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6932007270225247e-05,
+      "loss": 1.4044,
+      "step": 53255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6931462597441176e-05,
+      "loss": 1.1369,
+      "step": 53260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6930917885074726e-05,
+      "loss": 1.2918,
+      "step": 53265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6930373133129008e-05,
+      "loss": 1.0625,
+      "step": 53270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.692982834160713e-05,
+      "loss": 1.1614,
+      "step": 53275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6929283510512203e-05,
+      "loss": 1.195,
+      "step": 53280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6928738639847343e-05,
+      "loss": 1.3142,
+      "step": 53285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6928193729615657e-05,
+      "loss": 1.4846,
+      "step": 53290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.692764877982026e-05,
+      "loss": 1.2711,
+      "step": 53295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6927103790464263e-05,
+      "loss": 1.2031,
+      "step": 53300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.692655876155078e-05,
+      "loss": 1.4668,
+      "step": 53305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6926013693082916e-05,
+      "loss": 1.2216,
+      "step": 53310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6925468585063788e-05,
+      "loss": 1.3343,
+      "step": 53315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6924923437496513e-05,
+      "loss": 1.4141,
+      "step": 53320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.69243782503842e-05,
+      "loss": 0.936,
+      "step": 53325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6923833023729963e-05,
+      "loss": 0.9831,
+      "step": 53330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6923287757536915e-05,
+      "loss": 1.4481,
+      "step": 53335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6922742451808168e-05,
+      "loss": 1.0673,
+      "step": 53340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6922197106546842e-05,
+      "loss": 1.2571,
+      "step": 53345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6921651721756046e-05,
+      "loss": 1.7778,
+      "step": 53350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6921106297438892e-05,
+      "loss": 1.4223,
+      "step": 53355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6920560833598502e-05,
+      "loss": 1.0819,
+      "step": 53360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.692001533023799e-05,
+      "loss": 1.0243,
+      "step": 53365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6919469787360463e-05,
+      "loss": 1.3579,
+      "step": 53370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6918924204969048e-05,
+      "loss": 1.0849,
+      "step": 53375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.691837858306685e-05,
+      "loss": 1.0932,
+      "step": 53380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.691783292165699e-05,
+      "loss": 1.0132,
+      "step": 53385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6917287220742582e-05,
+      "loss": 1.1009,
+      "step": 53390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6916741480326745e-05,
+      "loss": 1.5721,
+      "step": 53395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6916195700412593e-05,
+      "loss": 1.4377,
+      "step": 53400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6915649881003246e-05,
+      "loss": 1.4619,
+      "step": 53405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6915104022101818e-05,
+      "loss": 1.2483,
+      "step": 53410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6914558123711426e-05,
+      "loss": 1.1482,
+      "step": 53415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.691401218583519e-05,
+      "loss": 1.3113,
+      "step": 53420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.691346620847622e-05,
+      "loss": 1.2608,
+      "step": 53425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6912920191637643e-05,
+      "loss": 1.3423,
+      "step": 53430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6912374135322576e-05,
+      "loss": 1.2168,
+      "step": 53435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6911828039534134e-05,
+      "loss": 1.1949,
+      "step": 53440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6911281904275434e-05,
+      "loss": 1.5307,
+      "step": 53445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6910735729549596e-05,
+      "loss": 1.2669,
+      "step": 53450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6910189515359744e-05,
+      "loss": 1.0686,
+      "step": 53455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.690964326170899e-05,
+      "loss": 1.1726,
+      "step": 53460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6909096968600457e-05,
+      "loss": 1.311,
+      "step": 53465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.690855063603727e-05,
+      "loss": 1.2158,
+      "step": 53470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6908004264022534e-05,
+      "loss": 1.3764,
+      "step": 53475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6907457852559383e-05,
+      "loss": 1.2138,
+      "step": 53480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.690691140165093e-05,
+      "loss": 1.2088,
+      "step": 53485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.69063649113003e-05,
+      "loss": 1.4381,
+      "step": 53490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6905818381510615e-05,
+      "loss": 1.5747,
+      "step": 53495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.690527181228499e-05,
+      "loss": 1.2922,
+      "step": 53500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6904725203626547e-05,
+      "loss": 1.3935,
+      "step": 53505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6904178555538413e-05,
+      "loss": 1.1779,
+      "step": 53510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.69036318680237e-05,
+      "loss": 1.0499,
+      "step": 53515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.690308514108554e-05,
+      "loss": 1.0984,
+      "step": 53520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6902538374727054e-05,
+      "loss": 1.4519,
+      "step": 53525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6901991568951354e-05,
+      "loss": 1.3277,
+      "step": 53530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6901444723761577e-05,
+      "loss": 1.2518,
+      "step": 53535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6900897839160834e-05,
+      "loss": 1.1092,
+      "step": 53540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6900350915152253e-05,
+      "loss": 1.3607,
+      "step": 53545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.689980395173896e-05,
+      "loss": 1.2952,
+      "step": 53550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6899256948924074e-05,
+      "loss": 1.2467,
+      "step": 53555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.689870990671072e-05,
+      "loss": 1.3054,
+      "step": 53560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.689816282510202e-05,
+      "loss": 1.308,
+      "step": 53565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.68976157041011e-05,
+      "loss": 1.337,
+      "step": 53570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.689706854371109e-05,
+      "loss": 0.9697,
+      "step": 53575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6896521343935103e-05,
+      "loss": 1.5263,
+      "step": 53580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6895974104776272e-05,
+      "loss": 1.2081,
+      "step": 53585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.689542682623772e-05,
+      "loss": 1.1856,
+      "step": 53590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.689487950832257e-05,
+      "loss": 1.2103,
+      "step": 53595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6894332151033953e-05,
+      "loss": 1.0996,
+      "step": 53600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.689378475437499e-05,
+      "loss": 1.516,
+      "step": 53605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6893237318348806e-05,
+      "loss": 1.1833,
+      "step": 53610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.689268984295853e-05,
+      "loss": 1.2904,
+      "step": 53615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.689214232820729e-05,
+      "loss": 1.1349,
+      "step": 53620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6891594774098213e-05,
+      "loss": 1.286,
+      "step": 53625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6891047180634418e-05,
+      "loss": 1.2511,
+      "step": 53630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6890499547819037e-05,
+      "loss": 1.4217,
+      "step": 53635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.68899518756552e-05,
+      "loss": 1.1114,
+      "step": 53640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.688940416414603e-05,
+      "loss": 1.0153,
+      "step": 53645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.688885641329466e-05,
+      "loss": 1.4163,
+      "step": 53650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6888308623104213e-05,
+      "loss": 0.9445,
+      "step": 53655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6887760793577818e-05,
+      "loss": 1.176,
+      "step": 53660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6887212924718604e-05,
+      "loss": 1.1169,
+      "step": 53665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.68866650165297e-05,
+      "loss": 1.0616,
+      "step": 53670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6886117069014236e-05,
+      "loss": 1.0491,
+      "step": 53675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.688556908217534e-05,
+      "loss": 1.1693,
+      "step": 53680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.688502105601614e-05,
+      "loss": 1.2313,
+      "step": 53685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6884472990539767e-05,
+      "loss": 1.4489,
+      "step": 53690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.688392488574935e-05,
+      "loss": 1.1804,
+      "step": 53695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.688337674164802e-05,
+      "loss": 1.293,
+      "step": 53700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6882828558238907e-05,
+      "loss": 1.1326,
+      "step": 53705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.688228033552514e-05,
+      "loss": 1.0185,
+      "step": 53710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.688173207350985e-05,
+      "loss": 1.4009,
+      "step": 53715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6881183772196174e-05,
+      "loss": 1.4302,
+      "step": 53720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.688063543158723e-05,
+      "loss": 1.1829,
+      "step": 53725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.688008705168616e-05,
+      "loss": 1.0426,
+      "step": 53730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.687953863249609e-05,
+      "loss": 1.0913,
+      "step": 53735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.687899017402016e-05,
+      "loss": 1.407,
+      "step": 53740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6878441676261493e-05,
+      "loss": 1.1019,
+      "step": 53745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6877893139223225e-05,
+      "loss": 1.1237,
+      "step": 53750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6877344562908487e-05,
+      "loss": 1.2613,
+      "step": 53755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6876795947320412e-05,
+      "loss": 1.1398,
+      "step": 53760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6876247292462135e-05,
+      "loss": 1.4323,
+      "step": 53765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6875698598336786e-05,
+      "loss": 1.0537,
+      "step": 53770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6875149864947502e-05,
+      "loss": 1.5329,
+      "step": 53775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.687460109229741e-05,
+      "loss": 1.3877,
+      "step": 53780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.687405228038965e-05,
+      "loss": 0.8074,
+      "step": 53785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6873503429227358e-05,
+      "loss": 1.3148,
+      "step": 53790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6872954538813663e-05,
+      "loss": 1.3087,
+      "step": 53795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.68724056091517e-05,
+      "loss": 1.3292,
+      "step": 53800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6871856640244605e-05,
+      "loss": 1.5196,
+      "step": 53805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6871307632095513e-05,
+      "loss": 1.0006,
+      "step": 53810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6870758584707555e-05,
+      "loss": 1.0515,
+      "step": 53815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6870209498083875e-05,
+      "loss": 1.2514,
+      "step": 53820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.68696603722276e-05,
+      "loss": 1.2401,
+      "step": 53825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.686911120714187e-05,
+      "loss": 1.1646,
+      "step": 53830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6868562002829824e-05,
+      "loss": 1.287,
+      "step": 53835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6868012759294592e-05,
+      "loss": 1.4166,
+      "step": 53840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6867463476539313e-05,
+      "loss": 1.2807,
+      "step": 53845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6866914154567126e-05,
+      "loss": 1.2439,
+      "step": 53850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6866364793381162e-05,
+      "loss": 1.1709,
+      "step": 53855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6865815392984563e-05,
+      "loss": 1.2797,
+      "step": 53860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.686526595338047e-05,
+      "loss": 1.7387,
+      "step": 53865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.686471647457201e-05,
+      "loss": 1.3561,
+      "step": 53870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6864166956562328e-05,
+      "loss": 1.1754,
+      "step": 53875
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6863617399354563e-05,
+      "loss": 1.0741,
+      "step": 53880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6863067802951846e-05,
+      "loss": 1.2899,
+      "step": 53885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6862518167357328e-05,
+      "loss": 1.5755,
+      "step": 53890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6861968492574136e-05,
+      "loss": 1.155,
+      "step": 53895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6861418778605414e-05,
+      "loss": 1.2174,
+      "step": 53900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6860869025454297e-05,
+      "loss": 1.1314,
+      "step": 53905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6860319233123932e-05,
+      "loss": 1.0589,
+      "step": 53910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6859769401617452e-05,
+      "loss": 0.9427,
+      "step": 53915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6859219530938e-05,
+      "loss": 1.0631,
+      "step": 53920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6858669621088718e-05,
+      "loss": 1.2643,
+      "step": 53925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.685811967207274e-05,
+      "loss": 1.1742,
+      "step": 53930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.685756968389321e-05,
+      "loss": 1.2348,
+      "step": 53935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.685701965655327e-05,
+      "loss": 1.3151,
+      "step": 53940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.685646959005606e-05,
+      "loss": 1.251,
+      "step": 53945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.685591948440472e-05,
+      "loss": 1.5527,
+      "step": 53950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6855369339602393e-05,
+      "loss": 1.1119,
+      "step": 53955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6854819155652216e-05,
+      "loss": 1.1624,
+      "step": 53960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.685426893255734e-05,
+      "loss": 0.9685,
+      "step": 53965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6853718670320902e-05,
+      "loss": 1.3925,
+      "step": 53970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6853168368946043e-05,
+      "loss": 1.2981,
+      "step": 53975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6852618028435907e-05,
+      "loss": 1.2228,
+      "step": 53980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6852067648793635e-05,
+      "loss": 1.0382,
+      "step": 53985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6851517230022367e-05,
+      "loss": 1.1239,
+      "step": 53990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6850966772125255e-05,
+      "loss": 1.2667,
+      "step": 53995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.685041627510544e-05,
+      "loss": 1.105,
+      "step": 54000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.684986573896606e-05,
+      "loss": 0.9438,
+      "step": 54005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6849315163710268e-05,
+      "loss": 1.1794,
+      "step": 54010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6848764549341194e-05,
+      "loss": 1.243,
+      "step": 54015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6848213895861996e-05,
+      "loss": 1.3701,
+      "step": 54020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6847663203275812e-05,
+      "loss": 1.2641,
+      "step": 54025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6847112471585786e-05,
+      "loss": 1.3279,
+      "step": 54030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.684656170079507e-05,
+      "loss": 1.3397,
+      "step": 54035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6846010890906803e-05,
+      "loss": 1.285,
+      "step": 54040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.684546004192413e-05,
+      "loss": 1.3573,
+      "step": 54045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.68449091538502e-05,
+      "loss": 1.101,
+      "step": 54050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6844358226688154e-05,
+      "loss": 1.2711,
+      "step": 54055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6843807260441146e-05,
+      "loss": 1.7234,
+      "step": 54060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6843256255112312e-05,
+      "loss": 1.1711,
+      "step": 54065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.684270521070481e-05,
+      "loss": 1.296,
+      "step": 54070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6842154127221774e-05,
+      "loss": 1.2142,
+      "step": 54075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6841603004666365e-05,
+      "loss": 1.3022,
+      "step": 54080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6841051843041718e-05,
+      "loss": 1.1638,
+      "step": 54085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.684050064235099e-05,
+      "loss": 0.7779,
+      "step": 54090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6839949402597316e-05,
+      "loss": 1.3885,
+      "step": 54095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.683939812378386e-05,
+      "loss": 1.6029,
+      "step": 54100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.683884680591376e-05,
+      "loss": 1.6373,
+      "step": 54105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6838295448990164e-05,
+      "loss": 1.1763,
+      "step": 54110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6837744053016226e-05,
+      "loss": 1.2697,
+      "step": 54115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6837192617995093e-05,
+      "loss": 1.4076,
+      "step": 54120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6836641143929908e-05,
+      "loss": 1.3,
+      "step": 54125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.683608963082383e-05,
+      "loss": 1.5341,
+      "step": 54130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6835538078680002e-05,
+      "loss": 1.0137,
+      "step": 54135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6834986487501572e-05,
+      "loss": 1.3826,
+      "step": 54140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6834434857291694e-05,
+      "loss": 0.9613,
+      "step": 54145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.683388318805352e-05,
+      "loss": 1.4877,
+      "step": 54150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6833331479790197e-05,
+      "loss": 1.1033,
+      "step": 54155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6832779732504872e-05,
+      "loss": 1.2978,
+      "step": 54160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.68322279462007e-05,
+      "loss": 0.9601,
+      "step": 54165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6831676120880835e-05,
+      "loss": 0.8494,
+      "step": 54170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6831124256548422e-05,
+      "loss": 1.1372,
+      "step": 54175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.683057235320662e-05,
+      "loss": 1.4803,
+      "step": 54180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.683002041085857e-05,
+      "loss": 0.9602,
+      "step": 54185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6829468429507434e-05,
+      "loss": 1.1795,
+      "step": 54190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.682891640915636e-05,
+      "loss": 1.0759,
+      "step": 54195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6828364349808498e-05,
+      "loss": 1.2063,
+      "step": 54200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6827812251467004e-05,
+      "loss": 0.9553,
+      "step": 54205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6827260114135028e-05,
+      "loss": 1.0956,
+      "step": 54210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6826707937815728e-05,
+      "loss": 1.3272,
+      "step": 54215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6826155722512255e-05,
+      "loss": 1.3914,
+      "step": 54220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6825603468227757e-05,
+      "loss": 1.5282,
+      "step": 54225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.682505117496539e-05,
+      "loss": 1.0786,
+      "step": 54230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6824498842728318e-05,
+      "loss": 1.4381,
+      "step": 54235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.682394647151968e-05,
+      "loss": 1.1789,
+      "step": 54240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6823394061342644e-05,
+      "loss": 0.9628,
+      "step": 54245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6822841612200354e-05,
+      "loss": 1.3511,
+      "step": 54250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6822289124095968e-05,
+      "loss": 1.4459,
+      "step": 54255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6821736597032644e-05,
+      "loss": 1.3305,
+      "step": 54260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6821184031013532e-05,
+      "loss": 1.2935,
+      "step": 54265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.682063142604179e-05,
+      "loss": 1.0583,
+      "step": 54270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.682007878212058e-05,
+      "loss": 0.9911,
+      "step": 54275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6819526099253046e-05,
+      "loss": 1.0945,
+      "step": 54280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6818973377442352e-05,
+      "loss": 1.0667,
+      "step": 54285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.681842061669165e-05,
+      "loss": 1.1832,
+      "step": 54290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6817867817004105e-05,
+      "loss": 1.1427,
+      "step": 54295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6817314978382862e-05,
+      "loss": 1.07,
+      "step": 54300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6816762100831087e-05,
+      "loss": 1.5145,
+      "step": 54305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.681620918435193e-05,
+      "loss": 1.3514,
+      "step": 54310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.681565622894856e-05,
+      "loss": 1.193,
+      "step": 54315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.681510323462412e-05,
+      "loss": 1.0357,
+      "step": 54320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6814550201381778e-05,
+      "loss": 1.2587,
+      "step": 54325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.681399712922469e-05,
+      "loss": 1.1185,
+      "step": 54330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.681344401815601e-05,
+      "loss": 1.0584,
+      "step": 54335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6812890868178903e-05,
+      "loss": 1.2484,
+      "step": 54340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6812337679296523e-05,
+      "loss": 1.4019,
+      "step": 54345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.681178445151203e-05,
+      "loss": 1.2467,
+      "step": 54350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6811231184828587e-05,
+      "loss": 1.0078,
+      "step": 54355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.681067787924935e-05,
+      "loss": 1.2452,
+      "step": 54360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.681012453477748e-05,
+      "loss": 1.6071,
+      "step": 54365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6809571151416135e-05,
+      "loss": 1.254,
+      "step": 54370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6809017729168473e-05,
+      "loss": 1.553,
+      "step": 54375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6808464268037662e-05,
+      "loss": 1.3805,
+      "step": 54380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6807910768026853e-05,
+      "loss": 1.4764,
+      "step": 54385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6807357229139217e-05,
+      "loss": 1.3244,
+      "step": 54390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.680680365137791e-05,
+      "loss": 1.197,
+      "step": 54395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6806250034746086e-05,
+      "loss": 1.3724,
+      "step": 54400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6805696379246923e-05,
+      "loss": 1.1868,
+      "step": 54405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6805142684883568e-05,
+      "loss": 1.4997,
+      "step": 54410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6804588951659188e-05,
+      "loss": 1.3538,
+      "step": 54415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6804035179576947e-05,
+      "loss": 1.2191,
+      "step": 54420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6803481368640005e-05,
+      "loss": 1.1212,
+      "step": 54425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6802927518851525e-05,
+      "loss": 1.4972,
+      "step": 54430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.680237363021467e-05,
+      "loss": 1.1076,
+      "step": 54435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.68018197027326e-05,
+      "loss": 1.4198,
+      "step": 54440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6801265736408488e-05,
+      "loss": 1.1661,
+      "step": 54445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6800711731245485e-05,
+      "loss": 1.4073,
+      "step": 54450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.680015768724676e-05,
+      "loss": 1.294,
+      "step": 54455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.679960360441548e-05,
+      "loss": 1.284,
+      "step": 54460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6799049482754807e-05,
+      "loss": 1.1975,
+      "step": 54465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6798495322267903e-05,
+      "loss": 1.0539,
+      "step": 54470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.679794112295793e-05,
+      "loss": 1.2803,
+      "step": 54475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6797386884828063e-05,
+      "loss": 1.3287,
+      "step": 54480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6796832607881457e-05,
+      "loss": 1.4543,
+      "step": 54485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6796278292121284e-05,
+      "loss": 1.196,
+      "step": 54490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6795723937550705e-05,
+      "loss": 1.1178,
+      "step": 54495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6795169544172888e-05,
+      "loss": 1.0382,
+      "step": 54500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6794615111990994e-05,
+      "loss": 1.4382,
+      "step": 54505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6794060641008198e-05,
+      "loss": 1.1878,
+      "step": 54510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.679350613122766e-05,
+      "loss": 1.2321,
+      "step": 54515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.679295158265255e-05,
+      "loss": 0.9603,
+      "step": 54520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6792396995286032e-05,
+      "loss": 1.3545,
+      "step": 54525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6791842369131272e-05,
+      "loss": 1.2138,
+      "step": 54530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6791287704191443e-05,
+      "loss": 1.1722,
+      "step": 54535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6790733000469702e-05,
+      "loss": 1.0558,
+      "step": 54540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6790178257969224e-05,
+      "loss": 1.2428,
+      "step": 54545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.678962347669318e-05,
+      "loss": 1.2335,
+      "step": 54550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6789068656644736e-05,
+      "loss": 1.7181,
+      "step": 54555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6788513797827054e-05,
+      "loss": 1.851,
+      "step": 54560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6787958900243307e-05,
+      "loss": 1.1942,
+      "step": 54565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6787403963896666e-05,
+      "loss": 1.4476,
+      "step": 54570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6786848988790295e-05,
+      "loss": 1.1073,
+      "step": 54575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.678629397492737e-05,
+      "loss": 1.4803,
+      "step": 54580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6785738922311054e-05,
+      "loss": 1.4759,
+      "step": 54585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.678518383094452e-05,
+      "loss": 1.0853,
+      "step": 54590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6784628700830932e-05,
+      "loss": 0.9634,
+      "step": 54595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.678407353197347e-05,
+      "loss": 1.1955,
+      "step": 54600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.67835183243753e-05,
+      "loss": 1.292,
+      "step": 54605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.678296307803959e-05,
+      "loss": 1.269,
+      "step": 54610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6782407792969514e-05,
+      "loss": 1.2068,
+      "step": 54615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.678185246916824e-05,
+      "loss": 1.3165,
+      "step": 54620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.678129710663894e-05,
+      "loss": 1.074,
+      "step": 54625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.678074170538479e-05,
+      "loss": 1.3477,
+      "step": 54630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6780186265408952e-05,
+      "loss": 1.0458,
+      "step": 54635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.677963078671461e-05,
+      "loss": 1.0055,
+      "step": 54640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6779075269304925e-05,
+      "loss": 1.4309,
+      "step": 54645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6778519713183077e-05,
+      "loss": 1.3723,
+      "step": 54650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6777964118352233e-05,
+      "loss": 1.038,
+      "step": 54655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6777408484815572e-05,
+      "loss": 1.2535,
+      "step": 54660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6776852812576257e-05,
+      "loss": 1.2672,
+      "step": 54665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6776297101637475e-05,
+      "loss": 0.9366,
+      "step": 54670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6775741352002385e-05,
+      "loss": 1.4298,
+      "step": 54675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6775185563674172e-05,
+      "loss": 1.1826,
+      "step": 54680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6774629736656006e-05,
+      "loss": 1.1973,
+      "step": 54685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.677407387095106e-05,
+      "loss": 0.9679,
+      "step": 54690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6773517966562508e-05,
+      "loss": 1.3859,
+      "step": 54695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6772962023493522e-05,
+      "loss": 1.1917,
+      "step": 54700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.677240604174728e-05,
+      "loss": 0.9684,
+      "step": 54705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6771850021326963e-05,
+      "loss": 1.6049,
+      "step": 54710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6771293962235734e-05,
+      "loss": 1.395,
+      "step": 54715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6770737864476777e-05,
+      "loss": 1.241,
+      "step": 54720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.677018172805327e-05,
+      "loss": 1.2892,
+      "step": 54725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6769625552968375e-05,
+      "loss": 1.1477,
+      "step": 54730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6769069339225284e-05,
+      "loss": 1.0628,
+      "step": 54735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.676851308682716e-05,
+      "loss": 0.9817,
+      "step": 54740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6767956795777194e-05,
+      "loss": 1.5397,
+      "step": 54745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.676740046607855e-05,
+      "loss": 1.0558,
+      "step": 54750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.676684409773441e-05,
+      "loss": 1.5219,
+      "step": 54755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6766287690747952e-05,
+      "loss": 1.151,
+      "step": 54760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.676573124512235e-05,
+      "loss": 1.1951,
+      "step": 54765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.676517476086078e-05,
+      "loss": 1.309,
+      "step": 54770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6764618237966427e-05,
+      "loss": 1.1841,
+      "step": 54775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6764061676442468e-05,
+      "loss": 1.2402,
+      "step": 54780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6763505076292075e-05,
+      "loss": 1.1511,
+      "step": 54785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6762948437518433e-05,
+      "loss": 1.5032,
+      "step": 54790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6762391760124717e-05,
+      "loss": 1.2525,
+      "step": 54795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6761835044114104e-05,
+      "loss": 1.1522,
+      "step": 54800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.676127828948978e-05,
+      "loss": 1.2457,
+      "step": 54805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.676072149625492e-05,
+      "loss": 1.0825,
+      "step": 54810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6760164664412702e-05,
+      "loss": 0.9734,
+      "step": 54815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.675960779396631e-05,
+      "loss": 1.3689,
+      "step": 54820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6759050884918922e-05,
+      "loss": 1.2827,
+      "step": 54825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6758493937273716e-05,
+      "loss": 1.1727,
+      "step": 54830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6757936951033875e-05,
+      "loss": 1.4748,
+      "step": 54835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6757379926202578e-05,
+      "loss": 1.0773,
+      "step": 54840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.675682286278301e-05,
+      "loss": 1.1211,
+      "step": 54845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6756265760778347e-05,
+      "loss": 1.2299,
+      "step": 54850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6755708620191776e-05,
+      "loss": 1.2807,
+      "step": 54855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6755151441026472e-05,
+      "loss": 1.4611,
+      "step": 54860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.675459422328562e-05,
+      "loss": 1.0331,
+      "step": 54865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6754036966972404e-05,
+      "loss": 1.3547,
+      "step": 54870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6753479672090002e-05,
+      "loss": 1.3093,
+      "step": 54875
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.67529223386416e-05,
+      "loss": 1.3766,
+      "step": 54880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6752364966630378e-05,
+      "loss": 1.3714,
+      "step": 54885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6751807556059522e-05,
+      "loss": 1.2661,
+      "step": 54890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6751250106932214e-05,
+      "loss": 1.5058,
+      "step": 54895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6750692619251636e-05,
+      "loss": 1.0649,
+      "step": 54900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6750135093020972e-05,
+      "loss": 1.2345,
+      "step": 54905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6749577528243404e-05,
+      "loss": 1.1728,
+      "step": 54910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.674901992492212e-05,
+      "loss": 1.4114,
+      "step": 54915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.67484622830603e-05,
+      "loss": 1.1586,
+      "step": 54920
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6747904602661137e-05,
+      "loss": 1.4276,
+      "step": 54925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.67473468837278e-05,
+      "loss": 1.3436,
+      "step": 54930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.674678912626349e-05,
+      "loss": 1.5302,
+      "step": 54935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6746231330271384e-05,
+      "loss": 1.3666,
+      "step": 54940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.674567349575467e-05,
+      "loss": 1.2607,
+      "step": 54945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.674511562271653e-05,
+      "loss": 1.4032,
+      "step": 54950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6744557711160153e-05,
+      "loss": 1.7512,
+      "step": 54955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6743999761088724e-05,
+      "loss": 1.7153,
+      "step": 54960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.674344177250543e-05,
+      "loss": 1.2301,
+      "step": 54965
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6742883745413455e-05,
+      "loss": 1.4081,
+      "step": 54970
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6742325679815984e-05,
+      "loss": 1.2965,
+      "step": 54975
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.674176757571621e-05,
+      "loss": 1.3274,
+      "step": 54980
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6741209433117317e-05,
+      "loss": 1.5259,
+      "step": 54985
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6740651252022493e-05,
+      "loss": 1.3669,
+      "step": 54990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6740093032434924e-05,
+      "loss": 1.2444,
+      "step": 54995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.67395347743578e-05,
+      "loss": 1.0792,
+      "step": 55000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6738976477794305e-05,
+      "loss": 1.2227,
+      "step": 55005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.673841814274763e-05,
+      "loss": 1.2031,
+      "step": 55010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.673785976922097e-05,
+      "loss": 1.2786,
+      "step": 55015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6737301357217496e-05,
+      "loss": 1.2745,
+      "step": 55020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6736742906740415e-05,
+      "loss": 1.5017,
+      "step": 55025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6736184417792903e-05,
+      "loss": 1.0239,
+      "step": 55030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.673562589037816e-05,
+      "loss": 1.4405,
+      "step": 55035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.673506732449937e-05,
+      "loss": 1.2752,
+      "step": 55040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.673450872015972e-05,
+      "loss": 1.0896,
+      "step": 55045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6733950077362405e-05,
+      "loss": 1.2741,
+      "step": 55050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6733391396110614e-05,
+      "loss": 1.2301,
+      "step": 55055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6732832676407533e-05,
+      "loss": 1.1125,
+      "step": 55060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.673227391825636e-05,
+      "loss": 1.1206,
+      "step": 55065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6731715121660283e-05,
+      "loss": 1.3606,
+      "step": 55070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6731156286622488e-05,
+      "loss": 1.2159,
+      "step": 55075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6730597413146172e-05,
+      "loss": 1.1151,
+      "step": 55080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6730038501234523e-05,
+      "loss": 1.0875,
+      "step": 55085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6729479550890736e-05,
+      "loss": 1.3782,
+      "step": 55090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6728920562118003e-05,
+      "loss": 1.4735,
+      "step": 55095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.672836153491951e-05,
+      "loss": 0.9457,
+      "step": 55100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.672780246929846e-05,
+      "loss": 1.0394,
+      "step": 55105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6727243365258034e-05,
+      "loss": 1.2598,
+      "step": 55110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6726684222801432e-05,
+      "loss": 0.8006,
+      "step": 55115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6726125041931845e-05,
+      "loss": 1.3939,
+      "step": 55120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6725565822652465e-05,
+      "loss": 1.3028,
+      "step": 55125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.672500656496649e-05,
+      "loss": 1.3506,
+      "step": 55130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6724447268877108e-05,
+      "loss": 1.1869,
+      "step": 55135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6723887934387515e-05,
+      "loss": 1.3402,
+      "step": 55140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.672332856150091e-05,
+      "loss": 1.2074,
+      "step": 55145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6722769150220477e-05,
+      "loss": 0.7907,
+      "step": 55150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.672220970054942e-05,
+      "loss": 1.075,
+      "step": 55155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6721650212490926e-05,
+      "loss": 1.1798,
+      "step": 55160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6721090686048197e-05,
+      "loss": 1.308,
+      "step": 55165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6720531121224426e-05,
+      "loss": 1.1698,
+      "step": 55170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6719971518022802e-05,
+      "loss": 1.1866,
+      "step": 55175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.671941187644653e-05,
+      "loss": 1.3591,
+      "step": 55180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.671885219649881e-05,
+      "loss": 1.3558,
+      "step": 55185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6718292478182818e-05,
+      "loss": 1.3578,
+      "step": 55190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.671773272150177e-05,
+      "loss": 1.4512,
+      "step": 55195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.671717292645885e-05,
+      "loss": 0.8912,
+      "step": 55200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6716613093057263e-05,
+      "loss": 1.1479,
+      "step": 55205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6716053221300204e-05,
+      "loss": 1.3201,
+      "step": 55210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6715493311190866e-05,
+      "loss": 1.1782,
+      "step": 55215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.671493336273245e-05,
+      "loss": 1.0428,
+      "step": 55220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6714373375928153e-05,
+      "loss": 1.0316,
+      "step": 55225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6713813350781173e-05,
+      "loss": 1.1986,
+      "step": 55230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.671325328729471e-05,
+      "loss": 1.3374,
+      "step": 55235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6712693185471956e-05,
+      "loss": 1.2782,
+      "step": 55240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6712133045316113e-05,
+      "loss": 1.135,
+      "step": 55245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6711572866830382e-05,
+      "loss": 1.2012,
+      "step": 55250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.671101265001796e-05,
+      "loss": 1.6094,
+      "step": 55255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6710452394882048e-05,
+      "loss": 1.12,
+      "step": 55260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6709892101425838e-05,
+      "loss": 1.2429,
+      "step": 55265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6709331769652538e-05,
+      "loss": 1.3621,
+      "step": 55270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.670877139956535e-05,
+      "loss": 1.2955,
+      "step": 55275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6708210991167463e-05,
+      "loss": 1.1439,
+      "step": 55280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6707650544462085e-05,
+      "loss": 0.8871,
+      "step": 55285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6707090059452417e-05,
+      "loss": 1.3488,
+      "step": 55290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6706529536141657e-05,
+      "loss": 1.6126,
+      "step": 55295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6705968974533003e-05,
+      "loss": 0.9887,
+      "step": 55300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6705408374629662e-05,
+      "loss": 1.1518,
+      "step": 55305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6704847736434827e-05,
+      "loss": 1.1,
+      "step": 55310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.670428705995171e-05,
+      "loss": 1.6816,
+      "step": 55315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.670372634518351e-05,
+      "loss": 1.1816,
+      "step": 55320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6703165592133422e-05,
+      "loss": 1.2425,
+      "step": 55325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.670260480080466e-05,
+      "loss": 1.2206,
+      "step": 55330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6702043971200413e-05,
+      "loss": 1.694,
+      "step": 55335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6701483103323894e-05,
+      "loss": 1.0586,
+      "step": 55340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.67009221971783e-05,
+      "loss": 1.4209,
+      "step": 55345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6700361252766836e-05,
+      "loss": 1.3175,
+      "step": 55350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6699800270092707e-05,
+      "loss": 1.1099,
+      "step": 55355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6699239249159113e-05,
+      "loss": 1.1241,
+      "step": 55360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.669867818996926e-05,
+      "loss": 1.3162,
+      "step": 55365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6698117092526354e-05,
+      "loss": 1.5197,
+      "step": 55370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6697555956833596e-05,
+      "loss": 1.3534,
+      "step": 55375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.669699478289419e-05,
+      "loss": 1.1663,
+      "step": 55380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.669643357071134e-05,
+      "loss": 1.1931,
+      "step": 55385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6695872320288257e-05,
+      "loss": 1.0332,
+      "step": 55390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.669531103162814e-05,
+      "loss": 1.4638,
+      "step": 55395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6694749704734196e-05,
+      "loss": 1.1808,
+      "step": 55400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6694188339609628e-05,
+      "loss": 1.0975,
+      "step": 55405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6693626936257646e-05,
+      "loss": 1.496,
+      "step": 55410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6693065494681457e-05,
+      "loss": 1.3932,
+      "step": 55415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.669250401488426e-05,
+      "loss": 1.1158,
+      "step": 55420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6691942496869266e-05,
+      "loss": 1.2709,
+      "step": 55425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6691380940639683e-05,
+      "loss": 1.0386,
+      "step": 55430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.669081934619871e-05,
+      "loss": 0.7695,
+      "step": 55435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6690257713549568e-05,
+      "loss": 1.1436,
+      "step": 55440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.668969604269545e-05,
+      "loss": 1.0778,
+      "step": 55445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.668913433363957e-05,
+      "loss": 1.1968,
+      "step": 55450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.668857258638514e-05,
+      "loss": 1.3306,
+      "step": 55455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6688010800935362e-05,
+      "loss": 1.5069,
+      "step": 55460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.668744897729344e-05,
+      "loss": 1.2014,
+      "step": 55465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6686887115462594e-05,
+      "loss": 1.2779,
+      "step": 55470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6686325215446023e-05,
+      "loss": 1.3543,
+      "step": 55475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6685763277246936e-05,
+      "loss": 1.1245,
+      "step": 55480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.668520130086855e-05,
+      "loss": 1.7469,
+      "step": 55485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6684639286314066e-05,
+      "loss": 1.0742,
+      "step": 55490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6684077233586695e-05,
+      "loss": 1.1416,
+      "step": 55495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6683515142689654e-05,
+      "loss": 0.9991,
+      "step": 55500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.668295301362614e-05,
+      "loss": 1.1535,
+      "step": 55505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.668239084639937e-05,
+      "loss": 1.1804,
+      "step": 55510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.668182864101256e-05,
+      "loss": 1.1954,
+      "step": 55515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6681266397468905e-05,
+      "loss": 1.0841,
+      "step": 55520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6680704115771635e-05,
+      "loss": 1.3482,
+      "step": 55525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6680141795923947e-05,
+      "loss": 1.2825,
+      "step": 55530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6679579437929056e-05,
+      "loss": 1.3918,
+      "step": 55535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6679017041790177e-05,
+      "loss": 1.3628,
+      "step": 55540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6678454607510516e-05,
+      "loss": 1.2562,
+      "step": 55545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6677892135093288e-05,
+      "loss": 0.9231,
+      "step": 55550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.66773296245417e-05,
+      "loss": 1.0552,
+      "step": 55555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6676767075858972e-05,
+      "loss": 1.5246,
+      "step": 55560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6676204489048315e-05,
+      "loss": 1.1262,
+      "step": 55565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.667564186411294e-05,
+      "loss": 1.1892,
+      "step": 55570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.667507920105605e-05,
+      "loss": 1.5072,
+      "step": 55575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.667451649988088e-05,
+      "loss": 1.1453,
+      "step": 55580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.667395376059062e-05,
+      "loss": 1.6164,
+      "step": 55585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6673390983188504e-05,
+      "loss": 1.5422,
+      "step": 55590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.667282816767773e-05,
+      "loss": 1.222,
+      "step": 55595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.667226531406152e-05,
+      "loss": 1.1455,
+      "step": 55600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.667170242234309e-05,
+      "loss": 1.4634,
+      "step": 55605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6671139492525644e-05,
+      "loss": 1.1407,
+      "step": 55610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6670576524612407e-05,
+      "loss": 1.2556,
+      "step": 55615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.667001351860659e-05,
+      "loss": 1.3347,
+      "step": 55620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6669450474511408e-05,
+      "loss": 1.2339,
+      "step": 55625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.666888739233008e-05,
+      "loss": 1.6526,
+      "step": 55630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6668324272065814e-05,
+      "loss": 1.3,
+      "step": 55635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6667761113721827e-05,
+      "loss": 1.2749,
+      "step": 55640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6667197917301344e-05,
+      "loss": 1.1273,
+      "step": 55645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.666663468280757e-05,
+      "loss": 1.1724,
+      "step": 55650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6666071410243732e-05,
+      "loss": 1.1996,
+      "step": 55655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6665508099613036e-05,
+      "loss": 1.3707,
+      "step": 55660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6664944750918707e-05,
+      "loss": 1.561,
+      "step": 55665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6664381364163957e-05,
+      "loss": 1.4748,
+      "step": 55670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6663817939352004e-05,
+      "loss": 1.2483,
+      "step": 55675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.666325447648607e-05,
+      "loss": 1.6268,
+      "step": 55680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6662690975569365e-05,
+      "loss": 1.1107,
+      "step": 55685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.666212743660511e-05,
+      "loss": 1.0002,
+      "step": 55690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6661563859596528e-05,
+      "loss": 1.1542,
+      "step": 55695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6661000244546833e-05,
+      "loss": 1.3625,
+      "step": 55700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6660436591459244e-05,
+      "loss": 1.571,
+      "step": 55705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6659872900336978e-05,
+      "loss": 0.947,
+      "step": 55710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6659309171183254e-05,
+      "loss": 1.2963,
+      "step": 55715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.66587454040013e-05,
+      "loss": 1.2789,
+      "step": 55720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.665818159879432e-05,
+      "loss": 1.4849,
+      "step": 55725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6657617755565543e-05,
+      "loss": 1.0648,
+      "step": 55730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6657053874318194e-05,
+      "loss": 1.0754,
+      "step": 55735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6656489955055485e-05,
+      "loss": 1.2497,
+      "step": 55740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6655925997780637e-05,
+      "loss": 1.4869,
+      "step": 55745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.665536200249687e-05,
+      "loss": 1.2648,
+      "step": 55750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6654797969207404e-05,
+      "loss": 1.4725,
+      "step": 55755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6654233897915467e-05,
+      "loss": 1.46,
+      "step": 55760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6653669788624274e-05,
+      "loss": 1.3656,
+      "step": 55765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.665310564133705e-05,
+      "loss": 1.283,
+      "step": 55770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6652541456057012e-05,
+      "loss": 1.2904,
+      "step": 55775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6651977232787384e-05,
+      "loss": 1.5503,
+      "step": 55780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.665141297153139e-05,
+      "loss": 1.2388,
+      "step": 55785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6650848672292247e-05,
+      "loss": 1.4125,
+      "step": 55790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6650284335073185e-05,
+      "loss": 1.1688,
+      "step": 55795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.664971995987742e-05,
+      "loss": 1.1896,
+      "step": 55800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.664915554670818e-05,
+      "loss": 1.0828,
+      "step": 55805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.664859109556868e-05,
+      "loss": 1.2328,
+      "step": 55810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6648026606462155e-05,
+      "loss": 1.406,
+      "step": 55815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.664746207939182e-05,
+      "loss": 1.2293,
+      "step": 55820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.66468975143609e-05,
+      "loss": 1.2232,
+      "step": 55825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6646332911372622e-05,
+      "loss": 1.192,
+      "step": 55830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6645768270430208e-05,
+      "loss": 1.1831,
+      "step": 55835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6645203591536883e-05,
+      "loss": 1.2728,
+      "step": 55840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.664463887469587e-05,
+      "loss": 1.1617,
+      "step": 55845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6644074119910396e-05,
+      "loss": 1.2943,
+      "step": 55850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6643509327183686e-05,
+      "loss": 1.4974,
+      "step": 55855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6642944496518964e-05,
+      "loss": 1.0352,
+      "step": 55860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6642379627919453e-05,
+      "loss": 1.0183,
+      "step": 55865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6641814721388385e-05,
+      "loss": 1.0788,
+      "step": 55870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6641249776928984e-05,
+      "loss": 1.3413,
+      "step": 55875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.664068479454447e-05,
+      "loss": 1.5619,
+      "step": 55880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.664011977423808e-05,
+      "loss": 1.0377,
+      "step": 55885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.663955471601303e-05,
+      "loss": 1.2567,
+      "step": 55890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6638989619872556e-05,
+      "loss": 1.7478,
+      "step": 55895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6638424485819876e-05,
+      "loss": 1.3636,
+      "step": 55900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6637859313858223e-05,
+      "loss": 1.3458,
+      "step": 55905
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6637294103990826e-05,
+      "loss": 1.2387,
+      "step": 55910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6636728856220906e-05,
+      "loss": 1.0182,
+      "step": 55915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6636163570551697e-05,
+      "loss": 1.2289,
+      "step": 55920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6635598246986423e-05,
+      "loss": 1.0062,
+      "step": 55925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.663503288552831e-05,
+      "loss": 0.9353,
+      "step": 55930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.66344674861806e-05,
+      "loss": 1.3683,
+      "step": 55935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6633902048946506e-05,
+      "loss": 1.3156,
+      "step": 55940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6633336573829264e-05,
+      "loss": 1.1129,
+      "step": 55945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6632771060832106e-05,
+      "loss": 1.3268,
+      "step": 55950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6632205509958253e-05,
+      "loss": 1.2527,
+      "step": 55955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.663163992121094e-05,
+      "loss": 0.7033,
+      "step": 55960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.66310742945934e-05,
+      "loss": 0.9825,
+      "step": 55965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6630508630108854e-05,
+      "loss": 1.439,
+      "step": 55970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6629942927760543e-05,
+      "loss": 1.1083,
+      "step": 55975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6629377187551686e-05,
+      "loss": 1.0182,
+      "step": 55980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6628811409485524e-05,
+      "loss": 1.1906,
+      "step": 55985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6628245593565284e-05,
+      "loss": 0.9677,
+      "step": 55990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6627679739794195e-05,
+      "loss": 1.1794,
+      "step": 55995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.662711384817549e-05,
+      "loss": 1.1597,
+      "step": 56000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.66265479187124e-05,
+      "loss": 1.2941,
+      "step": 56005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6625981951408156e-05,
+      "loss": 1.3391,
+      "step": 56010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6625415946265994e-05,
+      "loss": 1.0821,
+      "step": 56015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6624849903289143e-05,
+      "loss": 1.2236,
+      "step": 56020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6624283822480836e-05,
+      "loss": 1.5577,
+      "step": 56025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6623717703844306e-05,
+      "loss": 1.7515,
+      "step": 56030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6623151547382784e-05,
+      "loss": 1.3473,
+      "step": 56035
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6622585353099502e-05,
+      "loss": 1.1657,
+      "step": 56040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.66220191209977e-05,
+      "loss": 1.0745,
+      "step": 56045
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6621452851080605e-05,
+      "loss": 1.3735,
+      "step": 56050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6620886543351453e-05,
+      "loss": 1.3001,
+      "step": 56055
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6620320197813478e-05,
+      "loss": 1.2839,
+      "step": 56060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6619753814469915e-05,
+      "loss": 1.2797,
+      "step": 56065
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6619187393323994e-05,
+      "loss": 1.0543,
+      "step": 56070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6618620934378956e-05,
+      "loss": 0.98,
+      "step": 56075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6618054437638034e-05,
+      "loss": 1.4415,
+      "step": 56080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6617487903104457e-05,
+      "loss": 1.4291,
+      "step": 56085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6616921330781468e-05,
+      "loss": 1.3082,
+      "step": 56090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6616354720672298e-05,
+      "loss": 1.0779,
+      "step": 56095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6615788072780183e-05,
+      "loss": 1.1363,
+      "step": 56100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6615221387108362e-05,
+      "loss": 1.0511,
+      "step": 56105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6614654663660068e-05,
+      "loss": 0.9696,
+      "step": 56110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.661408790243854e-05,
+      "loss": 1.2048,
+      "step": 56115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.661352110344701e-05,
+      "loss": 1.3585,
+      "step": 56120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6612954266688716e-05,
+      "loss": 0.9004,
+      "step": 56125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6612387392166898e-05,
+      "loss": 1.3325,
+      "step": 56130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6611820479884792e-05,
+      "loss": 1.1122,
+      "step": 56135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6611253529845637e-05,
+      "loss": 1.2065,
+      "step": 56140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6610686542052668e-05,
+      "loss": 1.0763,
+      "step": 56145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.661011951650912e-05,
+      "loss": 1.0038,
+      "step": 56150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6609552453218233e-05,
+      "loss": 1.038,
+      "step": 56155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6608985352183254e-05,
+      "loss": 1.3524,
+      "step": 56160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6608418213407408e-05,
+      "loss": 1.0097,
+      "step": 56165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6607851036893943e-05,
+      "loss": 1.1158,
+      "step": 56170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6607283822646092e-05,
+      "loss": 1.1454,
+      "step": 56175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6606716570667095e-05,
+      "loss": 1.2965,
+      "step": 56180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.66061492809602e-05,
+      "loss": 1.116,
+      "step": 56185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.660558195352863e-05,
+      "loss": 0.9751,
+      "step": 56190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.660501458837564e-05,
+      "loss": 1.2223,
+      "step": 56195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6604447185504466e-05,
+      "loss": 1.4469,
+      "step": 56200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.660387974491834e-05,
+      "loss": 1.1872,
+      "step": 56205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6603312266620514e-05,
+      "loss": 1.2928,
+      "step": 56210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.660274475061422e-05,
+      "loss": 1.2296,
+      "step": 56215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6602177196902704e-05,
+      "loss": 1.2201,
+      "step": 56220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6601609605489205e-05,
+      "loss": 1.101,
+      "step": 56225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6601041976376967e-05,
+      "loss": 1.5543,
+      "step": 56230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6600474309569224e-05,
+      "loss": 1.3273,
+      "step": 56235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6599906605069224e-05,
+      "loss": 1.5435,
+      "step": 56240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.659933886288021e-05,
+      "loss": 1.1122,
+      "step": 56245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.659877108300542e-05,
+      "loss": 1.1275,
+      "step": 56250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.65982032654481e-05,
+      "loss": 0.9349,
+      "step": 56255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6597635410211484e-05,
+      "loss": 1.1707,
+      "step": 56260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6597067517298827e-05,
+      "loss": 1.1525,
+      "step": 56265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6596499586713366e-05,
+      "loss": 1.0921,
+      "step": 56270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.659593161845834e-05,
+      "loss": 1.2078,
+      "step": 56275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6595363612537005e-05,
+      "loss": 0.9448,
+      "step": 56280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6594795568952592e-05,
+      "loss": 1.1977,
+      "step": 56285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.659422748770835e-05,
+      "loss": 1.2815,
+      "step": 56290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.659365936880752e-05,
+      "loss": 1.0771,
+      "step": 56295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.659309121225335e-05,
+      "loss": 1.1428,
+      "step": 56300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6592523018049083e-05,
+      "loss": 1.4319,
+      "step": 56305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6591954786197964e-05,
+      "loss": 1.1716,
+      "step": 56310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6591386516703237e-05,
+      "loss": 1.1817,
+      "step": 56315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6590818209568148e-05,
+      "loss": 1.2274,
+      "step": 56320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6590249864795943e-05,
+      "loss": 1.3644,
+      "step": 56325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6589681482389864e-05,
+      "loss": 1.2528,
+      "step": 56330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.658911306235316e-05,
+      "loss": 1.4678,
+      "step": 56335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.658854460468908e-05,
+      "loss": 1.324,
+      "step": 56340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6587976109400867e-05,
+      "loss": 1.0384,
+      "step": 56345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6587407576491765e-05,
+      "loss": 1.3868,
+      "step": 56350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.658683900596502e-05,
+      "loss": 1.5274,
+      "step": 56355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6586270397823883e-05,
+      "loss": 1.4709,
+      "step": 56360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.65857017520716e-05,
+      "loss": 1.0527,
+      "step": 56365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.658513306871142e-05,
+      "loss": 1.1515,
+      "step": 56370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6584564347746587e-05,
+      "loss": 1.3794,
+      "step": 56375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.658399558918035e-05,
+      "loss": 1.5056,
+      "step": 56380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6583426793015954e-05,
+      "loss": 1.148,
+      "step": 56385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6582857959256653e-05,
+      "loss": 0.9169,
+      "step": 56390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6582289087905695e-05,
+      "loss": 1.2301,
+      "step": 56395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.658172017896632e-05,
+      "loss": 1.0754,
+      "step": 56400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6581151232441784e-05,
+      "loss": 1.2551,
+      "step": 56405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6580582248335338e-05,
+      "loss": 1.2112,
+      "step": 56410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6580013226650227e-05,
+      "loss": 1.228,
+      "step": 56415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6579444167389702e-05,
+      "loss": 1.0771,
+      "step": 56420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.657887507055701e-05,
+      "loss": 1.2827,
+      "step": 56425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6578305936155406e-05,
+      "loss": 1.35,
+      "step": 56430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.657773676418814e-05,
+      "loss": 1.0847,
+      "step": 56435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.657716755465845e-05,
+      "loss": 1.2628,
+      "step": 56440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.65765983075696e-05,
+      "loss": 1.169,
+      "step": 56445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6576029022924836e-05,
+      "loss": 1.1525,
+      "step": 56450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.657545970072741e-05,
+      "loss": 1.1282,
+      "step": 56455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6574890340980572e-05,
+      "loss": 1.5153,
+      "step": 56460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6574320943687575e-05,
+      "loss": 1.199,
+      "step": 56465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6573751508851668e-05,
+      "loss": 1.1389,
+      "step": 56470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6573182036476105e-05,
+      "loss": 1.2781,
+      "step": 56475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6572612526564134e-05,
+      "loss": 1.4014,
+      "step": 56480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6572042979119016e-05,
+      "loss": 1.3577,
+      "step": 56485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6571473394143995e-05,
+      "loss": 1.2823,
+      "step": 56490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6570903771642325e-05,
+      "loss": 1.103,
+      "step": 56495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6570334111617264e-05,
+      "loss": 1.1172,
+      "step": 56500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.656976441407206e-05,
+      "loss": 1.1184,
+      "step": 56505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6569194679009967e-05,
+      "loss": 1.2541,
+      "step": 56510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6568624906434238e-05,
+      "loss": 1.2083,
+      "step": 56515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6568055096348126e-05,
+      "loss": 1.2156,
+      "step": 56520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.656748524875489e-05,
+      "loss": 0.925,
+      "step": 56525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.656691536365778e-05,
+      "loss": 1.1082,
+      "step": 56530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.656634544106005e-05,
+      "loss": 1.6962,
+      "step": 56535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.656577548096496e-05,
+      "loss": 0.9301,
+      "step": 56540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.656520548337576e-05,
+      "loss": 1.2824,
+      "step": 56545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.65646354482957e-05,
+      "loss": 1.078,
+      "step": 56550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6564065375728047e-05,
+      "loss": 1.0487,
+      "step": 56555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6563495265676042e-05,
+      "loss": 1.1124,
+      "step": 56560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6562925118142955e-05,
+      "loss": 1.3526,
+      "step": 56565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6562354933132037e-05,
+      "loss": 1.6661,
+      "step": 56570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6561784710646542e-05,
+      "loss": 1.5595,
+      "step": 56575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6561214450689726e-05,
+      "loss": 1.0886,
+      "step": 56580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6560644153264848e-05,
+      "loss": 1.0645,
+      "step": 56585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6560073818375157e-05,
+      "loss": 1.028,
+      "step": 56590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6559503446023922e-05,
+      "loss": 1.2463,
+      "step": 56595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6558933036214394e-05,
+      "loss": 1.2744,
+      "step": 56600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.655836258894983e-05,
+      "loss": 0.9653,
+      "step": 56605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6557792104233487e-05,
+      "loss": 1.1689,
+      "step": 56610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6557221582068626e-05,
+      "loss": 1.2618,
+      "step": 56615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6556651022458502e-05,
+      "loss": 1.1844,
+      "step": 56620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6556080425406374e-05,
+      "loss": 0.9441,
+      "step": 56625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.65555097909155e-05,
+      "loss": 1.1209,
+      "step": 56630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6554939118989138e-05,
+      "loss": 1.5717,
+      "step": 56635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.655436840963055e-05,
+      "loss": 1.2525,
+      "step": 56640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6553797662842993e-05,
+      "loss": 1.4058,
+      "step": 56645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.655322687862973e-05,
+      "loss": 1.3877,
+      "step": 56650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.655265605699401e-05,
+      "loss": 0.8989,
+      "step": 56655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6552085197939104e-05,
+      "loss": 1.1339,
+      "step": 56660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6551514301468263e-05,
+      "loss": 1.0308,
+      "step": 56665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6550943367584756e-05,
+      "loss": 1.0858,
+      "step": 56670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.655037239629184e-05,
+      "loss": 1.2745,
+      "step": 56675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.654980138759277e-05,
+      "loss": 1.0076,
+      "step": 56680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6549230341490812e-05,
+      "loss": 1.2785,
+      "step": 56685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.654865925798923e-05,
+      "loss": 1.6648,
+      "step": 56690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.654808813709128e-05,
+      "loss": 1.1519,
+      "step": 56695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6547516978800222e-05,
+      "loss": 1.2901,
+      "step": 56700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6546945783119323e-05,
+      "loss": 1.1365,
+      "step": 56705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.654637455005184e-05,
+      "loss": 1.1228,
+      "step": 56710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6545803279601037e-05,
+      "loss": 1.2012,
+      "step": 56715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.654523197177018e-05,
+      "loss": 1.1034,
+      "step": 56720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6544660626562528e-05,
+      "loss": 1.4367,
+      "step": 56725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.654408924398134e-05,
+      "loss": 1.1971,
+      "step": 56730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6543517824029885e-05,
+      "loss": 1.3976,
+      "step": 56735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6542946366711425e-05,
+      "loss": 1.1725,
+      "step": 56740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.654237487202922e-05,
+      "loss": 1.4333,
+      "step": 56745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6541803339986536e-05,
+      "loss": 1.0952,
+      "step": 56750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.654123177058664e-05,
+      "loss": 1.2668,
+      "step": 56755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6540660163832787e-05,
+      "loss": 1.0253,
+      "step": 56760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.654008851972825e-05,
+      "loss": 1.0435,
+      "step": 56765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6539516838276287e-05,
+      "loss": 1.0229,
+      "step": 56770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.653894511948017e-05,
+      "loss": 1.4129,
+      "step": 56775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6538373363343155e-05,
+      "loss": 1.3944,
+      "step": 56780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6537801569868516e-05,
+      "loss": 1.4077,
+      "step": 56785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6537229739059507e-05,
+      "loss": 1.2186,
+      "step": 56790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6536657870919405e-05,
+      "loss": 1.0416,
+      "step": 56795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.653608596545147e-05,
+      "loss": 1.5589,
+      "step": 56800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.653551402265897e-05,
+      "loss": 1.2738,
+      "step": 56805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.653494204254517e-05,
+      "loss": 0.9693,
+      "step": 56810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.653437002511333e-05,
+      "loss": 1.1404,
+      "step": 56815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.653379797036673e-05,
+      "loss": 1.4269,
+      "step": 56820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6533225878308624e-05,
+      "loss": 1.4694,
+      "step": 56825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.653265374894229e-05,
+      "loss": 0.9491,
+      "step": 56830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6532081582270985e-05,
+      "loss": 1.774,
+      "step": 56835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.653150937829798e-05,
+      "loss": 1.264,
+      "step": 56840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6530937137026545e-05,
+      "loss": 1.4096,
+      "step": 56845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.653036485845995e-05,
+      "loss": 1.4467,
+      "step": 56850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6529792542601453e-05,
+      "loss": 1.3566,
+      "step": 56855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.652922018945433e-05,
+      "loss": 1.1384,
+      "step": 56860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.652864779902185e-05,
+      "loss": 1.4689,
+      "step": 56865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.652807537130728e-05,
+      "loss": 1.1611,
+      "step": 56870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.652750290631389e-05,
+      "loss": 1.2315,
+      "step": 56875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6526930404044943e-05,
+      "loss": 1.022,
+      "step": 56880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6526357864503717e-05,
+      "loss": 1.2307,
+      "step": 56885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6525785287693476e-05,
+      "loss": 1.3231,
+      "step": 56890
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.652521267361749e-05,
+      "loss": 1.6229,
+      "step": 56895
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.652464002227903e-05,
+      "loss": 1.0879,
+      "step": 56900
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6524067333681365e-05,
+      "loss": 1.1511,
+      "step": 56905
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.652349460782777e-05,
+      "loss": 1.3641,
+      "step": 56910
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6522921844721508e-05,
+      "loss": 1.0717,
+      "step": 56915
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6522349044365858e-05,
+      "loss": 1.2647,
+      "step": 56920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6521776206764086e-05,
+      "loss": 1.272,
+      "step": 56925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.652120333191946e-05,
+      "loss": 1.1977,
+      "step": 56930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6520630419835263e-05,
+      "loss": 1.0346,
+      "step": 56935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6520057470514754e-05,
+      "loss": 1.191,
+      "step": 56940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.651948448396121e-05,
+      "loss": 1.1251,
+      "step": 56945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6518911460177905e-05,
+      "loss": 1.1663,
+      "step": 56950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6518338399168106e-05,
+      "loss": 1.0709,
+      "step": 56955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6517765300935093e-05,
+      "loss": 1.178,
+      "step": 56960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6517192165482133e-05,
+      "loss": 1.1284,
+      "step": 56965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6516618992812498e-05,
+      "loss": 1.1183,
+      "step": 56970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.651604578292947e-05,
+      "loss": 1.2352,
+      "step": 56975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.651547253583631e-05,
+      "loss": 1.3623,
+      "step": 56980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6514899251536297e-05,
+      "loss": 1.2241,
+      "step": 56985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.651432593003271e-05,
+      "loss": 1.0759,
+      "step": 56990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6513752571328815e-05,
+      "loss": 0.785,
+      "step": 56995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6513179175427888e-05,
+      "loss": 1.7194,
+      "step": 57000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.651260574233321e-05,
+      "loss": 1.2059,
+      "step": 57005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6512032272048048e-05,
+      "loss": 1.1044,
+      "step": 57010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6511458764575677e-05,
+      "loss": 1.5603,
+      "step": 57015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6510885219919376e-05,
+      "loss": 1.3115,
+      "step": 57020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6510311638082418e-05,
+      "loss": 1.1464,
+      "step": 57025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6509738019068077e-05,
+      "loss": 1.1398,
+      "step": 57030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6509164362879634e-05,
+      "loss": 0.977,
+      "step": 57035
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.650859066952036e-05,
+      "loss": 1.0502,
+      "step": 57040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.650801693899353e-05,
+      "loss": 1.037,
+      "step": 57045
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6507443171302427e-05,
+      "loss": 1.2882,
+      "step": 57050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6506869366450324e-05,
+      "loss": 0.9853,
+      "step": 57055
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6506295524440498e-05,
+      "loss": 1.077,
+      "step": 57060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.650572164527622e-05,
+      "loss": 1.2014,
+      "step": 57065
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6505147728960777e-05,
+      "loss": 1.2004,
+      "step": 57070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6504573775497438e-05,
+      "loss": 1.0774,
+      "step": 57075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6503999784889484e-05,
+      "loss": 1.3006,
+      "step": 57080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6503425757140194e-05,
+      "loss": 1.1933,
+      "step": 57085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.650285169225285e-05,
+      "loss": 1.0132,
+      "step": 57090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.650227759023072e-05,
+      "loss": 1.1619,
+      "step": 57095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.650170345107709e-05,
+      "loss": 1.0575,
+      "step": 57100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.650112927479523e-05,
+      "loss": 1.1132,
+      "step": 57105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6500555061388434e-05,
+      "loss": 1.1873,
+      "step": 57110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6499980810859968e-05,
+      "loss": 1.0621,
+      "step": 57115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.649940652321312e-05,
+      "loss": 1.3738,
+      "step": 57120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6498832198451156e-05,
+      "loss": 1.395,
+      "step": 57125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.649825783657737e-05,
+      "loss": 1.0956,
+      "step": 57130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.649768343759504e-05,
+      "loss": 1.3566,
+      "step": 57135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6497109001507437e-05,
+      "loss": 1.2173,
+      "step": 57140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6496534528317848e-05,
+      "loss": 1.3034,
+      "step": 57145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6495960018029557e-05,
+      "loss": 1.3031,
+      "step": 57150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6495385470645838e-05,
+      "loss": 1.6115,
+      "step": 57155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6494810886169973e-05,
+      "loss": 1.385,
+      "step": 57160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6494236264605246e-05,
+      "loss": 1.1323,
+      "step": 57165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6493661605954934e-05,
+      "loss": 1.2089,
+      "step": 57170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6493086910222325e-05,
+      "loss": 1.4054,
+      "step": 57175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6492512177410696e-05,
+      "loss": 1.2086,
+      "step": 57180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6491937407523335e-05,
+      "loss": 1.3957,
+      "step": 57185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6491362600563512e-05,
+      "loss": 1.4917,
+      "step": 57190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.649078775653452e-05,
+      "loss": 1.3948,
+      "step": 57195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6490212875439643e-05,
+      "loss": 1.0443,
+      "step": 57200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6489637957282153e-05,
+      "loss": 1.6515,
+      "step": 57205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.648906300206534e-05,
+      "loss": 1.3593,
+      "step": 57210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6488488009792494e-05,
+      "loss": 1.252,
+      "step": 57215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.648791298046689e-05,
+      "loss": 1.2061,
+      "step": 57220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.648733791409181e-05,
+      "loss": 1.0394,
+      "step": 57225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6486762810670543e-05,
+      "loss": 1.2054,
+      "step": 57230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.648618767020637e-05,
+      "loss": 1.3474,
+      "step": 57235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.648561249270258e-05,
+      "loss": 1.3387,
+      "step": 57240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.648503727816245e-05,
+      "loss": 1.1063,
+      "step": 57245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.648446202658927e-05,
+      "loss": 1.2621,
+      "step": 57250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6483886737986327e-05,
+      "loss": 0.849,
+      "step": 57255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.64833114123569e-05,
+      "loss": 1.4546,
+      "step": 57260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6482736049704282e-05,
+      "loss": 1.2699,
+      "step": 57265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6482160650031752e-05,
+      "loss": 1.0314,
+      "step": 57270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6481585213342596e-05,
+      "loss": 1.4527,
+      "step": 57275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6481009739640105e-05,
+      "loss": 0.9513,
+      "step": 57280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6480434228927563e-05,
+      "loss": 1.0068,
+      "step": 57285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6479858681208253e-05,
+      "loss": 1.1458,
+      "step": 57290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.647928309648547e-05,
+      "loss": 1.0169,
+      "step": 57295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.647870747476249e-05,
+      "loss": 0.9912,
+      "step": 57300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.647813181604261e-05,
+      "loss": 1.152,
+      "step": 57305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.647755612032911e-05,
+      "loss": 1.1107,
+      "step": 57310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6476980387625282e-05,
+      "loss": 1.2971,
+      "step": 57315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6476404617934412e-05,
+      "loss": 1.2426,
+      "step": 57320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6475828811259787e-05,
+      "loss": 1.1919,
+      "step": 57325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.64752529676047e-05,
+      "loss": 1.2718,
+      "step": 57330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6474677086972432e-05,
+      "loss": 1.2902,
+      "step": 57335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6474101169366276e-05,
+      "loss": 1.4839,
+      "step": 57340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6473525214789523e-05,
+      "loss": 1.3725,
+      "step": 57345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6472949223245457e-05,
+      "loss": 0.9687,
+      "step": 57350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.647237319473737e-05,
+      "loss": 1.0546,
+      "step": 57355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6471797129268553e-05,
+      "loss": 1.3266,
+      "step": 57360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6471221026842288e-05,
+      "loss": 0.97,
+      "step": 57365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6470644887461875e-05,
+      "loss": 1.1153,
+      "step": 57370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.64700687111306e-05,
+      "loss": 1.1168,
+      "step": 57375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.646949249785175e-05,
+      "loss": 1.0598,
+      "step": 57380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.646891624762862e-05,
+      "loss": 1.4994,
+      "step": 57385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6468339960464496e-05,
+      "loss": 1.3707,
+      "step": 57390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6467763636362674e-05,
+      "loss": 1.2279,
+      "step": 57395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6467187275326445e-05,
+      "loss": 1.3017,
+      "step": 57400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.646661087735909e-05,
+      "loss": 1.0107,
+      "step": 57405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.646603444246392e-05,
+      "loss": 1.1732,
+      "step": 57410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.646545797064421e-05,
+      "loss": 1.1345,
+      "step": 57415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6464881461903258e-05,
+      "loss": 1.3004,
+      "step": 57420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6464304916244357e-05,
+      "loss": 1.3204,
+      "step": 57425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6463728333670794e-05,
+      "loss": 1.6184,
+      "step": 57430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6463151714185866e-05,
+      "loss": 1.1883,
+      "step": 57435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.646257505779287e-05,
+      "loss": 1.1279,
+      "step": 57440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.646199836449509e-05,
+      "loss": 1.1643,
+      "step": 57445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.646142163429583e-05,
+      "loss": 1.2136,
+      "step": 57450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6460844867198373e-05,
+      "loss": 1.1253,
+      "step": 57455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6460268063206017e-05,
+      "loss": 1.1749,
+      "step": 57460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6459691222322055e-05,
+      "loss": 1.4348,
+      "step": 57465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6459114344549782e-05,
+      "loss": 1.3382,
+      "step": 57470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.645853742989249e-05,
+      "loss": 1.4469,
+      "step": 57475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.645796047835348e-05,
+      "loss": 1.2397,
+      "step": 57480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.645738348993604e-05,
+      "loss": 1.3096,
+      "step": 57485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6456806464643472e-05,
+      "loss": 1.0491,
+      "step": 57490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6456229402479058e-05,
+      "loss": 1.3897,
+      "step": 57495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6455652303446105e-05,
+      "loss": 1.097,
+      "step": 57500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.645507516754791e-05,
+      "loss": 1.2859,
+      "step": 57505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6454497994787757e-05,
+      "loss": 1.095,
+      "step": 57510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.645392078516895e-05,
+      "loss": 1.1994,
+      "step": 57515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.645334353869479e-05,
+      "loss": 1.1811,
+      "step": 57520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.645276625536856e-05,
+      "loss": 1.2469,
+      "step": 57525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6452188935193568e-05,
+      "loss": 1.2782,
+      "step": 57530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6451611578173103e-05,
+      "loss": 1.317,
+      "step": 57535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.645103418431047e-05,
+      "loss": 1.7453,
+      "step": 57540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.645045675360896e-05,
+      "loss": 1.3685,
+      "step": 57545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6449879286071872e-05,
+      "loss": 1.2561,
+      "step": 57550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6449301781702507e-05,
+      "loss": 1.1026,
+      "step": 57555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.644872424050416e-05,
+      "loss": 1.424,
+      "step": 57560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6448146662480126e-05,
+      "loss": 1.2196,
+      "step": 57565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.644756904763371e-05,
+      "loss": 1.2316,
+      "step": 57570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6446991395968203e-05,
+      "loss": 1.3161,
+      "step": 57575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.644641370748691e-05,
+      "loss": 1.4536,
+      "step": 57580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6445835982193127e-05,
+      "loss": 1.0226,
+      "step": 57585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6445258220090156e-05,
+      "loss": 1.3922,
+      "step": 57590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.644468042118129e-05,
+      "loss": 0.9974,
+      "step": 57595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6444102585469837e-05,
+      "loss": 1.193,
+      "step": 57600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6443524712959085e-05,
+      "loss": 1.1985,
+      "step": 57605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.644294680365235e-05,
+      "loss": 1.1674,
+      "step": 57610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6442368857552916e-05,
+      "loss": 1.1339,
+      "step": 57615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6441790874664095e-05,
+      "loss": 1.3284,
+      "step": 57620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6441212854989183e-05,
+      "loss": 1.416,
+      "step": 57625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.644063479853148e-05,
+      "loss": 1.2587,
+      "step": 57630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.644005670529429e-05,
+      "loss": 1.2354,
+      "step": 57635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.643947857528091e-05,
+      "loss": 1.2403,
+      "step": 57640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6438900408494642e-05,
+      "loss": 1.4092,
+      "step": 57645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6438322204938793e-05,
+      "loss": 1.1515,
+      "step": 57650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6437743964616663e-05,
+      "loss": 1.4332,
+      "step": 57655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6437165687531548e-05,
+      "loss": 1.2594,
+      "step": 57660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6436587373686758e-05,
+      "loss": 1.2912,
+      "step": 57665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6436009023085593e-05,
+      "loss": 1.3216,
+      "step": 57670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.643543063573135e-05,
+      "loss": 1.3931,
+      "step": 57675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6434852211627343e-05,
+      "loss": 1.0841,
+      "step": 57680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6434273750776864e-05,
+      "loss": 1.21,
+      "step": 57685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.643369525318322e-05,
+      "loss": 1.1801,
+      "step": 57690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.643311671884972e-05,
+      "loss": 1.0806,
+      "step": 57695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6432538147779663e-05,
+      "loss": 1.1749,
+      "step": 57700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.643195953997635e-05,
+      "loss": 1.2681,
+      "step": 57705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.643138089544309e-05,
+      "loss": 1.1428,
+      "step": 57710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.643080221418319e-05,
+      "loss": 1.255,
+      "step": 57715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6430223496199948e-05,
+      "loss": 1.13,
+      "step": 57720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.642964474149667e-05,
+      "loss": 1.5282,
+      "step": 57725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6429065950076663e-05,
+      "loss": 1.3641,
+      "step": 57730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6428487121943232e-05,
+      "loss": 1.2858,
+      "step": 57735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6427908257099684e-05,
+      "loss": 1.1636,
+      "step": 57740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6427329355549314e-05,
+      "loss": 1.1107,
+      "step": 57745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6426750417295443e-05,
+      "loss": 1.1368,
+      "step": 57750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6426171442341374e-05,
+      "loss": 1.2872,
+      "step": 57755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6425592430690404e-05,
+      "loss": 1.3327,
+      "step": 57760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6425013382345846e-05,
+      "loss": 1.2732,
+      "step": 57765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.642443429731101e-05,
+      "loss": 1.1474,
+      "step": 57770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6423855175589195e-05,
+      "loss": 1.4029,
+      "step": 57775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.642327601718371e-05,
+      "loss": 1.1151,
+      "step": 57780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6422696822097866e-05,
+      "loss": 1.082,
+      "step": 57785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.642211759033497e-05,
+      "loss": 1.2515,
+      "step": 57790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6421538321898327e-05,
+      "loss": 1.129,
+      "step": 57795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6420959016791248e-05,
+      "loss": 1.2186,
+      "step": 57800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.642037967501704e-05,
+      "loss": 1.0744,
+      "step": 57805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.641980029657901e-05,
+      "loss": 1.3067,
+      "step": 57810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6419220881480464e-05,
+      "loss": 1.5234,
+      "step": 57815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6418641429724718e-05,
+      "loss": 1.3297,
+      "step": 57820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6418061941315075e-05,
+      "loss": 1.2066,
+      "step": 57825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6417482416254847e-05,
+      "loss": 1.5361,
+      "step": 57830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6416902854547347e-05,
+      "loss": 1.2014,
+      "step": 57835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6416323256195872e-05,
+      "loss": 0.8815,
+      "step": 57840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.641574362120375e-05,
+      "loss": 1.3463,
+      "step": 57845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6415163949574272e-05,
+      "loss": 1.1398,
+      "step": 57850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6414584241310762e-05,
+      "loss": 1.3873,
+      "step": 57855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6414004496416523e-05,
+      "loss": 1.2114,
+      "step": 57860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6413424714894876e-05,
+      "loss": 1.2156,
+      "step": 57865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6412844896749116e-05,
+      "loss": 1.0793,
+      "step": 57870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6412265041982568e-05,
+      "loss": 1.368,
+      "step": 57875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6411685150598534e-05,
+      "loss": 1.177,
+      "step": 57880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.641110522260033e-05,
+      "loss": 1.2207,
+      "step": 57885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6410525257991267e-05,
+      "loss": 1.2109,
+      "step": 57890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6409945256774657e-05,
+      "loss": 1.142,
+      "step": 57895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6409365218953812e-05,
+      "loss": 1.1624,
+      "step": 57900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6408785144532047e-05,
+      "loss": 1.1478,
+      "step": 57905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.640820503351267e-05,
+      "loss": 1.3772,
+      "step": 57910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6407624885898994e-05,
+      "loss": 1.3026,
+      "step": 57915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6407044701694336e-05,
+      "loss": 1.1445,
+      "step": 57920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6406464480902005e-05,
+      "loss": 1.4722,
+      "step": 57925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6405884223525315e-05,
+      "loss": 1.0041,
+      "step": 57930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6405303929567585e-05,
+      "loss": 1.2949,
+      "step": 57935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.640472359903212e-05,
+      "loss": 1.5004,
+      "step": 57940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.640414323192224e-05,
+      "loss": 0.9114,
+      "step": 57945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6403562828241256e-05,
+      "loss": 1.3219,
+      "step": 57950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6402982387992487e-05,
+      "loss": 1.0792,
+      "step": 57955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6402401911179244e-05,
+      "loss": 1.1805,
+      "step": 57960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6401821397804843e-05,
+      "loss": 1.1306,
+      "step": 57965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.64012408478726e-05,
+      "loss": 1.3109,
+      "step": 57970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.640066026138582e-05,
+      "loss": 1.3368,
+      "step": 57975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.640007963834784e-05,
+      "loss": 1.3638,
+      "step": 57980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6399498978761955e-05,
+      "loss": 1.4948,
+      "step": 57985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.639891828263149e-05,
+      "loss": 1.1701,
+      "step": 57990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.639833754995976e-05,
+      "loss": 1.0898,
+      "step": 57995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6397756780750082e-05,
+      "loss": 1.4895,
+      "step": 58000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6397175975005773e-05,
+      "loss": 1.423,
+      "step": 58005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6396595132730145e-05,
+      "loss": 1.1588,
+      "step": 58010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.639601425392652e-05,
+      "loss": 1.1697,
+      "step": 58015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6395433338598215e-05,
+      "loss": 1.372,
+      "step": 58020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6394852386748542e-05,
+      "loss": 1.1876,
+      "step": 58025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.639427139838083e-05,
+      "loss": 1.004,
+      "step": 58030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6393690373498383e-05,
+      "loss": 1.3434,
+      "step": 58035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6393109312104524e-05,
+      "loss": 1.4057,
+      "step": 58040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6392528214202577e-05,
+      "loss": 1.2484,
+      "step": 58045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.639194707979585e-05,
+      "loss": 1.4434,
+      "step": 58050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6391365908887672e-05,
+      "loss": 1.425,
+      "step": 58055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6390784701481355e-05,
+      "loss": 1.1145,
+      "step": 58060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.639020345758022e-05,
+      "loss": 1.5672,
+      "step": 58065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.638962217718759e-05,
+      "loss": 1.3737,
+      "step": 58070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6389040860306775e-05,
+      "loss": 1.0815,
+      "step": 58075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6388459506941103e-05,
+      "loss": 1.2481,
+      "step": 58080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6387878117093895e-05,
+      "loss": 1.328,
+      "step": 58085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.638729669076846e-05,
+      "loss": 1.411,
+      "step": 58090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.638671522796813e-05,
+      "loss": 1.1434,
+      "step": 58095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.638613372869622e-05,
+      "loss": 0.9015,
+      "step": 58100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.638555219295605e-05,
+      "loss": 1.0953,
+      "step": 58105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6384970620750943e-05,
+      "loss": 1.64,
+      "step": 58110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.638438901208422e-05,
+      "loss": 1.2874,
+      "step": 58115
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6383807366959203e-05,
+      "loss": 1.2753,
+      "step": 58120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6383225685379214e-05,
+      "loss": 1.6225,
+      "step": 58125
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6382643967347566e-05,
+      "loss": 1.1259,
+      "step": 58130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6382062212867595e-05,
+      "loss": 1.4138,
+      "step": 58135
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6381480421942616e-05,
+      "loss": 1.1569,
+      "step": 58140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.638089859457595e-05,
+      "loss": 1.078,
+      "step": 58145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6380316730770916e-05,
+      "loss": 1.0564,
+      "step": 58150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.637973483053085e-05,
+      "loss": 1.2978,
+      "step": 58155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.637915289385906e-05,
+      "loss": 1.5414,
+      "step": 58160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6378570920758882e-05,
+      "loss": 1.2985,
+      "step": 58165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.637798891123363e-05,
+      "loss": 1.2497,
+      "step": 58170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.637740686528663e-05,
+      "loss": 1.1533,
+      "step": 58175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.637682478292121e-05,
+      "loss": 1.0831,
+      "step": 58180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6376242664140688e-05,
+      "loss": 0.7607,
+      "step": 58185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6375660508948393e-05,
+      "loss": 1.0087,
+      "step": 58190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.637507831734764e-05,
+      "loss": 1.5422,
+      "step": 58195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6374496089341772e-05,
+      "loss": 1.2073,
+      "step": 58200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6373913824934096e-05,
+      "loss": 1.1461,
+      "step": 58205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6373331524127944e-05,
+      "loss": 1.1572,
+      "step": 58210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6372749186926643e-05,
+      "loss": 1.5027,
+      "step": 58215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6372166813333517e-05,
+      "loss": 1.1547,
+      "step": 58220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.637158440335189e-05,
+      "loss": 1.5022,
+      "step": 58225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6371001956985087e-05,
+      "loss": 1.0892,
+      "step": 58230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.637041947423644e-05,
+      "loss": 1.0965,
+      "step": 58235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.636983695510927e-05,
+      "loss": 1.2261,
+      "step": 58240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6369254399606905e-05,
+      "loss": 1.4306,
+      "step": 58245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6368671807732674e-05,
+      "loss": 1.1585,
+      "step": 58250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6368089179489897e-05,
+      "loss": 1.175,
+      "step": 58255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6367506514881908e-05,
+      "loss": 1.1801,
+      "step": 58260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.636692381391203e-05,
+      "loss": 1.294,
+      "step": 58265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6366341076583595e-05,
+      "loss": 1.155,
+      "step": 58270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.636575830289993e-05,
+      "loss": 1.1595,
+      "step": 58275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.636517549286436e-05,
+      "loss": 1.3154,
+      "step": 58280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.636459264648021e-05,
+      "loss": 1.2182,
+      "step": 58285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.636400976375082e-05,
+      "loss": 1.1781,
+      "step": 58290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6363426844679508e-05,
+      "loss": 1.1572,
+      "step": 58295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6362843889269604e-05,
+      "loss": 1.4097,
+      "step": 58300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.636226089752444e-05,
+      "loss": 1.4476,
+      "step": 58305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6361677869447347e-05,
+      "loss": 1.316,
+      "step": 58310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.636109480504165e-05,
+      "loss": 1.3227,
+      "step": 58315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6360511704310678e-05,
+      "loss": 1.2053,
+      "step": 58320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6359928567257767e-05,
+      "loss": 1.2895,
+      "step": 58325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.635934539388624e-05,
+      "loss": 1.2125,
+      "step": 58330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6358762184199433e-05,
+      "loss": 0.8767,
+      "step": 58335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6358178938200674e-05,
+      "loss": 1.1578,
+      "step": 58340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6357595655893292e-05,
+      "loss": 1.336,
+      "step": 58345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6357012337280617e-05,
+      "loss": 1.1756,
+      "step": 58350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6356428982365984e-05,
+      "loss": 0.9284,
+      "step": 58355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6355845591152727e-05,
+      "loss": 1.005,
+      "step": 58360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.635526216364417e-05,
+      "loss": 1.4847,
+      "step": 58365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6354678699843643e-05,
+      "loss": 1.0648,
+      "step": 58370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6354095199754492e-05,
+      "loss": 1.2349,
+      "step": 58375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6353511663380035e-05,
+      "loss": 1.0955,
+      "step": 58380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.635292809072361e-05,
+      "loss": 1.076,
+      "step": 58385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6352344481788545e-05,
+      "loss": 1.0564,
+      "step": 58390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.635176083657818e-05,
+      "loss": 1.5753,
+      "step": 58395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6351177155095847e-05,
+      "loss": 1.0217,
+      "step": 58400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6350593437344872e-05,
+      "loss": 1.2019,
+      "step": 58405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6350009683328593e-05,
+      "loss": 1.1723,
+      "step": 58410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6349425893050345e-05,
+      "loss": 1.2458,
+      "step": 58415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.634884206651346e-05,
+      "loss": 1.1602,
+      "step": 58420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6348258203721272e-05,
+      "loss": 1.265,
+      "step": 58425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6347674304677116e-05,
+      "loss": 0.8993,
+      "step": 58430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6347090369384323e-05,
+      "loss": 1.3719,
+      "step": 58435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6346506397846232e-05,
+      "loss": 1.1231,
+      "step": 58440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6345922390066176e-05,
+      "loss": 1.0083,
+      "step": 58445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.634533834604749e-05,
+      "loss": 1.1975,
+      "step": 58450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6344754265793508e-05,
+      "loss": 1.3555,
+      "step": 58455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6344170149307568e-05,
+      "loss": 1.1826,
+      "step": 58460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6343585996593002e-05,
+      "loss": 1.1697,
+      "step": 58465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.634300180765315e-05,
+      "loss": 1.8653,
+      "step": 58470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6342417582491345e-05,
+      "loss": 1.4961,
+      "step": 58475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6341833321110926e-05,
+      "loss": 0.9388,
+      "step": 58480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6341249023515226e-05,
+      "loss": 1.238,
+      "step": 58485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6340664689707584e-05,
+      "loss": 1.1566,
+      "step": 58490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6340080319691337e-05,
+      "loss": 1.293,
+      "step": 58495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6339495913469822e-05,
+      "loss": 1.2165,
+      "step": 58500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6338911471046376e-05,
+      "loss": 1.0832,
+      "step": 58505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6338326992424333e-05,
+      "loss": 0.9745,
+      "step": 58510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6337742477607035e-05,
+      "loss": 1.0824,
+      "step": 58515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.633715792659782e-05,
+      "loss": 1.3968,
+      "step": 58520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6336573339400025e-05,
+      "loss": 1.3301,
+      "step": 58525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6335988716016982e-05,
+      "loss": 1.5145,
+      "step": 58530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6335404056452043e-05,
+      "loss": 1.0291,
+      "step": 58535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6334819360708537e-05,
+      "loss": 1.2175,
+      "step": 58540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6334234628789806e-05,
+      "loss": 1.3146,
+      "step": 58545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6333649860699187e-05,
+      "loss": 1.3231,
+      "step": 58550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.633306505644002e-05,
+      "loss": 1.2279,
+      "step": 58555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6332480216015647e-05,
+      "loss": 1.1514,
+      "step": 58560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6331895339429404e-05,
+      "loss": 1.3433,
+      "step": 58565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6331310426684636e-05,
+      "loss": 0.9471,
+      "step": 58570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.633072547778468e-05,
+      "loss": 1.4535,
+      "step": 58575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6330140492732876e-05,
+      "loss": 1.3591,
+      "step": 58580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6329555471532566e-05,
+      "loss": 1.3029,
+      "step": 58585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6328970414187084e-05,
+      "loss": 1.2977,
+      "step": 58590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6328385320699783e-05,
+      "loss": 1.1981,
+      "step": 58595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6327800191073997e-05,
+      "loss": 1.5609,
+      "step": 58600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6327215025313064e-05,
+      "loss": 1.238,
+      "step": 58605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.632662982342034e-05,
+      "loss": 0.9435,
+      "step": 58610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6326044585399147e-05,
+      "loss": 1.3548,
+      "step": 58615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.632545931125284e-05,
+      "loss": 0.9307,
+      "step": 58620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.632487400098476e-05,
+      "loss": 1.1675,
+      "step": 58625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6324288654598245e-05,
+      "loss": 1.3052,
+      "step": 58630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.632370327209664e-05,
+      "loss": 1.0519,
+      "step": 58635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6323117853483292e-05,
+      "loss": 1.2083,
+      "step": 58640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.632253239876154e-05,
+      "loss": 1.2387,
+      "step": 58645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6321946907934724e-05,
+      "loss": 1.0236,
+      "step": 58650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.632136138100619e-05,
+      "loss": 1.1701,
+      "step": 58655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6320775817979286e-05,
+      "loss": 1.1352,
+      "step": 58660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.632019021885735e-05,
+      "loss": 1.2819,
+      "step": 58665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6319604583643732e-05,
+      "loss": 1.0907,
+      "step": 58670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.631901891234177e-05,
+      "loss": 1.1163,
+      "step": 58675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6318433204954814e-05,
+      "loss": 1.3345,
+      "step": 58680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.63178474614862e-05,
+      "loss": 0.9448,
+      "step": 58685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6317261681939284e-05,
+      "loss": 1.1127,
+      "step": 58690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6316675866317405e-05,
+      "loss": 1.316,
+      "step": 58695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.631609001462391e-05,
+      "loss": 1.1824,
+      "step": 58700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.631550412686214e-05,
+      "loss": 1.1084,
+      "step": 58705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.631491820303545e-05,
+      "loss": 1.2647,
+      "step": 58710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.631433224314718e-05,
+      "loss": 1.2602,
+      "step": 58715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6313746247200676e-05,
+      "loss": 1.0896,
+      "step": 58720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6313160215199283e-05,
+      "loss": 1.2069,
+      "step": 58725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.631257414714635e-05,
+      "loss": 1.37,
+      "step": 58730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6311988043045225e-05,
+      "loss": 1.216,
+      "step": 58735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.631140190289925e-05,
+      "loss": 1.192,
+      "step": 58740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.631081572671178e-05,
+      "loss": 1.0214,
+      "step": 58745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6310229514486153e-05,
+      "loss": 1.0597,
+      "step": 58750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6309643266225726e-05,
+      "loss": 1.1045,
+      "step": 58755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.630905698193384e-05,
+      "loss": 1.2397,
+      "step": 58760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6308470661613846e-05,
+      "loss": 1.1603,
+      "step": 58765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.630788430526909e-05,
+      "loss": 1.2644,
+      "step": 58770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6307297912902926e-05,
+      "loss": 1.0077,
+      "step": 58775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6306711484518697e-05,
+      "loss": 1.3352,
+      "step": 58780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.630612502011975e-05,
+      "loss": 1.4239,
+      "step": 58785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.630553851970944e-05,
+      "loss": 1.067,
+      "step": 58790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6304951983291112e-05,
+      "loss": 1.1087,
+      "step": 58795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.630436541086812e-05,
+      "loss": 1.2307,
+      "step": 58800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.630377880244381e-05,
+      "loss": 1.4245,
+      "step": 58805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.630319215802153e-05,
+      "loss": 0.9238,
+      "step": 58810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6302605477604636e-05,
+      "loss": 1.2541,
+      "step": 58815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.630201876119647e-05,
+      "loss": 1.0958,
+      "step": 58820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.630143200880039e-05,
+      "loss": 1.5921,
+      "step": 58825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6300845220419746e-05,
+      "loss": 1.5099,
+      "step": 58830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6300258396057884e-05,
+      "loss": 1.1411,
+      "step": 58835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.629967153571816e-05,
+      "loss": 1.0558,
+      "step": 58840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6299084639403922e-05,
+      "loss": 1.0314,
+      "step": 58845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6298497707118522e-05,
+      "loss": 1.2075,
+      "step": 58850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6297910738865315e-05,
+      "loss": 1.2433,
+      "step": 58855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.629732373464765e-05,
+      "loss": 1.4429,
+      "step": 58860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6296736694468877e-05,
+      "loss": 1.2465,
+      "step": 58865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.629614961833235e-05,
+      "loss": 0.9393,
+      "step": 58870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6295562506241428e-05,
+      "loss": 0.9781,
+      "step": 58875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6294975358199453e-05,
+      "loss": 1.4493,
+      "step": 58880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6294388174209783e-05,
+      "loss": 1.2484,
+      "step": 58885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6293800954275775e-05,
+      "loss": 1.3511,
+      "step": 58890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6293213698400773e-05,
+      "loss": 1.1094,
+      "step": 58895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.629262640658814e-05,
+      "loss": 1.2217,
+      "step": 58900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6292039078841223e-05,
+      "loss": 1.498,
+      "step": 58905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.629145171516338e-05,
+      "loss": 1.5639,
+      "step": 58910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6290864315557966e-05,
+      "loss": 1.2194,
+      "step": 58915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.629027688002833e-05,
+      "loss": 1.2098,
+      "step": 58920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.628968940857783e-05,
+      "loss": 0.8391,
+      "step": 58925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6289101901209824e-05,
+      "loss": 1.1089,
+      "step": 58930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.628851435792766e-05,
+      "loss": 1.0561,
+      "step": 58935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6287926778734697e-05,
+      "loss": 1.1586,
+      "step": 58940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.628733916363429e-05,
+      "loss": 1.0284,
+      "step": 58945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6286751512629795e-05,
+      "loss": 1.347,
+      "step": 58950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6286163825724572e-05,
+      "loss": 1.1695,
+      "step": 58955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6285576102921967e-05,
+      "loss": 1.4247,
+      "step": 58960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.628498834422534e-05,
+      "loss": 1.2453,
+      "step": 58965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6284400549638056e-05,
+      "loss": 0.8213,
+      "step": 58970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.628381271916346e-05,
+      "loss": 1.5581,
+      "step": 58975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6283224852804917e-05,
+      "loss": 1.464,
+      "step": 58980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.628263695056578e-05,
+      "loss": 1.2289,
+      "step": 58985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.62820490124494e-05,
+      "loss": 1.3626,
+      "step": 58990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6281461038459148e-05,
+      "loss": 1.1651,
+      "step": 58995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6280873028598375e-05,
+      "loss": 1.4691,
+      "step": 59000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.628028498287044e-05,
+      "loss": 1.1323,
+      "step": 59005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6279696901278694e-05,
+      "loss": 0.9843,
+      "step": 59010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.62791087838265e-05,
+      "loss": 1.2274,
+      "step": 59015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6278520630517226e-05,
+      "loss": 1.0784,
+      "step": 59020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6277932441354218e-05,
+      "loss": 1.1381,
+      "step": 59025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.627734421634084e-05,
+      "loss": 1.0407,
+      "step": 59030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6276755955480452e-05,
+      "loss": 1.3977,
+      "step": 59035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6276167658776407e-05,
+      "loss": 1.1287,
+      "step": 59040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6275579326232073e-05,
+      "loss": 1.1766,
+      "step": 59045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.62749909578508e-05,
+      "loss": 1.4412,
+      "step": 59050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.627440255363596e-05,
+      "loss": 1.4625,
+      "step": 59055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6273814113590908e-05,
+      "loss": 1.0619,
+      "step": 59060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6273225637718997e-05,
+      "loss": 1.1508,
+      "step": 59065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6272637126023597e-05,
+      "loss": 1.3185,
+      "step": 59070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6272048578508064e-05,
+      "loss": 1.2005,
+      "step": 59075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.627145999517576e-05,
+      "loss": 1.2803,
+      "step": 59080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6270871376030046e-05,
+      "loss": 1.7362,
+      "step": 59085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6270282721074286e-05,
+      "loss": 1.3062,
+      "step": 59090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6269694030311834e-05,
+      "loss": 2.188,
+      "step": 59095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6269105303746062e-05,
+      "loss": 1.3225,
+      "step": 59100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6268516541380325e-05,
+      "loss": 1.6942,
+      "step": 59105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6267927743217988e-05,
+      "loss": 1.4795,
+      "step": 59110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.626733890926241e-05,
+      "loss": 1.1683,
+      "step": 59115
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.626675003951696e-05,
+      "loss": 1.3546,
+      "step": 59120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6266161133984993e-05,
+      "loss": 1.2585,
+      "step": 59125
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6265572192669877e-05,
+      "loss": 1.1774,
+      "step": 59130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6264983215574975e-05,
+      "loss": 1.6682,
+      "step": 59135
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6264394202703648e-05,
+      "loss": 1.4181,
+      "step": 59140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.626380515405926e-05,
+      "loss": 1.2823,
+      "step": 59145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6263216069645178e-05,
+      "loss": 1.3628,
+      "step": 59150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.626262694946476e-05,
+      "loss": 1.4592,
+      "step": 59155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6262037793521376e-05,
+      "loss": 1.3853,
+      "step": 59160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6261448601818387e-05,
+      "loss": 1.3144,
+      "step": 59165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.626085937435916e-05,
+      "loss": 1.1782,
+      "step": 59170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.626027011114706e-05,
+      "loss": 1.6022,
+      "step": 59175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6259680812185447e-05,
+      "loss": 1.0812,
+      "step": 59180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6259091477477694e-05,
+      "loss": 1.16,
+      "step": 59185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6258502107027155e-05,
+      "loss": 1.399,
+      "step": 59190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.625791270083721e-05,
+      "loss": 1.3013,
+      "step": 59195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6257323258911217e-05,
+      "loss": 1.2645,
+      "step": 59200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6256733781252538e-05,
+      "loss": 1.4019,
+      "step": 59205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.625614426786455e-05,
+      "loss": 1.1239,
+      "step": 59210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.625555471875061e-05,
+      "loss": 1.2672,
+      "step": 59215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6254965133914086e-05,
+      "loss": 1.1508,
+      "step": 59220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6254375513358348e-05,
+      "loss": 1.4,
+      "step": 59225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6253785857086763e-05,
+      "loss": 1.3319,
+      "step": 59230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6253196165102698e-05,
+      "loss": 1.0476,
+      "step": 59235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.625260643740952e-05,
+      "loss": 1.1705,
+      "step": 59240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6252016674010594e-05,
+      "loss": 1.1848,
+      "step": 59245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6251426874909292e-05,
+      "loss": 1.1947,
+      "step": 59250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6250837040108977e-05,
+      "loss": 1.304,
+      "step": 59255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6250247169613022e-05,
+      "loss": 1.4084,
+      "step": 59260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6249657263424797e-05,
+      "loss": 1.2472,
+      "step": 59265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6249067321547666e-05,
+      "loss": 1.1714,
+      "step": 59270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6248477343984995e-05,
+      "loss": 1.2476,
+      "step": 59275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.624788733074016e-05,
+      "loss": 1.1874,
+      "step": 59280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6247297281816533e-05,
+      "loss": 1.2841,
+      "step": 59285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6246707197217474e-05,
+      "loss": 1.5347,
+      "step": 59290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6246117076946357e-05,
+      "loss": 1.3247,
+      "step": 59295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.624552692100655e-05,
+      "loss": 1.4434,
+      "step": 59300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.624493672940143e-05,
+      "loss": 1.3576,
+      "step": 59305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.624434650213436e-05,
+      "loss": 1.7308,
+      "step": 59310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6243756239208713e-05,
+      "loss": 1.1286,
+      "step": 59315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.624316594062786e-05,
+      "loss": 1.1915,
+      "step": 59320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6242575606395174e-05,
+      "loss": 1.3025,
+      "step": 59325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.624198523651402e-05,
+      "loss": 1.6064,
+      "step": 59330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6241394830987775e-05,
+      "loss": 1.321,
+      "step": 59335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6240804389819805e-05,
+      "loss": 1.361,
+      "step": 59340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6240213913013486e-05,
+      "loss": 1.2433,
+      "step": 59345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6239623400572197e-05,
+      "loss": 1.2834,
+      "step": 59350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6239032852499294e-05,
+      "loss": 1.418,
+      "step": 59355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.623844226879816e-05,
+      "loss": 1.0164,
+      "step": 59360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6237851649472167e-05,
+      "loss": 1.1406,
+      "step": 59365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6237260994524682e-05,
+      "loss": 1.4079,
+      "step": 59370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.623667030395909e-05,
+      "loss": 1.1231,
+      "step": 59375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.623607957777875e-05,
+      "loss": 0.9854,
+      "step": 59380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.623548881598704e-05,
+      "loss": 1.1864,
+      "step": 59385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 1.4172,
+      "step": 59390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6234307185583015e-05,
+      "loss": 1.266,
+      "step": 59395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6233716316977443e-05,
+      "loss": 1.0756,
+      "step": 59400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6233125412774e-05,
+      "loss": 1.1916,
+      "step": 59405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6232534472976056e-05,
+      "loss": 1.0872,
+      "step": 59410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.623194349758699e-05,
+      "loss": 1.2317,
+      "step": 59415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6231352486610173e-05,
+      "loss": 1.4578,
+      "step": 59420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6230761440048982e-05,
+      "loss": 1.2052,
+      "step": 59425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6230170357906796e-05,
+      "loss": 0.9588,
+      "step": 59430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.622957924018698e-05,
+      "loss": 1.2557,
+      "step": 59435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6228988086892923e-05,
+      "loss": 1.1655,
+      "step": 59440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.622839689802799e-05,
+      "loss": 0.9987,
+      "step": 59445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.622780567359556e-05,
+      "loss": 1.1818,
+      "step": 59450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.622721441359901e-05,
+      "loss": 1.4616,
+      "step": 59455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6226623118041716e-05,
+      "loss": 1.2529,
+      "step": 59460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6226031786927057e-05,
+      "loss": 1.2724,
+      "step": 59465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6225440420258407e-05,
+      "loss": 1.4162,
+      "step": 59470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6224849018039144e-05,
+      "loss": 1.3728,
+      "step": 59475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6224257580272645e-05,
+      "loss": 1.515,
+      "step": 59480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.622366610696229e-05,
+      "loss": 1.2287,
+      "step": 59485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.622307459811145e-05,
+      "loss": 1.2656,
+      "step": 59490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6222483053723508e-05,
+      "loss": 1.0257,
+      "step": 59495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.622189147380184e-05,
+      "loss": 1.2541,
+      "step": 59500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6221299858349826e-05,
+      "loss": 0.9614,
+      "step": 59505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6220708207370844e-05,
+      "loss": 0.9481,
+      "step": 59510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6220116520868275e-05,
+      "loss": 1.2411,
+      "step": 59515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6219524798845492e-05,
+      "loss": 1.1655,
+      "step": 59520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.621893304130588e-05,
+      "loss": 1.3482,
+      "step": 59525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6218341248252813e-05,
+      "loss": 1.1078,
+      "step": 59530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6217749419689673e-05,
+      "loss": 1.4865,
+      "step": 59535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6217157555619837e-05,
+      "loss": 1.4681,
+      "step": 59540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6216565656046693e-05,
+      "loss": 1.0959,
+      "step": 59545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.621597372097361e-05,
+      "loss": 1.1236,
+      "step": 59550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.621538175040398e-05,
+      "loss": 1.0488,
+      "step": 59555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6214789744341172e-05,
+      "loss": 1.1803,
+      "step": 59560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6214197702788576e-05,
+      "loss": 1.3781,
+      "step": 59565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6213605625749564e-05,
+      "loss": 1.9422,
+      "step": 59570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6213013513227526e-05,
+      "loss": 1.2314,
+      "step": 59575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6212421365225836e-05,
+      "loss": 1.2685,
+      "step": 59580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6211829181747882e-05,
+      "loss": 1.1824,
+      "step": 59585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6211236962797042e-05,
+      "loss": 1.3457,
+      "step": 59590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6210644708376694e-05,
+      "loss": 1.0953,
+      "step": 59595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.621005241849023e-05,
+      "loss": 1.3269,
+      "step": 59600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6209460093141022e-05,
+      "loss": 1.0736,
+      "step": 59605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6208867732332455e-05,
+      "loss": 1.2583,
+      "step": 59610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.620827533606792e-05,
+      "loss": 1.3063,
+      "step": 59615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.620768290435079e-05,
+      "loss": 1.3017,
+      "step": 59620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6207090437184457e-05,
+      "loss": 1.2103,
+      "step": 59625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6206497934572295e-05,
+      "loss": 1.4732,
+      "step": 59630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6205905396517692e-05,
+      "loss": 1.3217,
+      "step": 59635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.620531282302403e-05,
+      "loss": 1.2979,
+      "step": 59640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6204720214094695e-05,
+      "loss": 1.1506,
+      "step": 59645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.620412756973307e-05,
+      "loss": 1.0962,
+      "step": 59650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6203534889942543e-05,
+      "loss": 0.9902,
+      "step": 59655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6202942174726494e-05,
+      "loss": 1.076,
+      "step": 59660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6202349424088305e-05,
+      "loss": 1.11,
+      "step": 59665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6201756638031367e-05,
+      "loss": 1.2424,
+      "step": 59670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6201163816559062e-05,
+      "loss": 1.3112,
+      "step": 59675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.620057095967478e-05,
+      "loss": 1.8043,
+      "step": 59680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.61999780673819e-05,
+      "loss": 1.1514,
+      "step": 59685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.619938513968381e-05,
+      "loss": 1.4101,
+      "step": 59690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.61987921765839e-05,
+      "loss": 1.3127,
+      "step": 59695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6198199178085547e-05,
+      "loss": 1.2132,
+      "step": 59700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.619760614419215e-05,
+      "loss": 1.0455,
+      "step": 59705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.619701307490708e-05,
+      "loss": 0.9389,
+      "step": 59710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.619641997023374e-05,
+      "loss": 1.6152,
+      "step": 59715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6195826830175504e-05,
+      "loss": 1.4831,
+      "step": 59720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.619523365473577e-05,
+      "loss": 1.4322,
+      "step": 59725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6194640443917915e-05,
+      "loss": 1.2486,
+      "step": 59730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.619404719772533e-05,
+      "loss": 1.0008,
+      "step": 59735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6193453916161408e-05,
+      "loss": 1.15,
+      "step": 59740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.619286059922953e-05,
+      "loss": 1.173,
+      "step": 59745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.619226724693309e-05,
+      "loss": 1.0826,
+      "step": 59750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6191673859275473e-05,
+      "loss": 1.2779,
+      "step": 59755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6191080436260068e-05,
+      "loss": 1.2797,
+      "step": 59760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.619048697789026e-05,
+      "loss": 1.1757,
+      "step": 59765
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6189893484169447e-05,
+      "loss": 1.203,
+      "step": 59770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.618929995510101e-05,
+      "loss": 1.2009,
+      "step": 59775
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6188706390688342e-05,
+      "loss": 1.0161,
+      "step": 59780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.618811279093483e-05,
+      "loss": 1.1053,
+      "step": 59785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.618751915584387e-05,
+      "loss": 1.2948,
+      "step": 59790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.618692548541884e-05,
+      "loss": 1.1098,
+      "step": 59795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6186331779663148e-05,
+      "loss": 1.0466,
+      "step": 59800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.618573803858017e-05,
+      "loss": 1.1473,
+      "step": 59805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.61851442621733e-05,
+      "loss": 1.298,
+      "step": 59810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6184550450445932e-05,
+      "loss": 1.4293,
+      "step": 59815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.618395660340145e-05,
+      "loss": 1.1954,
+      "step": 59820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6183362721043252e-05,
+      "loss": 1.3377,
+      "step": 59825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6182768803374727e-05,
+      "loss": 1.4996,
+      "step": 59830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6182174850399267e-05,
+      "loss": 1.3722,
+      "step": 59835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6181580862120264e-05,
+      "loss": 1.1624,
+      "step": 59840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6180986838541105e-05,
+      "loss": 1.4893,
+      "step": 59845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.618039277966519e-05,
+      "loss": 1.1427,
+      "step": 59850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6179798685495912e-05,
+      "loss": 1.4635,
+      "step": 59855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6179204556036655e-05,
+      "loss": 1.4321,
+      "step": 59860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.617861039129082e-05,
+      "loss": 1.0738,
+      "step": 59865
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6178016191261795e-05,
+      "loss": 1.5745,
+      "step": 59870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.617742195595297e-05,
+      "loss": 1.1597,
+      "step": 59875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.617682768536775e-05,
+      "loss": 1.3202,
+      "step": 59880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6176233379509517e-05,
+      "loss": 1.2279,
+      "step": 59885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6175639038381673e-05,
+      "loss": 1.3193,
+      "step": 59890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6175044661987606e-05,
+      "loss": 1.2553,
+      "step": 59895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6174450250330713e-05,
+      "loss": 1.1992,
+      "step": 59900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6173855803414387e-05,
+      "loss": 1.118,
+      "step": 59905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6173261321242026e-05,
+      "loss": 1.2194,
+      "step": 59910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6172666803817024e-05,
+      "loss": 1.1345,
+      "step": 59915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6172072251142772e-05,
+      "loss": 1.2471,
+      "step": 59920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6171477663222667e-05,
+      "loss": 1.3981,
+      "step": 59925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6170883040060106e-05,
+      "loss": 1.0908,
+      "step": 59930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6170288381658484e-05,
+      "loss": 0.9669,
+      "step": 59935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6169693688021197e-05,
+      "loss": 1.3311,
+      "step": 59940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.616909895915164e-05,
+      "loss": 1.2586,
+      "step": 59945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6168504195053213e-05,
+      "loss": 0.8899,
+      "step": 59950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6167909395729307e-05,
+      "loss": 1.1972,
+      "step": 59955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6167314561183317e-05,
+      "loss": 1.1325,
+      "step": 59960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6166719691418647e-05,
+      "loss": 1.1235,
+      "step": 59965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6166124786438694e-05,
+      "loss": 1.3602,
+      "step": 59970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.616552984624685e-05,
+      "loss": 1.1697,
+      "step": 59975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.616493487084651e-05,
+      "loss": 1.1768,
+      "step": 59980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.616433986024108e-05,
+      "loss": 1.2223,
+      "step": 59985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6163744814433955e-05,
+      "loss": 1.2901,
+      "step": 59990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.616314973342853e-05,
+      "loss": 1.275,
+      "step": 59995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6162554617228207e-05,
+      "loss": 1.2857,
+      "step": 60000
+    },
+    {
+      "epoch": 0.29,
+      "eval_loss": 1.2266204357147217,
+      "eval_runtime": 6475.4057,
+      "eval_samples_per_second": 3.569,
+      "eval_steps_per_second": 1.784,
+      "step": 60000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.616195946583638e-05,
+      "loss": 1.5969,
+      "step": 60005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.616136427925645e-05,
+      "loss": 1.1492,
+      "step": 60010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.616076905749182e-05,
+      "loss": 1.575,
+      "step": 60015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6160173800545883e-05,
+      "loss": 1.2402,
+      "step": 60020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.615957850842204e-05,
+      "loss": 1.367,
+      "step": 60025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.615898318112369e-05,
+      "loss": 1.1973,
+      "step": 60030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6158387818654235e-05,
+      "loss": 1.5218,
+      "step": 60035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6157792421017074e-05,
+      "loss": 1.0773,
+      "step": 60040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6157196988215604e-05,
+      "loss": 0.9826,
+      "step": 60045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6156601520253226e-05,
+      "loss": 1.1045,
+      "step": 60050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6156006017133344e-05,
+      "loss": 1.1815,
+      "step": 60055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.615541047885936e-05,
+      "loss": 1.1286,
+      "step": 60060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6154814905434668e-05,
+      "loss": 1.0088,
+      "step": 60065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6154219296862673e-05,
+      "loss": 1.4104,
+      "step": 60070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6153623653146774e-05,
+      "loss": 1.2441,
+      "step": 60075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6153027974290378e-05,
+      "loss": 1.2493,
+      "step": 60080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6152432260296876e-05,
+      "loss": 1.2665,
+      "step": 60085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6151836511169684e-05,
+      "loss": 1.3225,
+      "step": 60090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.615124072691219e-05,
+      "loss": 1.2723,
+      "step": 60095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.615064490752781e-05,
+      "loss": 1.1765,
+      "step": 60100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.615004905301994e-05,
+      "loss": 1.4344,
+      "step": 60105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6149453163391978e-05,
+      "loss": 1.2297,
+      "step": 60110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.614885723864733e-05,
+      "loss": 0.9931,
+      "step": 60115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.61482612787894e-05,
+      "loss": 1.0879,
+      "step": 60120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6147665283821593e-05,
+      "loss": 1.0308,
+      "step": 60125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6147069253747312e-05,
+      "loss": 1.1406,
+      "step": 60130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.614647318856996e-05,
+      "loss": 1.1084,
+      "step": 60135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6145877088292937e-05,
+      "loss": 1.1866,
+      "step": 60140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.614528095291965e-05,
+      "loss": 1.2148,
+      "step": 60145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6144684782453503e-05,
+      "loss": 1.0428,
+      "step": 60150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6144088576897905e-05,
+      "loss": 1.1178,
+      "step": 60155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.614349233625625e-05,
+      "loss": 0.9232,
+      "step": 60160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6142896060531956e-05,
+      "loss": 1.1506,
+      "step": 60165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.614229974972842e-05,
+      "loss": 1.212,
+      "step": 60170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.614170340384905e-05,
+      "loss": 1.5154,
+      "step": 60175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6141107022897248e-05,
+      "loss": 1.202,
+      "step": 60180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.614051060687642e-05,
+      "loss": 1.3392,
+      "step": 60185
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6139914155789975e-05,
+      "loss": 1.2768,
+      "step": 60190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.613931766964132e-05,
+      "loss": 1.2262,
+      "step": 60195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6138721148433856e-05,
+      "loss": 1.406,
+      "step": 60200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6138124592171e-05,
+      "loss": 1.1814,
+      "step": 60205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6137528000856143e-05,
+      "loss": 1.2905,
+      "step": 60210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6136931374492703e-05,
+      "loss": 1.2342,
+      "step": 60215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.613633471308408e-05,
+      "loss": 0.9691,
+      "step": 60220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.613573801663369e-05,
+      "loss": 1.0887,
+      "step": 60225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6135141285144937e-05,
+      "loss": 1.0622,
+      "step": 60230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6134544518621224e-05,
+      "loss": 1.3605,
+      "step": 60235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6133947717065967e-05,
+      "loss": 1.0701,
+      "step": 60240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6133350880482567e-05,
+      "loss": 1.3862,
+      "step": 60245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.613275400887443e-05,
+      "loss": 1.3031,
+      "step": 60250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.613215710224497e-05,
+      "loss": 1.4711,
+      "step": 60255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.61315601605976e-05,
+      "loss": 1.2135,
+      "step": 60260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6130963183935723e-05,
+      "loss": 1.0584,
+      "step": 60265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6130366172262745e-05,
+      "loss": 1.0057,
+      "step": 60270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.612976912558208e-05,
+      "loss": 1.1784,
+      "step": 60275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.612917204389714e-05,
+      "loss": 1.2247,
+      "step": 60280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6128574927211323e-05,
+      "loss": 1.213,
+      "step": 60285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6127977775528055e-05,
+      "loss": 1.0604,
+      "step": 60290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.612738058885073e-05,
+      "loss": 1.3431,
+      "step": 60295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6126783367182775e-05,
+      "loss": 1.2551,
+      "step": 60300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6126186110527588e-05,
+      "loss": 1.3094,
+      "step": 60305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.612558881888858e-05,
+      "loss": 1.2102,
+      "step": 60310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6124991492269173e-05,
+      "loss": 1.6965,
+      "step": 60315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6124394130672763e-05,
+      "loss": 1.378,
+      "step": 60320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6123796734102773e-05,
+      "loss": 1.2308,
+      "step": 60325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6123199302562607e-05,
+      "loss": 0.9779,
+      "step": 60330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.612260183605568e-05,
+      "loss": 1.1647,
+      "step": 60335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6122004334585404e-05,
+      "loss": 1.2597,
+      "step": 60340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6121406798155193e-05,
+      "loss": 1.2011,
+      "step": 60345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6120809226768455e-05,
+      "loss": 1.6055,
+      "step": 60350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.61202116204286e-05,
+      "loss": 1.1374,
+      "step": 60355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6119613979139048e-05,
+      "loss": 1.031,
+      "step": 60360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.611901630290321e-05,
+      "loss": 1.1193,
+      "step": 60365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.61184185917245e-05,
+      "loss": 1.0538,
+      "step": 60370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.611782084560632e-05,
+      "loss": 1.4583,
+      "step": 60375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.61172230645521e-05,
+      "loss": 1.3407,
+      "step": 60380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6116625248565246e-05,
+      "loss": 0.9483,
+      "step": 60385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.611602739764917e-05,
+      "loss": 1.0946,
+      "step": 60390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.611542951180729e-05,
+      "loss": 1.2575,
+      "step": 60395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6114831591043018e-05,
+      "loss": 1.1271,
+      "step": 60400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6114233635359765e-05,
+      "loss": 1.2558,
+      "step": 60405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.611363564476095e-05,
+      "loss": 1.3219,
+      "step": 60410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6113037619249993e-05,
+      "loss": 1.2192,
+      "step": 60415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.61124395588303e-05,
+      "loss": 1.4454,
+      "step": 60420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6111841463505288e-05,
+      "loss": 1.1038,
+      "step": 60425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6111243333278378e-05,
+      "loss": 1.4835,
+      "step": 60430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6110645168152975e-05,
+      "loss": 1.2184,
+      "step": 60435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6110046968132508e-05,
+      "loss": 1.1112,
+      "step": 60440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6109448733220383e-05,
+      "loss": 1.2309,
+      "step": 60445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.610885046342002e-05,
+      "loss": 1.0288,
+      "step": 60450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6108252158734833e-05,
+      "loss": 1.3268,
+      "step": 60455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6107653819168246e-05,
+      "loss": 1.0695,
+      "step": 60460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.610705544472367e-05,
+      "loss": 1.281,
+      "step": 60465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.610645703540452e-05,
+      "loss": 1.195,
+      "step": 60470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6105858591214218e-05,
+      "loss": 1.205,
+      "step": 60475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6105260112156175e-05,
+      "loss": 1.2415,
+      "step": 60480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6104661598233816e-05,
+      "loss": 1.1299,
+      "step": 60485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.610406304945056e-05,
+      "loss": 1.4211,
+      "step": 60490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6103464465809815e-05,
+      "loss": 1.173,
+      "step": 60495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6102865847315005e-05,
+      "loss": 1.0202,
+      "step": 60500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6102267193969552e-05,
+      "loss": 1.0777,
+      "step": 60505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6101668505776872e-05,
+      "loss": 1.0596,
+      "step": 60510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.610106978274038e-05,
+      "loss": 1.2004,
+      "step": 60515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6100471024863497e-05,
+      "loss": 1.2897,
+      "step": 60520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.609987223214965e-05,
+      "loss": 1.0627,
+      "step": 60525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.609927340460224e-05,
+      "loss": 1.0373,
+      "step": 60530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6098674542224707e-05,
+      "loss": 1.55,
+      "step": 60535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.609807564502046e-05,
+      "loss": 1.157,
+      "step": 60540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6097476712992925e-05,
+      "loss": 1.4248,
+      "step": 60545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6096877746145515e-05,
+      "loss": 1.2128,
+      "step": 60550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6096278744481652e-05,
+      "loss": 1.094,
+      "step": 60555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.609567970800476e-05,
+      "loss": 1.0855,
+      "step": 60560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.609508063671826e-05,
+      "loss": 1.1394,
+      "step": 60565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.609448153062557e-05,
+      "loss": 1.1998,
+      "step": 60570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6093882389730114e-05,
+      "loss": 1.3645,
+      "step": 60575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6093283214035313e-05,
+      "loss": 1.2584,
+      "step": 60580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.609268400354458e-05,
+      "loss": 1.0724,
+      "step": 60585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6092084758261354e-05,
+      "loss": 1.2156,
+      "step": 60590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.609148547818904e-05,
+      "loss": 1.1432,
+      "step": 60595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6090886163331073e-05,
+      "loss": 0.9516,
+      "step": 60600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6090286813690867e-05,
+      "loss": 1.3454,
+      "step": 60605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6089687429271846e-05,
+      "loss": 1.2227,
+      "step": 60610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6089088010077438e-05,
+      "loss": 1.2931,
+      "step": 60615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6088488556111063e-05,
+      "loss": 1.2937,
+      "step": 60620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6087889067376137e-05,
+      "loss": 1.1215,
+      "step": 60625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6087289543876095e-05,
+      "loss": 1.1084,
+      "step": 60630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6086689985614357e-05,
+      "loss": 1.3695,
+      "step": 60635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6086090392594344e-05,
+      "loss": 1.074,
+      "step": 60640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6085490764819482e-05,
+      "loss": 1.3278,
+      "step": 60645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6084891102293192e-05,
+      "loss": 1.15,
+      "step": 60650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6084291405018904e-05,
+      "loss": 1.151,
+      "step": 60655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.608369167300004e-05,
+      "loss": 1.2459,
+      "step": 60660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.608309190624002e-05,
+      "loss": 1.4039,
+      "step": 60665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6082492104742277e-05,
+      "loss": 1.2808,
+      "step": 60670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6081892268510234e-05,
+      "loss": 1.4322,
+      "step": 60675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6081292397547314e-05,
+      "loss": 1.2704,
+      "step": 60680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.608069249185694e-05,
+      "loss": 1.2762,
+      "step": 60685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6080092551442542e-05,
+      "loss": 1.1363,
+      "step": 60690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.607949257630755e-05,
+      "loss": 1.1615,
+      "step": 60695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6078892566455378e-05,
+      "loss": 1.0792,
+      "step": 60700
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.607829252188947e-05,
+      "loss": 1.1027,
+      "step": 60705
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6077692442613236e-05,
+      "loss": 1.1784,
+      "step": 60710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6077092328630112e-05,
+      "loss": 1.3208,
+      "step": 60715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.607649217994352e-05,
+      "loss": 1.2734,
+      "step": 60720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.607589199655689e-05,
+      "loss": 1.0556,
+      "step": 60725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6075291778473644e-05,
+      "loss": 1.5492,
+      "step": 60730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6074691525697225e-05,
+      "loss": 1.3789,
+      "step": 60735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.607409123823104e-05,
+      "loss": 1.1168,
+      "step": 60740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.607349091607853e-05,
+      "loss": 1.1598,
+      "step": 60745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6072890559243125e-05,
+      "loss": 1.4195,
+      "step": 60750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6072290167728242e-05,
+      "loss": 1.2242,
+      "step": 60755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6071689741537322e-05,
+      "loss": 1.2854,
+      "step": 60760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6071089280673785e-05,
+      "loss": 1.0104,
+      "step": 60765
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.607048878514106e-05,
+      "loss": 1.489,
+      "step": 60770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6069888254942583e-05,
+      "loss": 1.0792,
+      "step": 60775
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.606928769008178e-05,
+      "loss": 1.1311,
+      "step": 60780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6068687090562078e-05,
+      "loss": 1.2852,
+      "step": 60785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.606808645638691e-05,
+      "loss": 1.0933,
+      "step": 60790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6067485787559702e-05,
+      "loss": 1.1473,
+      "step": 60795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.606688508408389e-05,
+      "loss": 1.251,
+      "step": 60800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6066284345962898e-05,
+      "loss": 1.3309,
+      "step": 60805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6065683573200163e-05,
+      "loss": 1.2932,
+      "step": 60810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.606508276579911e-05,
+      "loss": 1.2012,
+      "step": 60815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.606448192376317e-05,
+      "loss": 1.4023,
+      "step": 60820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.606388104709578e-05,
+      "loss": 1.0344,
+      "step": 60825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.606328013580037e-05,
+      "loss": 1.2081,
+      "step": 60830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6062679189880362e-05,
+      "loss": 1.2676,
+      "step": 60835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6062078209339202e-05,
+      "loss": 1.0451,
+      "step": 60840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.606147719418031e-05,
+      "loss": 1.3191,
+      "step": 60845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6060876144407122e-05,
+      "loss": 1.7273,
+      "step": 60850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.606027506002308e-05,
+      "loss": 1.0394,
+      "step": 60855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6059673941031596e-05,
+      "loss": 1.2617,
+      "step": 60860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6059072787436126e-05,
+      "loss": 1.219,
+      "step": 60865
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6058471599240085e-05,
+      "loss": 1.3345,
+      "step": 60870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6057870376446913e-05,
+      "loss": 1.2438,
+      "step": 60875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6057269119060043e-05,
+      "loss": 1.3039,
+      "step": 60880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.605666782708291e-05,
+      "loss": 1.1125,
+      "step": 60885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6056066500518945e-05,
+      "loss": 1.1088,
+      "step": 60890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6055465139371584e-05,
+      "loss": 1.2362,
+      "step": 60895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6054863743644258e-05,
+      "loss": 1.0228,
+      "step": 60900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6054262313340404e-05,
+      "loss": 1.26,
+      "step": 60905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6053660848463456e-05,
+      "loss": 0.9255,
+      "step": 60910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.605305934901685e-05,
+      "loss": 1.0316,
+      "step": 60915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.605245781500402e-05,
+      "loss": 1.2895,
+      "step": 60920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6051856246428397e-05,
+      "loss": 1.159,
+      "step": 60925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6051254643293423e-05,
+      "loss": 1.3632,
+      "step": 60930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6050653005602534e-05,
+      "loss": 1.2326,
+      "step": 60935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6050051333359154e-05,
+      "loss": 1.2334,
+      "step": 60940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6049449626566734e-05,
+      "loss": 1.063,
+      "step": 60945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.60488478852287e-05,
+      "loss": 1.4593,
+      "step": 60950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6048246109348487e-05,
+      "loss": 1.402,
+      "step": 60955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.604764429892954e-05,
+      "loss": 1.7788,
+      "step": 60960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.604704245397529e-05,
+      "loss": 1.2333,
+      "step": 60965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.604644057448918e-05,
+      "loss": 1.3801,
+      "step": 60970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6045838660474634e-05,
+      "loss": 1.1549,
+      "step": 60975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.60452367119351e-05,
+      "loss": 1.2323,
+      "step": 60980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6044634728874015e-05,
+      "loss": 1.2379,
+      "step": 60985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.604403271129481e-05,
+      "loss": 1.0558,
+      "step": 60990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6043430659200936e-05,
+      "loss": 1.5031,
+      "step": 60995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6042828572595816e-05,
+      "loss": 1.1873,
+      "step": 61000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6042226451482898e-05,
+      "loss": 1.1547,
+      "step": 61005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6041624295865618e-05,
+      "loss": 1.147,
+      "step": 61010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.604102210574741e-05,
+      "loss": 1.0162,
+      "step": 61015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6040419881131714e-05,
+      "loss": 1.1012,
+      "step": 61020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6039817622021977e-05,
+      "loss": 1.1222,
+      "step": 61025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.603921532842163e-05,
+      "loss": 1.3887,
+      "step": 61030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.603861300033412e-05,
+      "loss": 1.0546,
+      "step": 61035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.603801063776288e-05,
+      "loss": 1.2213,
+      "step": 61040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6037408240711346e-05,
+      "loss": 1.4335,
+      "step": 61045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.603680580918297e-05,
+      "loss": 1.0699,
+      "step": 61050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6036203343181184e-05,
+      "loss": 1.4573,
+      "step": 61055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.603560084270943e-05,
+      "loss": 1.1668,
+      "step": 61060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.603499830777115e-05,
+      "loss": 0.8665,
+      "step": 61065
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.603439573836978e-05,
+      "loss": 0.8536,
+      "step": 61070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6033793134508767e-05,
+      "loss": 1.4648,
+      "step": 61075
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6033190496191552e-05,
+      "loss": 1.3669,
+      "step": 61080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6032587823421573e-05,
+      "loss": 1.097,
+      "step": 61085
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6031985116202272e-05,
+      "loss": 1.2768,
+      "step": 61090
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6031382374537093e-05,
+      "loss": 1.0801,
+      "step": 61095
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6030779598429473e-05,
+      "loss": 1.4416,
+      "step": 61100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6030176787882867e-05,
+      "loss": 1.2663,
+      "step": 61105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6029573942900698e-05,
+      "loss": 1.3271,
+      "step": 61110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6028971063486426e-05,
+      "loss": 1.0743,
+      "step": 61115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.602836814964348e-05,
+      "loss": 1.1269,
+      "step": 61120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6027765201375314e-05,
+      "loss": 1.0504,
+      "step": 61125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6027162218685364e-05,
+      "loss": 1.4912,
+      "step": 61130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6026559201577078e-05,
+      "loss": 1.0296,
+      "step": 61135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6025956150053897e-05,
+      "loss": 1.1044,
+      "step": 61140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6025353064119268e-05,
+      "loss": 1.6089,
+      "step": 61145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.602474994377663e-05,
+      "loss": 1.0532,
+      "step": 61150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6024146789029426e-05,
+      "loss": 1.1332,
+      "step": 61155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6023543599881108e-05,
+      "loss": 1.3221,
+      "step": 61160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6022940376335114e-05,
+      "loss": 1.4833,
+      "step": 61165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6022337118394893e-05,
+      "loss": 1.2321,
+      "step": 61170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6021733826063884e-05,
+      "loss": 1.3295,
+      "step": 61175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.602113049934554e-05,
+      "loss": 1.1694,
+      "step": 61180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6020527138243303e-05,
+      "loss": 1.295,
+      "step": 61185
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6019923742760615e-05,
+      "loss": 1.1078,
+      "step": 61190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6019320312900925e-05,
+      "loss": 1.1569,
+      "step": 61195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.601871684866768e-05,
+      "loss": 1.215,
+      "step": 61200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6018113350064325e-05,
+      "loss": 1.2834,
+      "step": 61205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6017509817094302e-05,
+      "loss": 1.2621,
+      "step": 61210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6016906249761067e-05,
+      "loss": 1.386,
+      "step": 61215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6016302648068055e-05,
+      "loss": 1.4434,
+      "step": 61220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6015699012018724e-05,
+      "loss": 1.1963,
+      "step": 61225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6015095341616514e-05,
+      "loss": 1.148,
+      "step": 61230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.601449163686488e-05,
+      "loss": 1.4851,
+      "step": 61235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6013887897767257e-05,
+      "loss": 1.1725,
+      "step": 61240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.60132841243271e-05,
+      "loss": 0.8966,
+      "step": 61245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6012680316547856e-05,
+      "loss": 1.1061,
+      "step": 61250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6012076474432976e-05,
+      "loss": 1.1971,
+      "step": 61255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6011472597985905e-05,
+      "loss": 1.1115,
+      "step": 61260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.601086868721009e-05,
+      "loss": 1.2692,
+      "step": 61265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6010264742108984e-05,
+      "loss": 1.2811,
+      "step": 61270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6009660762686034e-05,
+      "loss": 1.1933,
+      "step": 61275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.600905674894469e-05,
+      "loss": 1.1737,
+      "step": 61280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6008452700888396e-05,
+      "loss": 1.379,
+      "step": 61285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.600784861852061e-05,
+      "loss": 1.3385,
+      "step": 61290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6007244501844773e-05,
+      "loss": 1.4418,
+      "step": 61295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.600664035086434e-05,
+      "loss": 1.3948,
+      "step": 61300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.600603616558276e-05,
+      "loss": 1.0926,
+      "step": 61305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6005431946003484e-05,
+      "loss": 1.0943,
+      "step": 61310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.600482769212996e-05,
+      "loss": 1.3356,
+      "step": 61315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6004223403965643e-05,
+      "loss": 1.4101,
+      "step": 61320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6003619081513978e-05,
+      "loss": 1.0097,
+      "step": 61325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6003014724778423e-05,
+      "loss": 1.3003,
+      "step": 61330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6002410333762418e-05,
+      "loss": 1.3489,
+      "step": 61335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.600180590846943e-05,
+      "loss": 1.1991,
+      "step": 61340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6001201448902896e-05,
+      "loss": 1.2559,
+      "step": 61345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6000596955066277e-05,
+      "loss": 1.0858,
+      "step": 61350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.599999242696302e-05,
+      "loss": 1.3959,
+      "step": 61355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5999387864596584e-05,
+      "loss": 1.3185,
+      "step": 61360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.599878326797041e-05,
+      "loss": 1.1438,
+      "step": 61365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.599817863708796e-05,
+      "loss": 1.3318,
+      "step": 61370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5997573971952682e-05,
+      "loss": 1.2275,
+      "step": 61375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5996969272568035e-05,
+      "loss": 1.1018,
+      "step": 61380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5996364538937465e-05,
+      "loss": 1.1101,
+      "step": 61385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5995759771064427e-05,
+      "loss": 1.2143,
+      "step": 61390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.599515496895238e-05,
+      "loss": 1.0954,
+      "step": 61395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.599455013260477e-05,
+      "loss": 1.3773,
+      "step": 61400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5993945262025057e-05,
+      "loss": 1.3724,
+      "step": 61405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5993340357216692e-05,
+      "loss": 1.1151,
+      "step": 61410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5992735418183127e-05,
+      "loss": 1.8078,
+      "step": 61415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.599213044492782e-05,
+      "loss": 1.1906,
+      "step": 61420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.599152543745423e-05,
+      "loss": 1.1174,
+      "step": 61425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5990920395765805e-05,
+      "loss": 1.3618,
+      "step": 61430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5990315319866003e-05,
+      "loss": 1.2017,
+      "step": 61435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5989710209758275e-05,
+      "loss": 0.9643,
+      "step": 61440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5989105065446088e-05,
+      "loss": 1.2468,
+      "step": 61445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5988499886932883e-05,
+      "loss": 1.5291,
+      "step": 61450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5987894674222122e-05,
+      "loss": 1.0162,
+      "step": 61455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5987289427317265e-05,
+      "loss": 1.0565,
+      "step": 61460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5986684146221763e-05,
+      "loss": 1.4356,
+      "step": 61465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5986078830939076e-05,
+      "loss": 1.2369,
+      "step": 61470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5985473481472657e-05,
+      "loss": 1.4729,
+      "step": 61475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.598486809782597e-05,
+      "loss": 1.061,
+      "step": 61480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5984262680002466e-05,
+      "loss": 1.0972,
+      "step": 61485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.59836572280056e-05,
+      "loss": 1.1336,
+      "step": 61490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5983051741838833e-05,
+      "loss": 1.222,
+      "step": 61495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5982446221505625e-05,
+      "loss": 1.4171,
+      "step": 61500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.598184066700943e-05,
+      "loss": 1.5155,
+      "step": 61505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5981235078353707e-05,
+      "loss": 0.9672,
+      "step": 61510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5980629455541915e-05,
+      "loss": 1.1132,
+      "step": 61515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5980023798577512e-05,
+      "loss": 1.4579,
+      "step": 61520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.597941810746396e-05,
+      "loss": 1.3117,
+      "step": 61525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.597881238220471e-05,
+      "loss": 1.3654,
+      "step": 61530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.597820662280323e-05,
+      "loss": 1.0957,
+      "step": 61535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5977600829262973e-05,
+      "loss": 1.3051,
+      "step": 61540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.59769950015874e-05,
+      "loss": 1.3481,
+      "step": 61545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5976389139779967e-05,
+      "loss": 1.0884,
+      "step": 61550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5975783243844144e-05,
+      "loss": 1.4536,
+      "step": 61555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5975177313783382e-05,
+      "loss": 1.0683,
+      "step": 61560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.597457134960114e-05,
+      "loss": 1.1623,
+      "step": 61565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5973965351300885e-05,
+      "loss": 1.3211,
+      "step": 61570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5973359318886078e-05,
+      "loss": 0.8446,
+      "step": 61575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5972753252360177e-05,
+      "loss": 1.0385,
+      "step": 61580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5972147151726634e-05,
+      "loss": 1.5547,
+      "step": 61585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5971541016988926e-05,
+      "loss": 1.1725,
+      "step": 61590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5970934848150505e-05,
+      "loss": 1.2138,
+      "step": 61595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5970328645214834e-05,
+      "loss": 0.9454,
+      "step": 61600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5969722408185378e-05,
+      "loss": 1.1137,
+      "step": 61605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5969116137065594e-05,
+      "loss": 1.9135,
+      "step": 61610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5968509831858944e-05,
+      "loss": 1.2055,
+      "step": 61615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5967903492568898e-05,
+      "loss": 1.0566,
+      "step": 61620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.596729711919891e-05,
+      "loss": 1.0555,
+      "step": 61625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5966690711752445e-05,
+      "loss": 1.1217,
+      "step": 61630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5966084270232966e-05,
+      "loss": 1.0576,
+      "step": 61635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5965477794643938e-05,
+      "loss": 1.2989,
+      "step": 61640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5964871284988827e-05,
+      "loss": 1.0847,
+      "step": 61645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.596426474127109e-05,
+      "loss": 0.9864,
+      "step": 61650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.596365816349419e-05,
+      "loss": 1.1521,
+      "step": 61655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.59630515516616e-05,
+      "loss": 1.1205,
+      "step": 61660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5962444905776775e-05,
+      "loss": 1.1809,
+      "step": 61665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5961838225843186e-05,
+      "loss": 1.2567,
+      "step": 61670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.596123151186429e-05,
+      "loss": 1.4497,
+      "step": 61675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5960624763843554e-05,
+      "loss": 0.9642,
+      "step": 61680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5960017981784454e-05,
+      "loss": 1.3675,
+      "step": 61685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5959411165690438e-05,
+      "loss": 1.1959,
+      "step": 61690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.595880431556498e-05,
+      "loss": 1.2271,
+      "step": 61695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5958197431411547e-05,
+      "loss": 1.0954,
+      "step": 61700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.59575905132336e-05,
+      "loss": 1.3998,
+      "step": 61705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5956983561034607e-05,
+      "loss": 1.3659,
+      "step": 61710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5956376574818035e-05,
+      "loss": 1.162,
+      "step": 61715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5955769554587347e-05,
+      "loss": 1.3429,
+      "step": 61720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5955162500346013e-05,
+      "loss": 1.3273,
+      "step": 61725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.59545554120975e-05,
+      "loss": 1.1147,
+      "step": 61730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.595394828984527e-05,
+      "loss": 1.0234,
+      "step": 61735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5953341133592794e-05,
+      "loss": 1.2344,
+      "step": 61740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5952733943343535e-05,
+      "loss": 1.1904,
+      "step": 61745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5952126719100967e-05,
+      "loss": 1.0714,
+      "step": 61750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5951519460868553e-05,
+      "loss": 1.335,
+      "step": 61755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5950912168649765e-05,
+      "loss": 1.3428,
+      "step": 61760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5950304842448066e-05,
+      "loss": 1.2053,
+      "step": 61765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5949697482266922e-05,
+      "loss": 1.3423,
+      "step": 61770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.594909008810981e-05,
+      "loss": 1.3012,
+      "step": 61775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.594848265998019e-05,
+      "loss": 1.0973,
+      "step": 61780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.594787519788154e-05,
+      "loss": 0.991,
+      "step": 61785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.594726770181732e-05,
+      "loss": 1.3236,
+      "step": 61790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5946660171791002e-05,
+      "loss": 0.925,
+      "step": 61795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5946052607806057e-05,
+      "loss": 1.0825,
+      "step": 61800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5945445009865953e-05,
+      "loss": 1.1332,
+      "step": 61805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.594483737797416e-05,
+      "loss": 1.169,
+      "step": 61810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.594422971213415e-05,
+      "loss": 1.4823,
+      "step": 61815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.594362201234939e-05,
+      "loss": 1.2085,
+      "step": 61820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.594301427862335e-05,
+      "loss": 1.42,
+      "step": 61825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5942406510959508e-05,
+      "loss": 1.2041,
+      "step": 61830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5941798709361323e-05,
+      "loss": 1.2442,
+      "step": 61835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5941190873832272e-05,
+      "loss": 1.4296,
+      "step": 61840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5940583004375824e-05,
+      "loss": 1.5392,
+      "step": 61845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5939975100995457e-05,
+      "loss": 1.228,
+      "step": 61850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5939367163694635e-05,
+      "loss": 1.2507,
+      "step": 61855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.593875919247683e-05,
+      "loss": 1.6487,
+      "step": 61860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.593815118734552e-05,
+      "loss": 1.3286,
+      "step": 61865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5937543148304172e-05,
+      "loss": 0.9929,
+      "step": 61870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5936935075356255e-05,
+      "loss": 1.1443,
+      "step": 61875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.593632696850525e-05,
+      "loss": 1.1681,
+      "step": 61880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.593571882775462e-05,
+      "loss": 1.2389,
+      "step": 61885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5935110653107847e-05,
+      "loss": 0.9846,
+      "step": 61890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.59345024445684e-05,
+      "loss": 1.2011,
+      "step": 61895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.593389420213975e-05,
+      "loss": 1.2114,
+      "step": 61900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5933285925825376e-05,
+      "loss": 1.4374,
+      "step": 61905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5932677615628745e-05,
+      "loss": 1.2183,
+      "step": 61910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5932069271553333e-05,
+      "loss": 1.1721,
+      "step": 61915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.593146089360262e-05,
+      "loss": 1.1611,
+      "step": 61920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.593085248178007e-05,
+      "loss": 1.21,
+      "step": 61925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5930244036089163e-05,
+      "loss": 1.3832,
+      "step": 61930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5929635556533376e-05,
+      "loss": 1.032,
+      "step": 61935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.592902704311618e-05,
+      "loss": 0.9935,
+      "step": 61940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5928418495841047e-05,
+      "loss": 1.1876,
+      "step": 61945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.592780991471146e-05,
+      "loss": 1.5008,
+      "step": 61950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5927201299730887e-05,
+      "loss": 1.2161,
+      "step": 61955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.592659265090281e-05,
+      "loss": 1.2254,
+      "step": 61960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5925983968230697e-05,
+      "loss": 1.4665,
+      "step": 61965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5925375251718033e-05,
+      "loss": 0.9761,
+      "step": 61970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5924766501368286e-05,
+      "loss": 0.9838,
+      "step": 61975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5924157717184936e-05,
+      "loss": 1.26,
+      "step": 61980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5923548899171463e-05,
+      "loss": 1.1752,
+      "step": 61985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5922940047331336e-05,
+      "loss": 1.209,
+      "step": 61990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5922331161668032e-05,
+      "loss": 1.1109,
+      "step": 61995
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.592172224218504e-05,
+      "loss": 1.3215,
+      "step": 62000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5921113288885823e-05,
+      "loss": 1.0682,
+      "step": 62005
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5920504301773863e-05,
+      "loss": 1.3363,
+      "step": 62010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5919895280852645e-05,
+      "loss": 1.1824,
+      "step": 62015
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5919286226125637e-05,
+      "loss": 0.956,
+      "step": 62020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5918677137596322e-05,
+      "loss": 1.3586,
+      "step": 62025
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.591806801526818e-05,
+      "loss": 1.0979,
+      "step": 62030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.591745885914468e-05,
+      "loss": 1.0276,
+      "step": 62035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5916849669229314e-05,
+      "loss": 1.5603,
+      "step": 62040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5916240445525552e-05,
+      "loss": 1.1134,
+      "step": 62045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5915631188036874e-05,
+      "loss": 1.2738,
+      "step": 62050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.591502189676676e-05,
+      "loss": 1.0845,
+      "step": 62055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.591441257171869e-05,
+      "loss": 1.2726,
+      "step": 62060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5913803212896142e-05,
+      "loss": 1.3491,
+      "step": 62065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.59131938203026e-05,
+      "loss": 1.0048,
+      "step": 62070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5912584393941536e-05,
+      "loss": 1.3387,
+      "step": 62075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5911974933816438e-05,
+      "loss": 1.1472,
+      "step": 62080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5911365439930785e-05,
+      "loss": 0.9806,
+      "step": 62085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.591075591228805e-05,
+      "loss": 1.1565,
+      "step": 62090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5910146350891728e-05,
+      "loss": 1.5442,
+      "step": 62095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.590953675574529e-05,
+      "loss": 1.5718,
+      "step": 62100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5908927126852213e-05,
+      "loss": 1.3392,
+      "step": 62105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5908317464215985e-05,
+      "loss": 1.3173,
+      "step": 62110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.590770776784009e-05,
+      "loss": 0.9635,
+      "step": 62115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5907098037728003e-05,
+      "loss": 1.0942,
+      "step": 62120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5906488273883212e-05,
+      "loss": 1.4638,
+      "step": 62125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5905878476309194e-05,
+      "loss": 1.1168,
+      "step": 62130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5905268645009434e-05,
+      "loss": 2.0022,
+      "step": 62135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5904658779987415e-05,
+      "loss": 1.1753,
+      "step": 62140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5904048881246617e-05,
+      "loss": 0.8797,
+      "step": 62145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5903438948790527e-05,
+      "loss": 1.3434,
+      "step": 62150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.590282898262262e-05,
+      "loss": 1.1134,
+      "step": 62155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.590221898274639e-05,
+      "loss": 1.1141,
+      "step": 62160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5901608949165313e-05,
+      "loss": 1.0328,
+      "step": 62165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5900998881882875e-05,
+      "loss": 1.2489,
+      "step": 62170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.590038878090256e-05,
+      "loss": 1.0184,
+      "step": 62175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.589977864622785e-05,
+      "loss": 1.1602,
+      "step": 62180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5899168477862234e-05,
+      "loss": 1.1839,
+      "step": 62185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5898558275809187e-05,
+      "loss": 1.3004,
+      "step": 62190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5897948040072204e-05,
+      "loss": 1.1154,
+      "step": 62195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5897337770654766e-05,
+      "loss": 1.5572,
+      "step": 62200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5896727467560354e-05,
+      "loss": 1.3791,
+      "step": 62205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.589611713079246e-05,
+      "loss": 1.0697,
+      "step": 62210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5895506760354565e-05,
+      "loss": 1.1259,
+      "step": 62215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5894896356250153e-05,
+      "loss": 1.0961,
+      "step": 62220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5894285918482718e-05,
+      "loss": 1.197,
+      "step": 62225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5893675447055738e-05,
+      "loss": 1.285,
+      "step": 62230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5893064941972696e-05,
+      "loss": 1.0365,
+      "step": 62235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.589245440323709e-05,
+      "loss": 1.1024,
+      "step": 62240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5891843830852396e-05,
+      "loss": 1.0812,
+      "step": 62245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5891233224822105e-05,
+      "loss": 1.2192,
+      "step": 62250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5890622585149706e-05,
+      "loss": 1.0664,
+      "step": 62255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5890011911838685e-05,
+      "loss": 1.3704,
+      "step": 62260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5889401204892527e-05,
+      "loss": 1.2645,
+      "step": 62265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5888790464314715e-05,
+      "loss": 1.2232,
+      "step": 62270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5888179690108748e-05,
+      "loss": 1.2326,
+      "step": 62275
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5887568882278106e-05,
+      "loss": 1.1758,
+      "step": 62280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.588695804082628e-05,
+      "loss": 1.1582,
+      "step": 62285
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5886347165756758e-05,
+      "loss": 1.4127,
+      "step": 62290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5885736257073027e-05,
+      "loss": 1.2179,
+      "step": 62295
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5885125314778575e-05,
+      "loss": 1.2036,
+      "step": 62300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5884514338876895e-05,
+      "loss": 1.2196,
+      "step": 62305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.588390332937147e-05,
+      "loss": 1.2496,
+      "step": 62310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5883292286265795e-05,
+      "loss": 1.337,
+      "step": 62315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5882681209563352e-05,
+      "loss": 1.7577,
+      "step": 62320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5882070099267636e-05,
+      "loss": 1.2305,
+      "step": 62325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5881458955382145e-05,
+      "loss": 1.7181,
+      "step": 62330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.588084777791035e-05,
+      "loss": 0.9944,
+      "step": 62335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5880236566855752e-05,
+      "loss": 1.1765,
+      "step": 62340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5879625322221844e-05,
+      "loss": 1.2085,
+      "step": 62345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5879014044012113e-05,
+      "loss": 1.2516,
+      "step": 62350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.587840273223005e-05,
+      "loss": 0.984,
+      "step": 62355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5877791386879145e-05,
+      "loss": 1.3315,
+      "step": 62360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5877180007962885e-05,
+      "loss": 1.0743,
+      "step": 62365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5876568595484774e-05,
+      "loss": 1.5652,
+      "step": 62370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5875957149448292e-05,
+      "loss": 1.2764,
+      "step": 62375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5875345669856935e-05,
+      "loss": 0.9949,
+      "step": 62380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5874734156714192e-05,
+      "loss": 1.088,
+      "step": 62385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.587412261002356e-05,
+      "loss": 0.8992,
+      "step": 62390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5873511029788526e-05,
+      "loss": 1.3097,
+      "step": 62395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5872899416012587e-05,
+      "loss": 1.0926,
+      "step": 62400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5872287768699233e-05,
+      "loss": 1.4681,
+      "step": 62405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5871676087851957e-05,
+      "loss": 1.1367,
+      "step": 62410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.587106437347425e-05,
+      "loss": 0.9068,
+      "step": 62415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5870452625569612e-05,
+      "loss": 1.3929,
+      "step": 62420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.586984084414153e-05,
+      "loss": 1.0096,
+      "step": 62425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5869229029193502e-05,
+      "loss": 1.1447,
+      "step": 62430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5868617180729017e-05,
+      "loss": 1.484,
+      "step": 62435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.586800529875157e-05,
+      "loss": 1.4275,
+      "step": 62440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.586739338326466e-05,
+      "loss": 1.3069,
+      "step": 62445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.586678143427178e-05,
+      "loss": 1.0645,
+      "step": 62450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5866169451776416e-05,
+      "loss": 1.6407,
+      "step": 62455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.586555743578207e-05,
+      "loss": 1.5152,
+      "step": 62460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.586494538629224e-05,
+      "loss": 1.2169,
+      "step": 62465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5864333303310416e-05,
+      "loss": 1.4896,
+      "step": 62470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.58637211868401e-05,
+      "loss": 1.2614,
+      "step": 62475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5863109036884773e-05,
+      "loss": 1.1389,
+      "step": 62480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5862496853447943e-05,
+      "loss": 1.4111,
+      "step": 62485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5861884636533104e-05,
+      "loss": 1.2843,
+      "step": 62490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.586127238614375e-05,
+      "loss": 1.1954,
+      "step": 62495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.586066010228338e-05,
+      "loss": 1.0462,
+      "step": 62500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5860047784955487e-05,
+      "loss": 1.4465,
+      "step": 62505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5859435434163568e-05,
+      "loss": 1.1618,
+      "step": 62510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5858823049911126e-05,
+      "loss": 1.4279,
+      "step": 62515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.585821063220165e-05,
+      "loss": 1.3355,
+      "step": 62520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.585759818103864e-05,
+      "loss": 1.0434,
+      "step": 62525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5856985696425593e-05,
+      "loss": 1.3344,
+      "step": 62530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5856373178366008e-05,
+      "loss": 1.2849,
+      "step": 62535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5855760626863384e-05,
+      "loss": 1.1084,
+      "step": 62540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5855148041921217e-05,
+      "loss": 0.9865,
+      "step": 62545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5854535423543004e-05,
+      "loss": 1.0727,
+      "step": 62550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5853922771732248e-05,
+      "loss": 1.0694,
+      "step": 62555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.585331008649244e-05,
+      "loss": 1.2754,
+      "step": 62560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5852697367827086e-05,
+      "loss": 1.3949,
+      "step": 62565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.585208461573968e-05,
+      "loss": 1.3002,
+      "step": 62570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5851471830233725e-05,
+      "loss": 1.2759,
+      "step": 62575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5850859011312717e-05,
+      "loss": 1.269,
+      "step": 62580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.585024615898016e-05,
+      "loss": 1.1065,
+      "step": 62585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.584963327323955e-05,
+      "loss": 1.4372,
+      "step": 62590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5849020354094387e-05,
+      "loss": 1.0654,
+      "step": 62595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.584840740154817e-05,
+      "loss": 1.0529,
+      "step": 62600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5847794415604404e-05,
+      "loss": 1.0709,
+      "step": 62605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5847181396266588e-05,
+      "loss": 1.0589,
+      "step": 62610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5846568343538218e-05,
+      "loss": 1.1739,
+      "step": 62615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.58459552574228e-05,
+      "loss": 1.3125,
+      "step": 62620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5845342137923832e-05,
+      "loss": 1.121,
+      "step": 62625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5844728985044814e-05,
+      "loss": 1.2004,
+      "step": 62630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5844115798789258e-05,
+      "loss": 1.1897,
+      "step": 62635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.584350257916065e-05,
+      "loss": 1.2806,
+      "step": 62640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.58428893261625e-05,
+      "loss": 1.1385,
+      "step": 62645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5842276039798314e-05,
+      "loss": 1.1464,
+      "step": 62650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5841662720071586e-05,
+      "loss": 1.2575,
+      "step": 62655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5841049366985824e-05,
+      "loss": 1.4001,
+      "step": 62660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5840435980544524e-05,
+      "loss": 1.1237,
+      "step": 62665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5839822560751197e-05,
+      "loss": 1.2464,
+      "step": 62670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5839209107609343e-05,
+      "loss": 0.9139,
+      "step": 62675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.583859562112246e-05,
+      "loss": 1.1293,
+      "step": 62680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5837982101294062e-05,
+      "loss": 1.4132,
+      "step": 62685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5837368548127642e-05,
+      "loss": 1.4297,
+      "step": 62690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5836754961626705e-05,
+      "loss": 1.0295,
+      "step": 62695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5836141341794763e-05,
+      "loss": 1.2746,
+      "step": 62700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5835527688635314e-05,
+      "loss": 1.1531,
+      "step": 62705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5834914002151865e-05,
+      "loss": 1.163,
+      "step": 62710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5834300282347915e-05,
+      "loss": 1.0031,
+      "step": 62715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5833686529226974e-05,
+      "loss": 1.2897,
+      "step": 62720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5833072742792546e-05,
+      "loss": 1.0434,
+      "step": 62725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5832458923048133e-05,
+      "loss": 1.0851,
+      "step": 62730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5831845069997244e-05,
+      "loss": 1.4528,
+      "step": 62735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5831231183643384e-05,
+      "loss": 1.4537,
+      "step": 62740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5830617263990056e-05,
+      "loss": 1.063,
+      "step": 62745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.583000331104077e-05,
+      "loss": 1.018,
+      "step": 62750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5829389324799027e-05,
+      "loss": 1.133,
+      "step": 62755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5828775305268335e-05,
+      "loss": 1.0916,
+      "step": 62760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5828161252452202e-05,
+      "loss": 1.4199,
+      "step": 62765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5827547166354132e-05,
+      "loss": 1.369,
+      "step": 62770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5826933046977636e-05,
+      "loss": 1.259,
+      "step": 62775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.582631889432622e-05,
+      "loss": 1.4432,
+      "step": 62780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5825704708403386e-05,
+      "loss": 1.0737,
+      "step": 62785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5825090489212645e-05,
+      "loss": 1.0685,
+      "step": 62790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.58244762367575e-05,
+      "loss": 1.1633,
+      "step": 62795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5823861951041467e-05,
+      "loss": 1.1712,
+      "step": 62800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5823247632068054e-05,
+      "loss": 1.17,
+      "step": 62805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5822633279840763e-05,
+      "loss": 1.1137,
+      "step": 62810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.58220188943631e-05,
+      "loss": 1.2174,
+      "step": 62815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5821404475638583e-05,
+      "loss": 1.121,
+      "step": 62820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.582079002367071e-05,
+      "loss": 1.3015,
+      "step": 62825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5820175538463e-05,
+      "loss": 1.5696,
+      "step": 62830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5819561020018953e-05,
+      "loss": 1.4741,
+      "step": 62835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5818946468342085e-05,
+      "loss": 1.1086,
+      "step": 62840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5818331883435905e-05,
+      "loss": 1.1448,
+      "step": 62845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.581771726530392e-05,
+      "loss": 1.1506,
+      "step": 62850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5817102613949636e-05,
+      "loss": 1.238,
+      "step": 62855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.581648792937657e-05,
+      "loss": 1.0174,
+      "step": 62860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.581587321158823e-05,
+      "loss": 1.0971,
+      "step": 62865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5815258460588127e-05,
+      "loss": 0.9437,
+      "step": 62870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5814643676379766e-05,
+      "loss": 1.7076,
+      "step": 62875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5814028858966667e-05,
+      "loss": 1.29,
+      "step": 62880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5813414008352333e-05,
+      "loss": 1.1409,
+      "step": 62885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.581279912454028e-05,
+      "loss": 1.5202,
+      "step": 62890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.581218420753402e-05,
+      "loss": 1.5236,
+      "step": 62895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5811569257337054e-05,
+      "loss": 1.016,
+      "step": 62900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5810954273952907e-05,
+      "loss": 1.0001,
+      "step": 62905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5810339257385086e-05,
+      "loss": 1.3085,
+      "step": 62910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5809724207637104e-05,
+      "loss": 1.3632,
+      "step": 62915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5809109124712468e-05,
+      "loss": 1.2811,
+      "step": 62920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5808494008614698e-05,
+      "loss": 1.3918,
+      "step": 62925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.58078788593473e-05,
+      "loss": 1.359,
+      "step": 62930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5807263676913792e-05,
+      "loss": 1.0878,
+      "step": 62935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5806648461317684e-05,
+      "loss": 1.295,
+      "step": 62940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5806033212562492e-05,
+      "loss": 1.6728,
+      "step": 62945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5805417930651727e-05,
+      "loss": 1.661,
+      "step": 62950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5804802615588903e-05,
+      "loss": 1.3768,
+      "step": 62955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5804187267377532e-05,
+      "loss": 1.0515,
+      "step": 62960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5803571886021135e-05,
+      "loss": 1.1475,
+      "step": 62965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5802956471523215e-05,
+      "loss": 1.1252,
+      "step": 62970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5802341023887296e-05,
+      "loss": 1.2566,
+      "step": 62975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.580172554311689e-05,
+      "loss": 1.0221,
+      "step": 62980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5801110029215502e-05,
+      "loss": 1.2665,
+      "step": 62985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5800494482186667e-05,
+      "loss": 1.1691,
+      "step": 62990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.579987890203388e-05,
+      "loss": 1.1715,
+      "step": 62995
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.579926328876067e-05,
+      "loss": 1.1613,
+      "step": 63000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.579864764237055e-05,
+      "loss": 1.1515,
+      "step": 63005
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5798031962867025e-05,
+      "loss": 1.4886,
+      "step": 63010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5797416250253627e-05,
+      "loss": 1.1489,
+      "step": 63015
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5796800504533858e-05,
+      "loss": 1.2254,
+      "step": 63020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5796184725711242e-05,
+      "loss": 1.2167,
+      "step": 63025
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5795568913789295e-05,
+      "loss": 1.2174,
+      "step": 63030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5794953068771532e-05,
+      "loss": 1.2658,
+      "step": 63035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5794337190661472e-05,
+      "loss": 1.138,
+      "step": 63040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.579372127946263e-05,
+      "loss": 1.3536,
+      "step": 63045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5793105335178516e-05,
+      "loss": 1.1104,
+      "step": 63050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.579248935781266e-05,
+      "loss": 1.112,
+      "step": 63055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5791873347368577e-05,
+      "loss": 1.6969,
+      "step": 63060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5791257303849778e-05,
+      "loss": 1.4583,
+      "step": 63065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5790641227259786e-05,
+      "loss": 1.4321,
+      "step": 63070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.579002511760212e-05,
+      "loss": 1.2255,
+      "step": 63075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5789408974880296e-05,
+      "loss": 1.3849,
+      "step": 63080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5788792799097833e-05,
+      "loss": 1.0703,
+      "step": 63085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5788176590258247e-05,
+      "loss": 1.2835,
+      "step": 63090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.578756034836506e-05,
+      "loss": 1.065,
+      "step": 63095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.578694407342179e-05,
+      "loss": 1.2562,
+      "step": 63100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.578632776543196e-05,
+      "loss": 1.1955,
+      "step": 63105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5785711424399082e-05,
+      "loss": 1.2656,
+      "step": 63110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5785095050326685e-05,
+      "loss": 1.1308,
+      "step": 63115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.578447864321828e-05,
+      "loss": 1.1636,
+      "step": 63120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5783862203077395e-05,
+      "loss": 1.3048,
+      "step": 63125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.578324572990754e-05,
+      "loss": 1.2581,
+      "step": 63130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5782629223712243e-05,
+      "loss": 1.2797,
+      "step": 63135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5782012684495024e-05,
+      "loss": 1.2293,
+      "step": 63140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5781396112259404e-05,
+      "loss": 1.0169,
+      "step": 63145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.57807795070089e-05,
+      "loss": 0.9411,
+      "step": 63150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.578016286874704e-05,
+      "loss": 1.0818,
+      "step": 63155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.577954619747734e-05,
+      "loss": 1.1657,
+      "step": 63160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.577892949320332e-05,
+      "loss": 1.1402,
+      "step": 63165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5778312755928503e-05,
+      "loss": 1.3199,
+      "step": 63170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5777695985656417e-05,
+      "loss": 1.319,
+      "step": 63175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5777079182390576e-05,
+      "loss": 0.998,
+      "step": 63180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5776462346134508e-05,
+      "loss": 0.9486,
+      "step": 63185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5775845476891732e-05,
+      "loss": 1.2323,
+      "step": 63190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.577522857466577e-05,
+      "loss": 1.191,
+      "step": 63195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5774611639460147e-05,
+      "loss": 1.0824,
+      "step": 63200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.577399467127839e-05,
+      "loss": 1.263,
+      "step": 63205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.577337767012401e-05,
+      "loss": 1.5436,
+      "step": 63210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5772760636000546e-05,
+      "loss": 1.4685,
+      "step": 63215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.577214356891151e-05,
+      "loss": 1.3188,
+      "step": 63220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5771526468860435e-05,
+      "loss": 1.028,
+      "step": 63225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5770909335850833e-05,
+      "loss": 1.0738,
+      "step": 63230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5770292169886236e-05,
+      "loss": 1.0812,
+      "step": 63235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.576967497097017e-05,
+      "loss": 1.4685,
+      "step": 63240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5769057739106154e-05,
+      "loss": 1.016,
+      "step": 63245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5768440474297717e-05,
+      "loss": 1.2593,
+      "step": 63250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5767823176548383e-05,
+      "loss": 1.296,
+      "step": 63255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5767205845861674e-05,
+      "loss": 1.5276,
+      "step": 63260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.576658848224112e-05,
+      "loss": 1.1883,
+      "step": 63265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5765971085690242e-05,
+      "loss": 1.207,
+      "step": 63270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5765353656212567e-05,
+      "loss": 1.0337,
+      "step": 63275
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5764736193811625e-05,
+      "loss": 1.1564,
+      "step": 63280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5764118698490937e-05,
+      "loss": 1.5202,
+      "step": 63285
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.576350117025403e-05,
+      "loss": 1.3384,
+      "step": 63290
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5762883609104433e-05,
+      "loss": 1.1882,
+      "step": 63295
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.576226601504567e-05,
+      "loss": 1.3638,
+      "step": 63300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5761648388081273e-05,
+      "loss": 1.0506,
+      "step": 63305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.576103072821476e-05,
+      "loss": 0.9903,
+      "step": 63310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5760413035449665e-05,
+      "loss": 1.321,
+      "step": 63315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5759795309789514e-05,
+      "loss": 1.4973,
+      "step": 63320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.575917755123783e-05,
+      "loss": 1.1261,
+      "step": 63325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5758559759798145e-05,
+      "loss": 1.1184,
+      "step": 63330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.575794193547399e-05,
+      "loss": 1.4879,
+      "step": 63335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5757324078268888e-05,
+      "loss": 1.163,
+      "step": 63340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.575670618818637e-05,
+      "loss": 1.2102,
+      "step": 63345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5756088265229963e-05,
+      "loss": 1.1553,
+      "step": 63350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5755470309403195e-05,
+      "loss": 1.1639,
+      "step": 63355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5754852320709597e-05,
+      "loss": 1.0748,
+      "step": 63360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5754234299152696e-05,
+      "loss": 1.2855,
+      "step": 63365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5753616244736023e-05,
+      "loss": 1.3801,
+      "step": 63370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5752998157463107e-05,
+      "loss": 1.234,
+      "step": 63375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5752380037337477e-05,
+      "loss": 1.5989,
+      "step": 63380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5751761884362662e-05,
+      "loss": 1.4943,
+      "step": 63385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5751143698542197e-05,
+      "loss": 1.0592,
+      "step": 63390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.57505254798796e-05,
+      "loss": 1.1436,
+      "step": 63395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5749907228378418e-05,
+      "loss": 1.3654,
+      "step": 63400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5749288944042166e-05,
+      "loss": 1.5014,
+      "step": 63405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.574867062687439e-05,
+      "loss": 1.133,
+      "step": 63410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5748052276878605e-05,
+      "loss": 1.2236,
+      "step": 63415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5747433894058354e-05,
+      "loss": 1.1364,
+      "step": 63420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5746815478417166e-05,
+      "loss": 1.272,
+      "step": 63425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5746197029958566e-05,
+      "loss": 0.9208,
+      "step": 63430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.574557854868609e-05,
+      "loss": 1.5958,
+      "step": 63435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5744960034603278e-05,
+      "loss": 1.2915,
+      "step": 63440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5744341487713643e-05,
+      "loss": 1.1995,
+      "step": 63445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.574372290802074e-05,
+      "loss": 1.3839,
+      "step": 63450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5743104295528084e-05,
+      "loss": 1.0788,
+      "step": 63455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.574248565023921e-05,
+      "loss": 1.6351,
+      "step": 63460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.574186697215766e-05,
+      "loss": 1.6435,
+      "step": 63465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.574124826128696e-05,
+      "loss": 1.0672,
+      "step": 63470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5740629517630642e-05,
+      "loss": 1.1629,
+      "step": 63475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5740010741192246e-05,
+      "loss": 1.1473,
+      "step": 63480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5739391931975298e-05,
+      "loss": 1.2608,
+      "step": 63485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5738773089983338e-05,
+      "loss": 1.4889,
+      "step": 63490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5738154215219893e-05,
+      "loss": 1.3186,
+      "step": 63495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.57375353076885e-05,
+      "loss": 1.3068,
+      "step": 63500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5736916367392703e-05,
+      "loss": 0.998,
+      "step": 63505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.573629739433602e-05,
+      "loss": 1.3449,
+      "step": 63510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5735678388522e-05,
+      "loss": 1.1997,
+      "step": 63515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.573505934995417e-05,
+      "loss": 1.1726,
+      "step": 63520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.573444027863606e-05,
+      "loss": 1.0077,
+      "step": 63525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.573382117457122e-05,
+      "loss": 1.2137,
+      "step": 63530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.573320203776317e-05,
+      "loss": 1.0715,
+      "step": 63535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.573258286821546e-05,
+      "loss": 1.0493,
+      "step": 63540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.573196366593161e-05,
+      "loss": 1.1444,
+      "step": 63545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.573134443091517e-05,
+      "loss": 1.102,
+      "step": 63550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5730725163169672e-05,
+      "loss": 1.2812,
+      "step": 63555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.573010586269865e-05,
+      "loss": 1.2478,
+      "step": 63560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5729486529505636e-05,
+      "loss": 1.3305,
+      "step": 63565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5728867163594182e-05,
+      "loss": 1.149,
+      "step": 63570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5728247764967806e-05,
+      "loss": 1.6193,
+      "step": 63575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5727628333630057e-05,
+      "loss": 1.4518,
+      "step": 63580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5727008869584473e-05,
+      "loss": 1.1928,
+      "step": 63585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5726389372834585e-05,
+      "loss": 0.9628,
+      "step": 63590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.572576984338394e-05,
+      "loss": 1.3104,
+      "step": 63595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.572515028123606e-05,
+      "loss": 0.946,
+      "step": 63600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5724530686394498e-05,
+      "loss": 1.0344,
+      "step": 63605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5723911058862787e-05,
+      "loss": 1.4075,
+      "step": 63610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5723291398644464e-05,
+      "loss": 1.011,
+      "step": 63615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5722671705743066e-05,
+      "loss": 1.5228,
+      "step": 63620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5722051980162138e-05,
+      "loss": 0.9495,
+      "step": 63625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5721432221905216e-05,
+      "loss": 1.2059,
+      "step": 63630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.572081243097584e-05,
+      "loss": 1.1836,
+      "step": 63635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5720192607377542e-05,
+      "loss": 1.2066,
+      "step": 63640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.571957275111387e-05,
+      "loss": 1.1966,
+      "step": 63645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5718952862188364e-05,
+      "loss": 1.2437,
+      "step": 63650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.571833294060456e-05,
+      "loss": 1.1794,
+      "step": 63655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5717712986366e-05,
+      "loss": 1.324,
+      "step": 63660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5717092999476222e-05,
+      "loss": 1.2095,
+      "step": 63665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5716472979938767e-05,
+      "loss": 1.4393,
+      "step": 63670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.571585292775718e-05,
+      "loss": 1.2325,
+      "step": 63675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5715232842934994e-05,
+      "loss": 1.1523,
+      "step": 63680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.571461272547576e-05,
+      "loss": 1.206,
+      "step": 63685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5713992575383014e-05,
+      "loss": 0.9012,
+      "step": 63690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.571337239266029e-05,
+      "loss": 1.282,
+      "step": 63695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5712752177311143e-05,
+      "loss": 1.5227,
+      "step": 63700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5712131929339105e-05,
+      "loss": 1.2185,
+      "step": 63705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5711511648747722e-05,
+      "loss": 1.1987,
+      "step": 63710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.571089133554054e-05,
+      "loss": 1.8264,
+      "step": 63715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.571027098972109e-05,
+      "loss": 1.6921,
+      "step": 63720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5709650611292927e-05,
+      "loss": 1.2848,
+      "step": 63725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5709030200259585e-05,
+      "loss": 1.1524,
+      "step": 63730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.570840975662461e-05,
+      "loss": 1.1328,
+      "step": 63735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5707789280391547e-05,
+      "loss": 1.3297,
+      "step": 63740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5707168771563935e-05,
+      "loss": 1.0244,
+      "step": 63745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.570654823014532e-05,
+      "loss": 1.168,
+      "step": 63750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5705927656139245e-05,
+      "loss": 1.2633,
+      "step": 63755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5705307049549255e-05,
+      "loss": 1.3575,
+      "step": 63760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.570468641037889e-05,
+      "loss": 1.3536,
+      "step": 63765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5704065738631705e-05,
+      "loss": 1.3146,
+      "step": 63770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.570344503431123e-05,
+      "loss": 1.0042,
+      "step": 63775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.570282429742102e-05,
+      "loss": 1.355,
+      "step": 63780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5702203527964613e-05,
+      "loss": 1.2619,
+      "step": 63785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.570158272594556e-05,
+      "loss": 1.1116,
+      "step": 63790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.57009618913674e-05,
+      "loss": 1.3789,
+      "step": 63795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5700341024233682e-05,
+      "loss": 1.8281,
+      "step": 63800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5699720124547954e-05,
+      "loss": 1.0729,
+      "step": 63805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5699099192313756e-05,
+      "loss": 1.4604,
+      "step": 63810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5698478227534634e-05,
+      "loss": 1.4702,
+      "step": 63815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.569785723021414e-05,
+      "loss": 1.1726,
+      "step": 63820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5697236200355816e-05,
+      "loss": 1.4614,
+      "step": 63825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5696615137963208e-05,
+      "loss": 1.0985,
+      "step": 63830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5695994043039862e-05,
+      "loss": 0.9047,
+      "step": 63835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.569537291558933e-05,
+      "loss": 1.3353,
+      "step": 63840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5694751755615155e-05,
+      "loss": 1.3626,
+      "step": 63845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.569413056312088e-05,
+      "loss": 1.3427,
+      "step": 63850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.569350933811006e-05,
+      "loss": 1.5698,
+      "step": 63855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.569288808058624e-05,
+      "loss": 1.6834,
+      "step": 63860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5692266790552966e-05,
+      "loss": 1.431,
+      "step": 63865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5691645468013788e-05,
+      "loss": 1.3121,
+      "step": 63870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5691024112972252e-05,
+      "loss": 1.1456,
+      "step": 63875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.569040272543191e-05,
+      "loss": 1.4844,
+      "step": 63880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5689781305396304e-05,
+      "loss": 1.1095,
+      "step": 63885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.568915985286899e-05,
+      "loss": 1.3482,
+      "step": 63890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.568853836785351e-05,
+      "loss": 1.1096,
+      "step": 63895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.568791685035342e-05,
+      "loss": 1.342,
+      "step": 63900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5687295300372263e-05,
+      "loss": 1.1974,
+      "step": 63905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5686673717913592e-05,
+      "loss": 1.2565,
+      "step": 63910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5686052102980955e-05,
+      "loss": 1.4973,
+      "step": 63915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.56854304555779e-05,
+      "loss": 0.9825,
+      "step": 63920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5684808775707984e-05,
+      "loss": 1.2527,
+      "step": 63925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.568418706337475e-05,
+      "loss": 1.2055,
+      "step": 63930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.568356531858175e-05,
+      "loss": 1.2999,
+      "step": 63935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5682943541332535e-05,
+      "loss": 1.374,
+      "step": 63940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5682321731630656e-05,
+      "loss": 1.2746,
+      "step": 63945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5681699889479665e-05,
+      "loss": 1.4504,
+      "step": 63950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.568107801488311e-05,
+      "loss": 1.1782,
+      "step": 63955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5680456107844543e-05,
+      "loss": 1.4675,
+      "step": 63960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.567983416836752e-05,
+      "loss": 1.2992,
+      "step": 63965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5679212196455587e-05,
+      "loss": 1.0227,
+      "step": 63970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5678590192112296e-05,
+      "loss": 1.0965,
+      "step": 63975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5677968155341205e-05,
+      "loss": 1.0843,
+      "step": 63980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.567734608614586e-05,
+      "loss": 1.4963,
+      "step": 63985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5676723984529815e-05,
+      "loss": 0.9652,
+      "step": 63990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.567610185049662e-05,
+      "loss": 1.4445,
+      "step": 63995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5675479684049834e-05,
+      "loss": 1.2462,
+      "step": 64000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5674857485193005e-05,
+      "loss": 1.004,
+      "step": 64005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.567423525392969e-05,
+      "loss": 1.1284,
+      "step": 64010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5673612990263435e-05,
+      "loss": 1.0448,
+      "step": 64015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5672990694197802e-05,
+      "loss": 1.4272,
+      "step": 64020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.567236836573634e-05,
+      "loss": 1.3684,
+      "step": 64025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5671746004882603e-05,
+      "loss": 1.2009,
+      "step": 64030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5671123611640146e-05,
+      "loss": 1.3506,
+      "step": 64035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5670501186012523e-05,
+      "loss": 0.9845,
+      "step": 64040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.566987872800329e-05,
+      "loss": 1.5071,
+      "step": 64045
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5669256237615997e-05,
+      "loss": 1.1031,
+      "step": 64050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5668633714854206e-05,
+      "loss": 1.3003,
+      "step": 64055
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5668011159721462e-05,
+      "loss": 1.1256,
+      "step": 64060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.566738857222133e-05,
+      "loss": 1.3087,
+      "step": 64065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5666765952357362e-05,
+      "loss": 1.3127,
+      "step": 64070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.566614330013311e-05,
+      "loss": 1.5488,
+      "step": 64075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.566552061555213e-05,
+      "loss": 1.4949,
+      "step": 64080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5664897898617984e-05,
+      "loss": 1.1566,
+      "step": 64085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5664275149334223e-05,
+      "loss": 1.2699,
+      "step": 64090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.56636523677044e-05,
+      "loss": 1.1971,
+      "step": 64095
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5663029553732084e-05,
+      "loss": 0.8558,
+      "step": 64100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5662406707420817e-05,
+      "loss": 1.0275,
+      "step": 64105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5661783828774163e-05,
+      "loss": 0.9933,
+      "step": 64110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.566116091779568e-05,
+      "loss": 1.3226,
+      "step": 64115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5660537974488924e-05,
+      "loss": 1.6066,
+      "step": 64120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.565991499885745e-05,
+      "loss": 1.1505,
+      "step": 64125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5659291990904817e-05,
+      "loss": 1.2915,
+      "step": 64130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5658668950634586e-05,
+      "loss": 1.3765,
+      "step": 64135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.565804587805031e-05,
+      "loss": 1.3023,
+      "step": 64140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5657422773155544e-05,
+      "loss": 1.1148,
+      "step": 64145
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5656799635953858e-05,
+      "loss": 1.1237,
+      "step": 64150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5656176466448798e-05,
+      "loss": 1.0487,
+      "step": 64155
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5655553264643933e-05,
+      "loss": 1.0446,
+      "step": 64160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5654930030542816e-05,
+      "loss": 1.483,
+      "step": 64165
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5654306764149005e-05,
+      "loss": 1.287,
+      "step": 64170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.565368346546606e-05,
+      "loss": 1.0558,
+      "step": 64175
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5653060134497544e-05,
+      "loss": 1.2962,
+      "step": 64180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5652436771247013e-05,
+      "loss": 1.3921,
+      "step": 64185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.565181337571803e-05,
+      "loss": 1.3539,
+      "step": 64190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.565118994791415e-05,
+      "loss": 0.9637,
+      "step": 64195
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5650566487838935e-05,
+      "loss": 1.053,
+      "step": 64200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.564994299549595e-05,
+      "loss": 0.8878,
+      "step": 64205
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.564931947088875e-05,
+      "loss": 1.2069,
+      "step": 64210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5648695914020898e-05,
+      "loss": 1.1798,
+      "step": 64215
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5648072324895953e-05,
+      "loss": 1.2273,
+      "step": 64220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.564744870351748e-05,
+      "loss": 1.0025,
+      "step": 64225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5646825049889033e-05,
+      "loss": 1.5419,
+      "step": 64230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.564620136401418e-05,
+      "loss": 1.1788,
+      "step": 64235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5645577645896485e-05,
+      "loss": 1.137,
+      "step": 64240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5644953895539497e-05,
+      "loss": 1.4922,
+      "step": 64245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.564433011294679e-05,
+      "loss": 1.3651,
+      "step": 64250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.564370629812192e-05,
+      "loss": 1.4503,
+      "step": 64255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5643082451068453e-05,
+      "loss": 1.8787,
+      "step": 64260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.564245857178995e-05,
+      "loss": 1.2834,
+      "step": 64265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5641834660289975e-05,
+      "loss": 0.8064,
+      "step": 64270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5641210716572092e-05,
+      "loss": 1.2646,
+      "step": 64275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.564058674063986e-05,
+      "loss": 1.2581,
+      "step": 64280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.563996273249684e-05,
+      "loss": 1.0107,
+      "step": 64285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5639338692146598e-05,
+      "loss": 1.0252,
+      "step": 64290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5638714619592707e-05,
+      "loss": 1.3349,
+      "step": 64295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.563809051483872e-05,
+      "loss": 1.081,
+      "step": 64300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.56374663778882e-05,
+      "loss": 1.097,
+      "step": 64305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5636842208744714e-05,
+      "loss": 0.932,
+      "step": 64310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5636218007411833e-05,
+      "loss": 1.2444,
+      "step": 64315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5635593773893113e-05,
+      "loss": 1.3986,
+      "step": 64320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.563496950819212e-05,
+      "loss": 1.3793,
+      "step": 64325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5634345210312423e-05,
+      "loss": 1.0975,
+      "step": 64330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5633720880257585e-05,
+      "loss": 1.5135,
+      "step": 64335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5633096518031168e-05,
+      "loss": 1.0193,
+      "step": 64340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.563247212363674e-05,
+      "loss": 1.0827,
+      "step": 64345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5631847697077867e-05,
+      "loss": 1.2578,
+      "step": 64350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5631223238358114e-05,
+      "loss": 1.3692,
+      "step": 64355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.563059874748105e-05,
+      "loss": 1.4355,
+      "step": 64360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5629974224450236e-05,
+      "loss": 1.1803,
+      "step": 64365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5629349669269243e-05,
+      "loss": 1.1569,
+      "step": 64370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5628725081941635e-05,
+      "loss": 1.28,
+      "step": 64375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.562810046247098e-05,
+      "loss": 1.128,
+      "step": 64380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5627475810860843e-05,
+      "loss": 1.065,
+      "step": 64385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5626851127114794e-05,
+      "loss": 1.2422,
+      "step": 64390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5626226411236397e-05,
+      "loss": 1.2444,
+      "step": 64395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.562560166322922e-05,
+      "loss": 1.4537,
+      "step": 64400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5624976883096837e-05,
+      "loss": 1.6815,
+      "step": 64405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5624352070842808e-05,
+      "loss": 1.2793,
+      "step": 64410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5623727226470702e-05,
+      "loss": 1.1842,
+      "step": 64415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5623102349984088e-05,
+      "loss": 1.1969,
+      "step": 64420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.562247744138654e-05,
+      "loss": 1.2243,
+      "step": 64425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5621852500681615e-05,
+      "loss": 0.9557,
+      "step": 64430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5621227527872894e-05,
+      "loss": 1.2511,
+      "step": 64435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.562060252296394e-05,
+      "loss": 1.5845,
+      "step": 64440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.561997748595832e-05,
+      "loss": 1.0103,
+      "step": 64445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5619352416859605e-05,
+      "loss": 1.1439,
+      "step": 64450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.561872731567137e-05,
+      "loss": 1.4323,
+      "step": 64455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5618102182397176e-05,
+      "loss": 1.0409,
+      "step": 64460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.56174770170406e-05,
+      "loss": 0.9687,
+      "step": 64465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5616851819605203e-05,
+      "loss": 1.2983,
+      "step": 64470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5616226590094565e-05,
+      "loss": 1.3075,
+      "step": 64475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5615601328512255e-05,
+      "loss": 1.1392,
+      "step": 64480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5614976034861836e-05,
+      "loss": 1.4633,
+      "step": 64485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5614350709146887e-05,
+      "loss": 1.0881,
+      "step": 64490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5613725351370975e-05,
+      "loss": 1.0815,
+      "step": 64495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5613099961537672e-05,
+      "loss": 1.1801,
+      "step": 64500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.561247453965055e-05,
+      "loss": 1.1264,
+      "step": 64505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5611849085713177e-05,
+      "loss": 1.1694,
+      "step": 64510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5611223599729132e-05,
+      "loss": 1.0835,
+      "step": 64515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5610598081701976e-05,
+      "loss": 1.3402,
+      "step": 64520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.560997253163529e-05,
+      "loss": 1.1064,
+      "step": 64525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5609346949532644e-05,
+      "loss": 1.2479,
+      "step": 64530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5608721335397613e-05,
+      "loss": 1.1214,
+      "step": 64535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5608095689233765e-05,
+      "loss": 1.3154,
+      "step": 64540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5607470011044673e-05,
+      "loss": 1.5059,
+      "step": 64545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.560684430083391e-05,
+      "loss": 1.21,
+      "step": 64550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.560621855860505e-05,
+      "loss": 1.4415,
+      "step": 64555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.560559278436167e-05,
+      "loss": 1.1611,
+      "step": 64560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.560496697810734e-05,
+      "loss": 1.0599,
+      "step": 64565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5604341139845632e-05,
+      "loss": 0.9963,
+      "step": 64570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.560371526958012e-05,
+      "loss": 1.207,
+      "step": 64575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.560308936731438e-05,
+      "loss": 1.0464,
+      "step": 64580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.560246343305199e-05,
+      "loss": 1.1113,
+      "step": 64585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5601837466796518e-05,
+      "loss": 1.0373,
+      "step": 64590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5601211468551544e-05,
+      "loss": 1.0103,
+      "step": 64595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5600585438320638e-05,
+      "loss": 1.0334,
+      "step": 64600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5599959376107373e-05,
+      "loss": 1.2335,
+      "step": 64605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5599333281915335e-05,
+      "loss": 1.2296,
+      "step": 64610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5598707155748088e-05,
+      "loss": 1.475,
+      "step": 64615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.559808099760921e-05,
+      "loss": 0.98,
+      "step": 64620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5597454807502284e-05,
+      "loss": 1.1268,
+      "step": 64625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5596828585430873e-05,
+      "loss": 1.2341,
+      "step": 64630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5596202331398568e-05,
+      "loss": 1.0136,
+      "step": 64635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5595576045408936e-05,
+      "loss": 1.185,
+      "step": 64640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5594949727465552e-05,
+      "loss": 1.0661,
+      "step": 64645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5594323377572e-05,
+      "loss": 1.2888,
+      "step": 64650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.559369699573185e-05,
+      "loss": 1.1374,
+      "step": 64655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.559307058194868e-05,
+      "loss": 0.9881,
+      "step": 64660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5592444136226073e-05,
+      "loss": 1.3752,
+      "step": 64665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.55918176585676e-05,
+      "loss": 1.119,
+      "step": 64670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.559119114897684e-05,
+      "loss": 1.0828,
+      "step": 64675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5590564607457374e-05,
+      "loss": 1.0867,
+      "step": 64680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5589938034012773e-05,
+      "loss": 1.0097,
+      "step": 64685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5589311428646624e-05,
+      "loss": 1.0065,
+      "step": 64690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5588684791362494e-05,
+      "loss": 0.9807,
+      "step": 64695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5588058122163973e-05,
+      "loss": 1.0879,
+      "step": 64700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5587431421054634e-05,
+      "loss": 1.0238,
+      "step": 64705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5586804688038055e-05,
+      "loss": 0.9737,
+      "step": 64710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5586177923117818e-05,
+      "loss": 1.1988,
+      "step": 64715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.55855511262975e-05,
+      "loss": 1.2695,
+      "step": 64720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.558492429758068e-05,
+      "loss": 1.2348,
+      "step": 64725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.558429743697094e-05,
+      "loss": 1.2304,
+      "step": 64730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5583670544471853e-05,
+      "loss": 1.2283,
+      "step": 64735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.558304362008701e-05,
+      "loss": 1.3942,
+      "step": 64740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5582416663819984e-05,
+      "loss": 1.2445,
+      "step": 64745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5581789675674354e-05,
+      "loss": 1.3892,
+      "step": 64750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5581162655653705e-05,
+      "loss": 1.006,
+      "step": 64755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5580535603761612e-05,
+      "loss": 1.6209,
+      "step": 64760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5579908520001663e-05,
+      "loss": 0.9037,
+      "step": 64765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5579281404377432e-05,
+      "loss": 1.5188,
+      "step": 64770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5578654256892507e-05,
+      "loss": 1.018,
+      "step": 64775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5578027077550463e-05,
+      "loss": 1.3563,
+      "step": 64780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.557739986635489e-05,
+      "loss": 1.2785,
+      "step": 64785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5576772623309357e-05,
+      "loss": 1.6279,
+      "step": 64790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5576145348417453e-05,
+      "loss": 1.1876,
+      "step": 64795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5575518041682762e-05,
+      "loss": 1.179,
+      "step": 64800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5574890703108864e-05,
+      "loss": 1.2921,
+      "step": 64805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5574263332699346e-05,
+      "loss": 1.0334,
+      "step": 64810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.557363593045778e-05,
+      "loss": 1.0255,
+      "step": 64815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5573008496387755e-05,
+      "loss": 1.3676,
+      "step": 64820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5572381030492856e-05,
+      "loss": 1.5002,
+      "step": 64825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5571753532776663e-05,
+      "loss": 1.4074,
+      "step": 64830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5571126003242762e-05,
+      "loss": 1.1444,
+      "step": 64835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5570498441894738e-05,
+      "loss": 1.097,
+      "step": 64840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5569870848736167e-05,
+      "loss": 1.2365,
+      "step": 64845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.556924322377064e-05,
+      "loss": 1.2084,
+      "step": 64850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5568615567001737e-05,
+      "loss": 1.1479,
+      "step": 64855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5567987878433047e-05,
+      "loss": 1.3506,
+      "step": 64860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.556736015806815e-05,
+      "loss": 1.0889,
+      "step": 64865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5566732405910628e-05,
+      "loss": 1.118,
+      "step": 64870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5566104621964076e-05,
+      "loss": 1.1168,
+      "step": 64875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5565476806232068e-05,
+      "loss": 1.0387,
+      "step": 64880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5564848958718197e-05,
+      "loss": 1.3653,
+      "step": 64885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.556422107942605e-05,
+      "loss": 1.2341,
+      "step": 64890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5563593168359198e-05,
+      "loss": 0.9177,
+      "step": 64895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.556296522552124e-05,
+      "loss": 1.1378,
+      "step": 64900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.556233725091576e-05,
+      "loss": 1.249,
+      "step": 64905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5561709244546345e-05,
+      "loss": 1.1244,
+      "step": 64910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5561081206416577e-05,
+      "loss": 1.103,
+      "step": 64915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.556045313653004e-05,
+      "loss": 1.1425,
+      "step": 64920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.555982503489033e-05,
+      "loss": 1.3342,
+      "step": 64925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.555919690150103e-05,
+      "loss": 1.3547,
+      "step": 64930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.555856873636572e-05,
+      "loss": 1.0941,
+      "step": 64935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5557940539487995e-05,
+      "loss": 1.5256,
+      "step": 64940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.555731231087144e-05,
+      "loss": 1.2726,
+      "step": 64945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5556684050519643e-05,
+      "loss": 1.001,
+      "step": 64950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5556055758436196e-05,
+      "loss": 1.7569,
+      "step": 64955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5555427434624674e-05,
+      "loss": 1.1164,
+      "step": 64960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.555479907908868e-05,
+      "loss": 1.2435,
+      "step": 64965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.555417069183179e-05,
+      "loss": 1.5697,
+      "step": 64970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.55535422728576e-05,
+      "loss": 0.9792,
+      "step": 64975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.55529138221697e-05,
+      "loss": 1.2125,
+      "step": 64980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5552285339771674e-05,
+      "loss": 1.2606,
+      "step": 64985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.555165682566711e-05,
+      "loss": 1.111,
+      "step": 64990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.55510282798596e-05,
+      "loss": 1.2746,
+      "step": 64995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5550399702352734e-05,
+      "loss": 1.3304,
+      "step": 65000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5549771093150103e-05,
+      "loss": 1.115,
+      "step": 65005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.554914245225529e-05,
+      "loss": 1.1155,
+      "step": 65010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.554851377967189e-05,
+      "loss": 1.2526,
+      "step": 65015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5547885075403496e-05,
+      "loss": 1.1299,
+      "step": 65020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.554725633945369e-05,
+      "loss": 1.2849,
+      "step": 65025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5546627571826066e-05,
+      "loss": 1.081,
+      "step": 65030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5545998772524217e-05,
+      "loss": 1.3254,
+      "step": 65035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.554536994155173e-05,
+      "loss": 1.4302,
+      "step": 65040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.55447410789122e-05,
+      "loss": 0.9537,
+      "step": 65045
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.554411218460922e-05,
+      "loss": 1.1413,
+      "step": 65050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5543483258646374e-05,
+      "loss": 1.2892,
+      "step": 65055
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5542854301027253e-05,
+      "loss": 1.2593,
+      "step": 65060
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5542225311755457e-05,
+      "loss": 1.2362,
+      "step": 65065
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5541596290834573e-05,
+      "loss": 1.3359,
+      "step": 65070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5540967238268195e-05,
+      "loss": 1.2575,
+      "step": 65075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5540338154059913e-05,
+      "loss": 1.3368,
+      "step": 65080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5539709038213316e-05,
+      "loss": 1.3094,
+      "step": 65085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5539079890732005e-05,
+      "loss": 1.1736,
+      "step": 65090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.553845071161957e-05,
+      "loss": 1.2121,
+      "step": 65095
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.55378215008796e-05,
+      "loss": 1.2955,
+      "step": 65100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.553719225851569e-05,
+      "loss": 1.0028,
+      "step": 65105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.553656298453144e-05,
+      "loss": 1.269,
+      "step": 65110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.553593367893043e-05,
+      "loss": 1.3783,
+      "step": 65115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5535304341716263e-05,
+      "loss": 1.4146,
+      "step": 65120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5534674972892534e-05,
+      "loss": 1.1298,
+      "step": 65125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.553404557246283e-05,
+      "loss": 1.4405,
+      "step": 65130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5533416140430754e-05,
+      "loss": 0.9368,
+      "step": 65135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.553278667679989e-05,
+      "loss": 1.0675,
+      "step": 65140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5532157181573844e-05,
+      "loss": 1.0146,
+      "step": 65145
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5531527654756203e-05,
+      "loss": 1.1852,
+      "step": 65150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5530898096350562e-05,
+      "loss": 1.2541,
+      "step": 65155
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.553026850636052e-05,
+      "loss": 1.0633,
+      "step": 65160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.552963888478967e-05,
+      "loss": 1.2184,
+      "step": 65165
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5529009231641605e-05,
+      "loss": 0.9249,
+      "step": 65170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5528379546919926e-05,
+      "loss": 1.2135,
+      "step": 65175
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.552774983062823e-05,
+      "loss": 1.2107,
+      "step": 65180
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5527120082770105e-05,
+      "loss": 1.4152,
+      "step": 65185
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5526490303349153e-05,
+      "loss": 1.1235,
+      "step": 65190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5525860492368967e-05,
+      "loss": 1.015,
+      "step": 65195
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5525230649833144e-05,
+      "loss": 1.1048,
+      "step": 65200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5524600775745286e-05,
+      "loss": 1.0254,
+      "step": 65205
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5523970870108988e-05,
+      "loss": 1.1378,
+      "step": 65210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5523340932927838e-05,
+      "loss": 0.9882,
+      "step": 65215
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5522710964205448e-05,
+      "loss": 1.8187,
+      "step": 65220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5522080963945402e-05,
+      "loss": 1.1058,
+      "step": 65225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.552145093215131e-05,
+      "loss": 1.0599,
+      "step": 65230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.552082086882676e-05,
+      "loss": 1.4046,
+      "step": 65235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5520190773975353e-05,
+      "loss": 1.1022,
+      "step": 65240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5519560647600687e-05,
+      "loss": 1.4254,
+      "step": 65245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5518930489706363e-05,
+      "loss": 1.302,
+      "step": 65250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5518300300295975e-05,
+      "loss": 1.1709,
+      "step": 65255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5517670079373127e-05,
+      "loss": 1.2754,
+      "step": 65260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5517039826941415e-05,
+      "loss": 1.0837,
+      "step": 65265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5516409543004437e-05,
+      "loss": 0.8967,
+      "step": 65270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5515779227565793e-05,
+      "loss": 1.1455,
+      "step": 65275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5515148880629084e-05,
+      "loss": 1.5419,
+      "step": 65280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5514518502197907e-05,
+      "loss": 1.305,
+      "step": 65285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5513888092275865e-05,
+      "loss": 1.0971,
+      "step": 65290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.551325765086656e-05,
+      "loss": 1.0375,
+      "step": 65295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.551262717797358e-05,
+      "loss": 1.1409,
+      "step": 65300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.551199667360054e-05,
+      "loss": 1.0871,
+      "step": 65305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5511366137751032e-05,
+      "loss": 1.2485,
+      "step": 65310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.551073557042866e-05,
+      "loss": 1.3954,
+      "step": 65315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.551010497163702e-05,
+      "loss": 1.7874,
+      "step": 65320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5509474341379726e-05,
+      "loss": 1.2846,
+      "step": 65325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5508843679660362e-05,
+      "loss": 1.6552,
+      "step": 65330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5508212986482544e-05,
+      "loss": 1.0362,
+      "step": 65335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.550758226184986e-05,
+      "loss": 1.2271,
+      "step": 65340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5506951505765922e-05,
+      "loss": 1.3128,
+      "step": 65345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5506320718234333e-05,
+      "loss": 1.362,
+      "step": 65350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5505689899258688e-05,
+      "loss": 1.2846,
+      "step": 65355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.550505904884259e-05,
+      "loss": 1.0699,
+      "step": 65360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5504428166989645e-05,
+      "loss": 1.3549,
+      "step": 65365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5503797253703458e-05,
+      "loss": 1.4311,
+      "step": 65370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.550316630898763e-05,
+      "loss": 1.2224,
+      "step": 65375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5502535332845752e-05,
+      "loss": 1.05,
+      "step": 65380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5501904325281447e-05,
+      "loss": 1.1351,
+      "step": 65385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5501273286298305e-05,
+      "loss": 1.1115,
+      "step": 65390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5500642215899933e-05,
+      "loss": 1.2159,
+      "step": 65395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.550001111408994e-05,
+      "loss": 0.9024,
+      "step": 65400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5499379980871924e-05,
+      "loss": 1.2062,
+      "step": 65405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.549874881624949e-05,
+      "loss": 1.2433,
+      "step": 65410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.549811762022624e-05,
+      "loss": 1.1223,
+      "step": 65415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5497486392805787e-05,
+      "loss": 1.2547,
+      "step": 65420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5496855133991723e-05,
+      "loss": 1.5485,
+      "step": 65425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5496223843787665e-05,
+      "loss": 1.3794,
+      "step": 65430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5495592522197207e-05,
+      "loss": 0.9898,
+      "step": 65435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5494961169223966e-05,
+      "loss": 1.2571,
+      "step": 65440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5494329784871538e-05,
+      "loss": 1.0758,
+      "step": 65445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5493698369143533e-05,
+      "loss": 0.9871,
+      "step": 65450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5493066922043555e-05,
+      "loss": 1.3285,
+      "step": 65455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.549243544357521e-05,
+      "loss": 1.472,
+      "step": 65460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5491803933742105e-05,
+      "loss": 1.1211,
+      "step": 65465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5491172392547843e-05,
+      "loss": 1.1558,
+      "step": 65470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5490540819996035e-05,
+      "loss": 1.0927,
+      "step": 65475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5489909216090292e-05,
+      "loss": 1.032,
+      "step": 65480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5489277580834207e-05,
+      "loss": 1.3302,
+      "step": 65485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5488645914231397e-05,
+      "loss": 1.3714,
+      "step": 65490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.548801421628547e-05,
+      "loss": 0.994,
+      "step": 65495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5487382487000022e-05,
+      "loss": 1.304,
+      "step": 65500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5486750726378673e-05,
+      "loss": 1.6474,
+      "step": 65505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.548611893442503e-05,
+      "loss": 1.2459,
+      "step": 65510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5485487111142693e-05,
+      "loss": 1.4147,
+      "step": 65515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5484855256535273e-05,
+      "loss": 1.312,
+      "step": 65520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.548422337060638e-05,
+      "loss": 1.2962,
+      "step": 65525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5483591453359627e-05,
+      "loss": 1.2515,
+      "step": 65530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5482959504798614e-05,
+      "loss": 1.0406,
+      "step": 65535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.548232752492695e-05,
+      "loss": 1.4646,
+      "step": 65540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.548169551374825e-05,
+      "loss": 1.0907,
+      "step": 65545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5481063471266122e-05,
+      "loss": 1.2073,
+      "step": 65550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5480431397484173e-05,
+      "loss": 1.2614,
+      "step": 65555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5479799292406013e-05,
+      "loss": 1.2849,
+      "step": 65560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5479167156035248e-05,
+      "loss": 1.2072,
+      "step": 65565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5478534988375496e-05,
+      "loss": 1.1346,
+      "step": 65570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5477902789430358e-05,
+      "loss": 0.9662,
+      "step": 65575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5477270559203454e-05,
+      "loss": 1.4417,
+      "step": 65580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5476638297698387e-05,
+      "loss": 1.2862,
+      "step": 65585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.547600600491877e-05,
+      "loss": 0.9943,
+      "step": 65590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.547537368086821e-05,
+      "loss": 0.9138,
+      "step": 65595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5474741325550325e-05,
+      "loss": 1.195,
+      "step": 65600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5474108938968724e-05,
+      "loss": 1.2795,
+      "step": 65605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.547347652112701e-05,
+      "loss": 1.3909,
+      "step": 65610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5472844072028808e-05,
+      "loss": 1.2725,
+      "step": 65615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.547221159167772e-05,
+      "loss": 1.0841,
+      "step": 65620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5471579080077362e-05,
+      "loss": 1.2939,
+      "step": 65625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5470946537231345e-05,
+      "loss": 1.0023,
+      "step": 65630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5470313963143277e-05,
+      "loss": 1.8093,
+      "step": 65635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5469681357816776e-05,
+      "loss": 1.5874,
+      "step": 65640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5469048721255455e-05,
+      "loss": 1.1557,
+      "step": 65645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.546841605346292e-05,
+      "loss": 1.2148,
+      "step": 65650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5467783354442788e-05,
+      "loss": 1.234,
+      "step": 65655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5467150624198674e-05,
+      "loss": 1.3185,
+      "step": 65660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5466517862734192e-05,
+      "loss": 1.3455,
+      "step": 65665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.546588507005295e-05,
+      "loss": 1.2157,
+      "step": 65670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5465252246158563e-05,
+      "loss": 1.3445,
+      "step": 65675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.546461939105465e-05,
+      "loss": 1.1458,
+      "step": 65680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.546398650474482e-05,
+      "loss": 1.3248,
+      "step": 65685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5463353587232685e-05,
+      "loss": 1.1563,
+      "step": 65690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5462720638521867e-05,
+      "loss": 1.6418,
+      "step": 65695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5462087658615974e-05,
+      "loss": 1.06,
+      "step": 65700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5461454647518624e-05,
+      "loss": 1.1537,
+      "step": 65705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.546082160523343e-05,
+      "loss": 1.3978,
+      "step": 65710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5460188531764003e-05,
+      "loss": 1.0909,
+      "step": 65715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.545955542711397e-05,
+      "loss": 1.1139,
+      "step": 65720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.545892229128693e-05,
+      "loss": 1.2139,
+      "step": 65725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5458289124286517e-05,
+      "loss": 1.3903,
+      "step": 65730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5457655926116334e-05,
+      "loss": 1.3365,
+      "step": 65735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.545702269678e-05,
+      "loss": 1.343,
+      "step": 65740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5456389436281128e-05,
+      "loss": 1.3433,
+      "step": 65745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5455756144623346e-05,
+      "loss": 0.9905,
+      "step": 65750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5455122821810255e-05,
+      "loss": 1.0996,
+      "step": 65755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5454489467845478e-05,
+      "loss": 1.151,
+      "step": 65760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5453856082732636e-05,
+      "loss": 1.317,
+      "step": 65765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5453222666475342e-05,
+      "loss": 1.2107,
+      "step": 65770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5452589219077213e-05,
+      "loss": 1.0841,
+      "step": 65775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5451955740541865e-05,
+      "loss": 1.3088,
+      "step": 65780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.545132223087292e-05,
+      "loss": 1.5704,
+      "step": 65785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5450688690073993e-05,
+      "loss": 1.2812,
+      "step": 65790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.54500551181487e-05,
+      "loss": 1.4332,
+      "step": 65795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5449421515100663e-05,
+      "loss": 1.2702,
+      "step": 65800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5448787880933494e-05,
+      "loss": 1.4087,
+      "step": 65805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.544815421565082e-05,
+      "loss": 1.1344,
+      "step": 65810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.544752051925625e-05,
+      "loss": 1.3024,
+      "step": 65815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5446886791753413e-05,
+      "loss": 1.2817,
+      "step": 65820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5446253033145923e-05,
+      "loss": 1.3594,
+      "step": 65825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5445619243437396e-05,
+      "loss": 1.3034,
+      "step": 65830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5444985422631454e-05,
+      "loss": 1.4458,
+      "step": 65835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.544435157073172e-05,
+      "loss": 1.5975,
+      "step": 65840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5443717687741807e-05,
+      "loss": 1.2325,
+      "step": 65845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.544308377366534e-05,
+      "loss": 1.4591,
+      "step": 65850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.544244982850594e-05,
+      "loss": 0.8455,
+      "step": 65855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.544181585226722e-05,
+      "loss": 1.1222,
+      "step": 65860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5441181844952806e-05,
+      "loss": 1.2538,
+      "step": 65865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5440547806566315e-05,
+      "loss": 1.2345,
+      "step": 65870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.543991373711137e-05,
+      "loss": 1.271,
+      "step": 65875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.543927963659159e-05,
+      "loss": 1.0701,
+      "step": 65880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5438645505010606e-05,
+      "loss": 1.4507,
+      "step": 65885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5438011342372025e-05,
+      "loss": 1.1029,
+      "step": 65890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5437377148679476e-05,
+      "loss": 1.455,
+      "step": 65895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5436742923936575e-05,
+      "loss": 1.0832,
+      "step": 65900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5436108668146953e-05,
+      "loss": 1.1154,
+      "step": 65905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5435474381314223e-05,
+      "loss": 1.4387,
+      "step": 65910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5434840063442015e-05,
+      "loss": 1.1377,
+      "step": 65915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5434205714533944e-05,
+      "loss": 1.1973,
+      "step": 65920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5433571334593635e-05,
+      "loss": 1.3416,
+      "step": 65925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5432936923624713e-05,
+      "loss": 1.4133,
+      "step": 65930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5432302481630798e-05,
+      "loss": 1.1893,
+      "step": 65935
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5431668008615513e-05,
+      "loss": 0.9746,
+      "step": 65940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5431033504582486e-05,
+      "loss": 1.2788,
+      "step": 65945
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5430398969535333e-05,
+      "loss": 1.1378,
+      "step": 65950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.542976440347768e-05,
+      "loss": 1.6963,
+      "step": 65955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5429129806413157e-05,
+      "loss": 1.1075,
+      "step": 65960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.542849517834538e-05,
+      "loss": 1.2985,
+      "step": 65965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5427860519277972e-05,
+      "loss": 1.149,
+      "step": 65970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5427225829214564e-05,
+      "loss": 1.0791,
+      "step": 65975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.542659110815878e-05,
+      "loss": 1.2907,
+      "step": 65980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.542595635611424e-05,
+      "loss": 1.1406,
+      "step": 65985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.542532157308457e-05,
+      "loss": 1.2626,
+      "step": 65990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.54246867590734e-05,
+      "loss": 1.6925,
+      "step": 65995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5424051914084346e-05,
+      "loss": 1.1479,
+      "step": 66000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.542341703812104e-05,
+      "loss": 1.2762,
+      "step": 66005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.542278213118711e-05,
+      "loss": 1.1102,
+      "step": 66010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5422147193286167e-05,
+      "loss": 1.3562,
+      "step": 66015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5421512224421855e-05,
+      "loss": 0.9295,
+      "step": 66020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5420877224597793e-05,
+      "loss": 0.9984,
+      "step": 66025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.54202421938176e-05,
+      "loss": 1.5244,
+      "step": 66030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5419607132084915e-05,
+      "loss": 1.3693,
+      "step": 66035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5418972039403355e-05,
+      "loss": 1.0801,
+      "step": 66040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5418336915776552e-05,
+      "loss": 1.2495,
+      "step": 66045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.541770176120813e-05,
+      "loss": 1.3161,
+      "step": 66050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5417066575701717e-05,
+      "loss": 1.173,
+      "step": 66055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5416431359260942e-05,
+      "loss": 1.1134,
+      "step": 66060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.541579611188943e-05,
+      "loss": 1.095,
+      "step": 66065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5415160833590804e-05,
+      "loss": 1.3137,
+      "step": 66070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5414525524368704e-05,
+      "loss": 1.3036,
+      "step": 66075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5413890184226746e-05,
+      "loss": 1.2725,
+      "step": 66080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5413254813168568e-05,
+      "loss": 1.5915,
+      "step": 66085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.541261941119779e-05,
+      "loss": 1.2041,
+      "step": 66090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5411983978318047e-05,
+      "loss": 1.0264,
+      "step": 66095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5411348514532963e-05,
+      "loss": 1.3385,
+      "step": 66100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5410713019846165e-05,
+      "loss": 0.9448,
+      "step": 66105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.541007749426129e-05,
+      "loss": 1.5106,
+      "step": 66110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5409441937781964e-05,
+      "loss": 1.1346,
+      "step": 66115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.540880635041181e-05,
+      "loss": 1.0584,
+      "step": 66120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5408170732154463e-05,
+      "loss": 1.0952,
+      "step": 66125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.540753508301356e-05,
+      "loss": 1.2628,
+      "step": 66130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5406899402992713e-05,
+      "loss": 1.4197,
+      "step": 66135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.540626369209557e-05,
+      "loss": 1.0854,
+      "step": 66140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.540562795032575e-05,
+      "loss": 1.4251,
+      "step": 66145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5404992177686886e-05,
+      "loss": 1.2753,
+      "step": 66150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5404356374182612e-05,
+      "loss": 1.1255,
+      "step": 66155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5403720539816555e-05,
+      "loss": 1.4904,
+      "step": 66160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5403084674592345e-05,
+      "loss": 1.4551,
+      "step": 66165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5402448778513618e-05,
+      "loss": 1.2811,
+      "step": 66170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5401812851584006e-05,
+      "loss": 1.2129,
+      "step": 66175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5401176893807135e-05,
+      "loss": 1.0864,
+      "step": 66180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5400540905186637e-05,
+      "loss": 1.3146,
+      "step": 66185
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5399904885726146e-05,
+      "loss": 1.2463,
+      "step": 66190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5399268835429296e-05,
+      "loss": 1.2173,
+      "step": 66195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5398632754299715e-05,
+      "loss": 1.2931,
+      "step": 66200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.539799664234104e-05,
+      "loss": 2.0617,
+      "step": 66205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5397360499556895e-05,
+      "loss": 1.4784,
+      "step": 66210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5396724325950923e-05,
+      "loss": 1.0987,
+      "step": 66215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5396088121526755e-05,
+      "loss": 1.4263,
+      "step": 66220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5395451886288016e-05,
+      "loss": 1.0056,
+      "step": 66225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5394815620238346e-05,
+      "loss": 0.9383,
+      "step": 66230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.539417932338138e-05,
+      "loss": 1.21,
+      "step": 66235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5393542995720744e-05,
+      "loss": 1.3974,
+      "step": 66240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.539290663726008e-05,
+      "loss": 1.1998,
+      "step": 66245
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5392270248003017e-05,
+      "loss": 1.392,
+      "step": 66250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5391633827953192e-05,
+      "loss": 1.3386,
+      "step": 66255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5390997377114237e-05,
+      "loss": 1.1611,
+      "step": 66260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.539036089548979e-05,
+      "loss": 1.2714,
+      "step": 66265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.538972438308348e-05,
+      "loss": 1.0739,
+      "step": 66270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5389087839898947e-05,
+      "loss": 1.1354,
+      "step": 66275
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5388451265939824e-05,
+      "loss": 1.4236,
+      "step": 66280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5387814661209745e-05,
+      "loss": 1.0444,
+      "step": 66285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5387178025712343e-05,
+      "loss": 1.161,
+      "step": 66290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.538654135945126e-05,
+      "loss": 1.3877,
+      "step": 66295
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.538590466243013e-05,
+      "loss": 1.2855,
+      "step": 66300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5385267934652586e-05,
+      "loss": 1.4075,
+      "step": 66305
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.538463117612227e-05,
+      "loss": 1.3474,
+      "step": 66310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5383994386842807e-05,
+      "loss": 1.0972,
+      "step": 66315
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5383357566817845e-05,
+      "loss": 1.281,
+      "step": 66320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5382720716051014e-05,
+      "loss": 1.0625,
+      "step": 66325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5382083834545953e-05,
+      "loss": 1.3561,
+      "step": 66330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5381446922306296e-05,
+      "loss": 1.1306,
+      "step": 66335
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5380809979335687e-05,
+      "loss": 1.0491,
+      "step": 66340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5380173005637758e-05,
+      "loss": 1.4117,
+      "step": 66345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5379536001216145e-05,
+      "loss": 1.2165,
+      "step": 66350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.537889896607449e-05,
+      "loss": 1.129,
+      "step": 66355
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5378261900216426e-05,
+      "loss": 1.3307,
+      "step": 66360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5377624803645598e-05,
+      "loss": 1.3867,
+      "step": 66365
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5376987676365637e-05,
+      "loss": 1.0793,
+      "step": 66370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5376350518380183e-05,
+      "loss": 1.0351,
+      "step": 66375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5375713329692882e-05,
+      "loss": 1.1587,
+      "step": 66380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.537507611030736e-05,
+      "loss": 1.3061,
+      "step": 66385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5374438860227268e-05,
+      "loss": 1.4411,
+      "step": 66390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5373801579456237e-05,
+      "loss": 1.2534,
+      "step": 66395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5373164267997903e-05,
+      "loss": 1.1354,
+      "step": 66400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5372526925855916e-05,
+      "loss": 1.5938,
+      "step": 66405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5371889553033914e-05,
+      "loss": 1.233,
+      "step": 66410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.537125214953553e-05,
+      "loss": 1.4074,
+      "step": 66415
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5370614715364406e-05,
+      "loss": 1.2511,
+      "step": 66420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5369977250524188e-05,
+      "loss": 1.1746,
+      "step": 66425
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.536933975501851e-05,
+      "loss": 1.6388,
+      "step": 66430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5368702228851012e-05,
+      "loss": 1.1071,
+      "step": 66435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5368064672025336e-05,
+      "loss": 1.2631,
+      "step": 66440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5367427084545125e-05,
+      "loss": 1.3975,
+      "step": 66445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5366789466414017e-05,
+      "loss": 1.2721,
+      "step": 66450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5366151817635654e-05,
+      "loss": 1.483,
+      "step": 66455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5365514138213685e-05,
+      "loss": 1.4991,
+      "step": 66460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5364876428151736e-05,
+      "loss": 1.3945,
+      "step": 66465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5364238687453463e-05,
+      "loss": 1.0868,
+      "step": 66470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5363600916122497e-05,
+      "loss": 0.888,
+      "step": 66475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5362963114162488e-05,
+      "loss": 1.2937,
+      "step": 66480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5362325281577074e-05,
+      "loss": 1.1675,
+      "step": 66485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5361687418369898e-05,
+      "loss": 1.2277,
+      "step": 66490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5361049524544605e-05,
+      "loss": 1.1152,
+      "step": 66495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5360411600104835e-05,
+      "loss": 1.2226,
+      "step": 66500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5359773645054232e-05,
+      "loss": 1.0052,
+      "step": 66505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.535913565939644e-05,
+      "loss": 1.2304,
+      "step": 66510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.53584976431351e-05,
+      "loss": 1.4462,
+      "step": 66515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5357859596273853e-05,
+      "loss": 1.0555,
+      "step": 66520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5357221518816347e-05,
+      "loss": 1.2038,
+      "step": 66525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5356583410766227e-05,
+      "loss": 1.1967,
+      "step": 66530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5355945272127132e-05,
+      "loss": 1.3182,
+      "step": 66535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5355307102902713e-05,
+      "loss": 1.2371,
+      "step": 66540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5354668903096608e-05,
+      "loss": 1.1494,
+      "step": 66545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5354030672712463e-05,
+      "loss": 1.3069,
+      "step": 66550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5353392411753923e-05,
+      "loss": 1.3612,
+      "step": 66555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.535275412022463e-05,
+      "loss": 1.4111,
+      "step": 66560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.535211579812824e-05,
+      "loss": 1.2871,
+      "step": 66565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5351477445468384e-05,
+      "loss": 0.9052,
+      "step": 66570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5350839062248718e-05,
+      "loss": 1.2216,
+      "step": 66575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5350200648472877e-05,
+      "loss": 1.4199,
+      "step": 66580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5349562204144514e-05,
+      "loss": 1.1111,
+      "step": 66585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.534892372926728e-05,
+      "loss": 0.9968,
+      "step": 66590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5348285223844806e-05,
+      "loss": 1.1287,
+      "step": 66595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5347646687880753e-05,
+      "loss": 1.3682,
+      "step": 66600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5347008121378757e-05,
+      "loss": 1.3097,
+      "step": 66605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.534636952434247e-05,
+      "loss": 1.1749,
+      "step": 66610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.534573089677554e-05,
+      "loss": 1.1165,
+      "step": 66615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.534509223868161e-05,
+      "loss": 1.1757,
+      "step": 66620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5344453550064328e-05,
+      "loss": 1.1475,
+      "step": 66625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.534381483092734e-05,
+      "loss": 1.6118,
+      "step": 66630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5343176081274295e-05,
+      "loss": 1.0628,
+      "step": 66635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5342537301108844e-05,
+      "loss": 0.9087,
+      "step": 66640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.534189849043463e-05,
+      "loss": 1.3653,
+      "step": 66645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5341259649255304e-05,
+      "loss": 1.3161,
+      "step": 66650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.534062077757451e-05,
+      "loss": 1.4971,
+      "step": 66655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5339981875395897e-05,
+      "loss": 1.1384,
+      "step": 66660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.533934294272312e-05,
+      "loss": 1.2842,
+      "step": 66665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.533870397955982e-05,
+      "loss": 1.1463,
+      "step": 66670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.533806498590965e-05,
+      "loss": 1.1585,
+      "step": 66675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.533742596177626e-05,
+      "loss": 1.4642,
+      "step": 66680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5336786907163295e-05,
+      "loss": 1.2462,
+      "step": 66685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5336147822074402e-05,
+      "loss": 0.96,
+      "step": 66690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5335508706513245e-05,
+      "loss": 1.2725,
+      "step": 66695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5334869560483453e-05,
+      "loss": 1.5154,
+      "step": 66700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.533423038398869e-05,
+      "loss": 1.0306,
+      "step": 66705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5333591177032604e-05,
+      "loss": 1.0851,
+      "step": 66710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5332951939618845e-05,
+      "loss": 1.1634,
+      "step": 66715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.533231267175106e-05,
+      "loss": 1.4524,
+      "step": 66720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5331673373432903e-05,
+      "loss": 1.2117,
+      "step": 66725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5331034044668024e-05,
+      "loss": 1.4259,
+      "step": 66730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.533039468546007e-05,
+      "loss": 0.9967,
+      "step": 66735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5329755295812695e-05,
+      "loss": 1.211,
+      "step": 66740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5329115875729556e-05,
+      "loss": 1.4534,
+      "step": 66745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5328476425214295e-05,
+      "loss": 1.3615,
+      "step": 66750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.532783694427057e-05,
+      "loss": 1.8337,
+      "step": 66755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.532719743290203e-05,
+      "loss": 0.9637,
+      "step": 66760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5326557891112327e-05,
+      "loss": 1.1358,
+      "step": 66765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5325918318905107e-05,
+      "loss": 1.2057,
+      "step": 66770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5325278716284037e-05,
+      "loss": 0.9564,
+      "step": 66775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.532463908325276e-05,
+      "loss": 1.3039,
+      "step": 66780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5323999419814928e-05,
+      "loss": 1.4601,
+      "step": 66785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.53233597259742e-05,
+      "loss": 1.2669,
+      "step": 66790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5322720001734218e-05,
+      "loss": 1.5885,
+      "step": 66795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5322080247098646e-05,
+      "loss": 1.2159,
+      "step": 66800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5321440462071135e-05,
+      "loss": 1.2337,
+      "step": 66805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.532080064665533e-05,
+      "loss": 1.0998,
+      "step": 66810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.53201608008549e-05,
+      "loss": 1.3776,
+      "step": 66815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5319520924673484e-05,
+      "loss": 1.4999,
+      "step": 66820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.531888101811475e-05,
+      "loss": 0.9655,
+      "step": 66825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5318241081182335e-05,
+      "loss": 1.1764,
+      "step": 66830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5317601113879907e-05,
+      "loss": 1.4473,
+      "step": 66835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5316961116211117e-05,
+      "loss": 1.1981,
+      "step": 66840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.531632108817962e-05,
+      "loss": 1.2354,
+      "step": 66845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5315681029789072e-05,
+      "loss": 1.1365,
+      "step": 66850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5315040941043126e-05,
+      "loss": 1.1366,
+      "step": 66855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.531440082194543e-05,
+      "loss": 1.318,
+      "step": 66860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5313760672499657e-05,
+      "loss": 1.0864,
+      "step": 66865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5313120492709445e-05,
+      "loss": 0.9824,
+      "step": 66870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.531248028257846e-05,
+      "loss": 1.2176,
+      "step": 66875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.531184004211036e-05,
+      "loss": 1.2073,
+      "step": 66880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.531119977130879e-05,
+      "loss": 1.1075,
+      "step": 66885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.531055947017741e-05,
+      "loss": 1.0373,
+      "step": 66890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5309919138719886e-05,
+      "loss": 0.9865,
+      "step": 66895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5309278776939865e-05,
+      "loss": 1.1195,
+      "step": 66900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.530863838484101e-05,
+      "loss": 1.0834,
+      "step": 66905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5307997962426968e-05,
+      "loss": 1.5256,
+      "step": 66910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5307357509701406e-05,
+      "loss": 1.237,
+      "step": 66915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.530671702666798e-05,
+      "loss": 1.2389,
+      "step": 66920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5306076513330342e-05,
+      "loss": 1.2103,
+      "step": 66925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5305435969692154e-05,
+      "loss": 0.8544,
+      "step": 66930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5304795395757075e-05,
+      "loss": 1.2074,
+      "step": 66935
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.530415479152876e-05,
+      "loss": 1.0621,
+      "step": 66940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.530351415701087e-05,
+      "loss": 0.9858,
+      "step": 66945
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.530287349220706e-05,
+      "loss": 1.0161,
+      "step": 66950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.530223279712099e-05,
+      "loss": 1.3412,
+      "step": 66955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.530159207175632e-05,
+      "loss": 1.3149,
+      "step": 66960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5300951316116708e-05,
+      "loss": 1.2099,
+      "step": 66965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.530031053020581e-05,
+      "loss": 1.3466,
+      "step": 66970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.529966971402729e-05,
+      "loss": 1.442,
+      "step": 66975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5299028867584803e-05,
+      "loss": 1.3684,
+      "step": 66980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.529838799088202e-05,
+      "loss": 1.2976,
+      "step": 66985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5297747083922584e-05,
+      "loss": 1.5804,
+      "step": 66990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.529710614671016e-05,
+      "loss": 0.9423,
+      "step": 66995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.529646517924842e-05,
+      "loss": 1.1805,
+      "step": 67000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5295824181541006e-05,
+      "loss": 0.879,
+      "step": 67005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5295183153591592e-05,
+      "loss": 1.4256,
+      "step": 67010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5294542095403834e-05,
+      "loss": 1.1378,
+      "step": 67015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.529390100698139e-05,
+      "loss": 1.6002,
+      "step": 67020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.529325988832793e-05,
+      "loss": 1.1319,
+      "step": 67025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5292618739447104e-05,
+      "loss": 1.2987,
+      "step": 67030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.529197756034258e-05,
+      "loss": 1.1324,
+      "step": 67035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5291336351018018e-05,
+      "loss": 1.1495,
+      "step": 67040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.529069511147708e-05,
+      "loss": 1.1713,
+      "step": 67045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5290053841723426e-05,
+      "loss": 1.0125,
+      "step": 67050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.528941254176072e-05,
+      "loss": 1.4541,
+      "step": 67055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5288771211592624e-05,
+      "loss": 1.2952,
+      "step": 67060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5288129851222802e-05,
+      "loss": 1.4199,
+      "step": 67065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.528748846065491e-05,
+      "loss": 1.3596,
+      "step": 67070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5286847039892615e-05,
+      "loss": 1.4341,
+      "step": 67075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5286205588939585e-05,
+      "loss": 1.1861,
+      "step": 67080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.528556410779947e-05,
+      "loss": 1.178,
+      "step": 67085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5284922596475946e-05,
+      "loss": 1.7983,
+      "step": 67090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5284281054972675e-05,
+      "loss": 1.0812,
+      "step": 67095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.528363948329331e-05,
+      "loss": 1.2194,
+      "step": 67100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5282997881441524e-05,
+      "loss": 1.475,
+      "step": 67105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.528235624942098e-05,
+      "loss": 1.1333,
+      "step": 67110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.528171458723534e-05,
+      "loss": 1.389,
+      "step": 67115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5281072894888268e-05,
+      "loss": 1.341,
+      "step": 67120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.528043117238343e-05,
+      "loss": 0.9785,
+      "step": 67125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5279789419724492e-05,
+      "loss": 1.0338,
+      "step": 67130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.527914763691511e-05,
+      "loss": 1.1953,
+      "step": 67135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5278505823958963e-05,
+      "loss": 1.0259,
+      "step": 67140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5277863980859705e-05,
+      "loss": 1.2136,
+      "step": 67145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5277222107621005e-05,
+      "loss": 1.183,
+      "step": 67150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5276580204246533e-05,
+      "loss": 1.0052,
+      "step": 67155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5275938270739944e-05,
+      "loss": 1.44,
+      "step": 67160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5275296307104916e-05,
+      "loss": 1.2693,
+      "step": 67165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5274654313345103e-05,
+      "loss": 1.0622,
+      "step": 67170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5274012289464178e-05,
+      "loss": 1.0272,
+      "step": 67175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5273370235465804e-05,
+      "loss": 1.1635,
+      "step": 67180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5272728151353655e-05,
+      "loss": 1.3776,
+      "step": 67185
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.527208603713139e-05,
+      "loss": 1.4529,
+      "step": 67190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.527144389280268e-05,
+      "loss": 1.3762,
+      "step": 67195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.527080171837119e-05,
+      "loss": 1.1752,
+      "step": 67200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5270159513840584e-05,
+      "loss": 1.5051,
+      "step": 67205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5269517279214535e-05,
+      "loss": 1.1657,
+      "step": 67210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5268875014496705e-05,
+      "loss": 1.4397,
+      "step": 67215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5268232719690764e-05,
+      "loss": 1.2304,
+      "step": 67220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.526759039480039e-05,
+      "loss": 1.417,
+      "step": 67225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.526694803982923e-05,
+      "loss": 1.3917,
+      "step": 67230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.526630565478097e-05,
+      "loss": 1.4556,
+      "step": 67235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5265663239659276e-05,
+      "loss": 1.1319,
+      "step": 67240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5265020794467806e-05,
+      "loss": 1.5539,
+      "step": 67245
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5264378319210237e-05,
+      "loss": 1.2944,
+      "step": 67250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5263735813890235e-05,
+      "loss": 1.7369,
+      "step": 67255
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5263093278511474e-05,
+      "loss": 1.4271,
+      "step": 67260
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.526245071307762e-05,
+      "loss": 1.0804,
+      "step": 67265
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.526180811759234e-05,
+      "loss": 1.3796,
+      "step": 67270
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.52611654920593e-05,
+      "loss": 1.5543,
+      "step": 67275
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5260522836482184e-05,
+      "loss": 1.1012,
+      "step": 67280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.525988015086465e-05,
+      "loss": 1.0295,
+      "step": 67285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5259237435210372e-05,
+      "loss": 0.8764,
+      "step": 67290
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.525859468952302e-05,
+      "loss": 1.3474,
+      "step": 67295
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5257951913806262e-05,
+      "loss": 1.1551,
+      "step": 67300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.525730910806377e-05,
+      "loss": 1.3102,
+      "step": 67305
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5256666272299217e-05,
+      "loss": 1.157,
+      "step": 67310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5256023406516274e-05,
+      "loss": 1.3454,
+      "step": 67315
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5255380510718607e-05,
+      "loss": 1.4134,
+      "step": 67320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5254737584909892e-05,
+      "loss": 1.3108,
+      "step": 67325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5254094629093801e-05,
+      "loss": 1.5573,
+      "step": 67330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5253451643274e-05,
+      "loss": 1.3304,
+      "step": 67335
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.525280862745417e-05,
+      "loss": 1.4437,
+      "step": 67340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.525216558163797e-05,
+      "loss": 0.9377,
+      "step": 67345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5251522505829085e-05,
+      "loss": 1.011,
+      "step": 67350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.525087940003118e-05,
+      "loss": 1.1908,
+      "step": 67355
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.525023626424793e-05,
+      "loss": 1.2277,
+      "step": 67360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.524959309848301e-05,
+      "loss": 1.0954,
+      "step": 67365
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5248949902740085e-05,
+      "loss": 1.0891,
+      "step": 67370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5248306677022836e-05,
+      "loss": 1.4438,
+      "step": 67375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.524766342133493e-05,
+      "loss": 1.3793,
+      "step": 67380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5247020135680046e-05,
+      "loss": 1.3919,
+      "step": 67385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5246376820061854e-05,
+      "loss": 1.1571,
+      "step": 67390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5245733474484027e-05,
+      "loss": 1.0346,
+      "step": 67395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5245090098950247e-05,
+      "loss": 1.1294,
+      "step": 67400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5244446693464175e-05,
+      "loss": 1.0946,
+      "step": 67405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5243803258029493e-05,
+      "loss": 1.0601,
+      "step": 67410
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5243159792649872e-05,
+      "loss": 1.0749,
+      "step": 67415
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5242516297328994e-05,
+      "loss": 1.1332,
+      "step": 67420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5241872772070524e-05,
+      "loss": 1.2636,
+      "step": 67425
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5241229216878141e-05,
+      "loss": 1.4025,
+      "step": 67430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5240585631755522e-05,
+      "loss": 1.1753,
+      "step": 67435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.523994201670634e-05,
+      "loss": 1.0709,
+      "step": 67440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5239298371734273e-05,
+      "loss": 1.4493,
+      "step": 67445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5238654696842989e-05,
+      "loss": 1.2834,
+      "step": 67450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5238010992036171e-05,
+      "loss": 1.1204,
+      "step": 67455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5237367257317495e-05,
+      "loss": 1.1115,
+      "step": 67460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5236723492690634e-05,
+      "loss": 1.1353,
+      "step": 67465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5236079698159265e-05,
+      "loss": 1.1574,
+      "step": 67470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5235435873727068e-05,
+      "loss": 1.229,
+      "step": 67475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5234792019397713e-05,
+      "loss": 1.1978,
+      "step": 67480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.523414813517488e-05,
+      "loss": 1.1303,
+      "step": 67485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5233504221062245e-05,
+      "loss": 1.3615,
+      "step": 67490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5232860277063486e-05,
+      "loss": 1.3734,
+      "step": 67495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5232216303182284e-05,
+      "loss": 1.2657,
+      "step": 67500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.523157229942231e-05,
+      "loss": 1.2611,
+      "step": 67505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5230928265787241e-05,
+      "loss": 1.4995,
+      "step": 67510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5230284202280765e-05,
+      "loss": 1.2239,
+      "step": 67515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5229640108906546e-05,
+      "loss": 1.2634,
+      "step": 67520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5228995985668275e-05,
+      "loss": 1.276,
+      "step": 67525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5228351832569623e-05,
+      "loss": 1.2977,
+      "step": 67530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.522770764961427e-05,
+      "loss": 1.1136,
+      "step": 67535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5227063436805896e-05,
+      "loss": 1.3544,
+      "step": 67540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5226419194148174e-05,
+      "loss": 1.1213,
+      "step": 67545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5225774921644791e-05,
+      "loss": 0.87,
+      "step": 67550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5225130619299421e-05,
+      "loss": 1.3756,
+      "step": 67555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5224486287115746e-05,
+      "loss": 0.8472,
+      "step": 67560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5223841925097442e-05,
+      "loss": 1.3194,
+      "step": 67565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5223197533248193e-05,
+      "loss": 1.2005,
+      "step": 67570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5222553111571677e-05,
+      "loss": 1.2659,
+      "step": 67575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5221908660071574e-05,
+      "loss": 1.4904,
+      "step": 67580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5221264178751561e-05,
+      "loss": 1.0484,
+      "step": 67585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5220619667615321e-05,
+      "loss": 1.2987,
+      "step": 67590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.521997512666654e-05,
+      "loss": 1.0761,
+      "step": 67595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5219330555908891e-05,
+      "loss": 1.1417,
+      "step": 67600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5218685955346055e-05,
+      "loss": 1.1449,
+      "step": 67605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5218041324981718e-05,
+      "loss": 1.2086,
+      "step": 67610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5217396664819557e-05,
+      "loss": 1.3384,
+      "step": 67615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5216751974863256e-05,
+      "loss": 1.2952,
+      "step": 67620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5216107255116492e-05,
+      "loss": 1.3836,
+      "step": 67625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5215462505582954e-05,
+      "loss": 1.0693,
+      "step": 67630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.521481772626632e-05,
+      "loss": 1.1816,
+      "step": 67635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.521417291717027e-05,
+      "loss": 1.1853,
+      "step": 67640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5213528078298492e-05,
+      "loss": 1.2698,
+      "step": 67645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5212883209654656e-05,
+      "loss": 1.17,
+      "step": 67650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.521223831124246e-05,
+      "loss": 1.1495,
+      "step": 67655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.521159338306558e-05,
+      "loss": 1.4649,
+      "step": 67660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5210948425127696e-05,
+      "loss": 0.9457,
+      "step": 67665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5210303437432497e-05,
+      "loss": 1.3801,
+      "step": 67670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.520965841998366e-05,
+      "loss": 1.2847,
+      "step": 67675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5209013372784874e-05,
+      "loss": 1.2814,
+      "step": 67680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5208368295839817e-05,
+      "loss": 1.3398,
+      "step": 67685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5207723189152176e-05,
+      "loss": 1.0837,
+      "step": 67690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5207078052725636e-05,
+      "loss": 1.0668,
+      "step": 67695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.520643288656388e-05,
+      "loss": 0.9878,
+      "step": 67700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5205787690670591e-05,
+      "loss": 1.1254,
+      "step": 67705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5205142465049455e-05,
+      "loss": 1.1903,
+      "step": 67710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5204497209704156e-05,
+      "loss": 1.1499,
+      "step": 67715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5203851924638375e-05,
+      "loss": 1.336,
+      "step": 67720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5203206609855806e-05,
+      "loss": 1.2779,
+      "step": 67725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5202561265360127e-05,
+      "loss": 1.8069,
+      "step": 67730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5201915891155024e-05,
+      "loss": 1.3005,
+      "step": 67735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5201270487244184e-05,
+      "loss": 1.1138,
+      "step": 67740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.520062505363129e-05,
+      "loss": 1.3384,
+      "step": 67745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5199979590320034e-05,
+      "loss": 1.2419,
+      "step": 67750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5199334097314094e-05,
+      "loss": 1.0067,
+      "step": 67755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.519868857461716e-05,
+      "loss": 1.0415,
+      "step": 67760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.519804302223292e-05,
+      "loss": 0.9851,
+      "step": 67765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5197397440165058e-05,
+      "loss": 1.4232,
+      "step": 67770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5196751828417263e-05,
+      "loss": 1.541,
+      "step": 67775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5196106186993216e-05,
+      "loss": 1.7217,
+      "step": 67780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.519546051589661e-05,
+      "loss": 1.5021,
+      "step": 67785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5194814815131129e-05,
+      "loss": 1.2979,
+      "step": 67790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5194169084700463e-05,
+      "loss": 1.2878,
+      "step": 67795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5193523324608296e-05,
+      "loss": 1.1633,
+      "step": 67800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.519287753485832e-05,
+      "loss": 0.9776,
+      "step": 67805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5192231715454217e-05,
+      "loss": 1.2639,
+      "step": 67810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5191585866399682e-05,
+      "loss": 1.5464,
+      "step": 67815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5190939987698396e-05,
+      "loss": 1.1713,
+      "step": 67820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5190294079354052e-05,
+      "loss": 1.0982,
+      "step": 67825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5189648141370338e-05,
+      "loss": 1.4605,
+      "step": 67830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5189002173750944e-05,
+      "loss": 0.9938,
+      "step": 67835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5188356176499555e-05,
+      "loss": 1.0465,
+      "step": 67840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.518771014961986e-05,
+      "loss": 1.5901,
+      "step": 67845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5187064093115549e-05,
+      "loss": 1.1327,
+      "step": 67850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5186418006990314e-05,
+      "loss": 1.2697,
+      "step": 67855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.518577189124784e-05,
+      "loss": 1.5469,
+      "step": 67860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5185125745891823e-05,
+      "loss": 1.3841,
+      "step": 67865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.518447957092595e-05,
+      "loss": 1.4111,
+      "step": 67870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5183833366353907e-05,
+      "loss": 1.586,
+      "step": 67875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5183187132179388e-05,
+      "loss": 1.3622,
+      "step": 67880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5182540868406082e-05,
+      "loss": 1.1441,
+      "step": 67885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5181894575037681e-05,
+      "loss": 1.3182,
+      "step": 67890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5181248252077875e-05,
+      "loss": 1.0867,
+      "step": 67895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5180601899530355e-05,
+      "loss": 1.1426,
+      "step": 67900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5179955517398813e-05,
+      "loss": 1.1406,
+      "step": 67905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5179309105686937e-05,
+      "loss": 1.261,
+      "step": 67910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.517866266439842e-05,
+      "loss": 1.236,
+      "step": 67915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5178016193536955e-05,
+      "loss": 1.1665,
+      "step": 67920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5177369693106232e-05,
+      "loss": 0.9975,
+      "step": 67925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5176723163109946e-05,
+      "loss": 1.1508,
+      "step": 67930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5176076603551781e-05,
+      "loss": 1.2382,
+      "step": 67935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.517543001443544e-05,
+      "loss": 0.9965,
+      "step": 67940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5174783395764606e-05,
+      "loss": 1.2027,
+      "step": 67945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5174136747542977e-05,
+      "loss": 1.3452,
+      "step": 67950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5173490069774243e-05,
+      "loss": 1.4286,
+      "step": 67955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5172843362462102e-05,
+      "loss": 1.3292,
+      "step": 67960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5172196625610238e-05,
+      "loss": 1.2545,
+      "step": 67965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5171549859222352e-05,
+      "loss": 1.1018,
+      "step": 67970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5170903063302139e-05,
+      "loss": 1.3106,
+      "step": 67975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.517025623785328e-05,
+      "loss": 1.1412,
+      "step": 67980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5169609382879483e-05,
+      "loss": 1.3154,
+      "step": 67985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5168962498384434e-05,
+      "loss": 1.3146,
+      "step": 67990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.516831558437183e-05,
+      "loss": 1.4255,
+      "step": 67995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5167668640845361e-05,
+      "loss": 1.4407,
+      "step": 68000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5167021667808727e-05,
+      "loss": 0.9971,
+      "step": 68005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5166374665265621e-05,
+      "loss": 1.1157,
+      "step": 68010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5165727633219737e-05,
+      "loss": 1.1839,
+      "step": 68015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5165080571674769e-05,
+      "loss": 1.1651,
+      "step": 68020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5164433480634413e-05,
+      "loss": 1.0084,
+      "step": 68025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5163786360102363e-05,
+      "loss": 1.0133,
+      "step": 68030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5163139210082318e-05,
+      "loss": 1.1853,
+      "step": 68035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5162492030577968e-05,
+      "loss": 1.293,
+      "step": 68040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5161844821593015e-05,
+      "loss": 1.7034,
+      "step": 68045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5161197583131149e-05,
+      "loss": 1.2644,
+      "step": 68050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5160550315196072e-05,
+      "loss": 1.7268,
+      "step": 68055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5159903017791472e-05,
+      "loss": 1.4025,
+      "step": 68060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5159255690921056e-05,
+      "loss": 1.1352,
+      "step": 68065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.515860833458851e-05,
+      "loss": 1.0542,
+      "step": 68070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5157960948797539e-05,
+      "loss": 1.0718,
+      "step": 68075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5157313533551839e-05,
+      "loss": 1.139,
+      "step": 68080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5156666088855102e-05,
+      "loss": 0.9338,
+      "step": 68085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5156018614711025e-05,
+      "loss": 1.1354,
+      "step": 68090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.515537111112331e-05,
+      "loss": 1.695,
+      "step": 68095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5154723578095655e-05,
+      "loss": 1.3538,
+      "step": 68100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5154076015631758e-05,
+      "loss": 1.1391,
+      "step": 68105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5153428423735314e-05,
+      "loss": 1.1796,
+      "step": 68110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5152780802410019e-05,
+      "loss": 1.2088,
+      "step": 68115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5152133151659576e-05,
+      "loss": 1.1016,
+      "step": 68120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5151485471487681e-05,
+      "loss": 1.1511,
+      "step": 68125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5150837761898034e-05,
+      "loss": 1.1933,
+      "step": 68130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5150190022894334e-05,
+      "loss": 1.1769,
+      "step": 68135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5149542254480276e-05,
+      "loss": 1.3746,
+      "step": 68140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5148894456659564e-05,
+      "loss": 1.1973,
+      "step": 68145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5148246629435895e-05,
+      "loss": 1.2549,
+      "step": 68150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5147598772812968e-05,
+      "loss": 1.3579,
+      "step": 68155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5146950886794485e-05,
+      "loss": 1.1458,
+      "step": 68160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5146302971384141e-05,
+      "loss": 1.5225,
+      "step": 68165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5145655026585643e-05,
+      "loss": 1.1725,
+      "step": 68170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5145007052402685e-05,
+      "loss": 1.4319,
+      "step": 68175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5144359048838967e-05,
+      "loss": 1.3085,
+      "step": 68180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5143711015898199e-05,
+      "loss": 1.461,
+      "step": 68185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5143062953584066e-05,
+      "loss": 1.5269,
+      "step": 68190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5142414861900284e-05,
+      "loss": 1.3949,
+      "step": 68195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5141766740850545e-05,
+      "loss": 1.6486,
+      "step": 68200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.514111859043855e-05,
+      "loss": 1.276,
+      "step": 68205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5140470410668007e-05,
+      "loss": 1.0075,
+      "step": 68210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5139822201542612e-05,
+      "loss": 1.0917,
+      "step": 68215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5139173963066065e-05,
+      "loss": 1.2938,
+      "step": 68220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5138525695242074e-05,
+      "loss": 1.212,
+      "step": 68225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5137877398074335e-05,
+      "loss": 1.1239,
+      "step": 68230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5137229071566553e-05,
+      "loss": 1.4484,
+      "step": 68235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5136580715722432e-05,
+      "loss": 1.0431,
+      "step": 68240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5135932330545671e-05,
+      "loss": 1.1453,
+      "step": 68245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5135283916039973e-05,
+      "loss": 1.1593,
+      "step": 68250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5134635472209045e-05,
+      "loss": 1.1093,
+      "step": 68255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.513398699905658e-05,
+      "loss": 1.146,
+      "step": 68260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5133338496586296e-05,
+      "loss": 1.345,
+      "step": 68265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5132689964801884e-05,
+      "loss": 1.1843,
+      "step": 68270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.513204140370705e-05,
+      "loss": 1.2225,
+      "step": 68275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5131392813305503e-05,
+      "loss": 1.1604,
+      "step": 68280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5130744193600941e-05,
+      "loss": 1.164,
+      "step": 68285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5130095544597075e-05,
+      "loss": 1.5263,
+      "step": 68290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5129446866297594e-05,
+      "loss": 1.3133,
+      "step": 68295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5128798158706222e-05,
+      "loss": 1.2231,
+      "step": 68300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.512814942182665e-05,
+      "loss": 1.3828,
+      "step": 68305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5127500655662587e-05,
+      "loss": 1.1439,
+      "step": 68310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.512685186021774e-05,
+      "loss": 0.9976,
+      "step": 68315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5126203035495806e-05,
+      "loss": 1.4805,
+      "step": 68320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5125554181500498e-05,
+      "loss": 1.2299,
+      "step": 68325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5124905298235519e-05,
+      "loss": 1.3282,
+      "step": 68330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5124256385704574e-05,
+      "loss": 1.1554,
+      "step": 68335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.512360744391137e-05,
+      "loss": 1.0447,
+      "step": 68340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5122958472859613e-05,
+      "loss": 1.5366,
+      "step": 68345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5122309472553005e-05,
+      "loss": 1.2528,
+      "step": 68350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5121660442995257e-05,
+      "loss": 1.2184,
+      "step": 68355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5121011384190072e-05,
+      "loss": 1.1712,
+      "step": 68360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5120362296141156e-05,
+      "loss": 1.4066,
+      "step": 68365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.511971317885222e-05,
+      "loss": 1.0079,
+      "step": 68370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5119064032326969e-05,
+      "loss": 1.2148,
+      "step": 68375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5118414856569104e-05,
+      "loss": 1.3131,
+      "step": 68380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5117765651582343e-05,
+      "loss": 1.314,
+      "step": 68385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5117116417370384e-05,
+      "loss": 1.0178,
+      "step": 68390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.511646715393694e-05,
+      "loss": 1.2382,
+      "step": 68395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5115817861285713e-05,
+      "loss": 1.0446,
+      "step": 68400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5115168539420418e-05,
+      "loss": 1.2554,
+      "step": 68405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5114519188344758e-05,
+      "loss": 1.0854,
+      "step": 68410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5113869808062442e-05,
+      "loss": 1.3664,
+      "step": 68415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.511322039857718e-05,
+      "loss": 1.1527,
+      "step": 68420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5112570959892679e-05,
+      "loss": 1.0848,
+      "step": 68425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.511192149201265e-05,
+      "loss": 1.0423,
+      "step": 68430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5111271994940796e-05,
+      "loss": 1.0504,
+      "step": 68435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5110622468680834e-05,
+      "loss": 0.9125,
+      "step": 68440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5109972913236466e-05,
+      "loss": 1.1634,
+      "step": 68445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5109323328611405e-05,
+      "loss": 1.3008,
+      "step": 68450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.510867371480936e-05,
+      "loss": 1.1765,
+      "step": 68455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.510802407183404e-05,
+      "loss": 1.3551,
+      "step": 68460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5107374399689157e-05,
+      "loss": 1.395,
+      "step": 68465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5106724698378414e-05,
+      "loss": 1.2708,
+      "step": 68470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5106074967905533e-05,
+      "loss": 1.1445,
+      "step": 68475
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5105425208274214e-05,
+      "loss": 1.4943,
+      "step": 68480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.510477541948817e-05,
+      "loss": 1.1026,
+      "step": 68485
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5104125601551117e-05,
+      "loss": 1.1123,
+      "step": 68490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5103475754466756e-05,
+      "loss": 1.6118,
+      "step": 68495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5102825878238809e-05,
+      "loss": 1.0446,
+      "step": 68500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5102175972870977e-05,
+      "loss": 1.4579,
+      "step": 68505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5101526038366978e-05,
+      "loss": 1.2919,
+      "step": 68510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5100876074730524e-05,
+      "loss": 1.0255,
+      "step": 68515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.510022608196532e-05,
+      "loss": 1.0183,
+      "step": 68520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5099576060075083e-05,
+      "loss": 1.6265,
+      "step": 68525
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5098926009063523e-05,
+      "loss": 1.5737,
+      "step": 68530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5098275928934356e-05,
+      "loss": 1.3051,
+      "step": 68535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5097625819691289e-05,
+      "loss": 1.7743,
+      "step": 68540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5096975681338037e-05,
+      "loss": 1.1019,
+      "step": 68545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5096325513878314e-05,
+      "loss": 0.9618,
+      "step": 68550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.509567531731583e-05,
+      "loss": 1.5808,
+      "step": 68555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5095025091654298e-05,
+      "loss": 1.4055,
+      "step": 68560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5094374836897434e-05,
+      "loss": 1.2834,
+      "step": 68565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5093724553048951e-05,
+      "loss": 1.2862,
+      "step": 68570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.509307424011256e-05,
+      "loss": 1.351,
+      "step": 68575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5092423898091974e-05,
+      "loss": 1.1355,
+      "step": 68580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.509177352699091e-05,
+      "loss": 1.0306,
+      "step": 68585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.509112312681308e-05,
+      "loss": 1.1377,
+      "step": 68590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5090472697562199e-05,
+      "loss": 1.1655,
+      "step": 68595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.508982223924198e-05,
+      "loss": 1.093,
+      "step": 68600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5089171751856144e-05,
+      "loss": 1.1741,
+      "step": 68605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5088521235408392e-05,
+      "loss": 1.3523,
+      "step": 68610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.508787068990245e-05,
+      "loss": 1.1227,
+      "step": 68615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5087220115342033e-05,
+      "loss": 1.1726,
+      "step": 68620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.508656951173085e-05,
+      "loss": 1.2206,
+      "step": 68625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5085918879072619e-05,
+      "loss": 1.2597,
+      "step": 68630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5085268217371057e-05,
+      "loss": 0.9757,
+      "step": 68635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5084617526629876e-05,
+      "loss": 1.4451,
+      "step": 68640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5083966806852796e-05,
+      "loss": 1.2311,
+      "step": 68645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5083316058043531e-05,
+      "loss": 1.4135,
+      "step": 68650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5082665280205797e-05,
+      "loss": 1.2518,
+      "step": 68655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5082014473343311e-05,
+      "loss": 1.0521,
+      "step": 68660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.508136363745979e-05,
+      "loss": 1.0034,
+      "step": 68665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5080712772558945e-05,
+      "loss": 1.2654,
+      "step": 68670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5080061878644503e-05,
+      "loss": 1.2056,
+      "step": 68675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5079410955720174e-05,
+      "loss": 1.2095,
+      "step": 68680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5078760003789677e-05,
+      "loss": 1.3203,
+      "step": 68685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5078109022856727e-05,
+      "loss": 1.1827,
+      "step": 68690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.507745801292504e-05,
+      "loss": 1.3612,
+      "step": 68695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5076806973998342e-05,
+      "loss": 1.414,
+      "step": 68700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.507615590608034e-05,
+      "loss": 1.173,
+      "step": 68705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5075504809174763e-05,
+      "loss": 1.1408,
+      "step": 68710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5074853683285324e-05,
+      "loss": 1.0767,
+      "step": 68715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5074202528415738e-05,
+      "loss": 1.2509,
+      "step": 68720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5073551344569728e-05,
+      "loss": 1.2698,
+      "step": 68725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5072900131751009e-05,
+      "loss": 0.9206,
+      "step": 68730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5072248889963304e-05,
+      "loss": 1.1579,
+      "step": 68735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5071597619210327e-05,
+      "loss": 1.0534,
+      "step": 68740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5070946319495802e-05,
+      "loss": 0.9857,
+      "step": 68745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5070294990823447e-05,
+      "loss": 1.1773,
+      "step": 68750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5069643633196979e-05,
+      "loss": 1.0495,
+      "step": 68755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.506899224662012e-05,
+      "loss": 1.2356,
+      "step": 68760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5068340831096589e-05,
+      "loss": 1.5103,
+      "step": 68765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5067689386630106e-05,
+      "loss": 1.0603,
+      "step": 68770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5067037913224389e-05,
+      "loss": 1.07,
+      "step": 68775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5066386410883162e-05,
+      "loss": 1.1902,
+      "step": 68780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5065734879610143e-05,
+      "loss": 1.4393,
+      "step": 68785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5065083319409053e-05,
+      "loss": 1.3668,
+      "step": 68790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5064431730283615e-05,
+      "loss": 1.2664,
+      "step": 68795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5063780112237542e-05,
+      "loss": 0.9375,
+      "step": 68800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5063128465274567e-05,
+      "loss": 1.2481,
+      "step": 68805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5062476789398401e-05,
+      "loss": 1.1898,
+      "step": 68810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5061825084612773e-05,
+      "loss": 1.2662,
+      "step": 68815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5061173350921403e-05,
+      "loss": 1.136,
+      "step": 68820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5060521588328008e-05,
+      "loss": 1.1666,
+      "step": 68825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5059869796836313e-05,
+      "loss": 1.135,
+      "step": 68830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5059217976450039e-05,
+      "loss": 1.2907,
+      "step": 68835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.505856612717291e-05,
+      "loss": 0.927,
+      "step": 68840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5057914249008647e-05,
+      "loss": 1.2034,
+      "step": 68845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5057262341960977e-05,
+      "loss": 1.4472,
+      "step": 68850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5056610406033615e-05,
+      "loss": 1.4787,
+      "step": 68855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.505595844123029e-05,
+      "loss": 1.1806,
+      "step": 68860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5055306447554721e-05,
+      "loss": 1.0242,
+      "step": 68865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5054654425010636e-05,
+      "loss": 0.9626,
+      "step": 68870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5054002373601753e-05,
+      "loss": 1.1979,
+      "step": 68875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.50533502933318e-05,
+      "loss": 1.3316,
+      "step": 68880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5052698184204498e-05,
+      "loss": 1.3846,
+      "step": 68885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5052046046223572e-05,
+      "loss": 1.263,
+      "step": 68890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5051393879392744e-05,
+      "loss": 1.2085,
+      "step": 68895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5050741683715742e-05,
+      "loss": 1.1769,
+      "step": 68900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5050089459196286e-05,
+      "loss": 1.2442,
+      "step": 68905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5049437205838107e-05,
+      "loss": 1.2292,
+      "step": 68910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5048784923644921e-05,
+      "loss": 1.2253,
+      "step": 68915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.504813261262046e-05,
+      "loss": 1.5115,
+      "step": 68920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5047480272768451e-05,
+      "loss": 1.0684,
+      "step": 68925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5046827904092607e-05,
+      "loss": 1.2537,
+      "step": 68930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5046175506596667e-05,
+      "loss": 1.3496,
+      "step": 68935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5045523080284347e-05,
+      "loss": 1.1523,
+      "step": 68940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5044870625159377e-05,
+      "loss": 1.2835,
+      "step": 68945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5044218141225485e-05,
+      "loss": 1.0313,
+      "step": 68950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5043565628486394e-05,
+      "loss": 1.0309,
+      "step": 68955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5042913086945826e-05,
+      "loss": 0.9153,
+      "step": 68960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.504226051660752e-05,
+      "loss": 1.4839,
+      "step": 68965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5041607917475188e-05,
+      "loss": 1.3763,
+      "step": 68970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5040955289552566e-05,
+      "loss": 1.3243,
+      "step": 68975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5040302632843375e-05,
+      "loss": 1.2595,
+      "step": 68980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5039649947351347e-05,
+      "loss": 1.272,
+      "step": 68985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5038997233080209e-05,
+      "loss": 1.2747,
+      "step": 68990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5038344490033683e-05,
+      "loss": 1.1858,
+      "step": 68995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5037691718215501e-05,
+      "loss": 1.3371,
+      "step": 69000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5037038917629392e-05,
+      "loss": 1.3091,
+      "step": 69005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5036386088279079e-05,
+      "loss": 1.0705,
+      "step": 69010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5035733230168297e-05,
+      "loss": 1.3765,
+      "step": 69015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5035080343300766e-05,
+      "loss": 0.9369,
+      "step": 69020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5034427427680217e-05,
+      "loss": 1.1194,
+      "step": 69025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5033774483310381e-05,
+      "loss": 1.0923,
+      "step": 69030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5033121510194986e-05,
+      "loss": 0.9825,
+      "step": 69035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5032468508337758e-05,
+      "loss": 1.0478,
+      "step": 69040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.503181547774243e-05,
+      "loss": 1.2842,
+      "step": 69045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5031162418412728e-05,
+      "loss": 0.9608,
+      "step": 69050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5030509330352384e-05,
+      "loss": 1.0906,
+      "step": 69055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5029856213565125e-05,
+      "loss": 1.2521,
+      "step": 69060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5029203068054682e-05,
+      "loss": 1.2983,
+      "step": 69065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5028549893824786e-05,
+      "loss": 1.2163,
+      "step": 69070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5027896690879165e-05,
+      "loss": 1.2651,
+      "step": 69075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5027243459221548e-05,
+      "loss": 1.1035,
+      "step": 69080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.502659019885567e-05,
+      "loss": 1.165,
+      "step": 69085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5025936909785254e-05,
+      "loss": 1.4659,
+      "step": 69090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5025283592014038e-05,
+      "loss": 1.2718,
+      "step": 69095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.502463024554575e-05,
+      "loss": 1.3192,
+      "step": 69100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5023976870384117e-05,
+      "loss": 1.2385,
+      "step": 69105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.502332346653288e-05,
+      "loss": 1.2005,
+      "step": 69110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.502267003399576e-05,
+      "loss": 1.1244,
+      "step": 69115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5022016572776494e-05,
+      "loss": 0.9256,
+      "step": 69120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5021363082878813e-05,
+      "loss": 1.3694,
+      "step": 69125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5020709564306444e-05,
+      "loss": 1.1899,
+      "step": 69130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.502005601706313e-05,
+      "loss": 1.0795,
+      "step": 69135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5019402441152588e-05,
+      "loss": 1.2719,
+      "step": 69140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5018748836578562e-05,
+      "loss": 1.1502,
+      "step": 69145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5018095203344783e-05,
+      "loss": 1.0269,
+      "step": 69150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.501744154145498e-05,
+      "loss": 1.3489,
+      "step": 69155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5016787850912888e-05,
+      "loss": 1.1699,
+      "step": 69160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5016134131722236e-05,
+      "loss": 1.5808,
+      "step": 69165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5015480383886762e-05,
+      "loss": 1.0446,
+      "step": 69170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5014826607410198e-05,
+      "loss": 1.5406,
+      "step": 69175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5014172802296276e-05,
+      "loss": 1.1773,
+      "step": 69180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5013518968548733e-05,
+      "loss": 1.0439,
+      "step": 69185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5012865106171295e-05,
+      "loss": 1.0077,
+      "step": 69190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5012211215167704e-05,
+      "loss": 1.2512,
+      "step": 69195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5011557295541692e-05,
+      "loss": 1.5192,
+      "step": 69200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5010903347296992e-05,
+      "loss": 0.9658,
+      "step": 69205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5010249370437335e-05,
+      "loss": 1.2043,
+      "step": 69210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5009595364966467e-05,
+      "loss": 1.2169,
+      "step": 69215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.500894133088811e-05,
+      "loss": 0.9885,
+      "step": 69220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5008287268206006e-05,
+      "loss": 1.3946,
+      "step": 69225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5007633176923885e-05,
+      "loss": 0.9929,
+      "step": 69230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5006979057045485e-05,
+      "loss": 1.1416,
+      "step": 69235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5006324908574544e-05,
+      "loss": 1.3104,
+      "step": 69240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5005670731514794e-05,
+      "loss": 1.1173,
+      "step": 69245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5005016525869971e-05,
+      "loss": 1.5396,
+      "step": 69250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5004362291643812e-05,
+      "loss": 1.3392,
+      "step": 69255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5003708028840054e-05,
+      "loss": 0.9922,
+      "step": 69260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5003053737462434e-05,
+      "loss": 0.9707,
+      "step": 69265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5002399417514682e-05,
+      "loss": 0.9967,
+      "step": 69270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.500174506900054e-05,
+      "loss": 1.1413,
+      "step": 69275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5001090691923742e-05,
+      "loss": 1.1201,
+      "step": 69280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5000436286288032e-05,
+      "loss": 1.4443,
+      "step": 69285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4999781852097135e-05,
+      "loss": 1.2505,
+      "step": 69290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4999127389354797e-05,
+      "loss": 0.9787,
+      "step": 69295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4998472898064753e-05,
+      "loss": 1.1563,
+      "step": 69300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4997818378230741e-05,
+      "loss": 1.0076,
+      "step": 69305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4997163829856498e-05,
+      "loss": 1.0201,
+      "step": 69310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4996509252945758e-05,
+      "loss": 1.2366,
+      "step": 69315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4995854647502269e-05,
+      "loss": 1.4117,
+      "step": 69320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.499520001352976e-05,
+      "loss": 0.958,
+      "step": 69325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4994545351031968e-05,
+      "loss": 1.5896,
+      "step": 69330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4993890660012643e-05,
+      "loss": 1.3103,
+      "step": 69335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4993235940475511e-05,
+      "loss": 1.4312,
+      "step": 69340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4992581192424321e-05,
+      "loss": 1.2964,
+      "step": 69345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4991926415862802e-05,
+      "loss": 1.1959,
+      "step": 69350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.49912716107947e-05,
+      "loss": 1.463,
+      "step": 69355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4990616777223755e-05,
+      "loss": 1.1433,
+      "step": 69360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4989961915153699e-05,
+      "loss": 1.0936,
+      "step": 69365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4989307024588282e-05,
+      "loss": 1.0345,
+      "step": 69370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4988652105531233e-05,
+      "loss": 1.0593,
+      "step": 69375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4987997157986303e-05,
+      "loss": 1.2735,
+      "step": 69380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4987342181957224e-05,
+      "loss": 1.1838,
+      "step": 69385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4986687177447737e-05,
+      "loss": 1.2857,
+      "step": 69390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4986032144461587e-05,
+      "loss": 1.0099,
+      "step": 69395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4985377083002508e-05,
+      "loss": 1.0607,
+      "step": 69400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4984721993074248e-05,
+      "loss": 1.2443,
+      "step": 69405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4984066874680543e-05,
+      "loss": 1.4039,
+      "step": 69410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4983411727825137e-05,
+      "loss": 1.0883,
+      "step": 69415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4982756552511766e-05,
+      "loss": 1.0766,
+      "step": 69420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4982101348744177e-05,
+      "loss": 1.1487,
+      "step": 69425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.498144611652611e-05,
+      "loss": 1.1623,
+      "step": 69430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4980790855861304e-05,
+      "loss": 1.2712,
+      "step": 69435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4980135566753509e-05,
+      "loss": 0.8943,
+      "step": 69440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4979480249206458e-05,
+      "loss": 1.4256,
+      "step": 69445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4978824903223897e-05,
+      "loss": 1.2687,
+      "step": 69450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4978169528809566e-05,
+      "loss": 1.1614,
+      "step": 69455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.497751412596721e-05,
+      "loss": 1.3372,
+      "step": 69460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4976858694700576e-05,
+      "loss": 1.0877,
+      "step": 69465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4976203235013397e-05,
+      "loss": 1.0991,
+      "step": 69470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.497554774690942e-05,
+      "loss": 1.1743,
+      "step": 69475
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.497489223039239e-05,
+      "loss": 1.3199,
+      "step": 69480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4974236685466052e-05,
+      "loss": 1.3,
+      "step": 69485
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4973581112134148e-05,
+      "loss": 1.188,
+      "step": 69490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4972925510400418e-05,
+      "loss": 1.1379,
+      "step": 69495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4972269880268608e-05,
+      "loss": 0.9842,
+      "step": 69500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4971614221742466e-05,
+      "loss": 1.198,
+      "step": 69505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.497095853482573e-05,
+      "loss": 0.9988,
+      "step": 69510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4970302819522149e-05,
+      "loss": 1.1068,
+      "step": 69515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4969647075835465e-05,
+      "loss": 1.4904,
+      "step": 69520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4968991303769422e-05,
+      "loss": 1.3056,
+      "step": 69525
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.496833550332777e-05,
+      "loss": 1.1733,
+      "step": 69530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4967679674514247e-05,
+      "loss": 1.1029,
+      "step": 69535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4967023817332598e-05,
+      "loss": 1.2868,
+      "step": 69540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4966367931786577e-05,
+      "loss": 1.1064,
+      "step": 69545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4965712017879921e-05,
+      "loss": 1.6982,
+      "step": 69550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.496505607561638e-05,
+      "loss": 1.2436,
+      "step": 69555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4964400104999697e-05,
+      "loss": 1.3778,
+      "step": 69560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.496374410603362e-05,
+      "loss": 1.4264,
+      "step": 69565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4963088078721897e-05,
+      "loss": 1.251,
+      "step": 69570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4962432023068265e-05,
+      "loss": 1.002,
+      "step": 69575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4961775939076482e-05,
+      "loss": 1.2434,
+      "step": 69580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.496111982675029e-05,
+      "loss": 1.4107,
+      "step": 69585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4960463686093434e-05,
+      "loss": 1.203,
+      "step": 69590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4959807517109662e-05,
+      "loss": 1.0396,
+      "step": 69595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4959151319802722e-05,
+      "loss": 1.303,
+      "step": 69600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.495849509417636e-05,
+      "loss": 1.1668,
+      "step": 69605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4957838840234327e-05,
+      "loss": 1.1692,
+      "step": 69610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4957182557980363e-05,
+      "loss": 1.0289,
+      "step": 69615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4956526247418223e-05,
+      "loss": 1.2628,
+      "step": 69620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4955869908551653e-05,
+      "loss": 1.3984,
+      "step": 69625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4955213541384399e-05,
+      "loss": 0.9403,
+      "step": 69630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.495455714592021e-05,
+      "loss": 1.3336,
+      "step": 69635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4953900722162838e-05,
+      "loss": 1.168,
+      "step": 69640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4953244270116023e-05,
+      "loss": 1.0521,
+      "step": 69645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4952587789783524e-05,
+      "loss": 1.473,
+      "step": 69650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4951931281169082e-05,
+      "loss": 1.3895,
+      "step": 69655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4951274744276452e-05,
+      "loss": 1.2558,
+      "step": 69660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4950618179109377e-05,
+      "loss": 1.0077,
+      "step": 69665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4949961585671611e-05,
+      "loss": 1.2614,
+      "step": 69670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4949304963966901e-05,
+      "loss": 0.9712,
+      "step": 69675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4948648313998999e-05,
+      "loss": 1.2698,
+      "step": 69680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4947991635771653e-05,
+      "loss": 1.324,
+      "step": 69685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4947334929288613e-05,
+      "loss": 1.3321,
+      "step": 69690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4946678194553629e-05,
+      "loss": 1.1607,
+      "step": 69695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4946021431570454e-05,
+      "loss": 1.2021,
+      "step": 69700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4945364640342836e-05,
+      "loss": 1.0774,
+      "step": 69705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4944707820874524e-05,
+      "loss": 1.1369,
+      "step": 69710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4944050973169273e-05,
+      "loss": 1.3358,
+      "step": 69715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4943394097230833e-05,
+      "loss": 1.2204,
+      "step": 69720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.494273719306295e-05,
+      "loss": 1.3153,
+      "step": 69725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4942080260669381e-05,
+      "loss": 1.1359,
+      "step": 69730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4941423300053876e-05,
+      "loss": 1.2083,
+      "step": 69735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4940766311220186e-05,
+      "loss": 1.7869,
+      "step": 69740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4940109294172063e-05,
+      "loss": 1.4135,
+      "step": 69745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4939452248913258e-05,
+      "loss": 1.3093,
+      "step": 69750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4938795175447525e-05,
+      "loss": 1.3729,
+      "step": 69755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4938138073778614e-05,
+      "loss": 1.0807,
+      "step": 69760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4937480943910278e-05,
+      "loss": 1.1065,
+      "step": 69765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4936823785846276e-05,
+      "loss": 0.9405,
+      "step": 69770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4936166599590347e-05,
+      "loss": 1.0945,
+      "step": 69775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4935509385146256e-05,
+      "loss": 1.1027,
+      "step": 69780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.493485214251775e-05,
+      "loss": 0.9683,
+      "step": 69785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4934194871708588e-05,
+      "loss": 1.2085,
+      "step": 69790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4933537572722515e-05,
+      "loss": 1.3834,
+      "step": 69795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.493288024556329e-05,
+      "loss": 1.3047,
+      "step": 69800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.493222289023467e-05,
+      "loss": 1.3723,
+      "step": 69805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4931565506740397e-05,
+      "loss": 1.5584,
+      "step": 69810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4930908095084236e-05,
+      "loss": 1.3056,
+      "step": 69815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4930250655269938e-05,
+      "loss": 1.2115,
+      "step": 69820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4929593187301258e-05,
+      "loss": 0.9451,
+      "step": 69825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4928935691181947e-05,
+      "loss": 1.1321,
+      "step": 69830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4928278166915764e-05,
+      "loss": 1.3658,
+      "step": 69835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4927620614506461e-05,
+      "loss": 1.0829,
+      "step": 69840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4926963033957795e-05,
+      "loss": 1.2316,
+      "step": 69845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.492630542527352e-05,
+      "loss": 1.1341,
+      "step": 69850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4925647788457388e-05,
+      "loss": 1.0703,
+      "step": 69855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4924990123513163e-05,
+      "loss": 1.32,
+      "step": 69860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4924332430444595e-05,
+      "loss": 1.0992,
+      "step": 69865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4923674709255435e-05,
+      "loss": 1.313,
+      "step": 69870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4923016959949452e-05,
+      "loss": 1.3764,
+      "step": 69875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4922359182530387e-05,
+      "loss": 1.0189,
+      "step": 69880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.492170137700201e-05,
+      "loss": 1.3485,
+      "step": 69885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4921043543368065e-05,
+      "loss": 1.2084,
+      "step": 69890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4920385681632316e-05,
+      "loss": 1.232,
+      "step": 69895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.491972779179852e-05,
+      "loss": 1.1035,
+      "step": 69900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4919069873870433e-05,
+      "loss": 1.345,
+      "step": 69905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4918411927851811e-05,
+      "loss": 1.2984,
+      "step": 69910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4917753953746408e-05,
+      "loss": 1.1179,
+      "step": 69915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.491709595155799e-05,
+      "loss": 1.0652,
+      "step": 69920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4916437921290306e-05,
+      "loss": 1.2296,
+      "step": 69925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4915779862947117e-05,
+      "loss": 1.0126,
+      "step": 69930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4915121776532182e-05,
+      "loss": 1.1581,
+      "step": 69935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4914463662049259e-05,
+      "loss": 1.1121,
+      "step": 69940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4913805519502105e-05,
+      "loss": 1.2968,
+      "step": 69945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4913147348894477e-05,
+      "loss": 1.2445,
+      "step": 69950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4912489150230136e-05,
+      "loss": 1.0913,
+      "step": 69955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4911830923512838e-05,
+      "loss": 0.9746,
+      "step": 69960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4911172668746348e-05,
+      "loss": 0.9264,
+      "step": 69965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4910514385934419e-05,
+      "loss": 1.076,
+      "step": 69970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4909856075080806e-05,
+      "loss": 1.1719,
+      "step": 69975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4909197736189282e-05,
+      "loss": 1.5822,
+      "step": 69980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4908539369263593e-05,
+      "loss": 1.0302,
+      "step": 69985
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.490788097430751e-05,
+      "loss": 1.0678,
+      "step": 69990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.490722255132478e-05,
+      "loss": 1.3453,
+      "step": 69995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4906564100319174e-05,
+      "loss": 1.014,
+      "step": 70000
+    },
+    {
+      "epoch": 0.34,
+      "eval_loss": 1.225610375404358,
+      "eval_runtime": 6475.9559,
+      "eval_samples_per_second": 3.569,
+      "eval_steps_per_second": 1.784,
+      "step": 70000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.490590562129445e-05,
+      "loss": 1.0557,
+      "step": 70005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.490524711425436e-05,
+      "loss": 1.3124,
+      "step": 70010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4904588579202674e-05,
+      "loss": 1.3012,
+      "step": 70015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4903930016143149e-05,
+      "loss": 1.0314,
+      "step": 70020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4903271425079547e-05,
+      "loss": 1.4507,
+      "step": 70025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4902612806015629e-05,
+      "loss": 1.1585,
+      "step": 70030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4901954158955151e-05,
+      "loss": 1.1201,
+      "step": 70035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4901295483901881e-05,
+      "loss": 1.1167,
+      "step": 70040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4900636780859577e-05,
+      "loss": 1.4045,
+      "step": 70045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4899978049832003e-05,
+      "loss": 1.3352,
+      "step": 70050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4899319290822913e-05,
+      "loss": 1.1862,
+      "step": 70055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4898660503836082e-05,
+      "loss": 1.2992,
+      "step": 70060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4898001688875262e-05,
+      "loss": 1.1167,
+      "step": 70065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4897342845944219e-05,
+      "loss": 1.2385,
+      "step": 70070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4896683975046711e-05,
+      "loss": 1.0299,
+      "step": 70075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4896025076186505e-05,
+      "loss": 0.9954,
+      "step": 70080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4895366149367365e-05,
+      "loss": 1.4806,
+      "step": 70085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4894707194593045e-05,
+      "loss": 1.0615,
+      "step": 70090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4894048211867322e-05,
+      "loss": 1.5696,
+      "step": 70095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4893389201193946e-05,
+      "loss": 1.0675,
+      "step": 70100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.489273016257669e-05,
+      "loss": 1.2806,
+      "step": 70105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4892071096019311e-05,
+      "loss": 1.1234,
+      "step": 70110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4891412001525573e-05,
+      "loss": 1.2339,
+      "step": 70115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4890752879099243e-05,
+      "loss": 1.404,
+      "step": 70120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4890093728744085e-05,
+      "loss": 1.3321,
+      "step": 70125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4889434550463861e-05,
+      "loss": 2.1127,
+      "step": 70130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4888775344262335e-05,
+      "loss": 1.061,
+      "step": 70135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4888116110143273e-05,
+      "loss": 1.659,
+      "step": 70140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4887456848110442e-05,
+      "loss": 1.1343,
+      "step": 70145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4886797558167598e-05,
+      "loss": 1.4706,
+      "step": 70150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4886138240318514e-05,
+      "loss": 1.2866,
+      "step": 70155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4885478894566952e-05,
+      "loss": 1.687,
+      "step": 70160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4884819520916681e-05,
+      "loss": 1.3501,
+      "step": 70165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.488416011937146e-05,
+      "loss": 1.2961,
+      "step": 70170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.488350068993506e-05,
+      "loss": 1.2255,
+      "step": 70175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4882841232611242e-05,
+      "loss": 1.2018,
+      "step": 70180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4882181747403774e-05,
+      "loss": 1.339,
+      "step": 70185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4881522234316425e-05,
+      "loss": 1.6076,
+      "step": 70190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4880862693352957e-05,
+      "loss": 1.0938,
+      "step": 70195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4880203124517138e-05,
+      "loss": 1.6981,
+      "step": 70200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4879543527812736e-05,
+      "loss": 1.4036,
+      "step": 70205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4878883903243513e-05,
+      "loss": 1.1225,
+      "step": 70210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4878224250813242e-05,
+      "loss": 1.1366,
+      "step": 70215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4877564570525684e-05,
+      "loss": 1.0953,
+      "step": 70220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4876904862384609e-05,
+      "loss": 1.2622,
+      "step": 70225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4876245126393787e-05,
+      "loss": 1.3725,
+      "step": 70230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4875585362556979e-05,
+      "loss": 1.262,
+      "step": 70235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4874925570877959e-05,
+      "loss": 1.3309,
+      "step": 70240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4874265751360491e-05,
+      "loss": 1.5537,
+      "step": 70245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4873605904008344e-05,
+      "loss": 1.0931,
+      "step": 70250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4872946028825285e-05,
+      "loss": 1.1838,
+      "step": 70255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4872286125815086e-05,
+      "loss": 1.3204,
+      "step": 70260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.487162619498151e-05,
+      "loss": 1.0874,
+      "step": 70265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4870966236328327e-05,
+      "loss": 0.9896,
+      "step": 70270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.487030624985931e-05,
+      "loss": 1.0456,
+      "step": 70275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4869646235578223e-05,
+      "loss": 1.3376,
+      "step": 70280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4868986193488838e-05,
+      "loss": 1.0516,
+      "step": 70285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4868326123594923e-05,
+      "loss": 1.316,
+      "step": 70290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4867666025900249e-05,
+      "loss": 1.1676,
+      "step": 70295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.486700590040858e-05,
+      "loss": 1.1652,
+      "step": 70300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.486634574712369e-05,
+      "loss": 1.2514,
+      "step": 70305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4865685566049353e-05,
+      "loss": 1.512,
+      "step": 70310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.486502535718933e-05,
+      "loss": 1.3522,
+      "step": 70315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4864365120547397e-05,
+      "loss": 1.4708,
+      "step": 70320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.486370485612732e-05,
+      "loss": 1.5934,
+      "step": 70325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4863044563932878e-05,
+      "loss": 1.7214,
+      "step": 70330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4862384243967832e-05,
+      "loss": 1.2474,
+      "step": 70335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4861723896235958e-05,
+      "loss": 0.988,
+      "step": 70340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4861063520741024e-05,
+      "loss": 1.296,
+      "step": 70345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4860403117486806e-05,
+      "loss": 1.0986,
+      "step": 70350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.485974268647707e-05,
+      "loss": 0.9611,
+      "step": 70355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.485908222771559e-05,
+      "loss": 1.5185,
+      "step": 70360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4858421741206136e-05,
+      "loss": 0.9451,
+      "step": 70365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.485776122695248e-05,
+      "loss": 1.2545,
+      "step": 70370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4857100684958396e-05,
+      "loss": 1.4544,
+      "step": 70375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4856440115227655e-05,
+      "loss": 1.4384,
+      "step": 70380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4855779517764029e-05,
+      "loss": 1.3616,
+      "step": 70385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.485511889257129e-05,
+      "loss": 1.1377,
+      "step": 70390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4854458239653209e-05,
+      "loss": 1.3126,
+      "step": 70395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4853797559013563e-05,
+      "loss": 1.3781,
+      "step": 70400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.485313685065612e-05,
+      "loss": 1.0424,
+      "step": 70405
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4852476114584657e-05,
+      "loss": 1.3601,
+      "step": 70410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4851815350802947e-05,
+      "loss": 1.4373,
+      "step": 70415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4851154559314758e-05,
+      "loss": 1.3022,
+      "step": 70420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4850493740123873e-05,
+      "loss": 1.1617,
+      "step": 70425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4849832893234053e-05,
+      "loss": 1.455,
+      "step": 70430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4849172018649082e-05,
+      "loss": 1.379,
+      "step": 70435
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.484851111637273e-05,
+      "loss": 1.1636,
+      "step": 70440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4847850186408772e-05,
+      "loss": 1.182,
+      "step": 70445
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4847189228760984e-05,
+      "loss": 1.1218,
+      "step": 70450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4846528243433136e-05,
+      "loss": 1.1004,
+      "step": 70455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4845867230429006e-05,
+      "loss": 1.4335,
+      "step": 70460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4845206189752366e-05,
+      "loss": 1.0248,
+      "step": 70465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4844545121406994e-05,
+      "loss": 1.3756,
+      "step": 70470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4843884025396665e-05,
+      "loss": 1.2537,
+      "step": 70475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4843222901725152e-05,
+      "loss": 1.182,
+      "step": 70480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.484256175039623e-05,
+      "loss": 1.3521,
+      "step": 70485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4841900571413674e-05,
+      "loss": 1.3777,
+      "step": 70490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4841239364781265e-05,
+      "loss": 1.231,
+      "step": 70495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4840578130502774e-05,
+      "loss": 1.6246,
+      "step": 70500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4839916868581983e-05,
+      "loss": 1.4157,
+      "step": 70505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4839255579022658e-05,
+      "loss": 1.4467,
+      "step": 70510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4838594261828581e-05,
+      "loss": 1.2242,
+      "step": 70515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4837932917003532e-05,
+      "loss": 1.2397,
+      "step": 70520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.483727154455128e-05,
+      "loss": 1.1002,
+      "step": 70525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4836610144475606e-05,
+      "loss": 1.3437,
+      "step": 70530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4835948716780289e-05,
+      "loss": 1.2916,
+      "step": 70535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4835287261469102e-05,
+      "loss": 1.0776,
+      "step": 70540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4834625778545826e-05,
+      "loss": 1.6908,
+      "step": 70545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4833964268014235e-05,
+      "loss": 1.5411,
+      "step": 70550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4833302729878109e-05,
+      "loss": 1.5653,
+      "step": 70555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4832641164141224e-05,
+      "loss": 1.1187,
+      "step": 70560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.483197957080736e-05,
+      "loss": 1.2127,
+      "step": 70565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4831317949880292e-05,
+      "loss": 1.2691,
+      "step": 70570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4830656301363802e-05,
+      "loss": 1.4945,
+      "step": 70575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4829994625261663e-05,
+      "loss": 1.2741,
+      "step": 70580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4829332921577659e-05,
+      "loss": 1.4251,
+      "step": 70585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4828671190315567e-05,
+      "loss": 1.2686,
+      "step": 70590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4828009431479165e-05,
+      "loss": 1.0339,
+      "step": 70595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4827347645072232e-05,
+      "loss": 1.0603,
+      "step": 70600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4826685831098547e-05,
+      "loss": 1.1433,
+      "step": 70605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4826023989561893e-05,
+      "loss": 1.1626,
+      "step": 70610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4825362120466045e-05,
+      "loss": 0.8025,
+      "step": 70615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4824700223814779e-05,
+      "loss": 1.2269,
+      "step": 70620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4824038299611887e-05,
+      "loss": 1.3726,
+      "step": 70625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4823376347861136e-05,
+      "loss": 1.2562,
+      "step": 70630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4822714368566314e-05,
+      "loss": 1.3984,
+      "step": 70635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4822052361731199e-05,
+      "loss": 1.1496,
+      "step": 70640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4821390327359573e-05,
+      "loss": 1.183,
+      "step": 70645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4820728265455215e-05,
+      "loss": 1.3061,
+      "step": 70650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.48200661760219e-05,
+      "loss": 1.1466,
+      "step": 70655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4819404059063423e-05,
+      "loss": 1.4245,
+      "step": 70660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4818741914583552e-05,
+      "loss": 1.1251,
+      "step": 70665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4818079742586074e-05,
+      "loss": 1.0733,
+      "step": 70670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4817417543074767e-05,
+      "loss": 1.3533,
+      "step": 70675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4816755316053418e-05,
+      "loss": 1.1543,
+      "step": 70680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4816093061525805e-05,
+      "loss": 1.3008,
+      "step": 70685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.481543077949571e-05,
+      "loss": 1.4394,
+      "step": 70690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4814768469966913e-05,
+      "loss": 1.5227,
+      "step": 70695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4814106132943197e-05,
+      "loss": 1.7589,
+      "step": 70700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4813443768428353e-05,
+      "loss": 1.4284,
+      "step": 70705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4812781376426153e-05,
+      "loss": 0.9808,
+      "step": 70710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.481211895694038e-05,
+      "loss": 1.1442,
+      "step": 70715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4811456509974822e-05,
+      "loss": 1.2093,
+      "step": 70720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4810794035533258e-05,
+      "loss": 1.3046,
+      "step": 70725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4810131533619475e-05,
+      "loss": 1.1278,
+      "step": 70730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.480946900423725e-05,
+      "loss": 1.2588,
+      "step": 70735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4808806447390371e-05,
+      "loss": 1.1035,
+      "step": 70740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4808143863082625e-05,
+      "loss": 1.0353,
+      "step": 70745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4807481251317786e-05,
+      "loss": 1.1282,
+      "step": 70750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.480681861209965e-05,
+      "loss": 1.3015,
+      "step": 70755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4806155945431988e-05,
+      "loss": 1.3869,
+      "step": 70760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4805493251318592e-05,
+      "loss": 0.9847,
+      "step": 70765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4804830529763246e-05,
+      "loss": 1.024,
+      "step": 70770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4804167780769734e-05,
+      "loss": 1.161,
+      "step": 70775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4803505004341839e-05,
+      "loss": 1.1823,
+      "step": 70780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4802842200483347e-05,
+      "loss": 1.3626,
+      "step": 70785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4802179369198043e-05,
+      "loss": 1.1299,
+      "step": 70790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4801516510489711e-05,
+      "loss": 1.1324,
+      "step": 70795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.480085362436214e-05,
+      "loss": 1.1596,
+      "step": 70800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4800190710819106e-05,
+      "loss": 1.2162,
+      "step": 70805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.479952776986441e-05,
+      "loss": 0.9396,
+      "step": 70810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4798864801501824e-05,
+      "loss": 1.4387,
+      "step": 70815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4798201805735137e-05,
+      "loss": 1.0942,
+      "step": 70820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4797538782568142e-05,
+      "loss": 1.4109,
+      "step": 70825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4796875732004617e-05,
+      "loss": 1.2997,
+      "step": 70830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4796212654048354e-05,
+      "loss": 1.2713,
+      "step": 70835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4795549548703134e-05,
+      "loss": 1.3592,
+      "step": 70840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4794886415972749e-05,
+      "loss": 1.7746,
+      "step": 70845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4794223255860985e-05,
+      "loss": 1.1093,
+      "step": 70850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4793560068371623e-05,
+      "loss": 1.2492,
+      "step": 70855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4792896853508461e-05,
+      "loss": 1.0121,
+      "step": 70860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4792233611275273e-05,
+      "loss": 1.1034,
+      "step": 70865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4791570341675856e-05,
+      "loss": 1.328,
+      "step": 70870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4790907044713999e-05,
+      "loss": 1.4272,
+      "step": 70875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4790243720393482e-05,
+      "loss": 1.1435,
+      "step": 70880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4789580368718097e-05,
+      "loss": 1.347,
+      "step": 70885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4788916989691636e-05,
+      "loss": 1.0248,
+      "step": 70890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4788253583317879e-05,
+      "loss": 1.2038,
+      "step": 70895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.478759014960062e-05,
+      "loss": 1.1641,
+      "step": 70900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4786926688543645e-05,
+      "loss": 1.385,
+      "step": 70905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4786263200150748e-05,
+      "loss": 1.4148,
+      "step": 70910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4785599684425708e-05,
+      "loss": 1.3459,
+      "step": 70915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4784936141372326e-05,
+      "loss": 1.1064,
+      "step": 70920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4784272570994378e-05,
+      "loss": 1.1822,
+      "step": 70925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4783608973295667e-05,
+      "loss": 1.5293,
+      "step": 70930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4782945348279972e-05,
+      "loss": 1.1154,
+      "step": 70935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4782281695951088e-05,
+      "loss": 1.1686,
+      "step": 70940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4781618016312803e-05,
+      "loss": 1.2065,
+      "step": 70945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4780954309368906e-05,
+      "loss": 1.2188,
+      "step": 70950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4780290575123193e-05,
+      "loss": 1.3008,
+      "step": 70955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4779626813579445e-05,
+      "loss": 1.0818,
+      "step": 70960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4778963024741459e-05,
+      "loss": 1.6531,
+      "step": 70965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4778299208613023e-05,
+      "loss": 1.4379,
+      "step": 70970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4777635365197928e-05,
+      "loss": 1.3745,
+      "step": 70975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4776971494499967e-05,
+      "loss": 1.0739,
+      "step": 70980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.477630759652293e-05,
+      "loss": 1.1599,
+      "step": 70985
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4775643671270606e-05,
+      "loss": 1.1579,
+      "step": 70990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.477497971874679e-05,
+      "loss": 1.6356,
+      "step": 70995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.477431573895527e-05,
+      "loss": 1.0788,
+      "step": 71000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4773651731899838e-05,
+      "loss": 1.0247,
+      "step": 71005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4772987697584289e-05,
+      "loss": 1.4816,
+      "step": 71010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4772323636012414e-05,
+      "loss": 1.3642,
+      "step": 71015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4771659547188003e-05,
+      "loss": 1.0637,
+      "step": 71020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.477099543111485e-05,
+      "loss": 1.115,
+      "step": 71025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4770331287796745e-05,
+      "loss": 1.2129,
+      "step": 71030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4769667117237485e-05,
+      "loss": 1.5179,
+      "step": 71035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4769002919440857e-05,
+      "loss": 1.2387,
+      "step": 71040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4768338694410662e-05,
+      "loss": 1.8877,
+      "step": 71045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4767674442150685e-05,
+      "loss": 1.3006,
+      "step": 71050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4767010162664724e-05,
+      "loss": 1.273,
+      "step": 71055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4766345855956572e-05,
+      "loss": 1.351,
+      "step": 71060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4765681522030017e-05,
+      "loss": 1.3875,
+      "step": 71065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4765017160888862e-05,
+      "loss": 1.4219,
+      "step": 71070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4764352772536893e-05,
+      "loss": 1.4252,
+      "step": 71075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.476368835697791e-05,
+      "loss": 1.3779,
+      "step": 71080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4763023914215705e-05,
+      "loss": 1.0715,
+      "step": 71085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4762359444254069e-05,
+      "loss": 1.1597,
+      "step": 71090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4761694947096802e-05,
+      "loss": 1.3819,
+      "step": 71095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4761030422747693e-05,
+      "loss": 1.64,
+      "step": 71100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4760365871210538e-05,
+      "loss": 1.0439,
+      "step": 71105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4759701292489136e-05,
+      "loss": 1.2702,
+      "step": 71110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4759036686587281e-05,
+      "loss": 1.1829,
+      "step": 71115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4758372053508765e-05,
+      "loss": 1.2089,
+      "step": 71120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4757707393257386e-05,
+      "loss": 1.2224,
+      "step": 71125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4757042705836937e-05,
+      "loss": 1.3937,
+      "step": 71130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4756377991251217e-05,
+      "loss": 1.3047,
+      "step": 71135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4755713249504023e-05,
+      "loss": 1.1531,
+      "step": 71140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4755048480599144e-05,
+      "loss": 1.0789,
+      "step": 71145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4754383684540384e-05,
+      "loss": 1.3401,
+      "step": 71150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4753718861331534e-05,
+      "loss": 1.6422,
+      "step": 71155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.475305401097639e-05,
+      "loss": 1.6247,
+      "step": 71160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4752389133478756e-05,
+      "loss": 1.2169,
+      "step": 71165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.475172422884242e-05,
+      "loss": 1.4241,
+      "step": 71170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4751059297071186e-05,
+      "loss": 1.1325,
+      "step": 71175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4750394338168848e-05,
+      "loss": 1.2618,
+      "step": 71180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.47497293521392e-05,
+      "loss": 1.4776,
+      "step": 71185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4749064338986047e-05,
+      "loss": 1.0797,
+      "step": 71190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.474839929871318e-05,
+      "loss": 1.5917,
+      "step": 71195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4747734231324402e-05,
+      "loss": 1.4453,
+      "step": 71200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4747069136823504e-05,
+      "loss": 1.0581,
+      "step": 71205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.474640401521429e-05,
+      "loss": 1.4374,
+      "step": 71210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4745738866500557e-05,
+      "loss": 1.0775,
+      "step": 71215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.47450736906861e-05,
+      "loss": 1.0876,
+      "step": 71220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4744408487774722e-05,
+      "loss": 1.3013,
+      "step": 71225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.474374325777022e-05,
+      "loss": 1.2242,
+      "step": 71230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4743078000676389e-05,
+      "loss": 1.603,
+      "step": 71235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4742412716497034e-05,
+      "loss": 1.638,
+      "step": 71240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4741747405235956e-05,
+      "loss": 1.6786,
+      "step": 71245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4741082066896942e-05,
+      "loss": 1.146,
+      "step": 71250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4740416701483804e-05,
+      "loss": 1.2113,
+      "step": 71255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4739751309000338e-05,
+      "loss": 1.2595,
+      "step": 71260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.473908588945034e-05,
+      "loss": 1.257,
+      "step": 71265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4738420442837616e-05,
+      "loss": 1.0785,
+      "step": 71270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4737754969165958e-05,
+      "loss": 1.423,
+      "step": 71275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4737089468439172e-05,
+      "loss": 1.1497,
+      "step": 71280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.473642394066106e-05,
+      "loss": 1.3467,
+      "step": 71285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4735758385835417e-05,
+      "loss": 1.4389,
+      "step": 71290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.473509280396605e-05,
+      "loss": 1.0224,
+      "step": 71295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4734427195056752e-05,
+      "loss": 1.0614,
+      "step": 71300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.473376155911133e-05,
+      "loss": 1.0227,
+      "step": 71305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4733095896133585e-05,
+      "loss": 1.5095,
+      "step": 71310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4732430206127314e-05,
+      "loss": 1.254,
+      "step": 71315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4731764489096326e-05,
+      "loss": 1.0339,
+      "step": 71320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4731098745044414e-05,
+      "loss": 1.1755,
+      "step": 71325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4730432973975384e-05,
+      "loss": 1.5328,
+      "step": 71330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4729767175893038e-05,
+      "loss": 1.6961,
+      "step": 71335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4729101350801176e-05,
+      "loss": 1.2546,
+      "step": 71340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.47284354987036e-05,
+      "loss": 1.1947,
+      "step": 71345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.472776961960412e-05,
+      "loss": 1.2705,
+      "step": 71350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.472710371350653e-05,
+      "loss": 1.2471,
+      "step": 71355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4726437780414632e-05,
+      "loss": 1.1542,
+      "step": 71360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4725771820332237e-05,
+      "loss": 1.3284,
+      "step": 71365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.472510583326314e-05,
+      "loss": 1.1076,
+      "step": 71370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.472443981921115e-05,
+      "loss": 1.1842,
+      "step": 71375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4723773778180064e-05,
+      "loss": 1.105,
+      "step": 71380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4723107710173692e-05,
+      "loss": 1.3015,
+      "step": 71385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4722441615195832e-05,
+      "loss": 0.7824,
+      "step": 71390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4721775493250294e-05,
+      "loss": 1.4929,
+      "step": 71395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4721109344340875e-05,
+      "loss": 1.2327,
+      "step": 71400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4720443168471384e-05,
+      "loss": 1.1627,
+      "step": 71405
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4719776965645624e-05,
+      "loss": 0.9189,
+      "step": 71410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.47191107358674e-05,
+      "loss": 1.0545,
+      "step": 71415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4718444479140512e-05,
+      "loss": 1.1048,
+      "step": 71420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.471777819546877e-05,
+      "loss": 1.2475,
+      "step": 71425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4717111884855977e-05,
+      "loss": 1.0422,
+      "step": 71430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.471644554730594e-05,
+      "loss": 1.2144,
+      "step": 71435
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4715779182822458e-05,
+      "loss": 1.0023,
+      "step": 71440
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4715112791409344e-05,
+      "loss": 1.1576,
+      "step": 71445
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4714446373070397e-05,
+      "loss": 1.6208,
+      "step": 71450
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4713779927809428e-05,
+      "loss": 0.9944,
+      "step": 71455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.471311345563024e-05,
+      "loss": 1.2084,
+      "step": 71460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4712446956536637e-05,
+      "loss": 1.3126,
+      "step": 71465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4711780430532431e-05,
+      "loss": 1.2775,
+      "step": 71470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.471111387762142e-05,
+      "loss": 1.3812,
+      "step": 71475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.471044729780742e-05,
+      "loss": 1.0218,
+      "step": 71480
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4709780691094229e-05,
+      "loss": 1.1377,
+      "step": 71485
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4709114057485656e-05,
+      "loss": 1.5928,
+      "step": 71490
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4708447396985514e-05,
+      "loss": 1.1158,
+      "step": 71495
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4707780709597601e-05,
+      "loss": 1.2641,
+      "step": 71500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4707113995325728e-05,
+      "loss": 1.1348,
+      "step": 71505
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4706447254173704e-05,
+      "loss": 1.2902,
+      "step": 71510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4705780486145334e-05,
+      "loss": 1.3653,
+      "step": 71515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4705113691244426e-05,
+      "loss": 1.2043,
+      "step": 71520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4704446869474787e-05,
+      "loss": 1.197,
+      "step": 71525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4703780020840228e-05,
+      "loss": 1.1344,
+      "step": 71530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4703113145344556e-05,
+      "loss": 1.1449,
+      "step": 71535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4702446242991576e-05,
+      "loss": 1.6656,
+      "step": 71540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4701779313785098e-05,
+      "loss": 1.014,
+      "step": 71545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4701112357728936e-05,
+      "loss": 1.3154,
+      "step": 71550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4700445374826891e-05,
+      "loss": 1.2291,
+      "step": 71555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4699778365082775e-05,
+      "loss": 1.3862,
+      "step": 71560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4699111328500397e-05,
+      "loss": 1.1325,
+      "step": 71565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4698444265083562e-05,
+      "loss": 1.1478,
+      "step": 71570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.469777717483609e-05,
+      "loss": 1.2058,
+      "step": 71575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4697110057761776e-05,
+      "loss": 0.9968,
+      "step": 71580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4696442913864442e-05,
+      "loss": 1.1815,
+      "step": 71585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4695775743147893e-05,
+      "loss": 1.303,
+      "step": 71590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4695108545615939e-05,
+      "loss": 1.5059,
+      "step": 71595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.469444132127239e-05,
+      "loss": 1.3343,
+      "step": 71600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4693774070121054e-05,
+      "loss": 1.468,
+      "step": 71605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4693106792165746e-05,
+      "loss": 1.2826,
+      "step": 71610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4692439487410273e-05,
+      "loss": 1.0234,
+      "step": 71615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4691772155858445e-05,
+      "loss": 1.2146,
+      "step": 71620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4691104797514078e-05,
+      "loss": 1.2586,
+      "step": 71625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4690437412380976e-05,
+      "loss": 1.2694,
+      "step": 71630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4689770000462955e-05,
+      "loss": 1.4186,
+      "step": 71635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4689102561763826e-05,
+      "loss": 1.1068,
+      "step": 71640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.46884350962874e-05,
+      "loss": 1.5669,
+      "step": 71645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4687767604037482e-05,
+      "loss": 1.2537,
+      "step": 71650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4687100085017898e-05,
+      "loss": 1.3818,
+      "step": 71655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4686432539232446e-05,
+      "loss": 1.1797,
+      "step": 71660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4685764966684946e-05,
+      "loss": 1.157,
+      "step": 71665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4685097367379207e-05,
+      "loss": 1.2991,
+      "step": 71670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.468442974131904e-05,
+      "loss": 1.2476,
+      "step": 71675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4683762088508263e-05,
+      "loss": 1.0006,
+      "step": 71680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4683094408950682e-05,
+      "loss": 1.2298,
+      "step": 71685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4682426702650116e-05,
+      "loss": 1.4536,
+      "step": 71690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4681758969610376e-05,
+      "loss": 0.993,
+      "step": 71695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.468109120983527e-05,
+      "loss": 1.2947,
+      "step": 71700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4680423423328621e-05,
+      "loss": 1.2828,
+      "step": 71705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.467975561009423e-05,
+      "loss": 1.2667,
+      "step": 71710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4679087770135923e-05,
+      "loss": 1.1553,
+      "step": 71715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4678419903457503e-05,
+      "loss": 1.3273,
+      "step": 71720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4677752010062793e-05,
+      "loss": 1.5556,
+      "step": 71725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4677084089955604e-05,
+      "loss": 1.1573,
+      "step": 71730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4676416143139748e-05,
+      "loss": 1.2319,
+      "step": 71735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.467574816961904e-05,
+      "loss": 1.1764,
+      "step": 71740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4675080169397296e-05,
+      "loss": 1.2461,
+      "step": 71745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4674412142478329e-05,
+      "loss": 1.2984,
+      "step": 71750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4673744088865954e-05,
+      "loss": 1.4367,
+      "step": 71755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4673076008563989e-05,
+      "loss": 1.3182,
+      "step": 71760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4672407901576243e-05,
+      "loss": 1.2865,
+      "step": 71765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4671739767906537e-05,
+      "loss": 1.1361,
+      "step": 71770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4671071607558683e-05,
+      "loss": 1.4429,
+      "step": 71775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4670403420536496e-05,
+      "loss": 1.4934,
+      "step": 71780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4669735206843801e-05,
+      "loss": 1.2227,
+      "step": 71785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4669066966484398e-05,
+      "loss": 0.9989,
+      "step": 71790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4668398699462115e-05,
+      "loss": 1.7479,
+      "step": 71795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4667730405780765e-05,
+      "loss": 1.1233,
+      "step": 71800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4667062085444162e-05,
+      "loss": 1.6332,
+      "step": 71805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4666393738456125e-05,
+      "loss": 1.0622,
+      "step": 71810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4665725364820471e-05,
+      "loss": 0.9668,
+      "step": 71815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4665056964541014e-05,
+      "loss": 1.0191,
+      "step": 71820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4664388537621575e-05,
+      "loss": 1.1592,
+      "step": 71825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4663720084065966e-05,
+      "loss": 1.2993,
+      "step": 71830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4663051603878009e-05,
+      "loss": 1.0171,
+      "step": 71835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.466238309706152e-05,
+      "loss": 1.5097,
+      "step": 71840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4661714563620313e-05,
+      "loss": 1.1262,
+      "step": 71845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.466104600355821e-05,
+      "loss": 1.1331,
+      "step": 71850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4660377416879028e-05,
+      "loss": 1.202,
+      "step": 71855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4659708803586582e-05,
+      "loss": 1.318,
+      "step": 71860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4659040163684695e-05,
+      "loss": 1.1201,
+      "step": 71865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4658371497177182e-05,
+      "loss": 1.3136,
+      "step": 71870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4657702804067859e-05,
+      "loss": 1.0386,
+      "step": 71875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4657034084360553e-05,
+      "loss": 1.2926,
+      "step": 71880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4656365338059076e-05,
+      "loss": 1.3179,
+      "step": 71885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4655696565167249e-05,
+      "loss": 1.1246,
+      "step": 71890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.465502776568889e-05,
+      "loss": 1.0746,
+      "step": 71895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4654358939627816e-05,
+      "loss": 1.2529,
+      "step": 71900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4653690086987854e-05,
+      "loss": 1.4356,
+      "step": 71905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4653021207772812e-05,
+      "loss": 1.0755,
+      "step": 71910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4652352301986523e-05,
+      "loss": 1.0266,
+      "step": 71915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4651683369632794e-05,
+      "loss": 1.3454,
+      "step": 71920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4651014410715457e-05,
+      "loss": 1.2086,
+      "step": 71925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4650345425238323e-05,
+      "loss": 1.0759,
+      "step": 71930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4649676413205215e-05,
+      "loss": 1.3385,
+      "step": 71935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4649007374619954e-05,
+      "loss": 1.3396,
+      "step": 71940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4648338309486363e-05,
+      "loss": 1.167,
+      "step": 71945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4647669217808258e-05,
+      "loss": 1.2983,
+      "step": 71950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4647000099589464e-05,
+      "loss": 1.2082,
+      "step": 71955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4646330954833796e-05,
+      "loss": 1.179,
+      "step": 71960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4645661783545083e-05,
+      "loss": 1.2529,
+      "step": 71965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.464499258572714e-05,
+      "loss": 1.0883,
+      "step": 71970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4644323361383792e-05,
+      "loss": 1.4235,
+      "step": 71975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.464365411051886e-05,
+      "loss": 1.1197,
+      "step": 71980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4642984833136167e-05,
+      "loss": 1.03,
+      "step": 71985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4642315529239528e-05,
+      "loss": 1.3201,
+      "step": 71990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4641646198832778e-05,
+      "loss": 0.899,
+      "step": 71995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4640976841919728e-05,
+      "loss": 1.078,
+      "step": 72000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.46403074585042e-05,
+      "loss": 1.2145,
+      "step": 72005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4639638048590028e-05,
+      "loss": 1.1294,
+      "step": 72010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4638968612181019e-05,
+      "loss": 1.1953,
+      "step": 72015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.463829914928101e-05,
+      "loss": 1.2267,
+      "step": 72020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4637629659893817e-05,
+      "loss": 1.1709,
+      "step": 72025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4636960144023265e-05,
+      "loss": 1.2892,
+      "step": 72030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4636290601673176e-05,
+      "loss": 1.2709,
+      "step": 72035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4635621032847374e-05,
+      "loss": 1.0441,
+      "step": 72040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4634951437549682e-05,
+      "loss": 1.3773,
+      "step": 72045
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4634281815783924e-05,
+      "loss": 1.053,
+      "step": 72050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4633612167553926e-05,
+      "loss": 1.1957,
+      "step": 72055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4632942492863512e-05,
+      "loss": 0.9944,
+      "step": 72060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4632272791716499e-05,
+      "loss": 1.4697,
+      "step": 72065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.463160306411672e-05,
+      "loss": 1.2527,
+      "step": 72070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4630933310067995e-05,
+      "loss": 1.3074,
+      "step": 72075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4630263529574153e-05,
+      "loss": 1.0996,
+      "step": 72080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4629593722639012e-05,
+      "loss": 1.0765,
+      "step": 72085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4628923889266404e-05,
+      "loss": 1.3973,
+      "step": 72090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4628254029460149e-05,
+      "loss": 1.1357,
+      "step": 72095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4627584143224078e-05,
+      "loss": 1.2702,
+      "step": 72100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4626914230562008e-05,
+      "loss": 1.2143,
+      "step": 72105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4626244291477769e-05,
+      "loss": 1.2632,
+      "step": 72110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4625574325975192e-05,
+      "loss": 1.2705,
+      "step": 72115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4624904334058092e-05,
+      "loss": 1.3098,
+      "step": 72120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4624234315730304e-05,
+      "loss": 1.1619,
+      "step": 72125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4623564270995649e-05,
+      "loss": 1.0641,
+      "step": 72130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4622894199857958e-05,
+      "loss": 1.1253,
+      "step": 72135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4622224102321054e-05,
+      "loss": 1.3913,
+      "step": 72140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.462155397838876e-05,
+      "loss": 0.9672,
+      "step": 72145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4620883828064911e-05,
+      "loss": 1.6032,
+      "step": 72150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.462021365135333e-05,
+      "loss": 0.9667,
+      "step": 72155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4619543448257842e-05,
+      "loss": 1.1425,
+      "step": 72160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4618873218782279e-05,
+      "loss": 1.4451,
+      "step": 72165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4618202962930464e-05,
+      "loss": 1.0887,
+      "step": 72170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4617532680706225e-05,
+      "loss": 1.1212,
+      "step": 72175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4616862372113393e-05,
+      "loss": 1.0683,
+      "step": 72180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4616192037155794e-05,
+      "loss": 1.2007,
+      "step": 72185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4615521675837251e-05,
+      "loss": 1.4195,
+      "step": 72190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4614851288161602e-05,
+      "loss": 1.2137,
+      "step": 72195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.461418087413267e-05,
+      "loss": 1.2833,
+      "step": 72200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.461351043375428e-05,
+      "loss": 1.0154,
+      "step": 72205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4612839967030269e-05,
+      "loss": 1.2986,
+      "step": 72210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4612169473964456e-05,
+      "loss": 1.2244,
+      "step": 72215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4611498954560677e-05,
+      "loss": 0.9769,
+      "step": 72220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4610828408822757e-05,
+      "loss": 1.4927,
+      "step": 72225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4610157836754526e-05,
+      "loss": 1.0374,
+      "step": 72230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4609487238359815e-05,
+      "loss": 1.3951,
+      "step": 72235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4608816613642457e-05,
+      "loss": 1.0405,
+      "step": 72240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4608145962606273e-05,
+      "loss": 1.5887,
+      "step": 72245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4607475285255098e-05,
+      "loss": 1.4863,
+      "step": 72250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4606804581592759e-05,
+      "loss": 1.1246,
+      "step": 72255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4606133851623093e-05,
+      "loss": 1.2302,
+      "step": 72260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4605463095349921e-05,
+      "loss": 1.3552,
+      "step": 72265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4604792312777078e-05,
+      "loss": 1.2437,
+      "step": 72270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4604121503908395e-05,
+      "loss": 1.2594,
+      "step": 72275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4603450668747702e-05,
+      "loss": 1.4024,
+      "step": 72280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4602779807298831e-05,
+      "loss": 1.3745,
+      "step": 72285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4602108919565609e-05,
+      "loss": 1.1199,
+      "step": 72290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4601438005551868e-05,
+      "loss": 1.3099,
+      "step": 72295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4600767065261448e-05,
+      "loss": 1.1497,
+      "step": 72300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4600096098698168e-05,
+      "loss": 0.7746,
+      "step": 72305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4599425105865867e-05,
+      "loss": 1.1813,
+      "step": 72310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4598754086768377e-05,
+      "loss": 1.395,
+      "step": 72315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4598083041409523e-05,
+      "loss": 1.3165,
+      "step": 72320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4597411969793146e-05,
+      "loss": 1.1877,
+      "step": 72325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.459674087192307e-05,
+      "loss": 1.4288,
+      "step": 72330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4596069747803135e-05,
+      "loss": 0.9812,
+      "step": 72335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.459539859743717e-05,
+      "loss": 1.2908,
+      "step": 72340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4594727420829002e-05,
+      "loss": 1.2535,
+      "step": 72345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4594056217982475e-05,
+      "loss": 1.1588,
+      "step": 72350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4593384988901411e-05,
+      "loss": 1.784,
+      "step": 72355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4592713733589652e-05,
+      "loss": 1.0353,
+      "step": 72360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4592042452051027e-05,
+      "loss": 1.5635,
+      "step": 72365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4591371144289369e-05,
+      "loss": 1.1882,
+      "step": 72370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.459069981030851e-05,
+      "loss": 1.2063,
+      "step": 72375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4590028450112291e-05,
+      "loss": 1.1666,
+      "step": 72380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4589357063704537e-05,
+      "loss": 1.2182,
+      "step": 72385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4588685651089087e-05,
+      "loss": 1.1838,
+      "step": 72390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4588014212269772e-05,
+      "loss": 1.167,
+      "step": 72395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.458734274725043e-05,
+      "loss": 1.244,
+      "step": 72400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4586671256034895e-05,
+      "loss": 1.0107,
+      "step": 72405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4585999738626999e-05,
+      "loss": 1.7152,
+      "step": 72410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4585328195030575e-05,
+      "loss": 1.2566,
+      "step": 72415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4584656625249465e-05,
+      "loss": 1.1534,
+      "step": 72420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4583985029287498e-05,
+      "loss": 1.0966,
+      "step": 72425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4583313407148511e-05,
+      "loss": 1.6674,
+      "step": 72430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4582641758836338e-05,
+      "loss": 1.2439,
+      "step": 72435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4581970084354817e-05,
+      "loss": 1.2612,
+      "step": 72440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4581298383707785e-05,
+      "loss": 1.0958,
+      "step": 72445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.458062665689907e-05,
+      "loss": 1.198,
+      "step": 72450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4579954903932516e-05,
+      "loss": 1.2768,
+      "step": 72455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4579283124811956e-05,
+      "loss": 1.604,
+      "step": 72460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4578611319541228e-05,
+      "loss": 1.182,
+      "step": 72465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4577939488124164e-05,
+      "loss": 1.3901,
+      "step": 72470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4577267630564604e-05,
+      "loss": 1.6329,
+      "step": 72475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4576595746866385e-05,
+      "loss": 1.144,
+      "step": 72480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.457592383703334e-05,
+      "loss": 1.1441,
+      "step": 72485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4575251901069311e-05,
+      "loss": 1.2502,
+      "step": 72490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4574579938978135e-05,
+      "loss": 1.1551,
+      "step": 72495
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4573907950763643e-05,
+      "loss": 0.872,
+      "step": 72500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4573235936429679e-05,
+      "loss": 1.1258,
+      "step": 72505
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4572563895980078e-05,
+      "loss": 1.1135,
+      "step": 72510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4571891829418676e-05,
+      "loss": 1.3744,
+      "step": 72515
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4571219736749314e-05,
+      "loss": 1.5778,
+      "step": 72520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.457054761797583e-05,
+      "loss": 1.0835,
+      "step": 72525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4569875473102056e-05,
+      "loss": 1.0091,
+      "step": 72530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.456920330213184e-05,
+      "loss": 1.2135,
+      "step": 72535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4568531105069015e-05,
+      "loss": 1.3151,
+      "step": 72540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4567858881917416e-05,
+      "loss": 1.2542,
+      "step": 72545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4567186632680892e-05,
+      "loss": 1.5168,
+      "step": 72550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.456651435736327e-05,
+      "loss": 1.5534,
+      "step": 72555
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.45658420559684e-05,
+      "loss": 1.2468,
+      "step": 72560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4565169728500113e-05,
+      "loss": 1.2191,
+      "step": 72565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4564497374962252e-05,
+      "loss": 1.298,
+      "step": 72570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4563824995358657e-05,
+      "loss": 1.3377,
+      "step": 72575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4563152589693166e-05,
+      "loss": 1.2927,
+      "step": 72580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.456248015796962e-05,
+      "loss": 1.318,
+      "step": 72585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4561807700191857e-05,
+      "loss": 1.1993,
+      "step": 72590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.456113521636372e-05,
+      "loss": 1.0692,
+      "step": 72595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4560462706489046e-05,
+      "loss": 1.1815,
+      "step": 72600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.455979017057168e-05,
+      "loss": 1.2874,
+      "step": 72605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4559117608615455e-05,
+      "loss": 1.1742,
+      "step": 72610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4558445020624221e-05,
+      "loss": 1.447,
+      "step": 72615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4557772406601812e-05,
+      "loss": 1.3862,
+      "step": 72620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.455709976655207e-05,
+      "loss": 1.2092,
+      "step": 72625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4556427100478838e-05,
+      "loss": 1.4087,
+      "step": 72630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4555754408385956e-05,
+      "loss": 1.128,
+      "step": 72635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4555081690277268e-05,
+      "loss": 1.0786,
+      "step": 72640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4554408946156615e-05,
+      "loss": 1.1641,
+      "step": 72645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4553736176027834e-05,
+      "loss": 1.2743,
+      "step": 72650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4553063379894772e-05,
+      "loss": 1.2552,
+      "step": 72655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4552390557761266e-05,
+      "loss": 1.0358,
+      "step": 72660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4551717709631165e-05,
+      "loss": 1.1349,
+      "step": 72665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4551044835508306e-05,
+      "loss": 1.459,
+      "step": 72670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4550371935396535e-05,
+      "loss": 1.3217,
+      "step": 72675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4549699009299691e-05,
+      "loss": 0.956,
+      "step": 72680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4549026057221621e-05,
+      "loss": 1.3206,
+      "step": 72685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4548353079166162e-05,
+      "loss": 1.1645,
+      "step": 72690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4547680075137162e-05,
+      "loss": 1.15,
+      "step": 72695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4547007045138464e-05,
+      "loss": 1.247,
+      "step": 72700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4546333989173909e-05,
+      "loss": 1.1697,
+      "step": 72705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4545660907247341e-05,
+      "loss": 1.0822,
+      "step": 72710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4544987799362607e-05,
+      "loss": 1.6594,
+      "step": 72715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4544314665523545e-05,
+      "loss": 1.2924,
+      "step": 72720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4543641505734002e-05,
+      "loss": 1.127,
+      "step": 72725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4542968319997823e-05,
+      "loss": 1.1046,
+      "step": 72730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4542295108318856e-05,
+      "loss": 1.1608,
+      "step": 72735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4541621870700936e-05,
+      "loss": 1.2379,
+      "step": 72740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.454094860714791e-05,
+      "loss": 1.1658,
+      "step": 72745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.454027531766363e-05,
+      "loss": 0.9788,
+      "step": 72750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4539602002251933e-05,
+      "loss": 1.2581,
+      "step": 72755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4538928660916672e-05,
+      "loss": 1.4459,
+      "step": 72760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.453825529366168e-05,
+      "loss": 1.1159,
+      "step": 72765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4537581900490814e-05,
+      "loss": 1.3189,
+      "step": 72770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4536908481407914e-05,
+      "loss": 1.3851,
+      "step": 72775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4536235036416825e-05,
+      "loss": 1.4951,
+      "step": 72780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4535561565521395e-05,
+      "loss": 1.4429,
+      "step": 72785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4534888068725471e-05,
+      "loss": 1.6854,
+      "step": 72790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4534214546032895e-05,
+      "loss": 1.7265,
+      "step": 72795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4533540997447513e-05,
+      "loss": 1.0106,
+      "step": 72800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4532867422973176e-05,
+      "loss": 1.1086,
+      "step": 72805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4532193822613726e-05,
+      "loss": 1.5351,
+      "step": 72810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4531520196373014e-05,
+      "loss": 1.257,
+      "step": 72815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4530846544254887e-05,
+      "loss": 1.3623,
+      "step": 72820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4530172866263186e-05,
+      "loss": 1.4851,
+      "step": 72825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4529499162401762e-05,
+      "loss": 1.172,
+      "step": 72830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4528825432674459e-05,
+      "loss": 1.3114,
+      "step": 72835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.452815167708513e-05,
+      "loss": 1.1652,
+      "step": 72840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4527477895637619e-05,
+      "loss": 1.2595,
+      "step": 72845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.452680408833577e-05,
+      "loss": 1.3086,
+      "step": 72850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4526130255183441e-05,
+      "loss": 1.1408,
+      "step": 72855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.452545639618447e-05,
+      "loss": 1.0987,
+      "step": 72860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4524782511342712e-05,
+      "loss": 0.9745,
+      "step": 72865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4524108600662007e-05,
+      "loss": 0.9911,
+      "step": 72870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4523434664146215e-05,
+      "loss": 1.2146,
+      "step": 72875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4522760701799175e-05,
+      "loss": 1.1613,
+      "step": 72880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4522086713624739e-05,
+      "loss": 1.3242,
+      "step": 72885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4521412699626755e-05,
+      "loss": 1.6415,
+      "step": 72890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4520738659809073e-05,
+      "loss": 1.0643,
+      "step": 72895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4520064594175542e-05,
+      "loss": 1.2642,
+      "step": 72900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.451939050273001e-05,
+      "loss": 1.475,
+      "step": 72905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.451871638547633e-05,
+      "loss": 1.3523,
+      "step": 72910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4518042242418346e-05,
+      "loss": 1.7597,
+      "step": 72915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4517368073559914e-05,
+      "loss": 1.2043,
+      "step": 72920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.451669387890488e-05,
+      "loss": 1.2507,
+      "step": 72925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.451601965845709e-05,
+      "loss": 1.4489,
+      "step": 72930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4515345412220404e-05,
+      "loss": 1.1007,
+      "step": 72935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4514671140198663e-05,
+      "loss": 1.1474,
+      "step": 72940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4513996842395727e-05,
+      "loss": 1.098,
+      "step": 72945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.451332251881544e-05,
+      "loss": 1.1016,
+      "step": 72950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.451264816946165e-05,
+      "loss": 1.2256,
+      "step": 72955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4511973794338215e-05,
+      "loss": 1.1908,
+      "step": 72960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4511299393448981e-05,
+      "loss": 1.195,
+      "step": 72965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4510624966797804e-05,
+      "loss": 1.024,
+      "step": 72970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4509950514388526e-05,
+      "loss": 1.2187,
+      "step": 72975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.450927603622501e-05,
+      "loss": 1.2476,
+      "step": 72980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4508601532311105e-05,
+      "loss": 1.6039,
+      "step": 72985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4507927002650655e-05,
+      "loss": 1.3347,
+      "step": 72990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4507252447247519e-05,
+      "loss": 1.0707,
+      "step": 72995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4506577866105547e-05,
+      "loss": 1.1977,
+      "step": 73000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4505903259228592e-05,
+      "loss": 1.5573,
+      "step": 73005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4505228626620507e-05,
+      "loss": 1.4274,
+      "step": 73010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4504553968285139e-05,
+      "loss": 1.1938,
+      "step": 73015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.450387928422635e-05,
+      "loss": 1.2227,
+      "step": 73020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4503204574447987e-05,
+      "loss": 1.2592,
+      "step": 73025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4502529838953903e-05,
+      "loss": 1.5405,
+      "step": 73030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4501855077747951e-05,
+      "loss": 1.1047,
+      "step": 73035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4501180290833987e-05,
+      "loss": 1.7511,
+      "step": 73040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.450050547821586e-05,
+      "loss": 1.1369,
+      "step": 73045
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.449983063989743e-05,
+      "loss": 1.0961,
+      "step": 73050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4499155775882546e-05,
+      "loss": 0.953,
+      "step": 73055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4498480886175057e-05,
+      "loss": 0.9021,
+      "step": 73060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4497805970778829e-05,
+      "loss": 1.2207,
+      "step": 73065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4497131029697706e-05,
+      "loss": 1.3241,
+      "step": 73070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.449645606293555e-05,
+      "loss": 1.4151,
+      "step": 73075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4495781070496207e-05,
+      "loss": 1.3464,
+      "step": 73080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4495106052383539e-05,
+      "loss": 1.2311,
+      "step": 73085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4494431008601397e-05,
+      "loss": 1.3436,
+      "step": 73090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4493755939153635e-05,
+      "loss": 1.6695,
+      "step": 73095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4493080844044112e-05,
+      "loss": 1.1984,
+      "step": 73100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4492405723276679e-05,
+      "loss": 1.2858,
+      "step": 73105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4491730576855195e-05,
+      "loss": 1.3645,
+      "step": 73110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4491055404783511e-05,
+      "loss": 1.4801,
+      "step": 73115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4490380207065486e-05,
+      "loss": 1.0408,
+      "step": 73120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4489704983704975e-05,
+      "loss": 1.0561,
+      "step": 73125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4489029734705835e-05,
+      "loss": 1.4189,
+      "step": 73130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4488354460071921e-05,
+      "loss": 1.077,
+      "step": 73135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4487679159807085e-05,
+      "loss": 1.1517,
+      "step": 73140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4487003833915192e-05,
+      "loss": 1.4018,
+      "step": 73145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4486328482400092e-05,
+      "loss": 1.4441,
+      "step": 73150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4485653105265645e-05,
+      "loss": 1.0767,
+      "step": 73155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4484977702515704e-05,
+      "loss": 1.5031,
+      "step": 73160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4484302274154124e-05,
+      "loss": 1.3788,
+      "step": 73165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4483626820184773e-05,
+      "loss": 1.0278,
+      "step": 73170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4482951340611497e-05,
+      "loss": 1.045,
+      "step": 73175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4482275835438162e-05,
+      "loss": 1.3856,
+      "step": 73180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4481600304668619e-05,
+      "loss": 1.2068,
+      "step": 73185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4480924748306727e-05,
+      "loss": 1.0257,
+      "step": 73190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4480249166356346e-05,
+      "loss": 1.2537,
+      "step": 73195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4479573558821329e-05,
+      "loss": 1.2047,
+      "step": 73200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4478897925705542e-05,
+      "loss": 1.1896,
+      "step": 73205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4478222267012839e-05,
+      "loss": 1.1783,
+      "step": 73210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4477546582747076e-05,
+      "loss": 1.5056,
+      "step": 73215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4476870872912114e-05,
+      "loss": 1.294,
+      "step": 73220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4476195137511811e-05,
+      "loss": 1.2412,
+      "step": 73225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4475519376550027e-05,
+      "loss": 1.1145,
+      "step": 73230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.447484359003062e-05,
+      "loss": 1.3896,
+      "step": 73235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4474167777957449e-05,
+      "loss": 1.4659,
+      "step": 73240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4473491940334375e-05,
+      "loss": 1.1183,
+      "step": 73245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4472816077165256e-05,
+      "loss": 1.3626,
+      "step": 73250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4472140188453949e-05,
+      "loss": 1.5717,
+      "step": 73255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4471464274204316e-05,
+      "loss": 0.9925,
+      "step": 73260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.447078833442022e-05,
+      "loss": 1.2789,
+      "step": 73265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4470112369105515e-05,
+      "loss": 1.5851,
+      "step": 73270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4469436378264068e-05,
+      "loss": 1.2268,
+      "step": 73275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.446876036189973e-05,
+      "loss": 1.3028,
+      "step": 73280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4468084320016372e-05,
+      "loss": 1.0572,
+      "step": 73285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4467408252617847e-05,
+      "loss": 1.5734,
+      "step": 73290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4466732159708017e-05,
+      "loss": 1.4346,
+      "step": 73295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4466056041290749e-05,
+      "loss": 0.9667,
+      "step": 73300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.446537989736989e-05,
+      "loss": 1.1965,
+      "step": 73305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4464703727949316e-05,
+      "loss": 1.2592,
+      "step": 73310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4464027533032883e-05,
+      "loss": 1.2309,
+      "step": 73315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.446335131262445e-05,
+      "loss": 1.305,
+      "step": 73320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.446267506672788e-05,
+      "loss": 1.2301,
+      "step": 73325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4461998795347037e-05,
+      "loss": 1.3985,
+      "step": 73330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.446132249848578e-05,
+      "loss": 1.143,
+      "step": 73335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.446064617614797e-05,
+      "loss": 1.2289,
+      "step": 73340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4459969828337474e-05,
+      "loss": 1.2012,
+      "step": 73345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4459293455058149e-05,
+      "loss": 1.1794,
+      "step": 73350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4458617056313861e-05,
+      "loss": 1.0892,
+      "step": 73355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4457940632108473e-05,
+      "loss": 1.2389,
+      "step": 73360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4457264182445845e-05,
+      "loss": 0.9375,
+      "step": 73365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.445658770732984e-05,
+      "loss": 1.2091,
+      "step": 73370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4455911206764324e-05,
+      "loss": 1.2566,
+      "step": 73375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4455234680753161e-05,
+      "loss": 1.375,
+      "step": 73380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4454558129300207e-05,
+      "loss": 1.4908,
+      "step": 73385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.445388155240933e-05,
+      "loss": 1.126,
+      "step": 73390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4453204950084399e-05,
+      "loss": 1.41,
+      "step": 73395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4452528322329266e-05,
+      "loss": 1.2824,
+      "step": 73400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4451851669147808e-05,
+      "loss": 1.045,
+      "step": 73405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4451174990543876e-05,
+      "loss": 1.452,
+      "step": 73410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4450498286521345e-05,
+      "loss": 1.6196,
+      "step": 73415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4449821557084073e-05,
+      "loss": 1.127,
+      "step": 73420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4449144802235928e-05,
+      "loss": 1.2142,
+      "step": 73425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4448468021980772e-05,
+      "loss": 1.8405,
+      "step": 73430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4447791216322473e-05,
+      "loss": 1.4041,
+      "step": 73435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4447114385264894e-05,
+      "loss": 1.1654,
+      "step": 73440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4446437528811896e-05,
+      "loss": 1.3488,
+      "step": 73445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4445760646967354e-05,
+      "loss": 1.1403,
+      "step": 73450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4445083739735122e-05,
+      "loss": 1.2876,
+      "step": 73455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4444406807119076e-05,
+      "loss": 1.2836,
+      "step": 73460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4443729849123074e-05,
+      "loss": 1.1753,
+      "step": 73465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4443052865750982e-05,
+      "loss": 1.1568,
+      "step": 73470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4442375857006671e-05,
+      "loss": 1.1632,
+      "step": 73475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4441698822894003e-05,
+      "loss": 1.1853,
+      "step": 73480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.444102176341685e-05,
+      "loss": 1.1932,
+      "step": 73485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4440344678579072e-05,
+      "loss": 1.1832,
+      "step": 73490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4439667568384535e-05,
+      "loss": 0.9739,
+      "step": 73495
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4438990432837112e-05,
+      "loss": 1.0828,
+      "step": 73500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4438313271940662e-05,
+      "loss": 1.5072,
+      "step": 73505
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4437636085699059e-05,
+      "loss": 1.0953,
+      "step": 73510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4436958874116168e-05,
+      "loss": 1.0542,
+      "step": 73515
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4436281637195855e-05,
+      "loss": 1.0144,
+      "step": 73520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4435604374941986e-05,
+      "loss": 1.0391,
+      "step": 73525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.443492708735843e-05,
+      "loss": 1.172,
+      "step": 73530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4434249774449056e-05,
+      "loss": 1.3899,
+      "step": 73535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4433572436217733e-05,
+      "loss": 1.5234,
+      "step": 73540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4432895072668322e-05,
+      "loss": 1.3533,
+      "step": 73545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.44322176838047e-05,
+      "loss": 1.401,
+      "step": 73550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4431540269630731e-05,
+      "loss": 1.0074,
+      "step": 73555
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.443086283015028e-05,
+      "loss": 1.1885,
+      "step": 73560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4430185365367221e-05,
+      "loss": 1.4768,
+      "step": 73565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4429507875285422e-05,
+      "loss": 1.4352,
+      "step": 73570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.442883035990875e-05,
+      "loss": 1.249,
+      "step": 73575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4428152819241077e-05,
+      "loss": 1.3865,
+      "step": 73580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4427475253286267e-05,
+      "loss": 1.2667,
+      "step": 73585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4426797662048191e-05,
+      "loss": 1.465,
+      "step": 73590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.442612004553072e-05,
+      "loss": 1.026,
+      "step": 73595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4425442403737721e-05,
+      "loss": 1.018,
+      "step": 73600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4424764736673071e-05,
+      "loss": 1.2105,
+      "step": 73605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.442408704434063e-05,
+      "loss": 1.2248,
+      "step": 73610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4423409326744273e-05,
+      "loss": 1.214,
+      "step": 73615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4422731583887873e-05,
+      "loss": 1.1392,
+      "step": 73620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4422053815775292e-05,
+      "loss": 1.3146,
+      "step": 73625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4421376022410411e-05,
+      "loss": 1.179,
+      "step": 73630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.442069820379709e-05,
+      "loss": 1.1143,
+      "step": 73635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4420020359939206e-05,
+      "loss": 1.2199,
+      "step": 73640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4419342490840628e-05,
+      "loss": 1.2856,
+      "step": 73645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.441866459650523e-05,
+      "loss": 1.3251,
+      "step": 73650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.441798667693688e-05,
+      "loss": 1.1864,
+      "step": 73655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4417308732139445e-05,
+      "loss": 1.2161,
+      "step": 73660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4416630762116807e-05,
+      "loss": 1.1464,
+      "step": 73665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.441595276687283e-05,
+      "loss": 1.0559,
+      "step": 73670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4415274746411387e-05,
+      "loss": 1.3441,
+      "step": 73675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4414596700736347e-05,
+      "loss": 1.2685,
+      "step": 73680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4413918629851592e-05,
+      "loss": 0.8816,
+      "step": 73685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4413240533760985e-05,
+      "loss": 1.5311,
+      "step": 73690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.44125624124684e-05,
+      "loss": 1.2338,
+      "step": 73695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4411884265977712e-05,
+      "loss": 1.4215,
+      "step": 73700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4411206094292788e-05,
+      "loss": 1.0823,
+      "step": 73705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.441052789741751e-05,
+      "loss": 1.1212,
+      "step": 73710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.440984967535574e-05,
+      "loss": 1.3886,
+      "step": 73715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4409171428111361e-05,
+      "loss": 1.1346,
+      "step": 73720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4408493155688239e-05,
+      "loss": 1.1387,
+      "step": 73725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4407814858090252e-05,
+      "loss": 1.2688,
+      "step": 73730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4407136535321273e-05,
+      "loss": 1.0054,
+      "step": 73735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4406458187385169e-05,
+      "loss": 1.2245,
+      "step": 73740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4405779814285821e-05,
+      "loss": 1.3442,
+      "step": 73745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4405101416027101e-05,
+      "loss": 1.2167,
+      "step": 73750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4404422992612883e-05,
+      "loss": 1.4337,
+      "step": 73755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.440374454404704e-05,
+      "loss": 1.123,
+      "step": 73760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4403066070333447e-05,
+      "loss": 1.3658,
+      "step": 73765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.440238757147598e-05,
+      "loss": 1.0957,
+      "step": 73770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.440170904747851e-05,
+      "loss": 1.5608,
+      "step": 73775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4401030498344915e-05,
+      "loss": 1.1052,
+      "step": 73780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4400351924079065e-05,
+      "loss": 1.4855,
+      "step": 73785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4399673324684847e-05,
+      "loss": 1.5005,
+      "step": 73790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4398994700166122e-05,
+      "loss": 1.1066,
+      "step": 73795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.439831605052677e-05,
+      "loss": 0.9151,
+      "step": 73800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4397637375770672e-05,
+      "loss": 1.2876,
+      "step": 73805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4396958675901695e-05,
+      "loss": 0.8963,
+      "step": 73810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4396279950923724e-05,
+      "loss": 1.8162,
+      "step": 73815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4395601200840625e-05,
+      "loss": 1.0967,
+      "step": 73820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4394922425656284e-05,
+      "loss": 1.081,
+      "step": 73825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.439424362537457e-05,
+      "loss": 1.1941,
+      "step": 73830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.439356479999936e-05,
+      "loss": 1.4242,
+      "step": 73835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4392885949534532e-05,
+      "loss": 1.4771,
+      "step": 73840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4392207073983964e-05,
+      "loss": 1.2971,
+      "step": 73845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.439152817335153e-05,
+      "loss": 1.097,
+      "step": 73850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.439084924764111e-05,
+      "loss": 1.29,
+      "step": 73855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4390170296856579e-05,
+      "loss": 1.1837,
+      "step": 73860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4389491321001811e-05,
+      "loss": 1.4903,
+      "step": 73865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4388812320080691e-05,
+      "loss": 1.3526,
+      "step": 73870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.438813329409709e-05,
+      "loss": 1.1136,
+      "step": 73875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4387454243054888e-05,
+      "loss": 1.3743,
+      "step": 73880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4386775166957962e-05,
+      "loss": 1.6515,
+      "step": 73885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4386096065810191e-05,
+      "loss": 1.256,
+      "step": 73890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.438541693961545e-05,
+      "loss": 1.0854,
+      "step": 73895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4384737788377622e-05,
+      "loss": 1.0372,
+      "step": 73900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4384058612100582e-05,
+      "loss": 1.2834,
+      "step": 73905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4383379410788212e-05,
+      "loss": 1.2442,
+      "step": 73910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4382700184444382e-05,
+      "loss": 1.2536,
+      "step": 73915
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4382020933072984e-05,
+      "loss": 1.5208,
+      "step": 73920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4381341656677886e-05,
+      "loss": 0.9513,
+      "step": 73925
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4380662355262969e-05,
+      "loss": 1.441,
+      "step": 73930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4379983028832119e-05,
+      "loss": 1.349,
+      "step": 73935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4379303677389205e-05,
+      "loss": 1.1983,
+      "step": 73940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4378624300938113e-05,
+      "loss": 1.2015,
+      "step": 73945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.437794489948272e-05,
+      "loss": 1.1402,
+      "step": 73950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.437726547302691e-05,
+      "loss": 0.978,
+      "step": 73955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4376586021574557e-05,
+      "loss": 0.953,
+      "step": 73960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4375906545129545e-05,
+      "loss": 0.9229,
+      "step": 73965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4375227043695755e-05,
+      "loss": 1.3237,
+      "step": 73970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4374547517277062e-05,
+      "loss": 1.5238,
+      "step": 73975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4373867965877351e-05,
+      "loss": 1.3253,
+      "step": 73980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4373188389500502e-05,
+      "loss": 1.3799,
+      "step": 73985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4372508788150395e-05,
+      "loss": 1.3521,
+      "step": 73990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4371829161830913e-05,
+      "loss": 1.2146,
+      "step": 73995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4371149510545932e-05,
+      "loss": 1.0888,
+      "step": 74000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4370469834299336e-05,
+      "loss": 1.102,
+      "step": 74005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4369790133095007e-05,
+      "loss": 1.4488,
+      "step": 74010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4369110406936829e-05,
+      "loss": 1.133,
+      "step": 74015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4368430655828676e-05,
+      "loss": 1.1853,
+      "step": 74020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.436775087977444e-05,
+      "loss": 1.2024,
+      "step": 74025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4367071078777995e-05,
+      "loss": 1.3989,
+      "step": 74030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4366391252843222e-05,
+      "loss": 1.2169,
+      "step": 74035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4365711401974009e-05,
+      "loss": 1.1547,
+      "step": 74040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4365031526174233e-05,
+      "loss": 1.6453,
+      "step": 74045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4364351625447783e-05,
+      "loss": 1.2512,
+      "step": 74050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4363671699798538e-05,
+      "loss": 1.0163,
+      "step": 74055
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.436299174923038e-05,
+      "loss": 1.5952,
+      "step": 74060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.436231177374719e-05,
+      "loss": 0.8819,
+      "step": 74065
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4361631773352858e-05,
+      "loss": 1.1278,
+      "step": 74070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.436095174805126e-05,
+      "loss": 0.9577,
+      "step": 74075
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.436027169784628e-05,
+      "loss": 1.2736,
+      "step": 74080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4359591622741807e-05,
+      "loss": 1.7032,
+      "step": 74085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4358911522741719e-05,
+      "loss": 1.0039,
+      "step": 74090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4358231397849902e-05,
+      "loss": 1.0211,
+      "step": 74095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4357551248070238e-05,
+      "loss": 1.138,
+      "step": 74100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4356871073406615e-05,
+      "loss": 1.6397,
+      "step": 74105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4356190873862917e-05,
+      "loss": 1.1829,
+      "step": 74110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4355510649443021e-05,
+      "loss": 1.1903,
+      "step": 74115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4354830400150821e-05,
+      "loss": 1.2746,
+      "step": 74120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4354150125990192e-05,
+      "loss": 1.1284,
+      "step": 74125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4353469826965027e-05,
+      "loss": 1.2384,
+      "step": 74130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.435278950307921e-05,
+      "loss": 1.5283,
+      "step": 74135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.435210915433662e-05,
+      "loss": 1.3965,
+      "step": 74140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.435142878074115e-05,
+      "loss": 1.3309,
+      "step": 74145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4350748382296677e-05,
+      "loss": 1.3177,
+      "step": 74150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4350067959007093e-05,
+      "loss": 1.2521,
+      "step": 74155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.434938751087628e-05,
+      "loss": 1.2203,
+      "step": 74160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4348707037908126e-05,
+      "loss": 1.0182,
+      "step": 74165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4348026540106517e-05,
+      "loss": 1.2563,
+      "step": 74170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4347346017475335e-05,
+      "loss": 1.2043,
+      "step": 74175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4346665470018471e-05,
+      "loss": 1.4249,
+      "step": 74180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4345984897739808e-05,
+      "loss": 1.0583,
+      "step": 74185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4345304300643234e-05,
+      "loss": 1.1722,
+      "step": 74190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4344623678732636e-05,
+      "loss": 1.529,
+      "step": 74195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.43439430320119e-05,
+      "loss": 0.9828,
+      "step": 74200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4343262360484913e-05,
+      "loss": 1.2607,
+      "step": 74205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4342581664155561e-05,
+      "loss": 1.4608,
+      "step": 74210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4341900943027734e-05,
+      "loss": 1.0777,
+      "step": 74215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4341220197105313e-05,
+      "loss": 1.398,
+      "step": 74220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4340539426392196e-05,
+      "loss": 1.3197,
+      "step": 74225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4339858630892258e-05,
+      "loss": 1.6957,
+      "step": 74230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4339177810609394e-05,
+      "loss": 1.3551,
+      "step": 74235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4338496965547497e-05,
+      "loss": 1.0703,
+      "step": 74240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4337816095710441e-05,
+      "loss": 1.2103,
+      "step": 74245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4337135201102128e-05,
+      "loss": 1.3878,
+      "step": 74250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4336454281726438e-05,
+      "loss": 1.1772,
+      "step": 74255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.433577333758726e-05,
+      "loss": 1.2947,
+      "step": 74260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4335092368688486e-05,
+      "loss": 1.4552,
+      "step": 74265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4334411375034003e-05,
+      "loss": 0.9831,
+      "step": 74270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.43337303566277e-05,
+      "loss": 1.2959,
+      "step": 74275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4333049313473465e-05,
+      "loss": 1.2923,
+      "step": 74280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4332368245575188e-05,
+      "loss": 1.2211,
+      "step": 74285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.433168715293676e-05,
+      "loss": 1.1808,
+      "step": 74290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4331006035562065e-05,
+      "loss": 1.0924,
+      "step": 74295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4330324893455e-05,
+      "loss": 1.5328,
+      "step": 74300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4329643726619447e-05,
+      "loss": 1.0221,
+      "step": 74305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.43289625350593e-05,
+      "loss": 1.354,
+      "step": 74310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4328281318778452e-05,
+      "loss": 1.0038,
+      "step": 74315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4327600077780787e-05,
+      "loss": 1.156,
+      "step": 74320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4326918812070196e-05,
+      "loss": 1.4671,
+      "step": 74325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4326237521650578e-05,
+      "loss": 1.2999,
+      "step": 74330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4325556206525813e-05,
+      "loss": 1.4035,
+      "step": 74335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4324874866699791e-05,
+      "loss": 1.6794,
+      "step": 74340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4324193502176415e-05,
+      "loss": 1.154,
+      "step": 74345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4323512112959564e-05,
+      "loss": 1.0719,
+      "step": 74350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4322830699053137e-05,
+      "loss": 1.2192,
+      "step": 74355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4322149260461017e-05,
+      "loss": 1.2048,
+      "step": 74360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4321467797187101e-05,
+      "loss": 1.004,
+      "step": 74365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4320786309235282e-05,
+      "loss": 1.5485,
+      "step": 74370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4320104796609448e-05,
+      "loss": 1.0526,
+      "step": 74375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4319423259313492e-05,
+      "loss": 1.3868,
+      "step": 74380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4318741697351307e-05,
+      "loss": 1.3362,
+      "step": 74385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4318060110726781e-05,
+      "loss": 1.1953,
+      "step": 74390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4317378499443812e-05,
+      "loss": 1.5331,
+      "step": 74395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.431669686350629e-05,
+      "loss": 1.3554,
+      "step": 74400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4316015202918105e-05,
+      "loss": 1.0932,
+      "step": 74405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4315333517683154e-05,
+      "loss": 1.2748,
+      "step": 74410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4314651807805326e-05,
+      "loss": 1.1042,
+      "step": 74415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4313970073288515e-05,
+      "loss": 1.119,
+      "step": 74420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4313288314136617e-05,
+      "loss": 1.4633,
+      "step": 74425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.431260653035352e-05,
+      "loss": 1.3837,
+      "step": 74430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4311924721943125e-05,
+      "loss": 1.4552,
+      "step": 74435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4311242888909315e-05,
+      "loss": 1.7708,
+      "step": 74440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4310561031255992e-05,
+      "loss": 1.2236,
+      "step": 74445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4309879148987049e-05,
+      "loss": 1.2208,
+      "step": 74450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4309197242106372e-05,
+      "loss": 0.9857,
+      "step": 74455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4308515310617867e-05,
+      "loss": 1.2138,
+      "step": 74460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4307833354525417e-05,
+      "loss": 1.0961,
+      "step": 74465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4307151373832926e-05,
+      "loss": 1.0372,
+      "step": 74470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4306469368544284e-05,
+      "loss": 1.2056,
+      "step": 74475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4305787338663382e-05,
+      "loss": 1.3027,
+      "step": 74480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4305105284194123e-05,
+      "loss": 1.2095,
+      "step": 74485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4304423205140395e-05,
+      "loss": 1.095,
+      "step": 74490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4303741101506094e-05,
+      "loss": 1.2502,
+      "step": 74495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4303058973295118e-05,
+      "loss": 1.1324,
+      "step": 74500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4302376820511361e-05,
+      "loss": 1.2116,
+      "step": 74505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.430169464315872e-05,
+      "loss": 1.1578,
+      "step": 74510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4301012441241084e-05,
+      "loss": 1.4826,
+      "step": 74515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4300330214762359e-05,
+      "loss": 1.4198,
+      "step": 74520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4299647963726431e-05,
+      "loss": 1.0799,
+      "step": 74525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4298965688137202e-05,
+      "loss": 1.3252,
+      "step": 74530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4298283387998568e-05,
+      "loss": 1.2308,
+      "step": 74535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4297601063314423e-05,
+      "loss": 1.3966,
+      "step": 74540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4296918714088666e-05,
+      "loss": 1.224,
+      "step": 74545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.429623634032519e-05,
+      "loss": 0.966,
+      "step": 74550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4295553942027896e-05,
+      "loss": 0.9399,
+      "step": 74555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4294871519200675e-05,
+      "loss": 1.0539,
+      "step": 74560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4294189071847431e-05,
+      "loss": 1.0515,
+      "step": 74565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4293506599972055e-05,
+      "loss": 1.0787,
+      "step": 74570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4292824103578449e-05,
+      "loss": 1.241,
+      "step": 74575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.429214158267051e-05,
+      "loss": 1.3091,
+      "step": 74580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4291459037252131e-05,
+      "loss": 1.1315,
+      "step": 74585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4290776467327215e-05,
+      "loss": 1.0756,
+      "step": 74590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4290093872899654e-05,
+      "loss": 1.0689,
+      "step": 74595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4289411253973354e-05,
+      "loss": 1.1461,
+      "step": 74600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4288728610552207e-05,
+      "loss": 1.4985,
+      "step": 74605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4288045942640112e-05,
+      "loss": 1.1649,
+      "step": 74610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4287363250240968e-05,
+      "loss": 1.264,
+      "step": 74615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4286680533358676e-05,
+      "loss": 1.1341,
+      "step": 74620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.428599779199713e-05,
+      "loss": 1.0516,
+      "step": 74625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4285315026160231e-05,
+      "loss": 1.2788,
+      "step": 74630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4284632235851884e-05,
+      "loss": 0.9471,
+      "step": 74635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4283949421075978e-05,
+      "loss": 1.1271,
+      "step": 74640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4283266581836416e-05,
+      "loss": 1.0591,
+      "step": 74645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4282583718137102e-05,
+      "loss": 1.0622,
+      "step": 74650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4281900829981925e-05,
+      "loss": 1.1491,
+      "step": 74655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4281217917374796e-05,
+      "loss": 1.2312,
+      "step": 74660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4280534980319609e-05,
+      "loss": 1.1808,
+      "step": 74665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4279852018820265e-05,
+      "loss": 1.3591,
+      "step": 74670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4279169032880666e-05,
+      "loss": 1.2969,
+      "step": 74675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4278486022504707e-05,
+      "loss": 1.3011,
+      "step": 74680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4277802987696295e-05,
+      "loss": 1.487,
+      "step": 74685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4277119928459322e-05,
+      "loss": 1.1778,
+      "step": 74690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4276436844797696e-05,
+      "loss": 1.5644,
+      "step": 74695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4275753736715318e-05,
+      "loss": 1.1166,
+      "step": 74700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4275070604216085e-05,
+      "loss": 1.1912,
+      "step": 74705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.42743874473039e-05,
+      "loss": 1.2373,
+      "step": 74710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4273704265982661e-05,
+      "loss": 1.0997,
+      "step": 74715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4273021060256275e-05,
+      "loss": 1.056,
+      "step": 74720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.427233783012864e-05,
+      "loss": 1.0663,
+      "step": 74725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4271654575603657e-05,
+      "loss": 1.1745,
+      "step": 74730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.427097129668523e-05,
+      "loss": 0.981,
+      "step": 74735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4270287993377258e-05,
+      "loss": 1.1892,
+      "step": 74740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4269604665683648e-05,
+      "loss": 1.3847,
+      "step": 74745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4268921313608296e-05,
+      "loss": 1.1727,
+      "step": 74750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.426823793715511e-05,
+      "loss": 1.127,
+      "step": 74755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4267554536327985e-05,
+      "loss": 0.8945,
+      "step": 74760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4266871111130834e-05,
+      "loss": 1.1976,
+      "step": 74765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4266187661567548e-05,
+      "loss": 1.1774,
+      "step": 74770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4265504187642038e-05,
+      "loss": 1.5199,
+      "step": 74775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.426482068935821e-05,
+      "loss": 1.502,
+      "step": 74780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4264137166719956e-05,
+      "loss": 1.0948,
+      "step": 74785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.426345361973119e-05,
+      "loss": 1.3665,
+      "step": 74790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4262770048395805e-05,
+      "loss": 1.1696,
+      "step": 74795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4262086452717711e-05,
+      "loss": 1.1345,
+      "step": 74800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4261402832700815e-05,
+      "loss": 1.1672,
+      "step": 74805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4260719188349015e-05,
+      "loss": 1.3716,
+      "step": 74810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4260035519666217e-05,
+      "loss": 1.1422,
+      "step": 74815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4259351826656325e-05,
+      "loss": 1.2373,
+      "step": 74820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4258668109323241e-05,
+      "loss": 0.946,
+      "step": 74825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4257984367670874e-05,
+      "loss": 1.085,
+      "step": 74830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4257300601703125e-05,
+      "loss": 0.991,
+      "step": 74835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4256616811423903e-05,
+      "loss": 1.2139,
+      "step": 74840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4255932996837105e-05,
+      "loss": 1.3339,
+      "step": 74845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4255249157946645e-05,
+      "loss": 1.4104,
+      "step": 74850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4254565294756417e-05,
+      "loss": 1.3748,
+      "step": 74855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.425388140727034e-05,
+      "loss": 1.2501,
+      "step": 74860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4253197495492308e-05,
+      "loss": 1.1847,
+      "step": 74865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4252513559426234e-05,
+      "loss": 0.9467,
+      "step": 74870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4251829599076016e-05,
+      "loss": 1.2251,
+      "step": 74875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4251145614445565e-05,
+      "loss": 1.1467,
+      "step": 74880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.425046160553879e-05,
+      "loss": 0.9883,
+      "step": 74885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4249777572359589e-05,
+      "loss": 1.2255,
+      "step": 74890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4249093514911875e-05,
+      "loss": 0.9253,
+      "step": 74895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4248409433199548e-05,
+      "loss": 1.0473,
+      "step": 74900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.424772532722652e-05,
+      "loss": 1.3185,
+      "step": 74905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4247041196996699e-05,
+      "loss": 1.4376,
+      "step": 74910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4246357042513985e-05,
+      "loss": 1.0014,
+      "step": 74915
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.424567286378229e-05,
+      "loss": 0.9036,
+      "step": 74920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4244988660805517e-05,
+      "loss": 1.2241,
+      "step": 74925
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4244304433587576e-05,
+      "loss": 0.9874,
+      "step": 74930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4243620182132377e-05,
+      "loss": 1.1498,
+      "step": 74935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4242935906443822e-05,
+      "loss": 1.1373,
+      "step": 74940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4242251606525821e-05,
+      "loss": 0.9513,
+      "step": 74945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4241567282382281e-05,
+      "loss": 1.6743,
+      "step": 74950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4240882934017112e-05,
+      "loss": 0.9829,
+      "step": 74955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4240198561434218e-05,
+      "loss": 1.0929,
+      "step": 74960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4239514164637514e-05,
+      "loss": 1.2577,
+      "step": 74965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4238829743630898e-05,
+      "loss": 0.9874,
+      "step": 74970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.423814529841829e-05,
+      "loss": 1.0302,
+      "step": 74975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4237460829003591e-05,
+      "loss": 1.0435,
+      "step": 74980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4236776335390707e-05,
+      "loss": 1.6934,
+      "step": 74985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.423609181758356e-05,
+      "loss": 1.2315,
+      "step": 74990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4235407275586042e-05,
+      "loss": 1.3808,
+      "step": 74995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4234722709402075e-05,
+      "loss": 1.5715,
+      "step": 75000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4234038119035563e-05,
+      "loss": 1.1024,
+      "step": 75005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4233353504490418e-05,
+      "loss": 1.3914,
+      "step": 75010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4232668865770545e-05,
+      "loss": 1.3237,
+      "step": 75015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4231984202879856e-05,
+      "loss": 1.4095,
+      "step": 75020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4231299515822261e-05,
+      "loss": 1.3909,
+      "step": 75025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4230614804601671e-05,
+      "loss": 1.4987,
+      "step": 75030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4229930069221993e-05,
+      "loss": 1.131,
+      "step": 75035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.422924530968714e-05,
+      "loss": 1.2231,
+      "step": 75040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4228560526001025e-05,
+      "loss": 1.2748,
+      "step": 75045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4227875718167552e-05,
+      "loss": 1.2125,
+      "step": 75050
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4227190886190636e-05,
+      "loss": 1.8343,
+      "step": 75055
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4226506030074182e-05,
+      "loss": 1.1455,
+      "step": 75060
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.422582114982211e-05,
+      "loss": 1.2287,
+      "step": 75065
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4225136245438327e-05,
+      "loss": 1.031,
+      "step": 75070
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4224451316926742e-05,
+      "loss": 1.048,
+      "step": 75075
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4223766364291267e-05,
+      "loss": 1.2708,
+      "step": 75080
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4223081387535816e-05,
+      "loss": 1.2494,
+      "step": 75085
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4222396386664295e-05,
+      "loss": 1.1839,
+      "step": 75090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4221711361680624e-05,
+      "loss": 0.9481,
+      "step": 75095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4221026312588708e-05,
+      "loss": 1.4635,
+      "step": 75100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4220341239392462e-05,
+      "loss": 1.1987,
+      "step": 75105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.42196561420958e-05,
+      "loss": 1.2948,
+      "step": 75110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4218971020702628e-05,
+      "loss": 1.4766,
+      "step": 75115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4218285875216865e-05,
+      "loss": 1.1281,
+      "step": 75120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.421760070564242e-05,
+      "loss": 1.2603,
+      "step": 75125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4216915511983204e-05,
+      "loss": 1.1659,
+      "step": 75130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4216230294243137e-05,
+      "loss": 1.1456,
+      "step": 75135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4215545052426125e-05,
+      "loss": 1.1723,
+      "step": 75140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4214859786536083e-05,
+      "loss": 1.6493,
+      "step": 75145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4214174496576925e-05,
+      "loss": 1.048,
+      "step": 75150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4213489182552565e-05,
+      "loss": 1.4391,
+      "step": 75155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4212803844466914e-05,
+      "loss": 1.1109,
+      "step": 75160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4212118482323889e-05,
+      "loss": 1.6113,
+      "step": 75165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4211433096127399e-05,
+      "loss": 1.198,
+      "step": 75170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4210747685881366e-05,
+      "loss": 1.7064,
+      "step": 75175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4210062251589696e-05,
+      "loss": 1.5998,
+      "step": 75180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4209376793256302e-05,
+      "loss": 1.1691,
+      "step": 75185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.420869131088511e-05,
+      "loss": 1.0307,
+      "step": 75190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.420800580448002e-05,
+      "loss": 1.1919,
+      "step": 75195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.420732027404496e-05,
+      "loss": 1.084,
+      "step": 75200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4206634719583833e-05,
+      "loss": 1.411,
+      "step": 75205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.420594914110056e-05,
+      "loss": 1.1301,
+      "step": 75210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4205263538599059e-05,
+      "loss": 1.1166,
+      "step": 75215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4204577912083238e-05,
+      "loss": 1.208,
+      "step": 75220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.420389226155702e-05,
+      "loss": 1.2064,
+      "step": 75225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.420320658702431e-05,
+      "loss": 1.4437,
+      "step": 75230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4202520888489033e-05,
+      "loss": 1.1453,
+      "step": 75235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.42018351659551e-05,
+      "loss": 1.4431,
+      "step": 75240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.420114941942643e-05,
+      "loss": 1.0999,
+      "step": 75245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4200463648906938e-05,
+      "loss": 1.1397,
+      "step": 75250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4199777854400536e-05,
+      "loss": 1.4801,
+      "step": 75255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4199092035911147e-05,
+      "loss": 1.1553,
+      "step": 75260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4198406193442682e-05,
+      "loss": 1.1089,
+      "step": 75265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4197720326999057e-05,
+      "loss": 1.1204,
+      "step": 75270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4197034436584196e-05,
+      "loss": 1.2459,
+      "step": 75275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4196348522202008e-05,
+      "loss": 1.7908,
+      "step": 75280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4195662583856415e-05,
+      "loss": 1.1013,
+      "step": 75285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4194976621551328e-05,
+      "loss": 1.1559,
+      "step": 75290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4194290635290675e-05,
+      "loss": 1.2987,
+      "step": 75295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4193604625078358e-05,
+      "loss": 1.2069,
+      "step": 75300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4192918590918312e-05,
+      "loss": 1.0705,
+      "step": 75305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4192232532814438e-05,
+      "loss": 1.1922,
+      "step": 75310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4191546450770665e-05,
+      "loss": 1.3446,
+      "step": 75315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.419086034479091e-05,
+      "loss": 1.2842,
+      "step": 75320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4190174214879085e-05,
+      "loss": 1.132,
+      "step": 75325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4189488061039111e-05,
+      "loss": 1.2075,
+      "step": 75330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4188801883274907e-05,
+      "loss": 1.2246,
+      "step": 75335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4188115681590392e-05,
+      "loss": 0.9674,
+      "step": 75340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4187429455989486e-05,
+      "loss": 1.2431,
+      "step": 75345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4186743206476104e-05,
+      "loss": 1.3102,
+      "step": 75350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4186056933054167e-05,
+      "loss": 1.2571,
+      "step": 75355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4185370635727591e-05,
+      "loss": 1.1859,
+      "step": 75360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4184684314500299e-05,
+      "loss": 1.2415,
+      "step": 75365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.418399796937621e-05,
+      "loss": 1.2476,
+      "step": 75370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.418331160035924e-05,
+      "loss": 1.1983,
+      "step": 75375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4182625207453313e-05,
+      "loss": 1.2854,
+      "step": 75380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4181938790662343e-05,
+      "loss": 0.9437,
+      "step": 75385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4181252349990258e-05,
+      "loss": 1.4542,
+      "step": 75390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4180565885440967e-05,
+      "loss": 1.0798,
+      "step": 75395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4179879397018404e-05,
+      "loss": 1.2311,
+      "step": 75400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4179192884726475e-05,
+      "loss": 1.2288,
+      "step": 75405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4178506348569113e-05,
+      "loss": 1.1473,
+      "step": 75410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4177819788550225e-05,
+      "loss": 1.3066,
+      "step": 75415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4177133204673743e-05,
+      "loss": 1.3561,
+      "step": 75420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4176446596943586e-05,
+      "loss": 1.2956,
+      "step": 75425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4175759965363667e-05,
+      "loss": 1.2225,
+      "step": 75430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4175073309937918e-05,
+      "loss": 1.0545,
+      "step": 75435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4174386630670252e-05,
+      "loss": 1.6245,
+      "step": 75440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4173699927564593e-05,
+      "loss": 1.1086,
+      "step": 75445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4173013200624861e-05,
+      "loss": 1.0932,
+      "step": 75450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4172326449854983e-05,
+      "loss": 1.3643,
+      "step": 75455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4171639675258875e-05,
+      "loss": 1.3392,
+      "step": 75460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4170952876840461e-05,
+      "loss": 1.2282,
+      "step": 75465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4170266054603663e-05,
+      "loss": 1.1169,
+      "step": 75470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.41695792085524e-05,
+      "loss": 1.0636,
+      "step": 75475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.41688923386906e-05,
+      "loss": 1.0419,
+      "step": 75480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4168205445022186e-05,
+      "loss": 0.9968,
+      "step": 75485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4167518527551071e-05,
+      "loss": 1.4497,
+      "step": 75490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4166831586281187e-05,
+      "loss": 1.146,
+      "step": 75495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4166144621216449e-05,
+      "loss": 1.2194,
+      "step": 75500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4165457632360791e-05,
+      "loss": 1.0379,
+      "step": 75505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4164770619718122e-05,
+      "loss": 1.1697,
+      "step": 75510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.416408358329238e-05,
+      "loss": 1.2351,
+      "step": 75515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4163396523087478e-05,
+      "loss": 1.2682,
+      "step": 75520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.416270943910734e-05,
+      "loss": 1.1938,
+      "step": 75525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4162022331355899e-05,
+      "loss": 1.6743,
+      "step": 75530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4161335199837065e-05,
+      "loss": 1.3241,
+      "step": 75535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4160648044554772e-05,
+      "loss": 1.1295,
+      "step": 75540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.415996086551294e-05,
+      "loss": 1.1445,
+      "step": 75545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4159273662715496e-05,
+      "loss": 1.1342,
+      "step": 75550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4158586436166359e-05,
+      "loss": 1.1277,
+      "step": 75555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4157899185869458e-05,
+      "loss": 1.2836,
+      "step": 75560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4157211911828719e-05,
+      "loss": 1.2683,
+      "step": 75565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4156524614048061e-05,
+      "loss": 1.2055,
+      "step": 75570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4155837292531414e-05,
+      "loss": 1.1327,
+      "step": 75575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4155149947282698e-05,
+      "loss": 1.2005,
+      "step": 75580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4154462578305843e-05,
+      "loss": 1.0705,
+      "step": 75585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4153775185604773e-05,
+      "loss": 1.2741,
+      "step": 75590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4153087769183409e-05,
+      "loss": 1.211,
+      "step": 75595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4152400329045683e-05,
+      "loss": 1.124,
+      "step": 75600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4151712865195517e-05,
+      "loss": 0.9182,
+      "step": 75605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4151025377636838e-05,
+      "loss": 1.1282,
+      "step": 75610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.415033786637357e-05,
+      "loss": 1.2189,
+      "step": 75615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4149650331409641e-05,
+      "loss": 1.1524,
+      "step": 75620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.414896277274898e-05,
+      "loss": 1.0073,
+      "step": 75625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4148275190395505e-05,
+      "loss": 1.2835,
+      "step": 75630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.414758758435315e-05,
+      "loss": 1.0162,
+      "step": 75635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.414689995462584e-05,
+      "loss": 1.2108,
+      "step": 75640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.41462123012175e-05,
+      "loss": 1.51,
+      "step": 75645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4145524624132056e-05,
+      "loss": 1.5581,
+      "step": 75650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4144836923373439e-05,
+      "loss": 1.1248,
+      "step": 75655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4144149198945574e-05,
+      "loss": 1.3067,
+      "step": 75660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4143461450852388e-05,
+      "loss": 1.0959,
+      "step": 75665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4142773679097805e-05,
+      "loss": 1.3506,
+      "step": 75670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.414208588368576e-05,
+      "loss": 1.1594,
+      "step": 75675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4141398064620175e-05,
+      "loss": 1.3445,
+      "step": 75680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.414071022190498e-05,
+      "loss": 1.0899,
+      "step": 75685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4140022355544103e-05,
+      "loss": 1.1662,
+      "step": 75690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4139334465541473e-05,
+      "loss": 1.216,
+      "step": 75695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4138646551901015e-05,
+      "loss": 1.4557,
+      "step": 75700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4137958614626659e-05,
+      "loss": 1.1275,
+      "step": 75705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4137270653722333e-05,
+      "loss": 1.3081,
+      "step": 75710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.413658266919197e-05,
+      "loss": 1.1107,
+      "step": 75715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4135894661039491e-05,
+      "loss": 1.2274,
+      "step": 75720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.413520662926883e-05,
+      "loss": 1.1522,
+      "step": 75725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4134518573883918e-05,
+      "loss": 1.0569,
+      "step": 75730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4133830494888675e-05,
+      "loss": 1.269,
+      "step": 75735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4133142392287043e-05,
+      "loss": 1.2039,
+      "step": 75740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.413245426608294e-05,
+      "loss": 1.2206,
+      "step": 75745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4131766116280301e-05,
+      "loss": 1.373,
+      "step": 75750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4131077942883058e-05,
+      "loss": 1.4752,
+      "step": 75755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4130389745895137e-05,
+      "loss": 1.2755,
+      "step": 75760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.412970152532047e-05,
+      "loss": 0.9759,
+      "step": 75765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4129013281162983e-05,
+      "loss": 1.063,
+      "step": 75770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4128325013426612e-05,
+      "loss": 1.0235,
+      "step": 75775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4127636722115285e-05,
+      "loss": 1.3295,
+      "step": 75780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4126948407232929e-05,
+      "loss": 1.2244,
+      "step": 75785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4126260068783481e-05,
+      "loss": 1.3865,
+      "step": 75790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4125571706770869e-05,
+      "loss": 1.0409,
+      "step": 75795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4124883321199023e-05,
+      "loss": 1.1694,
+      "step": 75800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4124194912071874e-05,
+      "loss": 1.2407,
+      "step": 75805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4123506479393354e-05,
+      "loss": 1.4073,
+      "step": 75810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4122818023167394e-05,
+      "loss": 1.4254,
+      "step": 75815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4122129543397927e-05,
+      "loss": 1.0675,
+      "step": 75820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.412144104008888e-05,
+      "loss": 1.0355,
+      "step": 75825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4120752513244189e-05,
+      "loss": 1.3663,
+      "step": 75830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4120063962867789e-05,
+      "loss": 1.2285,
+      "step": 75835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4119375388963603e-05,
+      "loss": 1.2961,
+      "step": 75840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4118686791535571e-05,
+      "loss": 1.1504,
+      "step": 75845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4117998170587618e-05,
+      "loss": 1.2103,
+      "step": 75850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4117309526123684e-05,
+      "loss": 1.2838,
+      "step": 75855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4116620858147698e-05,
+      "loss": 1.3463,
+      "step": 75860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.411593216666359e-05,
+      "loss": 1.2417,
+      "step": 75865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4115243451675297e-05,
+      "loss": 1.2715,
+      "step": 75870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4114554713186754e-05,
+      "loss": 1.2443,
+      "step": 75875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4113865951201886e-05,
+      "loss": 1.417,
+      "step": 75880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4113177165724635e-05,
+      "loss": 1.0897,
+      "step": 75885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4112488356758927e-05,
+      "loss": 1.4757,
+      "step": 75890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4111799524308698e-05,
+      "loss": 1.0901,
+      "step": 75895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4111110668377883e-05,
+      "loss": 0.9964,
+      "step": 75900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4110421788970416e-05,
+      "loss": 1.2823,
+      "step": 75905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4109732886090228e-05,
+      "loss": 1.3495,
+      "step": 75910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4109043959741258e-05,
+      "loss": 1.1295,
+      "step": 75915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4108355009927435e-05,
+      "loss": 1.1346,
+      "step": 75920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4107666036652694e-05,
+      "loss": 1.2993,
+      "step": 75925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4106977039920972e-05,
+      "loss": 1.1256,
+      "step": 75930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4106288019736201e-05,
+      "loss": 1.2286,
+      "step": 75935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4105598976102317e-05,
+      "loss": 1.5273,
+      "step": 75940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4104909909023254e-05,
+      "loss": 1.2219,
+      "step": 75945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4104220818502952e-05,
+      "loss": 1.25,
+      "step": 75950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4103531704545337e-05,
+      "loss": 1.0932,
+      "step": 75955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4102842567154349e-05,
+      "loss": 1.0543,
+      "step": 75960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4102153406333928e-05,
+      "loss": 1.2636,
+      "step": 75965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4101464222087997e-05,
+      "loss": 1.5345,
+      "step": 75970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4100775014420504e-05,
+      "loss": 1.2536,
+      "step": 75975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.410008578333538e-05,
+      "loss": 1.1458,
+      "step": 75980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4099396528836562e-05,
+      "loss": 1.2251,
+      "step": 75985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4098707250927983e-05,
+      "loss": 1.0943,
+      "step": 75990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.409801794961358e-05,
+      "loss": 1.4062,
+      "step": 75995
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4097328624897293e-05,
+      "loss": 1.2232,
+      "step": 76000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4096639276783055e-05,
+      "loss": 1.1798,
+      "step": 76005
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4095949905274805e-05,
+      "loss": 0.9829,
+      "step": 76010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4095260510376471e-05,
+      "loss": 1.0002,
+      "step": 76015
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4094571092092007e-05,
+      "loss": 1.0831,
+      "step": 76020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4093881650425333e-05,
+      "loss": 1.0387,
+      "step": 76025
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4093192185380395e-05,
+      "loss": 1.4634,
+      "step": 76030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4092502696961128e-05,
+      "loss": 1.1834,
+      "step": 76035
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4091813185171468e-05,
+      "loss": 1.1711,
+      "step": 76040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4091123650015356e-05,
+      "loss": 1.126,
+      "step": 76045
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4090434091496725e-05,
+      "loss": 1.2244,
+      "step": 76050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4089744509619516e-05,
+      "loss": 1.0694,
+      "step": 76055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4089054904387669e-05,
+      "loss": 1.4422,
+      "step": 76060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4088365275805117e-05,
+      "loss": 1.2345,
+      "step": 76065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4087675623875802e-05,
+      "loss": 1.3204,
+      "step": 76070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4086985948603658e-05,
+      "loss": 1.2096,
+      "step": 76075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4086296249992626e-05,
+      "loss": 1.209,
+      "step": 76080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4085606528046647e-05,
+      "loss": 1.0098,
+      "step": 76085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4084916782769655e-05,
+      "loss": 0.9823,
+      "step": 76090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4084227014165595e-05,
+      "loss": 1.1426,
+      "step": 76095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4083537222238397e-05,
+      "loss": 1.0712,
+      "step": 76100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4082847406992008e-05,
+      "loss": 2.3063,
+      "step": 76105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4082157568430363e-05,
+      "loss": 1.5205,
+      "step": 76110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4081467706557405e-05,
+      "loss": 1.8769,
+      "step": 76115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.408077782137707e-05,
+      "loss": 1.3763,
+      "step": 76120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4080087912893298e-05,
+      "loss": 1.1319,
+      "step": 76125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4079397981110031e-05,
+      "loss": 1.2491,
+      "step": 76130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4078708026031204e-05,
+      "loss": 1.1716,
+      "step": 76135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4078018047660764e-05,
+      "loss": 1.2084,
+      "step": 76140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4077328046002645e-05,
+      "loss": 1.2831,
+      "step": 76145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4076638021060794e-05,
+      "loss": 1.3281,
+      "step": 76150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4075947972839141e-05,
+      "loss": 1.3728,
+      "step": 76155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4075257901341637e-05,
+      "loss": 1.2425,
+      "step": 76160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4074567806572218e-05,
+      "loss": 1.1739,
+      "step": 76165
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4073877688534822e-05,
+      "loss": 1.0178,
+      "step": 76170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.40731875472334e-05,
+      "loss": 1.0902,
+      "step": 76175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4072497382671882e-05,
+      "loss": 1.3622,
+      "step": 76180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4071807194854214e-05,
+      "loss": 0.9262,
+      "step": 76185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4071116983784337e-05,
+      "loss": 0.8823,
+      "step": 76190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4070426749466193e-05,
+      "loss": 1.3342,
+      "step": 76195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.406973649190372e-05,
+      "loss": 1.4655,
+      "step": 76200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4069046211100868e-05,
+      "loss": 1.1354,
+      "step": 76205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.406835590706157e-05,
+      "loss": 1.0915,
+      "step": 76210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4067665579789772e-05,
+      "loss": 1.1121,
+      "step": 76215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4066975229289414e-05,
+      "loss": 1.3986,
+      "step": 76220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4066284855564445e-05,
+      "loss": 1.3058,
+      "step": 76225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4065594458618798e-05,
+      "loss": 1.61,
+      "step": 76230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4064904038456421e-05,
+      "loss": 1.2079,
+      "step": 76235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4064213595081256e-05,
+      "loss": 1.0828,
+      "step": 76240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4063523128497247e-05,
+      "loss": 1.2966,
+      "step": 76245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4062832638708332e-05,
+      "loss": 1.2959,
+      "step": 76250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4062142125718462e-05,
+      "loss": 0.9532,
+      "step": 76255
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.406145158953157e-05,
+      "loss": 1.505,
+      "step": 76260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.406076103015161e-05,
+      "loss": 1.065,
+      "step": 76265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4060070447582518e-05,
+      "loss": 1.3759,
+      "step": 76270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4059379841828239e-05,
+      "loss": 1.5679,
+      "step": 76275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.405868921289272e-05,
+      "loss": 1.1413,
+      "step": 76280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.40579985607799e-05,
+      "loss": 1.0897,
+      "step": 76285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4057307885493728e-05,
+      "loss": 1.0828,
+      "step": 76290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4056617187038146e-05,
+      "loss": 1.2468,
+      "step": 76295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4055926465417096e-05,
+      "loss": 1.222,
+      "step": 76300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4055235720634526e-05,
+      "loss": 1.0981,
+      "step": 76305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4054544952694378e-05,
+      "loss": 1.08,
+      "step": 76310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4053854161600597e-05,
+      "loss": 1.1826,
+      "step": 76315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.405316334735713e-05,
+      "loss": 1.2457,
+      "step": 76320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.405247250996792e-05,
+      "loss": 1.1517,
+      "step": 76325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4051781649436911e-05,
+      "loss": 1.0212,
+      "step": 76330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4051090765768052e-05,
+      "loss": 1.3083,
+      "step": 76335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4050399858965284e-05,
+      "loss": 0.9489,
+      "step": 76340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4049708929032552e-05,
+      "loss": 1.0387,
+      "step": 76345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4049017975973809e-05,
+      "loss": 1.3618,
+      "step": 76350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4048326999792991e-05,
+      "loss": 1.3525,
+      "step": 76355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4047636000494053e-05,
+      "loss": 1.1394,
+      "step": 76360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4046944978080934e-05,
+      "loss": 1.1174,
+      "step": 76365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.404625393255758e-05,
+      "loss": 1.3221,
+      "step": 76370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4045562863927944e-05,
+      "loss": 1.1757,
+      "step": 76375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4044871772195964e-05,
+      "loss": 1.2911,
+      "step": 76380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4044180657365594e-05,
+      "loss": 1.5673,
+      "step": 76385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4043489519440777e-05,
+      "loss": 0.9524,
+      "step": 76390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4042798358425458e-05,
+      "loss": 1.1612,
+      "step": 76395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4042107174323587e-05,
+      "loss": 1.0457,
+      "step": 76400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4041415967139109e-05,
+      "loss": 1.1003,
+      "step": 76405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4040724736875975e-05,
+      "loss": 1.1959,
+      "step": 76410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4040033483538128e-05,
+      "loss": 0.9803,
+      "step": 76415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4039342207129516e-05,
+      "loss": 1.3364,
+      "step": 76420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4038650907654087e-05,
+      "loss": 1.2909,
+      "step": 76425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4037959585115792e-05,
+      "loss": 1.3474,
+      "step": 76430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4037268239518574e-05,
+      "loss": 1.4976,
+      "step": 76435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4036576870866382e-05,
+      "loss": 1.3731,
+      "step": 76440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4035885479163167e-05,
+      "loss": 1.4512,
+      "step": 76445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4035194064412872e-05,
+      "loss": 1.4008,
+      "step": 76450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4034502626619453e-05,
+      "loss": 1.1499,
+      "step": 76455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.403381116578685e-05,
+      "loss": 1.4638,
+      "step": 76460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4033119681919019e-05,
+      "loss": 1.2211,
+      "step": 76465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4032428175019904e-05,
+      "loss": 1.3418,
+      "step": 76470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4031736645093452e-05,
+      "loss": 1.1878,
+      "step": 76475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4031045092143622e-05,
+      "loss": 1.3476,
+      "step": 76480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4030353516174348e-05,
+      "loss": 1.2676,
+      "step": 76485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4029661917189593e-05,
+      "loss": 1.0781,
+      "step": 76490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4028970295193303e-05,
+      "loss": 1.0383,
+      "step": 76495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4028278650189421e-05,
+      "loss": 1.2089,
+      "step": 76500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4027586982181907e-05,
+      "loss": 1.6871,
+      "step": 76505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.40268952911747e-05,
+      "loss": 1.14,
+      "step": 76510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4026203577171755e-05,
+      "loss": 1.1864,
+      "step": 76515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4025511840177024e-05,
+      "loss": 1.5531,
+      "step": 76520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4024820080194457e-05,
+      "loss": 1.1251,
+      "step": 76525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4024128297228e-05,
+      "loss": 1.5609,
+      "step": 76530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4023436491281605e-05,
+      "loss": 1.0701,
+      "step": 76535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4022744662359227e-05,
+      "loss": 1.2903,
+      "step": 76540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.402205281046481e-05,
+      "loss": 1.1232,
+      "step": 76545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.402136093560231e-05,
+      "loss": 1.2469,
+      "step": 76550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4020669037775673e-05,
+      "loss": 1.1,
+      "step": 76555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4019977116988858e-05,
+      "loss": 1.3043,
+      "step": 76560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.401928517324581e-05,
+      "loss": 1.4037,
+      "step": 76565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4018593206550484e-05,
+      "loss": 1.0793,
+      "step": 76570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4017901216906826e-05,
+      "loss": 1.2922,
+      "step": 76575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4017209204318789e-05,
+      "loss": 1.0585,
+      "step": 76580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4016517168790331e-05,
+      "loss": 1.4356,
+      "step": 76585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4015825110325398e-05,
+      "loss": 1.4936,
+      "step": 76590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4015133028927943e-05,
+      "loss": 1.1536,
+      "step": 76595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.401444092460192e-05,
+      "loss": 1.0686,
+      "step": 76600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.401374879735128e-05,
+      "loss": 1.3245,
+      "step": 76605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4013056647179978e-05,
+      "loss": 1.0149,
+      "step": 76610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.401236447409196e-05,
+      "loss": 1.1903,
+      "step": 76615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4011672278091184e-05,
+      "loss": 1.4879,
+      "step": 76620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4010980059181603e-05,
+      "loss": 1.0562,
+      "step": 76625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4010287817367168e-05,
+      "loss": 1.2267,
+      "step": 76630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.400959555265183e-05,
+      "loss": 1.0536,
+      "step": 76635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4008903265039549e-05,
+      "loss": 1.2865,
+      "step": 76640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.400821095453427e-05,
+      "loss": 1.2291,
+      "step": 76645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4007518621139955e-05,
+      "loss": 1.2499,
+      "step": 76650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.400682626486055e-05,
+      "loss": 1.3045,
+      "step": 76655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4006133885700011e-05,
+      "loss": 1.1013,
+      "step": 76660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4005441483662296e-05,
+      "loss": 1.2189,
+      "step": 76665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4004749058751356e-05,
+      "loss": 1.4592,
+      "step": 76670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4004056610971141e-05,
+      "loss": 1.2451,
+      "step": 76675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4003364140325614e-05,
+      "loss": 1.3283,
+      "step": 76680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4002671646818722e-05,
+      "loss": 1.6623,
+      "step": 76685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4001979130454424e-05,
+      "loss": 1.3366,
+      "step": 76690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4001286591236669e-05,
+      "loss": 1.4072,
+      "step": 76695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4000594029169417e-05,
+      "loss": 1.3793,
+      "step": 76700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3999901444256625e-05,
+      "loss": 1.5826,
+      "step": 76705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3999208836502239e-05,
+      "loss": 1.1791,
+      "step": 76710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3998516205910222e-05,
+      "loss": 1.1408,
+      "step": 76715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3997823552484526e-05,
+      "loss": 1.3257,
+      "step": 76720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3997130876229108e-05,
+      "loss": 1.132,
+      "step": 76725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3996438177147923e-05,
+      "loss": 1.1795,
+      "step": 76730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3995745455244927e-05,
+      "loss": 1.3258,
+      "step": 76735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3995052710524074e-05,
+      "loss": 1.6494,
+      "step": 76740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3994359942989323e-05,
+      "loss": 1.5844,
+      "step": 76745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3993667152644625e-05,
+      "loss": 1.1619,
+      "step": 76750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3992974339493941e-05,
+      "loss": 1.0943,
+      "step": 76755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3992281503541227e-05,
+      "loss": 1.2164,
+      "step": 76760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.399158864479044e-05,
+      "loss": 1.5207,
+      "step": 76765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.399089576324553e-05,
+      "loss": 1.2086,
+      "step": 76770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3990202858910464e-05,
+      "loss": 1.1806,
+      "step": 76775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3989509931789188e-05,
+      "loss": 1.0382,
+      "step": 76780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.398881698188567e-05,
+      "loss": 1.1211,
+      "step": 76785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3988124009203858e-05,
+      "loss": 1.3359,
+      "step": 76790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3987431013747716e-05,
+      "loss": 1.3055,
+      "step": 76795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3986737995521193e-05,
+      "loss": 1.4898,
+      "step": 76800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3986044954528257e-05,
+      "loss": 1.2251,
+      "step": 76805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.398535189077286e-05,
+      "loss": 1.2823,
+      "step": 76810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3984658804258958e-05,
+      "loss": 1.1977,
+      "step": 76815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.398396569499051e-05,
+      "loss": 1.1487,
+      "step": 76820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3983272562971479e-05,
+      "loss": 1.1933,
+      "step": 76825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3982579408205816e-05,
+      "loss": 1.2758,
+      "step": 76830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3981886230697486e-05,
+      "loss": 1.5393,
+      "step": 76835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3981193030450442e-05,
+      "loss": 1.422,
+      "step": 76840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3980499807468641e-05,
+      "loss": 1.2944,
+      "step": 76845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3979806561756051e-05,
+      "loss": 1.3645,
+      "step": 76850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3979113293316622e-05,
+      "loss": 1.2367,
+      "step": 76855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3978420002154316e-05,
+      "loss": 1.0377,
+      "step": 76860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3977726688273092e-05,
+      "loss": 1.434,
+      "step": 76865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.397703335167691e-05,
+      "loss": 0.9553,
+      "step": 76870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3976339992369727e-05,
+      "loss": 1.0006,
+      "step": 76875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3975646610355507e-05,
+      "loss": 1.2035,
+      "step": 76880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3974953205638202e-05,
+      "loss": 1.0372,
+      "step": 76885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3974259778221782e-05,
+      "loss": 1.3811,
+      "step": 76890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3973566328110196e-05,
+      "loss": 1.2509,
+      "step": 76895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3972872855307412e-05,
+      "loss": 1.115,
+      "step": 76900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3972179359817384e-05,
+      "loss": 1.2706,
+      "step": 76905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3971485841644077e-05,
+      "loss": 1.2357,
+      "step": 76910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3970792300791454e-05,
+      "loss": 1.0311,
+      "step": 76915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3970098737263464e-05,
+      "loss": 1.3417,
+      "step": 76920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.396940515106408e-05,
+      "loss": 1.1078,
+      "step": 76925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3968711542197256e-05,
+      "loss": 1.1185,
+      "step": 76930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3968017910666954e-05,
+      "loss": 1.454,
+      "step": 76935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3967324256477134e-05,
+      "loss": 1.2727,
+      "step": 76940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3966630579631764e-05,
+      "loss": 1.2124,
+      "step": 76945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3965936880134795e-05,
+      "loss": 1.3675,
+      "step": 76950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3965243157990195e-05,
+      "loss": 1.3821,
+      "step": 76955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3964549413201924e-05,
+      "loss": 1.181,
+      "step": 76960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3963855645773944e-05,
+      "loss": 1.2223,
+      "step": 76965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3963161855710214e-05,
+      "loss": 1.3328,
+      "step": 76970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3962468043014701e-05,
+      "loss": 1.096,
+      "step": 76975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3961774207691364e-05,
+      "loss": 1.4404,
+      "step": 76980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3961080349744162e-05,
+      "loss": 1.3295,
+      "step": 76985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3960386469177063e-05,
+      "loss": 1.1187,
+      "step": 76990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.395969256599403e-05,
+      "loss": 1.0941,
+      "step": 76995
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3958998640199017e-05,
+      "loss": 1.0325,
+      "step": 77000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3958304691795998e-05,
+      "loss": 1.1676,
+      "step": 77005
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.395761072078893e-05,
+      "loss": 1.5006,
+      "step": 77010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.395691672718177e-05,
+      "loss": 0.9485,
+      "step": 77015
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3956222710978496e-05,
+      "loss": 1.5043,
+      "step": 77020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3955528672183053e-05,
+      "loss": 1.3317,
+      "step": 77025
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.395483461079942e-05,
+      "loss": 1.5026,
+      "step": 77030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3954140526831552e-05,
+      "loss": 1.4589,
+      "step": 77035
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3953446420283417e-05,
+      "loss": 1.4091,
+      "step": 77040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3952752291158973e-05,
+      "loss": 1.25,
+      "step": 77045
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.395205813946219e-05,
+      "loss": 1.736,
+      "step": 77050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3951363965197029e-05,
+      "loss": 1.0874,
+      "step": 77055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3950669768367454e-05,
+      "loss": 1.3081,
+      "step": 77060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.394997554897743e-05,
+      "loss": 1.1153,
+      "step": 77065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3949281307030922e-05,
+      "loss": 1.4158,
+      "step": 77070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3948587042531894e-05,
+      "loss": 1.074,
+      "step": 77075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3947892755484309e-05,
+      "loss": 1.4009,
+      "step": 77080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3947198445892133e-05,
+      "loss": 1.2478,
+      "step": 77085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3946504113759332e-05,
+      "loss": 1.202,
+      "step": 77090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3945809759089866e-05,
+      "loss": 1.3533,
+      "step": 77095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.394511538188771e-05,
+      "loss": 1.6416,
+      "step": 77100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3944420982156816e-05,
+      "loss": 1.2711,
+      "step": 77105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3943726559901163e-05,
+      "loss": 1.1144,
+      "step": 77110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3943032115124711e-05,
+      "loss": 1.5343,
+      "step": 77115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3942337647831418e-05,
+      "loss": 0.9202,
+      "step": 77120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3941643158025265e-05,
+      "loss": 1.1605,
+      "step": 77125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3940948645710202e-05,
+      "loss": 1.4154,
+      "step": 77130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3940254110890206e-05,
+      "loss": 1.2205,
+      "step": 77135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3939559553569239e-05,
+      "loss": 0.9387,
+      "step": 77140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3938864973751268e-05,
+      "loss": 0.9707,
+      "step": 77145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3938170371440262e-05,
+      "loss": 1.3066,
+      "step": 77150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3937475746640181e-05,
+      "loss": 1.1246,
+      "step": 77155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3936781099355e-05,
+      "loss": 1.2146,
+      "step": 77160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3936086429588677e-05,
+      "loss": 1.1716,
+      "step": 77165
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3935391737345185e-05,
+      "loss": 1.1348,
+      "step": 77170
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3934697022628491e-05,
+      "loss": 1.1578,
+      "step": 77175
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.393400228544256e-05,
+      "loss": 0.9223,
+      "step": 77180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.393330752579136e-05,
+      "loss": 1.1395,
+      "step": 77185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3932612743678857e-05,
+      "loss": 1.2206,
+      "step": 77190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3931917939109023e-05,
+      "loss": 1.4427,
+      "step": 77195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.393122311208582e-05,
+      "loss": 1.0382,
+      "step": 77200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.393052826261322e-05,
+      "loss": 1.2769,
+      "step": 77205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.392983339069519e-05,
+      "loss": 1.5094,
+      "step": 77210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3929138496335695e-05,
+      "loss": 1.0978,
+      "step": 77215
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.392844357953871e-05,
+      "loss": 1.1663,
+      "step": 77220
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3927748640308197e-05,
+      "loss": 1.1353,
+      "step": 77225
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3927053678648128e-05,
+      "loss": 1.3038,
+      "step": 77230
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3926358694562468e-05,
+      "loss": 1.3611,
+      "step": 77235
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3925663688055188e-05,
+      "loss": 1.1172,
+      "step": 77240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.392496865913026e-05,
+      "loss": 1.3612,
+      "step": 77245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3924273607791647e-05,
+      "loss": 1.7158,
+      "step": 77250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3923578534043323e-05,
+      "loss": 1.206,
+      "step": 77255
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3922883437889254e-05,
+      "loss": 1.2381,
+      "step": 77260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3922188319333412e-05,
+      "loss": 1.3054,
+      "step": 77265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3921493178379763e-05,
+      "loss": 1.5326,
+      "step": 77270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3920798015032282e-05,
+      "loss": 1.245,
+      "step": 77275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3920102829294932e-05,
+      "loss": 1.2476,
+      "step": 77280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3919407621171688e-05,
+      "loss": 1.4012,
+      "step": 77285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.391871239066652e-05,
+      "loss": 1.3454,
+      "step": 77290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3918017137783393e-05,
+      "loss": 1.2068,
+      "step": 77295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3917321862526283e-05,
+      "loss": 1.3996,
+      "step": 77300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3916626564899157e-05,
+      "loss": 0.9315,
+      "step": 77305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3915931244905988e-05,
+      "loss": 0.9545,
+      "step": 77310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3915235902550745e-05,
+      "loss": 1.6412,
+      "step": 77315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3914540537837398e-05,
+      "loss": 1.3768,
+      "step": 77320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3913845150769922e-05,
+      "loss": 1.3441,
+      "step": 77325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3913149741352283e-05,
+      "loss": 1.181,
+      "step": 77330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3912454309588457e-05,
+      "loss": 1.0135,
+      "step": 77335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3911758855482407e-05,
+      "loss": 1.2064,
+      "step": 77340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3911063379038115e-05,
+      "loss": 1.3458,
+      "step": 77345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3910367880259547e-05,
+      "loss": 1.3217,
+      "step": 77350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3909672359150672e-05,
+      "loss": 1.1599,
+      "step": 77355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3908976815715467e-05,
+      "loss": 0.8688,
+      "step": 77360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3908281249957901e-05,
+      "loss": 1.0578,
+      "step": 77365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3907585661881951e-05,
+      "loss": 1.0201,
+      "step": 77370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3906890051491582e-05,
+      "loss": 1.0944,
+      "step": 77375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3906194418790768e-05,
+      "loss": 0.9511,
+      "step": 77380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3905498763783485e-05,
+      "loss": 1.0942,
+      "step": 77385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3904803086473703e-05,
+      "loss": 1.1388,
+      "step": 77390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3904107386865396e-05,
+      "loss": 1.4233,
+      "step": 77395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3903411664962532e-05,
+      "loss": 1.2923,
+      "step": 77400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3902715920769094e-05,
+      "loss": 1.3022,
+      "step": 77405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3902020154289047e-05,
+      "loss": 1.3694,
+      "step": 77410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3901324365526363e-05,
+      "loss": 1.2933,
+      "step": 77415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3900628554485023e-05,
+      "loss": 1.4032,
+      "step": 77420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.389993272116899e-05,
+      "loss": 1.1852,
+      "step": 77425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3899236865582251e-05,
+      "loss": 1.4881,
+      "step": 77430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3898540987728766e-05,
+      "loss": 1.101,
+      "step": 77435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3897845087612518e-05,
+      "loss": 0.9445,
+      "step": 77440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3897149165237478e-05,
+      "loss": 1.1796,
+      "step": 77445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.389645322060762e-05,
+      "loss": 1.2529,
+      "step": 77450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3895757253726921e-05,
+      "loss": 1.2573,
+      "step": 77455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3895061264599347e-05,
+      "loss": 1.0721,
+      "step": 77460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3894365253228883e-05,
+      "loss": 1.1391,
+      "step": 77465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3893669219619496e-05,
+      "loss": 0.964,
+      "step": 77470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3892973163775166e-05,
+      "loss": 1.3035,
+      "step": 77475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3892277085699864e-05,
+      "loss": 1.1323,
+      "step": 77480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3891580985397566e-05,
+      "loss": 1.2308,
+      "step": 77485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3890884862872249e-05,
+      "loss": 0.9248,
+      "step": 77490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3890188718127885e-05,
+      "loss": 1.2032,
+      "step": 77495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3889492551168453e-05,
+      "loss": 1.4605,
+      "step": 77500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3888796361997923e-05,
+      "loss": 1.0509,
+      "step": 77505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3888100150620278e-05,
+      "loss": 1.3056,
+      "step": 77510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3887403917039489e-05,
+      "loss": 1.2544,
+      "step": 77515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3886707661259533e-05,
+      "loss": 1.1536,
+      "step": 77520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3886011383284383e-05,
+      "loss": 0.9256,
+      "step": 77525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.388531508311802e-05,
+      "loss": 1.1627,
+      "step": 77530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3884618760764419e-05,
+      "loss": 1.4291,
+      "step": 77535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3883922416227553e-05,
+      "loss": 1.4144,
+      "step": 77540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3883226049511402e-05,
+      "loss": 0.8993,
+      "step": 77545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3882529660619946e-05,
+      "loss": 0.9598,
+      "step": 77550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.388183324955715e-05,
+      "loss": 1.0622,
+      "step": 77555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3881136816327005e-05,
+      "loss": 0.9183,
+      "step": 77560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3880440360933476e-05,
+      "loss": 1.0133,
+      "step": 77565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3879743883380547e-05,
+      "loss": 1.357,
+      "step": 77570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3879047383672195e-05,
+      "loss": 1.1926,
+      "step": 77575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3878350861812396e-05,
+      "loss": 1.5227,
+      "step": 77580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3877654317805127e-05,
+      "loss": 1.2179,
+      "step": 77585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3876957751654368e-05,
+      "loss": 1.1658,
+      "step": 77590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3876261163364091e-05,
+      "loss": 0.8851,
+      "step": 77595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3875564552938281e-05,
+      "loss": 1.547,
+      "step": 77600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3874867920380913e-05,
+      "loss": 1.1488,
+      "step": 77605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3874171265695964e-05,
+      "loss": 1.2061,
+      "step": 77610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3873474588887414e-05,
+      "loss": 0.9456,
+      "step": 77615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.387277788995924e-05,
+      "loss": 1.7601,
+      "step": 77620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.387208116891542e-05,
+      "loss": 1.068,
+      "step": 77625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3871384425759937e-05,
+      "loss": 1.2864,
+      "step": 77630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3870687660496765e-05,
+      "loss": 1.3686,
+      "step": 77635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3869990873129884e-05,
+      "loss": 1.0755,
+      "step": 77640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3869294063663272e-05,
+      "loss": 1.2622,
+      "step": 77645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3868597232100914e-05,
+      "loss": 1.1985,
+      "step": 77650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3867900378446785e-05,
+      "loss": 1.1921,
+      "step": 77655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.386720350270486e-05,
+      "loss": 1.2065,
+      "step": 77660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3866506604879128e-05,
+      "loss": 1.2076,
+      "step": 77665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.386580968497356e-05,
+      "loss": 1.4506,
+      "step": 77670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3865112742992141e-05,
+      "loss": 1.1327,
+      "step": 77675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3864415778938849e-05,
+      "loss": 1.2255,
+      "step": 77680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3863718792817667e-05,
+      "loss": 1.4575,
+      "step": 77685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3863021784632569e-05,
+      "loss": 1.2466,
+      "step": 77690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.386232475438754e-05,
+      "loss": 1.0898,
+      "step": 77695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.386162770208656e-05,
+      "loss": 1.2859,
+      "step": 77700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3860930627733608e-05,
+      "loss": 1.1864,
+      "step": 77705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3860233531332665e-05,
+      "loss": 1.524,
+      "step": 77710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3859536412887713e-05,
+      "loss": 0.9262,
+      "step": 77715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3858839272402733e-05,
+      "loss": 1.1766,
+      "step": 77720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3858142109881705e-05,
+      "loss": 1.1784,
+      "step": 77725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3857444925328606e-05,
+      "loss": 1.0511,
+      "step": 77730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.385674771874743e-05,
+      "loss": 1.1465,
+      "step": 77735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3856050490142143e-05,
+      "loss": 1.2437,
+      "step": 77740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.385535323951674e-05,
+      "loss": 1.2332,
+      "step": 77745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.385465596687519e-05,
+      "loss": 1.3825,
+      "step": 77750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3853958672221483e-05,
+      "loss": 1.3064,
+      "step": 77755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3853261355559605e-05,
+      "loss": 1.3528,
+      "step": 77760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3852564016893527e-05,
+      "loss": 1.2402,
+      "step": 77765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3851866656227238e-05,
+      "loss": 1.0531,
+      "step": 77770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3851169273564715e-05,
+      "loss": 1.3215,
+      "step": 77775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3850471868909944e-05,
+      "loss": 1.5862,
+      "step": 77780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3849774442266913e-05,
+      "loss": 1.1562,
+      "step": 77785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3849076993639597e-05,
+      "loss": 1.3664,
+      "step": 77790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3848379523031981e-05,
+      "loss": 1.2398,
+      "step": 77795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3847682030448046e-05,
+      "loss": 1.3315,
+      "step": 77800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3846984515891779e-05,
+      "loss": 1.2592,
+      "step": 77805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3846286979367163e-05,
+      "loss": 1.3464,
+      "step": 77810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3845589420878178e-05,
+      "loss": 0.962,
+      "step": 77815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.384489184042881e-05,
+      "loss": 1.4135,
+      "step": 77820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3844194238023038e-05,
+      "loss": 1.5753,
+      "step": 77825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3843496613664853e-05,
+      "loss": 1.537,
+      "step": 77830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3842798967358231e-05,
+      "loss": 1.2264,
+      "step": 77835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3842101299107165e-05,
+      "loss": 1.0385,
+      "step": 77840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.384140360891563e-05,
+      "loss": 1.3276,
+      "step": 77845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3840705896787618e-05,
+      "loss": 1.0812,
+      "step": 77850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3840008162727108e-05,
+      "loss": 1.2566,
+      "step": 77855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3839310406738085e-05,
+      "loss": 0.857,
+      "step": 77860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3838612628824537e-05,
+      "loss": 1.5789,
+      "step": 77865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3837914828990444e-05,
+      "loss": 1.2939,
+      "step": 77870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3837217007239794e-05,
+      "loss": 1.5099,
+      "step": 77875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3836519163576569e-05,
+      "loss": 1.0824,
+      "step": 77880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.383582129800476e-05,
+      "loss": 0.9496,
+      "step": 77885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3835123410528347e-05,
+      "loss": 1.5651,
+      "step": 77890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3834425501151313e-05,
+      "loss": 1.2447,
+      "step": 77895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3833727569877651e-05,
+      "loss": 1.0347,
+      "step": 77900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3833029616711342e-05,
+      "loss": 1.3262,
+      "step": 77905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3832331641656373e-05,
+      "loss": 1.3394,
+      "step": 77910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3831633644716726e-05,
+      "loss": 1.3116,
+      "step": 77915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3830935625896393e-05,
+      "loss": 1.5073,
+      "step": 77920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3830237585199356e-05,
+      "loss": 1.1467,
+      "step": 77925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3829539522629603e-05,
+      "loss": 1.4455,
+      "step": 77930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3828841438191119e-05,
+      "loss": 1.0521,
+      "step": 77935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3828143331887888e-05,
+      "loss": 1.3803,
+      "step": 77940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3827445203723906e-05,
+      "loss": 1.2478,
+      "step": 77945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3826747053703149e-05,
+      "loss": 1.1897,
+      "step": 77950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.382604888182961e-05,
+      "loss": 1.4601,
+      "step": 77955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3825350688107274e-05,
+      "loss": 1.1523,
+      "step": 77960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3824652472540124e-05,
+      "loss": 1.3452,
+      "step": 77965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3823954235132156e-05,
+      "loss": 1.0651,
+      "step": 77970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3823255975887349e-05,
+      "loss": 1.2524,
+      "step": 77975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3822557694809698e-05,
+      "loss": 1.2214,
+      "step": 77980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3821859391903183e-05,
+      "loss": 1.2093,
+      "step": 77985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3821161067171797e-05,
+      "loss": 1.3177,
+      "step": 77990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3820462720619528e-05,
+      "loss": 1.2059,
+      "step": 77995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3819764352250358e-05,
+      "loss": 1.2052,
+      "step": 78000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.381906596206828e-05,
+      "loss": 1.0322,
+      "step": 78005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3818367550077283e-05,
+      "loss": 1.5801,
+      "step": 78010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3817669116281354e-05,
+      "loss": 1.3393,
+      "step": 78015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.381697066068448e-05,
+      "loss": 1.0411,
+      "step": 78020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.381627218329065e-05,
+      "loss": 1.2985,
+      "step": 78025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3815573684103854e-05,
+      "loss": 1.1407,
+      "step": 78030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.381487516312808e-05,
+      "loss": 1.119,
+      "step": 78035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3814176620367317e-05,
+      "loss": 1.3454,
+      "step": 78040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3813478055825552e-05,
+      "loss": 1.5237,
+      "step": 78045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3812779469506778e-05,
+      "loss": 1.3022,
+      "step": 78050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3812080861414985e-05,
+      "loss": 1.1815,
+      "step": 78055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3811382231554156e-05,
+      "loss": 1.1569,
+      "step": 78060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3810683579928288e-05,
+      "loss": 1.5777,
+      "step": 78065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3809984906541363e-05,
+      "loss": 1.2268,
+      "step": 78070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3809286211397379e-05,
+      "loss": 1.204,
+      "step": 78075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3808587494500318e-05,
+      "loss": 1.263,
+      "step": 78080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3807888755854175e-05,
+      "loss": 1.1324,
+      "step": 78085
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3807189995462939e-05,
+      "loss": 1.1838,
+      "step": 78090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3806491213330602e-05,
+      "loss": 1.508,
+      "step": 78095
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3805792409461157e-05,
+      "loss": 1.0158,
+      "step": 78100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3805093583858583e-05,
+      "loss": 1.4201,
+      "step": 78105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.380439473652688e-05,
+      "loss": 1.1612,
+      "step": 78110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3803695867470037e-05,
+      "loss": 1.2537,
+      "step": 78115
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3802996976692047e-05,
+      "loss": 1.1673,
+      "step": 78120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3802298064196898e-05,
+      "loss": 1.4226,
+      "step": 78125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3801599129988579e-05,
+      "loss": 1.3148,
+      "step": 78130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3800900174071088e-05,
+      "loss": 1.0998,
+      "step": 78135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3800201196448411e-05,
+      "loss": 1.2296,
+      "step": 78140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3799502197124541e-05,
+      "loss": 1.5818,
+      "step": 78145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3798803176103468e-05,
+      "loss": 1.2813,
+      "step": 78150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.379810413338919e-05,
+      "loss": 1.3027,
+      "step": 78155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3797405068985692e-05,
+      "loss": 1.1427,
+      "step": 78160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3796705982896968e-05,
+      "loss": 1.2124,
+      "step": 78165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3796006875127015e-05,
+      "loss": 1.3527,
+      "step": 78170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3795307745679813e-05,
+      "loss": 1.3179,
+      "step": 78175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3794608594559372e-05,
+      "loss": 1.4625,
+      "step": 78180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3793909421769666e-05,
+      "loss": 1.3285,
+      "step": 78185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3793210227314701e-05,
+      "loss": 1.0578,
+      "step": 78190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3792511011198469e-05,
+      "loss": 1.3042,
+      "step": 78195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3791811773424955e-05,
+      "loss": 1.2669,
+      "step": 78200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3791112513998158e-05,
+      "loss": 1.4912,
+      "step": 78205
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3790413232922066e-05,
+      "loss": 1.1562,
+      "step": 78210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3789713930200678e-05,
+      "loss": 1.443,
+      "step": 78215
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3789014605837987e-05,
+      "loss": 1.5026,
+      "step": 78220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3788315259837983e-05,
+      "loss": 1.0385,
+      "step": 78225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3787615892204661e-05,
+      "loss": 1.0034,
+      "step": 78230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3786916502942017e-05,
+      "loss": 1.2507,
+      "step": 78235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3786217092054042e-05,
+      "loss": 1.3027,
+      "step": 78240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3785517659544729e-05,
+      "loss": 1.0231,
+      "step": 78245
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3784818205418076e-05,
+      "loss": 1.2571,
+      "step": 78250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3784118729678076e-05,
+      "loss": 1.2782,
+      "step": 78255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3783419232328721e-05,
+      "loss": 1.3087,
+      "step": 78260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3782719713374006e-05,
+      "loss": 1.362,
+      "step": 78265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.378202017281793e-05,
+      "loss": 1.0847,
+      "step": 78270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3781320610664484e-05,
+      "loss": 1.0999,
+      "step": 78275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3780621026917662e-05,
+      "loss": 1.2121,
+      "step": 78280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3779921421581463e-05,
+      "loss": 1.3758,
+      "step": 78285
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3779221794659876e-05,
+      "loss": 1.579,
+      "step": 78290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3778522146156903e-05,
+      "loss": 1.1848,
+      "step": 78295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3777822476076537e-05,
+      "loss": 1.0958,
+      "step": 78300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.377712278442277e-05,
+      "loss": 1.1556,
+      "step": 78305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3776423071199601e-05,
+      "loss": 1.2472,
+      "step": 78310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3775723336411026e-05,
+      "loss": 1.1031,
+      "step": 78315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3775023580061038e-05,
+      "loss": 1.1939,
+      "step": 78320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3774323802153638e-05,
+      "loss": 1.2273,
+      "step": 78325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3773624002692816e-05,
+      "loss": 1.1779,
+      "step": 78330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3772924181682574e-05,
+      "loss": 1.1489,
+      "step": 78335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3772224339126905e-05,
+      "loss": 1.1941,
+      "step": 78340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3771524475029805e-05,
+      "loss": 1.2745,
+      "step": 78345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3770824589395274e-05,
+      "loss": 1.1822,
+      "step": 78350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3770124682227303e-05,
+      "loss": 1.0359,
+      "step": 78355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3769424753529895e-05,
+      "loss": 1.3076,
+      "step": 78360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3768724803307043e-05,
+      "loss": 1.3287,
+      "step": 78365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3768024831562748e-05,
+      "loss": 1.2668,
+      "step": 78370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3767324838300997e-05,
+      "loss": 1.5421,
+      "step": 78375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3766624823525803e-05,
+      "loss": 1.1392,
+      "step": 78380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3765924787241151e-05,
+      "loss": 1.3306,
+      "step": 78385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3765224729451048e-05,
+      "loss": 1.0368,
+      "step": 78390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.376452465015948e-05,
+      "loss": 1.2989,
+      "step": 78395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3763824549370454e-05,
+      "loss": 1.1769,
+      "step": 78400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3763124427087969e-05,
+      "loss": 1.1403,
+      "step": 78405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3762424283316013e-05,
+      "loss": 1.2035,
+      "step": 78410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3761724118058596e-05,
+      "loss": 1.3508,
+      "step": 78415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.376102393131971e-05,
+      "loss": 1.3524,
+      "step": 78420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3760323723103354e-05,
+      "loss": 1.2927,
+      "step": 78425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3759623493413528e-05,
+      "loss": 1.4205,
+      "step": 78430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3758923242254228e-05,
+      "loss": 1.2484,
+      "step": 78435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3758222969629454e-05,
+      "loss": 1.2494,
+      "step": 78440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3757522675543208e-05,
+      "loss": 1.2271,
+      "step": 78445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3756822359999484e-05,
+      "loss": 1.1301,
+      "step": 78450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3756122023002285e-05,
+      "loss": 1.2358,
+      "step": 78455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3755421664555609e-05,
+      "loss": 1.1723,
+      "step": 78460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3754721284663455e-05,
+      "loss": 1.0283,
+      "step": 78465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3754020883329824e-05,
+      "loss": 1.1649,
+      "step": 78470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3753320460558714e-05,
+      "loss": 1.4255,
+      "step": 78475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3752620016354123e-05,
+      "loss": 1.171,
+      "step": 78480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3751919550720058e-05,
+      "loss": 1.2461,
+      "step": 78485
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.375121906366051e-05,
+      "loss": 1.1035,
+      "step": 78490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.375051855517949e-05,
+      "loss": 1.174,
+      "step": 78495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3749818025280985e-05,
+      "loss": 1.4173,
+      "step": 78500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3749117473969005e-05,
+      "loss": 1.3143,
+      "step": 78505
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3748416901247548e-05,
+      "loss": 1.4282,
+      "step": 78510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3747716307120614e-05,
+      "loss": 1.3595,
+      "step": 78515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3747015691592203e-05,
+      "loss": 1.3069,
+      "step": 78520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3746315054666317e-05,
+      "loss": 1.2349,
+      "step": 78525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3745614396346958e-05,
+      "loss": 1.0719,
+      "step": 78530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3744913716638127e-05,
+      "loss": 1.3139,
+      "step": 78535
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3744213015543824e-05,
+      "loss": 1.4852,
+      "step": 78540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3743512293068051e-05,
+      "loss": 1.1087,
+      "step": 78545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.374281154921481e-05,
+      "loss": 1.2437,
+      "step": 78550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3742110783988101e-05,
+      "loss": 1.2109,
+      "step": 78555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3741409997391926e-05,
+      "loss": 1.674,
+      "step": 78560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3740709189430287e-05,
+      "loss": 1.2569,
+      "step": 78565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3740008360107188e-05,
+      "loss": 1.2262,
+      "step": 78570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3739307509426631e-05,
+      "loss": 1.3338,
+      "step": 78575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3738606637392615e-05,
+      "loss": 0.9608,
+      "step": 78580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3737905744009143e-05,
+      "loss": 1.3589,
+      "step": 78585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3737204829280221e-05,
+      "loss": 1.2449,
+      "step": 78590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.373650389320985e-05,
+      "loss": 1.3297,
+      "step": 78595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.373580293580203e-05,
+      "loss": 1.1563,
+      "step": 78600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3735101957060768e-05,
+      "loss": 1.1943,
+      "step": 78605
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3734400956990064e-05,
+      "loss": 1.4106,
+      "step": 78610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3733699935593922e-05,
+      "loss": 1.4266,
+      "step": 78615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3732998892876344e-05,
+      "loss": 1.7608,
+      "step": 78620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3732297828841335e-05,
+      "loss": 1.2755,
+      "step": 78625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3731596743492898e-05,
+      "loss": 1.3891,
+      "step": 78630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3730895636835037e-05,
+      "loss": 1.1391,
+      "step": 78635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3730194508871756e-05,
+      "loss": 1.2135,
+      "step": 78640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3729493359607057e-05,
+      "loss": 1.1257,
+      "step": 78645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3728792189044946e-05,
+      "loss": 1.1749,
+      "step": 78650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3728090997189426e-05,
+      "loss": 1.0999,
+      "step": 78655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.37273897840445e-05,
+      "loss": 1.3581,
+      "step": 78660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3726688549614176e-05,
+      "loss": 0.9412,
+      "step": 78665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3725987293902453e-05,
+      "loss": 1.1864,
+      "step": 78670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.372528601691334e-05,
+      "loss": 0.9209,
+      "step": 78675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.372458471865084e-05,
+      "loss": 1.1965,
+      "step": 78680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3723883399118958e-05,
+      "loss": 0.8047,
+      "step": 78685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3723182058321696e-05,
+      "loss": 1.7431,
+      "step": 78690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3722480696263068e-05,
+      "loss": 1.3268,
+      "step": 78695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.372177931294707e-05,
+      "loss": 1.0522,
+      "step": 78700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3721077908377706e-05,
+      "loss": 1.0853,
+      "step": 78705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3720376482558994e-05,
+      "loss": 1.0264,
+      "step": 78710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3719675035494926e-05,
+      "loss": 1.1931,
+      "step": 78715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3718973567189516e-05,
+      "loss": 1.8795,
+      "step": 78720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3718272077646762e-05,
+      "loss": 1.3688,
+      "step": 78725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3717570566870678e-05,
+      "loss": 1.1798,
+      "step": 78730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3716869034865266e-05,
+      "loss": 1.267,
+      "step": 78735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.371616748163453e-05,
+      "loss": 1.2615,
+      "step": 78740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3715465907182483e-05,
+      "loss": 1.2211,
+      "step": 78745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3714764311513125e-05,
+      "loss": 1.246,
+      "step": 78750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3714062694630465e-05,
+      "loss": 0.9788,
+      "step": 78755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3713361056538512e-05,
+      "loss": 1.1998,
+      "step": 78760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3712659397241269e-05,
+      "loss": 1.184,
+      "step": 78765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3711957716742742e-05,
+      "loss": 1.2333,
+      "step": 78770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.371125601504694e-05,
+      "loss": 1.6834,
+      "step": 78775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3710554292157872e-05,
+      "loss": 1.1701,
+      "step": 78780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.370985254807954e-05,
+      "loss": 1.0808,
+      "step": 78785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3709150782815958e-05,
+      "loss": 1.2051,
+      "step": 78790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.370844899637113e-05,
+      "loss": 1.3817,
+      "step": 78795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3707747188749064e-05,
+      "loss": 1.0085,
+      "step": 78800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3707045359953765e-05,
+      "loss": 1.2889,
+      "step": 78805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3706343509989241e-05,
+      "loss": 1.1374,
+      "step": 78810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3705641638859508e-05,
+      "loss": 1.6941,
+      "step": 78815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3704939746568565e-05,
+      "loss": 1.1449,
+      "step": 78820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3704237833120424e-05,
+      "loss": 1.2791,
+      "step": 78825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3703535898519092e-05,
+      "loss": 0.8973,
+      "step": 78830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3702833942768576e-05,
+      "loss": 1.0144,
+      "step": 78835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3702131965872893e-05,
+      "loss": 1.2763,
+      "step": 78840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3701429967836037e-05,
+      "loss": 1.4348,
+      "step": 78845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3700727948662032e-05,
+      "loss": 1.2565,
+      "step": 78850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3700025908354878e-05,
+      "loss": 1.5564,
+      "step": 78855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3699323846918583e-05,
+      "loss": 1.2024,
+      "step": 78860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3698621764357163e-05,
+      "loss": 1.3717,
+      "step": 78865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3697919660674621e-05,
+      "loss": 1.0182,
+      "step": 78870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3697217535874969e-05,
+      "loss": 1.0732,
+      "step": 78875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3696515389962217e-05,
+      "loss": 1.2145,
+      "step": 78880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3695813222940375e-05,
+      "loss": 1.3658,
+      "step": 78885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.369511103481345e-05,
+      "loss": 1.3787,
+      "step": 78890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3694408825585454e-05,
+      "loss": 1.4145,
+      "step": 78895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3693706595260397e-05,
+      "loss": 1.4962,
+      "step": 78900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3693004343842288e-05,
+      "loss": 1.1186,
+      "step": 78905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.369230207133514e-05,
+      "loss": 1.3712,
+      "step": 78910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3691599777742957e-05,
+      "loss": 1.6861,
+      "step": 78915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.369089746306976e-05,
+      "loss": 1.1364,
+      "step": 78920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3690195127319546e-05,
+      "loss": 1.1998,
+      "step": 78925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3689492770496337e-05,
+      "loss": 1.1643,
+      "step": 78930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.368879039260414e-05,
+      "loss": 1.3978,
+      "step": 78935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3688087993646964e-05,
+      "loss": 1.1219,
+      "step": 78940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3687385573628826e-05,
+      "loss": 1.5773,
+      "step": 78945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3686683132553728e-05,
+      "loss": 1.0243,
+      "step": 78950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.368598067042569e-05,
+      "loss": 1.1195,
+      "step": 78955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.368527818724872e-05,
+      "loss": 1.2319,
+      "step": 78960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3684575683026826e-05,
+      "loss": 1.1901,
+      "step": 78965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3683873157764026e-05,
+      "loss": 0.8841,
+      "step": 78970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3683170611464327e-05,
+      "loss": 1.29,
+      "step": 78975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3682468044131745e-05,
+      "loss": 1.4609,
+      "step": 78980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3681765455770288e-05,
+      "loss": 1.5288,
+      "step": 78985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.368106284638397e-05,
+      "loss": 1.079,
+      "step": 78990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3680360215976802e-05,
+      "loss": 1.1158,
+      "step": 78995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3679657564552802e-05,
+      "loss": 1.2376,
+      "step": 79000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3678954892115976e-05,
+      "loss": 1.2298,
+      "step": 79005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3678252198670339e-05,
+      "loss": 1.22,
+      "step": 79010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3677549484219904e-05,
+      "loss": 0.9726,
+      "step": 79015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3676846748768679e-05,
+      "loss": 0.9937,
+      "step": 79020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3676143992320689e-05,
+      "loss": 1.0512,
+      "step": 79025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3675441214879934e-05,
+      "loss": 1.3184,
+      "step": 79030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3674738416450435e-05,
+      "loss": 1.201,
+      "step": 79035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3674035597036203e-05,
+      "loss": 1.4494,
+      "step": 79040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3673332756641251e-05,
+      "loss": 1.2728,
+      "step": 79045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3672629895269596e-05,
+      "loss": 1.2109,
+      "step": 79050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3671927012925244e-05,
+      "loss": 1.4106,
+      "step": 79055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3671224109612217e-05,
+      "loss": 1.2307,
+      "step": 79060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3670521185334527e-05,
+      "loss": 1.1663,
+      "step": 79065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3669818240096185e-05,
+      "loss": 1.4116,
+      "step": 79070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3669115273901206e-05,
+      "loss": 0.9987,
+      "step": 79075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3668412286753605e-05,
+      "loss": 1.3291,
+      "step": 79080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.36677092786574e-05,
+      "loss": 1.1371,
+      "step": 79085
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3667006249616601e-05,
+      "loss": 1.1207,
+      "step": 79090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3666303199635223e-05,
+      "loss": 1.0272,
+      "step": 79095
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3665600128717282e-05,
+      "loss": 1.3789,
+      "step": 79100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3664897036866793e-05,
+      "loss": 1.087,
+      "step": 79105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3664193924087771e-05,
+      "loss": 1.3135,
+      "step": 79110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.366349079038423e-05,
+      "loss": 1.3652,
+      "step": 79115
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.366278763576019e-05,
+      "loss": 1.3781,
+      "step": 79120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3662084460219655e-05,
+      "loss": 0.9648,
+      "step": 79125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3661381263766655e-05,
+      "loss": 1.0985,
+      "step": 79130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3660678046405194e-05,
+      "loss": 1.1845,
+      "step": 79135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3659974808139293e-05,
+      "loss": 1.1671,
+      "step": 79140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3659271548972974e-05,
+      "loss": 1.3764,
+      "step": 79145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3658568268910237e-05,
+      "loss": 1.1505,
+      "step": 79150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3657864967955115e-05,
+      "loss": 1.0472,
+      "step": 79155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3657161646111611e-05,
+      "loss": 1.3705,
+      "step": 79160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3656458303383751e-05,
+      "loss": 1.3938,
+      "step": 79165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3655754939775545e-05,
+      "loss": 1.4449,
+      "step": 79170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3655051555291012e-05,
+      "loss": 1.0347,
+      "step": 79175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.365434814993417e-05,
+      "loss": 1.1286,
+      "step": 79180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3653644723709034e-05,
+      "loss": 1.4848,
+      "step": 79185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3652941276619622e-05,
+      "loss": 1.4712,
+      "step": 79190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.365223780866995e-05,
+      "loss": 1.3365,
+      "step": 79195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3651534319864036e-05,
+      "loss": 1.2522,
+      "step": 79200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3650830810205898e-05,
+      "loss": 1.3436,
+      "step": 79205
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3650127279699551e-05,
+      "loss": 1.455,
+      "step": 79210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3649423728349015e-05,
+      "loss": 1.0515,
+      "step": 79215
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3648720156158303e-05,
+      "loss": 1.2777,
+      "step": 79220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3648016563131443e-05,
+      "loss": 1.2368,
+      "step": 79225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.364731294927244e-05,
+      "loss": 1.4382,
+      "step": 79230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3646609314585325e-05,
+      "loss": 1.1483,
+      "step": 79235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3645905659074106e-05,
+      "loss": 1.163,
+      "step": 79240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3645201982742801e-05,
+      "loss": 1.1051,
+      "step": 79245
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3644498285595438e-05,
+      "loss": 1.2454,
+      "step": 79250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3643794567636026e-05,
+      "loss": 1.2857,
+      "step": 79255
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3643090828868587e-05,
+      "loss": 1.4072,
+      "step": 79260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3642387069297142e-05,
+      "loss": 1.2557,
+      "step": 79265
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3641683288925707e-05,
+      "loss": 1.4736,
+      "step": 79270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3640979487758301e-05,
+      "loss": 1.1537,
+      "step": 79275
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3640275665798946e-05,
+      "loss": 1.1298,
+      "step": 79280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3639571823051655e-05,
+      "loss": 1.1368,
+      "step": 79285
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3638867959520456e-05,
+      "loss": 1.2502,
+      "step": 79290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.363816407520936e-05,
+      "loss": 1.3137,
+      "step": 79295
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3637460170122393e-05,
+      "loss": 1.4606,
+      "step": 79300
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3636756244263571e-05,
+      "loss": 1.049,
+      "step": 79305
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3636052297636915e-05,
+      "loss": 1.2709,
+      "step": 79310
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3635348330246445e-05,
+      "loss": 1.2023,
+      "step": 79315
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3634644342096182e-05,
+      "loss": 1.2525,
+      "step": 79320
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3633940333190138e-05,
+      "loss": 1.3816,
+      "step": 79325
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.363323630353235e-05,
+      "loss": 1.0168,
+      "step": 79330
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3632532253126822e-05,
+      "loss": 1.5404,
+      "step": 79335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3631828181977584e-05,
+      "loss": 1.3325,
+      "step": 79340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3631124090088654e-05,
+      "loss": 1.2389,
+      "step": 79345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3630419977464049e-05,
+      "loss": 0.94,
+      "step": 79350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3629715844107799e-05,
+      "loss": 1.2694,
+      "step": 79355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3629011690023913e-05,
+      "loss": 1.1399,
+      "step": 79360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3628307515216422e-05,
+      "loss": 1.5472,
+      "step": 79365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.362760331968934e-05,
+      "loss": 1.3191,
+      "step": 79370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3626899103446696e-05,
+      "loss": 1.2469,
+      "step": 79375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3626194866492506e-05,
+      "loss": 0.9965,
+      "step": 79380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3625490608830794e-05,
+      "loss": 1.4572,
+      "step": 79385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3624786330465578e-05,
+      "loss": 0.9718,
+      "step": 79390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3624082031400885e-05,
+      "loss": 1.2339,
+      "step": 79395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3623377711640732e-05,
+      "loss": 1.0704,
+      "step": 79400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3622673371189145e-05,
+      "loss": 1.0053,
+      "step": 79405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3621969010050147e-05,
+      "loss": 1.4165,
+      "step": 79410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3621264628227755e-05,
+      "loss": 1.4619,
+      "step": 79415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3620560225725993e-05,
+      "loss": 1.3352,
+      "step": 79420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3619855802548888e-05,
+      "loss": 1.5846,
+      "step": 79425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3619151358700455e-05,
+      "loss": 0.9852,
+      "step": 79430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3618446894184728e-05,
+      "loss": 1.073,
+      "step": 79435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3617742409005717e-05,
+      "loss": 1.5635,
+      "step": 79440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3617037903167454e-05,
+      "loss": 1.2525,
+      "step": 79445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3616333376673958e-05,
+      "loss": 1.7354,
+      "step": 79450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3615628829529252e-05,
+      "loss": 1.2188,
+      "step": 79455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3614924261737363e-05,
+      "loss": 1.3857,
+      "step": 79460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3614219673302312e-05,
+      "loss": 1.2865,
+      "step": 79465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3613515064228122e-05,
+      "loss": 1.3372,
+      "step": 79470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3612810434518817e-05,
+      "loss": 1.5976,
+      "step": 79475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3612105784178422e-05,
+      "loss": 1.7261,
+      "step": 79480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3611401113210962e-05,
+      "loss": 1.3962,
+      "step": 79485
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3610696421620455e-05,
+      "loss": 1.4715,
+      "step": 79490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3609991709410933e-05,
+      "loss": 1.4128,
+      "step": 79495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3609286976586415e-05,
+      "loss": 1.4725,
+      "step": 79500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3608582223150928e-05,
+      "loss": 1.2916,
+      "step": 79505
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3607877449108493e-05,
+      "loss": 1.3636,
+      "step": 79510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3607172654463142e-05,
+      "loss": 1.1177,
+      "step": 79515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3606467839218891e-05,
+      "loss": 1.0837,
+      "step": 79520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.360576300337977e-05,
+      "loss": 1.5852,
+      "step": 79525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3605058146949803e-05,
+      "loss": 1.1677,
+      "step": 79530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3604353269933015e-05,
+      "loss": 1.1484,
+      "step": 79535
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3603648372333433e-05,
+      "loss": 1.4308,
+      "step": 79540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3602943454155079e-05,
+      "loss": 1.5496,
+      "step": 79545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.360223851540198e-05,
+      "loss": 1.55,
+      "step": 79550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3601533556078162e-05,
+      "loss": 1.2135,
+      "step": 79555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3600828576187647e-05,
+      "loss": 1.2256,
+      "step": 79560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3600123575734469e-05,
+      "loss": 1.1337,
+      "step": 79565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3599418554722646e-05,
+      "loss": 1.2873,
+      "step": 79570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3598713513156208e-05,
+      "loss": 1.1261,
+      "step": 79575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3598008451039179e-05,
+      "loss": 1.2613,
+      "step": 79580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.359730336837559e-05,
+      "loss": 1.0622,
+      "step": 79585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3596598265169462e-05,
+      "loss": 1.0585,
+      "step": 79590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3595893141424823e-05,
+      "loss": 1.2618,
+      "step": 79595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3595187997145703e-05,
+      "loss": 1.3254,
+      "step": 79600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3594482832336122e-05,
+      "loss": 1.1633,
+      "step": 79605
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3593777647000112e-05,
+      "loss": 1.4985,
+      "step": 79610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.35930724411417e-05,
+      "loss": 1.387,
+      "step": 79615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.359236721476491e-05,
+      "loss": 1.2212,
+      "step": 79620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.359166196787377e-05,
+      "loss": 1.1696,
+      "step": 79625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.359095670047231e-05,
+      "loss": 1.1703,
+      "step": 79630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3590251412564556e-05,
+      "loss": 1.2395,
+      "step": 79635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3589546104154534e-05,
+      "loss": 1.5505,
+      "step": 79640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3588840775246277e-05,
+      "loss": 1.4155,
+      "step": 79645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3588135425843807e-05,
+      "loss": 1.2039,
+      "step": 79650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3587430055951151e-05,
+      "loss": 1.0185,
+      "step": 79655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3586724665572344e-05,
+      "loss": 1.2666,
+      "step": 79660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3586019254711407e-05,
+      "loss": 1.2466,
+      "step": 79665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3585313823372372e-05,
+      "loss": 1.3921,
+      "step": 79670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3584608371559267e-05,
+      "loss": 1.1225,
+      "step": 79675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3583902899276121e-05,
+      "loss": 1.0338,
+      "step": 79680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3583197406526962e-05,
+      "loss": 1.512,
+      "step": 79685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3582491893315815e-05,
+      "loss": 1.0631,
+      "step": 79690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3581786359646718e-05,
+      "loss": 1.2093,
+      "step": 79695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3581080805523689e-05,
+      "loss": 1.1954,
+      "step": 79700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3580375230950766e-05,
+      "loss": 1.126,
+      "step": 79705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3579669635931974e-05,
+      "loss": 1.5666,
+      "step": 79710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3578964020471341e-05,
+      "loss": 1.3852,
+      "step": 79715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3578258384572902e-05,
+      "loss": 1.2838,
+      "step": 79720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3577552728240682e-05,
+      "loss": 1.3252,
+      "step": 79725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3576847051478709e-05,
+      "loss": 1.1221,
+      "step": 79730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.357614135429102e-05,
+      "loss": 1.2694,
+      "step": 79735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3575435636681637e-05,
+      "loss": 1.4462,
+      "step": 79740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3574729898654594e-05,
+      "loss": 1.1092,
+      "step": 79745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3574024140213922e-05,
+      "loss": 1.3831,
+      "step": 79750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.357331836136365e-05,
+      "loss": 0.9356,
+      "step": 79755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3572612562107805e-05,
+      "loss": 1.5972,
+      "step": 79760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3571906742450424e-05,
+      "loss": 1.3246,
+      "step": 79765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3571200902395532e-05,
+      "loss": 1.5657,
+      "step": 79770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3570495041947165e-05,
+      "loss": 1.3275,
+      "step": 79775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3569789161109346e-05,
+      "loss": 1.0483,
+      "step": 79780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3569083259886116e-05,
+      "loss": 1.4233,
+      "step": 79785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.35683773382815e-05,
+      "loss": 1.1569,
+      "step": 79790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3567671396299526e-05,
+      "loss": 1.5478,
+      "step": 79795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3566965433944235e-05,
+      "loss": 0.9337,
+      "step": 79800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.356625945121965e-05,
+      "loss": 1.1469,
+      "step": 79805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3565553448129807e-05,
+      "loss": 1.4775,
+      "step": 79810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3564847424678736e-05,
+      "loss": 1.0767,
+      "step": 79815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3564141380870465e-05,
+      "loss": 1.1051,
+      "step": 79820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3563435316709033e-05,
+      "loss": 1.2745,
+      "step": 79825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3562729232198469e-05,
+      "loss": 1.1667,
+      "step": 79830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3562023127342806e-05,
+      "loss": 1.1628,
+      "step": 79835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3561317002146068e-05,
+      "loss": 1.1772,
+      "step": 79840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3560610856612305e-05,
+      "loss": 1.2896,
+      "step": 79845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3559904690745534e-05,
+      "loss": 1.4779,
+      "step": 79850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3559198504549792e-05,
+      "loss": 1.2615,
+      "step": 79855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.355849229802911e-05,
+      "loss": 1.275,
+      "step": 79860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3557786071187525e-05,
+      "loss": 1.0746,
+      "step": 79865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.355707982402907e-05,
+      "loss": 1.2333,
+      "step": 79870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3556373556557771e-05,
+      "loss": 1.6034,
+      "step": 79875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3555667268777672e-05,
+      "loss": 1.133,
+      "step": 79880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3554960960692797e-05,
+      "loss": 1.2288,
+      "step": 79885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3554254632307184e-05,
+      "loss": 1.16,
+      "step": 79890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3553548283624869e-05,
+      "loss": 1.1944,
+      "step": 79895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3552841914649875e-05,
+      "loss": 1.4057,
+      "step": 79900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3552135525386247e-05,
+      "loss": 1.074,
+      "step": 79905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3551429115838013e-05,
+      "loss": 1.2092,
+      "step": 79910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.355072268600921e-05,
+      "loss": 1.0866,
+      "step": 79915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.355001623590387e-05,
+      "loss": 1.3117,
+      "step": 79920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3549309765526028e-05,
+      "loss": 1.0697,
+      "step": 79925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3548603274879719e-05,
+      "loss": 1.3463,
+      "step": 79930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3547896763968975e-05,
+      "loss": 1.539,
+      "step": 79935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3547190232797833e-05,
+      "loss": 0.9758,
+      "step": 79940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3546483681370326e-05,
+      "loss": 1.0242,
+      "step": 79945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3545777109690493e-05,
+      "loss": 1.0325,
+      "step": 79950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3545070517762362e-05,
+      "loss": 1.1247,
+      "step": 79955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3544363905589972e-05,
+      "loss": 1.4985,
+      "step": 79960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.354365727317736e-05,
+      "loss": 1.3114,
+      "step": 79965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3542950620528555e-05,
+      "loss": 1.2398,
+      "step": 79970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3542243947647602e-05,
+      "loss": 1.2171,
+      "step": 79975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3541537254538525e-05,
+      "loss": 0.8905,
+      "step": 79980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.354083054120537e-05,
+      "loss": 1.4277,
+      "step": 79985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3540123807652169e-05,
+      "loss": 1.0656,
+      "step": 79990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.353941705388295e-05,
+      "loss": 1.2023,
+      "step": 79995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3538710279901763e-05,
+      "loss": 1.2777,
+      "step": 80000
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 1.2250977754592896,
+      "eval_runtime": 6476.3089,
+      "eval_samples_per_second": 3.568,
+      "eval_steps_per_second": 1.784,
+      "step": 80000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3538003485712632e-05,
+      "loss": 1.0356,
+      "step": 80005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3537296671319602e-05,
+      "loss": 1.2002,
+      "step": 80010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3536589836726703e-05,
+      "loss": 1.3668,
+      "step": 80015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3535882981937975e-05,
+      "loss": 1.0317,
+      "step": 80020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3535176106957456e-05,
+      "loss": 1.1679,
+      "step": 80025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3534469211789176e-05,
+      "loss": 1.3956,
+      "step": 80030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3533762296437178e-05,
+      "loss": 1.3687,
+      "step": 80035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3533055360905497e-05,
+      "loss": 1.3098,
+      "step": 80040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3532348405198165e-05,
+      "loss": 1.2163,
+      "step": 80045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3531641429319229e-05,
+      "loss": 1.1468,
+      "step": 80050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3530934433272719e-05,
+      "loss": 1.5289,
+      "step": 80055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3530227417062674e-05,
+      "loss": 1.1981,
+      "step": 80060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3529520380693133e-05,
+      "loss": 1.0967,
+      "step": 80065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.352881332416813e-05,
+      "loss": 1.1464,
+      "step": 80070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3528106247491706e-05,
+      "loss": 1.0533,
+      "step": 80075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3527399150667899e-05,
+      "loss": 1.238,
+      "step": 80080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3526692033700746e-05,
+      "loss": 1.2044,
+      "step": 80085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.352598489659428e-05,
+      "loss": 1.1597,
+      "step": 80090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3525277739352548e-05,
+      "loss": 1.385,
+      "step": 80095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3524570561979581e-05,
+      "loss": 1.3426,
+      "step": 80100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3523863364479426e-05,
+      "loss": 1.3929,
+      "step": 80105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3523156146856112e-05,
+      "loss": 1.3333,
+      "step": 80110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3522448909113682e-05,
+      "loss": 1.515,
+      "step": 80115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3521741651256172e-05,
+      "loss": 1.3285,
+      "step": 80120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3521034373287624e-05,
+      "loss": 1.1408,
+      "step": 80125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.352032707521208e-05,
+      "loss": 1.4424,
+      "step": 80130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3519619757033571e-05,
+      "loss": 1.0482,
+      "step": 80135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.351891241875614e-05,
+      "loss": 0.9791,
+      "step": 80140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.351820506038383e-05,
+      "loss": 1.085,
+      "step": 80145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3517497681920673e-05,
+      "loss": 1.3096,
+      "step": 80150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3516790283370714e-05,
+      "loss": 0.9536,
+      "step": 80155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3516082864737989e-05,
+      "loss": 1.4903,
+      "step": 80160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3515375426026542e-05,
+      "loss": 1.8543,
+      "step": 80165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.351466796724041e-05,
+      "loss": 1.6382,
+      "step": 80170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3513960488383634e-05,
+      "loss": 1.313,
+      "step": 80175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.351325298946025e-05,
+      "loss": 1.1523,
+      "step": 80180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3512545470474308e-05,
+      "loss": 1.2847,
+      "step": 80185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.351183793142984e-05,
+      "loss": 1.2731,
+      "step": 80190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3511130372330885e-05,
+      "loss": 0.9963,
+      "step": 80195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3510422793181493e-05,
+      "loss": 1.1019,
+      "step": 80200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3509715193985691e-05,
+      "loss": 1.3063,
+      "step": 80205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3509007574747535e-05,
+      "loss": 1.0429,
+      "step": 80210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3508299935471052e-05,
+      "loss": 1.3332,
+      "step": 80215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3507592276160292e-05,
+      "loss": 1.1008,
+      "step": 80220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3506884596819297e-05,
+      "loss": 1.3354,
+      "step": 80225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3506176897452101e-05,
+      "loss": 1.0321,
+      "step": 80230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3505469178062751e-05,
+      "loss": 1.1662,
+      "step": 80235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3504761438655285e-05,
+      "loss": 1.1666,
+      "step": 80240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3504053679233748e-05,
+      "loss": 1.4261,
+      "step": 80245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3503345899802178e-05,
+      "loss": 0.9944,
+      "step": 80250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.350263810036462e-05,
+      "loss": 1.1514,
+      "step": 80255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3501930280925115e-05,
+      "loss": 0.9549,
+      "step": 80260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3501222441487703e-05,
+      "loss": 1.0952,
+      "step": 80265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3500514582056427e-05,
+      "loss": 1.6822,
+      "step": 80270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.349980670263533e-05,
+      "loss": 1.353,
+      "step": 80275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.349909880322846e-05,
+      "loss": 0.9426,
+      "step": 80280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3498390883839844e-05,
+      "loss": 1.1283,
+      "step": 80285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3497682944473543e-05,
+      "loss": 1.009,
+      "step": 80290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3496974985133589e-05,
+      "loss": 1.1941,
+      "step": 80295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3496267005824022e-05,
+      "loss": 1.6695,
+      "step": 80300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3495559006548897e-05,
+      "loss": 1.5077,
+      "step": 80305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3494850987312244e-05,
+      "loss": 1.0545,
+      "step": 80310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3494142948118118e-05,
+      "loss": 1.1431,
+      "step": 80315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3493434888970551e-05,
+      "loss": 1.3436,
+      "step": 80320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3492726809873595e-05,
+      "loss": 1.2281,
+      "step": 80325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3492018710831291e-05,
+      "loss": 1.07,
+      "step": 80330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.349131059184768e-05,
+      "loss": 1.0156,
+      "step": 80335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3490602452926806e-05,
+      "loss": 1.3416,
+      "step": 80340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3489894294072719e-05,
+      "loss": 1.3264,
+      "step": 80345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3489186115289456e-05,
+      "loss": 1.1116,
+      "step": 80350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3488477916581064e-05,
+      "loss": 1.3251,
+      "step": 80355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3487769697951588e-05,
+      "loss": 1.2302,
+      "step": 80360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3487061459405068e-05,
+      "loss": 0.9996,
+      "step": 80365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3486353200945555e-05,
+      "loss": 1.5646,
+      "step": 80370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3485644922577089e-05,
+      "loss": 1.1452,
+      "step": 80375
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3484936624303713e-05,
+      "loss": 1.0977,
+      "step": 80380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.348422830612948e-05,
+      "loss": 1.1692,
+      "step": 80385
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3483519968058425e-05,
+      "loss": 1.1656,
+      "step": 80390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3482811610094597e-05,
+      "loss": 1.2099,
+      "step": 80395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3482103232242044e-05,
+      "loss": 1.1055,
+      "step": 80400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3481394834504806e-05,
+      "loss": 1.0615,
+      "step": 80405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3480686416886934e-05,
+      "loss": 1.1993,
+      "step": 80410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3479977979392467e-05,
+      "loss": 1.2752,
+      "step": 80415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3479269522025458e-05,
+      "loss": 1.1695,
+      "step": 80420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3478561044789947e-05,
+      "loss": 1.2482,
+      "step": 80425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.347785254768998e-05,
+      "loss": 1.0835,
+      "step": 80430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3477144030729607e-05,
+      "loss": 1.0525,
+      "step": 80435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.347643549391287e-05,
+      "loss": 1.2094,
+      "step": 80440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3475726937243816e-05,
+      "loss": 1.1695,
+      "step": 80445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3475018360726494e-05,
+      "loss": 1.4138,
+      "step": 80450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3474309764364946e-05,
+      "loss": 1.1581,
+      "step": 80455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.347360114816322e-05,
+      "loss": 1.3293,
+      "step": 80460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3472892512125367e-05,
+      "loss": 0.9747,
+      "step": 80465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3472183856255425e-05,
+      "loss": 1.1546,
+      "step": 80470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3471475180557447e-05,
+      "loss": 1.1617,
+      "step": 80475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.347076648503548e-05,
+      "loss": 1.3585,
+      "step": 80480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3470057769693568e-05,
+      "loss": 1.2629,
+      "step": 80485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3469349034535762e-05,
+      "loss": 0.885,
+      "step": 80490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3468640279566105e-05,
+      "loss": 1.1196,
+      "step": 80495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.346793150478865e-05,
+      "loss": 1.0237,
+      "step": 80500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3467222710207437e-05,
+      "loss": 1.3124,
+      "step": 80505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3466513895826517e-05,
+      "loss": 1.116,
+      "step": 80510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.346580506164994e-05,
+      "loss": 1.3743,
+      "step": 80515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.346509620768175e-05,
+      "loss": 1.2156,
+      "step": 80520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3464387333926e-05,
+      "loss": 1.199,
+      "step": 80525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3463678440386737e-05,
+      "loss": 1.9087,
+      "step": 80530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3462969527068e-05,
+      "loss": 1.092,
+      "step": 80535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.346226059397385e-05,
+      "loss": 1.2331,
+      "step": 80540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3461551641108326e-05,
+      "loss": 1.0024,
+      "step": 80545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3460842668475483e-05,
+      "loss": 1.2122,
+      "step": 80550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3460133676079366e-05,
+      "loss": 0.9863,
+      "step": 80555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3459424663924025e-05,
+      "loss": 1.4303,
+      "step": 80560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3458715632013508e-05,
+      "loss": 1.1013,
+      "step": 80565
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3458006580351866e-05,
+      "loss": 1.2108,
+      "step": 80570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3457297508943144e-05,
+      "loss": 1.0988,
+      "step": 80575
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3456588417791396e-05,
+      "loss": 1.3745,
+      "step": 80580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3455879306900667e-05,
+      "loss": 1.0807,
+      "step": 80585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.345517017627501e-05,
+      "loss": 1.5057,
+      "step": 80590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3454461025918473e-05,
+      "loss": 1.5938,
+      "step": 80595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3453751855835106e-05,
+      "loss": 1.4446,
+      "step": 80600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3453042666028957e-05,
+      "loss": 1.1816,
+      "step": 80605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3452333456504079e-05,
+      "loss": 1.3012,
+      "step": 80610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3451624227264514e-05,
+      "loss": 1.1186,
+      "step": 80615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3450914978314326e-05,
+      "loss": 1.3471,
+      "step": 80620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3450205709657552e-05,
+      "loss": 1.2739,
+      "step": 80625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3449496421298252e-05,
+      "loss": 1.561,
+      "step": 80630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3448787113240468e-05,
+      "loss": 1.4056,
+      "step": 80635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3448077785488258e-05,
+      "loss": 1.3152,
+      "step": 80640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3447368438045669e-05,
+      "loss": 1.4101,
+      "step": 80645
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3446659070916748e-05,
+      "loss": 1.2328,
+      "step": 80650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3445949684105555e-05,
+      "loss": 1.1263,
+      "step": 80655
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3445240277616133e-05,
+      "loss": 1.1311,
+      "step": 80660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3444530851452537e-05,
+      "loss": 1.3666,
+      "step": 80665
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3443821405618817e-05,
+      "loss": 1.0448,
+      "step": 80670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3443111940119024e-05,
+      "loss": 1.5053,
+      "step": 80675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.344240245495721e-05,
+      "loss": 1.0754,
+      "step": 80680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3441692950137428e-05,
+      "loss": 1.1847,
+      "step": 80685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3440983425663727e-05,
+      "loss": 1.2353,
+      "step": 80690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.344027388154016e-05,
+      "loss": 1.1547,
+      "step": 80695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3439564317770779e-05,
+      "loss": 1.0302,
+      "step": 80700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3438854734359637e-05,
+      "loss": 1.143,
+      "step": 80705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.343814513131078e-05,
+      "loss": 1.0633,
+      "step": 80710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3437435508628269e-05,
+      "loss": 0.9124,
+      "step": 80715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3436725866316152e-05,
+      "loss": 1.3443,
+      "step": 80720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3436016204378483e-05,
+      "loss": 1.1227,
+      "step": 80725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3435306522819311e-05,
+      "loss": 1.13,
+      "step": 80730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3434596821642692e-05,
+      "loss": 1.1839,
+      "step": 80735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3433887100852679e-05,
+      "loss": 1.5911,
+      "step": 80740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3433177360453321e-05,
+      "loss": 1.1228,
+      "step": 80745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3432467600448678e-05,
+      "loss": 1.4891,
+      "step": 80750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3431757820842795e-05,
+      "loss": 1.0287,
+      "step": 80755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.343104802163973e-05,
+      "loss": 1.2185,
+      "step": 80760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3430338202843537e-05,
+      "loss": 1.1646,
+      "step": 80765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3429628364458264e-05,
+      "loss": 1.3813,
+      "step": 80770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3428918506487971e-05,
+      "loss": 1.3339,
+      "step": 80775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.342820862893671e-05,
+      "loss": 1.2462,
+      "step": 80780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3427498731808532e-05,
+      "loss": 1.076,
+      "step": 80785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3426788815107495e-05,
+      "loss": 1.4905,
+      "step": 80790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.342607887883765e-05,
+      "loss": 1.1591,
+      "step": 80795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.342536892300305e-05,
+      "loss": 1.1833,
+      "step": 80800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3424658947607754e-05,
+      "loss": 1.1878,
+      "step": 80805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.342394895265581e-05,
+      "loss": 0.9856,
+      "step": 80810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3423238938151278e-05,
+      "loss": 1.8321,
+      "step": 80815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3422528904098211e-05,
+      "loss": 1.2229,
+      "step": 80820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3421818850500661e-05,
+      "loss": 1.1531,
+      "step": 80825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3421108777362688e-05,
+      "loss": 1.4053,
+      "step": 80830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3420398684688343e-05,
+      "loss": 0.8733,
+      "step": 80835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3419688572481678e-05,
+      "loss": 1.1104,
+      "step": 80840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3418978440746756e-05,
+      "loss": 1.3478,
+      "step": 80845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3418268289487625e-05,
+      "loss": 1.4602,
+      "step": 80850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3417558118708347e-05,
+      "loss": 1.1321,
+      "step": 80855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3416847928412973e-05,
+      "loss": 1.3231,
+      "step": 80860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3416137718605559e-05,
+      "loss": 1.1491,
+      "step": 80865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3415427489290163e-05,
+      "loss": 1.3641,
+      "step": 80870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3414717240470836e-05,
+      "loss": 1.1969,
+      "step": 80875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.341400697215164e-05,
+      "loss": 1.1152,
+      "step": 80880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3413296684336626e-05,
+      "loss": 1.0143,
+      "step": 80885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3412586377029853e-05,
+      "loss": 1.2643,
+      "step": 80890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3411876050235375e-05,
+      "loss": 1.4068,
+      "step": 80895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3411165703957252e-05,
+      "loss": 1.209,
+      "step": 80900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3410455338199536e-05,
+      "loss": 1.3126,
+      "step": 80905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3409744952966286e-05,
+      "loss": 1.1367,
+      "step": 80910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.340903454826156e-05,
+      "loss": 1.2221,
+      "step": 80915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.340832412408941e-05,
+      "loss": 1.4221,
+      "step": 80920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3407613680453899e-05,
+      "loss": 1.502,
+      "step": 80925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3406903217359078e-05,
+      "loss": 1.0158,
+      "step": 80930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3406192734809012e-05,
+      "loss": 1.369,
+      "step": 80935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3405482232807749e-05,
+      "loss": 1.4183,
+      "step": 80940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3404771711359352e-05,
+      "loss": 1.2492,
+      "step": 80945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3404061170467882e-05,
+      "loss": 1.4433,
+      "step": 80950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3403350610137384e-05,
+      "loss": 1.252,
+      "step": 80955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3402640030371927e-05,
+      "loss": 1.1305,
+      "step": 80960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3401929431175566e-05,
+      "loss": 1.8943,
+      "step": 80965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3401218812552358e-05,
+      "loss": 1.1895,
+      "step": 80970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3400508174506363e-05,
+      "loss": 1.3837,
+      "step": 80975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3399797517041634e-05,
+      "loss": 1.0691,
+      "step": 80980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3399086840162231e-05,
+      "loss": 1.1962,
+      "step": 80985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3398376143872218e-05,
+      "loss": 1.1188,
+      "step": 80990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3397665428175648e-05,
+      "loss": 1.0616,
+      "step": 80995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3396954693076582e-05,
+      "loss": 1.1709,
+      "step": 81000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3396243938579077e-05,
+      "loss": 1.2141,
+      "step": 81005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3395533164687194e-05,
+      "loss": 1.383,
+      "step": 81010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3394822371404987e-05,
+      "loss": 1.5258,
+      "step": 81015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.339411155873652e-05,
+      "loss": 0.963,
+      "step": 81020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3393400726685848e-05,
+      "loss": 1.8074,
+      "step": 81025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3392689875257037e-05,
+      "loss": 1.2675,
+      "step": 81030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.339197900445414e-05,
+      "loss": 1.045,
+      "step": 81035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3391268114281217e-05,
+      "loss": 1.0688,
+      "step": 81040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3390557204742333e-05,
+      "loss": 1.2529,
+      "step": 81045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.338984627584154e-05,
+      "loss": 1.0944,
+      "step": 81050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3389135327582905e-05,
+      "loss": 1.3137,
+      "step": 81055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.338842435997048e-05,
+      "loss": 1.2963,
+      "step": 81060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.338771337300833e-05,
+      "loss": 1.2362,
+      "step": 81065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3387002366700518e-05,
+      "loss": 1.2895,
+      "step": 81070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3386291341051099e-05,
+      "loss": 1.1642,
+      "step": 81075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3385580296064137e-05,
+      "loss": 1.3251,
+      "step": 81080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3384869231743688e-05,
+      "loss": 1.2311,
+      "step": 81085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3384158148093816e-05,
+      "loss": 1.055,
+      "step": 81090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.338344704511858e-05,
+      "loss": 1.5652,
+      "step": 81095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3382735922822044e-05,
+      "loss": 1.183,
+      "step": 81100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3382024781208264e-05,
+      "loss": 1.0822,
+      "step": 81105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3381313620281305e-05,
+      "loss": 1.3772,
+      "step": 81110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3380602440045225e-05,
+      "loss": 1.0779,
+      "step": 81115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3379891240504088e-05,
+      "loss": 1.3275,
+      "step": 81120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3379180021661955e-05,
+      "loss": 1.4552,
+      "step": 81125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3378468783522883e-05,
+      "loss": 1.2967,
+      "step": 81130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3377757526090942e-05,
+      "loss": 1.5726,
+      "step": 81135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3377046249370188e-05,
+      "loss": 1.1919,
+      "step": 81140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3376334953364678e-05,
+      "loss": 1.0838,
+      "step": 81145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3375623638078485e-05,
+      "loss": 1.1558,
+      "step": 81150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3374912303515662e-05,
+      "loss": 1.3499,
+      "step": 81155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3374200949680279e-05,
+      "loss": 1.0571,
+      "step": 81160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3373489576576388e-05,
+      "loss": 1.116,
+      "step": 81165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3372778184208062e-05,
+      "loss": 1.1544,
+      "step": 81170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3372066772579357e-05,
+      "loss": 1.3415,
+      "step": 81175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3371355341694335e-05,
+      "loss": 1.6755,
+      "step": 81180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3370643891557063e-05,
+      "loss": 1.0925,
+      "step": 81185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.33699324221716e-05,
+      "loss": 1.0515,
+      "step": 81190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3369220933542012e-05,
+      "loss": 1.3085,
+      "step": 81195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3368509425672361e-05,
+      "loss": 1.0722,
+      "step": 81200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3367797898566706e-05,
+      "loss": 1.4992,
+      "step": 81205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3367086352229115e-05,
+      "loss": 1.072,
+      "step": 81210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3366374786663652e-05,
+      "loss": 1.242,
+      "step": 81215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3365663201874377e-05,
+      "loss": 1.0976,
+      "step": 81220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3364951597865354e-05,
+      "loss": 1.2022,
+      "step": 81225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.336423997464065e-05,
+      "loss": 1.3484,
+      "step": 81230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3363528332204323e-05,
+      "loss": 1.1208,
+      "step": 81235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3362816670560443e-05,
+      "loss": 1.2011,
+      "step": 81240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.336210498971307e-05,
+      "loss": 1.5685,
+      "step": 81245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3361393289666267e-05,
+      "loss": 1.4243,
+      "step": 81250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3360681570424103e-05,
+      "loss": 1.0427,
+      "step": 81255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3359969831990638e-05,
+      "loss": 1.3608,
+      "step": 81260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3359258074369941e-05,
+      "loss": 1.2356,
+      "step": 81265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3358546297566071e-05,
+      "loss": 1.3694,
+      "step": 81270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3357834501583095e-05,
+      "loss": 1.3087,
+      "step": 81275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3357122686425082e-05,
+      "loss": 1.0663,
+      "step": 81280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3356410852096087e-05,
+      "loss": 1.4912,
+      "step": 81285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3355698998600184e-05,
+      "loss": 0.9015,
+      "step": 81290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3354987125941434e-05,
+      "loss": 1.1679,
+      "step": 81295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3354275234123903e-05,
+      "loss": 1.1101,
+      "step": 81300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3353563323151657e-05,
+      "loss": 1.273,
+      "step": 81305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3352851393028761e-05,
+      "loss": 1.2788,
+      "step": 81310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3352139443759279e-05,
+      "loss": 1.4409,
+      "step": 81315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3351427475347278e-05,
+      "loss": 0.9492,
+      "step": 81320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3350715487796827e-05,
+      "loss": 1.0976,
+      "step": 81325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3350003481111982e-05,
+      "loss": 1.0989,
+      "step": 81330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.334929145529682e-05,
+      "loss": 1.0313,
+      "step": 81335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3348579410355402e-05,
+      "loss": 0.9448,
+      "step": 81340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3347867346291795e-05,
+      "loss": 1.2292,
+      "step": 81345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3347155263110063e-05,
+      "loss": 1.3731,
+      "step": 81350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3346443160814273e-05,
+      "loss": 1.2711,
+      "step": 81355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3345731039408496e-05,
+      "loss": 1.7715,
+      "step": 81360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3345018898896791e-05,
+      "loss": 1.3915,
+      "step": 81365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3344306739283233e-05,
+      "loss": 1.0878,
+      "step": 81370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3343594560571884e-05,
+      "loss": 1.1254,
+      "step": 81375
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.334288236276681e-05,
+      "loss": 1.2529,
+      "step": 81380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3342170145872083e-05,
+      "loss": 1.2467,
+      "step": 81385
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3341457909891762e-05,
+      "loss": 1.1565,
+      "step": 81390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.334074565482992e-05,
+      "loss": 1.1993,
+      "step": 81395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3340033380690626e-05,
+      "loss": 1.1497,
+      "step": 81400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3339321087477943e-05,
+      "loss": 1.4169,
+      "step": 81405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3338608775195942e-05,
+      "loss": 1.3382,
+      "step": 81410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.333789644384869e-05,
+      "loss": 1.1282,
+      "step": 81415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3337184093440251e-05,
+      "loss": 1.0108,
+      "step": 81420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.33364717239747e-05,
+      "loss": 1.4169,
+      "step": 81425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3335759335456099e-05,
+      "loss": 1.1946,
+      "step": 81430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3335046927888518e-05,
+      "loss": 1.1267,
+      "step": 81435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3334334501276024e-05,
+      "loss": 1.0019,
+      "step": 81440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3333622055622688e-05,
+      "loss": 1.1356,
+      "step": 81445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3332909590932577e-05,
+      "loss": 1.073,
+      "step": 81450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.333219710720976e-05,
+      "loss": 1.3445,
+      "step": 81455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3331484604458302e-05,
+      "loss": 1.1593,
+      "step": 81460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3330772082682279e-05,
+      "loss": 1.324,
+      "step": 81465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3330059541885754e-05,
+      "loss": 1.1654,
+      "step": 81470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.33293469820728e-05,
+      "loss": 0.919,
+      "step": 81475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3328634403247485e-05,
+      "loss": 1.3216,
+      "step": 81480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3327921805413872e-05,
+      "loss": 1.0186,
+      "step": 81485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.332720918857604e-05,
+      "loss": 1.0354,
+      "step": 81490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.332649655273805e-05,
+      "loss": 1.1581,
+      "step": 81495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.332578389790398e-05,
+      "loss": 1.3474,
+      "step": 81500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.332507122407789e-05,
+      "loss": 1.2814,
+      "step": 81505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.332435853126386e-05,
+      "loss": 1.4137,
+      "step": 81510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3323645819465955e-05,
+      "loss": 1.4039,
+      "step": 81515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3322933088688239e-05,
+      "loss": 1.0432,
+      "step": 81520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3322220338934793e-05,
+      "loss": 1.2431,
+      "step": 81525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.332150757020968e-05,
+      "loss": 1.1661,
+      "step": 81530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3320794782516973e-05,
+      "loss": 1.2803,
+      "step": 81535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.332008197586074e-05,
+      "loss": 1.1562,
+      "step": 81540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3319369150245056e-05,
+      "loss": 1.2123,
+      "step": 81545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3318656305673987e-05,
+      "loss": 1.1169,
+      "step": 81550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3317943442151606e-05,
+      "loss": 1.2685,
+      "step": 81555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3317230559681984e-05,
+      "loss": 1.2623,
+      "step": 81560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3316517658269189e-05,
+      "loss": 1.1348,
+      "step": 81565
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3315804737917297e-05,
+      "loss": 1.0624,
+      "step": 81570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3315091798630375e-05,
+      "loss": 1.2765,
+      "step": 81575
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3314378840412494e-05,
+      "loss": 1.2327,
+      "step": 81580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3313665863267733e-05,
+      "loss": 1.0371,
+      "step": 81585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3312952867200153e-05,
+      "loss": 1.6506,
+      "step": 81590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3312239852213832e-05,
+      "loss": 1.3516,
+      "step": 81595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3311526818312839e-05,
+      "loss": 1.348,
+      "step": 81600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3310813765501247e-05,
+      "loss": 1.1581,
+      "step": 81605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3310100693783125e-05,
+      "loss": 1.1032,
+      "step": 81610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.330938760316255e-05,
+      "loss": 1.4187,
+      "step": 81615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3308674493643594e-05,
+      "loss": 1.1557,
+      "step": 81620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3307961365230326e-05,
+      "loss": 1.1377,
+      "step": 81625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3307248217926819e-05,
+      "loss": 1.5065,
+      "step": 81630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3306535051737146e-05,
+      "loss": 1.5277,
+      "step": 81635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3305821866665378e-05,
+      "loss": 1.0803,
+      "step": 81640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.330510866271559e-05,
+      "loss": 1.2613,
+      "step": 81645
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3304395439891856e-05,
+      "loss": 1.0462,
+      "step": 81650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3303682198198244e-05,
+      "loss": 1.2682,
+      "step": 81655
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3302968937638832e-05,
+      "loss": 1.2436,
+      "step": 81660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3302255658217689e-05,
+      "loss": 1.1762,
+      "step": 81665
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3301542359938888e-05,
+      "loss": 1.1172,
+      "step": 81670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3300829042806508e-05,
+      "loss": 1.1603,
+      "step": 81675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.330011570682462e-05,
+      "loss": 1.2815,
+      "step": 81680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.329940235199729e-05,
+      "loss": 1.0892,
+      "step": 81685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3298688978328605e-05,
+      "loss": 0.9658,
+      "step": 81690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3297975585822627e-05,
+      "loss": 1.2777,
+      "step": 81695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3297262174483439e-05,
+      "loss": 1.4062,
+      "step": 81700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3296548744315105e-05,
+      "loss": 1.3909,
+      "step": 81705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3295835295321706e-05,
+      "loss": 1.013,
+      "step": 81710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3295121827507318e-05,
+      "loss": 0.9964,
+      "step": 81715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3294408340876009e-05,
+      "loss": 1.1379,
+      "step": 81720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3293694835431858e-05,
+      "loss": 1.1657,
+      "step": 81725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3292981311178938e-05,
+      "loss": 1.2966,
+      "step": 81730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3292267768121323e-05,
+      "loss": 1.3329,
+      "step": 81735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.329155420626309e-05,
+      "loss": 1.2364,
+      "step": 81740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.329084062560831e-05,
+      "loss": 1.1091,
+      "step": 81745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3290127026161062e-05,
+      "loss": 1.1116,
+      "step": 81750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3289413407925418e-05,
+      "loss": 1.3908,
+      "step": 81755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3288699770905454e-05,
+      "loss": 1.3196,
+      "step": 81760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3287986115105242e-05,
+      "loss": 1.0169,
+      "step": 81765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3287272440528867e-05,
+      "loss": 1.177,
+      "step": 81770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3286558747180393e-05,
+      "loss": 1.1748,
+      "step": 81775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3285845035063907e-05,
+      "loss": 1.4271,
+      "step": 81780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3285131304183474e-05,
+      "loss": 1.2246,
+      "step": 81785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3284417554543174e-05,
+      "loss": 1.0276,
+      "step": 81790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3283703786147087e-05,
+      "loss": 1.0802,
+      "step": 81795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.328298999899928e-05,
+      "loss": 1.1089,
+      "step": 81800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.328227619310384e-05,
+      "loss": 1.112,
+      "step": 81805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3281562368464834e-05,
+      "loss": 1.1962,
+      "step": 81810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.328084852508634e-05,
+      "loss": 1.3408,
+      "step": 81815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3280134662972444e-05,
+      "loss": 1.5755,
+      "step": 81820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3279420782127206e-05,
+      "loss": 1.5247,
+      "step": 81825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3278706882554717e-05,
+      "loss": 1.3149,
+      "step": 81830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3277992964259045e-05,
+      "loss": 1.1887,
+      "step": 81835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3277279027244273e-05,
+      "loss": 1.2478,
+      "step": 81840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3276565071514474e-05,
+      "loss": 1.5635,
+      "step": 81845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3275851097073723e-05,
+      "loss": 1.0972,
+      "step": 81850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3275137103926105e-05,
+      "loss": 1.2592,
+      "step": 81855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.327442309207569e-05,
+      "loss": 1.8655,
+      "step": 81860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.327370906152656e-05,
+      "loss": 1.1544,
+      "step": 81865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3272995012282787e-05,
+      "loss": 0.8962,
+      "step": 81870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3272280944348456e-05,
+      "loss": 1.2359,
+      "step": 81875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.327156685772764e-05,
+      "loss": 1.0594,
+      "step": 81880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3270852752424416e-05,
+      "loss": 1.1143,
+      "step": 81885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3270138628442865e-05,
+      "loss": 1.0338,
+      "step": 81890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.326942448578706e-05,
+      "loss": 1.3548,
+      "step": 81895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.326871032446109e-05,
+      "loss": 1.3173,
+      "step": 81900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3267996144469018e-05,
+      "loss": 1.3521,
+      "step": 81905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3267281945814933e-05,
+      "loss": 1.373,
+      "step": 81910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3266567728502913e-05,
+      "loss": 1.1798,
+      "step": 81915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3265853492537032e-05,
+      "loss": 1.059,
+      "step": 81920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3265139237921375e-05,
+      "loss": 1.1656,
+      "step": 81925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3264424964660011e-05,
+      "loss": 1.1154,
+      "step": 81930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3263710672757028e-05,
+      "loss": 1.3863,
+      "step": 81935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3262996362216503e-05,
+      "loss": 1.1595,
+      "step": 81940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.326228203304251e-05,
+      "loss": 0.8974,
+      "step": 81945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3261567685239134e-05,
+      "loss": 0.8547,
+      "step": 81950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3260853318810454e-05,
+      "loss": 1.3599,
+      "step": 81955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3260138933760545e-05,
+      "loss": 1.5135,
+      "step": 81960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3259424530093489e-05,
+      "loss": 1.5296,
+      "step": 81965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3258710107813369e-05,
+      "loss": 1.2957,
+      "step": 81970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3257995666924258e-05,
+      "loss": 1.0896,
+      "step": 81975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3257281207430241e-05,
+      "loss": 1.3243,
+      "step": 81980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3256566729335396e-05,
+      "loss": 1.5083,
+      "step": 81985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3255852232643804e-05,
+      "loss": 1.265,
+      "step": 81990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3255137717359546e-05,
+      "loss": 1.1749,
+      "step": 81995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3254423183486695e-05,
+      "loss": 1.0783,
+      "step": 82000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3253708631029346e-05,
+      "loss": 1.2222,
+      "step": 82005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3252994059991561e-05,
+      "loss": 1.3122,
+      "step": 82010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3252279470377437e-05,
+      "loss": 1.2931,
+      "step": 82015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.325156486219105e-05,
+      "loss": 1.1296,
+      "step": 82020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3250850235436472e-05,
+      "loss": 2.0589,
+      "step": 82025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3250135590117794e-05,
+      "loss": 1.3369,
+      "step": 82030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3249420926239092e-05,
+      "loss": 1.1666,
+      "step": 82035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.324870624380445e-05,
+      "loss": 1.0992,
+      "step": 82040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.324799154281795e-05,
+      "loss": 1.0781,
+      "step": 82045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.324727682328367e-05,
+      "loss": 0.9149,
+      "step": 82050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3246562085205693e-05,
+      "loss": 0.9914,
+      "step": 82055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.32458473285881e-05,
+      "loss": 1.1661,
+      "step": 82060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3245132553434972e-05,
+      "loss": 1.2015,
+      "step": 82065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3244417759750393e-05,
+      "loss": 1.2031,
+      "step": 82070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3243702947538442e-05,
+      "loss": 1.256,
+      "step": 82075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3242988116803203e-05,
+      "loss": 1.4285,
+      "step": 82080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3242273267548758e-05,
+      "loss": 1.1152,
+      "step": 82085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.324155839977919e-05,
+      "loss": 1.2736,
+      "step": 82090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3240843513498576e-05,
+      "loss": 1.2346,
+      "step": 82095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3240128608711009e-05,
+      "loss": 1.6604,
+      "step": 82100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.323941368542056e-05,
+      "loss": 1.1285,
+      "step": 82105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.323869874363132e-05,
+      "loss": 1.2025,
+      "step": 82110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3237983783347363e-05,
+      "loss": 1.256,
+      "step": 82115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.323726880457278e-05,
+      "loss": 1.1491,
+      "step": 82120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3236553807311653e-05,
+      "loss": 1.1582,
+      "step": 82125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.323583879156806e-05,
+      "loss": 1.0443,
+      "step": 82130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.323512375734609e-05,
+      "loss": 1.3299,
+      "step": 82135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.323440870464982e-05,
+      "loss": 0.899,
+      "step": 82140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3233693633483338e-05,
+      "loss": 1.1621,
+      "step": 82145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3232978543850726e-05,
+      "loss": 1.2054,
+      "step": 82150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3232263435756068e-05,
+      "loss": 1.3032,
+      "step": 82155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3231548309203448e-05,
+      "loss": 1.3739,
+      "step": 82160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3230833164196948e-05,
+      "loss": 1.315,
+      "step": 82165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3230118000740652e-05,
+      "loss": 1.1647,
+      "step": 82170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3229402818838645e-05,
+      "loss": 1.1646,
+      "step": 82175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3228687618495014e-05,
+      "loss": 1.3272,
+      "step": 82180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3227972399713838e-05,
+      "loss": 1.2964,
+      "step": 82185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3227257162499204e-05,
+      "loss": 1.0359,
+      "step": 82190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3226541906855195e-05,
+      "loss": 1.0864,
+      "step": 82195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3225826632785895e-05,
+      "loss": 1.1656,
+      "step": 82200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3225111340295393e-05,
+      "loss": 1.0444,
+      "step": 82205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.322439602938777e-05,
+      "loss": 1.3764,
+      "step": 82210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3223680700067112e-05,
+      "loss": 0.9798,
+      "step": 82215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.32229653523375e-05,
+      "loss": 0.9161,
+      "step": 82220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3222249986203025e-05,
+      "loss": 1.2852,
+      "step": 82225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3221534601667769e-05,
+      "loss": 1.1225,
+      "step": 82230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3220819198735818e-05,
+      "loss": 1.0014,
+      "step": 82235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3220103777411254e-05,
+      "loss": 1.2841,
+      "step": 82240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.321938833769817e-05,
+      "loss": 1.4665,
+      "step": 82245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3218672879600645e-05,
+      "loss": 1.269,
+      "step": 82250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3217957403122765e-05,
+      "loss": 1.3619,
+      "step": 82255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3217241908268622e-05,
+      "loss": 1.4357,
+      "step": 82260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3216526395042293e-05,
+      "loss": 1.1661,
+      "step": 82265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.321581086344787e-05,
+      "loss": 1.4497,
+      "step": 82270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3215095313489436e-05,
+      "loss": 1.15,
+      "step": 82275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.321437974517108e-05,
+      "loss": 1.0811,
+      "step": 82280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3213664158496888e-05,
+      "loss": 1.482,
+      "step": 82285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3212948553470943e-05,
+      "loss": 0.9933,
+      "step": 82290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3212232930097334e-05,
+      "loss": 1.2489,
+      "step": 82295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.321151728838015e-05,
+      "loss": 1.5734,
+      "step": 82300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3210801628323469e-05,
+      "loss": 1.2745,
+      "step": 82305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3210085949931389e-05,
+      "loss": 1.5127,
+      "step": 82310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.320937025320799e-05,
+      "loss": 1.3977,
+      "step": 82315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3208654538157362e-05,
+      "loss": 1.1471,
+      "step": 82320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.320793880478359e-05,
+      "loss": 1.0471,
+      "step": 82325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3207223053090759e-05,
+      "loss": 1.0073,
+      "step": 82330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3206507283082963e-05,
+      "loss": 1.447,
+      "step": 82335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3205791494764284e-05,
+      "loss": 1.3146,
+      "step": 82340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.320507568813881e-05,
+      "loss": 1.3169,
+      "step": 82345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3204359863210634e-05,
+      "loss": 0.8986,
+      "step": 82350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3203644019983837e-05,
+      "loss": 1.0952,
+      "step": 82355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3202928158462514e-05,
+      "loss": 1.3757,
+      "step": 82360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3202212278650741e-05,
+      "loss": 1.2586,
+      "step": 82365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.320149638055262e-05,
+      "loss": 1.1966,
+      "step": 82370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3200780464172227e-05,
+      "loss": 1.1415,
+      "step": 82375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.320006452951366e-05,
+      "loss": 1.2941,
+      "step": 82380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3199348576581001e-05,
+      "loss": 1.0057,
+      "step": 82385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3198632605378344e-05,
+      "loss": 1.0559,
+      "step": 82390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3197916615909772e-05,
+      "loss": 1.0389,
+      "step": 82395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3197200608179376e-05,
+      "loss": 1.0958,
+      "step": 82400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3196484582191246e-05,
+      "loss": 1.3171,
+      "step": 82405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3195768537949468e-05,
+      "loss": 1.1519,
+      "step": 82410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3195052475458137e-05,
+      "loss": 1.0063,
+      "step": 82415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3194336394721336e-05,
+      "loss": 1.3337,
+      "step": 82420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3193620295743154e-05,
+      "loss": 1.2206,
+      "step": 82425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3192904178527685e-05,
+      "loss": 1.0636,
+      "step": 82430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3192188043079013e-05,
+      "loss": 1.1801,
+      "step": 82435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3191471889401233e-05,
+      "loss": 1.1353,
+      "step": 82440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3190755717498429e-05,
+      "loss": 1.2056,
+      "step": 82445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3190039527374696e-05,
+      "loss": 1.0252,
+      "step": 82450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3189323319034122e-05,
+      "loss": 1.5239,
+      "step": 82455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3188607092480794e-05,
+      "loss": 1.105,
+      "step": 82460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3187890847718809e-05,
+      "loss": 1.2905,
+      "step": 82465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3187174584752246e-05,
+      "loss": 1.0688,
+      "step": 82470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3186458303585207e-05,
+      "loss": 1.4363,
+      "step": 82475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3185742004221776e-05,
+      "loss": 1.0624,
+      "step": 82480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3185025686666046e-05,
+      "loss": 1.231,
+      "step": 82485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3184309350922104e-05,
+      "loss": 1.2515,
+      "step": 82490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3183592996994044e-05,
+      "loss": 1.0591,
+      "step": 82495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3182876624885954e-05,
+      "loss": 1.375,
+      "step": 82500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.318216023460193e-05,
+      "loss": 1.1751,
+      "step": 82505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3181443826146057e-05,
+      "loss": 1.2108,
+      "step": 82510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3180727399522427e-05,
+      "loss": 1.3284,
+      "step": 82515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3180010954735136e-05,
+      "loss": 1.4389,
+      "step": 82520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3179294491788272e-05,
+      "loss": 1.3698,
+      "step": 82525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3178578010685923e-05,
+      "loss": 1.2554,
+      "step": 82530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3177861511432188e-05,
+      "loss": 1.2269,
+      "step": 82535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3177144994031153e-05,
+      "loss": 1.1017,
+      "step": 82540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3176428458486911e-05,
+      "loss": 1.1133,
+      "step": 82545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3175711904803554e-05,
+      "loss": 1.0422,
+      "step": 82550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3174995332985174e-05,
+      "loss": 1.2909,
+      "step": 82555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3174278743035865e-05,
+      "loss": 1.0814,
+      "step": 82560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3173562134959714e-05,
+      "loss": 1.0781,
+      "step": 82565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.317284550876082e-05,
+      "loss": 1.3228,
+      "step": 82570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3172128864443266e-05,
+      "loss": 1.1826,
+      "step": 82575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3171412202011153e-05,
+      "loss": 1.3528,
+      "step": 82580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3170695521468572e-05,
+      "loss": 1.1291,
+      "step": 82585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3169978822819612e-05,
+      "loss": 1.064,
+      "step": 82590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.316926210606837e-05,
+      "loss": 1.4443,
+      "step": 82595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3168545371218934e-05,
+      "loss": 1.2637,
+      "step": 82600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3167828618275402e-05,
+      "loss": 1.299,
+      "step": 82605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3167111847241864e-05,
+      "loss": 1.0587,
+      "step": 82610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3166395058122414e-05,
+      "loss": 1.4133,
+      "step": 82615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3165678250921144e-05,
+      "loss": 1.2673,
+      "step": 82620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3164961425642154e-05,
+      "loss": 1.1035,
+      "step": 82625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3164244582289527e-05,
+      "loss": 1.4388,
+      "step": 82630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.316352772086736e-05,
+      "loss": 1.2566,
+      "step": 82635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3162810841379752e-05,
+      "loss": 1.2415,
+      "step": 82640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.316209394383079e-05,
+      "loss": 1.1616,
+      "step": 82645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3161377028224576e-05,
+      "loss": 1.6612,
+      "step": 82650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3160660094565192e-05,
+      "loss": 1.5841,
+      "step": 82655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3159943142856742e-05,
+      "loss": 1.4226,
+      "step": 82660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.315922617310332e-05,
+      "loss": 1.2141,
+      "step": 82665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3158509185309012e-05,
+      "loss": 1.1835,
+      "step": 82670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3157792179477919e-05,
+      "loss": 1.4524,
+      "step": 82675
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3157075155614137e-05,
+      "loss": 1.5064,
+      "step": 82680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3156358113721756e-05,
+      "loss": 1.0022,
+      "step": 82685
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3155641053804873e-05,
+      "loss": 1.1091,
+      "step": 82690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3154923975867581e-05,
+      "loss": 1.5163,
+      "step": 82695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3154206879913978e-05,
+      "loss": 0.9795,
+      "step": 82700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3153489765948156e-05,
+      "loss": 1.3931,
+      "step": 82705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3152772633974214e-05,
+      "loss": 1.1103,
+      "step": 82710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3152055483996241e-05,
+      "loss": 1.4465,
+      "step": 82715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3151338316018336e-05,
+      "loss": 1.1803,
+      "step": 82720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3150621130044596e-05,
+      "loss": 0.9178,
+      "step": 82725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3149903926079115e-05,
+      "loss": 1.0309,
+      "step": 82730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.314918670412599e-05,
+      "loss": 1.2432,
+      "step": 82735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3148469464189308e-05,
+      "loss": 1.9327,
+      "step": 82740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3147752206273178e-05,
+      "loss": 1.2556,
+      "step": 82745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3147034930381688e-05,
+      "loss": 1.4739,
+      "step": 82750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3146317636518939e-05,
+      "loss": 1.1145,
+      "step": 82755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3145600324689021e-05,
+      "loss": 1.2654,
+      "step": 82760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3144882994896034e-05,
+      "loss": 1.1117,
+      "step": 82765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3144165647144074e-05,
+      "loss": 1.1853,
+      "step": 82770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3143448281437234e-05,
+      "loss": 1.2401,
+      "step": 82775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3142730897779617e-05,
+      "loss": 1.1898,
+      "step": 82780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3142013496175317e-05,
+      "loss": 1.2394,
+      "step": 82785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3141296076628428e-05,
+      "loss": 1.186,
+      "step": 82790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3140578639143046e-05,
+      "loss": 1.1775,
+      "step": 82795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3139861183723275e-05,
+      "loss": 1.4024,
+      "step": 82800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3139143710373207e-05,
+      "loss": 1.2151,
+      "step": 82805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3138426219096943e-05,
+      "loss": 1.4189,
+      "step": 82810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.313770870989857e-05,
+      "loss": 1.133,
+      "step": 82815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3136991182782197e-05,
+      "loss": 1.2932,
+      "step": 82820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.313627363775192e-05,
+      "loss": 1.0156,
+      "step": 82825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.313555607481183e-05,
+      "loss": 1.1432,
+      "step": 82830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3134838493966029e-05,
+      "loss": 1.2792,
+      "step": 82835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3134120895218614e-05,
+      "loss": 1.1807,
+      "step": 82840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3133403278573682e-05,
+      "loss": 1.0981,
+      "step": 82845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3132685644035334e-05,
+      "loss": 1.1816,
+      "step": 82850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3131967991607663e-05,
+      "loss": 1.1315,
+      "step": 82855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3131250321294775e-05,
+      "loss": 1.1749,
+      "step": 82860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3130532633100761e-05,
+      "loss": 1.252,
+      "step": 82865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3129814927029721e-05,
+      "loss": 0.9122,
+      "step": 82870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3129097203085757e-05,
+      "loss": 1.1099,
+      "step": 82875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.312837946127296e-05,
+      "loss": 1.2603,
+      "step": 82880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3127661701595438e-05,
+      "loss": 1.0315,
+      "step": 82885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3126943924057282e-05,
+      "loss": 1.2463,
+      "step": 82890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3126226128662598e-05,
+      "loss": 1.2482,
+      "step": 82895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3125508315415481e-05,
+      "loss": 0.7692,
+      "step": 82900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3124790484320028e-05,
+      "loss": 1.0707,
+      "step": 82905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3124072635380343e-05,
+      "loss": 1.3371,
+      "step": 82910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.312335476860052e-05,
+      "loss": 1.3292,
+      "step": 82915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3122636883984665e-05,
+      "loss": 1.2384,
+      "step": 82920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3121918981536871e-05,
+      "loss": 1.2162,
+      "step": 82925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3121201061261245e-05,
+      "loss": 1.2647,
+      "step": 82930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3120483123161878e-05,
+      "loss": 1.2995,
+      "step": 82935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3119765167242875e-05,
+      "loss": 1.5123,
+      "step": 82940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3119047193508334e-05,
+      "loss": 1.0609,
+      "step": 82945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3118329201962357e-05,
+      "loss": 1.338,
+      "step": 82950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3117611192609042e-05,
+      "loss": 1.2588,
+      "step": 82955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3116893165452491e-05,
+      "loss": 1.0992,
+      "step": 82960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3116175120496803e-05,
+      "loss": 1.0282,
+      "step": 82965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3115457057746082e-05,
+      "loss": 1.1481,
+      "step": 82970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3114738977204422e-05,
+      "loss": 1.0878,
+      "step": 82975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3114020878875931e-05,
+      "loss": 1.3537,
+      "step": 82980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3113302762764701e-05,
+      "loss": 1.378,
+      "step": 82985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3112584628874841e-05,
+      "loss": 1.5246,
+      "step": 82990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3111866477210447e-05,
+      "loss": 1.2742,
+      "step": 82995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3111148307775624e-05,
+      "loss": 1.2452,
+      "step": 83000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3110430120574469e-05,
+      "loss": 1.1023,
+      "step": 83005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3109711915611084e-05,
+      "loss": 1.0707,
+      "step": 83010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3108993692889573e-05,
+      "loss": 1.238,
+      "step": 83015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3108275452414034e-05,
+      "loss": 1.3678,
+      "step": 83020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3107557194188573e-05,
+      "loss": 1.4836,
+      "step": 83025
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3106838918217288e-05,
+      "loss": 1.4163,
+      "step": 83030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3106120624504281e-05,
+      "loss": 1.3275,
+      "step": 83035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3105402313053657e-05,
+      "loss": 1.0452,
+      "step": 83040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3104683983869512e-05,
+      "loss": 1.2407,
+      "step": 83045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3103965636955955e-05,
+      "loss": 1.3306,
+      "step": 83050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3103247272317081e-05,
+      "loss": 1.122,
+      "step": 83055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3102528889957e-05,
+      "loss": 1.3326,
+      "step": 83060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3101810489879809e-05,
+      "loss": 1.2386,
+      "step": 83065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.310109207208961e-05,
+      "loss": 1.3978,
+      "step": 83070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.310037363659051e-05,
+      "loss": 1.2335,
+      "step": 83075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3099655183386605e-05,
+      "loss": 1.1903,
+      "step": 83080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3098936712482009e-05,
+      "loss": 1.1233,
+      "step": 83085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.309821822388081e-05,
+      "loss": 1.3192,
+      "step": 83090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3097499717587121e-05,
+      "loss": 1.1622,
+      "step": 83095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3096781193605045e-05,
+      "loss": 1.5734,
+      "step": 83100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3096062651938679e-05,
+      "loss": 1.4869,
+      "step": 83105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3095344092592133e-05,
+      "loss": 1.1485,
+      "step": 83110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3094625515569505e-05,
+      "loss": 1.0286,
+      "step": 83115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3093906920874904e-05,
+      "loss": 1.6113,
+      "step": 83120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3093188308512428e-05,
+      "loss": 1.1424,
+      "step": 83125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3092469678486183e-05,
+      "loss": 1.0134,
+      "step": 83130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3091751030800274e-05,
+      "loss": 1.2033,
+      "step": 83135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3091032365458802e-05,
+      "loss": 1.1945,
+      "step": 83140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3090313682465877e-05,
+      "loss": 0.9731,
+      "step": 83145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3089594981825596e-05,
+      "loss": 1.0512,
+      "step": 83150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3088876263542062e-05,
+      "loss": 1.1668,
+      "step": 83155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3088157527619388e-05,
+      "loss": 1.312,
+      "step": 83160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3087438774061674e-05,
+      "loss": 1.2656,
+      "step": 83165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3086720002873022e-05,
+      "loss": 1.3392,
+      "step": 83170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3086001214057537e-05,
+      "loss": 1.1664,
+      "step": 83175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3085282407619328e-05,
+      "loss": 1.218,
+      "step": 83180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3084563583562496e-05,
+      "loss": 1.0517,
+      "step": 83185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3083844741891148e-05,
+      "loss": 1.1234,
+      "step": 83190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3083125882609385e-05,
+      "loss": 1.2131,
+      "step": 83195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3082407005721316e-05,
+      "loss": 1.0665,
+      "step": 83200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.308168811123105e-05,
+      "loss": 0.7728,
+      "step": 83205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3080969199142681e-05,
+      "loss": 1.0541,
+      "step": 83210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3080250269460324e-05,
+      "loss": 1.3878,
+      "step": 83215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.307953132218808e-05,
+      "loss": 1.1827,
+      "step": 83220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3078812357330055e-05,
+      "loss": 1.3474,
+      "step": 83225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3078093374890357e-05,
+      "loss": 1.1632,
+      "step": 83230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3077374374873091e-05,
+      "loss": 1.2191,
+      "step": 83235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3076655357282361e-05,
+      "loss": 1.3166,
+      "step": 83240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3075936322122274e-05,
+      "loss": 0.9877,
+      "step": 83245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3075217269396938e-05,
+      "loss": 1.2228,
+      "step": 83250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3074498199110452e-05,
+      "loss": 1.2603,
+      "step": 83255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3073779111266935e-05,
+      "loss": 1.2075,
+      "step": 83260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3073060005870481e-05,
+      "loss": 1.481,
+      "step": 83265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3072340882925202e-05,
+      "loss": 1.1624,
+      "step": 83270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3071621742435208e-05,
+      "loss": 1.4471,
+      "step": 83275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3070902584404596e-05,
+      "loss": 1.0316,
+      "step": 83280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3070183408837485e-05,
+      "loss": 1.2883,
+      "step": 83285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3069464215737968e-05,
+      "loss": 1.5963,
+      "step": 83290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3068745005110164e-05,
+      "loss": 1.4277,
+      "step": 83295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3068025776958175e-05,
+      "loss": 1.5339,
+      "step": 83300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3067306531286108e-05,
+      "loss": 1.2141,
+      "step": 83305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3066587268098075e-05,
+      "loss": 1.2836,
+      "step": 83310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3065867987398173e-05,
+      "loss": 1.2113,
+      "step": 83315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3065148689190518e-05,
+      "loss": 1.5168,
+      "step": 83320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3064429373479218e-05,
+      "loss": 1.4212,
+      "step": 83325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3063710040268376e-05,
+      "loss": 1.0499,
+      "step": 83330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.30629906895621e-05,
+      "loss": 1.4096,
+      "step": 83335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3062271321364501e-05,
+      "loss": 1.8411,
+      "step": 83340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3061551935679688e-05,
+      "loss": 1.3949,
+      "step": 83345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3060832532511765e-05,
+      "loss": 1.37,
+      "step": 83350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.306011311186484e-05,
+      "loss": 1.1029,
+      "step": 83355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3059393673743027e-05,
+      "loss": 1.3335,
+      "step": 83360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3058674218150428e-05,
+      "loss": 1.1837,
+      "step": 83365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3057954745091157e-05,
+      "loss": 0.8742,
+      "step": 83370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3057235254569318e-05,
+      "loss": 1.3865,
+      "step": 83375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.305651574658902e-05,
+      "loss": 1.1629,
+      "step": 83380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3055796221154374e-05,
+      "loss": 1.137,
+      "step": 83385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3055076678269489e-05,
+      "loss": 1.0623,
+      "step": 83390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.305435711793847e-05,
+      "loss": 1.0251,
+      "step": 83395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.305363754016543e-05,
+      "loss": 1.1137,
+      "step": 83400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3052917944954479e-05,
+      "loss": 1.2233,
+      "step": 83405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3052198332309724e-05,
+      "loss": 1.1158,
+      "step": 83410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3051478702235279e-05,
+      "loss": 1.0781,
+      "step": 83415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3050759054735242e-05,
+      "loss": 1.7103,
+      "step": 83420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3050039389813733e-05,
+      "loss": 1.1433,
+      "step": 83425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.304931970747486e-05,
+      "loss": 1.0142,
+      "step": 83430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3048600007722729e-05,
+      "loss": 1.1122,
+      "step": 83435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3047880290561454e-05,
+      "loss": 1.3116,
+      "step": 83440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3047160555995142e-05,
+      "loss": 1.1571,
+      "step": 83445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3046440804027904e-05,
+      "loss": 1.0822,
+      "step": 83450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3045721034663851e-05,
+      "loss": 1.3269,
+      "step": 83455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3045001247907095e-05,
+      "loss": 1.0941,
+      "step": 83460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.304428144376174e-05,
+      "loss": 1.2144,
+      "step": 83465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3043561622231906e-05,
+      "loss": 0.9471,
+      "step": 83470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3042841783321696e-05,
+      "loss": 1.05,
+      "step": 83475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.304212192703522e-05,
+      "loss": 1.2195,
+      "step": 83480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3041402053376595e-05,
+      "loss": 1.1038,
+      "step": 83485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3040682162349925e-05,
+      "loss": 1.6081,
+      "step": 83490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3039962253959329e-05,
+      "loss": 1.0562,
+      "step": 83495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3039242328208908e-05,
+      "loss": 1.165,
+      "step": 83500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3038522385102781e-05,
+      "loss": 1.0951,
+      "step": 83505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3037802424645062e-05,
+      "loss": 1.0817,
+      "step": 83510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3037082446839848e-05,
+      "loss": 1.3455,
+      "step": 83515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3036362451691267e-05,
+      "loss": 1.4283,
+      "step": 83520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3035642439203418e-05,
+      "loss": 1.5623,
+      "step": 83525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.303492240938042e-05,
+      "loss": 1.1901,
+      "step": 83530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3034202362226384e-05,
+      "loss": 1.0893,
+      "step": 83535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3033482297745419e-05,
+      "loss": 1.3025,
+      "step": 83540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3032762215941639e-05,
+      "loss": 1.011,
+      "step": 83545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3032042116819157e-05,
+      "loss": 1.3747,
+      "step": 83550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3031322000382083e-05,
+      "loss": 1.3809,
+      "step": 83555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.303060186663453e-05,
+      "loss": 1.1724,
+      "step": 83560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.302988171558061e-05,
+      "loss": 1.1992,
+      "step": 83565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3029161547224437e-05,
+      "loss": 1.1288,
+      "step": 83570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.302844136157012e-05,
+      "loss": 1.2437,
+      "step": 83575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3027721158621773e-05,
+      "loss": 1.2104,
+      "step": 83580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3027000938383512e-05,
+      "loss": 1.2987,
+      "step": 83585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3026280700859448e-05,
+      "loss": 1.2629,
+      "step": 83590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3025560446053691e-05,
+      "loss": 1.1957,
+      "step": 83595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3024840173970361e-05,
+      "loss": 1.1719,
+      "step": 83600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3024119884613561e-05,
+      "loss": 1.2303,
+      "step": 83605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3023399577987414e-05,
+      "loss": 1.1615,
+      "step": 83610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3022679254096031e-05,
+      "loss": 1.8557,
+      "step": 83615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3021958912943517e-05,
+      "loss": 1.4828,
+      "step": 83620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3021238554533999e-05,
+      "loss": 1.1772,
+      "step": 83625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3020518178871579e-05,
+      "loss": 1.2086,
+      "step": 83630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3019797785960378e-05,
+      "loss": 0.9857,
+      "step": 83635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3019077375804508e-05,
+      "loss": 1.3396,
+      "step": 83640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3018356948408082e-05,
+      "loss": 1.2414,
+      "step": 83645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3017636503775213e-05,
+      "loss": 1.4007,
+      "step": 83650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3016916041910017e-05,
+      "loss": 1.1657,
+      "step": 83655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3016195562816609e-05,
+      "loss": 1.1672,
+      "step": 83660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.30154750664991e-05,
+      "loss": 1.2669,
+      "step": 83665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3014754552961607e-05,
+      "loss": 1.4664,
+      "step": 83670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3014034022208246e-05,
+      "loss": 1.3225,
+      "step": 83675
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3013313474243127e-05,
+      "loss": 1.3211,
+      "step": 83680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.301259290907037e-05,
+      "loss": 1.1241,
+      "step": 83685
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3011872326694084e-05,
+      "loss": 0.9473,
+      "step": 83690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3011151727118392e-05,
+      "loss": 1.4111,
+      "step": 83695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3010431110347397e-05,
+      "loss": 1.0222,
+      "step": 83700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3009710476385228e-05,
+      "loss": 1.3993,
+      "step": 83705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3008989825235991e-05,
+      "loss": 1.43,
+      "step": 83710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3008269156903801e-05,
+      "loss": 1.5215,
+      "step": 83715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3007548471392782e-05,
+      "loss": 1.3961,
+      "step": 83720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3006827768707036e-05,
+      "loss": 1.3882,
+      "step": 83725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3006107048850692e-05,
+      "loss": 1.1285,
+      "step": 83730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3005386311827857e-05,
+      "loss": 1.106,
+      "step": 83735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.300466555764265e-05,
+      "loss": 1.2661,
+      "step": 83740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3003944786299187e-05,
+      "loss": 1.3592,
+      "step": 83745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3003223997801585e-05,
+      "loss": 1.0687,
+      "step": 83750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3002503192153959e-05,
+      "loss": 1.1961,
+      "step": 83755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3001782369360423e-05,
+      "loss": 1.0706,
+      "step": 83760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3001061529425096e-05,
+      "loss": 1.2871,
+      "step": 83765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3000340672352091e-05,
+      "loss": 1.3376,
+      "step": 83770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2999619798145531e-05,
+      "loss": 1.1695,
+      "step": 83775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2998898906809527e-05,
+      "loss": 1.3119,
+      "step": 83780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2998177998348198e-05,
+      "loss": 1.1851,
+      "step": 83785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.299745707276566e-05,
+      "loss": 1.1639,
+      "step": 83790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2996736130066026e-05,
+      "loss": 1.218,
+      "step": 83795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2996015170253422e-05,
+      "loss": 0.9811,
+      "step": 83800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2995294193331956e-05,
+      "loss": 1.2981,
+      "step": 83805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2994573199305753e-05,
+      "loss": 1.1084,
+      "step": 83810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2993852188178923e-05,
+      "loss": 1.6099,
+      "step": 83815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2993131159955588e-05,
+      "loss": 1.4177,
+      "step": 83820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2992410114639867e-05,
+      "loss": 0.8728,
+      "step": 83825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.299168905223587e-05,
+      "loss": 1.0627,
+      "step": 83830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2990967972747722e-05,
+      "loss": 1.5341,
+      "step": 83835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2990246876179538e-05,
+      "loss": 1.053,
+      "step": 83840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2989525762535436e-05,
+      "loss": 1.4563,
+      "step": 83845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2988804631819535e-05,
+      "loss": 1.0692,
+      "step": 83850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.298808348403595e-05,
+      "loss": 1.251,
+      "step": 83855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.29873623191888e-05,
+      "loss": 1.6111,
+      "step": 83860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2986641137282208e-05,
+      "loss": 1.3744,
+      "step": 83865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2985919938320287e-05,
+      "loss": 1.3716,
+      "step": 83870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2985198722307158e-05,
+      "loss": 1.3206,
+      "step": 83875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2984477489246938e-05,
+      "loss": 1.3283,
+      "step": 83880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2983756239143746e-05,
+      "loss": 0.8886,
+      "step": 83885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2983034972001702e-05,
+      "loss": 1.185,
+      "step": 83890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2982313687824924e-05,
+      "loss": 1.17,
+      "step": 83895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2981592386617527e-05,
+      "loss": 1.5049,
+      "step": 83900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2980871068383638e-05,
+      "loss": 1.5021,
+      "step": 83905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2980149733127372e-05,
+      "loss": 0.9138,
+      "step": 83910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2979428380852847e-05,
+      "loss": 1.4881,
+      "step": 83915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2978707011564184e-05,
+      "loss": 1.2255,
+      "step": 83920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2977985625265499e-05,
+      "loss": 1.0668,
+      "step": 83925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.297726422196092e-05,
+      "loss": 1.2752,
+      "step": 83930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2976542801654555e-05,
+      "loss": 1.2734,
+      "step": 83935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2975821364350532e-05,
+      "loss": 1.3011,
+      "step": 83940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2975099910052969e-05,
+      "loss": 1.3352,
+      "step": 83945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2974378438765984e-05,
+      "loss": 1.193,
+      "step": 83950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2973656950493702e-05,
+      "loss": 1.195,
+      "step": 83955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2972935445240236e-05,
+      "loss": 1.3239,
+      "step": 83960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2972213923009709e-05,
+      "loss": 0.9991,
+      "step": 83965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2971492383806243e-05,
+      "loss": 1.061,
+      "step": 83970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2970770827633959e-05,
+      "loss": 1.4415,
+      "step": 83975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2970049254496975e-05,
+      "loss": 1.5,
+      "step": 83980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2969327664399412e-05,
+      "loss": 1.2929,
+      "step": 83985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.296860605734539e-05,
+      "loss": 1.5058,
+      "step": 83990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2967884433339031e-05,
+      "loss": 1.2965,
+      "step": 83995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2967162792384456e-05,
+      "loss": 1.2004,
+      "step": 84000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2966441134485783e-05,
+      "loss": 1.2109,
+      "step": 84005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.296571945964714e-05,
+      "loss": 1.1739,
+      "step": 84010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2964997767872642e-05,
+      "loss": 1.0397,
+      "step": 84015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.296427605916641e-05,
+      "loss": 1.1611,
+      "step": 84020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.296355433353257e-05,
+      "loss": 1.2712,
+      "step": 84025
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2962832590975241e-05,
+      "loss": 1.2774,
+      "step": 84030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2962110831498544e-05,
+      "loss": 1.1714,
+      "step": 84035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2961389055106599e-05,
+      "loss": 1.4316,
+      "step": 84040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.296066726180353e-05,
+      "loss": 1.1959,
+      "step": 84045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2959945451593463e-05,
+      "loss": 1.2065,
+      "step": 84050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.295922362448051e-05,
+      "loss": 1.5702,
+      "step": 84055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2958501780468801e-05,
+      "loss": 1.1581,
+      "step": 84060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2957779919562455e-05,
+      "loss": 1.5102,
+      "step": 84065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2957058041765596e-05,
+      "loss": 1.2982,
+      "step": 84070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2956336147082342e-05,
+      "loss": 1.0753,
+      "step": 84075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2955614235516822e-05,
+      "loss": 1.0951,
+      "step": 84080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2954892307073155e-05,
+      "loss": 1.0233,
+      "step": 84085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2954170361755461e-05,
+      "loss": 1.0717,
+      "step": 84090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2953448399567867e-05,
+      "loss": 1.4665,
+      "step": 84095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2952726420514494e-05,
+      "loss": 1.3817,
+      "step": 84100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2952004424599464e-05,
+      "loss": 1.1609,
+      "step": 84105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2951282411826902e-05,
+      "loss": 1.3431,
+      "step": 84110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.295056038220093e-05,
+      "loss": 1.0621,
+      "step": 84115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.294983833572567e-05,
+      "loss": 1.2476,
+      "step": 84120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2949116272405247e-05,
+      "loss": 1.0508,
+      "step": 84125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2948394192243787e-05,
+      "loss": 1.3641,
+      "step": 84130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2947672095245406e-05,
+      "loss": 1.0759,
+      "step": 84135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2946949981414237e-05,
+      "loss": 1.0931,
+      "step": 84140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2946227850754392e-05,
+      "loss": 1.4692,
+      "step": 84145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2945505703270004e-05,
+      "loss": 1.217,
+      "step": 84150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2944783538965198e-05,
+      "loss": 1.1912,
+      "step": 84155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2944061357844088e-05,
+      "loss": 1.0971,
+      "step": 84160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2943339159910808e-05,
+      "loss": 1.2623,
+      "step": 84165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2942616945169477e-05,
+      "loss": 1.1519,
+      "step": 84170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.294189471362422e-05,
+      "loss": 1.1101,
+      "step": 84175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2941172465279165e-05,
+      "loss": 1.1562,
+      "step": 84180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.294045020013843e-05,
+      "loss": 0.9686,
+      "step": 84185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2939727918206144e-05,
+      "loss": 1.0581,
+      "step": 84190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2939005619486431e-05,
+      "loss": 1.2433,
+      "step": 84195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2938283303983416e-05,
+      "loss": 1.3377,
+      "step": 84200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2937560971701222e-05,
+      "loss": 1.0185,
+      "step": 84205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2936838622643974e-05,
+      "loss": 1.0277,
+      "step": 84210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2936116256815799e-05,
+      "loss": 1.3968,
+      "step": 84215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.293539387422082e-05,
+      "loss": 1.3418,
+      "step": 84220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2934671474863163e-05,
+      "loss": 1.4889,
+      "step": 84225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2933949058746953e-05,
+      "loss": 1.0311,
+      "step": 84230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2933226625876319e-05,
+      "loss": 1.0867,
+      "step": 84235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2932504176255379e-05,
+      "loss": 1.1766,
+      "step": 84240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2931781709888267e-05,
+      "loss": 1.0094,
+      "step": 84245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2931059226779099e-05,
+      "loss": 1.1763,
+      "step": 84250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.293033672693201e-05,
+      "loss": 1.105,
+      "step": 84255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2929614210351126e-05,
+      "loss": 1.1596,
+      "step": 84260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2928891677040563e-05,
+      "loss": 1.0438,
+      "step": 84265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2928169127004455e-05,
+      "loss": 1.4128,
+      "step": 84270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2927446560246927e-05,
+      "loss": 1.4292,
+      "step": 84275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2926723976772104e-05,
+      "loss": 1.3589,
+      "step": 84280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2926001376584113e-05,
+      "loss": 1.7172,
+      "step": 84285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2925278759687084e-05,
+      "loss": 1.0634,
+      "step": 84290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2924556126085135e-05,
+      "loss": 1.17,
+      "step": 84295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2923833475782401e-05,
+      "loss": 1.1643,
+      "step": 84300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2923110808783003e-05,
+      "loss": 1.2471,
+      "step": 84305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2922388125091068e-05,
+      "loss": 1.2834,
+      "step": 84310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.292166542471073e-05,
+      "loss": 1.1477,
+      "step": 84315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2920942707646109e-05,
+      "loss": 1.353,
+      "step": 84320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2920219973901333e-05,
+      "loss": 1.3201,
+      "step": 84325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2919497223480533e-05,
+      "loss": 1.2776,
+      "step": 84330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2918774456387828e-05,
+      "loss": 1.1626,
+      "step": 84335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.291805167262736e-05,
+      "loss": 1.0639,
+      "step": 84340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2917328872203237e-05,
+      "loss": 1.454,
+      "step": 84345
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2916606055119604e-05,
+      "loss": 1.2423,
+      "step": 84350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2915883221380582e-05,
+      "loss": 1.0199,
+      "step": 84355
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2915160370990294e-05,
+      "loss": 1.313,
+      "step": 84360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2914437503952876e-05,
+      "loss": 1.091,
+      "step": 84365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2913714620272449e-05,
+      "loss": 1.2262,
+      "step": 84370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2912991719953145e-05,
+      "loss": 1.1671,
+      "step": 84375
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2912268802999095e-05,
+      "loss": 1.0055,
+      "step": 84380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2911545869414422e-05,
+      "loss": 0.9411,
+      "step": 84385
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2910822919203257e-05,
+      "loss": 1.2888,
+      "step": 84390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2910099952369723e-05,
+      "loss": 0.9868,
+      "step": 84395
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2909376968917959e-05,
+      "loss": 1.1385,
+      "step": 84400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2908653968852084e-05,
+      "loss": 1.2321,
+      "step": 84405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.290793095217623e-05,
+      "loss": 1.1484,
+      "step": 84410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2907207918894531e-05,
+      "loss": 0.9813,
+      "step": 84415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2906484869011106e-05,
+      "loss": 1.2083,
+      "step": 84420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2905761802530093e-05,
+      "loss": 1.1486,
+      "step": 84425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2905038719455614e-05,
+      "loss": 1.0755,
+      "step": 84430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2904315619791802e-05,
+      "loss": 1.2513,
+      "step": 84435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2903592503542786e-05,
+      "loss": 1.1446,
+      "step": 84440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2902869370712695e-05,
+      "loss": 1.1673,
+      "step": 84445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2902146221305657e-05,
+      "loss": 1.4947,
+      "step": 84450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2901423055325808e-05,
+      "loss": 1.0505,
+      "step": 84455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2900699872777271e-05,
+      "loss": 1.1721,
+      "step": 84460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2899976673664174e-05,
+      "loss": 1.1647,
+      "step": 84465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2899253457990657e-05,
+      "loss": 1.1951,
+      "step": 84470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2898530225760837e-05,
+      "loss": 1.2953,
+      "step": 84475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2897806976978852e-05,
+      "loss": 1.224,
+      "step": 84480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2897083711648833e-05,
+      "loss": 1.2877,
+      "step": 84485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2896360429774905e-05,
+      "loss": 1.4864,
+      "step": 84490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2895637131361204e-05,
+      "loss": 1.3539,
+      "step": 84495
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2894913816411854e-05,
+      "loss": 1.3996,
+      "step": 84500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.289419048493099e-05,
+      "loss": 0.9461,
+      "step": 84505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2893467136922743e-05,
+      "loss": 1.2523,
+      "step": 84510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2892743772391242e-05,
+      "loss": 1.0778,
+      "step": 84515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.289202039134062e-05,
+      "loss": 1.9109,
+      "step": 84520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2891296993775004e-05,
+      "loss": 0.973,
+      "step": 84525
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2890573579698528e-05,
+      "loss": 0.9335,
+      "step": 84530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2889850149115322e-05,
+      "loss": 1.3048,
+      "step": 84535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2889126702029517e-05,
+      "loss": 1.1752,
+      "step": 84540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2888403238445241e-05,
+      "loss": 1.1511,
+      "step": 84545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2887679758366636e-05,
+      "loss": 1.1316,
+      "step": 84550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2886956261797823e-05,
+      "loss": 1.1955,
+      "step": 84555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2886232748742935e-05,
+      "loss": 1.3796,
+      "step": 84560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.288550921920611e-05,
+      "loss": 1.4321,
+      "step": 84565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.288478567319147e-05,
+      "loss": 1.151,
+      "step": 84570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2884062110703158e-05,
+      "loss": 1.4833,
+      "step": 84575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2883338531745295e-05,
+      "loss": 1.4521,
+      "step": 84580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2882614936322022e-05,
+      "loss": 1.1901,
+      "step": 84585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2881891324437466e-05,
+      "loss": 1.7831,
+      "step": 84590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2881167696095758e-05,
+      "loss": 1.0979,
+      "step": 84595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2880444051301035e-05,
+      "loss": 1.0515,
+      "step": 84600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2879720390057428e-05,
+      "loss": 1.2153,
+      "step": 84605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.287899671236907e-05,
+      "loss": 1.119,
+      "step": 84610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2878273018240088e-05,
+      "loss": 1.4335,
+      "step": 84615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2877549307674624e-05,
+      "loss": 1.3552,
+      "step": 84620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2876825580676801e-05,
+      "loss": 1.2712,
+      "step": 84625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.287610183725076e-05,
+      "loss": 1.4147,
+      "step": 84630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.287537807740063e-05,
+      "loss": 1.232,
+      "step": 84635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.287465430113054e-05,
+      "loss": 1.1075,
+      "step": 84640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2873930508444635e-05,
+      "loss": 1.4721,
+      "step": 84645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2873206699347036e-05,
+      "loss": 1.637,
+      "step": 84650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2872482873841886e-05,
+      "loss": 1.3481,
+      "step": 84655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2871759031933309e-05,
+      "loss": 1.4181,
+      "step": 84660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2871035173625444e-05,
+      "loss": 1.1288,
+      "step": 84665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2870311298922426e-05,
+      "loss": 0.9594,
+      "step": 84670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2869587407828386e-05,
+      "loss": 1.3171,
+      "step": 84675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.286886350034746e-05,
+      "loss": 1.1834,
+      "step": 84680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2868139576483776e-05,
+      "loss": 1.4389,
+      "step": 84685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2867415636241476e-05,
+      "loss": 1.0328,
+      "step": 84690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2866691679624692e-05,
+      "loss": 1.1445,
+      "step": 84695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2865967706637553e-05,
+      "loss": 1.1575,
+      "step": 84700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2865243717284198e-05,
+      "loss": 1.3,
+      "step": 84705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.286451971156876e-05,
+      "loss": 1.1724,
+      "step": 84710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2863795689495375e-05,
+      "loss": 1.6984,
+      "step": 84715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2863071651068175e-05,
+      "loss": 1.0339,
+      "step": 84720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2862347596291299e-05,
+      "loss": 1.0173,
+      "step": 84725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2861623525168874e-05,
+      "loss": 1.02,
+      "step": 84730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2860899437705041e-05,
+      "loss": 1.1196,
+      "step": 84735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2860175333903934e-05,
+      "loss": 1.278,
+      "step": 84740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2859451213769686e-05,
+      "loss": 1.1035,
+      "step": 84745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2858727077306439e-05,
+      "loss": 1.2505,
+      "step": 84750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.285800292451832e-05,
+      "loss": 1.1279,
+      "step": 84755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2857278755409466e-05,
+      "loss": 1.1398,
+      "step": 84760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2856554569984014e-05,
+      "loss": 1.4104,
+      "step": 84765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2855830368246097e-05,
+      "loss": 1.3698,
+      "step": 84770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2855106150199856e-05,
+      "loss": 0.993,
+      "step": 84775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2854381915849419e-05,
+      "loss": 1.4949,
+      "step": 84780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.285365766519893e-05,
+      "loss": 1.1854,
+      "step": 84785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2852933398252519e-05,
+      "loss": 1.034,
+      "step": 84790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2852209115014325e-05,
+      "loss": 1.0811,
+      "step": 84795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2851484815488485e-05,
+      "loss": 1.4521,
+      "step": 84800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2850760499679126e-05,
+      "loss": 0.9902,
+      "step": 84805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2850036167590399e-05,
+      "loss": 1.2097,
+      "step": 84810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2849311819226429e-05,
+      "loss": 1.2281,
+      "step": 84815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2848587454591355e-05,
+      "loss": 1.192,
+      "step": 84820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2847863073689317e-05,
+      "loss": 1.58,
+      "step": 84825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2847138676524446e-05,
+      "loss": 1.1083,
+      "step": 84830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2846414263100884e-05,
+      "loss": 1.1843,
+      "step": 84835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2845689833422766e-05,
+      "loss": 1.4724,
+      "step": 84840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2844965387494228e-05,
+      "loss": 1.1654,
+      "step": 84845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2844240925319404e-05,
+      "loss": 1.1078,
+      "step": 84850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.284351644690244e-05,
+      "loss": 1.0587,
+      "step": 84855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2842791952247467e-05,
+      "loss": 1.4374,
+      "step": 84860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2842067441358621e-05,
+      "loss": 1.149,
+      "step": 84865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2841342914240043e-05,
+      "loss": 1.0799,
+      "step": 84870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2840618370895866e-05,
+      "loss": 1.4787,
+      "step": 84875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2839893811330234e-05,
+      "loss": 1.287,
+      "step": 84880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2839169235547275e-05,
+      "loss": 1.6369,
+      "step": 84885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2838444643551136e-05,
+      "loss": 0.987,
+      "step": 84890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2837720035345954e-05,
+      "loss": 1.0983,
+      "step": 84895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.283699541093586e-05,
+      "loss": 1.1018,
+      "step": 84900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2836270770325e-05,
+      "loss": 1.1115,
+      "step": 84905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2835546113517506e-05,
+      "loss": 1.3266,
+      "step": 84910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2834821440517518e-05,
+      "loss": 1.2609,
+      "step": 84915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2834096751329177e-05,
+      "loss": 1.1658,
+      "step": 84920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2833372045956618e-05,
+      "loss": 1.3965,
+      "step": 84925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2832647324403981e-05,
+      "loss": 1.3245,
+      "step": 84930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2831922586675406e-05,
+      "loss": 1.4921,
+      "step": 84935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2831197832775027e-05,
+      "loss": 1.5269,
+      "step": 84940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2830473062706987e-05,
+      "loss": 1.1355,
+      "step": 84945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2829748276475423e-05,
+      "loss": 0.9828,
+      "step": 84950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2829023474084474e-05,
+      "loss": 1.0085,
+      "step": 84955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.282829865553828e-05,
+      "loss": 1.1509,
+      "step": 84960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.282757382084098e-05,
+      "loss": 1.2732,
+      "step": 84965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2826848969996712e-05,
+      "loss": 1.0878,
+      "step": 84970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2826124103009616e-05,
+      "loss": 1.1383,
+      "step": 84975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2825399219883829e-05,
+      "loss": 1.1449,
+      "step": 84980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2824674320623499e-05,
+      "loss": 1.2377,
+      "step": 84985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2823949405232752e-05,
+      "loss": 1.2584,
+      "step": 84990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2823224473715739e-05,
+      "loss": 1.2024,
+      "step": 84995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2822499526076597e-05,
+      "loss": 1.2009,
+      "step": 85000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.282177456231946e-05,
+      "loss": 1.1764,
+      "step": 85005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.282104958244848e-05,
+      "loss": 0.8872,
+      "step": 85010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2820324586467783e-05,
+      "loss": 1.0547,
+      "step": 85015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2819599574381516e-05,
+      "loss": 0.9206,
+      "step": 85020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2818874546193823e-05,
+      "loss": 0.9745,
+      "step": 85025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2818149501908837e-05,
+      "loss": 1.2312,
+      "step": 85030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2817424441530702e-05,
+      "loss": 1.2038,
+      "step": 85035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2816699365063558e-05,
+      "loss": 1.0561,
+      "step": 85040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2815974272511547e-05,
+      "loss": 1.1139,
+      "step": 85045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.281524916387881e-05,
+      "loss": 1.1006,
+      "step": 85050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2814524039169483e-05,
+      "loss": 1.244,
+      "step": 85055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2813798898387708e-05,
+      "loss": 1.0564,
+      "step": 85060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2813073741537632e-05,
+      "loss": 1.1781,
+      "step": 85065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.281234856862339e-05,
+      "loss": 1.1462,
+      "step": 85070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2811623379649123e-05,
+      "loss": 1.3141,
+      "step": 85075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.281089817461898e-05,
+      "loss": 1.1816,
+      "step": 85080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.281017295353709e-05,
+      "loss": 1.4212,
+      "step": 85085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2809447716407605e-05,
+      "loss": 1.1033,
+      "step": 85090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2808722463234662e-05,
+      "loss": 1.1457,
+      "step": 85095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.28079971940224e-05,
+      "loss": 1.4578,
+      "step": 85100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2807271908774972e-05,
+      "loss": 1.152,
+      "step": 85105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2806546607496503e-05,
+      "loss": 1.1752,
+      "step": 85110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.280582129019115e-05,
+      "loss": 1.2899,
+      "step": 85115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.280509595686304e-05,
+      "loss": 1.3155,
+      "step": 85120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.280437060751633e-05,
+      "loss": 1.405,
+      "step": 85125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2803645242155151e-05,
+      "loss": 1.2962,
+      "step": 85130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2802919860783654e-05,
+      "loss": 1.2858,
+      "step": 85135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2802194463405975e-05,
+      "loss": 1.2815,
+      "step": 85140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2801469050026258e-05,
+      "loss": 1.1974,
+      "step": 85145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2800743620648647e-05,
+      "loss": 1.4548,
+      "step": 85150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2800018175277285e-05,
+      "loss": 1.1507,
+      "step": 85155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2799292713916311e-05,
+      "loss": 1.2097,
+      "step": 85160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.279856723656987e-05,
+      "loss": 1.3251,
+      "step": 85165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2797841743242106e-05,
+      "loss": 1.3848,
+      "step": 85170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.279711623393716e-05,
+      "loss": 1.3326,
+      "step": 85175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2796390708659174e-05,
+      "loss": 1.0005,
+      "step": 85180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2795665167412297e-05,
+      "loss": 1.2176,
+      "step": 85185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2794939610200663e-05,
+      "loss": 1.3948,
+      "step": 85190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2794214037028427e-05,
+      "loss": 1.1955,
+      "step": 85195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2793488447899722e-05,
+      "loss": 1.1093,
+      "step": 85200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2792762842818696e-05,
+      "loss": 1.3273,
+      "step": 85205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2792037221789492e-05,
+      "loss": 1.3105,
+      "step": 85210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2791311584816252e-05,
+      "loss": 1.2119,
+      "step": 85215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2790585931903126e-05,
+      "loss": 1.033,
+      "step": 85220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.278986026305425e-05,
+      "loss": 1.1953,
+      "step": 85225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2789134578273771e-05,
+      "loss": 1.0217,
+      "step": 85230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2788408877565837e-05,
+      "loss": 1.3379,
+      "step": 85235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2787683160934586e-05,
+      "loss": 1.1965,
+      "step": 85240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2786957428384166e-05,
+      "loss": 1.052,
+      "step": 85245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.278623167991872e-05,
+      "loss": 0.9967,
+      "step": 85250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2785505915542392e-05,
+      "loss": 0.9841,
+      "step": 85255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2784780135259326e-05,
+      "loss": 1.2426,
+      "step": 85260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2784054339073668e-05,
+      "loss": 1.2398,
+      "step": 85265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2783328526989563e-05,
+      "loss": 1.1286,
+      "step": 85270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2782602699011155e-05,
+      "loss": 1.1475,
+      "step": 85275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2781876855142588e-05,
+      "loss": 1.2602,
+      "step": 85280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2781150995388007e-05,
+      "loss": 1.1911,
+      "step": 85285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2780425119751562e-05,
+      "loss": 1.3118,
+      "step": 85290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.277969922823739e-05,
+      "loss": 1.0701,
+      "step": 85295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2778973320849643e-05,
+      "loss": 0.949,
+      "step": 85300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2778247397592465e-05,
+      "loss": 1.2293,
+      "step": 85305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2777521458469995e-05,
+      "loss": 1.198,
+      "step": 85310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2776795503486388e-05,
+      "loss": 1.1718,
+      "step": 85315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2776069532645783e-05,
+      "loss": 1.4782,
+      "step": 85320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2775343545952329e-05,
+      "loss": 1.1125,
+      "step": 85325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.277461754341017e-05,
+      "loss": 1.4609,
+      "step": 85330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2773891525023452e-05,
+      "loss": 1.3962,
+      "step": 85335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2773165490796324e-05,
+      "loss": 1.126,
+      "step": 85340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2772439440732925e-05,
+      "loss": 1.3611,
+      "step": 85345
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.277171337483741e-05,
+      "loss": 1.0524,
+      "step": 85350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.277098729311392e-05,
+      "loss": 1.14,
+      "step": 85355
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.27702611955666e-05,
+      "loss": 1.441,
+      "step": 85360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2769535082199599e-05,
+      "loss": 0.9965,
+      "step": 85365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2768808953017065e-05,
+      "loss": 1.1963,
+      "step": 85370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2768082808023141e-05,
+      "loss": 1.2301,
+      "step": 85375
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2767356647221975e-05,
+      "loss": 1.0846,
+      "step": 85380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2766630470617715e-05,
+      "loss": 1.2852,
+      "step": 85385
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2765904278214506e-05,
+      "loss": 1.4587,
+      "step": 85390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2765178070016498e-05,
+      "loss": 1.1195,
+      "step": 85395
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2764451846027836e-05,
+      "loss": 0.8596,
+      "step": 85400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2763725606252666e-05,
+      "loss": 1.3127,
+      "step": 85405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2762999350695135e-05,
+      "loss": 1.318,
+      "step": 85410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.276227307935939e-05,
+      "loss": 1.0551,
+      "step": 85415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2761546792249586e-05,
+      "loss": 1.0859,
+      "step": 85420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2760820489369861e-05,
+      "loss": 1.3931,
+      "step": 85425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2760094170724366e-05,
+      "loss": 1.3384,
+      "step": 85430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2759367836317249e-05,
+      "loss": 1.1384,
+      "step": 85435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2758641486152658e-05,
+      "loss": 1.019,
+      "step": 85440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2757915120234743e-05,
+      "loss": 1.2638,
+      "step": 85445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2757188738567643e-05,
+      "loss": 1.0903,
+      "step": 85450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2756462341155516e-05,
+      "loss": 1.3601,
+      "step": 85455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2755735928002508e-05,
+      "loss": 1.07,
+      "step": 85460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2755009499112763e-05,
+      "loss": 1.1285,
+      "step": 85465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2754283054490434e-05,
+      "loss": 1.4873,
+      "step": 85470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2753556594139664e-05,
+      "loss": 1.1304,
+      "step": 85475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2752830118064607e-05,
+      "loss": 1.2094,
+      "step": 85480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2752103626269411e-05,
+      "loss": 1.4533,
+      "step": 85485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2751377118758222e-05,
+      "loss": 1.0817,
+      "step": 85490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2750650595535187e-05,
+      "loss": 1.0799,
+      "step": 85495
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2749924056604463e-05,
+      "loss": 1.1171,
+      "step": 85500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.274919750197019e-05,
+      "loss": 1.5252,
+      "step": 85505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2748470931636519e-05,
+      "loss": 1.0568,
+      "step": 85510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2747744345607607e-05,
+      "loss": 1.0721,
+      "step": 85515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2747017743887589e-05,
+      "loss": 1.5705,
+      "step": 85520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.274629112648063e-05,
+      "loss": 1.5517,
+      "step": 85525
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2745564493390864e-05,
+      "loss": 1.4802,
+      "step": 85530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2744837844622452e-05,
+      "loss": 1.2538,
+      "step": 85535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2744111180179541e-05,
+      "loss": 0.9776,
+      "step": 85540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2743384500066275e-05,
+      "loss": 1.0998,
+      "step": 85545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2742657804286813e-05,
+      "loss": 1.2064,
+      "step": 85550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2741931092845296e-05,
+      "loss": 1.5371,
+      "step": 85555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2741204365745879e-05,
+      "loss": 1.2477,
+      "step": 85560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.274047762299271e-05,
+      "loss": 1.4125,
+      "step": 85565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.273975086458994e-05,
+      "loss": 0.9802,
+      "step": 85570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.273902409054172e-05,
+      "loss": 1.0027,
+      "step": 85575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2738297300852201e-05,
+      "loss": 1.4215,
+      "step": 85580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2737570495525529e-05,
+      "loss": 1.521,
+      "step": 85585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2736843674565856e-05,
+      "loss": 1.411,
+      "step": 85590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2736116837977335e-05,
+      "loss": 1.2084,
+      "step": 85595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2735389985764117e-05,
+      "loss": 1.2123,
+      "step": 85600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2734663117930348e-05,
+      "loss": 1.1503,
+      "step": 85605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2733936234480182e-05,
+      "loss": 1.1622,
+      "step": 85610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.273320933541777e-05,
+      "loss": 1.3773,
+      "step": 85615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2732482420747265e-05,
+      "loss": 1.1251,
+      "step": 85620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2731755490472813e-05,
+      "loss": 1.4719,
+      "step": 85625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2731028544598568e-05,
+      "loss": 1.181,
+      "step": 85630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.273030158312868e-05,
+      "loss": 1.1581,
+      "step": 85635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2729574606067305e-05,
+      "loss": 1.1897,
+      "step": 85640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2728847613418592e-05,
+      "loss": 1.475,
+      "step": 85645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2728120605186686e-05,
+      "loss": 1.1545,
+      "step": 85650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2727393581375746e-05,
+      "loss": 1.4031,
+      "step": 85655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2726666541989923e-05,
+      "loss": 1.0091,
+      "step": 85660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2725939487033368e-05,
+      "loss": 1.2556,
+      "step": 85665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2725212416510231e-05,
+      "loss": 1.0415,
+      "step": 85670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2724485330424665e-05,
+      "loss": 1.2396,
+      "step": 85675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2723758228780824e-05,
+      "loss": 1.2621,
+      "step": 85680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2723031111582857e-05,
+      "loss": 1.4578,
+      "step": 85685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2722303978834918e-05,
+      "loss": 1.4277,
+      "step": 85690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2721576830541158e-05,
+      "loss": 1.3356,
+      "step": 85695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2720849666705734e-05,
+      "loss": 1.0687,
+      "step": 85700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2720122487332794e-05,
+      "loss": 1.2636,
+      "step": 85705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2719395292426489e-05,
+      "loss": 1.3981,
+      "step": 85710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2718668081990976e-05,
+      "loss": 1.3106,
+      "step": 85715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2717940856030404e-05,
+      "loss": 1.4091,
+      "step": 85720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.271721361454893e-05,
+      "loss": 1.2284,
+      "step": 85725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2716486357550704e-05,
+      "loss": 0.8897,
+      "step": 85730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2715759085039883e-05,
+      "loss": 1.0533,
+      "step": 85735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2715031797020611e-05,
+      "loss": 1.1793,
+      "step": 85740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2714304493497052e-05,
+      "loss": 1.3573,
+      "step": 85745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2713577174473353e-05,
+      "loss": 1.4108,
+      "step": 85750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2712849839953666e-05,
+      "loss": 1.316,
+      "step": 85755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2712122489942151e-05,
+      "loss": 1.1973,
+      "step": 85760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2711395124442958e-05,
+      "loss": 1.2552,
+      "step": 85765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2710667743460237e-05,
+      "loss": 1.2242,
+      "step": 85770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.270994034699815e-05,
+      "loss": 1.1706,
+      "step": 85775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2709212935060841e-05,
+      "loss": 1.2666,
+      "step": 85780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2708485507652472e-05,
+      "loss": 1.4031,
+      "step": 85785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2707758064777194e-05,
+      "loss": 0.9423,
+      "step": 85790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2707030606439162e-05,
+      "loss": 1.2462,
+      "step": 85795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2706303132642526e-05,
+      "loss": 1.199,
+      "step": 85800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2705575643391445e-05,
+      "loss": 1.4698,
+      "step": 85805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2704848138690073e-05,
+      "loss": 1.1547,
+      "step": 85810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2704120618542563e-05,
+      "loss": 1.0701,
+      "step": 85815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.270339308295307e-05,
+      "loss": 1.3104,
+      "step": 85820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2702665531925746e-05,
+      "loss": 1.1876,
+      "step": 85825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2701937965464752e-05,
+      "loss": 1.0987,
+      "step": 85830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2701210383574236e-05,
+      "loss": 1.2635,
+      "step": 85835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2700482786258354e-05,
+      "loss": 1.6775,
+      "step": 85840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.269975517352127e-05,
+      "loss": 1.1828,
+      "step": 85845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2699027545367124e-05,
+      "loss": 1.1424,
+      "step": 85850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2698299901800086e-05,
+      "loss": 1.0341,
+      "step": 85855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2697572242824299e-05,
+      "loss": 1.4481,
+      "step": 85860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2696844568443925e-05,
+      "loss": 1.2375,
+      "step": 85865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2696116878663118e-05,
+      "loss": 1.4015,
+      "step": 85870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2695389173486036e-05,
+      "loss": 1.3741,
+      "step": 85875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.269466145291683e-05,
+      "loss": 1.1512,
+      "step": 85880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2693933716959657e-05,
+      "loss": 1.0304,
+      "step": 85885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2693205965618676e-05,
+      "loss": 1.361,
+      "step": 85890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2692478198898038e-05,
+      "loss": 1.2502,
+      "step": 85895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2691750416801902e-05,
+      "loss": 1.2356,
+      "step": 85900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2691022619334424e-05,
+      "loss": 1.0873,
+      "step": 85905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2690294806499759e-05,
+      "loss": 1.3747,
+      "step": 85910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2689566978302063e-05,
+      "loss": 1.3003,
+      "step": 85915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2688839134745494e-05,
+      "loss": 1.2106,
+      "step": 85920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2688111275834207e-05,
+      "loss": 1.1993,
+      "step": 85925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2687383401572357e-05,
+      "loss": 1.1616,
+      "step": 85930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2686655511964108e-05,
+      "loss": 0.936,
+      "step": 85935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2685927607013604e-05,
+      "loss": 1.0265,
+      "step": 85940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2685199686725012e-05,
+      "loss": 1.1585,
+      "step": 85945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2684471751102488e-05,
+      "loss": 1.0267,
+      "step": 85950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2683743800150181e-05,
+      "loss": 1.4048,
+      "step": 85955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2683015833872259e-05,
+      "loss": 1.2593,
+      "step": 85960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.268228785227287e-05,
+      "loss": 1.2789,
+      "step": 85965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2681559855356174e-05,
+      "loss": 1.3643,
+      "step": 85970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2680831843126332e-05,
+      "loss": 1.1988,
+      "step": 85975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2680103815587495e-05,
+      "loss": 1.5922,
+      "step": 85980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2679375772743825e-05,
+      "loss": 1.496,
+      "step": 85985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.267864771459948e-05,
+      "loss": 1.2068,
+      "step": 85990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2677919641158614e-05,
+      "loss": 1.0146,
+      "step": 85995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2677191552425388e-05,
+      "loss": 1.4813,
+      "step": 86000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2676463448403957e-05,
+      "loss": 1.1024,
+      "step": 86005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.267573532909848e-05,
+      "loss": 1.133,
+      "step": 86010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2675007194513116e-05,
+      "loss": 1.1847,
+      "step": 86015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2674279044652022e-05,
+      "loss": 1.0838,
+      "step": 86020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2673550879519356e-05,
+      "loss": 1.1485,
+      "step": 86025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2672822699119279e-05,
+      "loss": 1.2459,
+      "step": 86030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.267209450345594e-05,
+      "loss": 1.5236,
+      "step": 86035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2671366292533512e-05,
+      "loss": 1.5509,
+      "step": 86040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2670638066356142e-05,
+      "loss": 1.1179,
+      "step": 86045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2669909824927991e-05,
+      "loss": 1.2671,
+      "step": 86050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2669181568253222e-05,
+      "loss": 1.2439,
+      "step": 86055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2668453296335988e-05,
+      "loss": 1.4073,
+      "step": 86060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2667725009180452e-05,
+      "loss": 1.3217,
+      "step": 86065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2666996706790768e-05,
+      "loss": 0.8762,
+      "step": 86070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2666268389171102e-05,
+      "loss": 1.0909,
+      "step": 86075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.266554005632561e-05,
+      "loss": 1.4187,
+      "step": 86080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2664811708258448e-05,
+      "loss": 1.1496,
+      "step": 86085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2664083344973778e-05,
+      "loss": 1.1944,
+      "step": 86090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2663354966475762e-05,
+      "loss": 1.3562,
+      "step": 86095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2662626572768555e-05,
+      "loss": 1.5219,
+      "step": 86100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2661898163856319e-05,
+      "loss": 1.6577,
+      "step": 86105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2661169739743211e-05,
+      "loss": 0.9996,
+      "step": 86110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2660441300433393e-05,
+      "loss": 1.3301,
+      "step": 86115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2659712845931025e-05,
+      "loss": 1.4159,
+      "step": 86120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2658984376240266e-05,
+      "loss": 1.1612,
+      "step": 86125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2658255891365273e-05,
+      "loss": 1.2702,
+      "step": 86130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2657527391310213e-05,
+      "loss": 1.4472,
+      "step": 86135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2656798876079242e-05,
+      "loss": 0.9843,
+      "step": 86140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2656070345676521e-05,
+      "loss": 1.076,
+      "step": 86145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2655341800106208e-05,
+      "loss": 1.2833,
+      "step": 86150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2654613239372464e-05,
+      "loss": 1.2521,
+      "step": 86155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2653884663479455e-05,
+      "loss": 1.2928,
+      "step": 86160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2653156072431331e-05,
+      "loss": 1.4556,
+      "step": 86165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2652427466232264e-05,
+      "loss": 1.1172,
+      "step": 86170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2651698844886407e-05,
+      "loss": 1.1289,
+      "step": 86175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2650970208397924e-05,
+      "loss": 1.3665,
+      "step": 86180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2650241556770977e-05,
+      "loss": 0.9647,
+      "step": 86185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2649512890009721e-05,
+      "loss": 1.1608,
+      "step": 86190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2648784208118326e-05,
+      "loss": 1.7411,
+      "step": 86195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2648055511100944e-05,
+      "loss": 1.258,
+      "step": 86200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2647326798961745e-05,
+      "loss": 1.2594,
+      "step": 86205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2646598071704883e-05,
+      "loss": 1.3517,
+      "step": 86210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2645869329334523e-05,
+      "loss": 1.1881,
+      "step": 86215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2645140571854828e-05,
+      "loss": 1.4,
+      "step": 86220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2644411799269958e-05,
+      "loss": 1.0117,
+      "step": 86225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.264368301158407e-05,
+      "loss": 1.2282,
+      "step": 86230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2642954208801333e-05,
+      "loss": 1.1467,
+      "step": 86235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.264222539092591e-05,
+      "loss": 1.3341,
+      "step": 86240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2641496557961954e-05,
+      "loss": 1.2965,
+      "step": 86245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.264076770991363e-05,
+      "loss": 1.2009,
+      "step": 86250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2640038846785107e-05,
+      "loss": 1.1527,
+      "step": 86255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2639309968580538e-05,
+      "loss": 1.369,
+      "step": 86260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2638581075304093e-05,
+      "loss": 1.0963,
+      "step": 86265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2637852166959928e-05,
+      "loss": 1.2835,
+      "step": 86270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2637123243552209e-05,
+      "loss": 1.0153,
+      "step": 86275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2636394305085098e-05,
+      "loss": 1.5046,
+      "step": 86280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2635665351562759e-05,
+      "loss": 1.2615,
+      "step": 86285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2634936382989356e-05,
+      "loss": 1.3052,
+      "step": 86290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2634207399369042e-05,
+      "loss": 1.1985,
+      "step": 86295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.263347840070599e-05,
+      "loss": 1.1948,
+      "step": 86300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2632749387004362e-05,
+      "loss": 1.3956,
+      "step": 86305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2632020358268316e-05,
+      "loss": 1.0281,
+      "step": 86310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2631291314502021e-05,
+      "loss": 1.2377,
+      "step": 86315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2630562255709635e-05,
+      "loss": 1.045,
+      "step": 86320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2629833181895324e-05,
+      "loss": 1.3972,
+      "step": 86325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.262910409306325e-05,
+      "loss": 1.2629,
+      "step": 86330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.262837498921758e-05,
+      "loss": 1.3858,
+      "step": 86335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2627645870362473e-05,
+      "loss": 1.1699,
+      "step": 86340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2626916736502096e-05,
+      "loss": 1.1114,
+      "step": 86345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2626187587640612e-05,
+      "loss": 1.0937,
+      "step": 86350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2625458423782186e-05,
+      "loss": 0.995,
+      "step": 86355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2624729244930978e-05,
+      "loss": 1.2065,
+      "step": 86360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2624000051091152e-05,
+      "loss": 1.3097,
+      "step": 86365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2623270842266879e-05,
+      "loss": 1.4188,
+      "step": 86370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2622541618462316e-05,
+      "loss": 1.3703,
+      "step": 86375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.262181237968163e-05,
+      "loss": 0.9486,
+      "step": 86380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2621083125928988e-05,
+      "loss": 1.1622,
+      "step": 86385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2620353857208546e-05,
+      "loss": 1.2597,
+      "step": 86390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.261962457352448e-05,
+      "loss": 1.0707,
+      "step": 86395
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2618895274880943e-05,
+      "loss": 1.4371,
+      "step": 86400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.261816596128211e-05,
+      "loss": 1.1778,
+      "step": 86405
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2617436632732138e-05,
+      "loss": 1.273,
+      "step": 86410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2616707289235195e-05,
+      "loss": 0.977,
+      "step": 86415
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2615977930795449e-05,
+      "loss": 1.0989,
+      "step": 86420
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2615248557417062e-05,
+      "loss": 1.0059,
+      "step": 86425
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2614519169104197e-05,
+      "loss": 1.0118,
+      "step": 86430
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.261378976586102e-05,
+      "loss": 1.2991,
+      "step": 86435
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.26130603476917e-05,
+      "loss": 1.2757,
+      "step": 86440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.26123309146004e-05,
+      "loss": 1.0491,
+      "step": 86445
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2611601466591283e-05,
+      "loss": 1.0496,
+      "step": 86450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.261087200366852e-05,
+      "loss": 1.3904,
+      "step": 86455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2610142525836273e-05,
+      "loss": 1.089,
+      "step": 86460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2609413033098707e-05,
+      "loss": 0.9845,
+      "step": 86465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2608683525459988e-05,
+      "loss": 1.2063,
+      "step": 86470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2607954002924287e-05,
+      "loss": 1.3934,
+      "step": 86475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2607224465495763e-05,
+      "loss": 1.533,
+      "step": 86480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2606494913178584e-05,
+      "loss": 1.2986,
+      "step": 86485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2605765345976924e-05,
+      "loss": 1.4105,
+      "step": 86490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2605035763894933e-05,
+      "loss": 1.0887,
+      "step": 86495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2604306166936792e-05,
+      "loss": 1.339,
+      "step": 86500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.260357655510666e-05,
+      "loss": 1.5556,
+      "step": 86505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2602846928408707e-05,
+      "loss": 1.2135,
+      "step": 86510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2602117286847097e-05,
+      "loss": 1.2055,
+      "step": 86515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2601387630425998e-05,
+      "loss": 1.155,
+      "step": 86520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2600657959149578e-05,
+      "loss": 1.2435,
+      "step": 86525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2599928273022001e-05,
+      "loss": 1.2199,
+      "step": 86530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2599198572047435e-05,
+      "loss": 1.1389,
+      "step": 86535
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2598468856230048e-05,
+      "loss": 1.2501,
+      "step": 86540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2597739125574005e-05,
+      "loss": 1.1926,
+      "step": 86545
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2597009380083476e-05,
+      "loss": 1.1433,
+      "step": 86550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2596279619762626e-05,
+      "loss": 1.3207,
+      "step": 86555
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2595549844615623e-05,
+      "loss": 0.9717,
+      "step": 86560
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2594820054646632e-05,
+      "loss": 1.2854,
+      "step": 86565
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2594090249859826e-05,
+      "loss": 1.1361,
+      "step": 86570
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2593360430259367e-05,
+      "loss": 1.2676,
+      "step": 86575
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2592630595849429e-05,
+      "loss": 1.3533,
+      "step": 86580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2591900746634171e-05,
+      "loss": 1.2327,
+      "step": 86585
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2591170882617767e-05,
+      "loss": 1.0099,
+      "step": 86590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2590441003804389e-05,
+      "loss": 1.2369,
+      "step": 86595
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2589711110198191e-05,
+      "loss": 1.185,
+      "step": 86600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2588981201803356e-05,
+      "loss": 1.5502,
+      "step": 86605
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2588251278624041e-05,
+      "loss": 1.4606,
+      "step": 86610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2587521340664424e-05,
+      "loss": 1.3438,
+      "step": 86615
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2586791387928665e-05,
+      "loss": 1.458,
+      "step": 86620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2586061420420937e-05,
+      "loss": 1.218,
+      "step": 86625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2585331438145408e-05,
+      "loss": 1.0056,
+      "step": 86630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2584601441106245e-05,
+      "loss": 1.5007,
+      "step": 86635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.258387142930762e-05,
+      "loss": 1.4922,
+      "step": 86640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2583141402753696e-05,
+      "loss": 1.5184,
+      "step": 86645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2582411361448647e-05,
+      "loss": 1.1362,
+      "step": 86650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.258168130539664e-05,
+      "loss": 1.1241,
+      "step": 86655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2580951234601844e-05,
+      "loss": 1.3697,
+      "step": 86660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2580221149068428e-05,
+      "loss": 1.2222,
+      "step": 86665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2579491048800559e-05,
+      "loss": 1.54,
+      "step": 86670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2578760933802414e-05,
+      "loss": 1.5396,
+      "step": 86675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2578030804078152e-05,
+      "loss": 1.2731,
+      "step": 86680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2577300659631953e-05,
+      "loss": 1.3465,
+      "step": 86685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2576570500467979e-05,
+      "loss": 1.1558,
+      "step": 86690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2575840326590399e-05,
+      "loss": 1.1932,
+      "step": 86695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2575110138003388e-05,
+      "loss": 1.2938,
+      "step": 86700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2574379934711112e-05,
+      "loss": 1.2082,
+      "step": 86705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2573649716717742e-05,
+      "loss": 1.2587,
+      "step": 86710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2572919484027447e-05,
+      "loss": 1.1266,
+      "step": 86715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2572189236644398e-05,
+      "loss": 1.2186,
+      "step": 86720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2571458974572765e-05,
+      "loss": 1.247,
+      "step": 86725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2570728697816721e-05,
+      "loss": 1.2173,
+      "step": 86730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.256999840638043e-05,
+      "loss": 1.0985,
+      "step": 86735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2569268100268068e-05,
+      "loss": 1.0755,
+      "step": 86740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2568537779483802e-05,
+      "loss": 1.1909,
+      "step": 86745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2567807444031806e-05,
+      "loss": 1.3508,
+      "step": 86750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2567077093916248e-05,
+      "loss": 1.3821,
+      "step": 86755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2566346729141297e-05,
+      "loss": 1.3387,
+      "step": 86760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2565616349711128e-05,
+      "loss": 1.2684,
+      "step": 86765
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2564885955629906e-05,
+      "loss": 1.3174,
+      "step": 86770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2564155546901806e-05,
+      "loss": 1.0072,
+      "step": 86775
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2563425123531003e-05,
+      "loss": 1.0712,
+      "step": 86780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.256269468552166e-05,
+      "loss": 0.9312,
+      "step": 86785
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2561964232877957e-05,
+      "loss": 1.2127,
+      "step": 86790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2561233765604054e-05,
+      "loss": 1.2422,
+      "step": 86795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.256050328370413e-05,
+      "loss": 1.2979,
+      "step": 86800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2559772787182361e-05,
+      "loss": 1.0499,
+      "step": 86805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2559042276042906e-05,
+      "loss": 1.0743,
+      "step": 86810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2558311750289945e-05,
+      "loss": 1.2742,
+      "step": 86815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2557581209927647e-05,
+      "loss": 1.0892,
+      "step": 86820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2556850654960186e-05,
+      "loss": 1.2913,
+      "step": 86825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2556120085391736e-05,
+      "loss": 1.077,
+      "step": 86830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2555389501226461e-05,
+      "loss": 1.3012,
+      "step": 86835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2554658902468537e-05,
+      "loss": 1.1044,
+      "step": 86840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2553928289122137e-05,
+      "loss": 1.1089,
+      "step": 86845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2553197661191434e-05,
+      "loss": 1.2477,
+      "step": 86850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.25524670186806e-05,
+      "loss": 1.1259,
+      "step": 86855
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2551736361593807e-05,
+      "loss": 1.4594,
+      "step": 86860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2551005689935225e-05,
+      "loss": 1.1957,
+      "step": 86865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.255027500370903e-05,
+      "loss": 1.3193,
+      "step": 86870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2549544302919391e-05,
+      "loss": 1.0111,
+      "step": 86875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.254881358757048e-05,
+      "loss": 1.1684,
+      "step": 86880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2548082857666479e-05,
+      "loss": 1.1449,
+      "step": 86885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2547352113211552e-05,
+      "loss": 1.0712,
+      "step": 86890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2546621354209873e-05,
+      "loss": 1.368,
+      "step": 86895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2545890580665621e-05,
+      "loss": 1.038,
+      "step": 86900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2545159792582956e-05,
+      "loss": 1.2473,
+      "step": 86905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2544428989966069e-05,
+      "loss": 1.128,
+      "step": 86910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2543698172819117e-05,
+      "loss": 1.1916,
+      "step": 86915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2542967341146282e-05,
+      "loss": 1.5909,
+      "step": 86920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2542236494951736e-05,
+      "loss": 1.3722,
+      "step": 86925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2541505634239655e-05,
+      "loss": 1.2243,
+      "step": 86930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2540774759014209e-05,
+      "loss": 1.4953,
+      "step": 86935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2540043869279571e-05,
+      "loss": 1.0859,
+      "step": 86940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2539312965039914e-05,
+      "loss": 1.129,
+      "step": 86945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2538582046299417e-05,
+      "loss": 1.0717,
+      "step": 86950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2537851113062251e-05,
+      "loss": 1.2656,
+      "step": 86955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.253712016533259e-05,
+      "loss": 0.9616,
+      "step": 86960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.253638920311461e-05,
+      "loss": 0.8135,
+      "step": 86965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2535658226412482e-05,
+      "loss": 2.0265,
+      "step": 86970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.253492723523038e-05,
+      "loss": 1.1238,
+      "step": 86975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2534196229572482e-05,
+      "loss": 1.2177,
+      "step": 86980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2533465209442959e-05,
+      "loss": 1.2769,
+      "step": 86985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2532734174845989e-05,
+      "loss": 1.1626,
+      "step": 86990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2532003125785744e-05,
+      "loss": 1.5279,
+      "step": 86995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2531272062266395e-05,
+      "loss": 1.1953,
+      "step": 87000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.253054098429213e-05,
+      "loss": 1.0916,
+      "step": 87005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2529809891867107e-05,
+      "loss": 1.3173,
+      "step": 87010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2529078784995514e-05,
+      "loss": 1.1734,
+      "step": 87015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2528347663681517e-05,
+      "loss": 1.2355,
+      "step": 87020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2527616527929296e-05,
+      "loss": 1.1526,
+      "step": 87025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2526885377743028e-05,
+      "loss": 1.1225,
+      "step": 87030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2526154213126882e-05,
+      "loss": 1.3117,
+      "step": 87035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2525423034085038e-05,
+      "loss": 1.1691,
+      "step": 87040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.252469184062167e-05,
+      "loss": 1.2319,
+      "step": 87045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2523960632740955e-05,
+      "loss": 1.1,
+      "step": 87050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2523229410447064e-05,
+      "loss": 0.9154,
+      "step": 87055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.252249817374418e-05,
+      "loss": 1.0368,
+      "step": 87060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2521766922636472e-05,
+      "loss": 1.2882,
+      "step": 87065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2521035657128118e-05,
+      "loss": 1.1786,
+      "step": 87070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2520304377223296e-05,
+      "loss": 1.2081,
+      "step": 87075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2519573082926181e-05,
+      "loss": 1.3793,
+      "step": 87080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2518841774240947e-05,
+      "loss": 1.1574,
+      "step": 87085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2518110451171773e-05,
+      "loss": 1.1689,
+      "step": 87090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2517379113722834e-05,
+      "loss": 1.0912,
+      "step": 87095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2516647761898306e-05,
+      "loss": 1.2512,
+      "step": 87100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2515916395702363e-05,
+      "loss": 1.3373,
+      "step": 87105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2515185015139188e-05,
+      "loss": 1.2558,
+      "step": 87110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.251445362021295e-05,
+      "loss": 1.2327,
+      "step": 87115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2513722210927835e-05,
+      "loss": 1.191,
+      "step": 87120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2512990787288008e-05,
+      "loss": 1.1278,
+      "step": 87125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2512259349297654e-05,
+      "loss": 1.2325,
+      "step": 87130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2511527896960951e-05,
+      "loss": 1.2479,
+      "step": 87135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2510796430282071e-05,
+      "loss": 1.0391,
+      "step": 87140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.251006494926519e-05,
+      "loss": 1.1978,
+      "step": 87145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2509333453914491e-05,
+      "loss": 1.0377,
+      "step": 87150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2508601944234149e-05,
+      "loss": 1.312,
+      "step": 87155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2507870420228341e-05,
+      "loss": 1.1174,
+      "step": 87160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2507138881901241e-05,
+      "loss": 1.1113,
+      "step": 87165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2506407329257031e-05,
+      "loss": 1.3275,
+      "step": 87170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2505675762299889e-05,
+      "loss": 1.3413,
+      "step": 87175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2504944181033988e-05,
+      "loss": 1.1925,
+      "step": 87180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2504212585463509e-05,
+      "loss": 1.4814,
+      "step": 87185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.250348097559263e-05,
+      "loss": 1.2135,
+      "step": 87190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2502749351425529e-05,
+      "loss": 0.9941,
+      "step": 87195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2502017712966381e-05,
+      "loss": 1.4306,
+      "step": 87200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2501286060219367e-05,
+      "loss": 1.3668,
+      "step": 87205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2500554393188662e-05,
+      "loss": 1.2614,
+      "step": 87210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2499822711878451e-05,
+      "loss": 1.1514,
+      "step": 87215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2499091016292903e-05,
+      "loss": 1.1953,
+      "step": 87220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2498359306436207e-05,
+      "loss": 1.0982,
+      "step": 87225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2497627582312533e-05,
+      "loss": 1.1782,
+      "step": 87230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2496895843926058e-05,
+      "loss": 1.14,
+      "step": 87235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2496164091280972e-05,
+      "loss": 1.6711,
+      "step": 87240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.249543232438144e-05,
+      "loss": 1.2031,
+      "step": 87245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.249470054323165e-05,
+      "loss": 1.1501,
+      "step": 87250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2493968747835779e-05,
+      "loss": 1.0448,
+      "step": 87255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2493236938198003e-05,
+      "loss": 1.0824,
+      "step": 87260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2492505114322504e-05,
+      "loss": 1.0812,
+      "step": 87265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2491773276213462e-05,
+      "loss": 1.0789,
+      "step": 87270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2491041423875051e-05,
+      "loss": 1.0815,
+      "step": 87275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2490309557311458e-05,
+      "loss": 1.5756,
+      "step": 87280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2489577676526853e-05,
+      "loss": 1.2034,
+      "step": 87285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2488845781525424e-05,
+      "loss": 1.3374,
+      "step": 87290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2488113872311347e-05,
+      "loss": 1.5857,
+      "step": 87295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.24873819488888e-05,
+      "loss": 1.0723,
+      "step": 87300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2486650011261966e-05,
+      "loss": 0.9822,
+      "step": 87305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.248591805943502e-05,
+      "loss": 1.4782,
+      "step": 87310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2485186093412146e-05,
+      "loss": 1.1341,
+      "step": 87315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2484454113197523e-05,
+      "loss": 1.1363,
+      "step": 87320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.248372211879533e-05,
+      "loss": 1.2226,
+      "step": 87325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2482990110209749e-05,
+      "loss": 1.2773,
+      "step": 87330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.248225808744496e-05,
+      "loss": 1.1533,
+      "step": 87335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.248152605050514e-05,
+      "loss": 1.0453,
+      "step": 87340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2480793999394474e-05,
+      "loss": 0.98,
+      "step": 87345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2480061934117136e-05,
+      "loss": 1.0782,
+      "step": 87350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2479329854677313e-05,
+      "loss": 1.4983,
+      "step": 87355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2478597761079182e-05,
+      "loss": 1.0608,
+      "step": 87360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2477865653326928e-05,
+      "loss": 1.1593,
+      "step": 87365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2477133531424725e-05,
+      "loss": 1.0382,
+      "step": 87370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2476401395376757e-05,
+      "loss": 1.29,
+      "step": 87375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2475669245187207e-05,
+      "loss": 1.0507,
+      "step": 87380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2474937080860255e-05,
+      "loss": 1.2911,
+      "step": 87385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.247420490240008e-05,
+      "loss": 1.208,
+      "step": 87390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2473472709810864e-05,
+      "loss": 1.1157,
+      "step": 87395
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2472740503096787e-05,
+      "loss": 1.3873,
+      "step": 87400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2472008282262034e-05,
+      "loss": 1.1688,
+      "step": 87405
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2471276047310783e-05,
+      "loss": 1.1104,
+      "step": 87410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2470543798247217e-05,
+      "loss": 1.3618,
+      "step": 87415
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2469811535075516e-05,
+      "loss": 1.7843,
+      "step": 87420
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2469079257799866e-05,
+      "loss": 1.0207,
+      "step": 87425
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.246834696642444e-05,
+      "loss": 1.2256,
+      "step": 87430
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2467614660953428e-05,
+      "loss": 1.1128,
+      "step": 87435
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2466882341391012e-05,
+      "loss": 1.1527,
+      "step": 87440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2466150007741367e-05,
+      "loss": 1.217,
+      "step": 87445
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2465417660008681e-05,
+      "loss": 1.4503,
+      "step": 87450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2464685298197133e-05,
+      "loss": 1.2187,
+      "step": 87455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2463952922310905e-05,
+      "loss": 1.1636,
+      "step": 87460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2463220532354182e-05,
+      "loss": 1.074,
+      "step": 87465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2462488128331147e-05,
+      "loss": 1.025,
+      "step": 87470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2461755710245978e-05,
+      "loss": 1.0867,
+      "step": 87475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2461023278102861e-05,
+      "loss": 1.1978,
+      "step": 87480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2460290831905977e-05,
+      "loss": 1.6246,
+      "step": 87485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2459558371659507e-05,
+      "loss": 1.1395,
+      "step": 87490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.245882589736764e-05,
+      "loss": 1.5397,
+      "step": 87495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2458093409034552e-05,
+      "loss": 1.0898,
+      "step": 87500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2457360906664429e-05,
+      "loss": 1.1885,
+      "step": 87505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2456628390261454e-05,
+      "loss": 1.1705,
+      "step": 87510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2455895859829808e-05,
+      "loss": 1.3723,
+      "step": 87515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2455163315373678e-05,
+      "loss": 1.3178,
+      "step": 87520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2454430756897242e-05,
+      "loss": 1.1732,
+      "step": 87525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.245369818440469e-05,
+      "loss": 1.0541,
+      "step": 87530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.24529655979002e-05,
+      "loss": 1.2403,
+      "step": 87535
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2452232997387955e-05,
+      "loss": 1.2098,
+      "step": 87540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2451500382872144e-05,
+      "loss": 0.9746,
+      "step": 87545
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2450767754356942e-05,
+      "loss": 1.2569,
+      "step": 87550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2450035111846545e-05,
+      "loss": 1.0666,
+      "step": 87555
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2449302455345124e-05,
+      "loss": 1.2503,
+      "step": 87560
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2448569784856871e-05,
+      "loss": 0.9554,
+      "step": 87565
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2447837100385968e-05,
+      "loss": 1.2634,
+      "step": 87570
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2447104401936599e-05,
+      "loss": 1.1428,
+      "step": 87575
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2446371689512946e-05,
+      "loss": 1.2979,
+      "step": 87580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2445638963119196e-05,
+      "loss": 1.2146,
+      "step": 87585
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2444906222759531e-05,
+      "loss": 0.9409,
+      "step": 87590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2444173468438139e-05,
+      "loss": 1.0844,
+      "step": 87595
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2443440700159199e-05,
+      "loss": 1.3649,
+      "step": 87600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2442707917926899e-05,
+      "loss": 1.4102,
+      "step": 87605
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2441975121745425e-05,
+      "loss": 1.3648,
+      "step": 87610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2441242311618958e-05,
+      "loss": 1.2343,
+      "step": 87615
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2440509487551682e-05,
+      "loss": 1.2982,
+      "step": 87620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.243977664954779e-05,
+      "loss": 1.4946,
+      "step": 87625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2439043797611455e-05,
+      "loss": 1.1955,
+      "step": 87630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2438310931746874e-05,
+      "loss": 1.1992,
+      "step": 87635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2437578051958222e-05,
+      "loss": 1.3737,
+      "step": 87640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2436845158249689e-05,
+      "loss": 1.4585,
+      "step": 87645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.243611225062546e-05,
+      "loss": 1.234,
+      "step": 87650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2435379329089719e-05,
+      "loss": 1.456,
+      "step": 87655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2434646393646655e-05,
+      "loss": 1.2975,
+      "step": 87660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2433913444300445e-05,
+      "loss": 1.0313,
+      "step": 87665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2433180481055285e-05,
+      "loss": 1.3403,
+      "step": 87670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2432447503915355e-05,
+      "loss": 1.1775,
+      "step": 87675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2431714512884838e-05,
+      "loss": 1.0782,
+      "step": 87680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2430981507967928e-05,
+      "loss": 1.7248,
+      "step": 87685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2430248489168804e-05,
+      "loss": 1.149,
+      "step": 87690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2429515456491652e-05,
+      "loss": 1.4372,
+      "step": 87695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2428782409940663e-05,
+      "loss": 1.1802,
+      "step": 87700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.242804934952002e-05,
+      "loss": 1.2394,
+      "step": 87705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2427316275233908e-05,
+      "loss": 1.4944,
+      "step": 87710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2426583187086516e-05,
+      "loss": 1.3546,
+      "step": 87715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2425850085082028e-05,
+      "loss": 1.2041,
+      "step": 87720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2425116969224629e-05,
+      "loss": 1.2679,
+      "step": 87725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2424383839518512e-05,
+      "loss": 1.1486,
+      "step": 87730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2423650695967858e-05,
+      "loss": 1.2598,
+      "step": 87735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2422917538576852e-05,
+      "loss": 1.3595,
+      "step": 87740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2422184367349686e-05,
+      "loss": 1.1145,
+      "step": 87745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2421451182290542e-05,
+      "loss": 0.9424,
+      "step": 87750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2420717983403615e-05,
+      "loss": 1.2052,
+      "step": 87755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.241998477069308e-05,
+      "loss": 1.1318,
+      "step": 87760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2419251544163133e-05,
+      "loss": 1.1864,
+      "step": 87765
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.241851830381796e-05,
+      "loss": 1.1824,
+      "step": 87770
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2417785049661746e-05,
+      "loss": 1.1421,
+      "step": 87775
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.241705178169868e-05,
+      "loss": 1.1422,
+      "step": 87780
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2416318499932944e-05,
+      "loss": 1.3459,
+      "step": 87785
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2415585204368734e-05,
+      "loss": 1.046,
+      "step": 87790
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2414851895010231e-05,
+      "loss": 1.3122,
+      "step": 87795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2414118571861626e-05,
+      "loss": 1.114,
+      "step": 87800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2413385234927105e-05,
+      "loss": 1.2308,
+      "step": 87805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2412651884210858e-05,
+      "loss": 1.0823,
+      "step": 87810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2411918519717068e-05,
+      "loss": 1.3471,
+      "step": 87815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2411185141449928e-05,
+      "loss": 1.071,
+      "step": 87820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2410451749413623e-05,
+      "loss": 1.2774,
+      "step": 87825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.240971834361234e-05,
+      "loss": 1.0201,
+      "step": 87830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2408984924050276e-05,
+      "loss": 1.1894,
+      "step": 87835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2408251490731606e-05,
+      "loss": 1.3564,
+      "step": 87840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2407518043660527e-05,
+      "loss": 1.2249,
+      "step": 87845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2406784582841227e-05,
+      "loss": 1.2086,
+      "step": 87850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2406051108277888e-05,
+      "loss": 1.3411,
+      "step": 87855
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2405317619974706e-05,
+      "loss": 1.2918,
+      "step": 87860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2404584117935864e-05,
+      "loss": 1.2923,
+      "step": 87865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2403850602165556e-05,
+      "loss": 1.2728,
+      "step": 87870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.240311707266797e-05,
+      "loss": 1.1682,
+      "step": 87875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2402383529447288e-05,
+      "loss": 1.2997,
+      "step": 87880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.240164997250771e-05,
+      "loss": 1.1465,
+      "step": 87885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2400916401853413e-05,
+      "loss": 1.5634,
+      "step": 87890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2400182817488595e-05,
+      "loss": 1.3221,
+      "step": 87895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2399449219417442e-05,
+      "loss": 1.1501,
+      "step": 87900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2398715607644143e-05,
+      "loss": 1.164,
+      "step": 87905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2397981982172889e-05,
+      "loss": 1.1367,
+      "step": 87910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2397248343007866e-05,
+      "loss": 1.3683,
+      "step": 87915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2396514690153268e-05,
+      "loss": 1.2229,
+      "step": 87920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.239578102361328e-05,
+      "loss": 1.4119,
+      "step": 87925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2395047343392094e-05,
+      "loss": 1.1313,
+      "step": 87930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2394313649493901e-05,
+      "loss": 1.5717,
+      "step": 87935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2393579941922886e-05,
+      "loss": 0.9794,
+      "step": 87940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2392846220683245e-05,
+      "loss": 1.2692,
+      "step": 87945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2392112485779166e-05,
+      "loss": 1.2998,
+      "step": 87950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2391378737214837e-05,
+      "loss": 1.0336,
+      "step": 87955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2390644974994445e-05,
+      "loss": 0.9775,
+      "step": 87960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.238991119912219e-05,
+      "loss": 1.032,
+      "step": 87965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2389177409602254e-05,
+      "loss": 1.0079,
+      "step": 87970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2388443606438832e-05,
+      "loss": 1.4711,
+      "step": 87975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2387709789636114e-05,
+      "loss": 1.4282,
+      "step": 87980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2386975959198282e-05,
+      "loss": 1.0465,
+      "step": 87985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2386242115129541e-05,
+      "loss": 1.3417,
+      "step": 87990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2385508257434067e-05,
+      "loss": 1.3676,
+      "step": 87995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.238477438611606e-05,
+      "loss": 1.0826,
+      "step": 88000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2384040501179713e-05,
+      "loss": 1.0965,
+      "step": 88005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2383306602629209e-05,
+      "loss": 1.3795,
+      "step": 88010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2382572690468742e-05,
+      "loss": 0.9787,
+      "step": 88015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2381838764702505e-05,
+      "loss": 1.263,
+      "step": 88020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2381104825334687e-05,
+      "loss": 1.2559,
+      "step": 88025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2380370872369479e-05,
+      "loss": 1.3699,
+      "step": 88030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2379636905811075e-05,
+      "loss": 0.9786,
+      "step": 88035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2378902925663661e-05,
+      "loss": 1.0838,
+      "step": 88040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2378168931931436e-05,
+      "loss": 1.8236,
+      "step": 88045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2377434924618583e-05,
+      "loss": 1.2299,
+      "step": 88050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2376700903729299e-05,
+      "loss": 1.3333,
+      "step": 88055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2375966869267777e-05,
+      "loss": 1.2259,
+      "step": 88060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2375232821238203e-05,
+      "loss": 1.0583,
+      "step": 88065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2374498759644778e-05,
+      "loss": 1.042,
+      "step": 88070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2373764684491683e-05,
+      "loss": 1.2496,
+      "step": 88075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2373030595783113e-05,
+      "loss": 1.3597,
+      "step": 88080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2372296493523266e-05,
+      "loss": 1.4702,
+      "step": 88085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2371562377716328e-05,
+      "loss": 1.2333,
+      "step": 88090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2370828248366493e-05,
+      "loss": 1.1502,
+      "step": 88095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2370094105477954e-05,
+      "loss": 1.2866,
+      "step": 88100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2369359949054905e-05,
+      "loss": 1.1155,
+      "step": 88105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2368625779101536e-05,
+      "loss": 1.3731,
+      "step": 88110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2367891595622036e-05,
+      "loss": 1.6996,
+      "step": 88115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2367157398620606e-05,
+      "loss": 1.1407,
+      "step": 88120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.236642318810143e-05,
+      "loss": 1.1872,
+      "step": 88125
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2365688964068709e-05,
+      "loss": 0.9952,
+      "step": 88130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.236495472652663e-05,
+      "loss": 1.1864,
+      "step": 88135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2364220475479388e-05,
+      "loss": 1.16,
+      "step": 88140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2363486210931173e-05,
+      "loss": 1.0266,
+      "step": 88145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2362751932886181e-05,
+      "loss": 1.0884,
+      "step": 88150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2362017641348607e-05,
+      "loss": 1.1273,
+      "step": 88155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2361283336322638e-05,
+      "loss": 1.1679,
+      "step": 88160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2360549017812478e-05,
+      "loss": 1.2565,
+      "step": 88165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2359814685822306e-05,
+      "loss": 1.0299,
+      "step": 88170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.235908034035633e-05,
+      "loss": 1.3692,
+      "step": 88175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.235834598141873e-05,
+      "loss": 1.008,
+      "step": 88180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2357611609013706e-05,
+      "loss": 0.9805,
+      "step": 88185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2356877223145459e-05,
+      "loss": 1.3997,
+      "step": 88190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2356142823818166e-05,
+      "loss": 1.192,
+      "step": 88195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2355408411036037e-05,
+      "loss": 1.4312,
+      "step": 88200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2354673984803256e-05,
+      "loss": 1.1508,
+      "step": 88205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.235393954512402e-05,
+      "loss": 1.3649,
+      "step": 88210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2353205092002524e-05,
+      "loss": 0.9743,
+      "step": 88215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2352470625442962e-05,
+      "loss": 1.3527,
+      "step": 88220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2351736145449527e-05,
+      "loss": 1.2398,
+      "step": 88225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2351001652026414e-05,
+      "loss": 1.3421,
+      "step": 88230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2350267145177817e-05,
+      "loss": 1.339,
+      "step": 88235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.234953262490793e-05,
+      "loss": 1.2068,
+      "step": 88240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2348798091220948e-05,
+      "loss": 1.125,
+      "step": 88245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2348063544121065e-05,
+      "loss": 1.1332,
+      "step": 88250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.234732898361248e-05,
+      "loss": 1.1277,
+      "step": 88255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2346594409699379e-05,
+      "loss": 1.3012,
+      "step": 88260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2345859822385962e-05,
+      "loss": 1.4097,
+      "step": 88265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2345125221676427e-05,
+      "loss": 1.3946,
+      "step": 88270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2344390607574964e-05,
+      "loss": 1.2901,
+      "step": 88275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2343655980085769e-05,
+      "loss": 1.0627,
+      "step": 88280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.234292133921304e-05,
+      "loss": 1.2408,
+      "step": 88285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2342186684960965e-05,
+      "loss": 1.2051,
+      "step": 88290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2341452017333752e-05,
+      "loss": 1.1591,
+      "step": 88295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2340717336335582e-05,
+      "loss": 1.1654,
+      "step": 88300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.233998264197066e-05,
+      "loss": 1.3089,
+      "step": 88305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2339247934243179e-05,
+      "loss": 1.1431,
+      "step": 88310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2338513213157333e-05,
+      "loss": 1.3141,
+      "step": 88315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.233777847871732e-05,
+      "loss": 1.0491,
+      "step": 88320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2337043730927332e-05,
+      "loss": 1.1695,
+      "step": 88325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2336308969791568e-05,
+      "loss": 1.3443,
+      "step": 88330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2335574195314225e-05,
+      "loss": 1.0781,
+      "step": 88335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2334839407499497e-05,
+      "loss": 1.3924,
+      "step": 88340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2334104606351578e-05,
+      "loss": 1.2174,
+      "step": 88345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2333369791874668e-05,
+      "loss": 1.178,
+      "step": 88350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2332634964072962e-05,
+      "loss": 1.2705,
+      "step": 88355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2331900122950655e-05,
+      "loss": 1.5198,
+      "step": 88360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2331165268511944e-05,
+      "loss": 1.0224,
+      "step": 88365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2330430400761025e-05,
+      "loss": 1.2068,
+      "step": 88370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2329695519702096e-05,
+      "loss": 1.369,
+      "step": 88375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2328960625339352e-05,
+      "loss": 1.1018,
+      "step": 88380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2328225717676988e-05,
+      "loss": 1.264,
+      "step": 88385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2327490796719208e-05,
+      "loss": 0.9563,
+      "step": 88390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.23267558624702e-05,
+      "loss": 1.1371,
+      "step": 88395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2326020914934164e-05,
+      "loss": 1.3198,
+      "step": 88400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2325285954115297e-05,
+      "loss": 1.1015,
+      "step": 88405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.23245509800178e-05,
+      "loss": 1.2474,
+      "step": 88410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2323815992645864e-05,
+      "loss": 0.8905,
+      "step": 88415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.232308099200369e-05,
+      "loss": 1.0476,
+      "step": 88420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2322345978095479e-05,
+      "loss": 1.0996,
+      "step": 88425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2321610950925416e-05,
+      "loss": 1.2071,
+      "step": 88430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2320875910497707e-05,
+      "loss": 1.1438,
+      "step": 88435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2320140856816549e-05,
+      "loss": 1.2544,
+      "step": 88440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2319405789886138e-05,
+      "loss": 1.3078,
+      "step": 88445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2318670709710676e-05,
+      "loss": 1.2894,
+      "step": 88450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2317935616294354e-05,
+      "loss": 1.3031,
+      "step": 88455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2317200509641374e-05,
+      "loss": 1.2596,
+      "step": 88460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2316465389755931e-05,
+      "loss": 1.3535,
+      "step": 88465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2315730256642228e-05,
+      "loss": 1.2293,
+      "step": 88470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2314995110304455e-05,
+      "loss": 1.4048,
+      "step": 88475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.231425995074682e-05,
+      "loss": 1.1051,
+      "step": 88480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2313524777973513e-05,
+      "loss": 1.1081,
+      "step": 88485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2312789591988735e-05,
+      "loss": 1.1911,
+      "step": 88490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2312054392796689e-05,
+      "loss": 1.047,
+      "step": 88495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2311319180401562e-05,
+      "loss": 0.989,
+      "step": 88500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2310583954807565e-05,
+      "loss": 1.251,
+      "step": 88505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2309848716018887e-05,
+      "loss": 1.1714,
+      "step": 88510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.230911346403973e-05,
+      "loss": 0.8385,
+      "step": 88515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2308378198874299e-05,
+      "loss": 1.229,
+      "step": 88520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2307642920526781e-05,
+      "loss": 1.217,
+      "step": 88525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2306907629001381e-05,
+      "loss": 1.4103,
+      "step": 88530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2306172324302301e-05,
+      "loss": 1.6046,
+      "step": 88535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2305437006433734e-05,
+      "loss": 0.9739,
+      "step": 88540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2304701675399885e-05,
+      "loss": 1.1589,
+      "step": 88545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2303966331204946e-05,
+      "loss": 0.9701,
+      "step": 88550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2303230973853122e-05,
+      "loss": 1.1722,
+      "step": 88555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.230249560334861e-05,
+      "loss": 1.4967,
+      "step": 88560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2301760219695608e-05,
+      "loss": 1.0757,
+      "step": 88565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.230102482289832e-05,
+      "loss": 1.3082,
+      "step": 88570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2300289412960942e-05,
+      "loss": 1.2343,
+      "step": 88575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2299553989887674e-05,
+      "loss": 1.1045,
+      "step": 88580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2298818553682716e-05,
+      "loss": 1.0221,
+      "step": 88585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2298083104350266e-05,
+      "loss": 1.4092,
+      "step": 88590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2297347641894526e-05,
+      "loss": 1.0457,
+      "step": 88595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2296612166319696e-05,
+      "loss": 1.4285,
+      "step": 88600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2295876677629974e-05,
+      "loss": 1.0609,
+      "step": 88605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2295141175829566e-05,
+      "loss": 1.2669,
+      "step": 88610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2294405660922662e-05,
+      "loss": 1.3189,
+      "step": 88615
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.229367013291347e-05,
+      "loss": 1.3942,
+      "step": 88620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.229293459180619e-05,
+      "loss": 1.4431,
+      "step": 88625
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2292199037605015e-05,
+      "loss": 1.366,
+      "step": 88630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2291463470314154e-05,
+      "loss": 1.2218,
+      "step": 88635
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2290727889937803e-05,
+      "loss": 1.3342,
+      "step": 88640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2289992296480164e-05,
+      "loss": 1.06,
+      "step": 88645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2289256689945436e-05,
+      "loss": 1.1518,
+      "step": 88650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2288521070337824e-05,
+      "loss": 1.2548,
+      "step": 88655
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2287785437661526e-05,
+      "loss": 1.3103,
+      "step": 88660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.228704979192074e-05,
+      "loss": 1.2694,
+      "step": 88665
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.228631413311967e-05,
+      "loss": 1.2044,
+      "step": 88670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2285578461262514e-05,
+      "loss": 1.2506,
+      "step": 88675
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2284842776353482e-05,
+      "loss": 1.4493,
+      "step": 88680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2284107078396767e-05,
+      "loss": 1.4446,
+      "step": 88685
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.228337136739657e-05,
+      "loss": 1.2059,
+      "step": 88690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2282635643357094e-05,
+      "loss": 1.0935,
+      "step": 88695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2281899906282542e-05,
+      "loss": 1.1786,
+      "step": 88700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2281164156177117e-05,
+      "loss": 1.5498,
+      "step": 88705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2280428393045012e-05,
+      "loss": 1.1542,
+      "step": 88710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.227969261689044e-05,
+      "loss": 1.2326,
+      "step": 88715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2278956827717594e-05,
+      "loss": 1.4061,
+      "step": 88720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2278221025530677e-05,
+      "loss": 1.2434,
+      "step": 88725
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2277485210333896e-05,
+      "loss": 1.2253,
+      "step": 88730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2276749382131445e-05,
+      "loss": 1.3439,
+      "step": 88735
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2276013540927535e-05,
+      "loss": 1.2832,
+      "step": 88740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2275277686726361e-05,
+      "loss": 1.0562,
+      "step": 88745
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2274541819532132e-05,
+      "loss": 1.2508,
+      "step": 88750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2273805939349042e-05,
+      "loss": 1.5487,
+      "step": 88755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2273070046181297e-05,
+      "loss": 1.0668,
+      "step": 88760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.22723341400331e-05,
+      "loss": 1.1029,
+      "step": 88765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2271598220908653e-05,
+      "loss": 1.1839,
+      "step": 88770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2270862288812157e-05,
+      "loss": 1.1544,
+      "step": 88775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2270126343747819e-05,
+      "loss": 0.9752,
+      "step": 88780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2269390385719837e-05,
+      "loss": 1.1024,
+      "step": 88785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2268654414732414e-05,
+      "loss": 1.3129,
+      "step": 88790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2267918430789757e-05,
+      "loss": 1.316,
+      "step": 88795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2267182433896064e-05,
+      "loss": 1.0721,
+      "step": 88800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2266446424055538e-05,
+      "loss": 1.247,
+      "step": 88805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2265710401272388e-05,
+      "loss": 1.1476,
+      "step": 88810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.226497436555081e-05,
+      "loss": 1.0244,
+      "step": 88815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.226423831689501e-05,
+      "loss": 1.3844,
+      "step": 88820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2263502255309196e-05,
+      "loss": 0.9288,
+      "step": 88825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.226276618079756e-05,
+      "loss": 0.8893,
+      "step": 88830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.226203009336432e-05,
+      "loss": 1.2671,
+      "step": 88835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2261293993013663e-05,
+      "loss": 1.041,
+      "step": 88840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2260557879749806e-05,
+      "loss": 1.2058,
+      "step": 88845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2259821753576946e-05,
+      "loss": 1.3011,
+      "step": 88850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.225908561449929e-05,
+      "loss": 1.4207,
+      "step": 88855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2258349462521038e-05,
+      "loss": 1.4384,
+      "step": 88860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2257613297646398e-05,
+      "loss": 1.2002,
+      "step": 88865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2256877119879572e-05,
+      "loss": 1.257,
+      "step": 88870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.225614092922476e-05,
+      "loss": 1.0576,
+      "step": 88875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2255404725686173e-05,
+      "loss": 1.3588,
+      "step": 88880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2254668509268011e-05,
+      "loss": 0.9491,
+      "step": 88885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2253932279974481e-05,
+      "loss": 1.0885,
+      "step": 88890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2253196037809783e-05,
+      "loss": 1.2487,
+      "step": 88895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2252459782778125e-05,
+      "loss": 1.1817,
+      "step": 88900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2251723514883708e-05,
+      "loss": 1.3557,
+      "step": 88905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.225098723413074e-05,
+      "loss": 1.3711,
+      "step": 88910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2250250940523426e-05,
+      "loss": 1.4997,
+      "step": 88915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.224951463406597e-05,
+      "loss": 1.3148,
+      "step": 88920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2248778314762568e-05,
+      "loss": 0.8166,
+      "step": 88925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.224804198261744e-05,
+      "loss": 1.0818,
+      "step": 88930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2247305637634779e-05,
+      "loss": 1.1262,
+      "step": 88935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2246569279818799e-05,
+      "loss": 1.1811,
+      "step": 88940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2245832909173695e-05,
+      "loss": 1.1146,
+      "step": 88945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2245096525703677e-05,
+      "loss": 1.1276,
+      "step": 88950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2244360129412951e-05,
+      "loss": 1.2211,
+      "step": 88955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2243623720305724e-05,
+      "loss": 1.4502,
+      "step": 88960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2242887298386197e-05,
+      "loss": 1.3128,
+      "step": 88965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2242150863658577e-05,
+      "loss": 1.0866,
+      "step": 88970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2241414416127071e-05,
+      "loss": 1.1544,
+      "step": 88975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2240677955795882e-05,
+      "loss": 0.9582,
+      "step": 88980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2239941482669218e-05,
+      "loss": 1.2575,
+      "step": 88985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2239204996751281e-05,
+      "loss": 1.1184,
+      "step": 88990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2238468498046282e-05,
+      "loss": 1.042,
+      "step": 88995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2237731986558421e-05,
+      "loss": 1.5803,
+      "step": 89000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.223699546229191e-05,
+      "loss": 1.3101,
+      "step": 89005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.223625892525095e-05,
+      "loss": 1.2171,
+      "step": 89010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2235522375439746e-05,
+      "loss": 1.2039,
+      "step": 89015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2234785812862512e-05,
+      "loss": 1.3867,
+      "step": 89020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2234049237523447e-05,
+      "loss": 0.967,
+      "step": 89025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2233312649426755e-05,
+      "loss": 1.5948,
+      "step": 89030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2232576048576655e-05,
+      "loss": 1.1187,
+      "step": 89035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2231839434977338e-05,
+      "loss": 1.2703,
+      "step": 89040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2231102808633021e-05,
+      "loss": 1.1738,
+      "step": 89045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2230366169547904e-05,
+      "loss": 1.2039,
+      "step": 89050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2229629517726199e-05,
+      "loss": 1.1141,
+      "step": 89055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2228892853172109e-05,
+      "loss": 1.3092,
+      "step": 89060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2228156175889843e-05,
+      "loss": 1.1893,
+      "step": 89065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2227419485883607e-05,
+      "loss": 1.1004,
+      "step": 89070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2226682783157607e-05,
+      "loss": 1.0449,
+      "step": 89075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.222594606771605e-05,
+      "loss": 1.6009,
+      "step": 89080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 1.2309,
+      "step": 89085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2224472598703099e-05,
+      "loss": 0.9821,
+      "step": 89090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2223735845140117e-05,
+      "loss": 1.5911,
+      "step": 89095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2222999078878405e-05,
+      "loss": 1.0847,
+      "step": 89100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2222262299922174e-05,
+      "loss": 1.2159,
+      "step": 89105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2221525508275628e-05,
+      "loss": 1.1097,
+      "step": 89110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2220788703942982e-05,
+      "loss": 1.2976,
+      "step": 89115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2220051886928432e-05,
+      "loss": 1.2663,
+      "step": 89120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2219315057236196e-05,
+      "loss": 1.0404,
+      "step": 89125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2218578214870478e-05,
+      "loss": 1.1011,
+      "step": 89130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.221784135983548e-05,
+      "loss": 1.1277,
+      "step": 89135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.221710449213542e-05,
+      "loss": 1.2838,
+      "step": 89140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2216367611774498e-05,
+      "loss": 1.2203,
+      "step": 89145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2215630718756928e-05,
+      "loss": 1.0754,
+      "step": 89150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.221489381308691e-05,
+      "loss": 1.1522,
+      "step": 89155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2214156894768658e-05,
+      "loss": 1.0172,
+      "step": 89160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2213419963806386e-05,
+      "loss": 0.958,
+      "step": 89165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2212683020204288e-05,
+      "loss": 1.1759,
+      "step": 89170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2211946063966581e-05,
+      "loss": 1.3768,
+      "step": 89175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2211209095097474e-05,
+      "loss": 1.2442,
+      "step": 89180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2210472113601173e-05,
+      "loss": 1.3069,
+      "step": 89185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.220973511948189e-05,
+      "loss": 1.3308,
+      "step": 89190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2208998112743826e-05,
+      "loss": 1.2144,
+      "step": 89195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2208261093391198e-05,
+      "loss": 1.0758,
+      "step": 89200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.220752406142821e-05,
+      "loss": 1.1758,
+      "step": 89205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2206787016859074e-05,
+      "loss": 1.3495,
+      "step": 89210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2206049959687993e-05,
+      "loss": 1.3126,
+      "step": 89215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2205312889919187e-05,
+      "loss": 1.109,
+      "step": 89220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2204575807556855e-05,
+      "loss": 1.1534,
+      "step": 89225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2203838712605209e-05,
+      "loss": 1.1435,
+      "step": 89230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.220310160506846e-05,
+      "loss": 1.1409,
+      "step": 89235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2202364484950811e-05,
+      "loss": 1.2258,
+      "step": 89240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2201627352256482e-05,
+      "loss": 1.4601,
+      "step": 89245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2200890206989673e-05,
+      "loss": 1.1306,
+      "step": 89250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2200153049154598e-05,
+      "loss": 1.177,
+      "step": 89255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2199415878755468e-05,
+      "loss": 1.0768,
+      "step": 89260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2198678695796487e-05,
+      "loss": 1.1361,
+      "step": 89265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2197941500281871e-05,
+      "loss": 1.3996,
+      "step": 89270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2197204292215825e-05,
+      "loss": 1.0449,
+      "step": 89275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2196467071602562e-05,
+      "loss": 1.0702,
+      "step": 89280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2195729838446288e-05,
+      "loss": 1.551,
+      "step": 89285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.219499259275122e-05,
+      "loss": 1.1303,
+      "step": 89290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.219425533452156e-05,
+      "loss": 1.2112,
+      "step": 89295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2193518063761525e-05,
+      "loss": 1.3788,
+      "step": 89300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2192780780475319e-05,
+      "loss": 1.1846,
+      "step": 89305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2192043484667157e-05,
+      "loss": 1.2322,
+      "step": 89310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.219130617634125e-05,
+      "loss": 0.9603,
+      "step": 89315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2190568855501798e-05,
+      "loss": 1.1023,
+      "step": 89320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2189831522153028e-05,
+      "loss": 1.3219,
+      "step": 89325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.218909417629914e-05,
+      "loss": 1.2703,
+      "step": 89330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2188356817944346e-05,
+      "loss": 1.0656,
+      "step": 89335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2187619447092858e-05,
+      "loss": 1.2044,
+      "step": 89340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2186882063748883e-05,
+      "loss": 1.3201,
+      "step": 89345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2186144667916642e-05,
+      "loss": 1.2036,
+      "step": 89350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2185407259600332e-05,
+      "loss": 1.2902,
+      "step": 89355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2184669838804175e-05,
+      "loss": 1.3118,
+      "step": 89360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2183932405532381e-05,
+      "loss": 1.1267,
+      "step": 89365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2183194959789152e-05,
+      "loss": 1.2036,
+      "step": 89370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2182457501578713e-05,
+      "loss": 1.2629,
+      "step": 89375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.218172003090526e-05,
+      "loss": 1.0342,
+      "step": 89380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2180982547773015e-05,
+      "loss": 1.2877,
+      "step": 89385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2180245052186188e-05,
+      "loss": 1.2581,
+      "step": 89390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.217950754414899e-05,
+      "loss": 1.5136,
+      "step": 89395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2178770023665629e-05,
+      "loss": 1.0636,
+      "step": 89400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2178032490740322e-05,
+      "loss": 1.2546,
+      "step": 89405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2177294945377274e-05,
+      "loss": 1.1203,
+      "step": 89410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2176557387580703e-05,
+      "loss": 1.4474,
+      "step": 89415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2175819817354818e-05,
+      "loss": 1.1894,
+      "step": 89420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2175082234703835e-05,
+      "loss": 1.2239,
+      "step": 89425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2174344639631958e-05,
+      "loss": 1.0625,
+      "step": 89430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2173607032143407e-05,
+      "loss": 1.5361,
+      "step": 89435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2172869412242387e-05,
+      "loss": 1.3942,
+      "step": 89440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2172131779933118e-05,
+      "loss": 1.8331,
+      "step": 89445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2171394135219805e-05,
+      "loss": 1.136,
+      "step": 89450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2170656478106668e-05,
+      "loss": 1.2354,
+      "step": 89455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2169918808597908e-05,
+      "loss": 1.2029,
+      "step": 89460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2169181126697752e-05,
+      "loss": 1.2854,
+      "step": 89465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2168443432410402e-05,
+      "loss": 0.9556,
+      "step": 89470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2167705725740071e-05,
+      "loss": 1.1294,
+      "step": 89475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.216696800669098e-05,
+      "loss": 1.7089,
+      "step": 89480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.216623027526733e-05,
+      "loss": 1.0898,
+      "step": 89485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2165492531473344e-05,
+      "loss": 1.1838,
+      "step": 89490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.216475477531323e-05,
+      "loss": 1.0394,
+      "step": 89495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2164017006791202e-05,
+      "loss": 1.3576,
+      "step": 89500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2163279225911473e-05,
+      "loss": 1.5305,
+      "step": 89505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2162541432678254e-05,
+      "loss": 1.2978,
+      "step": 89510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2161803627095763e-05,
+      "loss": 1.1226,
+      "step": 89515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2161065809168212e-05,
+      "loss": 1.4674,
+      "step": 89520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2160327978899809e-05,
+      "loss": 1.2528,
+      "step": 89525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2159590136294774e-05,
+      "loss": 0.9897,
+      "step": 89530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2158852281357316e-05,
+      "loss": 1.5295,
+      "step": 89535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2158114414091651e-05,
+      "loss": 1.1419,
+      "step": 89540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.215737653450199e-05,
+      "loss": 1.4152,
+      "step": 89545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2156638642592554e-05,
+      "loss": 1.247,
+      "step": 89550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2155900738367546e-05,
+      "loss": 1.3391,
+      "step": 89555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2155162821831187e-05,
+      "loss": 1.5459,
+      "step": 89560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.215442489298769e-05,
+      "loss": 1.1081,
+      "step": 89565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2153686951841265e-05,
+      "loss": 1.2941,
+      "step": 89570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2152948998396137e-05,
+      "loss": 1.5964,
+      "step": 89575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2152211032656505e-05,
+      "loss": 1.2024,
+      "step": 89580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.215147305462659e-05,
+      "loss": 1.3347,
+      "step": 89585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.215073506431061e-05,
+      "loss": 0.9894,
+      "step": 89590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2149997061712775e-05,
+      "loss": 1.1324,
+      "step": 89595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.21492590468373e-05,
+      "loss": 1.1283,
+      "step": 89600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2148521019688401e-05,
+      "loss": 1.178,
+      "step": 89605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2147782980270291e-05,
+      "loss": 1.1573,
+      "step": 89610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2147044928587185e-05,
+      "loss": 1.2446,
+      "step": 89615
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2146306864643298e-05,
+      "loss": 1.3363,
+      "step": 89620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2145568788442843e-05,
+      "loss": 1.4887,
+      "step": 89625
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2144830699990038e-05,
+      "loss": 1.2409,
+      "step": 89630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2144092599289094e-05,
+      "loss": 1.5678,
+      "step": 89635
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.214335448634423e-05,
+      "loss": 1.556,
+      "step": 89640
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2142616361159658e-05,
+      "loss": 1.4523,
+      "step": 89645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2141878223739592e-05,
+      "loss": 1.4975,
+      "step": 89650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2141140074088253e-05,
+      "loss": 1.1974,
+      "step": 89655
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.214040191220985e-05,
+      "loss": 1.0306,
+      "step": 89660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2139663738108605e-05,
+      "loss": 1.3915,
+      "step": 89665
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2138925551788726e-05,
+      "loss": 1.102,
+      "step": 89670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.213818735325443e-05,
+      "loss": 1.0635,
+      "step": 89675
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2137449142509938e-05,
+      "loss": 1.2491,
+      "step": 89680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2136710919559458e-05,
+      "loss": 1.3457,
+      "step": 89685
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2135972684407211e-05,
+      "loss": 0.9505,
+      "step": 89690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2135234437057413e-05,
+      "loss": 1.3764,
+      "step": 89695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2134496177514277e-05,
+      "loss": 1.156,
+      "step": 89700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2133757905782017e-05,
+      "loss": 1.2988,
+      "step": 89705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2133019621864855e-05,
+      "loss": 1.1341,
+      "step": 89710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2132281325767003e-05,
+      "loss": 1.3335,
+      "step": 89715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2131543017492677e-05,
+      "loss": 1.2665,
+      "step": 89720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2130804697046094e-05,
+      "loss": 1.2216,
+      "step": 89725
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2130066364431471e-05,
+      "loss": 1.95,
+      "step": 89730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2129328019653024e-05,
+      "loss": 1.3682,
+      "step": 89735
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2128589662714967e-05,
+      "loss": 1.569,
+      "step": 89740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.212785129362152e-05,
+      "loss": 1.3258,
+      "step": 89745
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2127112912376895e-05,
+      "loss": 1.3641,
+      "step": 89750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2126374518985312e-05,
+      "loss": 1.0824,
+      "step": 89755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.212563611345099e-05,
+      "loss": 1.1034,
+      "step": 89760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2124897695778137e-05,
+      "loss": 1.0476,
+      "step": 89765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2124159265970979e-05,
+      "loss": 1.2425,
+      "step": 89770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2123420824033729e-05,
+      "loss": 1.0518,
+      "step": 89775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2122682369970602e-05,
+      "loss": 1.1823,
+      "step": 89780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.212194390378582e-05,
+      "loss": 1.2505,
+      "step": 89785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2121205425483593e-05,
+      "loss": 1.0109,
+      "step": 89790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2120466935068145e-05,
+      "loss": 1.1506,
+      "step": 89795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2119728432543688e-05,
+      "loss": 1.4149,
+      "step": 89800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2118989917914445e-05,
+      "loss": 1.2807,
+      "step": 89805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.211825139118463e-05,
+      "loss": 1.1597,
+      "step": 89810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2117512852358457e-05,
+      "loss": 1.2063,
+      "step": 89815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2116774301440148e-05,
+      "loss": 1.1644,
+      "step": 89820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.211603573843392e-05,
+      "loss": 1.1207,
+      "step": 89825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.211529716334399e-05,
+      "loss": 1.3611,
+      "step": 89830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2114558576174573e-05,
+      "loss": 1.3115,
+      "step": 89835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2113819976929891e-05,
+      "loss": 1.2632,
+      "step": 89840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.211308136561416e-05,
+      "loss": 1.2075,
+      "step": 89845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2112342742231598e-05,
+      "loss": 1.0297,
+      "step": 89850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2111604106786423e-05,
+      "loss": 1.2253,
+      "step": 89855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.211086545928285e-05,
+      "loss": 1.6757,
+      "step": 89860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2110126799725104e-05,
+      "loss": 1.2289,
+      "step": 89865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.21093881281174e-05,
+      "loss": 1.2006,
+      "step": 89870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.210864944446395e-05,
+      "loss": 1.0723,
+      "step": 89875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.210791074876898e-05,
+      "loss": 1.3911,
+      "step": 89880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2107172041036703e-05,
+      "loss": 1.1896,
+      "step": 89885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2106433321271347e-05,
+      "loss": 1.0818,
+      "step": 89890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2105694589477116e-05,
+      "loss": 1.2196,
+      "step": 89895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2104955845658242e-05,
+      "loss": 1.0929,
+      "step": 89900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2104217089818937e-05,
+      "loss": 1.3623,
+      "step": 89905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.210347832196342e-05,
+      "loss": 1.179,
+      "step": 89910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2102739542095911e-05,
+      "loss": 1.3814,
+      "step": 89915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2102000750220629e-05,
+      "loss": 1.1299,
+      "step": 89920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2101261946341789e-05,
+      "loss": 1.165,
+      "step": 89925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2100523130463614e-05,
+      "loss": 1.098,
+      "step": 89930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2099784302590325e-05,
+      "loss": 1.1052,
+      "step": 89935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2099045462726136e-05,
+      "loss": 1.3173,
+      "step": 89940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2098306610875268e-05,
+      "loss": 1.2799,
+      "step": 89945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2097567747041943e-05,
+      "loss": 1.2386,
+      "step": 89950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2096828871230375e-05,
+      "loss": 1.0271,
+      "step": 89955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2096089983444787e-05,
+      "loss": 1.2227,
+      "step": 89960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2095351083689397e-05,
+      "loss": 1.588,
+      "step": 89965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.209461217196843e-05,
+      "loss": 1.0915,
+      "step": 89970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2093873248286098e-05,
+      "loss": 1.2005,
+      "step": 89975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2093134312646621e-05,
+      "loss": 1.2591,
+      "step": 89980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2092395365054225e-05,
+      "loss": 1.3112,
+      "step": 89985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2091656405513123e-05,
+      "loss": 1.2284,
+      "step": 89990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2090917434027542e-05,
+      "loss": 1.3955,
+      "step": 89995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2090178450601695e-05,
+      "loss": 0.9019,
+      "step": 90000
+    },
+    {
+      "epoch": 0.43,
+      "eval_loss": 1.2240839004516602,
+      "eval_runtime": 6474.7236,
+      "eval_samples_per_second": 3.569,
+      "eval_steps_per_second": 1.785,
+      "step": 90000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2089439455239804e-05,
+      "loss": 1.2171,
+      "step": 90005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2088700447946093e-05,
+      "loss": 1.3638,
+      "step": 90010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2087961428724778e-05,
+      "loss": 1.346,
+      "step": 90015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2087222397580078e-05,
+      "loss": 1.6355,
+      "step": 90020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2086483354516216e-05,
+      "loss": 1.1529,
+      "step": 90025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2085744299537414e-05,
+      "loss": 1.327,
+      "step": 90030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2085005232647888e-05,
+      "loss": 0.8125,
+      "step": 90035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2084266153851862e-05,
+      "loss": 1.1579,
+      "step": 90040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2083527063153558e-05,
+      "loss": 1.0531,
+      "step": 90045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.208278796055719e-05,
+      "loss": 1.2524,
+      "step": 90050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2082048846066983e-05,
+      "loss": 1.1675,
+      "step": 90055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2081309719687158e-05,
+      "loss": 1.0117,
+      "step": 90060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2080570581421933e-05,
+      "loss": 1.1566,
+      "step": 90065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2079831431275535e-05,
+      "loss": 1.7776,
+      "step": 90070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2079092269252178e-05,
+      "loss": 1.6034,
+      "step": 90075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2078353095356085e-05,
+      "loss": 1.3545,
+      "step": 90080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2077613909591479e-05,
+      "loss": 1.4015,
+      "step": 90085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.207687471196258e-05,
+      "loss": 1.3483,
+      "step": 90090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2076135502473609e-05,
+      "loss": 1.2277,
+      "step": 90095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.207539628112879e-05,
+      "loss": 1.0862,
+      "step": 90100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.207465704793234e-05,
+      "loss": 1.0986,
+      "step": 90105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2073917802888481e-05,
+      "loss": 1.3383,
+      "step": 90110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.207317854600144e-05,
+      "loss": 1.0935,
+      "step": 90115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2072439277275432e-05,
+      "loss": 1.3619,
+      "step": 90120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2071699996714681e-05,
+      "loss": 1.1968,
+      "step": 90125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2070960704323409e-05,
+      "loss": 1.1481,
+      "step": 90130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2070221400105837e-05,
+      "loss": 1.2449,
+      "step": 90135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2069482084066188e-05,
+      "loss": 1.253,
+      "step": 90140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2068742756208684e-05,
+      "loss": 1.6731,
+      "step": 90145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2068003416537546e-05,
+      "loss": 1.6265,
+      "step": 90150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2067264065056994e-05,
+      "loss": 1.4557,
+      "step": 90155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2066524701771253e-05,
+      "loss": 1.2597,
+      "step": 90160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2065785326684544e-05,
+      "loss": 1.0494,
+      "step": 90165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2065045939801094e-05,
+      "loss": 1.3549,
+      "step": 90170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2064306541125117e-05,
+      "loss": 1.319,
+      "step": 90175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.206356713066084e-05,
+      "loss": 1.1853,
+      "step": 90180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2062827708412485e-05,
+      "loss": 1.2556,
+      "step": 90185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2062088274384273e-05,
+      "loss": 1.299,
+      "step": 90190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2061348828580433e-05,
+      "loss": 1.3916,
+      "step": 90195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2060609371005175e-05,
+      "loss": 1.0398,
+      "step": 90200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2059869901662735e-05,
+      "loss": 1.2045,
+      "step": 90205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2059130420557328e-05,
+      "loss": 1.0421,
+      "step": 90210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2058390927693178e-05,
+      "loss": 1.5002,
+      "step": 90215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.205765142307451e-05,
+      "loss": 1.3803,
+      "step": 90220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2056911906705544e-05,
+      "loss": 1.2379,
+      "step": 90225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2056172378590505e-05,
+      "loss": 1.4121,
+      "step": 90230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2055432838733616e-05,
+      "loss": 1.3563,
+      "step": 90235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2054693287139101e-05,
+      "loss": 1.2795,
+      "step": 90240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2053953723811183e-05,
+      "loss": 1.3467,
+      "step": 90245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2053214148754081e-05,
+      "loss": 1.3299,
+      "step": 90250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2052474561972026e-05,
+      "loss": 1.2151,
+      "step": 90255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2051734963469235e-05,
+      "loss": 1.2601,
+      "step": 90260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2050995353249934e-05,
+      "loss": 1.0119,
+      "step": 90265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2050255731318346e-05,
+      "loss": 1.0864,
+      "step": 90270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2049516097678695e-05,
+      "loss": 1.0281,
+      "step": 90275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2048776452335208e-05,
+      "loss": 1.2097,
+      "step": 90280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2048036795292101e-05,
+      "loss": 1.0038,
+      "step": 90285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2047297126553606e-05,
+      "loss": 1.3953,
+      "step": 90290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2046557446123939e-05,
+      "loss": 1.2912,
+      "step": 90295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2045817754007333e-05,
+      "loss": 1.1372,
+      "step": 90300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2045078050208004e-05,
+      "loss": 1.2391,
+      "step": 90305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2044338334730178e-05,
+      "loss": 1.3404,
+      "step": 90310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2043598607578087e-05,
+      "loss": 1.3078,
+      "step": 90315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2042858868755942e-05,
+      "loss": 1.2476,
+      "step": 90320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2042119118267978e-05,
+      "loss": 1.3494,
+      "step": 90325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2041379356118411e-05,
+      "loss": 1.5356,
+      "step": 90330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2040639582311475e-05,
+      "loss": 1.162,
+      "step": 90335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2039899796851389e-05,
+      "loss": 1.3974,
+      "step": 90340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2039159999742375e-05,
+      "loss": 0.9953,
+      "step": 90345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2038420190988665e-05,
+      "loss": 1.2579,
+      "step": 90350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2037680370594475e-05,
+      "loss": 1.1879,
+      "step": 90355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2036940538564035e-05,
+      "loss": 1.4412,
+      "step": 90360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.203620069490157e-05,
+      "loss": 0.9213,
+      "step": 90365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2035460839611303e-05,
+      "loss": 1.1481,
+      "step": 90370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2034720972697463e-05,
+      "loss": 1.2876,
+      "step": 90375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2033981094164268e-05,
+      "loss": 1.2777,
+      "step": 90380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2033241204015948e-05,
+      "loss": 1.3457,
+      "step": 90385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2032501302256727e-05,
+      "loss": 1.4647,
+      "step": 90390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2031761388890833e-05,
+      "loss": 1.2917,
+      "step": 90395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2031021463922483e-05,
+      "loss": 1.0665,
+      "step": 90400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2030281527355913e-05,
+      "loss": 1.4162,
+      "step": 90405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2029541579195344e-05,
+      "loss": 1.3014,
+      "step": 90410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2028801619444996e-05,
+      "loss": 1.4589,
+      "step": 90415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2028061648109104e-05,
+      "loss": 1.4192,
+      "step": 90420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2027321665191887e-05,
+      "loss": 1.016,
+      "step": 90425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2026581670697575e-05,
+      "loss": 1.4257,
+      "step": 90430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2025841664630389e-05,
+      "loss": 1.1824,
+      "step": 90435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2025101646994559e-05,
+      "loss": 0.9314,
+      "step": 90440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.202436161779431e-05,
+      "loss": 0.9828,
+      "step": 90445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2023621577033865e-05,
+      "loss": 1.1305,
+      "step": 90450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2022881524717454e-05,
+      "loss": 1.2005,
+      "step": 90455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2022141460849302e-05,
+      "loss": 1.0477,
+      "step": 90460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2021401385433633e-05,
+      "loss": 1.8219,
+      "step": 90465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2020661298474676e-05,
+      "loss": 1.1408,
+      "step": 90470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2019921199976656e-05,
+      "loss": 1.072,
+      "step": 90475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2019181089943801e-05,
+      "loss": 1.1709,
+      "step": 90480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2018440968380333e-05,
+      "loss": 1.2111,
+      "step": 90485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2017700835290484e-05,
+      "loss": 1.3438,
+      "step": 90490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2016960690678474e-05,
+      "loss": 1.261,
+      "step": 90495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2016220534548539e-05,
+      "loss": 1.5046,
+      "step": 90500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2015480366904897e-05,
+      "loss": 1.2144,
+      "step": 90505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2014740187751781e-05,
+      "loss": 1.1675,
+      "step": 90510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2013999997093414e-05,
+      "loss": 1.1948,
+      "step": 90515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2013259794934021e-05,
+      "loss": 0.8481,
+      "step": 90520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2012519581277835e-05,
+      "loss": 1.171,
+      "step": 90525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2011779356129077e-05,
+      "loss": 1.2193,
+      "step": 90530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2011039119491979e-05,
+      "loss": 1.298,
+      "step": 90535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2010298871370762e-05,
+      "loss": 1.1199,
+      "step": 90540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.200955861176966e-05,
+      "loss": 1.5049,
+      "step": 90545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2008818340692898e-05,
+      "loss": 1.1929,
+      "step": 90550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2008078058144703e-05,
+      "loss": 1.0119,
+      "step": 90555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2007337764129302e-05,
+      "loss": 1.0587,
+      "step": 90560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2006597458650924e-05,
+      "loss": 1.3346,
+      "step": 90565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2005857141713793e-05,
+      "loss": 1.3482,
+      "step": 90570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2005116813322139e-05,
+      "loss": 1.217,
+      "step": 90575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.200437647348019e-05,
+      "loss": 1.1728,
+      "step": 90580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2003636122192175e-05,
+      "loss": 1.5128,
+      "step": 90585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2002895759462318e-05,
+      "loss": 1.237,
+      "step": 90590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2002155385294849e-05,
+      "loss": 1.1116,
+      "step": 90595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2001414999693995e-05,
+      "loss": 1.2431,
+      "step": 90600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2000674602663988e-05,
+      "loss": 1.123,
+      "step": 90605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.199993419420905e-05,
+      "loss": 1.1585,
+      "step": 90610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1999193774333415e-05,
+      "loss": 1.2217,
+      "step": 90615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1998453343041305e-05,
+      "loss": 1.5444,
+      "step": 90620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.199771290033695e-05,
+      "loss": 1.117,
+      "step": 90625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1996972446224586e-05,
+      "loss": 1.1993,
+      "step": 90630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1996231980708431e-05,
+      "loss": 1.2022,
+      "step": 90635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.199549150379272e-05,
+      "loss": 1.1074,
+      "step": 90640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1994751015481679e-05,
+      "loss": 1.1018,
+      "step": 90645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1994010515779535e-05,
+      "loss": 1.2874,
+      "step": 90650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.199327000469052e-05,
+      "loss": 0.8718,
+      "step": 90655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.199252948221886e-05,
+      "loss": 1.3309,
+      "step": 90660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1991788948368786e-05,
+      "loss": 1.2752,
+      "step": 90665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1991048403144525e-05,
+      "loss": 1.1131,
+      "step": 90670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1990307846550306e-05,
+      "loss": 1.1566,
+      "step": 90675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1989567278590362e-05,
+      "loss": 0.9448,
+      "step": 90680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1988826699268916e-05,
+      "loss": 0.9922,
+      "step": 90685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.19880861085902e-05,
+      "loss": 1.2389,
+      "step": 90690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1987345506558443e-05,
+      "loss": 1.0502,
+      "step": 90695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1986604893177875e-05,
+      "loss": 1.4737,
+      "step": 90700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1985864268452723e-05,
+      "loss": 1.3513,
+      "step": 90705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.198512363238722e-05,
+      "loss": 1.2634,
+      "step": 90710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1984382984985594e-05,
+      "loss": 1.192,
+      "step": 90715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1983642326252071e-05,
+      "loss": 1.4603,
+      "step": 90720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1982901656190886e-05,
+      "loss": 0.985,
+      "step": 90725
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1982160974806262e-05,
+      "loss": 1.0834,
+      "step": 90730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.198142028210244e-05,
+      "loss": 1.3723,
+      "step": 90735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1980679578083633e-05,
+      "loss": 1.4984,
+      "step": 90740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1979938862754088e-05,
+      "loss": 1.4648,
+      "step": 90745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1979198136118023e-05,
+      "loss": 1.2684,
+      "step": 90750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1978457398179675e-05,
+      "loss": 1.2581,
+      "step": 90755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1977716648943272e-05,
+      "loss": 1.2745,
+      "step": 90760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1976975888413038e-05,
+      "loss": 1.0834,
+      "step": 90765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.197623511659321e-05,
+      "loss": 1.6666,
+      "step": 90770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1975494333488021e-05,
+      "loss": 0.9012,
+      "step": 90775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1974753539101694e-05,
+      "loss": 1.1178,
+      "step": 90780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.197401273343846e-05,
+      "loss": 1.2117,
+      "step": 90785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1973271916502553e-05,
+      "loss": 1.2546,
+      "step": 90790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1972531088298205e-05,
+      "loss": 1.3167,
+      "step": 90795
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.197179024882964e-05,
+      "loss": 1.4533,
+      "step": 90800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1971049398101093e-05,
+      "loss": 0.9373,
+      "step": 90805
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1970308536116793e-05,
+      "loss": 1.3105,
+      "step": 90810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1969567662880974e-05,
+      "loss": 1.0331,
+      "step": 90815
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1968826778397863e-05,
+      "loss": 1.1484,
+      "step": 90820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1968085882671691e-05,
+      "loss": 1.3489,
+      "step": 90825
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1967344975706691e-05,
+      "loss": 1.5033,
+      "step": 90830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1966604057507092e-05,
+      "loss": 0.936,
+      "step": 90835
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.196586312807713e-05,
+      "loss": 1.2978,
+      "step": 90840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1965122187421027e-05,
+      "loss": 1.444,
+      "step": 90845
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1964381235543022e-05,
+      "loss": 1.0613,
+      "step": 90850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1963640272447346e-05,
+      "loss": 1.276,
+      "step": 90855
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1962899298138222e-05,
+      "loss": 1.1885,
+      "step": 90860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1962158312619893e-05,
+      "loss": 1.0803,
+      "step": 90865
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1961417315896582e-05,
+      "loss": 1.416,
+      "step": 90870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1960676307972523e-05,
+      "loss": 1.5043,
+      "step": 90875
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1959935288851946e-05,
+      "loss": 1.3028,
+      "step": 90880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1959194258539088e-05,
+      "loss": 1.5329,
+      "step": 90885
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1958453217038175e-05,
+      "loss": 1.2144,
+      "step": 90890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1957712164353442e-05,
+      "loss": 1.3625,
+      "step": 90895
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1956971100489119e-05,
+      "loss": 1.2084,
+      "step": 90900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1956230025449438e-05,
+      "loss": 1.1701,
+      "step": 90905
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1955488939238633e-05,
+      "loss": 1.3257,
+      "step": 90910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1954747841860931e-05,
+      "loss": 1.5755,
+      "step": 90915
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.195400673332057e-05,
+      "loss": 1.1763,
+      "step": 90920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.195326561362178e-05,
+      "loss": 1.2252,
+      "step": 90925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.195252448276879e-05,
+      "loss": 1.16,
+      "step": 90930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1951783340765839e-05,
+      "loss": 1.4654,
+      "step": 90935
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1951042187617152e-05,
+      "loss": 1.3812,
+      "step": 90940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1950301023326966e-05,
+      "loss": 1.0985,
+      "step": 90945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1949559847899512e-05,
+      "loss": 1.3381,
+      "step": 90950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1948818661339023e-05,
+      "loss": 1.0922,
+      "step": 90955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1948077463649734e-05,
+      "loss": 1.3427,
+      "step": 90960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1947336254835868e-05,
+      "loss": 1.3096,
+      "step": 90965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.194659503490167e-05,
+      "loss": 1.5956,
+      "step": 90970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1945853803851367e-05,
+      "loss": 1.4145,
+      "step": 90975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1945112561689189e-05,
+      "loss": 1.1529,
+      "step": 90980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1944371308419376e-05,
+      "loss": 1.137,
+      "step": 90985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1943630044046154e-05,
+      "loss": 0.983,
+      "step": 90990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1942888768573762e-05,
+      "loss": 1.0399,
+      "step": 90995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.194214748200643e-05,
+      "loss": 1.1218,
+      "step": 91000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.194140618434839e-05,
+      "loss": 1.1236,
+      "step": 91005
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1940664875603876e-05,
+      "loss": 1.2876,
+      "step": 91010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1939923555777121e-05,
+      "loss": 1.3623,
+      "step": 91015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1939182224872362e-05,
+      "loss": 1.1598,
+      "step": 91020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1938440882893827e-05,
+      "loss": 1.1045,
+      "step": 91025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1937699529845753e-05,
+      "loss": 1.4656,
+      "step": 91030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1936958165732368e-05,
+      "loss": 1.2754,
+      "step": 91035
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1936216790557918e-05,
+      "loss": 1.1486,
+      "step": 91040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1935475404326621e-05,
+      "loss": 1.2095,
+      "step": 91045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1934734007042726e-05,
+      "loss": 1.2508,
+      "step": 91050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1933992598710456e-05,
+      "loss": 1.143,
+      "step": 91055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1933251179334047e-05,
+      "loss": 1.3154,
+      "step": 91060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1932509748917736e-05,
+      "loss": 1.1222,
+      "step": 91065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1931768307465752e-05,
+      "loss": 1.1184,
+      "step": 91070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1931026854982332e-05,
+      "loss": 0.8389,
+      "step": 91075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1930285391471712e-05,
+      "loss": 1.2699,
+      "step": 91080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1929543916938125e-05,
+      "loss": 1.1534,
+      "step": 91085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1928802431385802e-05,
+      "loss": 1.1787,
+      "step": 91090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1928060934818982e-05,
+      "loss": 1.452,
+      "step": 91095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1927319427241896e-05,
+      "loss": 0.8658,
+      "step": 91100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.192657790865878e-05,
+      "loss": 1.4903,
+      "step": 91105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1925836379073868e-05,
+      "loss": 0.999,
+      "step": 91110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1925094838491393e-05,
+      "loss": 1.5757,
+      "step": 91115
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1924353286915593e-05,
+      "loss": 1.3578,
+      "step": 91120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1923611724350698e-05,
+      "loss": 1.5154,
+      "step": 91125
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1922870150800949e-05,
+      "loss": 1.1817,
+      "step": 91130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1922128566270574e-05,
+      "loss": 1.2828,
+      "step": 91135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1921386970763811e-05,
+      "loss": 1.1071,
+      "step": 91140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1920645364284897e-05,
+      "loss": 1.1616,
+      "step": 91145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1919903746838064e-05,
+      "loss": 1.212,
+      "step": 91150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.191916211842755e-05,
+      "loss": 1.4859,
+      "step": 91155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1918420479057587e-05,
+      "loss": 1.1654,
+      "step": 91160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1917678828732408e-05,
+      "loss": 1.4468,
+      "step": 91165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1916937167456254e-05,
+      "loss": 1.4633,
+      "step": 91170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1916195495233358e-05,
+      "loss": 1.211,
+      "step": 91175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1915453812067953e-05,
+      "loss": 1.2764,
+      "step": 91180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.191471211796428e-05,
+      "loss": 1.1234,
+      "step": 91185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.191397041292657e-05,
+      "loss": 1.1256,
+      "step": 91190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1913228696959061e-05,
+      "loss": 1.3824,
+      "step": 91195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1912486970065982e-05,
+      "loss": 1.1622,
+      "step": 91200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1911745232251577e-05,
+      "loss": 1.3044,
+      "step": 91205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1911003483520079e-05,
+      "loss": 1.4169,
+      "step": 91210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1910261723875722e-05,
+      "loss": 1.3541,
+      "step": 91215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1909519953322743e-05,
+      "loss": 1.4287,
+      "step": 91220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.190877817186538e-05,
+      "loss": 1.0694,
+      "step": 91225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1908036379507866e-05,
+      "loss": 1.3217,
+      "step": 91230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1907294576254438e-05,
+      "loss": 1.2326,
+      "step": 91235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1906552762109332e-05,
+      "loss": 1.1118,
+      "step": 91240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.190581093707678e-05,
+      "loss": 1.2057,
+      "step": 91245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.190506910116103e-05,
+      "loss": 1.3618,
+      "step": 91250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1904327254366308e-05,
+      "loss": 1.4516,
+      "step": 91255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1903585396696852e-05,
+      "loss": 1.4879,
+      "step": 91260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.19028435281569e-05,
+      "loss": 1.0203,
+      "step": 91265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1902101648750686e-05,
+      "loss": 1.1094,
+      "step": 91270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1901359758482451e-05,
+      "loss": 1.1315,
+      "step": 91275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.190061785735643e-05,
+      "loss": 1.2076,
+      "step": 91280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1899875945376855e-05,
+      "loss": 1.6704,
+      "step": 91285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1899134022547969e-05,
+      "loss": 0.9434,
+      "step": 91290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1898392088874006e-05,
+      "loss": 1.2642,
+      "step": 91295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1897650144359208e-05,
+      "loss": 1.3181,
+      "step": 91300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1896908189007799e-05,
+      "loss": 1.2494,
+      "step": 91305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1896166222824028e-05,
+      "loss": 1.1139,
+      "step": 91310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1895424245812128e-05,
+      "loss": 1.0506,
+      "step": 91315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1894682257976336e-05,
+      "loss": 1.187,
+      "step": 91320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1893940259320886e-05,
+      "loss": 1.2144,
+      "step": 91325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1893198249850023e-05,
+      "loss": 1.2676,
+      "step": 91330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1892456229567977e-05,
+      "loss": 1.5721,
+      "step": 91335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.189171419847899e-05,
+      "loss": 1.296,
+      "step": 91340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1890972156587297e-05,
+      "loss": 1.3431,
+      "step": 91345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1890230103897133e-05,
+      "loss": 1.2413,
+      "step": 91350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1889488040412742e-05,
+      "loss": 0.8405,
+      "step": 91355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1888745966138358e-05,
+      "loss": 1.1437,
+      "step": 91360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1888003881078215e-05,
+      "loss": 0.9968,
+      "step": 91365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.188726178523656e-05,
+      "loss": 1.152,
+      "step": 91370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.188651967861762e-05,
+      "loss": 0.9894,
+      "step": 91375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1885777561225643e-05,
+      "loss": 1.0754,
+      "step": 91380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1885035433064855e-05,
+      "loss": 1.3701,
+      "step": 91385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1884293294139504e-05,
+      "loss": 1.0729,
+      "step": 91390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1883551144453826e-05,
+      "loss": 1.1631,
+      "step": 91395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1882808984012057e-05,
+      "loss": 1.0965,
+      "step": 91400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.188206681281844e-05,
+      "loss": 1.1307,
+      "step": 91405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1881324630877202e-05,
+      "loss": 0.889,
+      "step": 91410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1880582438192594e-05,
+      "loss": 1.07,
+      "step": 91415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1879840234768846e-05,
+      "loss": 1.0451,
+      "step": 91420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.18790980206102e-05,
+      "loss": 1.1475,
+      "step": 91425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1878355795720896e-05,
+      "loss": 1.0608,
+      "step": 91430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1877613560105166e-05,
+      "loss": 1.2512,
+      "step": 91435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1876871313767255e-05,
+      "loss": 1.1447,
+      "step": 91440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.18761290567114e-05,
+      "loss": 1.1909,
+      "step": 91445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1875386788941837e-05,
+      "loss": 1.1456,
+      "step": 91450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1874644510462807e-05,
+      "loss": 1.0479,
+      "step": 91455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.187390222127855e-05,
+      "loss": 1.157,
+      "step": 91460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1873159921393302e-05,
+      "loss": 1.197,
+      "step": 91465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1872417610811303e-05,
+      "loss": 1.096,
+      "step": 91470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1871675289536795e-05,
+      "loss": 1.1396,
+      "step": 91475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.187093295757401e-05,
+      "loss": 0.992,
+      "step": 91480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1870190614927196e-05,
+      "loss": 1.074,
+      "step": 91485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1869448261600584e-05,
+      "loss": 1.2219,
+      "step": 91490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1868705897598421e-05,
+      "loss": 1.1988,
+      "step": 91495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.186796352292494e-05,
+      "loss": 1.3759,
+      "step": 91500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1867221137584381e-05,
+      "loss": 1.2259,
+      "step": 91505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1866478741580986e-05,
+      "loss": 1.1038,
+      "step": 91510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1865736334918996e-05,
+      "loss": 1.2403,
+      "step": 91515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1864993917602647e-05,
+      "loss": 0.954,
+      "step": 91520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1864251489636177e-05,
+      "loss": 1.2199,
+      "step": 91525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1863509051023832e-05,
+      "loss": 1.1336,
+      "step": 91530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1862766601769846e-05,
+      "loss": 1.4342,
+      "step": 91535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.186202414187846e-05,
+      "loss": 1.2598,
+      "step": 91540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1861281671353917e-05,
+      "loss": 1.5118,
+      "step": 91545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1860539190200451e-05,
+      "loss": 1.5191,
+      "step": 91550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.185979669842231e-05,
+      "loss": 1.2646,
+      "step": 91555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1859054196023728e-05,
+      "loss": 1.1454,
+      "step": 91560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1858311683008945e-05,
+      "loss": 1.2721,
+      "step": 91565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1857569159382205e-05,
+      "loss": 1.1306,
+      "step": 91570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1856826625147741e-05,
+      "loss": 1.266,
+      "step": 91575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1856084080309807e-05,
+      "loss": 1.4885,
+      "step": 91580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1855341524872626e-05,
+      "loss": 1.2487,
+      "step": 91585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1854598958840454e-05,
+      "loss": 1.0767,
+      "step": 91590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1853856382217519e-05,
+      "loss": 1.3245,
+      "step": 91595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1853113795008069e-05,
+      "loss": 1.1,
+      "step": 91600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1852371197216344e-05,
+      "loss": 1.4874,
+      "step": 91605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.185162858884658e-05,
+      "loss": 1.1419,
+      "step": 91610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1850885969903022e-05,
+      "loss": 1.2248,
+      "step": 91615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.185014334038991e-05,
+      "loss": 1.0992,
+      "step": 91620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1849400700311486e-05,
+      "loss": 1.1663,
+      "step": 91625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1848658049671988e-05,
+      "loss": 1.2809,
+      "step": 91630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1847915388475659e-05,
+      "loss": 1.472,
+      "step": 91635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1847172716726738e-05,
+      "loss": 1.0682,
+      "step": 91640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1846430034429468e-05,
+      "loss": 1.4308,
+      "step": 91645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1845687341588088e-05,
+      "loss": 0.8908,
+      "step": 91650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1844944638206841e-05,
+      "loss": 1.4192,
+      "step": 91655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.184420192428997e-05,
+      "loss": 1.2292,
+      "step": 91660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1843459199841712e-05,
+      "loss": 1.1969,
+      "step": 91665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1842716464866311e-05,
+      "loss": 1.1713,
+      "step": 91670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1841973719368007e-05,
+      "loss": 1.1897,
+      "step": 91675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1841230963351042e-05,
+      "loss": 1.2742,
+      "step": 91680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1840488196819662e-05,
+      "loss": 1.2703,
+      "step": 91685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.18397454197781e-05,
+      "loss": 1.6233,
+      "step": 91690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1839002632230603e-05,
+      "loss": 1.56,
+      "step": 91695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1838259834181415e-05,
+      "loss": 1.5222,
+      "step": 91700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.183751702563477e-05,
+      "loss": 1.2793,
+      "step": 91705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.183677420659492e-05,
+      "loss": 0.8897,
+      "step": 91710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1836031377066094e-05,
+      "loss": 1.009,
+      "step": 91715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1835288537052547e-05,
+      "loss": 1.2977,
+      "step": 91720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1834545686558515e-05,
+      "loss": 1.3105,
+      "step": 91725
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1833802825588239e-05,
+      "loss": 1.069,
+      "step": 91730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1833059954145962e-05,
+      "loss": 1.3145,
+      "step": 91735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1832317072235929e-05,
+      "loss": 1.295,
+      "step": 91740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1831574179862377e-05,
+      "loss": 1.2121,
+      "step": 91745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1830831277029552e-05,
+      "loss": 1.5318,
+      "step": 91750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1830088363741698e-05,
+      "loss": 1.3415,
+      "step": 91755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1829345440003052e-05,
+      "loss": 1.1182,
+      "step": 91760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1828602505817863e-05,
+      "loss": 0.9885,
+      "step": 91765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.182785956119037e-05,
+      "loss": 1.5594,
+      "step": 91770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1827116606124813e-05,
+      "loss": 1.2049,
+      "step": 91775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.182637364062544e-05,
+      "loss": 1.1075,
+      "step": 91780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1825630664696486e-05,
+      "loss": 1.2007,
+      "step": 91785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1824887678342204e-05,
+      "loss": 1.0267,
+      "step": 91790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1824144681566829e-05,
+      "loss": 1.5508,
+      "step": 91795
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1823401674374609e-05,
+      "loss": 1.3268,
+      "step": 91800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1822658656769787e-05,
+      "loss": 1.3288,
+      "step": 91805
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.18219156287566e-05,
+      "loss": 1.7313,
+      "step": 91810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1821172590339295e-05,
+      "loss": 1.3877,
+      "step": 91815
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1820429541522116e-05,
+      "loss": 1.1858,
+      "step": 91820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1819686482309305e-05,
+      "loss": 1.3514,
+      "step": 91825
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1818943412705103e-05,
+      "loss": 1.0828,
+      "step": 91830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.181820033271376e-05,
+      "loss": 1.0645,
+      "step": 91835
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1817457242339513e-05,
+      "loss": 1.6748,
+      "step": 91840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1816714141586609e-05,
+      "loss": 1.2695,
+      "step": 91845
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1815971030459289e-05,
+      "loss": 1.2049,
+      "step": 91850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1815227908961797e-05,
+      "loss": 1.3977,
+      "step": 91855
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1814484777098378e-05,
+      "loss": 1.2492,
+      "step": 91860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1813741634873275e-05,
+      "loss": 1.1966,
+      "step": 91865
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1812998482290731e-05,
+      "loss": 1.2353,
+      "step": 91870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1812255319354993e-05,
+      "loss": 1.2945,
+      "step": 91875
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1811512146070301e-05,
+      "loss": 1.6673,
+      "step": 91880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.18107689624409e-05,
+      "loss": 1.3348,
+      "step": 91885
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1810025768471032e-05,
+      "loss": 1.2591,
+      "step": 91890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1809282564164947e-05,
+      "loss": 1.0722,
+      "step": 91895
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1808539349526886e-05,
+      "loss": 1.2036,
+      "step": 91900
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1807796124561089e-05,
+      "loss": 1.667,
+      "step": 91905
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1807052889271807e-05,
+      "loss": 1.4783,
+      "step": 91910
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1806309643663278e-05,
+      "loss": 1.0917,
+      "step": 91915
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1805566387739753e-05,
+      "loss": 1.1132,
+      "step": 91920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1804823121505469e-05,
+      "loss": 1.3913,
+      "step": 91925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1804079844964676e-05,
+      "loss": 1.0584,
+      "step": 91930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1803336558121616e-05,
+      "loss": 1.3735,
+      "step": 91935
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1802593260980535e-05,
+      "loss": 1.1034,
+      "step": 91940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1801849953545679e-05,
+      "loss": 1.3337,
+      "step": 91945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1801106635821288e-05,
+      "loss": 1.1715,
+      "step": 91950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.180036330781161e-05,
+      "loss": 1.5601,
+      "step": 91955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1799619969520891e-05,
+      "loss": 1.3818,
+      "step": 91960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1798876620953373e-05,
+      "loss": 1.2107,
+      "step": 91965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.17981332621133e-05,
+      "loss": 1.0598,
+      "step": 91970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.179738989300492e-05,
+      "loss": 1.1594,
+      "step": 91975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1796646513632477e-05,
+      "loss": 1.2806,
+      "step": 91980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1795903124000214e-05,
+      "loss": 0.8898,
+      "step": 91985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1795159724112385e-05,
+      "loss": 1.4185,
+      "step": 91990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.179441631397322e-05,
+      "loss": 1.0129,
+      "step": 91995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1793672893586978e-05,
+      "loss": 1.5044,
+      "step": 92000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1792929462957898e-05,
+      "loss": 1.2264,
+      "step": 92005
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1792186022090223e-05,
+      "loss": 1.3697,
+      "step": 92010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.179144257098821e-05,
+      "loss": 1.3974,
+      "step": 92015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.179069910965609e-05,
+      "loss": 1.4401,
+      "step": 92020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1789955638098118e-05,
+      "loss": 1.3997,
+      "step": 92025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1789212156318533e-05,
+      "loss": 0.9163,
+      "step": 92030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1788468664321586e-05,
+      "loss": 1.18,
+      "step": 92035
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1787725162111522e-05,
+      "loss": 1.1461,
+      "step": 92040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1786981649692584e-05,
+      "loss": 0.9985,
+      "step": 92045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1786238127069021e-05,
+      "loss": 1.2347,
+      "step": 92050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1785494594245078e-05,
+      "loss": 1.2004,
+      "step": 92055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1784751051224999e-05,
+      "loss": 1.1537,
+      "step": 92060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1784007498013033e-05,
+      "loss": 1.2742,
+      "step": 92065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1783263934613425e-05,
+      "loss": 1.4191,
+      "step": 92070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1782520361030419e-05,
+      "loss": 1.0515,
+      "step": 92075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1781776777268263e-05,
+      "loss": 1.3741,
+      "step": 92080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1781033183331204e-05,
+      "loss": 1.2968,
+      "step": 92085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1780289579223486e-05,
+      "loss": 0.9946,
+      "step": 92090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1779545964949362e-05,
+      "loss": 1.1473,
+      "step": 92095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.177880234051307e-05,
+      "loss": 1.3963,
+      "step": 92100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.177805870591886e-05,
+      "loss": 1.5151,
+      "step": 92105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1777315061170977e-05,
+      "loss": 1.1139,
+      "step": 92110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1776571406273669e-05,
+      "loss": 1.4234,
+      "step": 92115
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1775827741231187e-05,
+      "loss": 1.5073,
+      "step": 92120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1775084066047768e-05,
+      "loss": 1.3511,
+      "step": 92125
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1774340380727667e-05,
+      "loss": 1.2594,
+      "step": 92130
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1773596685275129e-05,
+      "loss": 1.2609,
+      "step": 92135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1772852979694398e-05,
+      "loss": 1.4292,
+      "step": 92140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1772109263989726e-05,
+      "loss": 1.3628,
+      "step": 92145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1771365538165354e-05,
+      "loss": 1.2441,
+      "step": 92150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1770621802225532e-05,
+      "loss": 1.0992,
+      "step": 92155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1769878056174509e-05,
+      "loss": 1.3157,
+      "step": 92160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1769134300016529e-05,
+      "loss": 1.1479,
+      "step": 92165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1768390533755843e-05,
+      "loss": 1.332,
+      "step": 92170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1767646757396693e-05,
+      "loss": 1.0989,
+      "step": 92175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.176690297094333e-05,
+      "loss": 1.2493,
+      "step": 92180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.17661591744e-05,
+      "loss": 1.5018,
+      "step": 92185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1765415367770954e-05,
+      "loss": 1.0844,
+      "step": 92190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1764671551060432e-05,
+      "loss": 1.3915,
+      "step": 92195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1763927724272692e-05,
+      "loss": 1.2642,
+      "step": 92200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1763183887411973e-05,
+      "loss": 1.3212,
+      "step": 92205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1762440040482526e-05,
+      "loss": 1.4756,
+      "step": 92210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1761696183488597e-05,
+      "loss": 1.24,
+      "step": 92215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1760952316434435e-05,
+      "loss": 1.1899,
+      "step": 92220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1760208439324292e-05,
+      "loss": 1.1484,
+      "step": 92225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.175946455216241e-05,
+      "loss": 1.2593,
+      "step": 92230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1758720654953037e-05,
+      "loss": 1.384,
+      "step": 92235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1757976747700425e-05,
+      "loss": 1.3589,
+      "step": 92240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.175723283040882e-05,
+      "loss": 1.4451,
+      "step": 92245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1756488903082475e-05,
+      "loss": 1.238,
+      "step": 92250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1755744965725626e-05,
+      "loss": 1.3643,
+      "step": 92255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1755001018342534e-05,
+      "loss": 1.4414,
+      "step": 92260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.175425706093744e-05,
+      "loss": 1.3939,
+      "step": 92265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1753513093514595e-05,
+      "loss": 1.1446,
+      "step": 92270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1752769116078246e-05,
+      "loss": 1.1176,
+      "step": 92275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1752025128632645e-05,
+      "loss": 1.2918,
+      "step": 92280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1751281131182037e-05,
+      "loss": 1.3735,
+      "step": 92285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1750537123730671e-05,
+      "loss": 1.2068,
+      "step": 92290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1749793106282799e-05,
+      "loss": 1.3284,
+      "step": 92295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1749049078842663e-05,
+      "loss": 1.087,
+      "step": 92300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.174830504141452e-05,
+      "loss": 1.2243,
+      "step": 92305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1747560994002613e-05,
+      "loss": 1.3752,
+      "step": 92310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1746816936611194e-05,
+      "loss": 1.1119,
+      "step": 92315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1746072869244509e-05,
+      "loss": 1.5433,
+      "step": 92320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1745328791906809e-05,
+      "loss": 1.6517,
+      "step": 92325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1744584704602347e-05,
+      "loss": 0.899,
+      "step": 92330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1743840607335361e-05,
+      "loss": 1.2782,
+      "step": 92335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.174309650011011e-05,
+      "loss": 1.2526,
+      "step": 92340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1742352382930844e-05,
+      "loss": 1.1952,
+      "step": 92345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1741608255801803e-05,
+      "loss": 1.2497,
+      "step": 92350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1740864118727249e-05,
+      "loss": 1.3169,
+      "step": 92355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.174011997171142e-05,
+      "loss": 1.1375,
+      "step": 92360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.173937581475857e-05,
+      "loss": 1.1993,
+      "step": 92365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.173863164787295e-05,
+      "loss": 1.6117,
+      "step": 92370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1737887471058806e-05,
+      "loss": 1.192,
+      "step": 92375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1737143284320394e-05,
+      "loss": 1.0866,
+      "step": 92380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1736399087661956e-05,
+      "loss": 1.424,
+      "step": 92385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1735654881087745e-05,
+      "loss": 1.6275,
+      "step": 92390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1734910664602012e-05,
+      "loss": 1.0285,
+      "step": 92395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1734166438209008e-05,
+      "loss": 1.1834,
+      "step": 92400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1733422201912978e-05,
+      "loss": 1.1301,
+      "step": 92405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1732677955718176e-05,
+      "loss": 1.1778,
+      "step": 92410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1731933699628852e-05,
+      "loss": 1.3573,
+      "step": 92415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1731189433649253e-05,
+      "loss": 1.0721,
+      "step": 92420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1730445157783634e-05,
+      "loss": 1.2045,
+      "step": 92425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1729700872036239e-05,
+      "loss": 1.3493,
+      "step": 92430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1728956576411323e-05,
+      "loss": 1.2996,
+      "step": 92435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1728212270913136e-05,
+      "loss": 1.1808,
+      "step": 92440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1727467955545925e-05,
+      "loss": 0.9462,
+      "step": 92445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1726723630313947e-05,
+      "loss": 1.2626,
+      "step": 92450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1725979295221443e-05,
+      "loss": 1.3873,
+      "step": 92455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1725234950272678e-05,
+      "loss": 1.0989,
+      "step": 92460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1724490595471882e-05,
+      "loss": 1.2055,
+      "step": 92465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1723746230823326e-05,
+      "loss": 1.0212,
+      "step": 92470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1723001856331247e-05,
+      "loss": 1.219,
+      "step": 92475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1722257471999903e-05,
+      "loss": 1.2355,
+      "step": 92480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1721513077833539e-05,
+      "loss": 1.3922,
+      "step": 92485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1720768673836413e-05,
+      "loss": 1.6588,
+      "step": 92490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1720024260012772e-05,
+      "loss": 1.3115,
+      "step": 92495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1719279836366868e-05,
+      "loss": 1.0743,
+      "step": 92500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1718535402902948e-05,
+      "loss": 0.9842,
+      "step": 92505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.171779095962527e-05,
+      "loss": 1.2784,
+      "step": 92510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1717046506538082e-05,
+      "loss": 1.0771,
+      "step": 92515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1716302043645633e-05,
+      "loss": 1.2048,
+      "step": 92520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1715557570952174e-05,
+      "loss": 1.1452,
+      "step": 92525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1714813088461963e-05,
+      "loss": 1.3327,
+      "step": 92530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1714068596179245e-05,
+      "loss": 1.2112,
+      "step": 92535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1713324094108273e-05,
+      "loss": 1.4518,
+      "step": 92540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.17125795822533e-05,
+      "loss": 0.9678,
+      "step": 92545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1711835060618575e-05,
+      "loss": 1.1824,
+      "step": 92550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1711090529208353e-05,
+      "loss": 1.1147,
+      "step": 92555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.171034598802688e-05,
+      "loss": 1.0582,
+      "step": 92560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1709601437078418e-05,
+      "loss": 1.1372,
+      "step": 92565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1708856876367208e-05,
+      "loss": 1.3921,
+      "step": 92570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1708112305897506e-05,
+      "loss": 1.2334,
+      "step": 92575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1707367725673564e-05,
+      "loss": 1.2001,
+      "step": 92580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1706623135699635e-05,
+      "loss": 1.7107,
+      "step": 92585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1705878535979971e-05,
+      "loss": 1.1846,
+      "step": 92590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1705133926518822e-05,
+      "loss": 1.1114,
+      "step": 92595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1704389307320439e-05,
+      "loss": 1.1435,
+      "step": 92600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1703644678389078e-05,
+      "loss": 1.1331,
+      "step": 92605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.170290003972899e-05,
+      "loss": 1.1479,
+      "step": 92610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1702155391344427e-05,
+      "loss": 1.519,
+      "step": 92615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1701410733239643e-05,
+      "loss": 1.2623,
+      "step": 92620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1700666065418884e-05,
+      "loss": 1.8439,
+      "step": 92625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1699921387886407e-05,
+      "loss": 1.6662,
+      "step": 92630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1699176700646471e-05,
+      "loss": 1.2472,
+      "step": 92635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1698432003703315e-05,
+      "loss": 1.1411,
+      "step": 92640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1697687297061205e-05,
+      "loss": 1.5674,
+      "step": 92645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1696942580724384e-05,
+      "loss": 1.5643,
+      "step": 92650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1696197854697106e-05,
+      "loss": 1.1722,
+      "step": 92655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1695453118983631e-05,
+      "loss": 1.4966,
+      "step": 92660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1694708373588203e-05,
+      "loss": 1.3223,
+      "step": 92665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.169396361851508e-05,
+      "loss": 1.0126,
+      "step": 92670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1693218853768514e-05,
+      "loss": 0.8533,
+      "step": 92675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1692474079352758e-05,
+      "loss": 1.2931,
+      "step": 92680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1691729295272065e-05,
+      "loss": 1.4952,
+      "step": 92685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1690984501530685e-05,
+      "loss": 1.0359,
+      "step": 92690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1690239698132877e-05,
+      "loss": 1.1894,
+      "step": 92695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1689494885082888e-05,
+      "loss": 1.0402,
+      "step": 92700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1688750062384977e-05,
+      "loss": 1.2722,
+      "step": 92705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1688005230043395e-05,
+      "loss": 1.1741,
+      "step": 92710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1687260388062396e-05,
+      "loss": 1.3149,
+      "step": 92715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.168651553644623e-05,
+      "loss": 1.1001,
+      "step": 92720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1685770675199155e-05,
+      "loss": 1.1976,
+      "step": 92725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1685025804325423e-05,
+      "loss": 1.0071,
+      "step": 92730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1684280923829284e-05,
+      "loss": 0.9624,
+      "step": 92735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1683536033714999e-05,
+      "loss": 1.1657,
+      "step": 92740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1682791133986817e-05,
+      "loss": 1.1651,
+      "step": 92745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.168204622464899e-05,
+      "loss": 1.1457,
+      "step": 92750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1681301305705779e-05,
+      "loss": 1.1845,
+      "step": 92755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1680556377161428e-05,
+      "loss": 1.3295,
+      "step": 92760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1679811439020202e-05,
+      "loss": 1.3563,
+      "step": 92765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1679066491286341e-05,
+      "loss": 1.0441,
+      "step": 92770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1678321533964114e-05,
+      "loss": 1.3769,
+      "step": 92775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1677576567057766e-05,
+      "loss": 1.2397,
+      "step": 92780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1676831590571553e-05,
+      "loss": 1.3984,
+      "step": 92785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1676086604509735e-05,
+      "loss": 1.1666,
+      "step": 92790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1675341608876553e-05,
+      "loss": 1.288,
+      "step": 92795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1674596603676274e-05,
+      "loss": 1.2487,
+      "step": 92800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1673851588913148e-05,
+      "loss": 1.3221,
+      "step": 92805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1673106564591427e-05,
+      "loss": 1.0814,
+      "step": 92810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1672361530715368e-05,
+      "loss": 1.1753,
+      "step": 92815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1671616487289226e-05,
+      "loss": 1.2537,
+      "step": 92820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1670871434317254e-05,
+      "loss": 1.0192,
+      "step": 92825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.167012637180371e-05,
+      "loss": 1.1686,
+      "step": 92830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1669381299752844e-05,
+      "loss": 1.1489,
+      "step": 92835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.166863621816891e-05,
+      "loss": 1.2702,
+      "step": 92840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.166789112705617e-05,
+      "loss": 1.1334,
+      "step": 92845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1667146026418874e-05,
+      "loss": 1.0573,
+      "step": 92850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1666400916261276e-05,
+      "loss": 1.135,
+      "step": 92855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1665655796587637e-05,
+      "loss": 1.2092,
+      "step": 92860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1664910667402199e-05,
+      "loss": 1.3717,
+      "step": 92865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1664165528709233e-05,
+      "loss": 1.0493,
+      "step": 92870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1663420380512982e-05,
+      "loss": 1.3164,
+      "step": 92875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1662675222817708e-05,
+      "loss": 1.1968,
+      "step": 92880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1661930055627665e-05,
+      "loss": 1.2981,
+      "step": 92885
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1661184878947106e-05,
+      "loss": 1.1536,
+      "step": 92890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.166043969278029e-05,
+      "loss": 1.231,
+      "step": 92895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1659694497131468e-05,
+      "loss": 1.3267,
+      "step": 92900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.16589492920049e-05,
+      "loss": 1.1419,
+      "step": 92905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1658204077404838e-05,
+      "loss": 0.9078,
+      "step": 92910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.165745885333554e-05,
+      "loss": 1.3379,
+      "step": 92915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1656713619801259e-05,
+      "loss": 1.033,
+      "step": 92920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1655968376806253e-05,
+      "loss": 1.1581,
+      "step": 92925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1655223124354775e-05,
+      "loss": 1.1987,
+      "step": 92930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1654477862451084e-05,
+      "loss": 0.9713,
+      "step": 92935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1653732591099436e-05,
+      "loss": 1.2076,
+      "step": 92940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1652987310304082e-05,
+      "loss": 1.1443,
+      "step": 92945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1652242020069283e-05,
+      "loss": 1.2971,
+      "step": 92950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1651496720399295e-05,
+      "loss": 1.0699,
+      "step": 92955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.165075141129837e-05,
+      "loss": 0.963,
+      "step": 92960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1650006092770769e-05,
+      "loss": 1.2212,
+      "step": 92965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1649260764820745e-05,
+      "loss": 1.3049,
+      "step": 92970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1648515427452556e-05,
+      "loss": 1.185,
+      "step": 92975
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1647770080670452e-05,
+      "loss": 1.2077,
+      "step": 92980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1647024724478699e-05,
+      "loss": 1.1658,
+      "step": 92985
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1646279358881551e-05,
+      "loss": 1.274,
+      "step": 92990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1645533983883259e-05,
+      "loss": 1.5047,
+      "step": 92995
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1644788599488085e-05,
+      "loss": 1.0643,
+      "step": 93000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.164404320570028e-05,
+      "loss": 1.6092,
+      "step": 93005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1643297802524108e-05,
+      "loss": 1.1436,
+      "step": 93010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.164255238996382e-05,
+      "loss": 1.1883,
+      "step": 93015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1641806968023675e-05,
+      "loss": 1.2008,
+      "step": 93020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1641061536707927e-05,
+      "loss": 1.0592,
+      "step": 93025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1640316096020835e-05,
+      "loss": 1.3541,
+      "step": 93030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1639570645966657e-05,
+      "loss": 1.5407,
+      "step": 93035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.163882518654965e-05,
+      "loss": 1.19,
+      "step": 93040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1638079717774068e-05,
+      "loss": 0.9128,
+      "step": 93045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1637334239644173e-05,
+      "loss": 1.1309,
+      "step": 93050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1636588752164217e-05,
+      "loss": 1.1203,
+      "step": 93055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1635843255338457e-05,
+      "loss": 1.2187,
+      "step": 93060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1635097749171153e-05,
+      "loss": 1.1768,
+      "step": 93065
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1634352233666563e-05,
+      "loss": 1.0957,
+      "step": 93070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1633606708828939e-05,
+      "loss": 0.9662,
+      "step": 93075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.163286117466255e-05,
+      "loss": 1.1042,
+      "step": 93080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1632115631171636e-05,
+      "loss": 1.1435,
+      "step": 93085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.163137007836047e-05,
+      "loss": 1.2972,
+      "step": 93090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1630624516233302e-05,
+      "loss": 1.1533,
+      "step": 93095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1629878944794388e-05,
+      "loss": 1.0934,
+      "step": 93100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1629133364047993e-05,
+      "loss": 1.3489,
+      "step": 93105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1628387773998368e-05,
+      "loss": 1.2816,
+      "step": 93110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1627642174649774e-05,
+      "loss": 1.4217,
+      "step": 93115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1626896566006466e-05,
+      "loss": 1.6762,
+      "step": 93120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1626150948072706e-05,
+      "loss": 1.4102,
+      "step": 93125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1625405320852747e-05,
+      "loss": 1.125,
+      "step": 93130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1624659684350853e-05,
+      "loss": 1.1124,
+      "step": 93135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1623914038571274e-05,
+      "loss": 1.62,
+      "step": 93140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1623168383518273e-05,
+      "loss": 1.4425,
+      "step": 93145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.162242271919611e-05,
+      "loss": 1.2307,
+      "step": 93150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.162167704560904e-05,
+      "loss": 1.4305,
+      "step": 93155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.162093136276132e-05,
+      "loss": 1.2026,
+      "step": 93160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1620185670657213e-05,
+      "loss": 1.2647,
+      "step": 93165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.161943996930097e-05,
+      "loss": 0.9434,
+      "step": 93170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1618694258696859e-05,
+      "loss": 1.5114,
+      "step": 93175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.161794853884913e-05,
+      "loss": 1.3084,
+      "step": 93180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1617202809762044e-05,
+      "loss": 1.1831,
+      "step": 93185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1616457071439864e-05,
+      "loss": 1.191,
+      "step": 93190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.161571132388684e-05,
+      "loss": 1.2712,
+      "step": 93195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.161496556710724e-05,
+      "loss": 1.2659,
+      "step": 93200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1614219801105313e-05,
+      "loss": 1.3197,
+      "step": 93205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1613474025885326e-05,
+      "loss": 0.9148,
+      "step": 93210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1612728241451534e-05,
+      "loss": 1.1877,
+      "step": 93215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1611982447808197e-05,
+      "loss": 1.0182,
+      "step": 93220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1611236644959572e-05,
+      "loss": 1.0663,
+      "step": 93225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1610490832909919e-05,
+      "loss": 1.147,
+      "step": 93230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1609745011663498e-05,
+      "loss": 1.5085,
+      "step": 93235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1608999181224567e-05,
+      "loss": 1.2091,
+      "step": 93240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1608253341597385e-05,
+      "loss": 1.5056,
+      "step": 93245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1607507492786212e-05,
+      "loss": 1.179,
+      "step": 93250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1606761634795305e-05,
+      "loss": 1.0386,
+      "step": 93255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1606015767628926e-05,
+      "loss": 1.7775,
+      "step": 93260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1605269891291333e-05,
+      "loss": 1.1795,
+      "step": 93265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1604524005786786e-05,
+      "loss": 1.179,
+      "step": 93270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.160377811111954e-05,
+      "loss": 1.2382,
+      "step": 93275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1603032207293865e-05,
+      "loss": 1.2468,
+      "step": 93280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1602286294314009e-05,
+      "loss": 1.318,
+      "step": 93285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1601540372184239e-05,
+      "loss": 1.0865,
+      "step": 93290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1600794440908814e-05,
+      "loss": 1.0983,
+      "step": 93295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1600048500491987e-05,
+      "loss": 1.3074,
+      "step": 93300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1599302550938028e-05,
+      "loss": 1.0636,
+      "step": 93305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1598556592251184e-05,
+      "loss": 1.0907,
+      "step": 93310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1597810624435726e-05,
+      "loss": 1.3966,
+      "step": 93315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.159706464749591e-05,
+      "loss": 1.0926,
+      "step": 93320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1596318661435998e-05,
+      "loss": 1.4884,
+      "step": 93325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1595572666260244e-05,
+      "loss": 1.1749,
+      "step": 93330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1594826661972913e-05,
+      "loss": 1.1817,
+      "step": 93335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1594080648578266e-05,
+      "loss": 1.1727,
+      "step": 93340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1593334626080558e-05,
+      "loss": 1.2115,
+      "step": 93345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1592588594484055e-05,
+      "loss": 1.2576,
+      "step": 93350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1591842553793014e-05,
+      "loss": 1.138,
+      "step": 93355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1591096504011694e-05,
+      "loss": 1.0698,
+      "step": 93360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.159035044514436e-05,
+      "loss": 1.2068,
+      "step": 93365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1589604377195267e-05,
+      "loss": 0.853,
+      "step": 93370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.158885830016868e-05,
+      "loss": 1.1241,
+      "step": 93375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1588112214068854e-05,
+      "loss": 1.1942,
+      "step": 93380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1587366118900058e-05,
+      "loss": 1.5494,
+      "step": 93385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1586620014666545e-05,
+      "loss": 1.2213,
+      "step": 93390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1585873901372577e-05,
+      "loss": 1.2046,
+      "step": 93395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.158512777902242e-05,
+      "loss": 1.345,
+      "step": 93400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1584381647620326e-05,
+      "loss": 1.2667,
+      "step": 93405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1583635507170567e-05,
+      "loss": 1.4373,
+      "step": 93410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.158288935767739e-05,
+      "loss": 1.3385,
+      "step": 93415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1582143199145069e-05,
+      "loss": 0.9737,
+      "step": 93420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1581397031577857e-05,
+      "loss": 1.181,
+      "step": 93425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1580650854980018e-05,
+      "loss": 1.1602,
+      "step": 93430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1579904669355813e-05,
+      "loss": 1.1978,
+      "step": 93435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1579158474709503e-05,
+      "loss": 1.5073,
+      "step": 93440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1578412271045348e-05,
+      "loss": 1.3857,
+      "step": 93445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1577666058367609e-05,
+      "loss": 1.1476,
+      "step": 93450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1576919836680549e-05,
+      "loss": 0.9235,
+      "step": 93455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1576173605988428e-05,
+      "loss": 1.2391,
+      "step": 93460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1575427366295508e-05,
+      "loss": 1.33,
+      "step": 93465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1574681117606053e-05,
+      "loss": 0.9809,
+      "step": 93470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1573934859924317e-05,
+      "loss": 1.2185,
+      "step": 93475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1573188593254572e-05,
+      "loss": 1.6375,
+      "step": 93480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.157244231760107e-05,
+      "loss": 1.305,
+      "step": 93485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1571696032968078e-05,
+      "loss": 1.4353,
+      "step": 93490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1570949739359856e-05,
+      "loss": 1.4013,
+      "step": 93495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1570203436780665e-05,
+      "loss": 1.3715,
+      "step": 93500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.156945712523477e-05,
+      "loss": 1.1358,
+      "step": 93505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1568710804726429e-05,
+      "loss": 1.1888,
+      "step": 93510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1567964475259907e-05,
+      "loss": 1.103,
+      "step": 93515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1567218136839463e-05,
+      "loss": 1.1753,
+      "step": 93520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.156647178946936e-05,
+      "loss": 1.0674,
+      "step": 93525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1565725433153866e-05,
+      "loss": 1.0582,
+      "step": 93530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.156497906789723e-05,
+      "loss": 1.2108,
+      "step": 93535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1564232693703725e-05,
+      "loss": 1.214,
+      "step": 93540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1563486310577612e-05,
+      "loss": 1.3162,
+      "step": 93545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.156273991852315e-05,
+      "loss": 1.5087,
+      "step": 93550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1561993517544601e-05,
+      "loss": 1.4295,
+      "step": 93555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.156124710764623e-05,
+      "loss": 1.1082,
+      "step": 93560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.15605006888323e-05,
+      "loss": 1.4598,
+      "step": 93565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1559754261107067e-05,
+      "loss": 1.3926,
+      "step": 93570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1559007824474805e-05,
+      "loss": 1.0481,
+      "step": 93575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1558261378939761e-05,
+      "loss": 1.3676,
+      "step": 93580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1557514924506215e-05,
+      "loss": 1.0604,
+      "step": 93585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1556768461178418e-05,
+      "loss": 1.5417,
+      "step": 93590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1556021988960636e-05,
+      "loss": 1.3974,
+      "step": 93595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1555275507857129e-05,
+      "loss": 1.4054,
+      "step": 93600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1554529017872163e-05,
+      "loss": 1.0166,
+      "step": 93605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1553782519010004e-05,
+      "loss": 1.3084,
+      "step": 93610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1553036011274906e-05,
+      "loss": 1.2584,
+      "step": 93615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.155228949467114e-05,
+      "loss": 1.2171,
+      "step": 93620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1551542969202965e-05,
+      "loss": 1.1388,
+      "step": 93625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1550796434874645e-05,
+      "loss": 1.018,
+      "step": 93630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1550049891690447e-05,
+      "loss": 1.0214,
+      "step": 93635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1549303339654624e-05,
+      "loss": 1.303,
+      "step": 93640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1548556778771451e-05,
+      "loss": 1.2893,
+      "step": 93645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1547810209045186e-05,
+      "loss": 1.5098,
+      "step": 93650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.154706363048009e-05,
+      "loss": 1.3418,
+      "step": 93655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1546317043080429e-05,
+      "loss": 1.1325,
+      "step": 93660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1545570446850464e-05,
+      "loss": 1.232,
+      "step": 93665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1544823841794463e-05,
+      "loss": 0.9318,
+      "step": 93670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1544077227916685e-05,
+      "loss": 1.2422,
+      "step": 93675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1543330605221398e-05,
+      "loss": 1.5214,
+      "step": 93680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1542583973712859e-05,
+      "loss": 1.2398,
+      "step": 93685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.154183733339534e-05,
+      "loss": 1.3319,
+      "step": 93690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1541090684273101e-05,
+      "loss": 1.5197,
+      "step": 93695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1540344026350403e-05,
+      "loss": 1.0516,
+      "step": 93700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1539597359631513e-05,
+      "loss": 1.3245,
+      "step": 93705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1538850684120692e-05,
+      "loss": 1.3796,
+      "step": 93710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.153810399982221e-05,
+      "loss": 1.0582,
+      "step": 93715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1537357306740322e-05,
+      "loss": 1.3828,
+      "step": 93720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1536610604879299e-05,
+      "loss": 0.9915,
+      "step": 93725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1535863894243407e-05,
+      "loss": 1.1275,
+      "step": 93730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1535117174836901e-05,
+      "loss": 1.0042,
+      "step": 93735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1534370446664053e-05,
+      "loss": 1.1631,
+      "step": 93740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.153362370972912e-05,
+      "loss": 1.3536,
+      "step": 93745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1532876964036375e-05,
+      "loss": 1.3153,
+      "step": 93750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1532130209590078e-05,
+      "loss": 1.1898,
+      "step": 93755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1531383446394493e-05,
+      "loss": 1.0142,
+      "step": 93760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1530636674453883e-05,
+      "loss": 1.1671,
+      "step": 93765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1529889893772516e-05,
+      "loss": 1.0361,
+      "step": 93770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1529143104354654e-05,
+      "loss": 1.3448,
+      "step": 93775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1528396306204564e-05,
+      "loss": 1.3327,
+      "step": 93780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1527649499326507e-05,
+      "loss": 1.2617,
+      "step": 93785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1526902683724753e-05,
+      "loss": 1.1513,
+      "step": 93790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.152615585940356e-05,
+      "loss": 1.0805,
+      "step": 93795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1525409026367198e-05,
+      "loss": 1.1327,
+      "step": 93800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.152466218461993e-05,
+      "loss": 1.1278,
+      "step": 93805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.152391533416602e-05,
+      "loss": 1.4449,
+      "step": 93810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1523168475009732e-05,
+      "loss": 1.4654,
+      "step": 93815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1522421607155337e-05,
+      "loss": 1.1985,
+      "step": 93820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.152167473060709e-05,
+      "loss": 1.1016,
+      "step": 93825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1520927845369267e-05,
+      "loss": 0.8919,
+      "step": 93830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.152018095144613e-05,
+      "loss": 1.4169,
+      "step": 93835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1519434048841936e-05,
+      "loss": 1.3563,
+      "step": 93840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.151868713756096e-05,
+      "loss": 1.5251,
+      "step": 93845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1517940217607461e-05,
+      "loss": 1.181,
+      "step": 93850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1517193288985708e-05,
+      "loss": 1.063,
+      "step": 93855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1516446351699967e-05,
+      "loss": 1.0612,
+      "step": 93860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1515699405754499e-05,
+      "loss": 1.3226,
+      "step": 93865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1514952451153576e-05,
+      "loss": 1.1545,
+      "step": 93870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1514205487901458e-05,
+      "loss": 1.1286,
+      "step": 93875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.151345851600241e-05,
+      "loss": 0.9849,
+      "step": 93880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1512711535460701e-05,
+      "loss": 1.3248,
+      "step": 93885
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1511964546280597e-05,
+      "loss": 1.0533,
+      "step": 93890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.151121754846636e-05,
+      "loss": 1.1855,
+      "step": 93895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.151047054202226e-05,
+      "loss": 1.2109,
+      "step": 93900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1509723526952561e-05,
+      "loss": 1.0949,
+      "step": 93905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1508976503261524e-05,
+      "loss": 0.9734,
+      "step": 93910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1508229470953428e-05,
+      "loss": 0.987,
+      "step": 93915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1507482430032522e-05,
+      "loss": 1.1213,
+      "step": 93920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1506735380503087e-05,
+      "loss": 1.3315,
+      "step": 93925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1505988322369376e-05,
+      "loss": 1.0743,
+      "step": 93930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1505241255635667e-05,
+      "loss": 0.929,
+      "step": 93935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1504494180306221e-05,
+      "loss": 1.1394,
+      "step": 93940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.15037470963853e-05,
+      "loss": 1.043,
+      "step": 93945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1503000003877178e-05,
+      "loss": 1.2383,
+      "step": 93950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1502252902786115e-05,
+      "loss": 1.1913,
+      "step": 93955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1501505793116383e-05,
+      "loss": 1.2658,
+      "step": 93960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1500758674872244e-05,
+      "loss": 1.0668,
+      "step": 93965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1500011548057962e-05,
+      "loss": 1.1341,
+      "step": 93970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1499264412677813e-05,
+      "loss": 1.4284,
+      "step": 93975
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1498517268736056e-05,
+      "loss": 1.3527,
+      "step": 93980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1497770116236957e-05,
+      "loss": 1.1441,
+      "step": 93985
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1497022955184786e-05,
+      "loss": 1.1984,
+      "step": 93990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.149627578558381e-05,
+      "loss": 0.9944,
+      "step": 93995
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1495528607438296e-05,
+      "loss": 1.2527,
+      "step": 94000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1494781420752507e-05,
+      "loss": 1.0592,
+      "step": 94005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1494034225530713e-05,
+      "loss": 1.3857,
+      "step": 94010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1493287021777178e-05,
+      "loss": 1.2522,
+      "step": 94015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1492539809496174e-05,
+      "loss": 1.088,
+      "step": 94020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1491792588691961e-05,
+      "loss": 1.5133,
+      "step": 94025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1491045359368816e-05,
+      "loss": 0.9361,
+      "step": 94030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1490298121530997e-05,
+      "loss": 1.1277,
+      "step": 94035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1489550875182768e-05,
+      "loss": 1.13,
+      "step": 94040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1488803620328413e-05,
+      "loss": 1.4789,
+      "step": 94045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1488056356972179e-05,
+      "loss": 1.3611,
+      "step": 94050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1487309085118348e-05,
+      "loss": 1.3498,
+      "step": 94055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1486561804771181e-05,
+      "loss": 1.3258,
+      "step": 94060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1485814515934947e-05,
+      "loss": 1.1586,
+      "step": 94065
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1485067218613914e-05,
+      "loss": 1.036,
+      "step": 94070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1484319912812347e-05,
+      "loss": 1.2593,
+      "step": 94075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1483572598534515e-05,
+      "loss": 1.2917,
+      "step": 94080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1482825275784686e-05,
+      "loss": 1.0831,
+      "step": 94085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1482077944567128e-05,
+      "loss": 1.2925,
+      "step": 94090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1481330604886108e-05,
+      "loss": 1.2178,
+      "step": 94095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.148058325674589e-05,
+      "loss": 1.6387,
+      "step": 94100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1479835900150748e-05,
+      "loss": 1.4647,
+      "step": 94105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1479088535104947e-05,
+      "loss": 1.2966,
+      "step": 94110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1478341161612753e-05,
+      "loss": 1.1945,
+      "step": 94115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1477593779678438e-05,
+      "loss": 0.9843,
+      "step": 94120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1476846389306269e-05,
+      "loss": 1.2888,
+      "step": 94125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1476098990500508e-05,
+      "loss": 1.5289,
+      "step": 94130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1475351583265434e-05,
+      "loss": 1.4864,
+      "step": 94135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1474604167605306e-05,
+      "loss": 1.1025,
+      "step": 94140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1473856743524391e-05,
+      "loss": 1.4026,
+      "step": 94145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.147310931102697e-05,
+      "loss": 1.3088,
+      "step": 94150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1472361870117294e-05,
+      "loss": 1.1969,
+      "step": 94155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1471614420799645e-05,
+      "loss": 1.2203,
+      "step": 94160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1470866963078285e-05,
+      "loss": 1.1825,
+      "step": 94165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1470119496957482e-05,
+      "loss": 1.191,
+      "step": 94170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.146937202244151e-05,
+      "loss": 1.3664,
+      "step": 94175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1468624539534627e-05,
+      "loss": 1.3243,
+      "step": 94180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1467877048241111e-05,
+      "loss": 1.0114,
+      "step": 94185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.146712954856523e-05,
+      "loss": 1.2539,
+      "step": 94190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1466382040511248e-05,
+      "loss": 1.4579,
+      "step": 94195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1465634524083436e-05,
+      "loss": 0.7751,
+      "step": 94200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1464886999286062e-05,
+      "loss": 1.0784,
+      "step": 94205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1464139466123396e-05,
+      "loss": 1.351,
+      "step": 94210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1463391924599708e-05,
+      "loss": 0.909,
+      "step": 94215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1462644374719263e-05,
+      "loss": 1.1473,
+      "step": 94220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1461896816486329e-05,
+      "loss": 1.0482,
+      "step": 94225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1461149249905184e-05,
+      "loss": 1.0325,
+      "step": 94230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1460401674980089e-05,
+      "loss": 1.1547,
+      "step": 94235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.145965409171531e-05,
+      "loss": 1.1315,
+      "step": 94240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1458906500115128e-05,
+      "loss": 0.998,
+      "step": 94245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.14581589001838e-05,
+      "loss": 1.2471,
+      "step": 94250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1457411291925604e-05,
+      "loss": 0.9835,
+      "step": 94255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1456663675344802e-05,
+      "loss": 1.0981,
+      "step": 94260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1455916050445671e-05,
+      "loss": 1.2061,
+      "step": 94265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1455168417232474e-05,
+      "loss": 1.2167,
+      "step": 94270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1454420775709483e-05,
+      "loss": 1.3487,
+      "step": 94275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.145367312588097e-05,
+      "loss": 1.204,
+      "step": 94280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1452925467751194e-05,
+      "loss": 1.3195,
+      "step": 94285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1452177801324438e-05,
+      "loss": 1.233,
+      "step": 94290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1451430126604965e-05,
+      "loss": 0.9859,
+      "step": 94295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1450682443597046e-05,
+      "loss": 1.3111,
+      "step": 94300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1449934752304948e-05,
+      "loss": 1.2604,
+      "step": 94305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1449187052732944e-05,
+      "loss": 1.1065,
+      "step": 94310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1448439344885303e-05,
+      "loss": 1.073,
+      "step": 94315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1447691628766293e-05,
+      "loss": 1.219,
+      "step": 94320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1446943904380186e-05,
+      "loss": 1.2405,
+      "step": 94325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1446196171731247e-05,
+      "loss": 1.1745,
+      "step": 94330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1445448430823758e-05,
+      "loss": 1.1974,
+      "step": 94335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1444700681661976e-05,
+      "loss": 1.2576,
+      "step": 94340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1443952924250175e-05,
+      "loss": 1.1788,
+      "step": 94345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.144320515859263e-05,
+      "loss": 1.2408,
+      "step": 94350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1442457384693601e-05,
+      "loss": 1.1803,
+      "step": 94355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1441709602557372e-05,
+      "loss": 1.2635,
+      "step": 94360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.14409618121882e-05,
+      "loss": 1.5572,
+      "step": 94365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1440214013590365e-05,
+      "loss": 1.1944,
+      "step": 94370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1439466206768133e-05,
+      "loss": 1.2141,
+      "step": 94375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1438718391725772e-05,
+      "loss": 1.55,
+      "step": 94380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1437970568467557e-05,
+      "loss": 1.2455,
+      "step": 94385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1437222736997758e-05,
+      "loss": 1.303,
+      "step": 94390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1436474897320641e-05,
+      "loss": 1.5758,
+      "step": 94395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1435727049440482e-05,
+      "loss": 1.387,
+      "step": 94400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1434979193361548e-05,
+      "loss": 1.2796,
+      "step": 94405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1434231329088112e-05,
+      "loss": 1.1808,
+      "step": 94410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1433483456624444e-05,
+      "loss": 1.3476,
+      "step": 94415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1432735575974814e-05,
+      "loss": 1.1615,
+      "step": 94420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1431987687143492e-05,
+      "loss": 1.2028,
+      "step": 94425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1431239790134752e-05,
+      "loss": 1.2814,
+      "step": 94430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1430491884952861e-05,
+      "loss": 1.2785,
+      "step": 94435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1429743971602094e-05,
+      "loss": 1.219,
+      "step": 94440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1428996050086721e-05,
+      "loss": 1.3505,
+      "step": 94445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1428248120411007e-05,
+      "loss": 1.6126,
+      "step": 94450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1427500182579233e-05,
+      "loss": 1.2322,
+      "step": 94455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1426752236595661e-05,
+      "loss": 1.1687,
+      "step": 94460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1426004282464571e-05,
+      "loss": 1.2163,
+      "step": 94465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1425256320190226e-05,
+      "loss": 1.2909,
+      "step": 94470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1424508349776901e-05,
+      "loss": 1.2444,
+      "step": 94475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1423760371228872e-05,
+      "loss": 1.0779,
+      "step": 94480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1423012384550398e-05,
+      "loss": 1.246,
+      "step": 94485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.142226438974576e-05,
+      "loss": 1.1768,
+      "step": 94490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1421516386819231e-05,
+      "loss": 1.3798,
+      "step": 94495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1420768375775078e-05,
+      "loss": 1.3654,
+      "step": 94500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1420020356617571e-05,
+      "loss": 1.5817,
+      "step": 94505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1419272329350985e-05,
+      "loss": 1.3474,
+      "step": 94510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1418524293979591e-05,
+      "loss": 1.2972,
+      "step": 94515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.141777625050766e-05,
+      "loss": 1.3304,
+      "step": 94520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1417028198939463e-05,
+      "loss": 1.6551,
+      "step": 94525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1416280139279272e-05,
+      "loss": 1.1872,
+      "step": 94530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1415532071531365e-05,
+      "loss": 1.0872,
+      "step": 94535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1414783995700006e-05,
+      "loss": 1.3495,
+      "step": 94540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1414035911789467e-05,
+      "loss": 1.3079,
+      "step": 94545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1413287819804027e-05,
+      "loss": 1.3118,
+      "step": 94550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1412539719747949e-05,
+      "loss": 1.1521,
+      "step": 94555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1411791611625512e-05,
+      "loss": 1.2987,
+      "step": 94560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1411043495440981e-05,
+      "loss": 1.7654,
+      "step": 94565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1410295371198638e-05,
+      "loss": 0.9745,
+      "step": 94570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.140954723890275e-05,
+      "loss": 1.4582,
+      "step": 94575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1408799098557583e-05,
+      "loss": 1.1436,
+      "step": 94580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1408050950167425e-05,
+      "loss": 1.2766,
+      "step": 94585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.140730279373653e-05,
+      "loss": 1.2431,
+      "step": 94590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1406554629269182e-05,
+      "loss": 1.3317,
+      "step": 94595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1405806456769652e-05,
+      "loss": 1.3012,
+      "step": 94600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1405058276242208e-05,
+      "loss": 1.1124,
+      "step": 94605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.140431008769113e-05,
+      "loss": 1.3665,
+      "step": 94610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1403561891120682e-05,
+      "loss": 1.2589,
+      "step": 94615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1402813686535143e-05,
+      "loss": 1.6099,
+      "step": 94620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1402065473938784e-05,
+      "loss": 1.3071,
+      "step": 94625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1401317253335876e-05,
+      "loss": 1.2083,
+      "step": 94630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1400569024730693e-05,
+      "loss": 1.0839,
+      "step": 94635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1399820788127507e-05,
+      "loss": 1.3141,
+      "step": 94640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1399072543530592e-05,
+      "loss": 1.2928,
+      "step": 94645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1398324290944218e-05,
+      "loss": 1.2586,
+      "step": 94650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1397576030372664e-05,
+      "loss": 1.4576,
+      "step": 94655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1396827761820196e-05,
+      "loss": 1.3771,
+      "step": 94660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1396079485291091e-05,
+      "loss": 1.2776,
+      "step": 94665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1395331200789622e-05,
+      "loss": 1.1201,
+      "step": 94670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.139458290832006e-05,
+      "loss": 1.2166,
+      "step": 94675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1393834607886682e-05,
+      "loss": 1.3649,
+      "step": 94680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1393086299493756e-05,
+      "loss": 1.3189,
+      "step": 94685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.139233798314556e-05,
+      "loss": 1.1379,
+      "step": 94690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1391589658846363e-05,
+      "loss": 1.283,
+      "step": 94695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1390841326600444e-05,
+      "loss": 1.1554,
+      "step": 94700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1390092986412069e-05,
+      "loss": 1.1615,
+      "step": 94705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1389344638285518e-05,
+      "loss": 1.1586,
+      "step": 94710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1388596282225061e-05,
+      "loss": 1.3265,
+      "step": 94715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1387847918234974e-05,
+      "loss": 1.3314,
+      "step": 94720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1387099546319526e-05,
+      "loss": 1.1815,
+      "step": 94725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1386351166482997e-05,
+      "loss": 1.0608,
+      "step": 94730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1385602778729654e-05,
+      "loss": 1.2964,
+      "step": 94735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1384854383063775e-05,
+      "loss": 1.4215,
+      "step": 94740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1384105979489632e-05,
+      "loss": 1.1654,
+      "step": 94745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1383357568011501e-05,
+      "loss": 1.1822,
+      "step": 94750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1382609148633655e-05,
+      "loss": 1.1443,
+      "step": 94755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1381860721360365e-05,
+      "loss": 1.0676,
+      "step": 94760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1381112286195905e-05,
+      "loss": 1.1507,
+      "step": 94765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1380363843144555e-05,
+      "loss": 1.4481,
+      "step": 94770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1379615392210583e-05,
+      "loss": 1.2508,
+      "step": 94775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1378866933398265e-05,
+      "loss": 1.2997,
+      "step": 94780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1378118466711879e-05,
+      "loss": 1.0776,
+      "step": 94785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1377369992155689e-05,
+      "loss": 1.57,
+      "step": 94790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1376621509733983e-05,
+      "loss": 1.7182,
+      "step": 94795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.137587301945102e-05,
+      "loss": 1.5555,
+      "step": 94800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1375124521311086e-05,
+      "loss": 1.1695,
+      "step": 94805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1374376015318452e-05,
+      "loss": 1.2764,
+      "step": 94810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1373627501477389e-05,
+      "loss": 0.9732,
+      "step": 94815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1372878979792177e-05,
+      "loss": 1.2445,
+      "step": 94820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1372130450267086e-05,
+      "loss": 1.3064,
+      "step": 94825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1371381912906392e-05,
+      "loss": 1.0454,
+      "step": 94830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.137063336771437e-05,
+      "loss": 1.2614,
+      "step": 94835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1369884814695295e-05,
+      "loss": 1.0672,
+      "step": 94840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1369136253853441e-05,
+      "loss": 1.2058,
+      "step": 94845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1368387685193083e-05,
+      "loss": 1.3988,
+      "step": 94850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1367639108718494e-05,
+      "loss": 1.1205,
+      "step": 94855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1366890524433951e-05,
+      "loss": 1.2522,
+      "step": 94860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1366141932343726e-05,
+      "loss": 1.0594,
+      "step": 94865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1365393332452094e-05,
+      "loss": 1.602,
+      "step": 94870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1364644724763338e-05,
+      "loss": 1.3682,
+      "step": 94875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1363896109281725e-05,
+      "loss": 1.4446,
+      "step": 94880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1363147486011528e-05,
+      "loss": 1.2834,
+      "step": 94885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1362398854957028e-05,
+      "loss": 1.1926,
+      "step": 94890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1361650216122498e-05,
+      "loss": 1.192,
+      "step": 94895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1360901569512212e-05,
+      "loss": 1.6458,
+      "step": 94900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1360152915130446e-05,
+      "loss": 1.478,
+      "step": 94905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1359404252981476e-05,
+      "loss": 1.27,
+      "step": 94910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1358655583069575e-05,
+      "loss": 1.7824,
+      "step": 94915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1357906905399023e-05,
+      "loss": 1.4746,
+      "step": 94920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1357158219974091e-05,
+      "loss": 1.3284,
+      "step": 94925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1356409526799055e-05,
+      "loss": 1.1378,
+      "step": 94930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1355660825878192e-05,
+      "loss": 1.0222,
+      "step": 94935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1354912117215775e-05,
+      "loss": 1.2609,
+      "step": 94940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1354163400816081e-05,
+      "loss": 1.4444,
+      "step": 94945
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1353414676683387e-05,
+      "loss": 1.2635,
+      "step": 94950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1352665944821966e-05,
+      "loss": 1.6417,
+      "step": 94955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1351917205236095e-05,
+      "loss": 1.5748,
+      "step": 94960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1351168457930048e-05,
+      "loss": 1.1896,
+      "step": 94965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1350419702908109e-05,
+      "loss": 1.2698,
+      "step": 94970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.134967094017454e-05,
+      "loss": 1.2335,
+      "step": 94975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.134892216973363e-05,
+      "loss": 1.2349,
+      "step": 94980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1348173391589646e-05,
+      "loss": 1.119,
+      "step": 94985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1347424605746864e-05,
+      "loss": 1.212,
+      "step": 94990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1346675812209567e-05,
+      "loss": 1.0679,
+      "step": 94995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1345927010982022e-05,
+      "loss": 1.1601,
+      "step": 95000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1345178202068516e-05,
+      "loss": 1.064,
+      "step": 95005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1344429385473314e-05,
+      "loss": 1.3001,
+      "step": 95010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1343680561200698e-05,
+      "loss": 1.1966,
+      "step": 95015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1342931729254946e-05,
+      "loss": 1.2452,
+      "step": 95020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.134218288964033e-05,
+      "loss": 1.2184,
+      "step": 95025
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1341434042361127e-05,
+      "loss": 1.3963,
+      "step": 95030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1340685187421613e-05,
+      "loss": 1.2857,
+      "step": 95035
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1339936324826068e-05,
+      "loss": 1.1112,
+      "step": 95040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1339187454578767e-05,
+      "loss": 1.3757,
+      "step": 95045
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1338438576683983e-05,
+      "loss": 1.088,
+      "step": 95050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1337689691145994e-05,
+      "loss": 1.4059,
+      "step": 95055
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.133694079796908e-05,
+      "loss": 1.0465,
+      "step": 95060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1336191897157513e-05,
+      "loss": 1.009,
+      "step": 95065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.133544298871557e-05,
+      "loss": 1.217,
+      "step": 95070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1334694072647534e-05,
+      "loss": 1.2898,
+      "step": 95075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1333945148957674e-05,
+      "loss": 1.4238,
+      "step": 95080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.133319621765027e-05,
+      "loss": 0.9921,
+      "step": 95085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1332447278729598e-05,
+      "loss": 1.2992,
+      "step": 95090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1331698332199937e-05,
+      "loss": 1.3219,
+      "step": 95095
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1330949378065562e-05,
+      "loss": 1.1844,
+      "step": 95100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1330200416330748e-05,
+      "loss": 1.1427,
+      "step": 95105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1329451446999776e-05,
+      "loss": 1.3017,
+      "step": 95110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1328702470076922e-05,
+      "loss": 1.2296,
+      "step": 95115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.132795348556646e-05,
+      "loss": 1.2942,
+      "step": 95120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1327204493472674e-05,
+      "loss": 1.4322,
+      "step": 95125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1326455493799831e-05,
+      "loss": 1.747,
+      "step": 95130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1325706486552215e-05,
+      "loss": 1.1657,
+      "step": 95135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1324957471734104e-05,
+      "loss": 1.2685,
+      "step": 95140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.132420844934977e-05,
+      "loss": 1.5051,
+      "step": 95145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1323459419403497e-05,
+      "loss": 1.6937,
+      "step": 95150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1322710381899556e-05,
+      "loss": 1.1924,
+      "step": 95155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1321961336842229e-05,
+      "loss": 1.4198,
+      "step": 95160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1321212284235792e-05,
+      "loss": 1.1239,
+      "step": 95165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1320463224084521e-05,
+      "loss": 1.2352,
+      "step": 95170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1319714156392693e-05,
+      "loss": 1.139,
+      "step": 95175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1318965081164593e-05,
+      "loss": 1.1673,
+      "step": 95180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.131821599840449e-05,
+      "loss": 1.1237,
+      "step": 95185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1317466908116664e-05,
+      "loss": 1.4344,
+      "step": 95190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1316717810305393e-05,
+      "loss": 1.1297,
+      "step": 95195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1315968704974955e-05,
+      "loss": 1.4223,
+      "step": 95200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1315219592129632e-05,
+      "loss": 1.1199,
+      "step": 95205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.131447047177369e-05,
+      "loss": 1.6322,
+      "step": 95210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1313721343911418e-05,
+      "loss": 1.366,
+      "step": 95215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1312972208547094e-05,
+      "loss": 1.2904,
+      "step": 95220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1312223065684989e-05,
+      "loss": 1.4992,
+      "step": 95225
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.131147391532939e-05,
+      "loss": 1.2876,
+      "step": 95230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1310724757484562e-05,
+      "loss": 1.3229,
+      "step": 95235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1309975592154794e-05,
+      "loss": 1.0491,
+      "step": 95240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.130922641934436e-05,
+      "loss": 1.1418,
+      "step": 95245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.130847723905754e-05,
+      "loss": 1.3562,
+      "step": 95250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1307728051298609e-05,
+      "loss": 1.3305,
+      "step": 95255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1306978856071851e-05,
+      "loss": 1.3304,
+      "step": 95260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.130622965338154e-05,
+      "loss": 1.3145,
+      "step": 95265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1305480443231954e-05,
+      "loss": 1.1302,
+      "step": 95270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1304731225627372e-05,
+      "loss": 1.084,
+      "step": 95275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1303982000572074e-05,
+      "loss": 1.1462,
+      "step": 95280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1303232768070338e-05,
+      "loss": 1.0702,
+      "step": 95285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.130248352812644e-05,
+      "loss": 0.9275,
+      "step": 95290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1301734280744658e-05,
+      "loss": 1.0664,
+      "step": 95295
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.130098502592928e-05,
+      "loss": 1.4311,
+      "step": 95300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1300235763684573e-05,
+      "loss": 1.228,
+      "step": 95305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1299486494014823e-05,
+      "loss": 1.0654,
+      "step": 95310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1298737216924303e-05,
+      "loss": 1.4151,
+      "step": 95315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1297987932417297e-05,
+      "loss": 1.0873,
+      "step": 95320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1297238640498085e-05,
+      "loss": 1.1513,
+      "step": 95325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1296489341170937e-05,
+      "loss": 1.1131,
+      "step": 95330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1295740034440143e-05,
+      "loss": 1.2046,
+      "step": 95335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1294990720309971e-05,
+      "loss": 1.224,
+      "step": 95340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1294241398784709e-05,
+      "loss": 1.2474,
+      "step": 95345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.129349206986863e-05,
+      "loss": 1.3441,
+      "step": 95350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1292742733566017e-05,
+      "loss": 1.4407,
+      "step": 95355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1291993389881148e-05,
+      "loss": 1.3615,
+      "step": 95360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.12912440388183e-05,
+      "loss": 1.2194,
+      "step": 95365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1290494680381756e-05,
+      "loss": 1.0705,
+      "step": 95370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1289745314575793e-05,
+      "loss": 1.2852,
+      "step": 95375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1288995941404692e-05,
+      "loss": 1.0636,
+      "step": 95380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.128824656087273e-05,
+      "loss": 1.4552,
+      "step": 95385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1287497172984186e-05,
+      "loss": 1.6562,
+      "step": 95390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1286747777743342e-05,
+      "loss": 0.9424,
+      "step": 95395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1285998375154471e-05,
+      "loss": 1.5722,
+      "step": 95400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1285248965221865e-05,
+      "loss": 1.5508,
+      "step": 95405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.128449954794979e-05,
+      "loss": 1.3644,
+      "step": 95410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.128375012334254e-05,
+      "loss": 1.0735,
+      "step": 95415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.128300069140438e-05,
+      "loss": 1.2174,
+      "step": 95420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1282251252139594e-05,
+      "loss": 1.2365,
+      "step": 95425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1281501805552468e-05,
+      "loss": 1.3918,
+      "step": 95430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1280752351647276e-05,
+      "loss": 1.4511,
+      "step": 95435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.12800028904283e-05,
+      "loss": 1.3414,
+      "step": 95440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1279253421899817e-05,
+      "loss": 1.0647,
+      "step": 95445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.127850394606611e-05,
+      "loss": 1.3691,
+      "step": 95450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1277754462931458e-05,
+      "loss": 1.5159,
+      "step": 95455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1277004972500141e-05,
+      "loss": 1.7523,
+      "step": 95460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1276255474776439e-05,
+      "loss": 1.2499,
+      "step": 95465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1275505969764633e-05,
+      "loss": 1.0732,
+      "step": 95470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1274756457469e-05,
+      "loss": 1.2786,
+      "step": 95475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1274006937893821e-05,
+      "loss": 1.3497,
+      "step": 95480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.127325741104338e-05,
+      "loss": 1.3103,
+      "step": 95485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1272507876921952e-05,
+      "loss": 1.1824,
+      "step": 95490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1271758335533823e-05,
+      "loss": 1.4844,
+      "step": 95495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1271008786883267e-05,
+      "loss": 1.4443,
+      "step": 95500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1270259230974566e-05,
+      "loss": 1.0259,
+      "step": 95505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1269509667812005e-05,
+      "loss": 1.1986,
+      "step": 95510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1268760097399856e-05,
+      "loss": 1.1313,
+      "step": 95515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.126801051974241e-05,
+      "loss": 1.3244,
+      "step": 95520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.126726093484394e-05,
+      "loss": 1.1309,
+      "step": 95525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1266511342708725e-05,
+      "loss": 1.1237,
+      "step": 95530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1265761743341056e-05,
+      "loss": 1.2788,
+      "step": 95535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.12650121367452e-05,
+      "loss": 1.1551,
+      "step": 95540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1264262522925447e-05,
+      "loss": 1.2057,
+      "step": 95545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1263512901886075e-05,
+      "loss": 1.0959,
+      "step": 95550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1262763273631365e-05,
+      "loss": 1.3768,
+      "step": 95555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1262013638165597e-05,
+      "loss": 1.2628,
+      "step": 95560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1261263995493053e-05,
+      "loss": 0.9309,
+      "step": 95565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1260514345618013e-05,
+      "loss": 0.9961,
+      "step": 95570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1259764688544757e-05,
+      "loss": 1.3441,
+      "step": 95575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1259015024277568e-05,
+      "loss": 1.5748,
+      "step": 95580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1258265352820725e-05,
+      "loss": 1.4699,
+      "step": 95585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1257515674178511e-05,
+      "loss": 1.4749,
+      "step": 95590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1256765988355207e-05,
+      "loss": 1.1708,
+      "step": 95595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1256016295355094e-05,
+      "loss": 1.0719,
+      "step": 95600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.125526659518245e-05,
+      "loss": 1.6739,
+      "step": 95605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1254516887841557e-05,
+      "loss": 1.3056,
+      "step": 95610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1253767173336704e-05,
+      "loss": 1.5974,
+      "step": 95615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1253017451672159e-05,
+      "loss": 1.2192,
+      "step": 95620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1252267722852217e-05,
+      "loss": 0.9677,
+      "step": 95625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.125151798688115e-05,
+      "loss": 1.2595,
+      "step": 95630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.125076824376324e-05,
+      "loss": 1.626,
+      "step": 95635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1250018493502774e-05,
+      "loss": 1.0055,
+      "step": 95640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1249268736104027e-05,
+      "loss": 1.2579,
+      "step": 95645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1248518971571285e-05,
+      "loss": 1.0449,
+      "step": 95650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.124776919990883e-05,
+      "loss": 1.0374,
+      "step": 95655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1247019421120937e-05,
+      "loss": 1.0233,
+      "step": 95660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1246269635211899e-05,
+      "loss": 1.1975,
+      "step": 95665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1245519842185985e-05,
+      "loss": 1.6094,
+      "step": 95670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1244770042047485e-05,
+      "loss": 1.2736,
+      "step": 95675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.124402023480068e-05,
+      "loss": 1.5693,
+      "step": 95680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1243270420449848e-05,
+      "loss": 1.3647,
+      "step": 95685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1242520598999275e-05,
+      "loss": 1.1533,
+      "step": 95690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.124177077045324e-05,
+      "loss": 1.3303,
+      "step": 95695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1241020934816028e-05,
+      "loss": 1.7145,
+      "step": 95700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1240271092091916e-05,
+      "loss": 1.1133,
+      "step": 95705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1239521242285191e-05,
+      "loss": 1.1169,
+      "step": 95710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.123877138540013e-05,
+      "loss": 1.2887,
+      "step": 95715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1238021521441023e-05,
+      "loss": 1.7547,
+      "step": 95720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1237271650412145e-05,
+      "loss": 1.5134,
+      "step": 95725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.123652177231778e-05,
+      "loss": 1.1488,
+      "step": 95730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1235771887162213e-05,
+      "loss": 1.4244,
+      "step": 95735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1235021994949717e-05,
+      "loss": 1.3123,
+      "step": 95740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.123427209568459e-05,
+      "loss": 1.1238,
+      "step": 95745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1233522189371097e-05,
+      "loss": 1.373,
+      "step": 95750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1232772276013535e-05,
+      "loss": 1.1535,
+      "step": 95755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1232022355616178e-05,
+      "loss": 1.3742,
+      "step": 95760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.123127242818331e-05,
+      "loss": 1.5445,
+      "step": 95765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1230522493719219e-05,
+      "loss": 1.5443,
+      "step": 95770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1229772552228177e-05,
+      "loss": 1.2246,
+      "step": 95775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1229022603714475e-05,
+      "loss": 1.4867,
+      "step": 95780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1228272648182392e-05,
+      "loss": 1.3637,
+      "step": 95785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1227522685636212e-05,
+      "loss": 1.2838,
+      "step": 95790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1226772716080216e-05,
+      "loss": 1.3344,
+      "step": 95795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.122602273951869e-05,
+      "loss": 1.3574,
+      "step": 95800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1225272755955913e-05,
+      "loss": 1.3352,
+      "step": 95805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.122452276539617e-05,
+      "loss": 1.1855,
+      "step": 95810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1223772767843743e-05,
+      "loss": 1.4471,
+      "step": 95815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1223022763302915e-05,
+      "loss": 1.1577,
+      "step": 95820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1222272751777971e-05,
+      "loss": 1.4725,
+      "step": 95825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1221522733273193e-05,
+      "loss": 1.2697,
+      "step": 95830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.122077270779286e-05,
+      "loss": 1.1361,
+      "step": 95835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1220022675341262e-05,
+      "loss": 1.2191,
+      "step": 95840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1219272635922673e-05,
+      "loss": 1.5504,
+      "step": 95845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1218522589541387e-05,
+      "loss": 1.6006,
+      "step": 95850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1217772536201676e-05,
+      "loss": 1.1995,
+      "step": 95855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1217022475907833e-05,
+      "loss": 1.1013,
+      "step": 95860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.121627240866414e-05,
+      "loss": 0.973,
+      "step": 95865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.121552233447487e-05,
+      "loss": 1.3558,
+      "step": 95870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1214772253344317e-05,
+      "loss": 1.2686,
+      "step": 95875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1214022165276762e-05,
+      "loss": 1.4648,
+      "step": 95880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1213272070276486e-05,
+      "loss": 1.1361,
+      "step": 95885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1212521968347776e-05,
+      "loss": 1.2233,
+      "step": 95890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1211771859494912e-05,
+      "loss": 1.1609,
+      "step": 95895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1211021743722179e-05,
+      "loss": 1.0699,
+      "step": 95900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1210271621033861e-05,
+      "loss": 1.2259,
+      "step": 95905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1209521491434241e-05,
+      "loss": 1.3533,
+      "step": 95910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.12087713549276e-05,
+      "loss": 1.6804,
+      "step": 95915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1208021211518228e-05,
+      "loss": 1.3352,
+      "step": 95920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1207271061210405e-05,
+      "loss": 1.2577,
+      "step": 95925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1206520904008413e-05,
+      "loss": 1.033,
+      "step": 95930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1205770739916539e-05,
+      "loss": 1.1111,
+      "step": 95935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1205020568939063e-05,
+      "loss": 1.2219,
+      "step": 95940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1204270391080274e-05,
+      "loss": 0.9567,
+      "step": 95945
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1203520206344451e-05,
+      "loss": 1.2549,
+      "step": 95950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1202770014735887e-05,
+      "loss": 1.2957,
+      "step": 95955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1202019816258852e-05,
+      "loss": 1.1146,
+      "step": 95960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1201269610917639e-05,
+      "loss": 1.2831,
+      "step": 95965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1200519398716534e-05,
+      "loss": 1.1154,
+      "step": 95970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1199769179659812e-05,
+      "loss": 1.181,
+      "step": 95975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1199018953751765e-05,
+      "loss": 1.2181,
+      "step": 95980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1198268720996674e-05,
+      "loss": 1.2686,
+      "step": 95985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1197518481398828e-05,
+      "loss": 1.2647,
+      "step": 95990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1196768234962503e-05,
+      "loss": 1.1134,
+      "step": 95995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1196017981691991e-05,
+      "loss": 1.1836,
+      "step": 96000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1195267721591572e-05,
+      "loss": 1.3807,
+      "step": 96005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.119451745466553e-05,
+      "loss": 1.1688,
+      "step": 96010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1193767180918151e-05,
+      "loss": 1.4375,
+      "step": 96015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1193016900353719e-05,
+      "loss": 1.5788,
+      "step": 96020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.119226661297652e-05,
+      "loss": 1.1389,
+      "step": 96025
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1191516318790841e-05,
+      "loss": 1.3416,
+      "step": 96030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1190766017800958e-05,
+      "loss": 1.1943,
+      "step": 96035
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1190015710011162e-05,
+      "loss": 1.1544,
+      "step": 96040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1189265395425735e-05,
+      "loss": 1.3216,
+      "step": 96045
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1188515074048964e-05,
+      "loss": 1.5862,
+      "step": 96050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1187764745885133e-05,
+      "loss": 1.6648,
+      "step": 96055
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1187014410938529e-05,
+      "loss": 1.2306,
+      "step": 96060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.118626406921343e-05,
+      "loss": 1.6761,
+      "step": 96065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1185513720714125e-05,
+      "loss": 1.3223,
+      "step": 96070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1184763365444902e-05,
+      "loss": 1.4248,
+      "step": 96075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1184013003410038e-05,
+      "loss": 1.3278,
+      "step": 96080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1183262634613828e-05,
+      "loss": 1.2305,
+      "step": 96085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1182512259060552e-05,
+      "loss": 1.3052,
+      "step": 96090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1181761876754493e-05,
+      "loss": 1.3532,
+      "step": 96095
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1181011487699938e-05,
+      "loss": 1.206,
+      "step": 96100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1180261091901173e-05,
+      "loss": 1.1097,
+      "step": 96105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1179510689362483e-05,
+      "loss": 1.9124,
+      "step": 96110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.117876028008815e-05,
+      "loss": 1.6469,
+      "step": 96115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1178009864082463e-05,
+      "loss": 1.2398,
+      "step": 96120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1177259441349709e-05,
+      "loss": 1.0695,
+      "step": 96125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1176509011894167e-05,
+      "loss": 1.4045,
+      "step": 96130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1175758575720126e-05,
+      "loss": 1.517,
+      "step": 96135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1175008132831872e-05,
+      "loss": 1.0687,
+      "step": 96140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1174257683233689e-05,
+      "loss": 1.119,
+      "step": 96145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1173507226929862e-05,
+      "loss": 1.1399,
+      "step": 96150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1172756763924682e-05,
+      "loss": 1.374,
+      "step": 96155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1172006294222424e-05,
+      "loss": 1.205,
+      "step": 96160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1171255817827383e-05,
+      "loss": 1.2056,
+      "step": 96165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1170505334743844e-05,
+      "loss": 1.3196,
+      "step": 96170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1169754844976084e-05,
+      "loss": 1.1729,
+      "step": 96175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.11690043485284e-05,
+      "loss": 1.1995,
+      "step": 96180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1168253845405068e-05,
+      "loss": 1.2658,
+      "step": 96185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.116750333561038e-05,
+      "loss": 1.2887,
+      "step": 96190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.116675281914862e-05,
+      "loss": 1.4591,
+      "step": 96195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1166002296024076e-05,
+      "loss": 1.5264,
+      "step": 96200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.116525176624103e-05,
+      "loss": 1.3415,
+      "step": 96205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.116450122980377e-05,
+      "loss": 1.6151,
+      "step": 96210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1163750686716581e-05,
+      "loss": 1.083,
+      "step": 96215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.116300013698375e-05,
+      "loss": 1.245,
+      "step": 96220
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1162249580609562e-05,
+      "loss": 1.3176,
+      "step": 96225
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1161499017598306e-05,
+      "loss": 1.22,
+      "step": 96230
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1160748447954265e-05,
+      "loss": 0.9814,
+      "step": 96235
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1159997871681725e-05,
+      "loss": 1.1057,
+      "step": 96240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1159247288784974e-05,
+      "loss": 1.3164,
+      "step": 96245
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1158496699268298e-05,
+      "loss": 1.1921,
+      "step": 96250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1157746103135978e-05,
+      "loss": 1.0629,
+      "step": 96255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1156995500392311e-05,
+      "loss": 1.1758,
+      "step": 96260
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1156244891041574e-05,
+      "loss": 1.0086,
+      "step": 96265
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1155494275088058e-05,
+      "loss": 1.2322,
+      "step": 96270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.115474365253605e-05,
+      "loss": 1.0115,
+      "step": 96275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.115399302338983e-05,
+      "loss": 1.0733,
+      "step": 96280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1153242387653695e-05,
+      "loss": 1.2003,
+      "step": 96285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1152491745331921e-05,
+      "loss": 1.2538,
+      "step": 96290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1151741096428803e-05,
+      "loss": 1.2766,
+      "step": 96295
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1150990440948622e-05,
+      "loss": 1.4621,
+      "step": 96300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1150239778895666e-05,
+      "loss": 1.2855,
+      "step": 96305
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1149489110274224e-05,
+      "loss": 1.319,
+      "step": 96310
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1148738435088578e-05,
+      "loss": 1.2652,
+      "step": 96315
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.114798775334302e-05,
+      "loss": 1.3575,
+      "step": 96320
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1147237065041834e-05,
+      "loss": 1.1222,
+      "step": 96325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1146486370189306e-05,
+      "loss": 1.2665,
+      "step": 96330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1145735668789725e-05,
+      "loss": 1.0745,
+      "step": 96335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1144984960847377e-05,
+      "loss": 1.1291,
+      "step": 96340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1144234246366552e-05,
+      "loss": 1.877,
+      "step": 96345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1143483525351528e-05,
+      "loss": 1.2041,
+      "step": 96350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1142732797806603e-05,
+      "loss": 1.4304,
+      "step": 96355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1141982063736056e-05,
+      "loss": 1.2006,
+      "step": 96360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1141231323144181e-05,
+      "loss": 1.4021,
+      "step": 96365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.114048057603526e-05,
+      "loss": 1.2423,
+      "step": 96370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.113972982241358e-05,
+      "loss": 1.5397,
+      "step": 96375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.113897906228343e-05,
+      "loss": 1.024,
+      "step": 96380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1138228295649096e-05,
+      "loss": 1.1351,
+      "step": 96385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.113747752251487e-05,
+      "loss": 0.9785,
+      "step": 96390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1136726742885031e-05,
+      "loss": 1.8031,
+      "step": 96395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1135975956763873e-05,
+      "loss": 1.0626,
+      "step": 96400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1135225164155681e-05,
+      "loss": 1.2006,
+      "step": 96405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1134474365064743e-05,
+      "loss": 1.8106,
+      "step": 96410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1133723559495346e-05,
+      "loss": 1.3885,
+      "step": 96415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1132972747451778e-05,
+      "loss": 1.2232,
+      "step": 96420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1132221928938327e-05,
+      "loss": 1.1844,
+      "step": 96425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.113147110395928e-05,
+      "loss": 1.2858,
+      "step": 96430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1130720272518922e-05,
+      "loss": 1.3512,
+      "step": 96435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1129969434621545e-05,
+      "loss": 1.2787,
+      "step": 96440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1129218590271433e-05,
+      "loss": 1.2952,
+      "step": 96445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1128467739472878e-05,
+      "loss": 1.2776,
+      "step": 96450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1127716882230162e-05,
+      "loss": 1.2904,
+      "step": 96455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1126966018547581e-05,
+      "loss": 1.3137,
+      "step": 96460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1126215148429412e-05,
+      "loss": 1.2098,
+      "step": 96465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1125464271879953e-05,
+      "loss": 1.1253,
+      "step": 96470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1124713388903488e-05,
+      "loss": 1.0838,
+      "step": 96475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1123962499504302e-05,
+      "loss": 1.087,
+      "step": 96480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1123211603686687e-05,
+      "loss": 1.1223,
+      "step": 96485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1122460701454927e-05,
+      "loss": 1.2871,
+      "step": 96490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1121709792813315e-05,
+      "loss": 1.196,
+      "step": 96495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1120958877766138e-05,
+      "loss": 1.0836,
+      "step": 96500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1120207956317682e-05,
+      "loss": 1.2012,
+      "step": 96505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1119457028472237e-05,
+      "loss": 1.1296,
+      "step": 96510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1118706094234086e-05,
+      "loss": 1.3481,
+      "step": 96515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1117955153607525e-05,
+      "loss": 1.251,
+      "step": 96520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1117204206596838e-05,
+      "loss": 1.0121,
+      "step": 96525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1116453253206314e-05,
+      "loss": 1.2227,
+      "step": 96530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1115702293440243e-05,
+      "loss": 1.3183,
+      "step": 96535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.111495132730291e-05,
+      "loss": 1.2528,
+      "step": 96540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1114200354798606e-05,
+      "loss": 1.3297,
+      "step": 96545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1113449375931618e-05,
+      "loss": 1.2351,
+      "step": 96550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1112698390706236e-05,
+      "loss": 1.2992,
+      "step": 96555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1111947399126746e-05,
+      "loss": 0.9862,
+      "step": 96560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1111196401197442e-05,
+      "loss": 1.1223,
+      "step": 96565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1110445396922606e-05,
+      "loss": 1.3343,
+      "step": 96570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1109694386306531e-05,
+      "loss": 0.97,
+      "step": 96575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1108943369353501e-05,
+      "loss": 1.1822,
+      "step": 96580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1108192346067809e-05,
+      "loss": 1.2161,
+      "step": 96585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1107441316453747e-05,
+      "loss": 1.0786,
+      "step": 96590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1106690280515595e-05,
+      "loss": 1.0943,
+      "step": 96595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1105939238257645e-05,
+      "loss": 1.2894,
+      "step": 96600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1105188189684191e-05,
+      "loss": 0.9412,
+      "step": 96605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1104437134799518e-05,
+      "loss": 1.5945,
+      "step": 96610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1103686073607915e-05,
+      "loss": 0.9985,
+      "step": 96615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.110293500611367e-05,
+      "loss": 0.9076,
+      "step": 96620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1102183932321071e-05,
+      "loss": 0.858,
+      "step": 96625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.110143285223441e-05,
+      "loss": 1.342,
+      "step": 96630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1100681765857975e-05,
+      "loss": 1.0627,
+      "step": 96635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1099930673196056e-05,
+      "loss": 1.1212,
+      "step": 96640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.109917957425294e-05,
+      "loss": 1.1619,
+      "step": 96645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1098428469032918e-05,
+      "loss": 1.1289,
+      "step": 96650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.109767735754028e-05,
+      "loss": 1.2708,
+      "step": 96655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1096926239779313e-05,
+      "loss": 1.124,
+      "step": 96660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1096175115754303e-05,
+      "loss": 1.1601,
+      "step": 96665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.109542398546955e-05,
+      "loss": 1.3967,
+      "step": 96670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1094672848929336e-05,
+      "loss": 1.2513,
+      "step": 96675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1093921706137949e-05,
+      "loss": 1.379,
+      "step": 96680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1093170557099682e-05,
+      "loss": 1.2043,
+      "step": 96685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.109241940181882e-05,
+      "loss": 0.9223,
+      "step": 96690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.109166824029966e-05,
+      "loss": 1.2744,
+      "step": 96695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1090917072546484e-05,
+      "loss": 1.5333,
+      "step": 96700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1090165898563586e-05,
+      "loss": 1.2161,
+      "step": 96705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1089414718355255e-05,
+      "loss": 1.1192,
+      "step": 96710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.108866353192578e-05,
+      "loss": 1.4691,
+      "step": 96715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1087912339279449e-05,
+      "loss": 1.1794,
+      "step": 96720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1087161140420556e-05,
+      "loss": 1.0843,
+      "step": 96725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1086409935353383e-05,
+      "loss": 1.4907,
+      "step": 96730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1085658724082232e-05,
+      "loss": 0.9522,
+      "step": 96735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.108490750661138e-05,
+      "loss": 1.4687,
+      "step": 96740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1084156282945126e-05,
+      "loss": 1.2002,
+      "step": 96745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1083405053087754e-05,
+      "loss": 1.2025,
+      "step": 96750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.108265381704356e-05,
+      "loss": 1.2354,
+      "step": 96755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1081902574816828e-05,
+      "loss": 1.2942,
+      "step": 96760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.108115132641185e-05,
+      "loss": 1.1277,
+      "step": 96765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1080400071832917e-05,
+      "loss": 1.1396,
+      "step": 96770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.107964881108432e-05,
+      "loss": 1.0632,
+      "step": 96775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1078897544170345e-05,
+      "loss": 1.3274,
+      "step": 96780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1078146271095284e-05,
+      "loss": 1.6789,
+      "step": 96785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1077394991863433e-05,
+      "loss": 1.3172,
+      "step": 96790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.107664370647907e-05,
+      "loss": 1.2595,
+      "step": 96795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.10758924149465e-05,
+      "loss": 0.8788,
+      "step": 96800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1075141117270001e-05,
+      "loss": 1.4883,
+      "step": 96805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1074389813453868e-05,
+      "loss": 1.263,
+      "step": 96810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1073638503502395e-05,
+      "loss": 1.396,
+      "step": 96815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1072887187419863e-05,
+      "loss": 1.2099,
+      "step": 96820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1072135865210573e-05,
+      "loss": 1.5819,
+      "step": 96825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1071384536878806e-05,
+      "loss": 0.9725,
+      "step": 96830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1070633202428862e-05,
+      "loss": 1.2257,
+      "step": 96835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1069881861865025e-05,
+      "loss": 1.0185,
+      "step": 96840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1069130515191584e-05,
+      "loss": 0.9509,
+      "step": 96845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1068379162412836e-05,
+      "loss": 1.1355,
+      "step": 96850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1067627803533068e-05,
+      "loss": 1.1702,
+      "step": 96855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.106687643855657e-05,
+      "loss": 1.2439,
+      "step": 96860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1066125067487634e-05,
+      "loss": 1.4373,
+      "step": 96865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1065373690330551e-05,
+      "loss": 1.0664,
+      "step": 96870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1064622307089609e-05,
+      "loss": 1.1835,
+      "step": 96875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1063870917769105e-05,
+      "loss": 1.0211,
+      "step": 96880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1063119522373324e-05,
+      "loss": 1.4645,
+      "step": 96885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1062368120906555e-05,
+      "loss": 1.2173,
+      "step": 96890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.10616167133731e-05,
+      "loss": 1.1797,
+      "step": 96895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1060865299777236e-05,
+      "loss": 1.2492,
+      "step": 96900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1060113880123265e-05,
+      "loss": 1.2752,
+      "step": 96905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1059362454415472e-05,
+      "loss": 1.1893,
+      "step": 96910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1058611022658148e-05,
+      "loss": 1.2326,
+      "step": 96915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1057859584855588e-05,
+      "loss": 1.2061,
+      "step": 96920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1057108141012078e-05,
+      "loss": 1.7347,
+      "step": 96925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1056356691131913e-05,
+      "loss": 1.0549,
+      "step": 96930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1055605235219384e-05,
+      "loss": 1.0957,
+      "step": 96935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1054853773278781e-05,
+      "loss": 1.3028,
+      "step": 96940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1054102305314397e-05,
+      "loss": 1.4084,
+      "step": 96945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.105335083133052e-05,
+      "loss": 0.9383,
+      "step": 96950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1052599351331445e-05,
+      "loss": 1.2041,
+      "step": 96955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1051847865321461e-05,
+      "loss": 1.546,
+      "step": 96960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.105109637330486e-05,
+      "loss": 1.1906,
+      "step": 96965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1050344875285931e-05,
+      "loss": 1.2494,
+      "step": 96970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.104959337126897e-05,
+      "loss": 1.1672,
+      "step": 96975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1048841861258266e-05,
+      "loss": 1.2747,
+      "step": 96980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1048090345258111e-05,
+      "loss": 1.1151,
+      "step": 96985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1047338823272798e-05,
+      "loss": 1.4096,
+      "step": 96990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1046587295306612e-05,
+      "loss": 1.093,
+      "step": 96995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1045835761363855e-05,
+      "loss": 1.2072,
+      "step": 97000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.104508422144881e-05,
+      "loss": 1.1773,
+      "step": 97005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1044332675565773e-05,
+      "loss": 1.2844,
+      "step": 97010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1043581123719037e-05,
+      "loss": 1.5294,
+      "step": 97015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1042829565912886e-05,
+      "loss": 1.4195,
+      "step": 97020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1042078002151623e-05,
+      "loss": 1.1434,
+      "step": 97025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.104132643243953e-05,
+      "loss": 1.2138,
+      "step": 97030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1040574856780905e-05,
+      "loss": 1.5064,
+      "step": 97035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1039823275180037e-05,
+      "loss": 1.287,
+      "step": 97040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1039071687641221e-05,
+      "loss": 1.0581,
+      "step": 97045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1038320094168744e-05,
+      "loss": 1.679,
+      "step": 97050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1037568494766903e-05,
+      "loss": 1.0792,
+      "step": 97055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1036816889439987e-05,
+      "loss": 1.4678,
+      "step": 97060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.103606527819229e-05,
+      "loss": 1.483,
+      "step": 97065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.10353136610281e-05,
+      "loss": 1.3017,
+      "step": 97070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1034562037951715e-05,
+      "loss": 1.2659,
+      "step": 97075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1033810408967422e-05,
+      "loss": 1.1075,
+      "step": 97080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1033058774079519e-05,
+      "loss": 1.1202,
+      "step": 97085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1032307133292293e-05,
+      "loss": 1.0348,
+      "step": 97090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1031555486610037e-05,
+      "loss": 1.4538,
+      "step": 97095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1030803834037042e-05,
+      "loss": 1.1355,
+      "step": 97100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1030052175577607e-05,
+      "loss": 1.2294,
+      "step": 97105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1029300511236018e-05,
+      "loss": 1.3278,
+      "step": 97110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1028548841016573e-05,
+      "loss": 1.5668,
+      "step": 97115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1027797164923557e-05,
+      "loss": 1.1156,
+      "step": 97120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1027045482961267e-05,
+      "loss": 1.3376,
+      "step": 97125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1026293795133997e-05,
+      "loss": 1.2215,
+      "step": 97130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1025542101446034e-05,
+      "loss": 1.1605,
+      "step": 97135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1024790401901674e-05,
+      "loss": 1.3479,
+      "step": 97140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1024038696505213e-05,
+      "loss": 0.9769,
+      "step": 97145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.102328698526094e-05,
+      "loss": 1.2537,
+      "step": 97150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.102253526817315e-05,
+      "loss": 1.1175,
+      "step": 97155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1021783545246125e-05,
+      "loss": 0.9719,
+      "step": 97160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1021031816484174e-05,
+      "loss": 1.1982,
+      "step": 97165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.102028008189158e-05,
+      "loss": 1.2174,
+      "step": 97170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1019528341472635e-05,
+      "loss": 1.2211,
+      "step": 97175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1018776595231639e-05,
+      "loss": 1.2892,
+      "step": 97180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.101802484317288e-05,
+      "loss": 1.6108,
+      "step": 97185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1017273085300651e-05,
+      "loss": 1.1703,
+      "step": 97190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1016521321619245e-05,
+      "loss": 1.3592,
+      "step": 97195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1015769552132956e-05,
+      "loss": 1.3792,
+      "step": 97200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1015017776846075e-05,
+      "loss": 1.0034,
+      "step": 97205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.10142659957629e-05,
+      "loss": 1.6728,
+      "step": 97210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1013514208887719e-05,
+      "loss": 1.6778,
+      "step": 97215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1012762416224824e-05,
+      "loss": 0.9872,
+      "step": 97220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1012010617778516e-05,
+      "loss": 1.2741,
+      "step": 97225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1011258813553078e-05,
+      "loss": 1.1831,
+      "step": 97230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1010507003552813e-05,
+      "loss": 1.5476,
+      "step": 97235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1009755187782005e-05,
+      "loss": 1.0681,
+      "step": 97240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1009003366244955e-05,
+      "loss": 1.4554,
+      "step": 97245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1008251538945951e-05,
+      "loss": 1.0542,
+      "step": 97250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.100749970588929e-05,
+      "loss": 1.2083,
+      "step": 97255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1006747867079265e-05,
+      "loss": 0.9601,
+      "step": 97260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1005996022520165e-05,
+      "loss": 1.4536,
+      "step": 97265
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1005244172216287e-05,
+      "loss": 1.0751,
+      "step": 97270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1004492316171926e-05,
+      "loss": 1.0316,
+      "step": 97275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1003740454391374e-05,
+      "loss": 1.7887,
+      "step": 97280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1002988586878922e-05,
+      "loss": 1.3473,
+      "step": 97285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1002236713638868e-05,
+      "loss": 1.3995,
+      "step": 97290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1001484834675501e-05,
+      "loss": 1.4991,
+      "step": 97295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1000732949993118e-05,
+      "loss": 1.1545,
+      "step": 97300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0999981059596013e-05,
+      "loss": 1.4235,
+      "step": 97305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0999229163488474e-05,
+      "loss": 1.1706,
+      "step": 97310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0998477261674805e-05,
+      "loss": 1.2671,
+      "step": 97315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.099772535415929e-05,
+      "loss": 1.2511,
+      "step": 97320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0996973440946224e-05,
+      "loss": 1.2405,
+      "step": 97325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.099622152203991e-05,
+      "loss": 1.4376,
+      "step": 97330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.099546959744463e-05,
+      "loss": 1.2407,
+      "step": 97335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0994717667164688e-05,
+      "loss": 1.1086,
+      "step": 97340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0993965731204367e-05,
+      "loss": 1.2035,
+      "step": 97345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0993213789567971e-05,
+      "loss": 1.4933,
+      "step": 97350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0992461842259791e-05,
+      "loss": 1.4516,
+      "step": 97355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0991709889284116e-05,
+      "loss": 1.0519,
+      "step": 97360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0990957930645248e-05,
+      "loss": 1.2177,
+      "step": 97365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0990205966347476e-05,
+      "loss": 1.0754,
+      "step": 97370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0989453996395093e-05,
+      "loss": 1.3875,
+      "step": 97375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0988702020792398e-05,
+      "loss": 1.1436,
+      "step": 97380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0987950039543684e-05,
+      "loss": 1.1235,
+      "step": 97385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0987198052653243e-05,
+      "loss": 1.1512,
+      "step": 97390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0986446060125368e-05,
+      "loss": 1.2182,
+      "step": 97395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0985694061964359e-05,
+      "loss": 1.3052,
+      "step": 97400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0984942058174504e-05,
+      "loss": 1.0941,
+      "step": 97405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0984190048760099e-05,
+      "loss": 1.1014,
+      "step": 97410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0983438033725441e-05,
+      "loss": 1.1526,
+      "step": 97415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0982686013074823e-05,
+      "loss": 1.2277,
+      "step": 97420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0981933986812541e-05,
+      "loss": 1.117,
+      "step": 97425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0981181954942884e-05,
+      "loss": 1.2364,
+      "step": 97430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0980429917470154e-05,
+      "loss": 1.1299,
+      "step": 97435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.097967787439864e-05,
+      "loss": 1.2283,
+      "step": 97440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0978925825732641e-05,
+      "loss": 1.2353,
+      "step": 97445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0978173771476443e-05,
+      "loss": 1.0468,
+      "step": 97450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0977421711634351e-05,
+      "loss": 1.269,
+      "step": 97455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0976669646210658e-05,
+      "loss": 1.3373,
+      "step": 97460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0975917575209648e-05,
+      "loss": 1.1237,
+      "step": 97465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0975165498635632e-05,
+      "loss": 1.0661,
+      "step": 97470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0974413416492893e-05,
+      "loss": 1.0776,
+      "step": 97475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0973661328785728e-05,
+      "loss": 1.3647,
+      "step": 97480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0972909235518433e-05,
+      "loss": 1.3319,
+      "step": 97485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0972157136695307e-05,
+      "loss": 1.2623,
+      "step": 97490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0971405032320636e-05,
+      "loss": 1.0815,
+      "step": 97495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0970652922398724e-05,
+      "loss": 1.2135,
+      "step": 97500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.096990080693386e-05,
+      "loss": 1.705,
+      "step": 97505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0969148685930338e-05,
+      "loss": 1.2768,
+      "step": 97510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.096839655939246e-05,
+      "loss": 1.1645,
+      "step": 97515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0967644427324515e-05,
+      "loss": 0.8853,
+      "step": 97520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.09668922897308e-05,
+      "loss": 1.2177,
+      "step": 97525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0966140146615612e-05,
+      "loss": 1.14,
+      "step": 97530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0965387997983239e-05,
+      "loss": 1.1295,
+      "step": 97535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0964635843837987e-05,
+      "loss": 1.3991,
+      "step": 97540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.096388368418414e-05,
+      "loss": 1.1056,
+      "step": 97545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0963131519026003e-05,
+      "loss": 1.0723,
+      "step": 97550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0962379348367866e-05,
+      "loss": 1.1717,
+      "step": 97555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0961627172214025e-05,
+      "loss": 1.1706,
+      "step": 97560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0960874990568778e-05,
+      "loss": 1.2757,
+      "step": 97565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0960122803436413e-05,
+      "loss": 1.252,
+      "step": 97570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0959370610821232e-05,
+      "loss": 1.7019,
+      "step": 97575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.095861841272753e-05,
+      "loss": 0.8844,
+      "step": 97580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0957866209159602e-05,
+      "loss": 1.4541,
+      "step": 97585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.095711400012174e-05,
+      "loss": 1.0134,
+      "step": 97590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0956361785618246e-05,
+      "loss": 1.2069,
+      "step": 97595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0955609565653408e-05,
+      "loss": 1.1688,
+      "step": 97600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0954857340231529e-05,
+      "loss": 1.2914,
+      "step": 97605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0954105109356898e-05,
+      "loss": 1.2545,
+      "step": 97610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0953352873033816e-05,
+      "loss": 1.4127,
+      "step": 97615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0952600631266575e-05,
+      "loss": 1.3582,
+      "step": 97620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0951848384059474e-05,
+      "loss": 1.2169,
+      "step": 97625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0951096131416806e-05,
+      "loss": 1.4309,
+      "step": 97630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0950343873342866e-05,
+      "loss": 1.2791,
+      "step": 97635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0949591609841948e-05,
+      "loss": 1.0682,
+      "step": 97640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0948839340918358e-05,
+      "loss": 1.2256,
+      "step": 97645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.094808706657638e-05,
+      "loss": 1.421,
+      "step": 97650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0947334786820317e-05,
+      "loss": 1.1165,
+      "step": 97655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0946582501654465e-05,
+      "loss": 0.9824,
+      "step": 97660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0945830211083114e-05,
+      "loss": 1.3936,
+      "step": 97665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0945077915110568e-05,
+      "loss": 1.0938,
+      "step": 97670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0944325613741113e-05,
+      "loss": 1.1397,
+      "step": 97675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0943573306979056e-05,
+      "loss": 0.947,
+      "step": 97680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0942820994828685e-05,
+      "loss": 1.0023,
+      "step": 97685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.09420686772943e-05,
+      "loss": 1.5429,
+      "step": 97690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0941316354380196e-05,
+      "loss": 1.3554,
+      "step": 97695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.094056402609067e-05,
+      "loss": 1.5096,
+      "step": 97700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0939811692430019e-05,
+      "loss": 1.1108,
+      "step": 97705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0939059353402535e-05,
+      "loss": 1.1136,
+      "step": 97710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0938307009012516e-05,
+      "loss": 1.0725,
+      "step": 97715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0937554659264262e-05,
+      "loss": 1.2904,
+      "step": 97720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0936802304162066e-05,
+      "loss": 1.3828,
+      "step": 97725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0936049943710226e-05,
+      "loss": 1.2049,
+      "step": 97730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0935297577913035e-05,
+      "loss": 1.5298,
+      "step": 97735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0934545206774793e-05,
+      "loss": 1.3917,
+      "step": 97740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0933792830299794e-05,
+      "loss": 1.4858,
+      "step": 97745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0933040448492337e-05,
+      "loss": 1.3746,
+      "step": 97750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.093228806135672e-05,
+      "loss": 1.1449,
+      "step": 97755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.093153566889723e-05,
+      "loss": 1.273,
+      "step": 97760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0930783271118178e-05,
+      "loss": 0.9599,
+      "step": 97765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0930030868023846e-05,
+      "loss": 1.2496,
+      "step": 97770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0929278459618541e-05,
+      "loss": 1.2195,
+      "step": 97775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0928526045906556e-05,
+      "loss": 1.2138,
+      "step": 97780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0927773626892186e-05,
+      "loss": 1.2248,
+      "step": 97785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0927021202579733e-05,
+      "loss": 1.1657,
+      "step": 97790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0926268772973487e-05,
+      "loss": 1.1111,
+      "step": 97795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.092551633807775e-05,
+      "loss": 1.2243,
+      "step": 97800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0924763897896818e-05,
+      "loss": 1.2349,
+      "step": 97805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0924011452434984e-05,
+      "loss": 1.0792,
+      "step": 97810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0923259001696548e-05,
+      "loss": 1.0401,
+      "step": 97815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.092250654568581e-05,
+      "loss": 1.3549,
+      "step": 97820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0921754084407059e-05,
+      "loss": 1.19,
+      "step": 97825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0921001617864599e-05,
+      "loss": 0.9981,
+      "step": 97830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0920249146062722e-05,
+      "loss": 1.4473,
+      "step": 97835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.091949666900573e-05,
+      "loss": 1.178,
+      "step": 97840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0918744186697918e-05,
+      "loss": 1.4454,
+      "step": 97845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0917991699143578e-05,
+      "loss": 1.2146,
+      "step": 97850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0917239206347017e-05,
+      "loss": 1.2751,
+      "step": 97855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0916486708312526e-05,
+      "loss": 1.2286,
+      "step": 97860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0915734205044397e-05,
+      "loss": 1.1106,
+      "step": 97865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.091498169654694e-05,
+      "loss": 1.4494,
+      "step": 97870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0914229182824442e-05,
+      "loss": 1.311,
+      "step": 97875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0913476663881206e-05,
+      "loss": 1.4507,
+      "step": 97880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0912724139721524e-05,
+      "loss": 1.023,
+      "step": 97885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.09119716103497e-05,
+      "loss": 1.2003,
+      "step": 97890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0911219075770023e-05,
+      "loss": 1.1864,
+      "step": 97895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.09104665359868e-05,
+      "loss": 1.2185,
+      "step": 97900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.090971399100432e-05,
+      "loss": 1.4876,
+      "step": 97905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0908961440826885e-05,
+      "loss": 1.3534,
+      "step": 97910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.090820888545879e-05,
+      "loss": 1.2942,
+      "step": 97915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0907456324904335e-05,
+      "loss": 1.227,
+      "step": 97920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0906703759167816e-05,
+      "loss": 1.2986,
+      "step": 97925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0905951188253531e-05,
+      "loss": 1.3874,
+      "step": 97930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0905198612165778e-05,
+      "loss": 1.1472,
+      "step": 97935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0904446030908853e-05,
+      "loss": 1.3986,
+      "step": 97940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0903693444487054e-05,
+      "loss": 1.2162,
+      "step": 97945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0902940852904683e-05,
+      "loss": 1.4025,
+      "step": 97950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0902188256166029e-05,
+      "loss": 1.657,
+      "step": 97955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0901435654275399e-05,
+      "loss": 1.146,
+      "step": 97960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0900683047237083e-05,
+      "loss": 1.0643,
+      "step": 97965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0899930435055384e-05,
+      "loss": 1.1998,
+      "step": 97970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.08991778177346e-05,
+      "loss": 1.1061,
+      "step": 97975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0898425195279025e-05,
+      "loss": 1.1179,
+      "step": 97980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0897672567692958e-05,
+      "loss": 1.2103,
+      "step": 97985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0896919934980701e-05,
+      "loss": 1.6338,
+      "step": 97990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0896167297146546e-05,
+      "loss": 1.061,
+      "step": 97995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0895414654194798e-05,
+      "loss": 1.2992,
+      "step": 98000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0894662006129746e-05,
+      "loss": 1.0235,
+      "step": 98005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0893909352955694e-05,
+      "loss": 1.4629,
+      "step": 98010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0893156694676939e-05,
+      "loss": 1.0351,
+      "step": 98015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0892404031297778e-05,
+      "loss": 1.1144,
+      "step": 98020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0891651362822511e-05,
+      "loss": 1.4204,
+      "step": 98025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0890898689255436e-05,
+      "loss": 1.2242,
+      "step": 98030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0890146010600848e-05,
+      "loss": 1.2186,
+      "step": 98035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0889393326863049e-05,
+      "loss": 1.2914,
+      "step": 98040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0888640638046337e-05,
+      "loss": 1.1721,
+      "step": 98045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0887887944155004e-05,
+      "loss": 1.4416,
+      "step": 98050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0887135245193362e-05,
+      "loss": 1.1614,
+      "step": 98055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0886382541165695e-05,
+      "loss": 1.3243,
+      "step": 98060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0885629832076306e-05,
+      "loss": 1.0709,
+      "step": 98065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0884877117929495e-05,
+      "loss": 1.0685,
+      "step": 98070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0884124398729559e-05,
+      "loss": 1.3306,
+      "step": 98075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0883371674480803e-05,
+      "loss": 1.1721,
+      "step": 98080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.088261894518751e-05,
+      "loss": 1.8203,
+      "step": 98085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0881866210853996e-05,
+      "loss": 1.5961,
+      "step": 98090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0881113471484548e-05,
+      "loss": 1.0826,
+      "step": 98095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.088036072708347e-05,
+      "loss": 1.1282,
+      "step": 98100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.087960797765506e-05,
+      "loss": 1.3386,
+      "step": 98105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.087885522320361e-05,
+      "loss": 1.1154,
+      "step": 98110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0878102463733429e-05,
+      "loss": 1.2159,
+      "step": 98115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0877349699248809e-05,
+      "loss": 1.2691,
+      "step": 98120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.087659692975405e-05,
+      "loss": 1.5313,
+      "step": 98125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.087584415525345e-05,
+      "loss": 1.3944,
+      "step": 98130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0875091375751312e-05,
+      "loss": 1.0366,
+      "step": 98135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0874338591251929e-05,
+      "loss": 1.0795,
+      "step": 98140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0873585801759603e-05,
+      "loss": 1.4897,
+      "step": 98145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0872833007278631e-05,
+      "loss": 1.2236,
+      "step": 98150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0872080207813314e-05,
+      "loss": 1.0735,
+      "step": 98155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0871327403367953e-05,
+      "loss": 1.3654,
+      "step": 98160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.087057459394684e-05,
+      "loss": 0.9982,
+      "step": 98165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0869821779554279e-05,
+      "loss": 1.5738,
+      "step": 98170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0869068960194568e-05,
+      "loss": 1.136,
+      "step": 98175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0868316135872004e-05,
+      "loss": 1.5747,
+      "step": 98180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0867563306590891e-05,
+      "loss": 1.1641,
+      "step": 98185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0866810472355522e-05,
+      "loss": 1.2015,
+      "step": 98190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.08660576331702e-05,
+      "loss": 0.9952,
+      "step": 98195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0865304789039226e-05,
+      "loss": 1.3596,
+      "step": 98200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0864551939966892e-05,
+      "loss": 1.4585,
+      "step": 98205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0863799085957505e-05,
+      "loss": 1.4929,
+      "step": 98210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0863046227015357e-05,
+      "loss": 1.411,
+      "step": 98215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0862293363144751e-05,
+      "loss": 1.2497,
+      "step": 98220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0861540494349989e-05,
+      "loss": 1.405,
+      "step": 98225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0860787620635366e-05,
+      "loss": 1.091,
+      "step": 98230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0860034742005184e-05,
+      "loss": 1.4548,
+      "step": 98235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0859281858463739e-05,
+      "loss": 1.3063,
+      "step": 98240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0858528970015334e-05,
+      "loss": 1.0883,
+      "step": 98245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0857776076664267e-05,
+      "loss": 1.3992,
+      "step": 98250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0857023178414833e-05,
+      "loss": 1.6494,
+      "step": 98255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0856270275271341e-05,
+      "loss": 1.2107,
+      "step": 98260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0855517367238085e-05,
+      "loss": 1.1866,
+      "step": 98265
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0854764454319362e-05,
+      "loss": 1.1688,
+      "step": 98270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0854011536519473e-05,
+      "loss": 1.0834,
+      "step": 98275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0853258613842724e-05,
+      "loss": 0.9266,
+      "step": 98280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0852505686293402e-05,
+      "loss": 1.0125,
+      "step": 98285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0851752753875822e-05,
+      "loss": 1.1928,
+      "step": 98290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.085099981659427e-05,
+      "loss": 1.4654,
+      "step": 98295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0850246874453053e-05,
+      "loss": 1.191,
+      "step": 98300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0849493927456472e-05,
+      "loss": 1.1335,
+      "step": 98305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0848740975608818e-05,
+      "loss": 1.3531,
+      "step": 98310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0847988018914402e-05,
+      "loss": 1.2543,
+      "step": 98315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0847235057377515e-05,
+      "loss": 0.9885,
+      "step": 98320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.084648209100246e-05,
+      "loss": 1.1859,
+      "step": 98325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0845729119793538e-05,
+      "loss": 1.0735,
+      "step": 98330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0844976143755048e-05,
+      "loss": 1.4469,
+      "step": 98335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.084422316289129e-05,
+      "loss": 1.0558,
+      "step": 98340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0843470177206564e-05,
+      "loss": 1.1178,
+      "step": 98345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0842717186705167e-05,
+      "loss": 1.1744,
+      "step": 98350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0841964191391405e-05,
+      "loss": 1.1523,
+      "step": 98355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0841211191269574e-05,
+      "loss": 1.3947,
+      "step": 98360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0840458186343975e-05,
+      "loss": 1.3726,
+      "step": 98365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0839705176618905e-05,
+      "loss": 1.1022,
+      "step": 98370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0838952162098668e-05,
+      "loss": 1.0112,
+      "step": 98375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0838199142787562e-05,
+      "loss": 1.1137,
+      "step": 98380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0837446118689892e-05,
+      "loss": 1.3841,
+      "step": 98385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0836693089809948e-05,
+      "loss": 1.0484,
+      "step": 98390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0835940056152044e-05,
+      "loss": 1.0911,
+      "step": 98395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0835187017720468e-05,
+      "loss": 1.09,
+      "step": 98400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0834433974519523e-05,
+      "loss": 1.375,
+      "step": 98405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0833680926553516e-05,
+      "loss": 1.5276,
+      "step": 98410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.083292787382674e-05,
+      "loss": 1.3999,
+      "step": 98415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0832174816343497e-05,
+      "loss": 1.417,
+      "step": 98420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0831421754108088e-05,
+      "loss": 0.9769,
+      "step": 98425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0830668687124816e-05,
+      "loss": 1.3799,
+      "step": 98430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0829915615397978e-05,
+      "loss": 1.4742,
+      "step": 98435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0829162538931876e-05,
+      "loss": 0.9898,
+      "step": 98440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0828409457730808e-05,
+      "loss": 1.236,
+      "step": 98445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.082765637179908e-05,
+      "loss": 1.3452,
+      "step": 98450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0826903281140985e-05,
+      "loss": 1.5327,
+      "step": 98455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.082615018576083e-05,
+      "loss": 1.29,
+      "step": 98460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0825397085662913e-05,
+      "loss": 1.611,
+      "step": 98465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0824643980851534e-05,
+      "loss": 0.9944,
+      "step": 98470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0823890871330996e-05,
+      "loss": 1.1959,
+      "step": 98475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0823137757105595e-05,
+      "loss": 1.2787,
+      "step": 98480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0822384638179635e-05,
+      "loss": 1.3922,
+      "step": 98485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0821631514557421e-05,
+      "loss": 1.2772,
+      "step": 98490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0820878386243243e-05,
+      "loss": 1.4085,
+      "step": 98495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0820125253241411e-05,
+      "loss": 1.1153,
+      "step": 98500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0819372115556224e-05,
+      "loss": 1.1198,
+      "step": 98505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0818618973191977e-05,
+      "loss": 1.1637,
+      "step": 98510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.081786582615298e-05,
+      "loss": 1.3834,
+      "step": 98515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0817112674443526e-05,
+      "loss": 1.0966,
+      "step": 98520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.081635951806792e-05,
+      "loss": 1.1593,
+      "step": 98525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0815606357030464e-05,
+      "loss": 1.011,
+      "step": 98530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0814853191335454e-05,
+      "loss": 1.3503,
+      "step": 98535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.08141000209872e-05,
+      "loss": 1.4522,
+      "step": 98540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.081334684598999e-05,
+      "loss": 1.4947,
+      "step": 98545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0812593666348134e-05,
+      "loss": 1.2254,
+      "step": 98550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0811840482065932e-05,
+      "loss": 1.1479,
+      "step": 98555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0811087293147685e-05,
+      "loss": 1.3116,
+      "step": 98560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.081033409959769e-05,
+      "loss": 1.2887,
+      "step": 98565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0809580901420256e-05,
+      "loss": 1.0962,
+      "step": 98570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0808827698619678e-05,
+      "loss": 1.204,
+      "step": 98575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0808074491200257e-05,
+      "loss": 1.2632,
+      "step": 98580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0807321279166296e-05,
+      "loss": 0.9512,
+      "step": 98585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0806568062522095e-05,
+      "loss": 1.1651,
+      "step": 98590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0805814841271962e-05,
+      "loss": 1.014,
+      "step": 98595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0805061615420191e-05,
+      "loss": 1.0559,
+      "step": 98600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0804308384971085e-05,
+      "loss": 1.2757,
+      "step": 98605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0803555149928944e-05,
+      "loss": 1.4113,
+      "step": 98610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0802801910298069e-05,
+      "loss": 1.265,
+      "step": 98615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0802048666082768e-05,
+      "loss": 1.3316,
+      "step": 98620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0801295417287335e-05,
+      "loss": 1.1337,
+      "step": 98625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0800542163916075e-05,
+      "loss": 1.0674,
+      "step": 98630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0799788905973288e-05,
+      "loss": 1.2605,
+      "step": 98635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0799035643463276e-05,
+      "loss": 1.1372,
+      "step": 98640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0798282376390342e-05,
+      "loss": 1.1019,
+      "step": 98645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0797529104758784e-05,
+      "loss": 1.4037,
+      "step": 98650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0796775828572907e-05,
+      "loss": 1.1587,
+      "step": 98655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0796022547837014e-05,
+      "loss": 1.089,
+      "step": 98660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.07952692625554e-05,
+      "loss": 0.9463,
+      "step": 98665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0794515972732373e-05,
+      "loss": 0.9414,
+      "step": 98670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0793762678372232e-05,
+      "loss": 1.2576,
+      "step": 98675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.079300937947928e-05,
+      "loss": 1.6024,
+      "step": 98680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0792256076057819e-05,
+      "loss": 1.32,
+      "step": 98685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0791502768112147e-05,
+      "loss": 1.2983,
+      "step": 98690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0790749455646567e-05,
+      "loss": 1.2385,
+      "step": 98695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0789996138665388e-05,
+      "loss": 1.2,
+      "step": 98700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0789242817172903e-05,
+      "loss": 1.3029,
+      "step": 98705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0788489491173415e-05,
+      "loss": 1.3081,
+      "step": 98710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0787736160671232e-05,
+      "loss": 1.4534,
+      "step": 98715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0786982825670647e-05,
+      "loss": 1.2328,
+      "step": 98720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0786229486175973e-05,
+      "loss": 1.2625,
+      "step": 98725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.07854761421915e-05,
+      "loss": 1.2524,
+      "step": 98730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0784722793721538e-05,
+      "loss": 1.2877,
+      "step": 98735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0783969440770385e-05,
+      "loss": 1.2949,
+      "step": 98740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0783216083342347e-05,
+      "loss": 0.865,
+      "step": 98745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0782462721441724e-05,
+      "loss": 1.0675,
+      "step": 98750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0781709355072818e-05,
+      "loss": 1.0278,
+      "step": 98755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.078095598423993e-05,
+      "loss": 1.163,
+      "step": 98760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0780202608947366e-05,
+      "loss": 1.459,
+      "step": 98765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0779449229199424e-05,
+      "loss": 1.1631,
+      "step": 98770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0778695845000406e-05,
+      "loss": 1.4806,
+      "step": 98775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0777942456354617e-05,
+      "loss": 1.3423,
+      "step": 98780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0777189063266359e-05,
+      "loss": 1.3093,
+      "step": 98785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0776435665739932e-05,
+      "loss": 1.4436,
+      "step": 98790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0775682263779641e-05,
+      "loss": 1.3438,
+      "step": 98795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0774928857389784e-05,
+      "loss": 1.4368,
+      "step": 98800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0774175446574671e-05,
+      "loss": 1.1475,
+      "step": 98805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0773422031338599e-05,
+      "loss": 1.176,
+      "step": 98810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0772668611685866e-05,
+      "loss": 1.1662,
+      "step": 98815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0771915187620787e-05,
+      "loss": 1.2575,
+      "step": 98820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.077116175914765e-05,
+      "loss": 1.0306,
+      "step": 98825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0770408326270772e-05,
+      "loss": 1.3578,
+      "step": 98830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0769654888994441e-05,
+      "loss": 1.3092,
+      "step": 98835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.076890144732297e-05,
+      "loss": 1.237,
+      "step": 98840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.076814800126066e-05,
+      "loss": 1.2576,
+      "step": 98845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0767394550811809e-05,
+      "loss": 1.3938,
+      "step": 98850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0766641095980724e-05,
+      "loss": 1.2387,
+      "step": 98855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0765887636771703e-05,
+      "loss": 1.3352,
+      "step": 98860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0765134173189055e-05,
+      "loss": 1.1252,
+      "step": 98865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0764380705237079e-05,
+      "loss": 1.1033,
+      "step": 98870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0763627232920078e-05,
+      "loss": 1.0775,
+      "step": 98875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0762873756242353e-05,
+      "loss": 1.2838,
+      "step": 98880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.076212027520821e-05,
+      "loss": 1.4381,
+      "step": 98885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.076136678982195e-05,
+      "loss": 1.8831,
+      "step": 98890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0760613300087875e-05,
+      "loss": 1.0953,
+      "step": 98895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.075985980601029e-05,
+      "loss": 1.5512,
+      "step": 98900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.07591063075935e-05,
+      "loss": 1.3646,
+      "step": 98905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0758352804841801e-05,
+      "loss": 1.1874,
+      "step": 98910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0757599297759498e-05,
+      "loss": 1.2003,
+      "step": 98915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0756845786350895e-05,
+      "loss": 1.5319,
+      "step": 98920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0756092270620301e-05,
+      "loss": 1.3287,
+      "step": 98925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.075533875057201e-05,
+      "loss": 1.2062,
+      "step": 98930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0754585226210331e-05,
+      "loss": 0.9776,
+      "step": 98935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.075383169753956e-05,
+      "loss": 1.2749,
+      "step": 98940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0753078164564007e-05,
+      "loss": 1.2448,
+      "step": 98945
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0752324627287975e-05,
+      "loss": 1.0434,
+      "step": 98950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0751571085715761e-05,
+      "loss": 1.2215,
+      "step": 98955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0750817539851675e-05,
+      "loss": 1.1195,
+      "step": 98960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0750063989700015e-05,
+      "loss": 1.0382,
+      "step": 98965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0749310435265086e-05,
+      "loss": 1.1297,
+      "step": 98970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0748556876551192e-05,
+      "loss": 1.1596,
+      "step": 98975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0747803313562634e-05,
+      "loss": 1.1228,
+      "step": 98980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.074704974630372e-05,
+      "loss": 1.2137,
+      "step": 98985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.074629617477875e-05,
+      "loss": 1.5093,
+      "step": 98990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0745542598992024e-05,
+      "loss": 1.3734,
+      "step": 98995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.074478901894785e-05,
+      "loss": 1.2844,
+      "step": 99000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0744035434650532e-05,
+      "loss": 1.3936,
+      "step": 99005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0743281846104373e-05,
+      "loss": 1.463,
+      "step": 99010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.074252825331367e-05,
+      "loss": 1.3756,
+      "step": 99015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0741774656282735e-05,
+      "loss": 1.2079,
+      "step": 99020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0741021055015864e-05,
+      "loss": 1.1505,
+      "step": 99025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.074026744951737e-05,
+      "loss": 0.9429,
+      "step": 99030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0739513839791544e-05,
+      "loss": 1.3753,
+      "step": 99035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0738760225842699e-05,
+      "loss": 1.3187,
+      "step": 99040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.073800660767514e-05,
+      "loss": 1.2433,
+      "step": 99045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0737252985293161e-05,
+      "loss": 1.2902,
+      "step": 99050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0736499358701074e-05,
+      "loss": 1.3417,
+      "step": 99055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0735745727903175e-05,
+      "loss": 0.9596,
+      "step": 99060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0734992092903776e-05,
+      "loss": 1.0929,
+      "step": 99065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0734238453707176e-05,
+      "loss": 1.156,
+      "step": 99070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.073348481031768e-05,
+      "loss": 1.4324,
+      "step": 99075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0732731162739591e-05,
+      "loss": 1.0936,
+      "step": 99080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0731977510977213e-05,
+      "loss": 1.1749,
+      "step": 99085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0731223855034848e-05,
+      "loss": 1.163,
+      "step": 99090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0730470194916804e-05,
+      "loss": 1.4237,
+      "step": 99095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0729716530627382e-05,
+      "loss": 1.3087,
+      "step": 99100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0728962862170886e-05,
+      "loss": 1.1787,
+      "step": 99105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.072820918955162e-05,
+      "loss": 1.3394,
+      "step": 99110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0727455512773885e-05,
+      "loss": 1.2163,
+      "step": 99115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0726701831841992e-05,
+      "loss": 1.1068,
+      "step": 99120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0725948146760238e-05,
+      "loss": 1.2539,
+      "step": 99125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0725194457532928e-05,
+      "loss": 1.1252,
+      "step": 99130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0724440764164373e-05,
+      "loss": 1.2181,
+      "step": 99135
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0723687066658865e-05,
+      "loss": 1.5127,
+      "step": 99140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.072293336502072e-05,
+      "loss": 1.4041,
+      "step": 99145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0722179659254237e-05,
+      "loss": 1.3619,
+      "step": 99150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0721425949363715e-05,
+      "loss": 1.285,
+      "step": 99155
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0720672235353468e-05,
+      "loss": 1.2332,
+      "step": 99160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0719918517227789e-05,
+      "loss": 0.9846,
+      "step": 99165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0719164794990993e-05,
+      "loss": 1.1898,
+      "step": 99170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0718411068647376e-05,
+      "loss": 1.2947,
+      "step": 99175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0717657338201248e-05,
+      "loss": 1.3533,
+      "step": 99180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0716903603656909e-05,
+      "loss": 1.0591,
+      "step": 99185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0716149865018665e-05,
+      "loss": 1.2902,
+      "step": 99190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0715396122290822e-05,
+      "loss": 1.3113,
+      "step": 99195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.071464237547768e-05,
+      "loss": 1.0802,
+      "step": 99200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0713888624583548e-05,
+      "loss": 1.0794,
+      "step": 99205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0713134869612727e-05,
+      "loss": 1.2691,
+      "step": 99210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.071238111056952e-05,
+      "loss": 1.0986,
+      "step": 99215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0711627347458236e-05,
+      "loss": 1.0215,
+      "step": 99220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0710873580283177e-05,
+      "loss": 1.4497,
+      "step": 99225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0710119809048647e-05,
+      "loss": 1.292,
+      "step": 99230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.070936603375895e-05,
+      "loss": 1.1527,
+      "step": 99235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0708612254418393e-05,
+      "loss": 1.4463,
+      "step": 99240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0707858471031281e-05,
+      "loss": 1.0732,
+      "step": 99245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.070710468360191e-05,
+      "loss": 1.255,
+      "step": 99250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0706350892134598e-05,
+      "loss": 1.268,
+      "step": 99255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0705597096633635e-05,
+      "loss": 1.2723,
+      "step": 99260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0704843297103342e-05,
+      "loss": 1.241,
+      "step": 99265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0704089493548007e-05,
+      "loss": 1.2663,
+      "step": 99270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0703335685971945e-05,
+      "loss": 1.0325,
+      "step": 99275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0702581874379458e-05,
+      "loss": 1.1232,
+      "step": 99280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0701828058774852e-05,
+      "loss": 1.0027,
+      "step": 99285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0701074239162428e-05,
+      "loss": 1.1481,
+      "step": 99290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0700320415546495e-05,
+      "loss": 1.223,
+      "step": 99295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0699566587931356e-05,
+      "loss": 0.9322,
+      "step": 99300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0698812756321314e-05,
+      "loss": 1.4501,
+      "step": 99305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0698058920720675e-05,
+      "loss": 0.8669,
+      "step": 99310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0697305081133744e-05,
+      "loss": 1.5101,
+      "step": 99315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0696551237564828e-05,
+      "loss": 1.1758,
+      "step": 99320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.069579739001823e-05,
+      "loss": 1.3388,
+      "step": 99325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0695043538498251e-05,
+      "loss": 1.2251,
+      "step": 99330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0694289683009203e-05,
+      "loss": 1.2331,
+      "step": 99335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0693535823555384e-05,
+      "loss": 1.1588,
+      "step": 99340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0692781960141107e-05,
+      "loss": 1.1085,
+      "step": 99345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.069202809277067e-05,
+      "loss": 1.2647,
+      "step": 99350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0691274221448378e-05,
+      "loss": 1.307,
+      "step": 99355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0690520346178544e-05,
+      "loss": 1.2268,
+      "step": 99360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0689766466965463e-05,
+      "loss": 1.4842,
+      "step": 99365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0689012583813447e-05,
+      "loss": 1.2171,
+      "step": 99370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0688258696726796e-05,
+      "loss": 1.1062,
+      "step": 99375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0687504805709818e-05,
+      "loss": 1.303,
+      "step": 99380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.068675091076682e-05,
+      "loss": 1.2045,
+      "step": 99385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0685997011902103e-05,
+      "loss": 1.0901,
+      "step": 99390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0685243109119976e-05,
+      "loss": 1.2533,
+      "step": 99395
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0684489202424739e-05,
+      "loss": 1.1483,
+      "step": 99400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0683735291820706e-05,
+      "loss": 1.3907,
+      "step": 99405
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0682981377312172e-05,
+      "loss": 1.0284,
+      "step": 99410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0682227458903446e-05,
+      "loss": 1.2084,
+      "step": 99415
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0681473536598839e-05,
+      "loss": 1.4299,
+      "step": 99420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.068071961040265e-05,
+      "loss": 1.1956,
+      "step": 99425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0679965680319184e-05,
+      "loss": 1.6479,
+      "step": 99430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0679211746352747e-05,
+      "loss": 1.1782,
+      "step": 99435
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.067845780850765e-05,
+      "loss": 1.3036,
+      "step": 99440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0677703866788191e-05,
+      "loss": 1.5326,
+      "step": 99445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0676949921198681e-05,
+      "loss": 1.0478,
+      "step": 99450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0676195971743422e-05,
+      "loss": 1.2942,
+      "step": 99455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0675442018426717e-05,
+      "loss": 0.941,
+      "step": 99460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0674688061252881e-05,
+      "loss": 1.1263,
+      "step": 99465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0673934100226206e-05,
+      "loss": 1.1614,
+      "step": 99470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0673180135351008e-05,
+      "loss": 0.9723,
+      "step": 99475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0672426166631592e-05,
+      "loss": 1.1714,
+      "step": 99480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0671672194072258e-05,
+      "loss": 1.5008,
+      "step": 99485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0670918217677317e-05,
+      "loss": 0.9456,
+      "step": 99490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0670164237451068e-05,
+      "loss": 1.2244,
+      "step": 99495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0669410253397823e-05,
+      "loss": 1.1276,
+      "step": 99500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0668656265521886e-05,
+      "loss": 1.3227,
+      "step": 99505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0667902273827561e-05,
+      "loss": 1.2213,
+      "step": 99510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0667148278319154e-05,
+      "loss": 1.4821,
+      "step": 99515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0666394279000973e-05,
+      "loss": 1.3904,
+      "step": 99520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0665640275877323e-05,
+      "loss": 1.0664,
+      "step": 99525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0664886268952507e-05,
+      "loss": 1.1861,
+      "step": 99530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0664132258230834e-05,
+      "loss": 1.036,
+      "step": 99535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0663378243716609e-05,
+      "loss": 1.2056,
+      "step": 99540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0662624225414138e-05,
+      "loss": 1.0274,
+      "step": 99545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0661870203327726e-05,
+      "loss": 1.2096,
+      "step": 99550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0661116177461679e-05,
+      "loss": 1.3357,
+      "step": 99555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0660362147820302e-05,
+      "loss": 1.1156,
+      "step": 99560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.06596081144079e-05,
+      "loss": 1.2928,
+      "step": 99565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0658854077228786e-05,
+      "loss": 1.2039,
+      "step": 99570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0658100036287256e-05,
+      "loss": 1.6137,
+      "step": 99575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0657345991587622e-05,
+      "loss": 0.9439,
+      "step": 99580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.065659194313419e-05,
+      "loss": 0.9653,
+      "step": 99585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0655837890931265e-05,
+      "loss": 1.6202,
+      "step": 99590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0655083834983156e-05,
+      "loss": 1.3898,
+      "step": 99595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.065432977529416e-05,
+      "loss": 1.1418,
+      "step": 99600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.065357571186859e-05,
+      "loss": 1.3373,
+      "step": 99605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0652821644710754e-05,
+      "loss": 1.111,
+      "step": 99610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0652067573824954e-05,
+      "loss": 1.5068,
+      "step": 99615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0651313499215496e-05,
+      "loss": 1.1894,
+      "step": 99620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0650559420886691e-05,
+      "loss": 1.1692,
+      "step": 99625
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.064980533884284e-05,
+      "loss": 1.0543,
+      "step": 99630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.064905125308825e-05,
+      "loss": 1.0171,
+      "step": 99635
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.064829716362723e-05,
+      "loss": 1.002,
+      "step": 99640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0647543070464082e-05,
+      "loss": 1.2448,
+      "step": 99645
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0646788973603117e-05,
+      "loss": 1.4891,
+      "step": 99650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0646034873048638e-05,
+      "loss": 1.0995,
+      "step": 99655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0645280768804953e-05,
+      "loss": 1.3208,
+      "step": 99660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0644526660876368e-05,
+      "loss": 1.0861,
+      "step": 99665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0643772549267188e-05,
+      "loss": 1.7625,
+      "step": 99670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0643018433981722e-05,
+      "loss": 1.2023,
+      "step": 99675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0642264315024273e-05,
+      "loss": 1.7702,
+      "step": 99680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.064151019239915e-05,
+      "loss": 1.322,
+      "step": 99685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0640756066110662e-05,
+      "loss": 1.4795,
+      "step": 99690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0640001936163106e-05,
+      "loss": 1.3746,
+      "step": 99695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0639247802560802e-05,
+      "loss": 1.2829,
+      "step": 99700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0638493665308043e-05,
+      "loss": 1.0953,
+      "step": 99705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0637739524409143e-05,
+      "loss": 1.2724,
+      "step": 99710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0636985379868409e-05,
+      "loss": 1.323,
+      "step": 99715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0636231231690147e-05,
+      "loss": 0.9585,
+      "step": 99720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.063547707987866e-05,
+      "loss": 1.0838,
+      "step": 99725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0634722924438259e-05,
+      "loss": 1.3633,
+      "step": 99730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0633968765373246e-05,
+      "loss": 1.1464,
+      "step": 99735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0633214602687933e-05,
+      "loss": 1.1283,
+      "step": 99740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0632460436386622e-05,
+      "loss": 1.1588,
+      "step": 99745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0631706266473623e-05,
+      "loss": 1.1861,
+      "step": 99750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0630952092953242e-05,
+      "loss": 1.4901,
+      "step": 99755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0630197915829785e-05,
+      "loss": 1.0965,
+      "step": 99760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0629443735107556e-05,
+      "loss": 1.4803,
+      "step": 99765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0628689550790869e-05,
+      "loss": 1.1931,
+      "step": 99770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0627935362884023e-05,
+      "loss": 1.3426,
+      "step": 99775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0627181171391332e-05,
+      "loss": 1.134,
+      "step": 99780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0626426976317095e-05,
+      "loss": 1.0902,
+      "step": 99785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0625672777665625e-05,
+      "loss": 1.0434,
+      "step": 99790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.062491857544123e-05,
+      "loss": 1.2042,
+      "step": 99795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0624164369648206e-05,
+      "loss": 1.1585,
+      "step": 99800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0623410160290875e-05,
+      "loss": 1.1244,
+      "step": 99805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0622655947373535e-05,
+      "loss": 1.2801,
+      "step": 99810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0621901730900492e-05,
+      "loss": 1.1556,
+      "step": 99815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0621147510876058e-05,
+      "loss": 1.2265,
+      "step": 99820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0620393287304538e-05,
+      "loss": 1.2877,
+      "step": 99825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0619639060190237e-05,
+      "loss": 1.5251,
+      "step": 99830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0618884829537465e-05,
+      "loss": 1.1126,
+      "step": 99835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0618130595350527e-05,
+      "loss": 1.1247,
+      "step": 99840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0617376357633731e-05,
+      "loss": 1.3412,
+      "step": 99845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0616622116391385e-05,
+      "loss": 1.5972,
+      "step": 99850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0615867871627795e-05,
+      "loss": 1.5487,
+      "step": 99855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0615113623347267e-05,
+      "loss": 1.0282,
+      "step": 99860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0614359371554108e-05,
+      "loss": 0.9943,
+      "step": 99865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0613605116252627e-05,
+      "loss": 0.9993,
+      "step": 99870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0612850857447134e-05,
+      "loss": 0.8551,
+      "step": 99875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0612096595141929e-05,
+      "loss": 0.9179,
+      "step": 99880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0611342329341327e-05,
+      "loss": 1.4944,
+      "step": 99885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0610588060049632e-05,
+      "loss": 1.5941,
+      "step": 99890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0609833787271144e-05,
+      "loss": 0.9014,
+      "step": 99895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0609079511010185e-05,
+      "loss": 1.4651,
+      "step": 99900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0608325231271048e-05,
+      "loss": 1.0254,
+      "step": 99905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.060757094805805e-05,
+      "loss": 1.7284,
+      "step": 99910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0606816661375496e-05,
+      "loss": 1.1312,
+      "step": 99915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.060606237122769e-05,
+      "loss": 1.2942,
+      "step": 99920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0605308077618943e-05,
+      "loss": 1.141,
+      "step": 99925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0604553780553563e-05,
+      "loss": 1.5027,
+      "step": 99930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0603799480035854e-05,
+      "loss": 1.2766,
+      "step": 99935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0603045176070126e-05,
+      "loss": 1.1188,
+      "step": 99940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0602290868660684e-05,
+      "loss": 1.2403,
+      "step": 99945
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.060153655781184e-05,
+      "loss": 1.2286,
+      "step": 99950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0600782243527896e-05,
+      "loss": 1.6069,
+      "step": 99955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0600027925813163e-05,
+      "loss": 1.5426,
+      "step": 99960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0599273604671949e-05,
+      "loss": 1.1251,
+      "step": 99965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0598519280108558e-05,
+      "loss": 1.1146,
+      "step": 99970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.05977649521273e-05,
+      "loss": 1.4784,
+      "step": 99975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0597010620732486e-05,
+      "loss": 1.1533,
+      "step": 99980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0596256285928416e-05,
+      "loss": 1.2922,
+      "step": 99985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0595501947719408e-05,
+      "loss": 1.0784,
+      "step": 99990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.059474760610976e-05,
+      "loss": 1.3717,
+      "step": 99995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.059399326110378e-05,
+      "loss": 1.1926,
+      "step": 100000
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 1.2235456705093384,
+      "eval_runtime": 6474.323,
+      "eval_samples_per_second": 3.569,
+      "eval_steps_per_second": 1.785,
+      "step": 100000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0593238912705786e-05,
+      "loss": 1.3475,
+      "step": 100005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0592484560920073e-05,
+      "loss": 1.0106,
+      "step": 100010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0591730205750956e-05,
+      "loss": 1.0692,
+      "step": 100015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0590975847202742e-05,
+      "loss": 1.3085,
+      "step": 100020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0590221485279737e-05,
+      "loss": 1.3372,
+      "step": 100025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0589467119986255e-05,
+      "loss": 1.0656,
+      "step": 100030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0588712751326591e-05,
+      "loss": 1.2178,
+      "step": 100035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0587958379305063e-05,
+      "loss": 1.4459,
+      "step": 100040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0587204003925978e-05,
+      "loss": 1.5086,
+      "step": 100045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0586449625193643e-05,
+      "loss": 1.1637,
+      "step": 100050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0585695243112361e-05,
+      "loss": 1.4481,
+      "step": 100055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0584940857686447e-05,
+      "loss": 1.6909,
+      "step": 100060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0584186468920207e-05,
+      "loss": 1.3721,
+      "step": 100065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0583432076817948e-05,
+      "loss": 1.2175,
+      "step": 100070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0582677681383977e-05,
+      "loss": 1.1728,
+      "step": 100075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0581923282622601e-05,
+      "loss": 1.4601,
+      "step": 100080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0581168880538134e-05,
+      "loss": 1.0997,
+      "step": 100085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.058041447513488e-05,
+      "loss": 1.3787,
+      "step": 100090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0579660066417141e-05,
+      "loss": 1.2918,
+      "step": 100095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0578905654389237e-05,
+      "loss": 0.8988,
+      "step": 100100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.057815123905547e-05,
+      "loss": 1.2793,
+      "step": 100105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.057739682042015e-05,
+      "loss": 1.3058,
+      "step": 100110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0576642398487578e-05,
+      "loss": 1.1512,
+      "step": 100115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.057588797326207e-05,
+      "loss": 1.1426,
+      "step": 100120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0575133544747934e-05,
+      "loss": 1.3265,
+      "step": 100125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0574379112949476e-05,
+      "loss": 1.5504,
+      "step": 100130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0573624677871003e-05,
+      "loss": 1.0807,
+      "step": 100135
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0572870239516824e-05,
+      "loss": 1.061,
+      "step": 100140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.057211579789125e-05,
+      "loss": 1.5565,
+      "step": 100145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0571361352998584e-05,
+      "loss": 1.0457,
+      "step": 100150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0570606904843139e-05,
+      "loss": 1.2049,
+      "step": 100155
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0569852453429224e-05,
+      "loss": 1.2116,
+      "step": 100160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0569097998761141e-05,
+      "loss": 1.1779,
+      "step": 100165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0568343540843204e-05,
+      "loss": 1.152,
+      "step": 100170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.056758907967972e-05,
+      "loss": 1.2854,
+      "step": 100175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0566834615274997e-05,
+      "loss": 1.2515,
+      "step": 100180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0566080147633342e-05,
+      "loss": 1.2165,
+      "step": 100185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0565325676759068e-05,
+      "loss": 0.9801,
+      "step": 100190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.056457120265648e-05,
+      "loss": 1.198,
+      "step": 100195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0563816725329884e-05,
+      "loss": 1.3838,
+      "step": 100200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0563062244783592e-05,
+      "loss": 1.2811,
+      "step": 100205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0562307761021911e-05,
+      "loss": 1.1437,
+      "step": 100210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0561553274049154e-05,
+      "loss": 0.957,
+      "step": 100215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0560798783869621e-05,
+      "loss": 1.0445,
+      "step": 100220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0560044290487628e-05,
+      "loss": 1.6798,
+      "step": 100225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0559289793907482e-05,
+      "loss": 1.1428,
+      "step": 100230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0558535294133487e-05,
+      "loss": 1.1645,
+      "step": 100235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0557780791169958e-05,
+      "loss": 1.1835,
+      "step": 100240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.05570262850212e-05,
+      "loss": 1.1942,
+      "step": 100245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0556271775691523e-05,
+      "loss": 1.2567,
+      "step": 100250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0555517263185233e-05,
+      "loss": 1.0414,
+      "step": 100255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0554762747506642e-05,
+      "loss": 1.4426,
+      "step": 100260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0554008228660057e-05,
+      "loss": 1.3737,
+      "step": 100265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0553253706649787e-05,
+      "loss": 1.5569,
+      "step": 100270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0552499181480141e-05,
+      "loss": 1.3879,
+      "step": 100275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0551744653155427e-05,
+      "loss": 1.0545,
+      "step": 100280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0550990121679956e-05,
+      "loss": 1.2857,
+      "step": 100285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0550235587058033e-05,
+      "loss": 1.0744,
+      "step": 100290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0549481049293971e-05,
+      "loss": 1.1623,
+      "step": 100295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0548726508392072e-05,
+      "loss": 1.1715,
+      "step": 100300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0547971964356652e-05,
+      "loss": 1.6718,
+      "step": 100305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0547217417192022e-05,
+      "loss": 1.3186,
+      "step": 100310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0546462866902478e-05,
+      "loss": 1.1196,
+      "step": 100315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0545708313492345e-05,
+      "loss": 1.0948,
+      "step": 100320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0544953756965916e-05,
+      "loss": 1.4032,
+      "step": 100325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0544199197327513e-05,
+      "loss": 1.2126,
+      "step": 100330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0543444634581443e-05,
+      "loss": 1.3244,
+      "step": 100335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0542690068732007e-05,
+      "loss": 1.1088,
+      "step": 100340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0541935499783517e-05,
+      "loss": 1.3485,
+      "step": 100345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0541180927740288e-05,
+      "loss": 1.3501,
+      "step": 100350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0540426352606624e-05,
+      "loss": 1.3483,
+      "step": 100355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0539671774386835e-05,
+      "loss": 1.1113,
+      "step": 100360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.053891719308523e-05,
+      "loss": 1.0755,
+      "step": 100365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0538162608706116e-05,
+      "loss": 1.0139,
+      "step": 100370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0537408021253805e-05,
+      "loss": 1.2043,
+      "step": 100375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0536653430732607e-05,
+      "loss": 1.2553,
+      "step": 100380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0535898837146826e-05,
+      "loss": 1.1793,
+      "step": 100385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0535144240500776e-05,
+      "loss": 1.3943,
+      "step": 100390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0534389640798765e-05,
+      "loss": 1.2776,
+      "step": 100395
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0533635038045102e-05,
+      "loss": 1.1739,
+      "step": 100400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0532880432244097e-05,
+      "loss": 1.2641,
+      "step": 100405
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0532125823400052e-05,
+      "loss": 1.1446,
+      "step": 100410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.053137121151729e-05,
+      "loss": 1.1268,
+      "step": 100415
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0530616596600107e-05,
+      "loss": 1.3934,
+      "step": 100420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0529861978652822e-05,
+      "loss": 1.0004,
+      "step": 100425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0529107357679738e-05,
+      "loss": 1.3892,
+      "step": 100430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0528352733685166e-05,
+      "loss": 1.1875,
+      "step": 100435
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0527598106673419e-05,
+      "loss": 1.0507,
+      "step": 100440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0526843476648799e-05,
+      "loss": 1.1282,
+      "step": 100445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0526088843615621e-05,
+      "loss": 1.1058,
+      "step": 100450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0525334207578192e-05,
+      "loss": 1.112,
+      "step": 100455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0524579568540823e-05,
+      "loss": 1.2027,
+      "step": 100460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0523824926507822e-05,
+      "loss": 1.1745,
+      "step": 100465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0523070281483498e-05,
+      "loss": 1.1304,
+      "step": 100470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0522315633472164e-05,
+      "loss": 1.2217,
+      "step": 100475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0521560982478125e-05,
+      "loss": 1.2519,
+      "step": 100480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0520806328505693e-05,
+      "loss": 1.1429,
+      "step": 100485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0520051671559175e-05,
+      "loss": 1.4882,
+      "step": 100490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0519297011642886e-05,
+      "loss": 1.3005,
+      "step": 100495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.051854234876113e-05,
+      "loss": 1.1354,
+      "step": 100500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0517787682918217e-05,
+      "loss": 1.3249,
+      "step": 100505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0517033014118458e-05,
+      "loss": 1.2218,
+      "step": 100510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0516278342366161e-05,
+      "loss": 1.2325,
+      "step": 100515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.051552366766564e-05,
+      "loss": 1.3229,
+      "step": 100520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.05147689900212e-05,
+      "loss": 1.3248,
+      "step": 100525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0514014309437153e-05,
+      "loss": 1.2005,
+      "step": 100530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.051325962591781e-05,
+      "loss": 1.519,
+      "step": 100535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0512504939467473e-05,
+      "loss": 0.9806,
+      "step": 100540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0511750250090462e-05,
+      "loss": 0.9567,
+      "step": 100545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.051099555779108e-05,
+      "loss": 1.0377,
+      "step": 100550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0510240862573638e-05,
+      "loss": 1.2825,
+      "step": 100555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0509486164442446e-05,
+      "loss": 1.0132,
+      "step": 100560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0508731463401813e-05,
+      "loss": 1.3224,
+      "step": 100565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0507976759456054e-05,
+      "loss": 1.1512,
+      "step": 100570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0507222052609472e-05,
+      "loss": 1.3499,
+      "step": 100575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0506467342866377e-05,
+      "loss": 1.1601,
+      "step": 100580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0505712630231084e-05,
+      "loss": 1.3631,
+      "step": 100585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.05049579147079e-05,
+      "loss": 1.3267,
+      "step": 100590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0504203196301132e-05,
+      "loss": 1.3656,
+      "step": 100595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0503448475015097e-05,
+      "loss": 1.4619,
+      "step": 100600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0502693750854095e-05,
+      "loss": 1.2645,
+      "step": 100605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0501939023822448e-05,
+      "loss": 1.2261,
+      "step": 100610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0501184293924454e-05,
+      "loss": 1.4337,
+      "step": 100615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0500429561164426e-05,
+      "loss": 1.1583,
+      "step": 100620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0499674825546682e-05,
+      "loss": 1.5146,
+      "step": 100625
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.049892008707552e-05,
+      "loss": 1.2701,
+      "step": 100630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0498165345755263e-05,
+      "loss": 1.1564,
+      "step": 100635
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0497410601590212e-05,
+      "loss": 1.3306,
+      "step": 100640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0496655854584674e-05,
+      "loss": 1.1186,
+      "step": 100645
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.049590110474297e-05,
+      "loss": 1.0651,
+      "step": 100650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0495146352069398e-05,
+      "loss": 1.1814,
+      "step": 100655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0494391596568279e-05,
+      "loss": 1.2452,
+      "step": 100660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0493636838243917e-05,
+      "loss": 1.3475,
+      "step": 100665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0492882077100622e-05,
+      "loss": 1.1179,
+      "step": 100670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0492127313142707e-05,
+      "loss": 1.4623,
+      "step": 100675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0491372546374477e-05,
+      "loss": 1.4151,
+      "step": 100680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0490617776800247e-05,
+      "loss": 1.6649,
+      "step": 100685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0489863004424327e-05,
+      "loss": 1.3918,
+      "step": 100690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0489108229251025e-05,
+      "loss": 1.2203,
+      "step": 100695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0488353451284652e-05,
+      "loss": 1.1774,
+      "step": 100700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0487598670529518e-05,
+      "loss": 1.4152,
+      "step": 100705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0486843886989934e-05,
+      "loss": 1.2635,
+      "step": 100710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0486089100670209e-05,
+      "loss": 1.3731,
+      "step": 100715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0485334311574653e-05,
+      "loss": 1.2392,
+      "step": 100720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0484579519707577e-05,
+      "loss": 1.1483,
+      "step": 100725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0483824725073293e-05,
+      "loss": 1.1021,
+      "step": 100730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0483069927676108e-05,
+      "loss": 1.3842,
+      "step": 100735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0482315127520333e-05,
+      "loss": 1.2212,
+      "step": 100740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0481560324610283e-05,
+      "loss": 1.1791,
+      "step": 100745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0480805518950261e-05,
+      "loss": 1.5289,
+      "step": 100750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0480050710544585e-05,
+      "loss": 1.0832,
+      "step": 100755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0479295899397555e-05,
+      "loss": 1.1638,
+      "step": 100760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0478541085513492e-05,
+      "loss": 1.4409,
+      "step": 100765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0477786268896701e-05,
+      "loss": 1.4612,
+      "step": 100770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0477031449551493e-05,
+      "loss": 1.5335,
+      "step": 100775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0476276627482181e-05,
+      "loss": 1.1726,
+      "step": 100780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0475521802693069e-05,
+      "loss": 1.1623,
+      "step": 100785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0474766975188475e-05,
+      "loss": 1.1766,
+      "step": 100790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0474012144972707e-05,
+      "loss": 1.3713,
+      "step": 100795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0473257312050074e-05,
+      "loss": 1.4815,
+      "step": 100800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0472502476424886e-05,
+      "loss": 1.1248,
+      "step": 100805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0471747638101454e-05,
+      "loss": 1.022,
+      "step": 100810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0470992797084092e-05,
+      "loss": 1.1625,
+      "step": 100815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0470237953377103e-05,
+      "loss": 1.1894,
+      "step": 100820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0469483106984809e-05,
+      "loss": 1.4265,
+      "step": 100825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0468728257911507e-05,
+      "loss": 1.432,
+      "step": 100830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.046797340616152e-05,
+      "loss": 1.2402,
+      "step": 100835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0467218551739154e-05,
+      "loss": 1.3308,
+      "step": 100840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0466463694648712e-05,
+      "loss": 1.3576,
+      "step": 100845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0465708834894519e-05,
+      "loss": 1.5787,
+      "step": 100850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0464953972480872e-05,
+      "loss": 1.0206,
+      "step": 100855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0464199107412092e-05,
+      "loss": 1.4352,
+      "step": 100860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0463444239692481e-05,
+      "loss": 1.3793,
+      "step": 100865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0462689369326357e-05,
+      "loss": 1.0446,
+      "step": 100870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.046193449631803e-05,
+      "loss": 1.1228,
+      "step": 100875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0461179620671805e-05,
+      "loss": 1.2044,
+      "step": 100880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0460424742391998e-05,
+      "loss": 0.9838,
+      "step": 100885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0459669861482917e-05,
+      "loss": 1.2479,
+      "step": 100890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0458914977948875e-05,
+      "loss": 1.6671,
+      "step": 100895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.045816009179418e-05,
+      "loss": 1.1122,
+      "step": 100900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0457405203023145e-05,
+      "loss": 1.0692,
+      "step": 100905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0456650311640082e-05,
+      "loss": 1.4202,
+      "step": 100910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.04558954176493e-05,
+      "loss": 1.4943,
+      "step": 100915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.045514052105511e-05,
+      "loss": 1.2131,
+      "step": 100920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0454385621861818e-05,
+      "loss": 0.9724,
+      "step": 100925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0453630720073747e-05,
+      "loss": 1.1537,
+      "step": 100930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0452875815695196e-05,
+      "loss": 1.2728,
+      "step": 100935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.045212090873048e-05,
+      "loss": 1.1002,
+      "step": 100940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0451365999183914e-05,
+      "loss": 1.2985,
+      "step": 100945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0450611087059801e-05,
+      "loss": 1.4209,
+      "step": 100950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0449856172362461e-05,
+      "loss": 1.1394,
+      "step": 100955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0449101255096197e-05,
+      "loss": 1.3477,
+      "step": 100960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0448346335265323e-05,
+      "loss": 1.4871,
+      "step": 100965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0447591412874152e-05,
+      "loss": 1.1027,
+      "step": 100970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0446836487926992e-05,
+      "loss": 1.2102,
+      "step": 100975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0446081560428161e-05,
+      "loss": 1.3473,
+      "step": 100980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0445326630381956e-05,
+      "loss": 1.601,
+      "step": 100985
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.04445716977927e-05,
+      "loss": 1.0142,
+      "step": 100990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.04438167626647e-05,
+      "loss": 1.016,
+      "step": 100995
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0443061825002269e-05,
+      "loss": 1.9044,
+      "step": 101000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0442306884809717e-05,
+      "loss": 1.1151,
+      "step": 101005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0441551942091353e-05,
+      "loss": 1.2763,
+      "step": 101010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.044079699685149e-05,
+      "loss": 1.1671,
+      "step": 101015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.044004204909444e-05,
+      "loss": 1.415,
+      "step": 101020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0439287098824513e-05,
+      "loss": 1.1448,
+      "step": 101025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0438532146046016e-05,
+      "loss": 1.1705,
+      "step": 101030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0437777190763272e-05,
+      "loss": 1.4434,
+      "step": 101035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0437022232980583e-05,
+      "loss": 1.0675,
+      "step": 101040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0436267272702259e-05,
+      "loss": 1.2437,
+      "step": 101045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0435512309932616e-05,
+      "loss": 1.3093,
+      "step": 101050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0434757344675961e-05,
+      "loss": 1.1535,
+      "step": 101055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0434002376936611e-05,
+      "loss": 1.099,
+      "step": 101060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0433247406718872e-05,
+      "loss": 0.9378,
+      "step": 101065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.043249243402706e-05,
+      "loss": 1.226,
+      "step": 101070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0431737458865481e-05,
+      "loss": 1.3331,
+      "step": 101075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.043098248123845e-05,
+      "loss": 1.3784,
+      "step": 101080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0430227501150278e-05,
+      "loss": 1.242,
+      "step": 101085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0429472518605272e-05,
+      "loss": 1.1077,
+      "step": 101090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.042871753360775e-05,
+      "loss": 1.3605,
+      "step": 101095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.042796254616202e-05,
+      "loss": 1.2138,
+      "step": 101100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0427207556272395e-05,
+      "loss": 1.2307,
+      "step": 101105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.042645256394318e-05,
+      "loss": 0.9792,
+      "step": 101110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0425697569178696e-05,
+      "loss": 1.2898,
+      "step": 101115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.042494257198325e-05,
+      "loss": 1.1188,
+      "step": 101120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0424187572361152e-05,
+      "loss": 1.2855,
+      "step": 101125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0423432570316716e-05,
+      "loss": 1.5243,
+      "step": 101130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0422677565854251e-05,
+      "loss": 1.3514,
+      "step": 101135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.042192255897807e-05,
+      "loss": 1.0824,
+      "step": 101140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0421167549692484e-05,
+      "loss": 1.2705,
+      "step": 101145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0420412538001803e-05,
+      "loss": 1.2248,
+      "step": 101150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0419657523910344e-05,
+      "loss": 1.0335,
+      "step": 101155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0418902507422412e-05,
+      "loss": 1.5613,
+      "step": 101160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0418147488542325e-05,
+      "loss": 0.8746,
+      "step": 101165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0417392467274388e-05,
+      "loss": 1.3499,
+      "step": 101170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0416637443622914e-05,
+      "loss": 1.244,
+      "step": 101175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.041588241759222e-05,
+      "loss": 0.9743,
+      "step": 101180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.041512738918661e-05,
+      "loss": 1.2933,
+      "step": 101185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0414372358410406e-05,
+      "loss": 1.6505,
+      "step": 101190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0413617325267904e-05,
+      "loss": 1.2734,
+      "step": 101195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.041286228976343e-05,
+      "loss": 1.2753,
+      "step": 101200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0412107251901289e-05,
+      "loss": 1.3388,
+      "step": 101205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0411352211685792e-05,
+      "loss": 1.3887,
+      "step": 101210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0410597169121256e-05,
+      "loss": 1.3989,
+      "step": 101215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.040984212421199e-05,
+      "loss": 1.2672,
+      "step": 101220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.04090870769623e-05,
+      "loss": 1.4844,
+      "step": 101225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0408332027376507e-05,
+      "loss": 1.3916,
+      "step": 101230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0407576975458917e-05,
+      "loss": 1.2644,
+      "step": 101235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0406821921213844e-05,
+      "loss": 1.3062,
+      "step": 101240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0406066864645599e-05,
+      "loss": 1.4504,
+      "step": 101245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0405311805758494e-05,
+      "loss": 1.3704,
+      "step": 101250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0404556744556838e-05,
+      "loss": 1.2936,
+      "step": 101255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0403801681044947e-05,
+      "loss": 1.1986,
+      "step": 101260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.040304661522713e-05,
+      "loss": 1.1017,
+      "step": 101265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0402291547107704e-05,
+      "loss": 1.3194,
+      "step": 101270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0401536476690971e-05,
+      "loss": 1.2794,
+      "step": 101275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0400781403981255e-05,
+      "loss": 1.1158,
+      "step": 101280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.040002632898286e-05,
+      "loss": 1.3631,
+      "step": 101285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0399271251700095e-05,
+      "loss": 1.3054,
+      "step": 101290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.039851617213728e-05,
+      "loss": 1.2109,
+      "step": 101295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0397761090298724e-05,
+      "loss": 1.1426,
+      "step": 101300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0397006006188736e-05,
+      "loss": 1.346,
+      "step": 101305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0396250919811633e-05,
+      "loss": 1.2839,
+      "step": 101310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0395495831171722e-05,
+      "loss": 1.1747,
+      "step": 101315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.039474074027332e-05,
+      "loss": 1.2177,
+      "step": 101320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0393985647120732e-05,
+      "loss": 1.3477,
+      "step": 101325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0393230551718276e-05,
+      "loss": 1.1133,
+      "step": 101330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.039247545407026e-05,
+      "loss": 1.1589,
+      "step": 101335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0391720354181003e-05,
+      "loss": 1.2501,
+      "step": 101340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0390965252054808e-05,
+      "loss": 1.5631,
+      "step": 101345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0390210147695991e-05,
+      "loss": 1.4363,
+      "step": 101350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0389455041108867e-05,
+      "loss": 1.4085,
+      "step": 101355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0388699932297742e-05,
+      "loss": 1.3453,
+      "step": 101360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0387944821266933e-05,
+      "loss": 1.2964,
+      "step": 101365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0387189708020749e-05,
+      "loss": 1.1756,
+      "step": 101370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0386434592563508e-05,
+      "loss": 1.2392,
+      "step": 101375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0385679474899512e-05,
+      "loss": 1.3419,
+      "step": 101380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.038492435503308e-05,
+      "loss": 1.656,
+      "step": 101385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0384169232968526e-05,
+      "loss": 1.2433,
+      "step": 101390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0383414108710156e-05,
+      "loss": 1.0051,
+      "step": 101395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0382658982262286e-05,
+      "loss": 1.2198,
+      "step": 101400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.038190385362923e-05,
+      "loss": 1.1116,
+      "step": 101405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0381148722815294e-05,
+      "loss": 1.189,
+      "step": 101410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0380393589824795e-05,
+      "loss": 1.4164,
+      "step": 101415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0379638454662045e-05,
+      "loss": 1.0905,
+      "step": 101420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0378883317331354e-05,
+      "loss": 1.0165,
+      "step": 101425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0378128177837035e-05,
+      "loss": 0.8811,
+      "step": 101430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0377373036183402e-05,
+      "loss": 1.0196,
+      "step": 101435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0376617892374765e-05,
+      "loss": 1.4616,
+      "step": 101440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.037586274641544e-05,
+      "loss": 1.4015,
+      "step": 101445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0375107598309733e-05,
+      "loss": 1.5028,
+      "step": 101450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0374352448061961e-05,
+      "loss": 1.0827,
+      "step": 101455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0373597295676436e-05,
+      "loss": 1.0941,
+      "step": 101460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0372842141157466e-05,
+      "loss": 1.0967,
+      "step": 101465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0372086984509373e-05,
+      "loss": 1.1319,
+      "step": 101470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0371331825736457e-05,
+      "loss": 1.0479,
+      "step": 101475
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0370576664843041e-05,
+      "loss": 1.277,
+      "step": 101480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0369821501833432e-05,
+      "loss": 1.02,
+      "step": 101485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0369066336711942e-05,
+      "loss": 1.3353,
+      "step": 101490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0368311169482887e-05,
+      "loss": 1.1767,
+      "step": 101495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0367556000150574e-05,
+      "loss": 1.4012,
+      "step": 101500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0366800828719321e-05,
+      "loss": 1.2417,
+      "step": 101505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0366045655193436e-05,
+      "loss": 1.0899,
+      "step": 101510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0365290479577234e-05,
+      "loss": 1.3077,
+      "step": 101515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0364535301875032e-05,
+      "loss": 1.3138,
+      "step": 101520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.036378012209113e-05,
+      "loss": 1.0103,
+      "step": 101525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.036302494022985e-05,
+      "loss": 1.2083,
+      "step": 101530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0362269756295503e-05,
+      "loss": 1.1485,
+      "step": 101535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0361514570292402e-05,
+      "loss": 1.0371,
+      "step": 101540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0360759382224856e-05,
+      "loss": 0.9686,
+      "step": 101545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0360004192097183e-05,
+      "loss": 1.4135,
+      "step": 101550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0359248999913689e-05,
+      "loss": 1.2823,
+      "step": 101555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0358493805678691e-05,
+      "loss": 1.5597,
+      "step": 101560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0357738609396502e-05,
+      "loss": 1.6,
+      "step": 101565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0356983411071431e-05,
+      "loss": 1.1549,
+      "step": 101570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0356228210707795e-05,
+      "loss": 1.1756,
+      "step": 101575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0355473008309904e-05,
+      "loss": 1.3578,
+      "step": 101580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0354717803882068e-05,
+      "loss": 1.3248,
+      "step": 101585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0353962597428606e-05,
+      "loss": 1.1228,
+      "step": 101590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0353207388953823e-05,
+      "loss": 1.1301,
+      "step": 101595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0352452178462042e-05,
+      "loss": 0.9735,
+      "step": 101600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0351696965957564e-05,
+      "loss": 1.0692,
+      "step": 101605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.035094175144471e-05,
+      "loss": 0.9605,
+      "step": 101610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.035018653492779e-05,
+      "loss": 1.1295,
+      "step": 101615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0349431316411113e-05,
+      "loss": 1.2958,
+      "step": 101620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0348676095899003e-05,
+      "loss": 1.2179,
+      "step": 101625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0347920873395756e-05,
+      "loss": 1.1405,
+      "step": 101630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0347165648905699e-05,
+      "loss": 1.2773,
+      "step": 101635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0346410422433138e-05,
+      "loss": 1.5441,
+      "step": 101640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0345655193982387e-05,
+      "loss": 1.0838,
+      "step": 101645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0344899963557759e-05,
+      "loss": 1.0263,
+      "step": 101650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0344144731163568e-05,
+      "loss": 1.4853,
+      "step": 101655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0343389496804124e-05,
+      "loss": 1.1522,
+      "step": 101660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0342634260483742e-05,
+      "loss": 1.5993,
+      "step": 101665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0341879022206733e-05,
+      "loss": 1.5553,
+      "step": 101670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.034112378197741e-05,
+      "loss": 1.272,
+      "step": 101675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.034036853980009e-05,
+      "loss": 1.157,
+      "step": 101680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.033961329567908e-05,
+      "loss": 1.4405,
+      "step": 101685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0338858049618696e-05,
+      "loss": 1.1775,
+      "step": 101690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0338102801623252e-05,
+      "loss": 0.9667,
+      "step": 101695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0337347551697057e-05,
+      "loss": 1.3055,
+      "step": 101700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0336592299844429e-05,
+      "loss": 0.8735,
+      "step": 101705
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0335837046069673e-05,
+      "loss": 1.8462,
+      "step": 101710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.033508179037711e-05,
+      "loss": 1.2407,
+      "step": 101715
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.033432653277105e-05,
+      "loss": 1.4523,
+      "step": 101720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0333571273255805e-05,
+      "loss": 1.2101,
+      "step": 101725
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0332816011835689e-05,
+      "loss": 1.1502,
+      "step": 101730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0332060748515014e-05,
+      "loss": 0.9822,
+      "step": 101735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0331305483298093e-05,
+      "loss": 1.3112,
+      "step": 101740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0330550216189241e-05,
+      "loss": 1.1105,
+      "step": 101745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0329794947192768e-05,
+      "loss": 1.1117,
+      "step": 101750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0329039676312988e-05,
+      "loss": 1.3126,
+      "step": 101755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0328284403554216e-05,
+      "loss": 1.2096,
+      "step": 101760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0327529128920764e-05,
+      "loss": 1.3608,
+      "step": 101765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0326773852416945e-05,
+      "loss": 1.2317,
+      "step": 101770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.032601857404707e-05,
+      "loss": 1.2555,
+      "step": 101775
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0325263293815453e-05,
+      "loss": 1.2732,
+      "step": 101780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0324508011726408e-05,
+      "loss": 1.2099,
+      "step": 101785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0323752727784249e-05,
+      "loss": 1.3328,
+      "step": 101790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0322997441993285e-05,
+      "loss": 1.0619,
+      "step": 101795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0322242154357835e-05,
+      "loss": 1.3833,
+      "step": 101800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0321486864882206e-05,
+      "loss": 1.2763,
+      "step": 101805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.032073157357072e-05,
+      "loss": 1.4039,
+      "step": 101810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0319976280427678e-05,
+      "loss": 1.1246,
+      "step": 101815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0319220985457402e-05,
+      "loss": 1.2838,
+      "step": 101820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0318465688664204e-05,
+      "loss": 1.4391,
+      "step": 101825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0317710390052389e-05,
+      "loss": 1.5898,
+      "step": 101830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0316955089626282e-05,
+      "loss": 1.5533,
+      "step": 101835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0316199787390191e-05,
+      "loss": 1.1552,
+      "step": 101840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.031544448334843e-05,
+      "loss": 1.2406,
+      "step": 101845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0314689177505308e-05,
+      "loss": 1.1289,
+      "step": 101850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0313933869865142e-05,
+      "loss": 1.3136,
+      "step": 101855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.031317856043225e-05,
+      "loss": 1.1135,
+      "step": 101860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0312423249210935e-05,
+      "loss": 0.8957,
+      "step": 101865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0311667936205516e-05,
+      "loss": 1.2335,
+      "step": 101870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0310912621420303e-05,
+      "loss": 1.3908,
+      "step": 101875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0310157304859617e-05,
+      "loss": 1.434,
+      "step": 101880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0309401986527763e-05,
+      "loss": 1.1069,
+      "step": 101885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0308646666429059e-05,
+      "loss": 1.0694,
+      "step": 101890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0307891344567813e-05,
+      "loss": 0.9393,
+      "step": 101895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0307136020948343e-05,
+      "loss": 1.0447,
+      "step": 101900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0306380695574964e-05,
+      "loss": 1.3088,
+      "step": 101905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0305625368451983e-05,
+      "loss": 1.3172,
+      "step": 101910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0304870039583721e-05,
+      "loss": 1.1036,
+      "step": 101915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0304114708974481e-05,
+      "loss": 1.1966,
+      "step": 101920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0303359376628585e-05,
+      "loss": 1.0956,
+      "step": 101925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0302604042550349e-05,
+      "loss": 0.9127,
+      "step": 101930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0301848706744072e-05,
+      "loss": 1.1823,
+      "step": 101935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0301093369214084e-05,
+      "loss": 1.0544,
+      "step": 101940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0300338029964688e-05,
+      "loss": 1.1836,
+      "step": 101945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.02995826890002e-05,
+      "loss": 1.2308,
+      "step": 101950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0298827346324936e-05,
+      "loss": 1.2459,
+      "step": 101955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0298072001943204e-05,
+      "loss": 1.1086,
+      "step": 101960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0297316655859321e-05,
+      "loss": 1.1908,
+      "step": 101965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0296561308077601e-05,
+      "loss": 1.1457,
+      "step": 101970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0295805958602358e-05,
+      "loss": 1.1398,
+      "step": 101975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0295050607437902e-05,
+      "loss": 1.4981,
+      "step": 101980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.029429525458855e-05,
+      "loss": 1.1261,
+      "step": 101985
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0293539900058611e-05,
+      "loss": 1.2677,
+      "step": 101990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0292784543852403e-05,
+      "loss": 1.3274,
+      "step": 101995
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0292029185974237e-05,
+      "loss": 1.3817,
+      "step": 102000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0291273826428427e-05,
+      "loss": 1.3062,
+      "step": 102005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.029051846521929e-05,
+      "loss": 1.1744,
+      "step": 102010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0289763102351131e-05,
+      "loss": 1.1295,
+      "step": 102015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.028900773782827e-05,
+      "loss": 1.652,
+      "step": 102020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0288252371655025e-05,
+      "loss": 1.3911,
+      "step": 102025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0287497003835698e-05,
+      "loss": 1.4924,
+      "step": 102030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0286741634374612e-05,
+      "loss": 1.1877,
+      "step": 102035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0285986263276073e-05,
+      "loss": 1.3803,
+      "step": 102040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0285230890544403e-05,
+      "loss": 0.9719,
+      "step": 102045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0284475516183909e-05,
+      "loss": 1.3118,
+      "step": 102050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0283720140198907e-05,
+      "loss": 1.4672,
+      "step": 102055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.028296476259371e-05,
+      "loss": 1.1646,
+      "step": 102060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0282209383372634e-05,
+      "loss": 1.2121,
+      "step": 102065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0281454002539988e-05,
+      "loss": 1.1858,
+      "step": 102070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0280698620100089e-05,
+      "loss": 0.9159,
+      "step": 102075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0279943236057251e-05,
+      "loss": 1.2644,
+      "step": 102080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0279187850415786e-05,
+      "loss": 1.0694,
+      "step": 102085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0278432463180008e-05,
+      "loss": 1.2027,
+      "step": 102090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0277677074354231e-05,
+      "loss": 1.415,
+      "step": 102095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.027692168394277e-05,
+      "loss": 1.3633,
+      "step": 102100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0276166291949936e-05,
+      "loss": 1.312,
+      "step": 102105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0275410898380041e-05,
+      "loss": 1.9032,
+      "step": 102110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0274655503237407e-05,
+      "loss": 1.1851,
+      "step": 102115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0273900106526338e-05,
+      "loss": 1.0732,
+      "step": 102120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0273144708251153e-05,
+      "loss": 1.2389,
+      "step": 102125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0272389308416168e-05,
+      "loss": 1.4631,
+      "step": 102130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0271633907025689e-05,
+      "loss": 1.1087,
+      "step": 102135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0270878504084039e-05,
+      "loss": 1.2004,
+      "step": 102140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0270123099595521e-05,
+      "loss": 1.3124,
+      "step": 102145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0269367693564458e-05,
+      "loss": 1.0472,
+      "step": 102150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0268612285995164e-05,
+      "loss": 0.9281,
+      "step": 102155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0267856876891945e-05,
+      "loss": 1.2346,
+      "step": 102160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.026710146625912e-05,
+      "loss": 1.0999,
+      "step": 102165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0266346054101002e-05,
+      "loss": 1.1319,
+      "step": 102170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0265590640421906e-05,
+      "loss": 1.2792,
+      "step": 102175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0264835225226145e-05,
+      "loss": 1.0926,
+      "step": 102180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.026407980851803e-05,
+      "loss": 0.9244,
+      "step": 102185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0263324390301879e-05,
+      "loss": 1.0218,
+      "step": 102190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0262568970582003e-05,
+      "loss": 1.3049,
+      "step": 102195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0261813549362718e-05,
+      "loss": 1.3146,
+      "step": 102200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0261058126648333e-05,
+      "loss": 1.1408,
+      "step": 102205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0260302702443172e-05,
+      "loss": 1.1563,
+      "step": 102210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0259547276751538e-05,
+      "loss": 1.0238,
+      "step": 102215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0258791849577754e-05,
+      "loss": 1.1932,
+      "step": 102220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0258036420926126e-05,
+      "loss": 1.1919,
+      "step": 102225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0257280990800969e-05,
+      "loss": 1.0807,
+      "step": 102230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0256525559206604e-05,
+      "loss": 1.4382,
+      "step": 102235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0255770126147335e-05,
+      "loss": 1.271,
+      "step": 102240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0255014691627488e-05,
+      "loss": 1.0773,
+      "step": 102245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0254259255651363e-05,
+      "loss": 1.3228,
+      "step": 102250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0253503818223282e-05,
+      "loss": 1.3359,
+      "step": 102255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0252748379347562e-05,
+      "loss": 1.3183,
+      "step": 102260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.025199293902851e-05,
+      "loss": 1.3343,
+      "step": 102265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0251237497270443e-05,
+      "loss": 1.2335,
+      "step": 102270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0250482054077675e-05,
+      "loss": 1.6612,
+      "step": 102275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.024972660945452e-05,
+      "loss": 1.1384,
+      "step": 102280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.024897116340529e-05,
+      "loss": 1.4479,
+      "step": 102285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0248215715934303e-05,
+      "loss": 1.2478,
+      "step": 102290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.024746026704587e-05,
+      "loss": 1.581,
+      "step": 102295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0246704816744307e-05,
+      "loss": 1.2226,
+      "step": 102300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0245949365033924e-05,
+      "loss": 1.1023,
+      "step": 102305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0245193911919038e-05,
+      "loss": 1.0864,
+      "step": 102310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0244438457403967e-05,
+      "loss": 1.1937,
+      "step": 102315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0243683001493014e-05,
+      "loss": 1.1059,
+      "step": 102320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0242927544190508e-05,
+      "loss": 1.052,
+      "step": 102325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0242172085500751e-05,
+      "loss": 1.0536,
+      "step": 102330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.024141662542806e-05,
+      "loss": 1.1941,
+      "step": 102335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0240661163976753e-05,
+      "loss": 1.1696,
+      "step": 102340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0239905701151138e-05,
+      "loss": 1.3431,
+      "step": 102345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0239150236955536e-05,
+      "loss": 1.2846,
+      "step": 102350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0238394771394254e-05,
+      "loss": 1.1997,
+      "step": 102355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0237639304471612e-05,
+      "loss": 1.1123,
+      "step": 102360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0236883836191924e-05,
+      "loss": 1.1604,
+      "step": 102365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0236128366559497e-05,
+      "loss": 1.1323,
+      "step": 102370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0235372895578652e-05,
+      "loss": 1.0853,
+      "step": 102375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.02346174232537e-05,
+      "loss": 1.2375,
+      "step": 102380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.023386194958896e-05,
+      "loss": 1.315,
+      "step": 102385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0233106474588739e-05,
+      "loss": 1.2153,
+      "step": 102390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0232350998257357e-05,
+      "loss": 1.0998,
+      "step": 102395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0231595520599123e-05,
+      "loss": 1.2049,
+      "step": 102400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0230840041618355e-05,
+      "loss": 1.2811,
+      "step": 102405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.023008456131937e-05,
+      "loss": 1.0613,
+      "step": 102410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0229329079706471e-05,
+      "loss": 1.5291,
+      "step": 102415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0228573596783987e-05,
+      "loss": 1.158,
+      "step": 102420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.022781811255622e-05,
+      "loss": 1.2309,
+      "step": 102425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0227062627027491e-05,
+      "loss": 1.4437,
+      "step": 102430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0226307140202111e-05,
+      "loss": 1.2644,
+      "step": 102435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0225551652084396e-05,
+      "loss": 1.1024,
+      "step": 102440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0224796162678662e-05,
+      "loss": 1.4193,
+      "step": 102445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0224040671989216e-05,
+      "loss": 0.8683,
+      "step": 102450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0223285180020381e-05,
+      "loss": 1.4273,
+      "step": 102455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0222529686776469e-05,
+      "loss": 1.2293,
+      "step": 102460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0221774192261791e-05,
+      "loss": 0.9461,
+      "step": 102465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0221018696480664e-05,
+      "loss": 1.363,
+      "step": 102470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0220263199437399e-05,
+      "loss": 1.1033,
+      "step": 102475
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0219507701136313e-05,
+      "loss": 1.0154,
+      "step": 102480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.021875220158172e-05,
+      "loss": 1.2098,
+      "step": 102485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0217996700777937e-05,
+      "loss": 1.2802,
+      "step": 102490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0217241198729272e-05,
+      "loss": 1.0446,
+      "step": 102495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0216485695440045e-05,
+      "loss": 1.3153,
+      "step": 102500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.021573019091457e-05,
+      "loss": 1.0477,
+      "step": 102505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0214974685157157e-05,
+      "loss": 1.016,
+      "step": 102510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0214219178172125e-05,
+      "loss": 1.384,
+      "step": 102515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0213463669963781e-05,
+      "loss": 1.3987,
+      "step": 102520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.021270816053645e-05,
+      "loss": 1.0457,
+      "step": 102525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0211952649894442e-05,
+      "loss": 1.3356,
+      "step": 102530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0211197138042066e-05,
+      "loss": 1.542,
+      "step": 102535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0210441624983643e-05,
+      "loss": 0.9999,
+      "step": 102540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0209686110723483e-05,
+      "loss": 1.0742,
+      "step": 102545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0208930595265904e-05,
+      "loss": 1.4475,
+      "step": 102550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0208175078615218e-05,
+      "loss": 1.1535,
+      "step": 102555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0207419560775742e-05,
+      "loss": 1.0797,
+      "step": 102560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.020666404175179e-05,
+      "loss": 1.5287,
+      "step": 102565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0205908521547671e-05,
+      "loss": 0.9728,
+      "step": 102570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0205153000167709e-05,
+      "loss": 1.3776,
+      "step": 102575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0204397477616207e-05,
+      "loss": 1.1863,
+      "step": 102580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0203641953897489e-05,
+      "loss": 1.1614,
+      "step": 102585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0202886429015866e-05,
+      "loss": 1.6828,
+      "step": 102590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.020213090297565e-05,
+      "loss": 1.4826,
+      "step": 102595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.020137537578116e-05,
+      "loss": 1.4947,
+      "step": 102600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0200619847436708e-05,
+      "loss": 1.1057,
+      "step": 102605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0199864317946609e-05,
+      "loss": 1.146,
+      "step": 102610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0199108787315177e-05,
+      "loss": 0.9905,
+      "step": 102615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0198353255546727e-05,
+      "loss": 1.1299,
+      "step": 102620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0197597722645572e-05,
+      "loss": 1.2919,
+      "step": 102625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.019684218861603e-05,
+      "loss": 1.0524,
+      "step": 102630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.019608665346241e-05,
+      "loss": 1.1396,
+      "step": 102635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.019533111718903e-05,
+      "loss": 1.0963,
+      "step": 102640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0194575579800207e-05,
+      "loss": 1.1732,
+      "step": 102645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0193820041300251e-05,
+      "loss": 0.8852,
+      "step": 102650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.019306450169348e-05,
+      "loss": 1.2432,
+      "step": 102655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0192308960984202e-05,
+      "loss": 1.2625,
+      "step": 102660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0191553419176743e-05,
+      "loss": 1.3124,
+      "step": 102665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0190797876275409e-05,
+      "loss": 1.6657,
+      "step": 102670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0190042332284514e-05,
+      "loss": 1.0087,
+      "step": 102675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0189286787208378e-05,
+      "loss": 1.3247,
+      "step": 102680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.018853124105131e-05,
+      "loss": 1.5315,
+      "step": 102685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0187775693817627e-05,
+      "loss": 1.2452,
+      "step": 102690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0187020145511645e-05,
+      "loss": 1.0789,
+      "step": 102695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0186264596137678e-05,
+      "loss": 1.0872,
+      "step": 102700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0185509045700042e-05,
+      "loss": 1.1305,
+      "step": 102705
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0184753494203046e-05,
+      "loss": 1.503,
+      "step": 102710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.018399794165101e-05,
+      "loss": 1.1251,
+      "step": 102715
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0183242388048247e-05,
+      "loss": 1.5343,
+      "step": 102720
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.018248683339907e-05,
+      "loss": 1.1974,
+      "step": 102725
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0181731277707796e-05,
+      "loss": 1.298,
+      "step": 102730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0180975720978737e-05,
+      "loss": 1.3409,
+      "step": 102735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.018022016321621e-05,
+      "loss": 1.1484,
+      "step": 102740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0179464604424527e-05,
+      "loss": 1.1966,
+      "step": 102745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.017870904460801e-05,
+      "loss": 1.1943,
+      "step": 102750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0177953483770963e-05,
+      "loss": 1.2592,
+      "step": 102755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0177197921917712e-05,
+      "loss": 1.3521,
+      "step": 102760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0176442359052558e-05,
+      "loss": 1.1879,
+      "step": 102765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0175686795179826e-05,
+      "loss": 1.1787,
+      "step": 102770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0174931230303833e-05,
+      "loss": 1.2624,
+      "step": 102775
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0174175664428883e-05,
+      "loss": 1.173,
+      "step": 102780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0173420097559296e-05,
+      "loss": 1.2465,
+      "step": 102785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.017266452969939e-05,
+      "loss": 1.0636,
+      "step": 102790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0171908960853476e-05,
+      "loss": 1.1292,
+      "step": 102795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0171153391025867e-05,
+      "loss": 1.5035,
+      "step": 102800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0170397820220884e-05,
+      "loss": 1.023,
+      "step": 102805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0169642248442836e-05,
+      "loss": 1.3355,
+      "step": 102810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0168886675696038e-05,
+      "loss": 1.2349,
+      "step": 102815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0168131101984806e-05,
+      "loss": 1.2299,
+      "step": 102820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0167375527313457e-05,
+      "loss": 1.0573,
+      "step": 102825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0166619951686303e-05,
+      "loss": 1.393,
+      "step": 102830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0165864375107662e-05,
+      "loss": 1.0478,
+      "step": 102835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0165108797581841e-05,
+      "loss": 1.2596,
+      "step": 102840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0164353219113163e-05,
+      "loss": 1.1353,
+      "step": 102845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0163597639705938e-05,
+      "loss": 1.4061,
+      "step": 102850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0162842059364486e-05,
+      "loss": 1.1349,
+      "step": 102855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0162086478093114e-05,
+      "loss": 1.3821,
+      "step": 102860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0161330895896146e-05,
+      "loss": 0.9774,
+      "step": 102865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0160575312777889e-05,
+      "loss": 1.182,
+      "step": 102870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0159819728742659e-05,
+      "loss": 1.1773,
+      "step": 102875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0159064143794777e-05,
+      "loss": 1.2642,
+      "step": 102880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0158308557938548e-05,
+      "loss": 1.359,
+      "step": 102885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0157552971178293e-05,
+      "loss": 1.3302,
+      "step": 102890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0156797383518327e-05,
+      "loss": 1.2036,
+      "step": 102895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0156041794962963e-05,
+      "loss": 1.6334,
+      "step": 102900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0155286205516516e-05,
+      "loss": 0.9768,
+      "step": 102905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0154530615183305e-05,
+      "loss": 1.1449,
+      "step": 102910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0153775023967638e-05,
+      "loss": 1.2415,
+      "step": 102915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0153019431873834e-05,
+      "loss": 1.2388,
+      "step": 102920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0152263838906205e-05,
+      "loss": 1.0032,
+      "step": 102925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0151508245069069e-05,
+      "loss": 1.3535,
+      "step": 102930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0150752650366737e-05,
+      "loss": 1.097,
+      "step": 102935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0149997054803531e-05,
+      "loss": 1.2961,
+      "step": 102940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0149241458383758e-05,
+      "loss": 1.3803,
+      "step": 102945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0148485861111736e-05,
+      "loss": 1.2762,
+      "step": 102950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0147730262991777e-05,
+      "loss": 1.2247,
+      "step": 102955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0146974664028206e-05,
+      "loss": 1.5194,
+      "step": 102960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0146219064225324e-05,
+      "loss": 1.0713,
+      "step": 102965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0145463463587458e-05,
+      "loss": 1.0853,
+      "step": 102970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0144707862118915e-05,
+      "loss": 1.1879,
+      "step": 102975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0143952259824008e-05,
+      "loss": 1.0287,
+      "step": 102980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0143196656707062e-05,
+      "loss": 0.8608,
+      "step": 102985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0142441052772382e-05,
+      "loss": 1.4582,
+      "step": 102990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.014168544802429e-05,
+      "loss": 1.4694,
+      "step": 102995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0140929842467098e-05,
+      "loss": 1.1127,
+      "step": 103000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0140174236105119e-05,
+      "loss": 1.323,
+      "step": 103005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0139418628942672e-05,
+      "loss": 1.5813,
+      "step": 103010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0138663020984066e-05,
+      "loss": 1.4634,
+      "step": 103015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0137907412233621e-05,
+      "loss": 0.9375,
+      "step": 103020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0137151802695652e-05,
+      "loss": 1.1697,
+      "step": 103025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0136396192374471e-05,
+      "loss": 1.192,
+      "step": 103030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0135640581274396e-05,
+      "loss": 1.3702,
+      "step": 103035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0134884969399738e-05,
+      "loss": 1.0953,
+      "step": 103040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0134129356754817e-05,
+      "loss": 1.3233,
+      "step": 103045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0133373743343944e-05,
+      "loss": 1.3088,
+      "step": 103050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0132618129171435e-05,
+      "loss": 1.2238,
+      "step": 103055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0131862514241601e-05,
+      "loss": 1.0102,
+      "step": 103060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.013110689855877e-05,
+      "loss": 1.243,
+      "step": 103065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0130351282127242e-05,
+      "loss": 1.3031,
+      "step": 103070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0129595664951336e-05,
+      "loss": 1.3177,
+      "step": 103075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0128840047035377e-05,
+      "loss": 1.1836,
+      "step": 103080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0128084428383664e-05,
+      "loss": 1.1973,
+      "step": 103085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0127328809000528e-05,
+      "loss": 0.9274,
+      "step": 103090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0126573188890267e-05,
+      "loss": 1.8056,
+      "step": 103095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.012581756805721e-05,
+      "loss": 1.3089,
+      "step": 103100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0125061946505668e-05,
+      "loss": 0.9617,
+      "step": 103105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0124306324239952e-05,
+      "loss": 1.2943,
+      "step": 103110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0123550701264383e-05,
+      "loss": 1.0002,
+      "step": 103115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0122795077583272e-05,
+      "loss": 1.4523,
+      "step": 103120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0122039453200934e-05,
+      "loss": 1.1917,
+      "step": 103125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0121283828121686e-05,
+      "loss": 1.183,
+      "step": 103130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0120528202349841e-05,
+      "loss": 1.2067,
+      "step": 103135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0119772575889718e-05,
+      "loss": 1.1671,
+      "step": 103140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0119016948745625e-05,
+      "loss": 1.0329,
+      "step": 103145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0118261320921886e-05,
+      "loss": 1.1206,
+      "step": 103150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0117505692422808e-05,
+      "loss": 1.273,
+      "step": 103155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.011675006325271e-05,
+      "loss": 1.4299,
+      "step": 103160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0115994433415903e-05,
+      "loss": 1.0917,
+      "step": 103165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0115238802916712e-05,
+      "loss": 1.0235,
+      "step": 103170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0114483171759442e-05,
+      "loss": 1.3663,
+      "step": 103175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.011372753994841e-05,
+      "loss": 1.4061,
+      "step": 103180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0112971907487937e-05,
+      "loss": 1.3033,
+      "step": 103185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0112216274382327e-05,
+      "loss": 1.4914,
+      "step": 103190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.011146064063591e-05,
+      "loss": 1.4672,
+      "step": 103195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0110705006252985e-05,
+      "loss": 1.3546,
+      "step": 103200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.010994937123788e-05,
+      "loss": 1.209,
+      "step": 103205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0109193735594905e-05,
+      "loss": 1.4408,
+      "step": 103210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.010843809932837e-05,
+      "loss": 1.2304,
+      "step": 103215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0107682462442598e-05,
+      "loss": 1.4685,
+      "step": 103220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0106926824941902e-05,
+      "loss": 1.6487,
+      "step": 103225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0106171186830595e-05,
+      "loss": 1.3237,
+      "step": 103230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0105415548112995e-05,
+      "loss": 1.3283,
+      "step": 103235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0104659908793417e-05,
+      "loss": 1.161,
+      "step": 103240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0103904268876173e-05,
+      "loss": 1.2441,
+      "step": 103245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0103148628365577e-05,
+      "loss": 1.4222,
+      "step": 103250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.010239298726595e-05,
+      "loss": 0.9243,
+      "step": 103255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0101637345581604e-05,
+      "loss": 1.3716,
+      "step": 103260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0100881703316852e-05,
+      "loss": 1.3406,
+      "step": 103265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0100126060476012e-05,
+      "loss": 1.5063,
+      "step": 103270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0099370417063399e-05,
+      "loss": 1.3248,
+      "step": 103275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0098614773083329e-05,
+      "loss": 1.1985,
+      "step": 103280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0097859128540111e-05,
+      "loss": 1.4499,
+      "step": 103285
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0097103483438068e-05,
+      "loss": 1.3093,
+      "step": 103290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.009634783778151e-05,
+      "loss": 1.4821,
+      "step": 103295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.009559219157476e-05,
+      "loss": 1.2936,
+      "step": 103300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0094836544822119e-05,
+      "loss": 1.1689,
+      "step": 103305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0094080897527914e-05,
+      "loss": 1.097,
+      "step": 103310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.009332524969646e-05,
+      "loss": 0.9727,
+      "step": 103315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0092569601332063e-05,
+      "loss": 1.1819,
+      "step": 103320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0091813952439045e-05,
+      "loss": 1.2999,
+      "step": 103325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0091058303021721e-05,
+      "loss": 1.1451,
+      "step": 103330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0090302653084404e-05,
+      "loss": 1.3999,
+      "step": 103335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0089547002631412e-05,
+      "loss": 1.2471,
+      "step": 103340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0088791351667057e-05,
+      "loss": 1.2911,
+      "step": 103345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0088035700195657e-05,
+      "loss": 1.1978,
+      "step": 103350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0087280048221526e-05,
+      "loss": 1.3383,
+      "step": 103355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0086524395748978e-05,
+      "loss": 1.2839,
+      "step": 103360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0085768742782328e-05,
+      "loss": 1.3817,
+      "step": 103365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0085013089325895e-05,
+      "loss": 1.2493,
+      "step": 103370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0084257435383988e-05,
+      "loss": 1.2205,
+      "step": 103375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0083501780960929e-05,
+      "loss": 1.1866,
+      "step": 103380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0082746126061029e-05,
+      "loss": 1.4799,
+      "step": 103385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0081990470688603e-05,
+      "loss": 0.962,
+      "step": 103390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.008123481484797e-05,
+      "loss": 1.4011,
+      "step": 103395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0080479158543436e-05,
+      "loss": 1.2217,
+      "step": 103400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.007972350177933e-05,
+      "loss": 1.0813,
+      "step": 103405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.007896784455996e-05,
+      "loss": 0.9773,
+      "step": 103410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0078212186889635e-05,
+      "loss": 1.0236,
+      "step": 103415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0077456528772682e-05,
+      "loss": 1.3768,
+      "step": 103420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0076700870213408e-05,
+      "loss": 1.1954,
+      "step": 103425
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.007594521121613e-05,
+      "loss": 1.194,
+      "step": 103430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0075189551785164e-05,
+      "loss": 1.2398,
+      "step": 103435
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.007443389192483e-05,
+      "loss": 1.1416,
+      "step": 103440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0073678231639433e-05,
+      "loss": 1.3705,
+      "step": 103445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0072922570933295e-05,
+      "loss": 1.0333,
+      "step": 103450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.007216690981073e-05,
+      "loss": 1.3408,
+      "step": 103455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0071411248276055e-05,
+      "loss": 1.2535,
+      "step": 103460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0070655586333582e-05,
+      "loss": 1.5927,
+      "step": 103465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0069899923987627e-05,
+      "loss": 1.2379,
+      "step": 103470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0069144261242505e-05,
+      "loss": 1.5167,
+      "step": 103475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0068388598102535e-05,
+      "loss": 1.0569,
+      "step": 103480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0067632934572026e-05,
+      "loss": 1.3569,
+      "step": 103485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.00668772706553e-05,
+      "loss": 1.17,
+      "step": 103490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0066121606356664e-05,
+      "loss": 1.2646,
+      "step": 103495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0065365941680442e-05,
+      "loss": 1.1062,
+      "step": 103500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0064610276630944e-05,
+      "loss": 1.2315,
+      "step": 103505
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0063854611212488e-05,
+      "loss": 1.3657,
+      "step": 103510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0063098945429388e-05,
+      "loss": 1.3514,
+      "step": 103515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0062343279285956e-05,
+      "loss": 1.258,
+      "step": 103520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0061587612786514e-05,
+      "loss": 1.3626,
+      "step": 103525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.006083194593537e-05,
+      "loss": 1.3792,
+      "step": 103530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0060076278736845e-05,
+      "loss": 1.064,
+      "step": 103535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0059320611195253e-05,
+      "loss": 1.1039,
+      "step": 103540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0058564943314905e-05,
+      "loss": 1.4786,
+      "step": 103545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0057809275100123e-05,
+      "loss": 1.245,
+      "step": 103550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0057053606555219e-05,
+      "loss": 1.2808,
+      "step": 103555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0056297937684508e-05,
+      "loss": 1.0233,
+      "step": 103560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0055542268492304e-05,
+      "loss": 1.4413,
+      "step": 103565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0054786598982924e-05,
+      "loss": 1.2433,
+      "step": 103570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0054030929160683e-05,
+      "loss": 1.6173,
+      "step": 103575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0053275259029898e-05,
+      "loss": 1.2161,
+      "step": 103580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0052519588594882e-05,
+      "loss": 1.2702,
+      "step": 103585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.005176391785995e-05,
+      "loss": 1.1673,
+      "step": 103590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0051008246829419e-05,
+      "loss": 0.991,
+      "step": 103595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.00502525755076e-05,
+      "loss": 1.869,
+      "step": 103600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0049496903898816e-05,
+      "loss": 1.2104,
+      "step": 103605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0048741232007376e-05,
+      "loss": 1.1685,
+      "step": 103610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0047985559837599e-05,
+      "loss": 1.0998,
+      "step": 103615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.00472298873938e-05,
+      "loss": 1.1995,
+      "step": 103620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0046474214680289e-05,
+      "loss": 1.4098,
+      "step": 103625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0045718541701389e-05,
+      "loss": 1.3711,
+      "step": 103630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0044962868461406e-05,
+      "loss": 1.1794,
+      "step": 103635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0044207194964665e-05,
+      "loss": 1.1959,
+      "step": 103640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0043451521215477e-05,
+      "loss": 1.2397,
+      "step": 103645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0042695847218155e-05,
+      "loss": 1.1956,
+      "step": 103650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.004194017297702e-05,
+      "loss": 1.0392,
+      "step": 103655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0041184498496383e-05,
+      "loss": 1.155,
+      "step": 103660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.004042882378056e-05,
+      "loss": 1.0955,
+      "step": 103665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0039673148833866e-05,
+      "loss": 1.6204,
+      "step": 103670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0038917473660619e-05,
+      "loss": 1.123,
+      "step": 103675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0038161798265132e-05,
+      "loss": 1.2745,
+      "step": 103680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0037406122651719e-05,
+      "loss": 1.2912,
+      "step": 103685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0036650446824698e-05,
+      "loss": 1.1367,
+      "step": 103690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.003589477078838e-05,
+      "loss": 1.212,
+      "step": 103695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0035139094547088e-05,
+      "loss": 1.2193,
+      "step": 103700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.003438341810513e-05,
+      "loss": 1.159,
+      "step": 103705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0033627741466827e-05,
+      "loss": 1.237,
+      "step": 103710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.003287206463649e-05,
+      "loss": 0.9538,
+      "step": 103715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0032116387618433e-05,
+      "loss": 1.022,
+      "step": 103720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.003136071041698e-05,
+      "loss": 1.2793,
+      "step": 103725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0030605033036435e-05,
+      "loss": 1.3264,
+      "step": 103730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0029849355481125e-05,
+      "loss": 1.2654,
+      "step": 103735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0029093677755352e-05,
+      "loss": 1.0349,
+      "step": 103740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0028337999863442e-05,
+      "loss": 1.4143,
+      "step": 103745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0027582321809705e-05,
+      "loss": 1.3427,
+      "step": 103750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0026826643598462e-05,
+      "loss": 1.2262,
+      "step": 103755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0026070965234023e-05,
+      "loss": 1.2709,
+      "step": 103760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0025315286720703e-05,
+      "loss": 1.1182,
+      "step": 103765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.002455960806282e-05,
+      "loss": 1.0951,
+      "step": 103770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0023803929264688e-05,
+      "loss": 1.2014,
+      "step": 103775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0023048250330623e-05,
+      "loss": 1.6147,
+      "step": 103780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.002229257126494e-05,
+      "loss": 1.0762,
+      "step": 103785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0021536892071955e-05,
+      "loss": 1.2667,
+      "step": 103790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0020781212755983e-05,
+      "loss": 1.1327,
+      "step": 103795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0020025533321336e-05,
+      "loss": 1.2801,
+      "step": 103800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0019269853772337e-05,
+      "loss": 1.0367,
+      "step": 103805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0018514174113295e-05,
+      "loss": 1.3549,
+      "step": 103810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0017758494348526e-05,
+      "loss": 1.1446,
+      "step": 103815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0017002814482348e-05,
+      "loss": 1.2538,
+      "step": 103820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0016247134519072e-05,
+      "loss": 1.1865,
+      "step": 103825
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.001549145446302e-05,
+      "loss": 1.5338,
+      "step": 103830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0014735774318501e-05,
+      "loss": 1.2559,
+      "step": 103835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0013980094089832e-05,
+      "loss": 1.1405,
+      "step": 103840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0013224413781332e-05,
+      "loss": 1.5487,
+      "step": 103845
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0012468733397311e-05,
+      "loss": 1.3042,
+      "step": 103850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0011713052942092e-05,
+      "loss": 1.2707,
+      "step": 103855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0010957372419979e-05,
+      "loss": 1.2436,
+      "step": 103860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0010201691835295e-05,
+      "loss": 1.1071,
+      "step": 103865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0009446011192355e-05,
+      "loss": 1.3385,
+      "step": 103870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0008690330495472e-05,
+      "loss": 1.0696,
+      "step": 103875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0007934649748963e-05,
+      "loss": 1.377,
+      "step": 103880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0007178968957143e-05,
+      "loss": 1.1837,
+      "step": 103885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0006423288124328e-05,
+      "loss": 1.2446,
+      "step": 103890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0005667607254831e-05,
+      "loss": 1.2982,
+      "step": 103895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0004911926352971e-05,
+      "loss": 1.2599,
+      "step": 103900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0004156245423058e-05,
+      "loss": 1.0404,
+      "step": 103905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0003400564469414e-05,
+      "loss": 1.1325,
+      "step": 103910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.000264488349635e-05,
+      "loss": 1.2986,
+      "step": 103915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0001889202508181e-05,
+      "loss": 1.0797,
+      "step": 103920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0001133521509226e-05,
+      "loss": 1.3377,
+      "step": 103925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0000377840503794e-05,
+      "loss": 1.0853,
+      "step": 103930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.999622159496208e-06,
+      "loss": 1.2361,
+      "step": 103935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.998866478490778e-06,
+      "loss": 1.0145,
+      "step": 103940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.998110797491824e-06,
+      "loss": 0.9744,
+      "step": 103945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.997355116503653e-06,
+      "loss": 1.1361,
+      "step": 103950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.996599435530588e-06,
+      "loss": 1.3139,
+      "step": 103955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.995843754576946e-06,
+      "loss": 1.1485,
+      "step": 103960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.99508807364703e-06,
+      "loss": 1.2776,
+      "step": 103965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.99433239274517e-06,
+      "loss": 1.0677,
+      "step": 103970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.993576711875676e-06,
+      "loss": 1.0273,
+      "step": 103975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.992821031042862e-06,
+      "loss": 1.2428,
+      "step": 103980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.99206535025104e-06,
+      "loss": 1.0328,
+      "step": 103985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.991309669504532e-06,
+      "loss": 1.497,
+      "step": 103990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.99055398880765e-06,
+      "loss": 1.3679,
+      "step": 103995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.989798308164706e-06,
+      "loss": 1.1348,
+      "step": 104000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.989042627580025e-06,
+      "loss": 0.9735,
+      "step": 104005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.988286947057913e-06,
+      "loss": 1.5232,
+      "step": 104010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.987531266602689e-06,
+      "loss": 1.1303,
+      "step": 104015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.986775586218671e-06,
+      "loss": 1.2546,
+      "step": 104020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.986019905910171e-06,
+      "loss": 1.2655,
+      "step": 104025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.9852642256815e-06,
+      "loss": 1.2666,
+      "step": 104030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.984508545536981e-06,
+      "loss": 1.0694,
+      "step": 104035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.98375286548093e-06,
+      "loss": 1.2744,
+      "step": 104040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.982997185517657e-06,
+      "loss": 1.3074,
+      "step": 104045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.982241505651476e-06,
+      "loss": 0.995,
+      "step": 104050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.98148582588671e-06,
+      "loss": 1.0532,
+      "step": 104055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.980730146227668e-06,
+      "loss": 1.0083,
+      "step": 104060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.979974466678665e-06,
+      "loss": 1.1896,
+      "step": 104065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.97921878724402e-06,
+      "loss": 1.2929,
+      "step": 104070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.97846310792805e-06,
+      "loss": 0.9084,
+      "step": 104075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.977707428735062e-06,
+      "loss": 1.1034,
+      "step": 104080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.976951749669379e-06,
+      "loss": 1.3182,
+      "step": 104085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.976196070735315e-06,
+      "loss": 1.2136,
+      "step": 104090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.975440391937182e-06,
+      "loss": 1.3745,
+      "step": 104095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.9746847132793e-06,
+      "loss": 1.0335,
+      "step": 104100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.973929034765982e-06,
+      "loss": 0.9344,
+      "step": 104105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.97317335640154e-06,
+      "loss": 1.1442,
+      "step": 104110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.972417678190297e-06,
+      "loss": 1.4616,
+      "step": 104115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.971662000136561e-06,
+      "loss": 0.9336,
+      "step": 104120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.970906322244651e-06,
+      "loss": 1.5297,
+      "step": 104125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.970150644518879e-06,
+      "loss": 0.9935,
+      "step": 104130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.969394966963566e-06,
+      "loss": 1.0838,
+      "step": 104135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.968639289583026e-06,
+      "loss": 0.9969,
+      "step": 104140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.967883612381566e-06,
+      "loss": 1.1734,
+      "step": 104145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.967127935363512e-06,
+      "loss": 1.3992,
+      "step": 104150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.966372258533178e-06,
+      "loss": 1.2869,
+      "step": 104155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.96561658189487e-06,
+      "loss": 1.3068,
+      "step": 104160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.964860905452914e-06,
+      "loss": 1.2706,
+      "step": 104165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.964105229211622e-06,
+      "loss": 1.0697,
+      "step": 104170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.963349553175306e-06,
+      "loss": 1.1689,
+      "step": 104175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.962593877348285e-06,
+      "loss": 1.3775,
+      "step": 104180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.961838201734873e-06,
+      "loss": 1.4479,
+      "step": 104185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.961082526339386e-06,
+      "loss": 1.0517,
+      "step": 104190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.960326851166136e-06,
+      "loss": 1.0669,
+      "step": 104195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.959571176219442e-06,
+      "loss": 1.2163,
+      "step": 104200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.958815501503622e-06,
+      "loss": 1.1937,
+      "step": 104205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.95805982702298e-06,
+      "loss": 1.2959,
+      "step": 104210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.957304152781846e-06,
+      "loss": 1.1897,
+      "step": 104215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.956548478784526e-06,
+      "loss": 1.0032,
+      "step": 104220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.955792805035336e-06,
+      "loss": 1.4253,
+      "step": 104225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.955037131538596e-06,
+      "loss": 1.1827,
+      "step": 104230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.954281458298616e-06,
+      "loss": 1.2224,
+      "step": 104235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.953525785319713e-06,
+      "loss": 1.4101,
+      "step": 104240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.952770112606204e-06,
+      "loss": 1.2771,
+      "step": 104245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.952014440162405e-06,
+      "loss": 1.4846,
+      "step": 104250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.951258767992626e-06,
+      "loss": 1.0579,
+      "step": 104255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.950503096101186e-06,
+      "loss": 1.1483,
+      "step": 104260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.949747424492404e-06,
+      "loss": 1.0667,
+      "step": 104265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.948991753170587e-06,
+      "loss": 1.5144,
+      "step": 104270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.948236082140054e-06,
+      "loss": 1.5243,
+      "step": 104275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.947480411405122e-06,
+      "loss": 1.239,
+      "step": 104280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.946724740970107e-06,
+      "loss": 1.3247,
+      "step": 104285
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.945969070839318e-06,
+      "loss": 1.4896,
+      "step": 104290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.94521340101708e-06,
+      "loss": 1.2191,
+      "step": 104295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.944457731507703e-06,
+      "loss": 1.2259,
+      "step": 104300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.943702062315494e-06,
+      "loss": 1.1105,
+      "step": 104305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.942946393444785e-06,
+      "loss": 1.3456,
+      "step": 104310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.94219072489988e-06,
+      "loss": 1.2685,
+      "step": 104315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.941435056685093e-06,
+      "loss": 1.3202,
+      "step": 104320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.94067938880475e-06,
+      "loss": 1.1773,
+      "step": 104325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.939923721263156e-06,
+      "loss": 1.4492,
+      "step": 104330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.939168054064631e-06,
+      "loss": 1.0025,
+      "step": 104335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.93841238721349e-06,
+      "loss": 1.213,
+      "step": 104340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.937656720714046e-06,
+      "loss": 1.0644,
+      "step": 104345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.936901054570617e-06,
+      "loss": 1.0015,
+      "step": 104350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.936145388787514e-06,
+      "loss": 1.3298,
+      "step": 104355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.935389723369059e-06,
+      "loss": 1.1062,
+      "step": 104360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.934634058319561e-06,
+      "loss": 1.1937,
+      "step": 104365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.933878393643336e-06,
+      "loss": 1.4039,
+      "step": 104370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.933122729344702e-06,
+      "loss": 1.3486,
+      "step": 104375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.932367065427977e-06,
+      "loss": 1.14,
+      "step": 104380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.931611401897469e-06,
+      "loss": 1.5608,
+      "step": 104385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.930855738757497e-06,
+      "loss": 1.1571,
+      "step": 104390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.930100076012376e-06,
+      "loss": 1.3788,
+      "step": 104395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.92934441366642e-06,
+      "loss": 1.3934,
+      "step": 104400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.928588751723949e-06,
+      "loss": 1.276,
+      "step": 104405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.927833090189273e-06,
+      "loss": 1.1162,
+      "step": 104410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.92707742906671e-06,
+      "loss": 1.1276,
+      "step": 104415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.92632176836057e-06,
+      "loss": 1.4274,
+      "step": 104420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.925566108075176e-06,
+      "loss": 1.244,
+      "step": 104425
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.924810448214837e-06,
+      "loss": 1.0687,
+      "step": 104430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.924054788783872e-06,
+      "loss": 0.882,
+      "step": 104435
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.923299129786597e-06,
+      "loss": 1.3279,
+      "step": 104440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.922543471227323e-06,
+      "loss": 1.3044,
+      "step": 104445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.921787813110365e-06,
+      "loss": 0.928,
+      "step": 104450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.921032155440044e-06,
+      "loss": 1.2855,
+      "step": 104455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.920276498220673e-06,
+      "loss": 1.1142,
+      "step": 104460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.919520841456562e-06,
+      "loss": 1.1971,
+      "step": 104465
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.918765185152034e-06,
+      "loss": 1.1541,
+      "step": 104470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.9180095293114e-06,
+      "loss": 1.0535,
+      "step": 104475
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.917253873938976e-06,
+      "loss": 1.0414,
+      "step": 104480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.916498219039073e-06,
+      "loss": 1.0585,
+      "step": 104485
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.915742564616014e-06,
+      "loss": 1.0999,
+      "step": 104490
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.91498691067411e-06,
+      "loss": 1.4286,
+      "step": 104495
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.914231257217674e-06,
+      "loss": 1.2377,
+      "step": 104500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.913475604251027e-06,
+      "loss": 0.8339,
+      "step": 104505
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.91271995177848e-06,
+      "loss": 1.0947,
+      "step": 104510
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.911964299804343e-06,
+      "loss": 1.3735,
+      "step": 104515
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.911208648332946e-06,
+      "loss": 1.0048,
+      "step": 104520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.91045299736859e-06,
+      "loss": 1.2308,
+      "step": 104525
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.909697346915596e-06,
+      "loss": 1.1823,
+      "step": 104530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.908941696978282e-06,
+      "loss": 1.1762,
+      "step": 104535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.908186047560959e-06,
+      "loss": 1.1774,
+      "step": 104540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.907430398667939e-06,
+      "loss": 1.1959,
+      "step": 104545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.906674750303545e-06,
+      "loss": 1.026,
+      "step": 104550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.905919102472089e-06,
+      "loss": 1.559,
+      "step": 104555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.905163455177884e-06,
+      "loss": 1.1524,
+      "step": 104560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.904407808425244e-06,
+      "loss": 1.0762,
+      "step": 104565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.903652162218492e-06,
+      "loss": 1.1202,
+      "step": 104570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.902896516561935e-06,
+      "loss": 1.1552,
+      "step": 104575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.902140871459889e-06,
+      "loss": 1.1335,
+      "step": 104580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.901385226916674e-06,
+      "loss": 1.3707,
+      "step": 104585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.900629582936605e-06,
+      "loss": 1.2457,
+      "step": 104590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.89987393952399e-06,
+      "loss": 1.2605,
+      "step": 104595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.89911829668315e-06,
+      "loss": 1.109,
+      "step": 104600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.898362654418401e-06,
+      "loss": 1.1099,
+      "step": 104605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.89760701273405e-06,
+      "loss": 1.4782,
+      "step": 104610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.896851371634424e-06,
+      "loss": 1.3009,
+      "step": 104615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.896095731123832e-06,
+      "loss": 1.4985,
+      "step": 104620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.89534009120659e-06,
+      "loss": 1.0841,
+      "step": 104625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.894584451887007e-06,
+      "loss": 1.1088,
+      "step": 104630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.893828813169407e-06,
+      "loss": 1.1101,
+      "step": 104635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.893073175058101e-06,
+      "loss": 1.6129,
+      "step": 104640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.892317537557402e-06,
+      "loss": 1.379,
+      "step": 104645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.891561900671632e-06,
+      "loss": 1.0093,
+      "step": 104650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.8908062644051e-06,
+      "loss": 1.0664,
+      "step": 104655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.890050628762122e-06,
+      "loss": 1.1904,
+      "step": 104660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.889294993747018e-06,
+      "loss": 1.2551,
+      "step": 104665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.888539359364096e-06,
+      "loss": 1.3983,
+      "step": 104670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.887783725617673e-06,
+      "loss": 1.2655,
+      "step": 104675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.887028092512066e-06,
+      "loss": 1.0499,
+      "step": 104680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.886272460051594e-06,
+      "loss": 1.0566,
+      "step": 104685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.88551682824056e-06,
+      "loss": 1.2302,
+      "step": 104690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.884761197083292e-06,
+      "loss": 1.3349,
+      "step": 104695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.884005566584098e-06,
+      "loss": 1.3024,
+      "step": 104700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.883249936747295e-06,
+      "loss": 1.3634,
+      "step": 104705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.882494307577195e-06,
+      "loss": 1.2009,
+      "step": 104710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.881738679078118e-06,
+      "loss": 1.149,
+      "step": 104715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.880983051254378e-06,
+      "loss": 1.3669,
+      "step": 104720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.880227424110286e-06,
+      "loss": 1.2588,
+      "step": 104725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.87947179765016e-06,
+      "loss": 1.1153,
+      "step": 104730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.878716171878316e-06,
+      "loss": 1.4409,
+      "step": 104735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.877960546799066e-06,
+      "loss": 1.3668,
+      "step": 104740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.877204922416732e-06,
+      "loss": 1.7003,
+      "step": 104745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.87644929873562e-06,
+      "loss": 1.3045,
+      "step": 104750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.875693675760048e-06,
+      "loss": 1.2162,
+      "step": 104755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.874938053494335e-06,
+      "loss": 1.1915,
+      "step": 104760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.874182431942793e-06,
+      "loss": 1.0309,
+      "step": 104765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.873426811109735e-06,
+      "loss": 1.0272,
+      "step": 104770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.872671190999477e-06,
+      "loss": 1.3663,
+      "step": 104775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.871915571616338e-06,
+      "loss": 1.0586,
+      "step": 104780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.871159952964628e-06,
+      "loss": 1.3518,
+      "step": 104785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.870404335048663e-06,
+      "loss": 1.2441,
+      "step": 104790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.869648717872761e-06,
+      "loss": 1.0235,
+      "step": 104795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.868893101441235e-06,
+      "loss": 1.1192,
+      "step": 104800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.868137485758399e-06,
+      "loss": 1.0326,
+      "step": 104805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.867381870828569e-06,
+      "loss": 1.2607,
+      "step": 104810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.866626256656061e-06,
+      "loss": 1.1381,
+      "step": 104815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.865870643245187e-06,
+      "loss": 1.1768,
+      "step": 104820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.865115030600264e-06,
+      "loss": 1.1793,
+      "step": 104825
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.864359418725607e-06,
+      "loss": 1.4861,
+      "step": 104830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.863603807625529e-06,
+      "loss": 1.1211,
+      "step": 104835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.862848197304351e-06,
+      "loss": 1.2349,
+      "step": 104840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.86209258776638e-06,
+      "loss": 1.3816,
+      "step": 104845
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.861336979015939e-06,
+      "loss": 0.9506,
+      "step": 104850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.860581371057331e-06,
+      "loss": 1.2759,
+      "step": 104855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.859825763894885e-06,
+      "loss": 1.0621,
+      "step": 104860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.859070157532907e-06,
+      "loss": 1.4136,
+      "step": 104865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.85831455197571e-06,
+      "loss": 1.2098,
+      "step": 104870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.85755894722762e-06,
+      "loss": 1.377,
+      "step": 104875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.856803343292941e-06,
+      "loss": 1.0434,
+      "step": 104880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.856047740175992e-06,
+      "loss": 0.8294,
+      "step": 104885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.855292137881089e-06,
+      "loss": 1.0743,
+      "step": 104890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.854536536412547e-06,
+      "loss": 1.0942,
+      "step": 104895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.853780935774677e-06,
+      "loss": 1.0906,
+      "step": 104900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.853025335971797e-06,
+      "loss": 1.5423,
+      "step": 104905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.852269737008225e-06,
+      "loss": 1.1246,
+      "step": 104910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.851514138888269e-06,
+      "loss": 1.225,
+      "step": 104915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.850758541616245e-06,
+      "loss": 1.2101,
+      "step": 104920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.850002945196472e-06,
+      "loss": 1.1835,
+      "step": 104925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.849247349633267e-06,
+      "loss": 1.4372,
+      "step": 104930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.848491754930934e-06,
+      "loss": 1.174,
+      "step": 104935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.847736161093799e-06,
+      "loss": 1.9801,
+      "step": 104940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.846980568126173e-06,
+      "loss": 1.149,
+      "step": 104945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.846224976032363e-06,
+      "loss": 1.0794,
+      "step": 104950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.845469384816698e-06,
+      "loss": 1.4285,
+      "step": 104955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.844713794483485e-06,
+      "loss": 1.2174,
+      "step": 104960
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.843958205037037e-06,
+      "loss": 1.1089,
+      "step": 104965
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.843202616481675e-06,
+      "loss": 1.6193,
+      "step": 104970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.84244702882171e-06,
+      "loss": 1.2,
+      "step": 104975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.841691442061453e-06,
+      "loss": 1.1288,
+      "step": 104980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.840935856205226e-06,
+      "loss": 1.0645,
+      "step": 104985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.840180271257344e-06,
+      "loss": 1.3578,
+      "step": 104990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.839424687222116e-06,
+      "loss": 1.0358,
+      "step": 104995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.838669104103857e-06,
+      "loss": 0.9864,
+      "step": 105000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.837913521906889e-06,
+      "loss": 1.0363,
+      "step": 105005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.83715794063552e-06,
+      "loss": 1.4886,
+      "step": 105010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.836402360294063e-06,
+      "loss": 1.0577,
+      "step": 105015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.835646780886839e-06,
+      "loss": 1.1297,
+      "step": 105020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.834891202418164e-06,
+      "loss": 0.9861,
+      "step": 105025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.834135624892343e-06,
+      "loss": 1.2032,
+      "step": 105030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.833380048313699e-06,
+      "loss": 1.4591,
+      "step": 105035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.832624472686545e-06,
+      "loss": 1.2188,
+      "step": 105040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.831868898015194e-06,
+      "loss": 1.1154,
+      "step": 105045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.831113324303964e-06,
+      "loss": 1.0913,
+      "step": 105050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.830357751557168e-06,
+      "loss": 1.4116,
+      "step": 105055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.829602179779121e-06,
+      "loss": 1.3349,
+      "step": 105060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.828846608974135e-06,
+      "loss": 1.0128,
+      "step": 105065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.828091039146528e-06,
+      "loss": 1.0424,
+      "step": 105070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.827335470300613e-06,
+      "loss": 1.1479,
+      "step": 105075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.826579902440704e-06,
+      "loss": 1.5626,
+      "step": 105080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.82582433557112e-06,
+      "loss": 1.2035,
+      "step": 105085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.825068769696172e-06,
+      "loss": 1.1135,
+      "step": 105090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.824313204820172e-06,
+      "loss": 1.279,
+      "step": 105095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.823557640947444e-06,
+      "loss": 1.2395,
+      "step": 105100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.822802078082295e-06,
+      "loss": 1.2263,
+      "step": 105105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.822046516229037e-06,
+      "loss": 1.0013,
+      "step": 105110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.821290955391991e-06,
+      "loss": 1.435,
+      "step": 105115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.820535395575475e-06,
+      "loss": 1.0419,
+      "step": 105120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.819779836783792e-06,
+      "loss": 1.3834,
+      "step": 105125
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.819024279021265e-06,
+      "loss": 1.3449,
+      "step": 105130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.818268722292207e-06,
+      "loss": 1.5406,
+      "step": 105135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.817513166600936e-06,
+      "loss": 1.13,
+      "step": 105140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.816757611951758e-06,
+      "loss": 1.38,
+      "step": 105145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.816002058348994e-06,
+      "loss": 1.2635,
+      "step": 105150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.815246505796959e-06,
+      "loss": 1.6985,
+      "step": 105155
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.814490954299963e-06,
+      "loss": 1.3601,
+      "step": 105160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.813735403862324e-06,
+      "loss": 1.3567,
+      "step": 105165
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.812979854488357e-06,
+      "loss": 1.1848,
+      "step": 105170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.812224306182373e-06,
+      "loss": 1.1302,
+      "step": 105175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.811468758948693e-06,
+      "loss": 1.248,
+      "step": 105180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.810713212791627e-06,
+      "loss": 1.1423,
+      "step": 105185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.809957667715488e-06,
+      "loss": 1.4337,
+      "step": 105190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.809202123724595e-06,
+      "loss": 1.1623,
+      "step": 105195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.808446580823262e-06,
+      "loss": 1.4906,
+      "step": 105200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.807691039015796e-06,
+      "loss": 1.1336,
+      "step": 105205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.806935498306522e-06,
+      "loss": 1.0756,
+      "step": 105210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.806179958699752e-06,
+      "loss": 1.2771,
+      "step": 105215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.805424420199798e-06,
+      "loss": 1.02,
+      "step": 105220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.80466888281097e-06,
+      "loss": 1.3312,
+      "step": 105225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.803913346537593e-06,
+      "loss": 1.4565,
+      "step": 105230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.803157811383977e-06,
+      "loss": 1.2077,
+      "step": 105235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.80240227735443e-06,
+      "loss": 1.1032,
+      "step": 105240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.801646744453277e-06,
+      "loss": 0.9231,
+      "step": 105245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.800891212684828e-06,
+      "loss": 0.9621,
+      "step": 105250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.800135682053393e-06,
+      "loss": 1.1241,
+      "step": 105255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.799380152563295e-06,
+      "loss": 1.2196,
+      "step": 105260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.798624624218842e-06,
+      "loss": 1.303,
+      "step": 105265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.79786909702435e-06,
+      "loss": 1.1902,
+      "step": 105270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.797113570984138e-06,
+      "loss": 1.2714,
+      "step": 105275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.796358046102514e-06,
+      "loss": 1.1414,
+      "step": 105280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.795602522383798e-06,
+      "loss": 1.45,
+      "step": 105285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.794846999832294e-06,
+      "loss": 1.6661,
+      "step": 105290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.79409147845233e-06,
+      "loss": 1.0939,
+      "step": 105295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.793335958248215e-06,
+      "loss": 1.1745,
+      "step": 105300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.792580439224258e-06,
+      "loss": 1.2333,
+      "step": 105305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.791824921384785e-06,
+      "loss": 1.1549,
+      "step": 105310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.7910694047341e-06,
+      "loss": 1.1602,
+      "step": 105315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.790313889276519e-06,
+      "loss": 1.254,
+      "step": 105320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.78955837501636e-06,
+      "loss": 1.1571,
+      "step": 105325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.78880286195794e-06,
+      "loss": 1.3549,
+      "step": 105330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.788047350105563e-06,
+      "loss": 1.1752,
+      "step": 105335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.787291839463552e-06,
+      "loss": 1.0676,
+      "step": 105340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.786536330036222e-06,
+      "loss": 1.064,
+      "step": 105345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.785780821827877e-06,
+      "loss": 1.3608,
+      "step": 105350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.785025314842845e-06,
+      "loss": 1.3875,
+      "step": 105355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.784269809085433e-06,
+      "loss": 1.1954,
+      "step": 105360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.78351430455996e-06,
+      "loss": 1.1813,
+      "step": 105365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.78275880127073e-06,
+      "loss": 1.2883,
+      "step": 105370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.782003299222067e-06,
+      "loss": 1.0355,
+      "step": 105375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.781247798418281e-06,
+      "loss": 1.356,
+      "step": 105380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.780492298863687e-06,
+      "loss": 1.2359,
+      "step": 105385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.779736800562603e-06,
+      "loss": 1.3949,
+      "step": 105390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.77898130351934e-06,
+      "loss": 1.1084,
+      "step": 105395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.77822580773821e-06,
+      "loss": 1.2741,
+      "step": 105400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.777470313223535e-06,
+      "loss": 1.3972,
+      "step": 105405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.77671481997962e-06,
+      "loss": 1.2779,
+      "step": 105410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.775959328010784e-06,
+      "loss": 1.2732,
+      "step": 105415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.77520383732134e-06,
+      "loss": 1.1137,
+      "step": 105420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.774448347915607e-06,
+      "loss": 1.3061,
+      "step": 105425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.773692859797892e-06,
+      "loss": 1.0841,
+      "step": 105430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.77293737297251e-06,
+      "loss": 1.1182,
+      "step": 105435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.772181887443782e-06,
+      "loss": 1.3051,
+      "step": 105440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.771426403216018e-06,
+      "loss": 0.7586,
+      "step": 105445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.77067092029353e-06,
+      "loss": 1.2065,
+      "step": 105450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.769915438680636e-06,
+      "loss": 1.2485,
+      "step": 105455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.76915995838165e-06,
+      "loss": 1.1912,
+      "step": 105460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.768404479400879e-06,
+      "loss": 1.19,
+      "step": 105465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.767649001742647e-06,
+      "loss": 1.0014,
+      "step": 105470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.766893525411265e-06,
+      "loss": 1.2932,
+      "step": 105475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.766138050411042e-06,
+      "loss": 0.9263,
+      "step": 105480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.765382576746301e-06,
+      "loss": 1.2224,
+      "step": 105485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.764627104421351e-06,
+      "loss": 1.1959,
+      "step": 105490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.763871633440505e-06,
+      "loss": 1.1614,
+      "step": 105495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.76311616380808e-06,
+      "loss": 1.1251,
+      "step": 105500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.762360695528391e-06,
+      "loss": 1.1259,
+      "step": 105505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.761605228605748e-06,
+      "loss": 1.2583,
+      "step": 105510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.760849763044466e-06,
+      "loss": 1.2496,
+      "step": 105515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.760094298848865e-06,
+      "loss": 1.3951,
+      "step": 105520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.759338836023252e-06,
+      "loss": 1.2569,
+      "step": 105525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.758583374571942e-06,
+      "loss": 1.0723,
+      "step": 105530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.75782791449925e-06,
+      "loss": 0.9873,
+      "step": 105535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.757072455809498e-06,
+      "loss": 1.1395,
+      "step": 105540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.756316998506986e-06,
+      "loss": 1.0711,
+      "step": 105545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.755561542596037e-06,
+      "loss": 1.4992,
+      "step": 105550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.754806088080966e-06,
+      "loss": 1.3915,
+      "step": 105555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.754050634966076e-06,
+      "loss": 1.2266,
+      "step": 105560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.753295183255697e-06,
+      "loss": 1.1908,
+      "step": 105565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.752539732954133e-06,
+      "loss": 1.742,
+      "step": 105570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.751784284065702e-06,
+      "loss": 1.9732,
+      "step": 105575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.751028836594713e-06,
+      "loss": 1.0829,
+      "step": 105580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.750273390545484e-06,
+      "loss": 1.3763,
+      "step": 105585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.749517945922331e-06,
+      "loss": 1.0906,
+      "step": 105590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.748762502729558e-06,
+      "loss": 1.0886,
+      "step": 105595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.748007060971495e-06,
+      "loss": 1.4028,
+      "step": 105600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.747251620652443e-06,
+      "loss": 0.8945,
+      "step": 105605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.746496181776718e-06,
+      "loss": 1.2952,
+      "step": 105610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.74574074434864e-06,
+      "loss": 1.0934,
+      "step": 105615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.744985308372519e-06,
+      "loss": 1.2355,
+      "step": 105620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.744229873852665e-06,
+      "loss": 1.0599,
+      "step": 105625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.7434744407934e-06,
+      "loss": 1.3112,
+      "step": 105630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.742719009199035e-06,
+      "loss": 1.1657,
+      "step": 105635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.741963579073878e-06,
+      "loss": 1.2651,
+      "step": 105640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.74120815042225e-06,
+      "loss": 1.0136,
+      "step": 105645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.740452723248465e-06,
+      "loss": 1.3388,
+      "step": 105650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.739697297556833e-06,
+      "loss": 1.2048,
+      "step": 105655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.738941873351667e-06,
+      "loss": 1.3944,
+      "step": 105660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.738186450637286e-06,
+      "loss": 1.1203,
+      "step": 105665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.737431029418002e-06,
+      "loss": 1.3928,
+      "step": 105670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.736675609698125e-06,
+      "loss": 1.6138,
+      "step": 105675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.735920191481973e-06,
+      "loss": 1.2441,
+      "step": 105680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.73516477477386e-06,
+      "loss": 1.1794,
+      "step": 105685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.734409359578096e-06,
+      "loss": 1.4346,
+      "step": 105690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.733653945899e-06,
+      "loss": 1.29,
+      "step": 105695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.732898533740884e-06,
+      "loss": 1.3836,
+      "step": 105700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.732143123108057e-06,
+      "loss": 1.1338,
+      "step": 105705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.731387714004841e-06,
+      "loss": 1.1313,
+      "step": 105710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.730632306435543e-06,
+      "loss": 1.1316,
+      "step": 105715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.72987690040448e-06,
+      "loss": 1.1387,
+      "step": 105720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.729121495915965e-06,
+      "loss": 1.3109,
+      "step": 105725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.728366092974314e-06,
+      "loss": 0.9221,
+      "step": 105730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.727610691583837e-06,
+      "loss": 1.0604,
+      "step": 105735
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.726855291748847e-06,
+      "loss": 1.1011,
+      "step": 105740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.726099893473665e-06,
+      "loss": 1.2947,
+      "step": 105745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.7253444967626e-06,
+      "loss": 1.2523,
+      "step": 105750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.72458910161996e-06,
+      "loss": 1.33,
+      "step": 105755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.723833708050067e-06,
+      "loss": 1.2199,
+      "step": 105760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.723078316057236e-06,
+      "loss": 1.2685,
+      "step": 105765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.72232292564577e-06,
+      "loss": 0.9414,
+      "step": 105770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.721567536819995e-06,
+      "loss": 1.0822,
+      "step": 105775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.720812149584217e-06,
+      "loss": 1.3049,
+      "step": 105780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.720056763942749e-06,
+      "loss": 1.2802,
+      "step": 105785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.719301379899913e-06,
+      "loss": 1.3602,
+      "step": 105790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.718545997460014e-06,
+      "loss": 1.3027,
+      "step": 105795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.717790616627371e-06,
+      "loss": 1.3474,
+      "step": 105800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.717035237406292e-06,
+      "loss": 1.3125,
+      "step": 105805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.716279859801096e-06,
+      "loss": 1.1131,
+      "step": 105810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.715524483816095e-06,
+      "loss": 1.2184,
+      "step": 105815
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.714769109455599e-06,
+      "loss": 1.2617,
+      "step": 105820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.714013736723929e-06,
+      "loss": 0.8975,
+      "step": 105825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.713258365625393e-06,
+      "loss": 1.7279,
+      "step": 105830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.712502996164303e-06,
+      "loss": 1.3478,
+      "step": 105835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.711747628344977e-06,
+      "loss": 1.129,
+      "step": 105840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.71099226217173e-06,
+      "loss": 1.7904,
+      "step": 105845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.710236897648869e-06,
+      "loss": 1.1288,
+      "step": 105850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.709481534780713e-06,
+      "loss": 1.0199,
+      "step": 105855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.708726173571576e-06,
+      "loss": 1.2965,
+      "step": 105860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.707970814025766e-06,
+      "loss": 1.3707,
+      "step": 105865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.7072154561476e-06,
+      "loss": 1.3476,
+      "step": 105870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.706460099941392e-06,
+      "loss": 1.1281,
+      "step": 105875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.705704745411455e-06,
+      "loss": 1.003,
+      "step": 105880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.7049493925621e-06,
+      "loss": 1.5313,
+      "step": 105885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.704194041397644e-06,
+      "loss": 1.3138,
+      "step": 105890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.703438691922402e-06,
+      "loss": 1.0373,
+      "step": 105895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.702683344140677e-06,
+      "loss": 1.4466,
+      "step": 105900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.701927998056799e-06,
+      "loss": 1.2542,
+      "step": 105905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.701172653675068e-06,
+      "loss": 1.3695,
+      "step": 105910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.7004173109998e-06,
+      "loss": 1.5696,
+      "step": 105915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.699661970035314e-06,
+      "loss": 1.2101,
+      "step": 105920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.69890663078592e-06,
+      "loss": 1.3844,
+      "step": 105925
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.698151293255926e-06,
+      "loss": 1.5554,
+      "step": 105930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.697395957449654e-06,
+      "loss": 1.2015,
+      "step": 105935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.696640623371416e-06,
+      "loss": 1.1397,
+      "step": 105940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.69588529102552e-06,
+      "loss": 1.3549,
+      "step": 105945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.695129960416284e-06,
+      "loss": 1.462,
+      "step": 105950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.69437463154802e-06,
+      "loss": 1.1691,
+      "step": 105955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.69361930442504e-06,
+      "loss": 1.4023,
+      "step": 105960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.692863979051657e-06,
+      "loss": 1.1672,
+      "step": 105965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.692108655432189e-06,
+      "loss": 1.1824,
+      "step": 105970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.691353333570948e-06,
+      "loss": 1.3038,
+      "step": 105975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.69059801347224e-06,
+      "loss": 1.237,
+      "step": 105980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.689842695140387e-06,
+      "loss": 1.0531,
+      "step": 105985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.6890873785797e-06,
+      "loss": 1.3257,
+      "step": 105990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.688332063794486e-06,
+      "loss": 1.4478,
+      "step": 105995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.687576750789069e-06,
+      "loss": 0.9838,
+      "step": 106000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.686821439567756e-06,
+      "loss": 1.4188,
+      "step": 106005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.686066130134861e-06,
+      "loss": 1.6089,
+      "step": 106010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.685310822494695e-06,
+      "loss": 1.1746,
+      "step": 106015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.684555516651575e-06,
+      "loss": 1.283,
+      "step": 106020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.683800212609812e-06,
+      "loss": 1.1526,
+      "step": 106025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.683044910373718e-06,
+      "loss": 1.3838,
+      "step": 106030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.682289609947613e-06,
+      "loss": 1.0766,
+      "step": 106035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.681534311335803e-06,
+      "loss": 0.9268,
+      "step": 106040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.6807790145426e-06,
+      "loss": 1.2492,
+      "step": 106045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.680023719572326e-06,
+      "loss": 1.2342,
+      "step": 106050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.679268426429288e-06,
+      "loss": 1.0736,
+      "step": 106055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.678513135117794e-06,
+      "loss": 1.257,
+      "step": 106060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.677757845642168e-06,
+      "loss": 1.3472,
+      "step": 106065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.677002558006718e-06,
+      "loss": 1.4622,
+      "step": 106070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.676247272215755e-06,
+      "loss": 1.0216,
+      "step": 106075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.675491988273593e-06,
+      "loss": 1.3666,
+      "step": 106080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.67473670618455e-06,
+      "loss": 0.8783,
+      "step": 106085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.673981425952936e-06,
+      "loss": 1.2115,
+      "step": 106090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.673226147583059e-06,
+      "loss": 1.0798,
+      "step": 106095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.672470871079237e-06,
+      "loss": 1.4123,
+      "step": 106100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.671715596445787e-06,
+      "loss": 1.3643,
+      "step": 106105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.670960323687014e-06,
+      "loss": 0.9932,
+      "step": 106110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.670205052807234e-06,
+      "loss": 1.0762,
+      "step": 106115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.669449783810762e-06,
+      "loss": 1.1991,
+      "step": 106120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.668694516701907e-06,
+      "loss": 1.0893,
+      "step": 106125
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.66793925148499e-06,
+      "loss": 1.5148,
+      "step": 106130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.667183988164315e-06,
+      "loss": 1.1426,
+      "step": 106135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.666428726744197e-06,
+      "loss": 1.0766,
+      "step": 106140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.665673467228951e-06,
+      "loss": 1.1233,
+      "step": 106145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.664918209622893e-06,
+      "loss": 1.1386,
+      "step": 106150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.664162953930329e-06,
+      "loss": 1.2038,
+      "step": 106155
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.663407700155575e-06,
+      "loss": 1.3149,
+      "step": 106160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.662652448302947e-06,
+      "loss": 1.0348,
+      "step": 106165
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.661897198376753e-06,
+      "loss": 0.8522,
+      "step": 106170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.661141950381306e-06,
+      "loss": 1.3015,
+      "step": 106175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.660386704320921e-06,
+      "loss": 1.3403,
+      "step": 106180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.659631460199913e-06,
+      "loss": 1.1524,
+      "step": 106185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.65887621802259e-06,
+      "loss": 1.0233,
+      "step": 106190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.65812097779327e-06,
+      "loss": 1.5059,
+      "step": 106195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.657365739516263e-06,
+      "loss": 1.389,
+      "step": 106200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.656610503195876e-06,
+      "loss": 1.1299,
+      "step": 106205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.655855268836437e-06,
+      "loss": 1.2968,
+      "step": 106210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.655100036442242e-06,
+      "loss": 1.3933,
+      "step": 106215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.654344806017613e-06,
+      "loss": 1.2755,
+      "step": 106220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.653589577566866e-06,
+      "loss": 1.1502,
+      "step": 106225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.652834351094304e-06,
+      "loss": 1.1965,
+      "step": 106230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.652079126604247e-06,
+      "loss": 1.3076,
+      "step": 106235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.651323904101e-06,
+      "loss": 1.2952,
+      "step": 106240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.650568683588888e-06,
+      "loss": 1.0346,
+      "step": 106245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.649813465072214e-06,
+      "loss": 1.1533,
+      "step": 106250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.64905824855529e-06,
+      "loss": 1.296,
+      "step": 106255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.648303034042439e-06,
+      "loss": 1.4819,
+      "step": 106260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.647547821537963e-06,
+      "loss": 1.1131,
+      "step": 106265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.646792611046177e-06,
+      "loss": 1.4404,
+      "step": 106270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.646037402571397e-06,
+      "loss": 1.3431,
+      "step": 106275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.645282196117936e-06,
+      "loss": 1.3968,
+      "step": 106280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.6445269916901e-06,
+      "loss": 1.0135,
+      "step": 106285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.643771789292209e-06,
+      "loss": 1.4271,
+      "step": 106290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.643016588928574e-06,
+      "loss": 1.0018,
+      "step": 106295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.642261390603503e-06,
+      "loss": 1.2805,
+      "step": 106300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.641506194321312e-06,
+      "loss": 1.3923,
+      "step": 106305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.640751000086313e-06,
+      "loss": 1.1427,
+      "step": 106310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.639995807902823e-06,
+      "loss": 1.0086,
+      "step": 106315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.639240617775146e-06,
+      "loss": 1.081,
+      "step": 106320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.638485429707602e-06,
+      "loss": 1.2764,
+      "step": 106325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.637730243704502e-06,
+      "loss": 1.1198,
+      "step": 106330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.63697505977015e-06,
+      "loss": 1.2936,
+      "step": 106335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.636219877908873e-06,
+      "loss": 1.3639,
+      "step": 106340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.635464698124975e-06,
+      "loss": 1.2827,
+      "step": 106345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.634709520422766e-06,
+      "loss": 1.1581,
+      "step": 106350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.633954344806566e-06,
+      "loss": 1.1808,
+      "step": 106355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.633199171280682e-06,
+      "loss": 1.3005,
+      "step": 106360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.632443999849426e-06,
+      "loss": 1.1662,
+      "step": 106365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.631688830517114e-06,
+      "loss": 1.6446,
+      "step": 106370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.630933663288062e-06,
+      "loss": 1.202,
+      "step": 106375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.630178498166573e-06,
+      "loss": 1.3348,
+      "step": 106380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.62942333515696e-06,
+      "loss": 1.3583,
+      "step": 106385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.628668174263545e-06,
+      "loss": 1.3267,
+      "step": 106390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.627913015490632e-06,
+      "loss": 0.9566,
+      "step": 106395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.627157858842534e-06,
+      "loss": 1.0981,
+      "step": 106400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.626402704323566e-06,
+      "loss": 1.1547,
+      "step": 106405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.625647551938042e-06,
+      "loss": 1.2735,
+      "step": 106410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.62489240169027e-06,
+      "loss": 1.0775,
+      "step": 106415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.624137253584564e-06,
+      "loss": 0.881,
+      "step": 106420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.623382107625237e-06,
+      "loss": 1.4191,
+      "step": 106425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.622626963816598e-06,
+      "loss": 1.0772,
+      "step": 106430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.621871822162966e-06,
+      "loss": 1.0277,
+      "step": 106435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.621116682668649e-06,
+      "loss": 1.3771,
+      "step": 106440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.62036154533796e-06,
+      "loss": 1.2761,
+      "step": 106445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.619606410175207e-06,
+      "loss": 1.2927,
+      "step": 106450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.618851277184708e-06,
+      "loss": 1.1804,
+      "step": 106455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.618096146370774e-06,
+      "loss": 1.5561,
+      "step": 106460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.617341017737714e-06,
+      "loss": 1.0009,
+      "step": 106465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.616585891289847e-06,
+      "loss": 1.2223,
+      "step": 106470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.615830767031477e-06,
+      "loss": 1.3632,
+      "step": 106475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.615075644966921e-06,
+      "loss": 1.157,
+      "step": 106480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.61432052510049e-06,
+      "loss": 1.2093,
+      "step": 106485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.613565407436498e-06,
+      "loss": 1.3449,
+      "step": 106490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.612810291979251e-06,
+      "loss": 1.1221,
+      "step": 106495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.612055178733069e-06,
+      "loss": 1.0717,
+      "step": 106500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.611300067702262e-06,
+      "loss": 0.7893,
+      "step": 106505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.610544958891137e-06,
+      "loss": 1.1677,
+      "step": 106510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.609789852304012e-06,
+      "loss": 1.2604,
+      "step": 106515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.609034747945195e-06,
+      "loss": 1.2396,
+      "step": 106520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.608279645819004e-06,
+      "loss": 1.4315,
+      "step": 106525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.607524545929741e-06,
+      "loss": 0.9327,
+      "step": 106530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.606769448281726e-06,
+      "loss": 1.0926,
+      "step": 106535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.606014352879273e-06,
+      "loss": 1.1016,
+      "step": 106540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.605259259726682e-06,
+      "loss": 1.2864,
+      "step": 106545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.60450416882828e-06,
+      "loss": 1.2086,
+      "step": 106550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.603749080188368e-06,
+      "loss": 1.2795,
+      "step": 106555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.602993993811262e-06,
+      "loss": 1.4596,
+      "step": 106560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.60223890970128e-06,
+      "loss": 1.265,
+      "step": 106565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.601483827862723e-06,
+      "loss": 0.9952,
+      "step": 106570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.600728748299905e-06,
+      "loss": 1.48,
+      "step": 106575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.599973671017143e-06,
+      "loss": 1.2465,
+      "step": 106580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.599218596018749e-06,
+      "loss": 1.673,
+      "step": 106585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.59846352330903e-06,
+      "loss": 1.4759,
+      "step": 106590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.597708452892298e-06,
+      "loss": 1.1652,
+      "step": 106595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.596953384772873e-06,
+      "loss": 1.1993,
+      "step": 106600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.596198318955056e-06,
+      "loss": 1.2661,
+      "step": 106605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.595443255443164e-06,
+      "loss": 1.163,
+      "step": 106610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.594688194241511e-06,
+      "loss": 1.3699,
+      "step": 106615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.593933135354406e-06,
+      "loss": 1.0494,
+      "step": 106620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.59317807878616e-06,
+      "loss": 1.1384,
+      "step": 106625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.592423024541085e-06,
+      "loss": 1.3051,
+      "step": 106630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.591667972623498e-06,
+      "loss": 1.2427,
+      "step": 106635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.5909129230377e-06,
+      "loss": 1.1948,
+      "step": 106640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.590157875788014e-06,
+      "loss": 1.2565,
+      "step": 106645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.589402830878747e-06,
+      "loss": 1.1254,
+      "step": 106650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.588647788314208e-06,
+      "loss": 1.1124,
+      "step": 106655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.587892748098715e-06,
+      "loss": 1.021,
+      "step": 106660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.587137710236574e-06,
+      "loss": 1.1725,
+      "step": 106665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.586382674732098e-06,
+      "loss": 1.036,
+      "step": 106670
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.5856276415896e-06,
+      "loss": 1.2718,
+      "step": 106675
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.584872610813392e-06,
+      "loss": 1.2027,
+      "step": 106680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.584117582407784e-06,
+      "loss": 1.6677,
+      "step": 106685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.583362556377086e-06,
+      "loss": 1.2636,
+      "step": 106690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.582607532725617e-06,
+      "loss": 1.5308,
+      "step": 106695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.58185251145768e-06,
+      "loss": 1.3357,
+      "step": 106700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.581097492577588e-06,
+      "loss": 1.3162,
+      "step": 106705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.580342476089658e-06,
+      "loss": 1.2328,
+      "step": 106710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.579587461998199e-06,
+      "loss": 1.3236,
+      "step": 106715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.578832450307517e-06,
+      "loss": 1.3489,
+      "step": 106720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.578077441021932e-06,
+      "loss": 1.2839,
+      "step": 106725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.577322434145752e-06,
+      "loss": 1.1593,
+      "step": 106730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.57656742968329e-06,
+      "loss": 1.1281,
+      "step": 106735
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.575812427638851e-06,
+      "loss": 1.0631,
+      "step": 106740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.575057428016753e-06,
+      "loss": 1.282,
+      "step": 106745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.574302430821307e-06,
+      "loss": 1.4639,
+      "step": 106750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.573547436056821e-06,
+      "loss": 1.318,
+      "step": 106755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.572792443727609e-06,
+      "loss": 1.2809,
+      "step": 106760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.572037453837982e-06,
+      "loss": 1.1433,
+      "step": 106765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.57128246639225e-06,
+      "loss": 1.187,
+      "step": 106770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.57052748139473e-06,
+      "loss": 1.1034,
+      "step": 106775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.569772498849727e-06,
+      "loss": 1.1625,
+      "step": 106780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.569017518761552e-06,
+      "loss": 1.1891,
+      "step": 106785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.568262541134522e-06,
+      "loss": 1.128,
+      "step": 106790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.567507565972945e-06,
+      "loss": 1.0541,
+      "step": 106795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.566752593281128e-06,
+      "loss": 1.4222,
+      "step": 106800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.56599762306339e-06,
+      "loss": 1.9189,
+      "step": 106805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.565242655324042e-06,
+      "loss": 1.6696,
+      "step": 106810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.564487690067389e-06,
+      "loss": 1.1359,
+      "step": 106815
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.563732727297742e-06,
+      "loss": 1.217,
+      "step": 106820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.562977767019422e-06,
+      "loss": 1.0943,
+      "step": 106825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.562222809236733e-06,
+      "loss": 1.0632,
+      "step": 106830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.561467853953984e-06,
+      "loss": 1.0916,
+      "step": 106835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.56071290117549e-06,
+      "loss": 1.1698,
+      "step": 106840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.559957950905566e-06,
+      "loss": 1.088,
+      "step": 106845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.559203003148513e-06,
+      "loss": 1.0948,
+      "step": 106850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.55844805790865e-06,
+      "loss": 1.2011,
+      "step": 106855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.557693115190288e-06,
+      "loss": 1.3419,
+      "step": 106860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.556938174997733e-06,
+      "loss": 1.2545,
+      "step": 106865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.556183237335303e-06,
+      "loss": 1.1301,
+      "step": 106870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.555428302207303e-06,
+      "loss": 1.1357,
+      "step": 106875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.554673369618049e-06,
+      "loss": 1.3664,
+      "step": 106880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.553918439571844e-06,
+      "loss": 0.9398,
+      "step": 106885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.55316351207301e-06,
+      "loss": 1.1129,
+      "step": 106890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.552408587125851e-06,
+      "loss": 1.038,
+      "step": 106895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.551653664734677e-06,
+      "loss": 1.2422,
+      "step": 106900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.550898744903808e-06,
+      "loss": 1.0956,
+      "step": 106905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.550143827637544e-06,
+      "loss": 1.5686,
+      "step": 106910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.549388912940199e-06,
+      "loss": 1.5868,
+      "step": 106915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.54863400081609e-06,
+      "loss": 1.1709,
+      "step": 106920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.547879091269523e-06,
+      "loss": 0.9916,
+      "step": 106925
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.547124184304807e-06,
+      "loss": 1.0304,
+      "step": 106930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.546369279926257e-06,
+      "loss": 1.1107,
+      "step": 106935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.545614378138185e-06,
+      "loss": 1.0203,
+      "step": 106940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.544859478944892e-06,
+      "loss": 1.2551,
+      "step": 106945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.544104582350703e-06,
+      "loss": 1.15,
+      "step": 106950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.54334968835992e-06,
+      "loss": 1.3014,
+      "step": 106955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.542594796976859e-06,
+      "loss": 1.4482,
+      "step": 106960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.541839908205822e-06,
+      "loss": 1.3971,
+      "step": 106965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.541085022051129e-06,
+      "loss": 1.0342,
+      "step": 106970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.540330138517088e-06,
+      "loss": 1.1498,
+      "step": 106975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.539575257608004e-06,
+      "loss": 1.0852,
+      "step": 106980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.538820379328199e-06,
+      "loss": 1.4836,
+      "step": 106985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.538065503681974e-06,
+      "loss": 1.4495,
+      "step": 106990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.537310630673643e-06,
+      "loss": 1.1396,
+      "step": 106995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.53655576030752e-06,
+      "loss": 1.5463,
+      "step": 107000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.535800892587913e-06,
+      "loss": 0.9568,
+      "step": 107005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.53504602751913e-06,
+      "loss": 1.0448,
+      "step": 107010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.534291165105485e-06,
+      "loss": 1.4351,
+      "step": 107015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.533536305351291e-06,
+      "loss": 1.2526,
+      "step": 107020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.532781448260853e-06,
+      "loss": 1.3424,
+      "step": 107025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.532026593838481e-06,
+      "loss": 1.1239,
+      "step": 107030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.531271742088495e-06,
+      "loss": 1.2137,
+      "step": 107035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.530516893015196e-06,
+      "loss": 1.2242,
+      "step": 107040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.529762046622897e-06,
+      "loss": 1.2158,
+      "step": 107045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.529007202915911e-06,
+      "loss": 1.6378,
+      "step": 107050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.52825236189855e-06,
+      "loss": 1.3065,
+      "step": 107055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.527497523575117e-06,
+      "loss": 1.3021,
+      "step": 107060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.52674268794993e-06,
+      "loss": 1.3288,
+      "step": 107065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.525987855027296e-06,
+      "loss": 0.9335,
+      "step": 107070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.525233024811525e-06,
+      "loss": 1.0239,
+      "step": 107075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.524478197306933e-06,
+      "loss": 1.3398,
+      "step": 107080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.523723372517824e-06,
+      "loss": 1.1329,
+      "step": 107085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.522968550448507e-06,
+      "loss": 1.1829,
+      "step": 107090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.522213731103302e-06,
+      "loss": 1.2418,
+      "step": 107095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.521458914486511e-06,
+      "loss": 1.2284,
+      "step": 107100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.520704100602446e-06,
+      "loss": 1.1851,
+      "step": 107105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.51994928945542e-06,
+      "loss": 1.2612,
+      "step": 107110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.519194481049742e-06,
+      "loss": 1.5269,
+      "step": 107115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.518439675389722e-06,
+      "loss": 1.2835,
+      "step": 107120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.517684872479667e-06,
+      "loss": 1.0428,
+      "step": 107125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.516930072323894e-06,
+      "loss": 1.3971,
+      "step": 107130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.51617527492671e-06,
+      "loss": 0.7546,
+      "step": 107135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.515420480292425e-06,
+      "loss": 1.0161,
+      "step": 107140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.51466568842535e-06,
+      "loss": 0.9547,
+      "step": 107145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.513910899329796e-06,
+      "loss": 1.6011,
+      "step": 107150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.513156113010069e-06,
+      "loss": 1.5686,
+      "step": 107155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.512401329470483e-06,
+      "loss": 1.1145,
+      "step": 107160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.511646548715351e-06,
+      "loss": 1.5864,
+      "step": 107165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.51089177074898e-06,
+      "loss": 1.0552,
+      "step": 107170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.510136995575674e-06,
+      "loss": 1.1735,
+      "step": 107175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.509382223199754e-06,
+      "loss": 1.3744,
+      "step": 107180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.508627453625528e-06,
+      "loss": 1.1274,
+      "step": 107185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.507872686857295e-06,
+      "loss": 1.1889,
+      "step": 107190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.507117922899382e-06,
+      "loss": 1.1136,
+      "step": 107195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.506363161756088e-06,
+      "loss": 1.3949,
+      "step": 107200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.505608403431723e-06,
+      "loss": 1.2388,
+      "step": 107205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.504853647930603e-06,
+      "loss": 1.1349,
+      "step": 107210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.504098895257034e-06,
+      "loss": 1.3507,
+      "step": 107215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.503344145415326e-06,
+      "loss": 1.3445,
+      "step": 107220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.502589398409791e-06,
+      "loss": 1.2598,
+      "step": 107225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.50183465424474e-06,
+      "loss": 1.126,
+      "step": 107230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.501079912924478e-06,
+      "loss": 1.1817,
+      "step": 107235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.50032517445332e-06,
+      "loss": 1.1144,
+      "step": 107240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.499570438835575e-06,
+      "loss": 1.1094,
+      "step": 107245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.49881570607555e-06,
+      "loss": 1.0217,
+      "step": 107250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.498060976177557e-06,
+      "loss": 1.2,
+      "step": 107255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.497306249145906e-06,
+      "loss": 0.9563,
+      "step": 107260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.49655152498491e-06,
+      "loss": 1.1798,
+      "step": 107265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.49579680369887e-06,
+      "loss": 1.0776,
+      "step": 107270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.495042085292103e-06,
+      "loss": 1.438,
+      "step": 107275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.49428736976892e-06,
+      "loss": 1.3109,
+      "step": 107280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.493532657133623e-06,
+      "loss": 1.1547,
+      "step": 107285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.492777947390533e-06,
+      "loss": 1.1835,
+      "step": 107290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.49202324054395e-06,
+      "loss": 0.9402,
+      "step": 107295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.491268536598187e-06,
+      "loss": 1.2035,
+      "step": 107300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.490513835557557e-06,
+      "loss": 1.1216,
+      "step": 107305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.489759137426367e-06,
+      "loss": 1.138,
+      "step": 107310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.489004442208927e-06,
+      "loss": 1.469,
+      "step": 107315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.48824974990954e-06,
+      "loss": 1.2376,
+      "step": 107320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.48749506053253e-06,
+      "loss": 1.1925,
+      "step": 107325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.486740374082196e-06,
+      "loss": 1.3595,
+      "step": 107330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.485985690562847e-06,
+      "loss": 1.0359,
+      "step": 107335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.485231009978804e-06,
+      "loss": 1.6194,
+      "step": 107340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.484476332334364e-06,
+      "loss": 1.3811,
+      "step": 107345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.48372165763384e-06,
+      "loss": 1.452,
+      "step": 107350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.482966985881545e-06,
+      "loss": 1.2302,
+      "step": 107355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.482212317081788e-06,
+      "loss": 1.331,
+      "step": 107360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.481457651238874e-06,
+      "loss": 1.2581,
+      "step": 107365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.480702988357118e-06,
+      "loss": 1.0924,
+      "step": 107370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.479948328440829e-06,
+      "loss": 1.3424,
+      "step": 107375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.479193671494308e-06,
+      "loss": 1.3255,
+      "step": 107380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.478439017521878e-06,
+      "loss": 1.3626,
+      "step": 107385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.477684366527838e-06,
+      "loss": 1.4912,
+      "step": 107390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.476929718516506e-06,
+      "loss": 1.3489,
+      "step": 107395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.476175073492181e-06,
+      "loss": 1.4536,
+      "step": 107400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.47542043145918e-06,
+      "loss": 1.6774,
+      "step": 107405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.474665792421811e-06,
+      "loss": 1.2011,
+      "step": 107410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.47391115638438e-06,
+      "loss": 1.1756,
+      "step": 107415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.473156523351205e-06,
+      "loss": 1.2742,
+      "step": 107420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.472401893326586e-06,
+      "loss": 1.0709,
+      "step": 107425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.471647266314834e-06,
+      "loss": 1.0435,
+      "step": 107430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.470892642320265e-06,
+      "loss": 1.4317,
+      "step": 107435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.470138021347182e-06,
+      "loss": 1.1086,
+      "step": 107440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.469383403399893e-06,
+      "loss": 1.3605,
+      "step": 107445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.468628788482712e-06,
+      "loss": 1.2727,
+      "step": 107450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.467874176599949e-06,
+      "loss": 1.2075,
+      "step": 107455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.46711956775591e-06,
+      "loss": 1.2183,
+      "step": 107460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.466364961954901e-06,
+      "loss": 1.16,
+      "step": 107465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.465610359201237e-06,
+      "loss": 1.1527,
+      "step": 107470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.464855759499228e-06,
+      "loss": 1.3137,
+      "step": 107475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.464101162853176e-06,
+      "loss": 1.132,
+      "step": 107480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.463346569267397e-06,
+      "loss": 1.5601,
+      "step": 107485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.4625919787462e-06,
+      "loss": 1.43,
+      "step": 107490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.461837391293886e-06,
+      "loss": 1.0328,
+      "step": 107495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.461082806914774e-06,
+      "loss": 1.2279,
+      "step": 107500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.46032822561317e-06,
+      "loss": 1.5464,
+      "step": 107505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.459573647393376e-06,
+      "loss": 1.4134,
+      "step": 107510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.458819072259715e-06,
+      "loss": 1.023,
+      "step": 107515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.458064500216484e-06,
+      "loss": 1.3835,
+      "step": 107520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.457309931267995e-06,
+      "loss": 1.4004,
+      "step": 107525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.45655536541856e-06,
+      "loss": 1.4532,
+      "step": 107530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.455800802672489e-06,
+      "loss": 1.1303,
+      "step": 107535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.455046243034086e-06,
+      "loss": 1.218,
+      "step": 107540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.454291686507658e-06,
+      "loss": 1.1775,
+      "step": 107545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.453537133097524e-06,
+      "loss": 1.0438,
+      "step": 107550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.452782582807985e-06,
+      "loss": 1.2621,
+      "step": 107555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.452028035643348e-06,
+      "loss": 1.0148,
+      "step": 107560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.45127349160793e-06,
+      "loss": 1.4313,
+      "step": 107565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.450518950706036e-06,
+      "loss": 1.1725,
+      "step": 107570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.449764412941969e-06,
+      "loss": 1.4622,
+      "step": 107575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.449009878320047e-06,
+      "loss": 1.4804,
+      "step": 107580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.448255346844578e-06,
+      "loss": 1.1881,
+      "step": 107585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.447500818519859e-06,
+      "loss": 1.0857,
+      "step": 107590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.446746293350215e-06,
+      "loss": 1.5482,
+      "step": 107595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.445991771339946e-06,
+      "loss": 1.3352,
+      "step": 107600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.44523725249336e-06,
+      "loss": 1.3495,
+      "step": 107605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.44448273681477e-06,
+      "loss": 1.0593,
+      "step": 107610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.44372822430848e-06,
+      "loss": 1.2956,
+      "step": 107615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.442973714978805e-06,
+      "loss": 1.0048,
+      "step": 107620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.442219208830042e-06,
+      "loss": 1.2269,
+      "step": 107625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.441464705866515e-06,
+      "loss": 1.2204,
+      "step": 107630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.440710206092523e-06,
+      "loss": 1.1497,
+      "step": 107635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.439955709512372e-06,
+      "loss": 1.2334,
+      "step": 107640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.43920121613038e-06,
+      "loss": 1.4094,
+      "step": 107645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.438446725950851e-06,
+      "loss": 1.1196,
+      "step": 107650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.437692238978089e-06,
+      "loss": 1.4158,
+      "step": 107655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.43693775521641e-06,
+      "loss": 1.1024,
+      "step": 107660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.436183274670121e-06,
+      "loss": 1.1437,
+      "step": 107665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.435428797343524e-06,
+      "loss": 1.2131,
+      "step": 107670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.434674323240934e-06,
+      "loss": 1.526,
+      "step": 107675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.433919852366661e-06,
+      "loss": 0.8574,
+      "step": 107680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.433165384725006e-06,
+      "loss": 1.4352,
+      "step": 107685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.432410920320283e-06,
+      "loss": 1.4013,
+      "step": 107690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.431656459156799e-06,
+      "loss": 1.438,
+      "step": 107695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.430902001238864e-06,
+      "loss": 1.1708,
+      "step": 107700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.43014754657078e-06,
+      "loss": 1.0702,
+      "step": 107705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.429393095156863e-06,
+      "loss": 1.2319,
+      "step": 107710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.428638647001418e-06,
+      "loss": 1.2765,
+      "step": 107715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.427884202108752e-06,
+      "loss": 1.0291,
+      "step": 107720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.42712976048318e-06,
+      "loss": 1.8215,
+      "step": 107725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.426375322129002e-06,
+      "loss": 1.2384,
+      "step": 107730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.425620887050526e-06,
+      "loss": 1.3362,
+      "step": 107735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.42486645525207e-06,
+      "loss": 1.3005,
+      "step": 107740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.424112026737933e-06,
+      "loss": 1.2521,
+      "step": 107745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.423357601512422e-06,
+      "loss": 1.1803,
+      "step": 107750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.422603179579855e-06,
+      "loss": 1.3362,
+      "step": 107755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.421848760944535e-06,
+      "loss": 1.1783,
+      "step": 107760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.421094345610766e-06,
+      "loss": 1.6073,
+      "step": 107765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.420339933582859e-06,
+      "loss": 0.982,
+      "step": 107770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.419585524865124e-06,
+      "loss": 1.2896,
+      "step": 107775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.418831119461871e-06,
+      "loss": 1.3531,
+      "step": 107780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.4180767173774e-06,
+      "loss": 1.5873,
+      "step": 107785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.417322318616026e-06,
+      "loss": 1.1784,
+      "step": 107790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.416567923182057e-06,
+      "loss": 1.1555,
+      "step": 107795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.415813531079796e-06,
+      "loss": 1.1416,
+      "step": 107800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.415059142313555e-06,
+      "loss": 1.2175,
+      "step": 107805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.414304756887642e-06,
+      "loss": 1.1188,
+      "step": 107810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.413550374806359e-06,
+      "loss": 1.1291,
+      "step": 107815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.412795996074026e-06,
+      "loss": 1.4237,
+      "step": 107820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.41204162069494e-06,
+      "loss": 1.2594,
+      "step": 107825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.411287248673415e-06,
+      "loss": 1.4432,
+      "step": 107830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.410532880013749e-06,
+      "loss": 1.2707,
+      "step": 107835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.409778514720266e-06,
+      "loss": 1.3577,
+      "step": 107840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.409024152797261e-06,
+      "loss": 1.2323,
+      "step": 107845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.408269794249045e-06,
+      "loss": 1.2388,
+      "step": 107850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.40751543907993e-06,
+      "loss": 1.2975,
+      "step": 107855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.406761087294219e-06,
+      "loss": 1.1375,
+      "step": 107860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.40600673889622e-06,
+      "loss": 1.2999,
+      "step": 107865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.405252393890243e-06,
+      "loss": 1.353,
+      "step": 107870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.404498052280597e-06,
+      "loss": 1.2179,
+      "step": 107875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.403743714071582e-06,
+      "loss": 1.3773,
+      "step": 107880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.402989379267516e-06,
+      "loss": 1.3059,
+      "step": 107885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.402235047872702e-06,
+      "loss": 1.4793,
+      "step": 107890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.401480719891442e-06,
+      "loss": 1.2867,
+      "step": 107895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.400726395328053e-06,
+      "loss": 1.3321,
+      "step": 107900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.39997207418684e-06,
+      "loss": 1.3425,
+      "step": 107905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.39921775647211e-06,
+      "loss": 1.3373,
+      "step": 107910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.398463442188163e-06,
+      "loss": 1.2917,
+      "step": 107915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.397709131339317e-06,
+      "loss": 1.2585,
+      "step": 107920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.396954823929879e-06,
+      "loss": 1.3245,
+      "step": 107925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.396200519964148e-06,
+      "loss": 1.2638,
+      "step": 107930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.39544621944644e-06,
+      "loss": 1.2509,
+      "step": 107935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.394691922381058e-06,
+      "loss": 1.1191,
+      "step": 107940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.393937628772309e-06,
+      "loss": 1.2248,
+      "step": 107945
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.393183338624508e-06,
+      "loss": 1.1819,
+      "step": 107950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.392429051941952e-06,
+      "loss": 1.4352,
+      "step": 107955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.391674768728952e-06,
+      "loss": 0.8758,
+      "step": 107960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.390920488989818e-06,
+      "loss": 1.1974,
+      "step": 107965
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.390166212728858e-06,
+      "loss": 1.209,
+      "step": 107970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.389411939950375e-06,
+      "loss": 1.7601,
+      "step": 107975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.388657670658674e-06,
+      "loss": 1.2849,
+      "step": 107980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.387903404858073e-06,
+      "loss": 1.3175,
+      "step": 107985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.387149142552871e-06,
+      "loss": 1.4855,
+      "step": 107990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.386394883747373e-06,
+      "loss": 1.2747,
+      "step": 107995
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.385640628445893e-06,
+      "loss": 1.4438,
+      "step": 108000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.384886376652738e-06,
+      "loss": 1.2548,
+      "step": 108005
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.384132128372208e-06,
+      "loss": 1.0465,
+      "step": 108010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.383377883608617e-06,
+      "loss": 1.2313,
+      "step": 108015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.382623642366274e-06,
+      "loss": 1.2026,
+      "step": 108020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.381869404649473e-06,
+      "loss": 1.2946,
+      "step": 108025
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.381115170462538e-06,
+      "loss": 1.0364,
+      "step": 108030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.380360939809765e-06,
+      "loss": 1.507,
+      "step": 108035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.379606712695462e-06,
+      "loss": 1.2656,
+      "step": 108040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.378852489123944e-06,
+      "loss": 1.1843,
+      "step": 108045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.37809826909951e-06,
+      "loss": 1.1433,
+      "step": 108050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.37734405262647e-06,
+      "loss": 1.3761,
+      "step": 108055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.376589839709125e-06,
+      "loss": 1.01,
+      "step": 108060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.375835630351796e-06,
+      "loss": 1.3089,
+      "step": 108065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.375081424558776e-06,
+      "loss": 1.2924,
+      "step": 108070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.374327222334376e-06,
+      "loss": 1.2464,
+      "step": 108075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.37357302368291e-06,
+      "loss": 0.9853,
+      "step": 108080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.372818828608675e-06,
+      "loss": 0.9965,
+      "step": 108085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.372064637115978e-06,
+      "loss": 1.288,
+      "step": 108090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.371310449209135e-06,
+      "loss": 1.253,
+      "step": 108095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.370556264892447e-06,
+      "loss": 1.3393,
+      "step": 108100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.369802084170218e-06,
+      "loss": 1.4962,
+      "step": 108105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.369047907046761e-06,
+      "loss": 1.1883,
+      "step": 108110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.368293733526379e-06,
+      "loss": 1.1481,
+      "step": 108115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.367539563613383e-06,
+      "loss": 1.3783,
+      "step": 108120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.36678539731207e-06,
+      "loss": 0.8967,
+      "step": 108125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.366031234626757e-06,
+      "loss": 1.2085,
+      "step": 108130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.365277075561748e-06,
+      "loss": 1.6026,
+      "step": 108135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.364522920121343e-06,
+      "loss": 1.4056,
+      "step": 108140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.363768768309857e-06,
+      "loss": 1.1925,
+      "step": 108145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.363014620131593e-06,
+      "loss": 1.3914,
+      "step": 108150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.362260475590857e-06,
+      "loss": 1.0492,
+      "step": 108155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.36150633469196e-06,
+      "loss": 1.0093,
+      "step": 108160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.360752197439203e-06,
+      "loss": 1.177,
+      "step": 108165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.359998063836894e-06,
+      "loss": 1.289,
+      "step": 108170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.359243933889342e-06,
+      "loss": 1.1299,
+      "step": 108175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.358489807600852e-06,
+      "loss": 1.3698,
+      "step": 108180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.357735684975728e-06,
+      "loss": 1.5224,
+      "step": 108185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.35698156601828e-06,
+      "loss": 1.0154,
+      "step": 108190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.356227450732815e-06,
+      "loss": 1.2014,
+      "step": 108195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.355473339123637e-06,
+      "loss": 1.3666,
+      "step": 108200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.354719231195048e-06,
+      "loss": 1.2211,
+      "step": 108205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.353965126951365e-06,
+      "loss": 1.2719,
+      "step": 108210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.353211026396888e-06,
+      "loss": 0.9334,
+      "step": 108215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.35245692953592e-06,
+      "loss": 1.2119,
+      "step": 108220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.351702836372773e-06,
+      "loss": 0.985,
+      "step": 108225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.350948746911754e-06,
+      "loss": 1.0998,
+      "step": 108230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.350194661157162e-06,
+      "loss": 1.3644,
+      "step": 108235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.349440579113312e-06,
+      "loss": 1.8781,
+      "step": 108240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.348686500784506e-06,
+      "loss": 1.2079,
+      "step": 108245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.347932426175047e-06,
+      "loss": 0.8478,
+      "step": 108250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.34717835528925e-06,
+      "loss": 1.0339,
+      "step": 108255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.346424288131413e-06,
+      "loss": 1.1703,
+      "step": 108260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.345670224705846e-06,
+      "loss": 1.2086,
+      "step": 108265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.344916165016848e-06,
+      "loss": 1.346,
+      "step": 108270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.344162109068737e-06,
+      "loss": 1.3136,
+      "step": 108275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.343408056865812e-06,
+      "loss": 1.2804,
+      "step": 108280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.34265400841238e-06,
+      "loss": 1.328,
+      "step": 108285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.341899963712747e-06,
+      "loss": 0.9846,
+      "step": 108290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.341145922771219e-06,
+      "loss": 1.2925,
+      "step": 108295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.340391885592101e-06,
+      "loss": 1.2295,
+      "step": 108300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.339637852179701e-06,
+      "loss": 1.2683,
+      "step": 108305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.338883822538326e-06,
+      "loss": 1.4824,
+      "step": 108310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.338129796672277e-06,
+      "loss": 1.4867,
+      "step": 108315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.337375774585863e-06,
+      "loss": 1.1212,
+      "step": 108320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.336621756283395e-06,
+      "loss": 1.3944,
+      "step": 108325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.335867741769164e-06,
+      "loss": 0.9569,
+      "step": 108330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.335113731047495e-06,
+      "loss": 1.2995,
+      "step": 108335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.334359724122679e-06,
+      "loss": 1.282,
+      "step": 108340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.33360572099903e-06,
+      "loss": 1.4222,
+      "step": 108345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.332851721680847e-06,
+      "loss": 0.9317,
+      "step": 108350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.332097726172442e-06,
+      "loss": 1.3826,
+      "step": 108355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.331343734478117e-06,
+      "loss": 1.3386,
+      "step": 108360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.330589746602179e-06,
+      "loss": 1.1982,
+      "step": 108365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.329835762548935e-06,
+      "loss": 1.3048,
+      "step": 108370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.32908178232269e-06,
+      "loss": 1.3614,
+      "step": 108375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.328327805927743e-06,
+      "loss": 1.3141,
+      "step": 108380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.327573833368413e-06,
+      "loss": 1.3737,
+      "step": 108385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.326819864648995e-06,
+      "loss": 1.102,
+      "step": 108390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.326065899773796e-06,
+      "loss": 1.0453,
+      "step": 108395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.325311938747124e-06,
+      "loss": 0.8608,
+      "step": 108400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.324557981573285e-06,
+      "loss": 1.1324,
+      "step": 108405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.323804028256583e-06,
+      "loss": 1.5218,
+      "step": 108410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.323050078801322e-06,
+      "loss": 1.2407,
+      "step": 108415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.32229613321181e-06,
+      "loss": 1.2333,
+      "step": 108420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.321542191492354e-06,
+      "loss": 1.1481,
+      "step": 108425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.320788253647255e-06,
+      "loss": 1.5246,
+      "step": 108430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.320034319680819e-06,
+      "loss": 1.3925,
+      "step": 108435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.319280389597355e-06,
+      "loss": 1.3051,
+      "step": 108440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.318526463401165e-06,
+      "loss": 1.1329,
+      "step": 108445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.317772541096555e-06,
+      "loss": 1.5278,
+      "step": 108450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.317018622687831e-06,
+      "loss": 1.0557,
+      "step": 108455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.316264708179296e-06,
+      "loss": 1.2609,
+      "step": 108460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.315510797575263e-06,
+      "loss": 1.2323,
+      "step": 108465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.314756890880029e-06,
+      "loss": 1.0275,
+      "step": 108470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.314002988097897e-06,
+      "loss": 0.9871,
+      "step": 108475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.313249089233181e-06,
+      "loss": 1.2474,
+      "step": 108480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.312495194290184e-06,
+      "loss": 0.9987,
+      "step": 108485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.311741303273208e-06,
+      "loss": 1.472,
+      "step": 108490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.310987416186556e-06,
+      "loss": 1.0292,
+      "step": 108495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.31023353303454e-06,
+      "loss": 1.5925,
+      "step": 108500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.309479653821461e-06,
+      "loss": 1.195,
+      "step": 108505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.308725778551622e-06,
+      "loss": 1.2346,
+      "step": 108510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.307971907229333e-06,
+      "loss": 1.4077,
+      "step": 108515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.307218039858898e-06,
+      "loss": 1.5822,
+      "step": 108520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.306464176444616e-06,
+      "loss": 1.1742,
+      "step": 108525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.3057103169908e-06,
+      "loss": 1.2074,
+      "step": 108530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.304956461501754e-06,
+      "loss": 1.1033,
+      "step": 108535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.304202609981773e-06,
+      "loss": 1.5726,
+      "step": 108540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.303448762435174e-06,
+      "loss": 1.6602,
+      "step": 108545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.302694918866258e-06,
+      "loss": 1.512,
+      "step": 108550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.30194107927933e-06,
+      "loss": 1.1818,
+      "step": 108555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.301187243678688e-06,
+      "loss": 1.0493,
+      "step": 108560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.300433412068647e-06,
+      "loss": 1.3569,
+      "step": 108565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.29967958445351e-06,
+      "loss": 1.8657,
+      "step": 108570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.298925760837572e-06,
+      "loss": 1.4382,
+      "step": 108575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.298171941225151e-06,
+      "loss": 1.0317,
+      "step": 108580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.297418125620544e-06,
+      "loss": 1.3628,
+      "step": 108585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.296664314028055e-06,
+      "loss": 1.0818,
+      "step": 108590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.295910506451996e-06,
+      "loss": 1.0762,
+      "step": 108595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.295156702896664e-06,
+      "loss": 1.1059,
+      "step": 108600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.294402903366363e-06,
+      "loss": 1.2166,
+      "step": 108605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.293649107865405e-06,
+      "loss": 1.0361,
+      "step": 108610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.292895316398094e-06,
+      "loss": 1.2679,
+      "step": 108615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.292141528968724e-06,
+      "loss": 1.271,
+      "step": 108620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.291387745581609e-06,
+      "loss": 0.9362,
+      "step": 108625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.290633966241053e-06,
+      "loss": 1.1446,
+      "step": 108630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.289880190951358e-06,
+      "loss": 1.0809,
+      "step": 108635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.289126419716824e-06,
+      "loss": 1.3433,
+      "step": 108640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.288372652541766e-06,
+      "loss": 1.2408,
+      "step": 108645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.287618889430483e-06,
+      "loss": 1.7263,
+      "step": 108650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.286865130387276e-06,
+      "loss": 1.0159,
+      "step": 108655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.286111375416455e-06,
+      "loss": 1.6481,
+      "step": 108660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.285357624522325e-06,
+      "loss": 0.9286,
+      "step": 108665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.284603877709178e-06,
+      "loss": 1.07,
+      "step": 108670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.283850134981337e-06,
+      "loss": 0.9595,
+      "step": 108675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.283096396343093e-06,
+      "loss": 1.3889,
+      "step": 108680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.282342661798752e-06,
+      "loss": 1.1104,
+      "step": 108685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.281588931352626e-06,
+      "loss": 1.0938,
+      "step": 108690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.28083520500901e-06,
+      "loss": 0.8841,
+      "step": 108695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.280081482772213e-06,
+      "loss": 1.0644,
+      "step": 108700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.279327764646535e-06,
+      "loss": 0.9905,
+      "step": 108705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.278574050636287e-06,
+      "loss": 1.0273,
+      "step": 108710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.277820340745768e-06,
+      "loss": 1.3589,
+      "step": 108715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.277066634979282e-06,
+      "loss": 1.4877,
+      "step": 108720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.276312933341136e-06,
+      "loss": 1.2211,
+      "step": 108725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.275559235835632e-06,
+      "loss": 1.5794,
+      "step": 108730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.27480554246707e-06,
+      "loss": 1.2424,
+      "step": 108735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.274051853239765e-06,
+      "loss": 1.2444,
+      "step": 108740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.273298168158013e-06,
+      "loss": 1.2998,
+      "step": 108745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.272544487226116e-06,
+      "loss": 1.5037,
+      "step": 108750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.271790810448384e-06,
+      "loss": 1.2644,
+      "step": 108755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.271037137829118e-06,
+      "loss": 1.0523,
+      "step": 108760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.270283469372618e-06,
+      "loss": 1.0893,
+      "step": 108765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.269529805083198e-06,
+      "loss": 0.9744,
+      "step": 108770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.268776144965153e-06,
+      "loss": 1.2959,
+      "step": 108775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.268022489022792e-06,
+      "loss": 1.0459,
+      "step": 108780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.26726883726041e-06,
+      "loss": 1.1236,
+      "step": 108785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.266515189682323e-06,
+      "loss": 1.2425,
+      "step": 108790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.265761546292826e-06,
+      "loss": 1.189,
+      "step": 108795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.265007907096224e-06,
+      "loss": 1.1563,
+      "step": 108800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.264254272096827e-06,
+      "loss": 1.1152,
+      "step": 108805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.263500641298931e-06,
+      "loss": 1.1934,
+      "step": 108810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.26274701470684e-06,
+      "loss": 1.1318,
+      "step": 108815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.261993392324864e-06,
+      "loss": 1.278,
+      "step": 108820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.261239774157304e-06,
+      "loss": 1.1113,
+      "step": 108825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.260486160208456e-06,
+      "loss": 1.1518,
+      "step": 108830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.259732550482634e-06,
+      "loss": 1.3875,
+      "step": 108835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.258978944984138e-06,
+      "loss": 1.2548,
+      "step": 108840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.25822534371727e-06,
+      "loss": 1.1717,
+      "step": 108845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.257471746686331e-06,
+      "loss": 1.3624,
+      "step": 108850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.25671815389563e-06,
+      "loss": 0.812,
+      "step": 108855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.255964565349471e-06,
+      "loss": 1.4742,
+      "step": 108860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.25521098105215e-06,
+      "loss": 0.9486,
+      "step": 108865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.254457401007977e-06,
+      "loss": 1.3096,
+      "step": 108870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.253703825221257e-06,
+      "loss": 1.3648,
+      "step": 108875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.25295025369628e-06,
+      "loss": 1.0442,
+      "step": 108880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.252196686437367e-06,
+      "loss": 1.0037,
+      "step": 108885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.251443123448811e-06,
+      "loss": 1.5478,
+      "step": 108890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.250689564734915e-06,
+      "loss": 1.3465,
+      "step": 108895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.24993601029999e-06,
+      "loss": 1.0761,
+      "step": 108900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.24918246014833e-06,
+      "loss": 1.1565,
+      "step": 108905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.24842891428424e-06,
+      "loss": 1.3743,
+      "step": 108910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.247675372712026e-06,
+      "loss": 1.4526,
+      "step": 108915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.246921835435996e-06,
+      "loss": 1.1165,
+      "step": 108920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.246168302460443e-06,
+      "loss": 1.8834,
+      "step": 108925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.24541477378967e-06,
+      "loss": 1.6986,
+      "step": 108930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.244661249427992e-06,
+      "loss": 1.7221,
+      "step": 108935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.243907729379704e-06,
+      "loss": 1.271,
+      "step": 108940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.243154213649103e-06,
+      "loss": 1.4151,
+      "step": 108945
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.242400702240503e-06,
+      "loss": 1.5514,
+      "step": 108950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.241647195158205e-06,
+      "loss": 1.3118,
+      "step": 108955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.240893692406506e-06,
+      "loss": 1.3486,
+      "step": 108960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.240140193989713e-06,
+      "loss": 1.0332,
+      "step": 108965
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.23938669991213e-06,
+      "loss": 1.247,
+      "step": 108970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.238633210178051e-06,
+      "loss": 1.459,
+      "step": 108975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.237879724791793e-06,
+      "loss": 1.0037,
+      "step": 108980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.237126243757649e-06,
+      "loss": 1.2281,
+      "step": 108985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.236372767079929e-06,
+      "loss": 1.4263,
+      "step": 108990
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.235619294762924e-06,
+      "loss": 1.0684,
+      "step": 108995
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.234865826810948e-06,
+      "loss": 1.2893,
+      "step": 109000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.234112363228299e-06,
+      "loss": 1.1709,
+      "step": 109005
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.233358904019278e-06,
+      "loss": 1.5233,
+      "step": 109010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.232605449188195e-06,
+      "loss": 1.1866,
+      "step": 109015
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.231851998739345e-06,
+      "loss": 1.3302,
+      "step": 109020
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.23109855267703e-06,
+      "loss": 1.1887,
+      "step": 109025
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.23034511100556e-06,
+      "loss": 0.9755,
+      "step": 109030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.229591673729233e-06,
+      "loss": 1.3408,
+      "step": 109035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.228838240852351e-06,
+      "loss": 0.9788,
+      "step": 109040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.228084812379217e-06,
+      "loss": 1.5889,
+      "step": 109045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.227331388314137e-06,
+      "loss": 1.4352,
+      "step": 109050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.226577968661406e-06,
+      "loss": 1.4757,
+      "step": 109055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.225824553425332e-06,
+      "loss": 1.2343,
+      "step": 109060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.22507114261022e-06,
+      "loss": 1.173,
+      "step": 109065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.224317736220366e-06,
+      "loss": 1.2836,
+      "step": 109070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.223564334260071e-06,
+      "loss": 1.6764,
+      "step": 109075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.222810936733645e-06,
+      "loss": 0.9748,
+      "step": 109080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.222057543645388e-06,
+      "loss": 1.6039,
+      "step": 109085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.221304154999598e-06,
+      "loss": 1.416,
+      "step": 109090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.220550770800581e-06,
+      "loss": 1.0925,
+      "step": 109095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.219797391052638e-06,
+      "loss": 1.1305,
+      "step": 109100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.21904401576007e-06,
+      "loss": 1.0267,
+      "step": 109105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.218290644927185e-06,
+      "loss": 1.1781,
+      "step": 109110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.217537278558278e-06,
+      "loss": 0.9461,
+      "step": 109115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.216783916657653e-06,
+      "loss": 1.1199,
+      "step": 109120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.216030559229617e-06,
+      "loss": 1.3201,
+      "step": 109125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.215277206278466e-06,
+      "loss": 1.2061,
+      "step": 109130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.214523857808503e-06,
+      "loss": 1.1724,
+      "step": 109135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.21377051382403e-06,
+      "loss": 1.2156,
+      "step": 109140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.213017174329355e-06,
+      "loss": 0.9858,
+      "step": 109145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.212263839328773e-06,
+      "loss": 1.3511,
+      "step": 109150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.211510508826587e-06,
+      "loss": 1.1145,
+      "step": 109155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.2107571828271e-06,
+      "loss": 1.512,
+      "step": 109160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.210003861334617e-06,
+      "loss": 1.4534,
+      "step": 109165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.209250544353433e-06,
+      "loss": 1.0907,
+      "step": 109170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.208497231887855e-06,
+      "loss": 1.3624,
+      "step": 109175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.207743923942188e-06,
+      "loss": 1.1095,
+      "step": 109180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.206990620520722e-06,
+      "loss": 1.1698,
+      "step": 109185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.20623732162777e-06,
+      "loss": 1.3849,
+      "step": 109190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.20548402726763e-06,
+      "loss": 1.2008,
+      "step": 109195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.2047307374446e-06,
+      "loss": 1.1573,
+      "step": 109200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.20397745216299e-06,
+      "loss": 1.1274,
+      "step": 109205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.203224171427094e-06,
+      "loss": 1.2296,
+      "step": 109210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.20247089524122e-06,
+      "loss": 1.0754,
+      "step": 109215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.20171762360966e-06,
+      "loss": 1.3945,
+      "step": 109220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.200964356536729e-06,
+      "loss": 1.0611,
+      "step": 109225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.200211094026716e-06,
+      "loss": 1.1617,
+      "step": 109230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.199457836083926e-06,
+      "loss": 1.3041,
+      "step": 109235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.19870458271267e-06,
+      "loss": 1.6695,
+      "step": 109240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.197951333917237e-06,
+      "loss": 1.0798,
+      "step": 109245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.197198089701933e-06,
+      "loss": 0.9901,
+      "step": 109250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.196444850071059e-06,
+      "loss": 1.1998,
+      "step": 109255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.195691615028922e-06,
+      "loss": 1.4775,
+      "step": 109260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.194938384579812e-06,
+      "loss": 1.278,
+      "step": 109265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.19418515872804e-06,
+      "loss": 1.0772,
+      "step": 109270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.193431937477906e-06,
+      "loss": 1.0459,
+      "step": 109275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.192678720833707e-06,
+      "loss": 1.2576,
+      "step": 109280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.191925508799746e-06,
+      "loss": 1.1834,
+      "step": 109285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.191172301380328e-06,
+      "loss": 1.1736,
+      "step": 109290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.19041909857975e-06,
+      "loss": 1.1387,
+      "step": 109295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.189665900402312e-06,
+      "loss": 1.5552,
+      "step": 109300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.188912706852319e-06,
+      "loss": 1.2869,
+      "step": 109305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.188159517934073e-06,
+      "loss": 1.3678,
+      "step": 109310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.187406333651868e-06,
+      "loss": 1.1735,
+      "step": 109315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.186653154010013e-06,
+      "loss": 1.3229,
+      "step": 109320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.185899979012807e-06,
+      "loss": 1.3563,
+      "step": 109325
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.185146808664546e-06,
+      "loss": 1.1485,
+      "step": 109330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.18439364296954e-06,
+      "loss": 1.3153,
+      "step": 109335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.183640481932083e-06,
+      "loss": 1.4596,
+      "step": 109340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.182887325556476e-06,
+      "loss": 1.4485,
+      "step": 109345
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.182134173847022e-06,
+      "loss": 1.2126,
+      "step": 109350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.181381026808027e-06,
+      "loss": 1.3306,
+      "step": 109355
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.18062788444378e-06,
+      "loss": 0.9397,
+      "step": 109360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.179874746758592e-06,
+      "loss": 1.3271,
+      "step": 109365
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.17912161375676e-06,
+      "loss": 1.1795,
+      "step": 109370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.178368485442586e-06,
+      "loss": 1.2079,
+      "step": 109375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.177615361820365e-06,
+      "loss": 1.2866,
+      "step": 109380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.176862242894407e-06,
+      "loss": 1.2467,
+      "step": 109385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.176109128669009e-06,
+      "loss": 1.0759,
+      "step": 109390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.175356019148469e-06,
+      "loss": 1.1203,
+      "step": 109395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.17460291433709e-06,
+      "loss": 1.1849,
+      "step": 109400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.173849814239171e-06,
+      "loss": 1.1811,
+      "step": 109405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.173096718859013e-06,
+      "loss": 0.989,
+      "step": 109410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.172343628200924e-06,
+      "loss": 1.0961,
+      "step": 109415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.171590542269193e-06,
+      "loss": 1.3632,
+      "step": 109420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.17083746106813e-06,
+      "loss": 1.4722,
+      "step": 109425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.170084384602023e-06,
+      "loss": 1.0976,
+      "step": 109430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.169331312875185e-06,
+      "loss": 1.2642,
+      "step": 109435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.168578245891913e-06,
+      "loss": 1.1833,
+      "step": 109440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.167825183656503e-06,
+      "loss": 1.1191,
+      "step": 109445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.167072126173265e-06,
+      "loss": 1.0627,
+      "step": 109450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.166319073446489e-06,
+      "loss": 1.244,
+      "step": 109455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.165566025480477e-06,
+      "loss": 1.196,
+      "step": 109460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.164812982279535e-06,
+      "loss": 1.1937,
+      "step": 109465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.164059943847961e-06,
+      "loss": 1.0471,
+      "step": 109470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.163306910190052e-06,
+      "loss": 1.1776,
+      "step": 109475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.162553881310112e-06,
+      "loss": 1.6247,
+      "step": 109480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.161800857212441e-06,
+      "loss": 1.0231,
+      "step": 109485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.161047837901335e-06,
+      "loss": 1.25,
+      "step": 109490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.160294823381097e-06,
+      "loss": 1.4449,
+      "step": 109495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.15954181365603e-06,
+      "loss": 1.4138,
+      "step": 109500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.158788808730432e-06,
+      "loss": 0.9079,
+      "step": 109505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.158035808608597e-06,
+      "loss": 0.9796,
+      "step": 109510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.157282813294834e-06,
+      "loss": 1.0168,
+      "step": 109515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.156529822793441e-06,
+      "loss": 1.226,
+      "step": 109520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.15577683710871e-06,
+      "loss": 1.1105,
+      "step": 109525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.155023856244955e-06,
+      "loss": 1.0741,
+      "step": 109530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.154270880206465e-06,
+      "loss": 1.1193,
+      "step": 109535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.15351790899754e-06,
+      "loss": 1.2163,
+      "step": 109540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.15276494262249e-06,
+      "loss": 1.3455,
+      "step": 109545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.152011981085603e-06,
+      "loss": 1.5598,
+      "step": 109550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.151259024391182e-06,
+      "loss": 1.1584,
+      "step": 109555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.150506072543532e-06,
+      "loss": 1.4893,
+      "step": 109560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.14975312554695e-06,
+      "loss": 1.1052,
+      "step": 109565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.149000183405733e-06,
+      "loss": 1.2689,
+      "step": 109570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.148247246124183e-06,
+      "loss": 1.3986,
+      "step": 109575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.147494313706601e-06,
+      "loss": 1.1496,
+      "step": 109580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.146741386157283e-06,
+      "loss": 0.9087,
+      "step": 109585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.145988463480529e-06,
+      "loss": 1.2843,
+      "step": 109590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.145235545680642e-06,
+      "loss": 1.1934,
+      "step": 109595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.144482632761922e-06,
+      "loss": 1.1716,
+      "step": 109600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.143729724728662e-06,
+      "loss": 1.2207,
+      "step": 109605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.142976821585169e-06,
+      "loss": 1.2625,
+      "step": 109610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.14222392333574e-06,
+      "loss": 1.0582,
+      "step": 109615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.141471029984668e-06,
+      "loss": 1.2286,
+      "step": 109620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.140718141536264e-06,
+      "loss": 1.1888,
+      "step": 109625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.139965257994821e-06,
+      "loss": 1.2861,
+      "step": 109630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.139212379364636e-06,
+      "loss": 1.2374,
+      "step": 109635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.138459505650013e-06,
+      "loss": 1.0813,
+      "step": 109640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.13770663685525e-06,
+      "loss": 0.9258,
+      "step": 109645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.13695377298465e-06,
+      "loss": 1.2374,
+      "step": 109650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.136200914042499e-06,
+      "loss": 1.2703,
+      "step": 109655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.135448060033111e-06,
+      "loss": 1.0179,
+      "step": 109660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.13469521096078e-06,
+      "loss": 1.1184,
+      "step": 109665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.1339423668298e-06,
+      "loss": 1.4762,
+      "step": 109670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.133189527644481e-06,
+      "loss": 1.0688,
+      "step": 109675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.132436693409114e-06,
+      "loss": 1.0697,
+      "step": 109680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.131683864127998e-06,
+      "loss": 1.4059,
+      "step": 109685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.130931039805435e-06,
+      "loss": 1.1845,
+      "step": 109690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.130178220445726e-06,
+      "loss": 1.1518,
+      "step": 109695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.129425406053163e-06,
+      "loss": 1.5667,
+      "step": 109700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.12867259663205e-06,
+      "loss": 1.2852,
+      "step": 109705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.12791979218669e-06,
+      "loss": 1.3743,
+      "step": 109710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.127166992721372e-06,
+      "loss": 1.1985,
+      "step": 109715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.1264141982404e-06,
+      "loss": 1.5392,
+      "step": 109720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.125661408748074e-06,
+      "loss": 1.375,
+      "step": 109725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.124908624248694e-06,
+      "loss": 0.9772,
+      "step": 109730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.124155844746551e-06,
+      "loss": 1.0547,
+      "step": 109735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.123403070245953e-06,
+      "loss": 1.5518,
+      "step": 109740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.122650300751194e-06,
+      "loss": 1.1243,
+      "step": 109745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.121897536266573e-06,
+      "loss": 1.2191,
+      "step": 109750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.121144776796393e-06,
+      "loss": 1.7122,
+      "step": 109755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.120392022344946e-06,
+      "loss": 1.4337,
+      "step": 109760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.119639272916531e-06,
+      "loss": 1.3455,
+      "step": 109765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.118886528515455e-06,
+      "loss": 1.1061,
+      "step": 109770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.118133789146008e-06,
+      "loss": 1.361,
+      "step": 109775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.117381054812488e-06,
+      "loss": 1.3903,
+      "step": 109780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.116628325519202e-06,
+      "loss": 1.4334,
+      "step": 109785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.115875601270443e-06,
+      "loss": 1.5097,
+      "step": 109790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.115122882070508e-06,
+      "loss": 1.2302,
+      "step": 109795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.114370167923696e-06,
+      "loss": 1.2472,
+      "step": 109800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.11361745883431e-06,
+      "loss": 1.5671,
+      "step": 109805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.112864754806645e-06,
+      "loss": 1.0261,
+      "step": 109810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.112112055844996e-06,
+      "loss": 1.0738,
+      "step": 109815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.111359361953667e-06,
+      "loss": 1.0729,
+      "step": 109820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.110606673136955e-06,
+      "loss": 1.2272,
+      "step": 109825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.109853989399154e-06,
+      "loss": 1.1372,
+      "step": 109830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.109101310744569e-06,
+      "loss": 1.2133,
+      "step": 109835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.108348637177492e-06,
+      "loss": 1.0131,
+      "step": 109840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.107595968702222e-06,
+      "loss": 0.9717,
+      "step": 109845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.106843305323065e-06,
+      "loss": 1.3437,
+      "step": 109850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.10609064704431e-06,
+      "loss": 1.0992,
+      "step": 109855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.105337993870256e-06,
+      "loss": 1.0876,
+      "step": 109860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.104585345805205e-06,
+      "loss": 1.2276,
+      "step": 109865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.103832702853457e-06,
+      "loss": 1.3012,
+      "step": 109870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.103080065019302e-06,
+      "loss": 1.1569,
+      "step": 109875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.102327432307042e-06,
+      "loss": 1.1287,
+      "step": 109880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.101574804720978e-06,
+      "loss": 1.0346,
+      "step": 109885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.100822182265405e-06,
+      "loss": 1.391,
+      "step": 109890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.100069564944618e-06,
+      "loss": 1.2068,
+      "step": 109895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.099316952762918e-06,
+      "loss": 1.2351,
+      "step": 109900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.098564345724606e-06,
+      "loss": 1.2081,
+      "step": 109905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.097811743833971e-06,
+      "loss": 1.6701,
+      "step": 109910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.09705914709532e-06,
+      "loss": 1.3086,
+      "step": 109915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.096306555512951e-06,
+      "loss": 1.2508,
+      "step": 109920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.095553969091148e-06,
+      "loss": 1.5516,
+      "step": 109925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.094801387834226e-06,
+      "loss": 1.322,
+      "step": 109930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.094048811746472e-06,
+      "loss": 1.3387,
+      "step": 109935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.09329624083219e-06,
+      "loss": 1.0766,
+      "step": 109940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.092543675095668e-06,
+      "loss": 1.2845,
+      "step": 109945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.091791114541213e-06,
+      "loss": 1.2989,
+      "step": 109950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.09103855917312e-06,
+      "loss": 1.2882,
+      "step": 109955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.090286008995681e-06,
+      "loss": 1.1565,
+      "step": 109960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.089533464013204e-06,
+      "loss": 0.9999,
+      "step": 109965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.088780924229979e-06,
+      "loss": 0.9183,
+      "step": 109970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.088028389650302e-06,
+      "loss": 1.1241,
+      "step": 109975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.087275860278478e-06,
+      "loss": 1.2736,
+      "step": 109980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.086523336118797e-06,
+      "loss": 1.1618,
+      "step": 109985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.08577081717556e-06,
+      "loss": 1.2536,
+      "step": 109990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.085018303453061e-06,
+      "loss": 1.2306,
+      "step": 109995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.084265794955605e-06,
+      "loss": 1.2298,
+      "step": 110000
+    },
+    {
+      "epoch": 0.53,
+      "eval_loss": 1.2232636213302612,
+      "eval_runtime": 6474.4804,
+      "eval_samples_per_second": 3.569,
+      "eval_steps_per_second": 1.785,
+      "step": 110000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.083513291687479e-06,
+      "loss": 1.0708,
+      "step": 110005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.082760793652986e-06,
+      "loss": 1.202,
+      "step": 110010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.082008300856424e-06,
+      "loss": 1.1616,
+      "step": 110015
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.081255813302087e-06,
+      "loss": 1.1232,
+      "step": 110020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.080503330994272e-06,
+      "loss": 1.6539,
+      "step": 110025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.07975085393728e-06,
+      "loss": 1.1088,
+      "step": 110030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.078998382135406e-06,
+      "loss": 1.1235,
+      "step": 110035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.078245915592943e-06,
+      "loss": 1.7338,
+      "step": 110040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.077493454314195e-06,
+      "loss": 1.1991,
+      "step": 110045
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.076740998303453e-06,
+      "loss": 1.0172,
+      "step": 110050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.075988547565016e-06,
+      "loss": 1.5212,
+      "step": 110055
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.075236102103187e-06,
+      "loss": 0.9144,
+      "step": 110060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.074483661922253e-06,
+      "loss": 1.3758,
+      "step": 110065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.073731227026513e-06,
+      "loss": 0.9874,
+      "step": 110070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.072978797420272e-06,
+      "loss": 1.5328,
+      "step": 110075
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.072226373107817e-06,
+      "loss": 1.0987,
+      "step": 110080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.071473954093449e-06,
+      "loss": 1.406,
+      "step": 110085
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.07072154038146e-06,
+      "loss": 1.1377,
+      "step": 110090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.069969131976158e-06,
+      "loss": 1.0875,
+      "step": 110095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.069216728881827e-06,
+      "loss": 1.4123,
+      "step": 110100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.06846433110277e-06,
+      "loss": 0.9244,
+      "step": 110105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.067711938643284e-06,
+      "loss": 1.1781,
+      "step": 110110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.066959551507666e-06,
+      "loss": 1.2056,
+      "step": 110115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.066207169700207e-06,
+      "loss": 1.1114,
+      "step": 110120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.065454793225209e-06,
+      "loss": 1.7384,
+      "step": 110125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.06470242208697e-06,
+      "loss": 0.9689,
+      "step": 110130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.063950056289778e-06,
+      "loss": 1.2661,
+      "step": 110135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.063197695837937e-06,
+      "loss": 1.371,
+      "step": 110140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.06244534073574e-06,
+      "loss": 1.1674,
+      "step": 110145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.061692990987482e-06,
+      "loss": 1.0187,
+      "step": 110150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.060940646597468e-06,
+      "loss": 1.0213,
+      "step": 110155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.060188307569985e-06,
+      "loss": 1.0142,
+      "step": 110160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.059435973909334e-06,
+      "loss": 1.3643,
+      "step": 110165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.058683645619804e-06,
+      "loss": 1.0277,
+      "step": 110170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.057931322705702e-06,
+      "loss": 0.9874,
+      "step": 110175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.057179005171317e-06,
+      "loss": 1.4289,
+      "step": 110180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.056426693020946e-06,
+      "loss": 1.4248,
+      "step": 110185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.055674386258888e-06,
+      "loss": 0.9742,
+      "step": 110190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.054922084889438e-06,
+      "loss": 1.3555,
+      "step": 110195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.054169788916887e-06,
+      "loss": 1.5055,
+      "step": 110200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.053417498345536e-06,
+      "loss": 1.3204,
+      "step": 110205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.052665213179685e-06,
+      "loss": 2.2494,
+      "step": 110210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.051912933423621e-06,
+      "loss": 1.1685,
+      "step": 110215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.051160659081645e-06,
+      "loss": 1.2836,
+      "step": 110220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.050408390158054e-06,
+      "loss": 1.0102,
+      "step": 110225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.04965612665714e-06,
+      "loss": 1.2084,
+      "step": 110230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.048903868583199e-06,
+      "loss": 1.2708,
+      "step": 110235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.04815161594053e-06,
+      "loss": 1.1324,
+      "step": 110240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.04739936873343e-06,
+      "loss": 1.3014,
+      "step": 110245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.046647126966188e-06,
+      "loss": 1.5992,
+      "step": 110250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.045894890643104e-06,
+      "loss": 1.3187,
+      "step": 110255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.045142659768476e-06,
+      "loss": 1.3937,
+      "step": 110260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.044390434346591e-06,
+      "loss": 1.1706,
+      "step": 110265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.043638214381759e-06,
+      "loss": 1.4705,
+      "step": 110270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.042885999878262e-06,
+      "loss": 1.2427,
+      "step": 110275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.0421337908404e-06,
+      "loss": 1.2133,
+      "step": 110280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.041381587272473e-06,
+      "loss": 1.5186,
+      "step": 110285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.04062938917877e-06,
+      "loss": 0.9368,
+      "step": 110290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.039877196563587e-06,
+      "loss": 1.3882,
+      "step": 110295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.039125009431225e-06,
+      "loss": 1.1417,
+      "step": 110300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.038372827785978e-06,
+      "loss": 1.2685,
+      "step": 110305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.037620651632137e-06,
+      "loss": 1.7205,
+      "step": 110310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.036868480973999e-06,
+      "loss": 1.2481,
+      "step": 110315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.036116315815862e-06,
+      "loss": 1.1078,
+      "step": 110320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.035364156162018e-06,
+      "loss": 1.3376,
+      "step": 110325
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.034612002016761e-06,
+      "loss": 1.0237,
+      "step": 110330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.033859853384391e-06,
+      "loss": 1.3297,
+      "step": 110335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.033107710269204e-06,
+      "loss": 1.2176,
+      "step": 110340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.032355572675486e-06,
+      "loss": 1.2096,
+      "step": 110345
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.031603440607542e-06,
+      "loss": 1.2652,
+      "step": 110350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.030851314069664e-06,
+      "loss": 1.1272,
+      "step": 110355
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.030099193066142e-06,
+      "loss": 1.4993,
+      "step": 110360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.02934707760128e-06,
+      "loss": 1.2624,
+      "step": 110365
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.028594967679366e-06,
+      "loss": 1.101,
+      "step": 110370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.0278428633047e-06,
+      "loss": 1.3557,
+      "step": 110375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.027090764481568e-06,
+      "loss": 1.3404,
+      "step": 110380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.026338671214274e-06,
+      "loss": 1.126,
+      "step": 110385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.025586583507112e-06,
+      "loss": 1.5293,
+      "step": 110390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.02483450136437e-06,
+      "loss": 1.0944,
+      "step": 110395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.024082424790353e-06,
+      "loss": 1.1782,
+      "step": 110400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.023330353789347e-06,
+      "loss": 1.3812,
+      "step": 110405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.022578288365649e-06,
+      "loss": 1.0969,
+      "step": 110410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.02182622852356e-06,
+      "loss": 1.2244,
+      "step": 110415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.021074174267364e-06,
+      "loss": 1.1756,
+      "step": 110420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.020322125601362e-06,
+      "loss": 1.36,
+      "step": 110425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.019570082529848e-06,
+      "loss": 0.996,
+      "step": 110430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.018818045057118e-06,
+      "loss": 1.5508,
+      "step": 110435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.018066013187462e-06,
+      "loss": 1.2284,
+      "step": 110440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.017313986925178e-06,
+      "loss": 1.4676,
+      "step": 110445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.01656196627456e-06,
+      "loss": 1.2057,
+      "step": 110450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.015809951239905e-06,
+      "loss": 1.2986,
+      "step": 110455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.015057941825499e-06,
+      "loss": 1.2158,
+      "step": 110460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.014305938035646e-06,
+      "loss": 1.1163,
+      "step": 110465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.013553939874635e-06,
+      "loss": 1.3989,
+      "step": 110470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.012801947346762e-06,
+      "loss": 1.0931,
+      "step": 110475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.012049960456319e-06,
+      "loss": 1.1559,
+      "step": 110480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.011297979207604e-06,
+      "loss": 1.0134,
+      "step": 110485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.010546003604907e-06,
+      "loss": 0.8751,
+      "step": 110490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.009794033652527e-06,
+      "loss": 1.0609,
+      "step": 110495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.009042069354756e-06,
+      "loss": 1.3098,
+      "step": 110500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.008290110715884e-06,
+      "loss": 1.4411,
+      "step": 110505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.007538157740212e-06,
+      "loss": 1.4584,
+      "step": 110510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.006786210432032e-06,
+      "loss": 1.2397,
+      "step": 110515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.006034268795636e-06,
+      "loss": 1.1713,
+      "step": 110520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.005282332835316e-06,
+      "loss": 1.2339,
+      "step": 110525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.004530402555373e-06,
+      "loss": 1.4107,
+      "step": 110530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.003778477960096e-06,
+      "loss": 1.2484,
+      "step": 110535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.003026559053776e-06,
+      "loss": 1.1755,
+      "step": 110540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.002274645840714e-06,
+      "loss": 1.1125,
+      "step": 110545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.001522738325202e-06,
+      "loss": 1.1261,
+      "step": 110550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.000770836511527e-06,
+      "loss": 1.4757,
+      "step": 110555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.00001894040399e-06,
+      "loss": 1.2062,
+      "step": 110560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.999267050006887e-06,
+      "loss": 0.942,
+      "step": 110565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.9985151653245e-06,
+      "loss": 1.1179,
+      "step": 110570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.997763286361136e-06,
+      "loss": 1.2213,
+      "step": 110575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.997011413121081e-06,
+      "loss": 1.328,
+      "step": 110580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.996259545608628e-06,
+      "loss": 1.2589,
+      "step": 110585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.995507683828077e-06,
+      "loss": 1.3629,
+      "step": 110590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.994755827783714e-06,
+      "loss": 1.5266,
+      "step": 110595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.99400397747984e-06,
+      "loss": 0.9589,
+      "step": 110600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.993252132920736e-06,
+      "loss": 1.3918,
+      "step": 110605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.992500294110713e-06,
+      "loss": 0.98,
+      "step": 110610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.99174846105405e-06,
+      "loss": 1.0559,
+      "step": 110615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.990996633755045e-06,
+      "loss": 1.2487,
+      "step": 110620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.990244812217997e-06,
+      "loss": 0.9362,
+      "step": 110625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.989492996447192e-06,
+      "loss": 1.2464,
+      "step": 110630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.988741186446922e-06,
+      "loss": 1.297,
+      "step": 110635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.987989382221487e-06,
+      "loss": 1.2165,
+      "step": 110640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.987237583775177e-06,
+      "loss": 1.1328,
+      "step": 110645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.986485791112284e-06,
+      "loss": 1.1916,
+      "step": 110650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.985734004237102e-06,
+      "loss": 1.3405,
+      "step": 110655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.984982223153929e-06,
+      "loss": 1.1693,
+      "step": 110660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.984230447867048e-06,
+      "loss": 1.03,
+      "step": 110665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.983478678380757e-06,
+      "loss": 1.2748,
+      "step": 110670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.982726914699352e-06,
+      "loss": 1.1382,
+      "step": 110675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.981975156827125e-06,
+      "loss": 1.1407,
+      "step": 110680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.981223404768363e-06,
+      "loss": 1.3576,
+      "step": 110685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.980471658527366e-06,
+      "loss": 1.2573,
+      "step": 110690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.979719918108424e-06,
+      "loss": 1.3797,
+      "step": 110695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.978968183515828e-06,
+      "loss": 1.4587,
+      "step": 110700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.978216454753877e-06,
+      "loss": 1.0288,
+      "step": 110705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.977464731826858e-06,
+      "loss": 1.4923,
+      "step": 110710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.976713014739062e-06,
+      "loss": 1.2226,
+      "step": 110715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.97596130349479e-06,
+      "loss": 1.5247,
+      "step": 110720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.975209598098327e-06,
+      "loss": 1.2874,
+      "step": 110725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.974457898553968e-06,
+      "loss": 1.1646,
+      "step": 110730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.973706204866006e-06,
+      "loss": 1.1989,
+      "step": 110735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.972954517038736e-06,
+      "loss": 1.1405,
+      "step": 110740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.972202835076446e-06,
+      "loss": 1.3257,
+      "step": 110745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.97145115898343e-06,
+      "loss": 1.2311,
+      "step": 110750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.970699488763985e-06,
+      "loss": 1.3598,
+      "step": 110755
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.969947824422398e-06,
+      "loss": 1.1533,
+      "step": 110760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.969196165962958e-06,
+      "loss": 1.1697,
+      "step": 110765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.968444513389967e-06,
+      "loss": 1.1634,
+      "step": 110770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.967692866707714e-06,
+      "loss": 1.1512,
+      "step": 110775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.966941225920485e-06,
+      "loss": 1.0934,
+      "step": 110780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.96618959103258e-06,
+      "loss": 0.9804,
+      "step": 110785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.965437962048289e-06,
+      "loss": 1.3345,
+      "step": 110790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.9646863389719e-06,
+      "loss": 1.4687,
+      "step": 110795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.963934721807714e-06,
+      "loss": 1.1347,
+      "step": 110800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.963183110560017e-06,
+      "loss": 1.1647,
+      "step": 110805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.962431505233102e-06,
+      "loss": 1.0586,
+      "step": 110810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.961679905831257e-06,
+      "loss": 1.2199,
+      "step": 110815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.960928312358782e-06,
+      "loss": 1.4739,
+      "step": 110820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.960176724819964e-06,
+      "loss": 1.0892,
+      "step": 110825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.959425143219096e-06,
+      "loss": 1.4556,
+      "step": 110830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.958673567560473e-06,
+      "loss": 1.4092,
+      "step": 110835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.957921997848382e-06,
+      "loss": 1.4102,
+      "step": 110840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.957170434087115e-06,
+      "loss": 1.2182,
+      "step": 110845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.956418876280967e-06,
+      "loss": 1.3318,
+      "step": 110850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.95566732443423e-06,
+      "loss": 1.1554,
+      "step": 110855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.954915778551191e-06,
+      "loss": 1.4551,
+      "step": 110860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.954164238636149e-06,
+      "loss": 1.3938,
+      "step": 110865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.953412704693391e-06,
+      "loss": 1.1824,
+      "step": 110870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.952661176727207e-06,
+      "loss": 1.3215,
+      "step": 110875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.951909654741892e-06,
+      "loss": 0.8213,
+      "step": 110880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.951158138741737e-06,
+      "loss": 1.9927,
+      "step": 110885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.950406628731035e-06,
+      "loss": 0.9782,
+      "step": 110890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.949655124714072e-06,
+      "loss": 1.0625,
+      "step": 110895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.948903626695144e-06,
+      "loss": 1.3077,
+      "step": 110900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.948152134678546e-06,
+      "loss": 1.1784,
+      "step": 110905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.947400648668556e-06,
+      "loss": 1.0749,
+      "step": 110910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.946649168669482e-06,
+      "loss": 1.1309,
+      "step": 110915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.945897694685605e-06,
+      "loss": 1.3535,
+      "step": 110920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.945146226721219e-06,
+      "loss": 1.4813,
+      "step": 110925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.944394764780619e-06,
+      "loss": 1.3076,
+      "step": 110930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.943643308868089e-06,
+      "loss": 1.287,
+      "step": 110935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.942891858987924e-06,
+      "loss": 1.3837,
+      "step": 110940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.942140415144415e-06,
+      "loss": 1.1023,
+      "step": 110945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.941388977341856e-06,
+      "loss": 0.9172,
+      "step": 110950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.940637545584533e-06,
+      "loss": 1.1132,
+      "step": 110955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.939886119876738e-06,
+      "loss": 1.321,
+      "step": 110960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.939134700222766e-06,
+      "loss": 1.0375,
+      "step": 110965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.938383286626906e-06,
+      "loss": 1.1513,
+      "step": 110970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.937631879093445e-06,
+      "loss": 1.1441,
+      "step": 110975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.93688047762668e-06,
+      "loss": 0.9705,
+      "step": 110980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.9361290822309e-06,
+      "loss": 1.6007,
+      "step": 110985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.935377692910393e-06,
+      "loss": 1.0236,
+      "step": 110990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.934626309669452e-06,
+      "loss": 1.22,
+      "step": 110995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.933874932512371e-06,
+      "loss": 1.167,
+      "step": 111000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.933123561443431e-06,
+      "loss": 1.1945,
+      "step": 111005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.932372196466935e-06,
+      "loss": 1.2271,
+      "step": 111010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.931620837587167e-06,
+      "loss": 1.1644,
+      "step": 111015
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.930869484808416e-06,
+      "loss": 1.3585,
+      "step": 111020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.93011813813498e-06,
+      "loss": 1.3118,
+      "step": 111025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.929366797571143e-06,
+      "loss": 1.2326,
+      "step": 111030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.928615463121195e-06,
+      "loss": 1.0312,
+      "step": 111035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.92786413478943e-06,
+      "loss": 1.1472,
+      "step": 111040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.927112812580139e-06,
+      "loss": 1.1318,
+      "step": 111045
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.926361496497611e-06,
+      "loss": 1.0466,
+      "step": 111050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.925610186546132e-06,
+      "loss": 1.0028,
+      "step": 111055
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.924858882730004e-06,
+      "loss": 1.2109,
+      "step": 111060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.924107585053506e-06,
+      "loss": 1.1759,
+      "step": 111065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.923356293520928e-06,
+      "loss": 1.5232,
+      "step": 111070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.92260500813657e-06,
+      "loss": 1.1195,
+      "step": 111075
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.921853728904718e-06,
+      "loss": 1.1323,
+      "step": 111080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.921102455829656e-06,
+      "loss": 1.3255,
+      "step": 111085
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.920351188915684e-06,
+      "loss": 1.2815,
+      "step": 111090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.919599928167085e-06,
+      "loss": 1.1478,
+      "step": 111095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.918848673588153e-06,
+      "loss": 1.1149,
+      "step": 111100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.918097425183176e-06,
+      "loss": 1.0477,
+      "step": 111105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.917346182956443e-06,
+      "loss": 1.2582,
+      "step": 111110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.916594946912249e-06,
+      "loss": 1.1609,
+      "step": 111115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.915843717054877e-06,
+      "loss": 1.2101,
+      "step": 111120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.915092493388621e-06,
+      "loss": 1.0972,
+      "step": 111125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.914341275917773e-06,
+      "loss": 1.2409,
+      "step": 111130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.913590064646615e-06,
+      "loss": 1.3405,
+      "step": 111135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.91283885957945e-06,
+      "loss": 1.4192,
+      "step": 111140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.912087660720554e-06,
+      "loss": 0.9888,
+      "step": 111145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.911336468074222e-06,
+      "loss": 1.3599,
+      "step": 111150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.910585281644746e-06,
+      "loss": 0.9622,
+      "step": 111155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.909834101436417e-06,
+      "loss": 1.0614,
+      "step": 111160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.909082927453518e-06,
+      "loss": 1.2819,
+      "step": 111165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.908331759700343e-06,
+      "loss": 1.2433,
+      "step": 111170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.907580598181182e-06,
+      "loss": 1.1204,
+      "step": 111175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.906829442900323e-06,
+      "loss": 1.2406,
+      "step": 111180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.906078293862054e-06,
+      "loss": 1.1096,
+      "step": 111185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.905327151070667e-06,
+      "loss": 1.1904,
+      "step": 111190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.904576014530454e-06,
+      "loss": 1.4133,
+      "step": 111195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.903824884245697e-06,
+      "loss": 1.0304,
+      "step": 111200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.90307376022069e-06,
+      "loss": 1.5451,
+      "step": 111205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.902322642459725e-06,
+      "loss": 1.2914,
+      "step": 111210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.90157153096708e-06,
+      "loss": 1.2831,
+      "step": 111215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.900820425747061e-06,
+      "loss": 1.0781,
+      "step": 111220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.900069326803947e-06,
+      "loss": 1.0983,
+      "step": 111225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.899318234142025e-06,
+      "loss": 0.8594,
+      "step": 111230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.898567147765593e-06,
+      "loss": 1.1438,
+      "step": 111235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.897816067678932e-06,
+      "loss": 1.225,
+      "step": 111240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.897064993886337e-06,
+      "loss": 1.2113,
+      "step": 111245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.896313926392088e-06,
+      "loss": 1.3091,
+      "step": 111250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.895562865200485e-06,
+      "loss": 1.3231,
+      "step": 111255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.89481181031581e-06,
+      "loss": 1.2862,
+      "step": 111260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.894060761742353e-06,
+      "loss": 1.3897,
+      "step": 111265
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.893309719484408e-06,
+      "loss": 1.1125,
+      "step": 111270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.892558683546258e-06,
+      "loss": 1.1265,
+      "step": 111275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.891807653932191e-06,
+      "loss": 1.0681,
+      "step": 111280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.8910566306465e-06,
+      "loss": 1.1307,
+      "step": 111285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.890305613693475e-06,
+      "loss": 1.0277,
+      "step": 111290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.889554603077397e-06,
+      "loss": 1.5693,
+      "step": 111295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.888803598802561e-06,
+      "loss": 1.0111,
+      "step": 111300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.888052600873258e-06,
+      "loss": 1.0497,
+      "step": 111305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.887301609293765e-06,
+      "loss": 1.1281,
+      "step": 111310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.886550624068384e-06,
+      "loss": 1.2237,
+      "step": 111315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.885799645201397e-06,
+      "loss": 0.9194,
+      "step": 111320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.885048672697095e-06,
+      "loss": 1.3679,
+      "step": 111325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.88429770655976e-06,
+      "loss": 1.2597,
+      "step": 111330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.883546746793687e-06,
+      "loss": 1.2139,
+      "step": 111335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.882795793403167e-06,
+      "loss": 1.1464,
+      "step": 111340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.882044846392476e-06,
+      "loss": 1.1264,
+      "step": 111345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.881293905765917e-06,
+      "loss": 1.5245,
+      "step": 111350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.880542971527769e-06,
+      "loss": 1.3076,
+      "step": 111355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.87979204368232e-06,
+      "loss": 1.3167,
+      "step": 111360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.879041122233866e-06,
+      "loss": 1.1641,
+      "step": 111365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.878290207186687e-06,
+      "loss": 1.1081,
+      "step": 111370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.877539298545073e-06,
+      "loss": 1.0727,
+      "step": 111375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.876788396313314e-06,
+      "loss": 1.2204,
+      "step": 111380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.876037500495703e-06,
+      "loss": 1.0472,
+      "step": 111385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.875286611096517e-06,
+      "loss": 1.3407,
+      "step": 111390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.874535728120049e-06,
+      "loss": 1.0769,
+      "step": 111395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.87378485157059e-06,
+      "loss": 1.3258,
+      "step": 111400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.873033981452424e-06,
+      "loss": 1.301,
+      "step": 111405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.872283117769838e-06,
+      "loss": 1.2564,
+      "step": 111410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.871532260527125e-06,
+      "loss": 1.1978,
+      "step": 111415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.87078140972857e-06,
+      "loss": 1.4935,
+      "step": 111420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.870030565378458e-06,
+      "loss": 1.2004,
+      "step": 111425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.86927972748108e-06,
+      "loss": 1.0756,
+      "step": 111430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.868528896040725e-06,
+      "loss": 1.3603,
+      "step": 111435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.867778071061673e-06,
+      "loss": 1.1656,
+      "step": 111440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.867027252548224e-06,
+      "loss": 1.1349,
+      "step": 111445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.866276440504656e-06,
+      "loss": 1.173,
+      "step": 111450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.865525634935257e-06,
+      "loss": 0.9811,
+      "step": 111455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.864774835844322e-06,
+      "loss": 1.2883,
+      "step": 111460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.86402404323613e-06,
+      "loss": 1.2283,
+      "step": 111465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.863273257114972e-06,
+      "loss": 1.2885,
+      "step": 111470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.862522477485133e-06,
+      "loss": 1.2762,
+      "step": 111475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.861771704350907e-06,
+      "loss": 1.3827,
+      "step": 111480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.861020937716574e-06,
+      "loss": 1.3408,
+      "step": 111485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.860270177586422e-06,
+      "loss": 1.1808,
+      "step": 111490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.859519423964743e-06,
+      "loss": 1.7796,
+      "step": 111495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.858768676855824e-06,
+      "loss": 1.1717,
+      "step": 111500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.858017936263944e-06,
+      "loss": 1.3359,
+      "step": 111505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.857267202193399e-06,
+      "loss": 1.1507,
+      "step": 111510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.856516474648474e-06,
+      "loss": 1.0993,
+      "step": 111515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.855765753633452e-06,
+      "loss": 0.9574,
+      "step": 111520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.855015039152624e-06,
+      "loss": 0.8463,
+      "step": 111525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.854264331210278e-06,
+      "loss": 1.0588,
+      "step": 111530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.853513629810699e-06,
+      "loss": 1.266,
+      "step": 111535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.85276293495817e-06,
+      "loss": 1.0752,
+      "step": 111540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.852012246656984e-06,
+      "loss": 1.37,
+      "step": 111545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.851261564911427e-06,
+      "loss": 1.1022,
+      "step": 111550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.850510889725778e-06,
+      "loss": 1.0986,
+      "step": 111555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.849760221104337e-06,
+      "loss": 1.253,
+      "step": 111560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.849009559051382e-06,
+      "loss": 1.1168,
+      "step": 111565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.848258903571199e-06,
+      "loss": 1.2059,
+      "step": 111570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.847508254668082e-06,
+      "loss": 1.3623,
+      "step": 111575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.84675761234631e-06,
+      "loss": 1.3322,
+      "step": 111580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.846006976610168e-06,
+      "loss": 1.189,
+      "step": 111585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.845256347463951e-06,
+      "loss": 1.367,
+      "step": 111590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.844505724911945e-06,
+      "loss": 1.0791,
+      "step": 111595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.843755108958427e-06,
+      "loss": 1.1615,
+      "step": 111600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.843004499607692e-06,
+      "loss": 1.1946,
+      "step": 111605
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.842253896864024e-06,
+      "loss": 1.1667,
+      "step": 111610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.841503300731709e-06,
+      "loss": 1.1428,
+      "step": 111615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.84075271121503e-06,
+      "loss": 1.1586,
+      "step": 111620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.840002128318279e-06,
+      "loss": 1.4541,
+      "step": 111625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.83925155204574e-06,
+      "loss": 0.9812,
+      "step": 111630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.838500982401696e-06,
+      "loss": 0.9987,
+      "step": 111635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.837750419390439e-06,
+      "loss": 1.1831,
+      "step": 111640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.836999863016255e-06,
+      "loss": 1.2272,
+      "step": 111645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.836249313283419e-06,
+      "loss": 1.1824,
+      "step": 111650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.835498770196232e-06,
+      "loss": 1.0648,
+      "step": 111655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.834748233758972e-06,
+      "loss": 1.3174,
+      "step": 111660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.833997703975925e-06,
+      "loss": 1.1077,
+      "step": 111665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.83324718085138e-06,
+      "loss": 1.3984,
+      "step": 111670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.832496664389621e-06,
+      "loss": 1.223,
+      "step": 111675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.831746154594934e-06,
+      "loss": 1.1859,
+      "step": 111680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.830995651471602e-06,
+      "loss": 1.3244,
+      "step": 111685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.830245155023918e-06,
+      "loss": 1.0979,
+      "step": 111690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.829494665256163e-06,
+      "loss": 1.2013,
+      "step": 111695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.828744182172617e-06,
+      "loss": 1.1176,
+      "step": 111700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.827993705777579e-06,
+      "loss": 1.2788,
+      "step": 111705
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.827243236075323e-06,
+      "loss": 1.4781,
+      "step": 111710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.82649277307014e-06,
+      "loss": 1.3506,
+      "step": 111715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.825742316766313e-06,
+      "loss": 1.3843,
+      "step": 111720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.824991867168133e-06,
+      "loss": 1.4314,
+      "step": 111725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.824241424279877e-06,
+      "loss": 1.0931,
+      "step": 111730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.823490988105836e-06,
+      "loss": 1.1219,
+      "step": 111735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.822740558650296e-06,
+      "loss": 1.0394,
+      "step": 111740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.821990135917536e-06,
+      "loss": 1.3642,
+      "step": 111745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.821239719911853e-06,
+      "loss": 1.8485,
+      "step": 111750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.820489310637522e-06,
+      "loss": 1.1554,
+      "step": 111755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.819738908098832e-06,
+      "loss": 1.3311,
+      "step": 111760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.818988512300065e-06,
+      "loss": 1.5251,
+      "step": 111765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.81823812324551e-06,
+      "loss": 1.2671,
+      "step": 111770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.817487740939453e-06,
+      "loss": 1.3245,
+      "step": 111775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.816737365386172e-06,
+      "loss": 1.2415,
+      "step": 111780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.815986996589963e-06,
+      "loss": 1.0245,
+      "step": 111785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.815236634555103e-06,
+      "loss": 1.3629,
+      "step": 111790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.814486279285877e-06,
+      "loss": 1.0601,
+      "step": 111795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.813735930786573e-06,
+      "loss": 1.3442,
+      "step": 111800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.812985589061478e-06,
+      "loss": 1.2288,
+      "step": 111805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.812235254114869e-06,
+      "loss": 1.3146,
+      "step": 111810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.811484925951037e-06,
+      "loss": 1.1068,
+      "step": 111815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.810734604574269e-06,
+      "loss": 1.3855,
+      "step": 111820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.809984289988843e-06,
+      "loss": 1.3283,
+      "step": 111825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.809233982199045e-06,
+      "loss": 1.3381,
+      "step": 111830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.808483681209164e-06,
+      "loss": 1.2162,
+      "step": 111835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.807733387023483e-06,
+      "loss": 1.0065,
+      "step": 111840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.806983099646283e-06,
+      "loss": 1.3059,
+      "step": 111845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.80623281908185e-06,
+      "loss": 1.0097,
+      "step": 111850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.805482545334475e-06,
+      "loss": 0.9486,
+      "step": 111855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.80473227840843e-06,
+      "loss": 1.0659,
+      "step": 111860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.803982018308012e-06,
+      "loss": 1.4249,
+      "step": 111865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.803231765037499e-06,
+      "loss": 1.0722,
+      "step": 111870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.802481518601174e-06,
+      "loss": 1.3964,
+      "step": 111875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.801731279003328e-06,
+      "loss": 1.3699,
+      "step": 111880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.800981046248238e-06,
+      "loss": 1.1468,
+      "step": 111885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.80023082034019e-06,
+      "loss": 1.6427,
+      "step": 111890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.79948060128347e-06,
+      "loss": 1.4066,
+      "step": 111895
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.798730389082363e-06,
+      "loss": 1.5636,
+      "step": 111900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.797980183741153e-06,
+      "loss": 1.1065,
+      "step": 111905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.797229985264117e-06,
+      "loss": 0.8373,
+      "step": 111910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.796479793655552e-06,
+      "loss": 1.0176,
+      "step": 111915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.79572960891973e-06,
+      "loss": 1.2747,
+      "step": 111920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.794979431060939e-06,
+      "loss": 1.0358,
+      "step": 111925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.794229260083464e-06,
+      "loss": 1.2271,
+      "step": 111930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.793479095991592e-06,
+      "loss": 1.1067,
+      "step": 111935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.792728938789598e-06,
+      "loss": 1.3152,
+      "step": 111940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.791978788481774e-06,
+      "loss": 0.9027,
+      "step": 111945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.791228645072405e-06,
+      "loss": 1.362,
+      "step": 111950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.79047850856576e-06,
+      "loss": 1.5368,
+      "step": 111955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.789728378966142e-06,
+      "loss": 1.1237,
+      "step": 111960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.788978256277824e-06,
+      "loss": 1.3274,
+      "step": 111965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.788228140505093e-06,
+      "loss": 1.4203,
+      "step": 111970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.787478031652229e-06,
+      "loss": 1.3621,
+      "step": 111975
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.786727929723516e-06,
+      "loss": 1.0597,
+      "step": 111980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.785977834723243e-06,
+      "loss": 1.0107,
+      "step": 111985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.785227746655683e-06,
+      "loss": 1.635,
+      "step": 111990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.784477665525133e-06,
+      "loss": 1.1915,
+      "step": 111995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.783727591335866e-06,
+      "loss": 1.2404,
+      "step": 112000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.782977524092167e-06,
+      "loss": 1.0177,
+      "step": 112005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.782227463798326e-06,
+      "loss": 1.1488,
+      "step": 112010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.781477410458618e-06,
+      "loss": 1.249,
+      "step": 112015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.780727364077327e-06,
+      "loss": 1.1071,
+      "step": 112020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.779977324658741e-06,
+      "loss": 1.3858,
+      "step": 112025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.779227292207144e-06,
+      "loss": 1.1565,
+      "step": 112030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.77847726672681e-06,
+      "loss": 1.4413,
+      "step": 112035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.77772724822203e-06,
+      "loss": 1.1933,
+      "step": 112040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.776977236697087e-06,
+      "loss": 1.1056,
+      "step": 112045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.77622723215626e-06,
+      "loss": 1.2228,
+      "step": 112050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.775477234603831e-06,
+      "loss": 0.938,
+      "step": 112055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.77472724404409e-06,
+      "loss": 1.1998,
+      "step": 112060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.773977260481315e-06,
+      "loss": 1.1796,
+      "step": 112065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.773227283919786e-06,
+      "loss": 1.5726,
+      "step": 112070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.772477314363791e-06,
+      "loss": 1.5097,
+      "step": 112075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.771727351817611e-06,
+      "loss": 1.1732,
+      "step": 112080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.770977396285526e-06,
+      "loss": 1.007,
+      "step": 112085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.770227447771826e-06,
+      "loss": 1.1183,
+      "step": 112090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.769477506280786e-06,
+      "loss": 1.2847,
+      "step": 112095
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.76872757181669e-06,
+      "loss": 1.1614,
+      "step": 112100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.767977644383825e-06,
+      "loss": 0.958,
+      "step": 112105
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.767227723986469e-06,
+      "loss": 1.0635,
+      "step": 112110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.766477810628905e-06,
+      "loss": 1.4083,
+      "step": 112115
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.765727904315412e-06,
+      "loss": 1.4716,
+      "step": 112120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.764978005050285e-06,
+      "loss": 1.5703,
+      "step": 112125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.764228112837794e-06,
+      "loss": 1.2864,
+      "step": 112130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.763478227682222e-06,
+      "loss": 0.9446,
+      "step": 112135
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.762728349587858e-06,
+      "loss": 1.2413,
+      "step": 112140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.761978478558982e-06,
+      "loss": 1.1885,
+      "step": 112145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.761228614599871e-06,
+      "loss": 1.1303,
+      "step": 112150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.760478757714812e-06,
+      "loss": 1.0762,
+      "step": 112155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.759728907908089e-06,
+      "loss": 1.6048,
+      "step": 112160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.758979065183976e-06,
+      "loss": 1.1681,
+      "step": 112165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.758229229546762e-06,
+      "loss": 1.4808,
+      "step": 112170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.757479401000728e-06,
+      "loss": 1.0648,
+      "step": 112175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.756729579550152e-06,
+      "loss": 1.1689,
+      "step": 112180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.755979765199322e-06,
+      "loss": 1.1794,
+      "step": 112185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.755229957952516e-06,
+      "loss": 1.3879,
+      "step": 112190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.754480157814021e-06,
+      "loss": 0.9499,
+      "step": 112195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.753730364788104e-06,
+      "loss": 1.1502,
+      "step": 112200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.752980578879065e-06,
+      "loss": 1.288,
+      "step": 112205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.752230800091175e-06,
+      "loss": 1.2858,
+      "step": 112210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.751481028428715e-06,
+      "loss": 1.3675,
+      "step": 112215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.750731263895977e-06,
+      "loss": 1.2672,
+      "step": 112220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.749981506497231e-06,
+      "loss": 1.455,
+      "step": 112225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.74923175623676e-06,
+      "loss": 1.3385,
+      "step": 112230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.748482013118853e-06,
+      "loss": 0.9882,
+      "step": 112235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.747732277147788e-06,
+      "loss": 1.69,
+      "step": 112240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.74698254832784e-06,
+      "loss": 1.1122,
+      "step": 112245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.7462328266633e-06,
+      "loss": 1.2732,
+      "step": 112250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.745483112158444e-06,
+      "loss": 1.28,
+      "step": 112255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.74473340481755e-06,
+      "loss": 1.5305,
+      "step": 112260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.74398370464491e-06,
+      "loss": 0.9508,
+      "step": 112265
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.743234011644795e-06,
+      "loss": 1.3301,
+      "step": 112270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.742484325821492e-06,
+      "loss": 1.4997,
+      "step": 112275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.741734647179277e-06,
+      "loss": 1.1539,
+      "step": 112280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.740984975722434e-06,
+      "loss": 1.1684,
+      "step": 112285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.740235311455247e-06,
+      "loss": 1.5781,
+      "step": 112290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.739485654381989e-06,
+      "loss": 1.3351,
+      "step": 112295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.73873600450695e-06,
+      "loss": 1.2515,
+      "step": 112300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.737986361834406e-06,
+      "loss": 1.1823,
+      "step": 112305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.737236726368637e-06,
+      "loss": 1.1689,
+      "step": 112310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.736487098113928e-06,
+      "loss": 1.0056,
+      "step": 112315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.735737477074556e-06,
+      "loss": 1.2967,
+      "step": 112320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.734987863254802e-06,
+      "loss": 1.2666,
+      "step": 112325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.734238256658948e-06,
+      "loss": 1.226,
+      "step": 112330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.733488657291278e-06,
+      "loss": 1.0346,
+      "step": 112335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.732739065156067e-06,
+      "loss": 1.0781,
+      "step": 112340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.731989480257593e-06,
+      "loss": 1.3457,
+      "step": 112345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.731239902600147e-06,
+      "loss": 1.3784,
+      "step": 112350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.730490332188e-06,
+      "loss": 1.1375,
+      "step": 112355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.729740769025436e-06,
+      "loss": 1.0514,
+      "step": 112360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.728991213116738e-06,
+      "loss": 1.2897,
+      "step": 112365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.728241664466184e-06,
+      "loss": 1.1163,
+      "step": 112370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.72749212307805e-06,
+      "loss": 1.1166,
+      "step": 112375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.726742588956622e-06,
+      "loss": 1.3483,
+      "step": 112380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.72599306210618e-06,
+      "loss": 1.2031,
+      "step": 112385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.725243542531001e-06,
+      "loss": 1.4198,
+      "step": 112390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.72449403023537e-06,
+      "loss": 1.4325,
+      "step": 112395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.723744525223563e-06,
+      "loss": 1.1992,
+      "step": 112400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.722995027499859e-06,
+      "loss": 1.379,
+      "step": 112405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.722245537068544e-06,
+      "loss": 1.1714,
+      "step": 112410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.721496053933893e-06,
+      "loss": 1.2019,
+      "step": 112415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.720746578100185e-06,
+      "loss": 1.1209,
+      "step": 112420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.719997109571702e-06,
+      "loss": 1.5014,
+      "step": 112425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.719247648352728e-06,
+      "loss": 1.2492,
+      "step": 112430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.718498194447535e-06,
+      "loss": 1.1847,
+      "step": 112435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.717748747860406e-06,
+      "loss": 1.255,
+      "step": 112440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.716999308595624e-06,
+      "loss": 1.2163,
+      "step": 112445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.716249876657466e-06,
+      "loss": 1.3869,
+      "step": 112450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.71550045205021e-06,
+      "loss": 1.1055,
+      "step": 112455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.714751034778137e-06,
+      "loss": 1.2895,
+      "step": 112460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.71400162484553e-06,
+      "loss": 1.187,
+      "step": 112465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.713252222256662e-06,
+      "loss": 1.1679,
+      "step": 112470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.712502827015818e-06,
+      "loss": 1.2019,
+      "step": 112475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.711753439127274e-06,
+      "loss": 1.0224,
+      "step": 112480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.711004058595314e-06,
+      "loss": 1.0511,
+      "step": 112485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.710254685424209e-06,
+      "loss": 1.0183,
+      "step": 112490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.709505319618246e-06,
+      "loss": 1.0874,
+      "step": 112495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.708755961181703e-06,
+      "loss": 1.3625,
+      "step": 112500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.708006610118855e-06,
+      "loss": 1.4867,
+      "step": 112505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.707257266433986e-06,
+      "loss": 1.1146,
+      "step": 112510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.706507930131373e-06,
+      "loss": 1.6844,
+      "step": 112515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.705758601215293e-06,
+      "loss": 1.1583,
+      "step": 112520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.705009279690032e-06,
+      "loss": 1.297,
+      "step": 112525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.704259965559862e-06,
+      "loss": 1.4162,
+      "step": 112530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.703510658829065e-06,
+      "loss": 0.9382,
+      "step": 112535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.702761359501917e-06,
+      "loss": 1.219,
+      "step": 112540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.702012067582706e-06,
+      "loss": 1.0425,
+      "step": 112545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.701262783075697e-06,
+      "loss": 1.3174,
+      "step": 112550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.70051350598518e-06,
+      "loss": 1.1925,
+      "step": 112555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.69976423631543e-06,
+      "loss": 1.1498,
+      "step": 112560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.699014974070726e-06,
+      "loss": 1.2503,
+      "step": 112565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.69826571925534e-06,
+      "loss": 1.3641,
+      "step": 112570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.697516471873562e-06,
+      "loss": 1.227,
+      "step": 112575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.696767231929668e-06,
+      "loss": 1.2263,
+      "step": 112580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.69601799942793e-06,
+      "loss": 1.4606,
+      "step": 112585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.69526877437263e-06,
+      "loss": 1.3989,
+      "step": 112590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.694519556768051e-06,
+      "loss": 1.0011,
+      "step": 112595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.693770346618461e-06,
+      "loss": 0.8213,
+      "step": 112600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.693021143928152e-06,
+      "loss": 1.0537,
+      "step": 112605
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.692271948701392e-06,
+      "loss": 1.3872,
+      "step": 112610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.691522760942461e-06,
+      "loss": 1.3222,
+      "step": 112615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.690773580655642e-06,
+      "loss": 1.2742,
+      "step": 112620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.69002440784521e-06,
+      "loss": 1.1476,
+      "step": 112625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.689275242515443e-06,
+      "loss": 1.0574,
+      "step": 112630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.688526084670614e-06,
+      "loss": 1.0613,
+      "step": 112635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.687776934315013e-06,
+      "loss": 1.4422,
+      "step": 112640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.687027791452909e-06,
+      "loss": 0.9759,
+      "step": 112645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.68627865608858e-06,
+      "loss": 0.994,
+      "step": 112650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.685529528226312e-06,
+      "loss": 1.3154,
+      "step": 112655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.684780407870375e-06,
+      "loss": 1.1641,
+      "step": 112660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.684031295025047e-06,
+      "loss": 1.3204,
+      "step": 112665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.68328218969461e-06,
+      "loss": 1.193,
+      "step": 112670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.682533091883341e-06,
+      "loss": 1.1842,
+      "step": 112675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.681784001595514e-06,
+      "loss": 1.7155,
+      "step": 112680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.68103491883541e-06,
+      "loss": 1.0043,
+      "step": 112685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.680285843607308e-06,
+      "loss": 1.1624,
+      "step": 112690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.679536775915479e-06,
+      "loss": 1.3953,
+      "step": 112695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.678787715764211e-06,
+      "loss": 1.2771,
+      "step": 112700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.678038663157773e-06,
+      "loss": 1.1156,
+      "step": 112705
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.677289618100449e-06,
+      "loss": 1.0831,
+      "step": 112710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.676540580596506e-06,
+      "loss": 1.2888,
+      "step": 112715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.675791550650232e-06,
+      "loss": 1.1758,
+      "step": 112720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.6750425282659e-06,
+      "loss": 1.4104,
+      "step": 112725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.674293513447786e-06,
+      "loss": 1.3097,
+      "step": 112730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.673544506200174e-06,
+      "loss": 0.9474,
+      "step": 112735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.672795506527331e-06,
+      "loss": 1.3574,
+      "step": 112740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.672046514433541e-06,
+      "loss": 1.3214,
+      "step": 112745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.671297529923081e-06,
+      "loss": 1.1452,
+      "step": 112750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.670548553000226e-06,
+      "loss": 0.9589,
+      "step": 112755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.669799583669252e-06,
+      "loss": 1.5129,
+      "step": 112760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.669050621934441e-06,
+      "loss": 1.1771,
+      "step": 112765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.668301667800068e-06,
+      "loss": 1.1124,
+      "step": 112770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.667552721270405e-06,
+      "loss": 1.1115,
+      "step": 112775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.666803782349732e-06,
+      "loss": 1.2892,
+      "step": 112780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.666054851042328e-06,
+      "loss": 1.2077,
+      "step": 112785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.66530592735247e-06,
+      "loss": 1.5279,
+      "step": 112790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.66455701128443e-06,
+      "loss": 1.2947,
+      "step": 112795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.663808102842489e-06,
+      "loss": 1.2327,
+      "step": 112800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.663059202030925e-06,
+      "loss": 1.1493,
+      "step": 112805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.662310308854009e-06,
+      "loss": 1.0849,
+      "step": 112810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.66156142331602e-06,
+      "loss": 1.294,
+      "step": 112815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.660812545421237e-06,
+      "loss": 1.3597,
+      "step": 112820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.660063675173932e-06,
+      "loss": 1.2013,
+      "step": 112825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.659314812578388e-06,
+      "loss": 1.0881,
+      "step": 112830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.658565957638875e-06,
+      "loss": 1.0888,
+      "step": 112835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.657817110359672e-06,
+      "loss": 1.241,
+      "step": 112840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.657068270745055e-06,
+      "loss": 1.2627,
+      "step": 112845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.656319438799305e-06,
+      "loss": 1.4279,
+      "step": 112850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.65557061452669e-06,
+      "loss": 1.1888,
+      "step": 112855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.654821797931487e-06,
+      "loss": 1.2352,
+      "step": 112860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.65407298901798e-06,
+      "loss": 1.0783,
+      "step": 112865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.653324187790438e-06,
+      "loss": 0.98,
+      "step": 112870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.652575394253138e-06,
+      "loss": 1.1019,
+      "step": 112875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.65182660841036e-06,
+      "loss": 1.107,
+      "step": 112880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.651077830266377e-06,
+      "loss": 1.2251,
+      "step": 112885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.650329059825461e-06,
+      "loss": 1.1078,
+      "step": 112890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.649580297091895e-06,
+      "loss": 1.1441,
+      "step": 112895
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.648831542069953e-06,
+      "loss": 1.0291,
+      "step": 112900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.648082794763905e-06,
+      "loss": 1.2573,
+      "step": 112905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.647334055178036e-06,
+      "loss": 1.0677,
+      "step": 112910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.646585323316616e-06,
+      "loss": 1.3068,
+      "step": 112915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.645836599183924e-06,
+      "loss": 1.074,
+      "step": 112920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.645087882784228e-06,
+      "loss": 1.0526,
+      "step": 112925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.644339174121811e-06,
+      "loss": 1.1695,
+      "step": 112930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.64359047320095e-06,
+      "loss": 0.9276,
+      "step": 112935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.64284178002591e-06,
+      "loss": 1.2204,
+      "step": 112940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.642093094600979e-06,
+      "loss": 1.1538,
+      "step": 112945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.641344416930427e-06,
+      "loss": 1.0249,
+      "step": 112950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.640595747018524e-06,
+      "loss": 1.1284,
+      "step": 112955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.639847084869557e-06,
+      "loss": 1.1793,
+      "step": 112960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.639098430487791e-06,
+      "loss": 1.1343,
+      "step": 112965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.638349783877504e-06,
+      "loss": 1.0714,
+      "step": 112970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.637601145042974e-06,
+      "loss": 1.3321,
+      "step": 112975
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.636852513988475e-06,
+      "loss": 1.2169,
+      "step": 112980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.636103890718279e-06,
+      "loss": 1.1378,
+      "step": 112985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.635355275236664e-06,
+      "loss": 1.1713,
+      "step": 112990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.634606667547908e-06,
+      "loss": 1.2293,
+      "step": 112995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.63385806765628e-06,
+      "loss": 1.0139,
+      "step": 113000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.633109475566052e-06,
+      "loss": 1.0258,
+      "step": 113005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.632360891281509e-06,
+      "loss": 1.3318,
+      "step": 113010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.631612314806923e-06,
+      "loss": 1.0095,
+      "step": 113015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.630863746146562e-06,
+      "loss": 1.6053,
+      "step": 113020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.630115185304707e-06,
+      "loss": 1.169,
+      "step": 113025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.629366632285634e-06,
+      "loss": 1.2766,
+      "step": 113030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.628618087093608e-06,
+      "loss": 1.1538,
+      "step": 113035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.627869549732917e-06,
+      "loss": 1.217,
+      "step": 113040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.627121020207828e-06,
+      "loss": 1.1377,
+      "step": 113045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.626372498522611e-06,
+      "loss": 1.2169,
+      "step": 113050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.625623984681553e-06,
+      "loss": 1.5605,
+      "step": 113055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.624875478688916e-06,
+      "loss": 1.2624,
+      "step": 113060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.624126980548983e-06,
+      "loss": 1.147,
+      "step": 113065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.623378490266022e-06,
+      "loss": 1.2568,
+      "step": 113070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.622630007844313e-06,
+      "loss": 1.1156,
+      "step": 113075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.621881533288126e-06,
+      "loss": 0.9059,
+      "step": 113080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.621133066601735e-06,
+      "loss": 1.1293,
+      "step": 113085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.62038460778942e-06,
+      "loss": 1.3063,
+      "step": 113090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.619636156855448e-06,
+      "loss": 1.2344,
+      "step": 113095
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.618887713804095e-06,
+      "loss": 1.1142,
+      "step": 113100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.618139278639638e-06,
+      "loss": 1.1308,
+      "step": 113105
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.61739085136635e-06,
+      "loss": 1.1312,
+      "step": 113110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.616642431988502e-06,
+      "loss": 1.3348,
+      "step": 113115
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.61589402051037e-06,
+      "loss": 1.5772,
+      "step": 113120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.615145616936227e-06,
+      "loss": 1.625,
+      "step": 113125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.614397221270346e-06,
+      "loss": 1.1006,
+      "step": 113130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.613648833517006e-06,
+      "loss": 1.4852,
+      "step": 113135
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.612900453680476e-06,
+      "loss": 1.4434,
+      "step": 113140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.612152081765031e-06,
+      "loss": 1.0533,
+      "step": 113145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.61140371777494e-06,
+      "loss": 1.0671,
+      "step": 113150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.610655361714486e-06,
+      "loss": 1.3079,
+      "step": 113155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.609907013587933e-06,
+      "loss": 1.5211,
+      "step": 113160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.609158673399557e-06,
+      "loss": 1.0019,
+      "step": 113165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.608410341153639e-06,
+      "loss": 1.1874,
+      "step": 113170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.607662016854443e-06,
+      "loss": 1.0815,
+      "step": 113175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.606913700506245e-06,
+      "loss": 1.3351,
+      "step": 113180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.60616539211332e-06,
+      "loss": 1.2896,
+      "step": 113185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.605417091679943e-06,
+      "loss": 1.0831,
+      "step": 113190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.60466879921038e-06,
+      "loss": 1.2472,
+      "step": 113195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.603920514708912e-06,
+      "loss": 1.3758,
+      "step": 113200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.60317223817981e-06,
+      "loss": 1.1747,
+      "step": 113205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.602423969627341e-06,
+      "loss": 1.3959,
+      "step": 113210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.601675709055784e-06,
+      "loss": 1.1032,
+      "step": 113215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.600927456469413e-06,
+      "loss": 1.4931,
+      "step": 113220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.600179211872498e-06,
+      "loss": 0.8704,
+      "step": 113225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.59943097526931e-06,
+      "loss": 1.1798,
+      "step": 113230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.598682746664127e-06,
+      "loss": 1.006,
+      "step": 113235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.59793452606122e-06,
+      "loss": 0.974,
+      "step": 113240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.597186313464857e-06,
+      "loss": 1.3417,
+      "step": 113245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.59643810887932e-06,
+      "loss": 1.1182,
+      "step": 113250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.595689912308873e-06,
+      "loss": 1.2254,
+      "step": 113255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.59494172375779e-06,
+      "loss": 1.13,
+      "step": 113260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.594193543230351e-06,
+      "loss": 1.1012,
+      "step": 113265
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.593445370730822e-06,
+      "loss": 1.4556,
+      "step": 113270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.59269720626347e-06,
+      "loss": 1.3692,
+      "step": 113275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.59194904983258e-06,
+      "loss": 1.021,
+      "step": 113280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.591200901442419e-06,
+      "loss": 1.0166,
+      "step": 113285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.590452761097256e-06,
+      "loss": 1.3079,
+      "step": 113290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.589704628801364e-06,
+      "loss": 1.1239,
+      "step": 113295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.58895650455902e-06,
+      "loss": 1.0567,
+      "step": 113300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.588208388374493e-06,
+      "loss": 1.4001,
+      "step": 113305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.587460280252053e-06,
+      "loss": 1.2336,
+      "step": 113310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.586712180195978e-06,
+      "loss": 1.2325,
+      "step": 113315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.585964088210537e-06,
+      "loss": 1.0063,
+      "step": 113320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.585216004299996e-06,
+      "loss": 1.4508,
+      "step": 113325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.584467928468636e-06,
+      "loss": 1.0863,
+      "step": 113330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.58371986072073e-06,
+      "loss": 1.0611,
+      "step": 113335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.582971801060535e-06,
+      "loss": 0.8934,
+      "step": 113340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.582223749492342e-06,
+      "loss": 1.2002,
+      "step": 113345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.581475706020412e-06,
+      "loss": 1.2909,
+      "step": 113350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.58072767064902e-06,
+      "loss": 1.2456,
+      "step": 113355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.579979643382432e-06,
+      "loss": 1.5574,
+      "step": 113360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.579231624224926e-06,
+      "loss": 1.0704,
+      "step": 113365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.578483613180772e-06,
+      "loss": 1.0067,
+      "step": 113370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.577735610254238e-06,
+      "loss": 1.1248,
+      "step": 113375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.576987615449605e-06,
+      "loss": 0.9565,
+      "step": 113380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.576239628771135e-06,
+      "loss": 1.2432,
+      "step": 113385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.5754916502231e-06,
+      "loss": 1.4664,
+      "step": 113390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.574743679809779e-06,
+      "loss": 1.1192,
+      "step": 113395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.573995717535434e-06,
+      "loss": 1.3258,
+      "step": 113400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.57324776340434e-06,
+      "loss": 1.5364,
+      "step": 113405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.572499817420769e-06,
+      "loss": 1.262,
+      "step": 113410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.571751879588996e-06,
+      "loss": 1.1499,
+      "step": 113415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.571003949913284e-06,
+      "loss": 1.606,
+      "step": 113420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.570256028397909e-06,
+      "loss": 1.2592,
+      "step": 113425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.56950811504714e-06,
+      "loss": 1.0908,
+      "step": 113430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.568760209865253e-06,
+      "loss": 1.3051,
+      "step": 113435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.568012312856511e-06,
+      "loss": 1.3706,
+      "step": 113440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.56726442402519e-06,
+      "loss": 1.1445,
+      "step": 113445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.566516543375561e-06,
+      "loss": 1.5585,
+      "step": 113450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.56576867091189e-06,
+      "loss": 1.2602,
+      "step": 113455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.565020806638454e-06,
+      "loss": 1.2319,
+      "step": 113460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.564272950559522e-06,
+      "loss": 1.2371,
+      "step": 113465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.56352510267936e-06,
+      "loss": 1.5367,
+      "step": 113470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.562777263002247e-06,
+      "loss": 1.1129,
+      "step": 113475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.562029431532446e-06,
+      "loss": 1.1125,
+      "step": 113480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.561281608274228e-06,
+      "loss": 1.3649,
+      "step": 113485
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.560533793231868e-06,
+      "loss": 1.4396,
+      "step": 113490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.559785986409638e-06,
+      "loss": 1.1278,
+      "step": 113495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.559038187811802e-06,
+      "loss": 1.7036,
+      "step": 113500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.55829039744263e-06,
+      "loss": 1.0878,
+      "step": 113505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.5575426153064e-06,
+      "loss": 1.1472,
+      "step": 113510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.556794841407376e-06,
+      "loss": 1.4514,
+      "step": 113515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.556047075749825e-06,
+      "loss": 1.1103,
+      "step": 113520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.555299318338027e-06,
+      "loss": 1.4496,
+      "step": 113525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.554551569176247e-06,
+      "loss": 1.0181,
+      "step": 113530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.553803828268753e-06,
+      "loss": 1.3358,
+      "step": 113535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.553056095619817e-06,
+      "loss": 1.0402,
+      "step": 113540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.552308371233714e-06,
+      "loss": 1.1301,
+      "step": 113545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.551560655114699e-06,
+      "loss": 1.5793,
+      "step": 113550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.55081294726706e-06,
+      "loss": 1.0775,
+      "step": 113555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.550065247695054e-06,
+      "loss": 1.0922,
+      "step": 113560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.549317556402956e-06,
+      "loss": 1.2052,
+      "step": 113565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.548569873395037e-06,
+      "loss": 1.2695,
+      "step": 113570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.547822198675564e-06,
+      "loss": 1.2793,
+      "step": 113575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.54707453224881e-06,
+      "loss": 1.0062,
+      "step": 113580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.546326874119034e-06,
+      "loss": 1.2771,
+      "step": 113585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.54557922429052e-06,
+      "loss": 1.2049,
+      "step": 113590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.544831582767529e-06,
+      "loss": 1.5706,
+      "step": 113595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.54408394955433e-06,
+      "loss": 1.1619,
+      "step": 113600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.5433363246552e-06,
+      "loss": 1.2305,
+      "step": 113605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.542588708074401e-06,
+      "loss": 1.0237,
+      "step": 113610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.541841099816201e-06,
+      "loss": 2.1003,
+      "step": 113615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.541093499884875e-06,
+      "loss": 1.2135,
+      "step": 113620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.540345908284693e-06,
+      "loss": 1.1279,
+      "step": 113625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.539598325019915e-06,
+      "loss": 0.9854,
+      "step": 113630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.53885075009482e-06,
+      "loss": 1.1184,
+      "step": 113635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.538103183513674e-06,
+      "loss": 1.3167,
+      "step": 113640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.537355625280744e-06,
+      "loss": 1.2012,
+      "step": 113645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.536608075400297e-06,
+      "loss": 1.4215,
+      "step": 113650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.535860533876606e-06,
+      "loss": 1.6447,
+      "step": 113655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.535113000713943e-06,
+      "loss": 1.2364,
+      "step": 113660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.534365475916568e-06,
+      "loss": 1.0975,
+      "step": 113665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.533617959488755e-06,
+      "loss": 1.3013,
+      "step": 113670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.532870451434777e-06,
+      "loss": 1.2151,
+      "step": 113675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.532122951758889e-06,
+      "loss": 1.1894,
+      "step": 113680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.531375460465375e-06,
+      "loss": 1.2203,
+      "step": 113685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.530627977558497e-06,
+      "loss": 1.0252,
+      "step": 113690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.52988050304252e-06,
+      "loss": 1.4974,
+      "step": 113695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.52913303692172e-06,
+      "loss": 1.2876,
+      "step": 113700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.528385579200359e-06,
+      "loss": 1.2297,
+      "step": 113705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.527638129882707e-06,
+      "loss": 1.14,
+      "step": 113710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.526890688973034e-06,
+      "loss": 1.2135,
+      "step": 113715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.52614325647561e-06,
+      "loss": 1.115,
+      "step": 113720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.5253958323947e-06,
+      "loss": 1.1301,
+      "step": 113725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.524648416734568e-06,
+      "loss": 1.1312,
+      "step": 113730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.523901009499494e-06,
+      "loss": 1.2732,
+      "step": 113735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.523153610693736e-06,
+      "loss": 1.3898,
+      "step": 113740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.522406220321563e-06,
+      "loss": 1.3322,
+      "step": 113745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.521658838387248e-06,
+      "loss": 1.41,
+      "step": 113750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.520911464895058e-06,
+      "loss": 1.1576,
+      "step": 113755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.520164099849254e-06,
+      "loss": 1.5541,
+      "step": 113760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.519416743254112e-06,
+      "loss": 1.362,
+      "step": 113765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.518669395113896e-06,
+      "loss": 1.2052,
+      "step": 113770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.517922055432874e-06,
+      "loss": 1.0465,
+      "step": 113775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.517174724215316e-06,
+      "loss": 1.4365,
+      "step": 113780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.516427401465486e-06,
+      "loss": 1.4162,
+      "step": 113785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.515680087187658e-06,
+      "loss": 1.1559,
+      "step": 113790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.51493278138609e-06,
+      "loss": 1.5894,
+      "step": 113795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.514185484065055e-06,
+      "loss": 1.3941,
+      "step": 113800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.513438195228822e-06,
+      "loss": 1.0789,
+      "step": 113805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.512690914881653e-06,
+      "loss": 1.1229,
+      "step": 113810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.511943643027823e-06,
+      "loss": 1.0469,
+      "step": 113815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.511196379671593e-06,
+      "loss": 1.2478,
+      "step": 113820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.510449124817232e-06,
+      "loss": 1.1216,
+      "step": 113825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.509701878469008e-06,
+      "loss": 0.9793,
+      "step": 113830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.508954640631191e-06,
+      "loss": 1.2063,
+      "step": 113835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.508207411308039e-06,
+      "loss": 1.2294,
+      "step": 113840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.507460190503829e-06,
+      "loss": 1.3792,
+      "step": 113845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.506712978222827e-06,
+      "loss": 1.0112,
+      "step": 113850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.50596577446929e-06,
+      "loss": 1.4893,
+      "step": 113855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.505218579247496e-06,
+      "loss": 1.273,
+      "step": 113860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.504471392561707e-06,
+      "loss": 1.5885,
+      "step": 113865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.503724214416194e-06,
+      "loss": 1.1238,
+      "step": 113870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.502977044815216e-06,
+      "loss": 1.1594,
+      "step": 113875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.502229883763044e-06,
+      "loss": 1.2732,
+      "step": 113880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.50148273126395e-06,
+      "loss": 1.0584,
+      "step": 113885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.500735587322189e-06,
+      "loss": 1.2129,
+      "step": 113890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.49998845194204e-06,
+      "loss": 0.9766,
+      "step": 113895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.499241325127761e-06,
+      "loss": 1.1342,
+      "step": 113900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.498494206883618e-06,
+      "loss": 1.0447,
+      "step": 113905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.497747097213887e-06,
+      "loss": 1.3394,
+      "step": 113910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.496999996122825e-06,
+      "loss": 1.1356,
+      "step": 113915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.4962529036147e-06,
+      "loss": 1.2202,
+      "step": 113920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.49550581969378e-06,
+      "loss": 1.2361,
+      "step": 113925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.494758744364337e-06,
+      "loss": 1.2897,
+      "step": 113930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.494011677630626e-06,
+      "loss": 1.0684,
+      "step": 113935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.493264619496916e-06,
+      "loss": 1.1252,
+      "step": 113940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.492517569967481e-06,
+      "loss": 1.3869,
+      "step": 113945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.491770529046579e-06,
+      "loss": 1.4976,
+      "step": 113950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.491023496738476e-06,
+      "loss": 1.0985,
+      "step": 113955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.490276473047444e-06,
+      "loss": 1.028,
+      "step": 113960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.489529457977745e-06,
+      "loss": 1.1934,
+      "step": 113965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.488782451533642e-06,
+      "loss": 1.2476,
+      "step": 113970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.488035453719405e-06,
+      "loss": 1.2333,
+      "step": 113975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.487288464539302e-06,
+      "loss": 1.5622,
+      "step": 113980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.48654148399759e-06,
+      "loss": 1.1016,
+      "step": 113985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.485794512098547e-06,
+      "loss": 1.694,
+      "step": 113990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.485047548846428e-06,
+      "loss": 1.4462,
+      "step": 113995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.484300594245501e-06,
+      "loss": 1.038,
+      "step": 114000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.483553648300035e-06,
+      "loss": 1.2328,
+      "step": 114005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.482806711014293e-06,
+      "loss": 1.278,
+      "step": 114010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.482059782392542e-06,
+      "loss": 1.3519,
+      "step": 114015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.481312862439042e-06,
+      "loss": 1.1758,
+      "step": 114020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.480565951158068e-06,
+      "loss": 1.1375,
+      "step": 114025
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.479819048553875e-06,
+      "loss": 1.4378,
+      "step": 114030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.479072154630733e-06,
+      "loss": 1.2852,
+      "step": 114035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.478325269392911e-06,
+      "loss": 1.3727,
+      "step": 114040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.477578392844668e-06,
+      "loss": 1.4572,
+      "step": 114045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.476831524990268e-06,
+      "loss": 1.301,
+      "step": 114050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.476084665833983e-06,
+      "loss": 1.1609,
+      "step": 114055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.475337815380076e-06,
+      "loss": 1.5465,
+      "step": 114060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.474590973632805e-06,
+      "loss": 1.1023,
+      "step": 114065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.473844140596443e-06,
+      "loss": 1.3542,
+      "step": 114070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.47309731627525e-06,
+      "loss": 1.076,
+      "step": 114075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.472350500673496e-06,
+      "loss": 1.4145,
+      "step": 114080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.471603693795438e-06,
+      "loss": 1.1158,
+      "step": 114085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.470856895645347e-06,
+      "loss": 1.0893,
+      "step": 114090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.470110106227489e-06,
+      "loss": 1.3664,
+      "step": 114095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.46936332554612e-06,
+      "loss": 1.4471,
+      "step": 114100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.46861655360551e-06,
+      "loss": 1.1636,
+      "step": 114105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.467869790409926e-06,
+      "loss": 1.6188,
+      "step": 114110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.467123035963625e-06,
+      "loss": 1.0896,
+      "step": 114115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.466376290270881e-06,
+      "loss": 1.3273,
+      "step": 114120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.465629553335952e-06,
+      "loss": 1.4762,
+      "step": 114125
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.4648828251631e-06,
+      "loss": 1.4333,
+      "step": 114130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.464136105756596e-06,
+      "loss": 1.5354,
+      "step": 114135
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.463389395120703e-06,
+      "loss": 1.4474,
+      "step": 114140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.462642693259678e-06,
+      "loss": 1.2708,
+      "step": 114145
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.461896000177792e-06,
+      "loss": 1.1577,
+      "step": 114150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.461149315879311e-06,
+      "loss": 1.3193,
+      "step": 114155
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.460402640368492e-06,
+      "loss": 1.0411,
+      "step": 114160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.4596559736496e-06,
+      "loss": 1.4279,
+      "step": 114165
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.458909315726902e-06,
+      "loss": 1.1102,
+      "step": 114170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.458162666604664e-06,
+      "loss": 1.2903,
+      "step": 114175
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.457416026287142e-06,
+      "loss": 1.3045,
+      "step": 114180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.456669394778604e-06,
+      "loss": 0.9681,
+      "step": 114185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.45592277208332e-06,
+      "loss": 1.0195,
+      "step": 114190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.45517615820554e-06,
+      "loss": 1.1306,
+      "step": 114195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.45442955314954e-06,
+      "loss": 1.4032,
+      "step": 114200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.453682956919576e-06,
+      "loss": 1.2272,
+      "step": 114205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.452936369519912e-06,
+      "loss": 1.0736,
+      "step": 114210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.45218979095482e-06,
+      "loss": 1.4145,
+      "step": 114215
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.451443221228552e-06,
+      "loss": 1.2837,
+      "step": 114220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.450696660345379e-06,
+      "loss": 1.3077,
+      "step": 114225
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.449950108309556e-06,
+      "loss": 1.2491,
+      "step": 114230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.449203565125356e-06,
+      "loss": 1.2387,
+      "step": 114235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.448457030797038e-06,
+      "loss": 0.9336,
+      "step": 114240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.447710505328861e-06,
+      "loss": 0.9824,
+      "step": 114245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.446963988725096e-06,
+      "loss": 1.2196,
+      "step": 114250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.446217480990002e-06,
+      "loss": 1.2912,
+      "step": 114255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.445470982127838e-06,
+      "loss": 1.3256,
+      "step": 114260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.444724492142873e-06,
+      "loss": 1.1086,
+      "step": 114265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.443978011039371e-06,
+      "loss": 1.5854,
+      "step": 114270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.443231538821585e-06,
+      "loss": 1.276,
+      "step": 114275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.442485075493788e-06,
+      "loss": 1.3456,
+      "step": 114280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.44173862106024e-06,
+      "loss": 1.5245,
+      "step": 114285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.440992175525197e-06,
+      "loss": 1.0765,
+      "step": 114290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.440245738892934e-06,
+      "loss": 1.5681,
+      "step": 114295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.439499311167704e-06,
+      "loss": 1.1908,
+      "step": 114300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.438752892353774e-06,
+      "loss": 1.2448,
+      "step": 114305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.438006482455402e-06,
+      "loss": 1.2609,
+      "step": 114310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.437260081476853e-06,
+      "loss": 1.369,
+      "step": 114315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.436513689422395e-06,
+      "loss": 1.1794,
+      "step": 114320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.435767306296275e-06,
+      "loss": 1.2595,
+      "step": 114325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.435020932102773e-06,
+      "loss": 1.3898,
+      "step": 114330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.43427456684614e-06,
+      "loss": 1.211,
+      "step": 114335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.433528210530638e-06,
+      "loss": 0.9,
+      "step": 114340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.432781863160539e-06,
+      "loss": 1.0466,
+      "step": 114345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.432035524740096e-06,
+      "loss": 1.2178,
+      "step": 114350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.431289195273571e-06,
+      "loss": 1.4426,
+      "step": 114355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.430542874765231e-06,
+      "loss": 1.1038,
+      "step": 114360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.429796563219338e-06,
+      "loss": 1.2215,
+      "step": 114365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.429050260640148e-06,
+      "loss": 1.1234,
+      "step": 114370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.428303967031924e-06,
+      "loss": 1.6012,
+      "step": 114375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.427557682398934e-06,
+      "loss": 1.4056,
+      "step": 114380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.426811406745433e-06,
+      "loss": 1.2331,
+      "step": 114385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.426065140075683e-06,
+      "loss": 1.1087,
+      "step": 114390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.425318882393952e-06,
+      "loss": 1.3096,
+      "step": 114395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.424572633704495e-06,
+      "loss": 1.1968,
+      "step": 114400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.423826394011573e-06,
+      "loss": 0.9288,
+      "step": 114405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.423080163319454e-06,
+      "loss": 1.0738,
+      "step": 114410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.422333941632395e-06,
+      "loss": 1.0425,
+      "step": 114415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.421587728954654e-06,
+      "loss": 1.4004,
+      "step": 114420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.4208415252905e-06,
+      "loss": 1.351,
+      "step": 114425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.42009533064419e-06,
+      "loss": 1.1251,
+      "step": 114430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.419349145019982e-06,
+      "loss": 0.9811,
+      "step": 114435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.418602968422145e-06,
+      "loss": 1.6305,
+      "step": 114440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.417856800854934e-06,
+      "loss": 1.1987,
+      "step": 114445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.41711064232261e-06,
+      "loss": 1.3669,
+      "step": 114450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.416364492829437e-06,
+      "loss": 1.0824,
+      "step": 114455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.415618352379676e-06,
+      "loss": 1.1048,
+      "step": 114460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.414872220977585e-06,
+      "loss": 1.0815,
+      "step": 114465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.414126098627423e-06,
+      "loss": 1.0426,
+      "step": 114470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.413379985333457e-06,
+      "loss": 1.2746,
+      "step": 114475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.412633881099947e-06,
+      "loss": 1.1305,
+      "step": 114480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.411887785931146e-06,
+      "loss": 1.1122,
+      "step": 114485
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.411141699831323e-06,
+      "loss": 1.2571,
+      "step": 114490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.410395622804737e-06,
+      "loss": 1.585,
+      "step": 114495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.409649554855643e-06,
+      "loss": 1.4669,
+      "step": 114500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.408903495988308e-06,
+      "loss": 1.1153,
+      "step": 114505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.40815744620699e-06,
+      "loss": 0.9727,
+      "step": 114510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.407411405515951e-06,
+      "loss": 1.1055,
+      "step": 114515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.406665373919443e-06,
+      "loss": 1.1031,
+      "step": 114520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.405919351421738e-06,
+      "loss": 1.1487,
+      "step": 114525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.405173338027092e-06,
+      "loss": 1.1481,
+      "step": 114530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.404427333739756e-06,
+      "loss": 1.387,
+      "step": 114535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.403681338564007e-06,
+      "loss": 1.6263,
+      "step": 114540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.402935352504092e-06,
+      "loss": 1.3976,
+      "step": 114545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.402189375564274e-06,
+      "loss": 1.1629,
+      "step": 114550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.401443407748818e-06,
+      "loss": 1.0909,
+      "step": 114555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.400697449061979e-06,
+      "loss": 1.5946,
+      "step": 114560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.399951499508014e-06,
+      "loss": 0.982,
+      "step": 114565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.39920555909119e-06,
+      "loss": 1.4057,
+      "step": 114570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.398459627815764e-06,
+      "loss": 1.3151,
+      "step": 114575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.397713705685991e-06,
+      "loss": 1.0717,
+      "step": 114580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.396967792706137e-06,
+      "loss": 1.2859,
+      "step": 114585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.396221888880461e-06,
+      "loss": 1.2265,
+      "step": 114590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.395475994213219e-06,
+      "loss": 0.9245,
+      "step": 114595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.394730108708669e-06,
+      "loss": 1.4244,
+      "step": 114600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.393984232371076e-06,
+      "loss": 1.4225,
+      "step": 114605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.3932383652047e-06,
+      "loss": 1.2365,
+      "step": 114610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.392492507213793e-06,
+      "loss": 1.3192,
+      "step": 114615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.391746658402619e-06,
+      "loss": 1.1637,
+      "step": 114620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.391000818775438e-06,
+      "loss": 1.055,
+      "step": 114625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.390254988336504e-06,
+      "loss": 1.089,
+      "step": 114630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.389509167090084e-06,
+      "loss": 1.229,
+      "step": 114635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.388763355040432e-06,
+      "loss": 1.4693,
+      "step": 114640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.388017552191805e-06,
+      "loss": 1.2422,
+      "step": 114645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.38727175854847e-06,
+      "loss": 1.3067,
+      "step": 114650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.386525974114676e-06,
+      "loss": 1.3829,
+      "step": 114655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.385780198894687e-06,
+      "loss": 1.6416,
+      "step": 114660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.385034432892762e-06,
+      "loss": 1.064,
+      "step": 114665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.384288676113163e-06,
+      "loss": 1.5403,
+      "step": 114670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.383542928560141e-06,
+      "loss": 1.5298,
+      "step": 114675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.382797190237958e-06,
+      "loss": 1.0906,
+      "step": 114680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.382051461150874e-06,
+      "loss": 1.5619,
+      "step": 114685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.381305741303145e-06,
+      "loss": 1.0965,
+      "step": 114690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.38056003069903e-06,
+      "loss": 1.2371,
+      "step": 114695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.37981432934279e-06,
+      "loss": 1.1817,
+      "step": 114700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.379068637238685e-06,
+      "loss": 1.1718,
+      "step": 114705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.378322954390962e-06,
+      "loss": 1.2467,
+      "step": 114710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.377577280803892e-06,
+      "loss": 1.3701,
+      "step": 114715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.376831616481732e-06,
+      "loss": 1.1695,
+      "step": 114720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.376085961428728e-06,
+      "loss": 1.1096,
+      "step": 114725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.375340315649152e-06,
+      "loss": 1.2915,
+      "step": 114730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.374594679147256e-06,
+      "loss": 1.0954,
+      "step": 114735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.373849051927299e-06,
+      "loss": 1.162,
+      "step": 114740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.373103433993537e-06,
+      "loss": 1.0711,
+      "step": 114745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.37235782535023e-06,
+      "loss": 1.0743,
+      "step": 114750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.371612226001635e-06,
+      "loss": 1.0714,
+      "step": 114755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.370866635952008e-06,
+      "loss": 1.3828,
+      "step": 114760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.370121055205613e-06,
+      "loss": 1.4,
+      "step": 114765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.369375483766701e-06,
+      "loss": 1.3643,
+      "step": 114770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.368629921639532e-06,
+      "loss": 1.1314,
+      "step": 114775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.367884368828367e-06,
+      "loss": 1.1949,
+      "step": 114780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.367138825337456e-06,
+      "loss": 1.0379,
+      "step": 114785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.36639329117106e-06,
+      "loss": 1.198,
+      "step": 114790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.365647766333439e-06,
+      "loss": 1.333,
+      "step": 114795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.36490225082885e-06,
+      "loss": 1.3635,
+      "step": 114800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.364156744661545e-06,
+      "loss": 1.169,
+      "step": 114805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.363411247835786e-06,
+      "loss": 1.307,
+      "step": 114810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.36266576035583e-06,
+      "loss": 1.352,
+      "step": 114815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.361920282225935e-06,
+      "loss": 1.2316,
+      "step": 114820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.361174813450352e-06,
+      "loss": 1.1445,
+      "step": 114825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.360429354033344e-06,
+      "loss": 1.4069,
+      "step": 114830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.359683903979168e-06,
+      "loss": 1.0651,
+      "step": 114835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.358938463292075e-06,
+      "loss": 1.2659,
+      "step": 114840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.358193031976329e-06,
+      "loss": 1.0388,
+      "step": 114845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.357447610036184e-06,
+      "loss": 1.4266,
+      "step": 114850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.356702197475894e-06,
+      "loss": 1.2207,
+      "step": 114855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.355956794299722e-06,
+      "loss": 1.0948,
+      "step": 114860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.35521140051192e-06,
+      "loss": 1.3785,
+      "step": 114865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.354466016116743e-06,
+      "loss": 0.937,
+      "step": 114870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.35372064111845e-06,
+      "loss": 1.3339,
+      "step": 114875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.352975275521303e-06,
+      "loss": 1.0307,
+      "step": 114880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.35222991932955e-06,
+      "loss": 1.1676,
+      "step": 114885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.351484572547447e-06,
+      "loss": 1.1915,
+      "step": 114890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.35073923517926e-06,
+      "loss": 1.1191,
+      "step": 114895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.349993907229235e-06,
+      "loss": 1.1207,
+      "step": 114900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.349248588701631e-06,
+      "loss": 1.2107,
+      "step": 114905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.348503279600707e-06,
+      "loss": 1.3299,
+      "step": 114910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.34775797993072e-06,
+      "loss": 1.2949,
+      "step": 114915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.34701268969592e-06,
+      "loss": 0.9831,
+      "step": 114920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.346267408900568e-06,
+      "loss": 1.1738,
+      "step": 114925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.34552213754892e-06,
+      "loss": 1.0188,
+      "step": 114930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.344776875645225e-06,
+      "loss": 1.2904,
+      "step": 114935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.34403162319375e-06,
+      "loss": 1.0468,
+      "step": 114940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.343286380198745e-06,
+      "loss": 0.9372,
+      "step": 114945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.342541146664462e-06,
+      "loss": 1.3051,
+      "step": 114950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.341795922595166e-06,
+      "loss": 1.1253,
+      "step": 114955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.341050707995103e-06,
+      "loss": 0.9584,
+      "step": 114960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.340305502868537e-06,
+      "loss": 1.2779,
+      "step": 114965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.33956030721971e-06,
+      "loss": 1.3086,
+      "step": 114970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.338815121052897e-06,
+      "loss": 1.4794,
+      "step": 114975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.338069944372338e-06,
+      "loss": 1.3446,
+      "step": 114980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.337324777182292e-06,
+      "loss": 1.338,
+      "step": 114985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.336579619487021e-06,
+      "loss": 1.2629,
+      "step": 114990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.335834471290772e-06,
+      "loss": 1.0456,
+      "step": 114995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.335089332597801e-06,
+      "loss": 1.2805,
+      "step": 115000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.334344203412368e-06,
+      "loss": 1.3132,
+      "step": 115005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.333599083738727e-06,
+      "loss": 1.2227,
+      "step": 115010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.332853973581129e-06,
+      "loss": 1.2475,
+      "step": 115015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.332108872943832e-06,
+      "loss": 1.2066,
+      "step": 115020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.331363781831092e-06,
+      "loss": 1.5682,
+      "step": 115025
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.330618700247163e-06,
+      "loss": 1.2559,
+      "step": 115030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.329873628196294e-06,
+      "loss": 1.0979,
+      "step": 115035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.329128565682747e-06,
+      "loss": 1.26,
+      "step": 115040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.328383512710778e-06,
+      "loss": 1.0729,
+      "step": 115045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.327638469284634e-06,
+      "loss": 1.3546,
+      "step": 115050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.326893435408576e-06,
+      "loss": 1.3276,
+      "step": 115055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.326148411086856e-06,
+      "loss": 1.1934,
+      "step": 115060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.325403396323728e-06,
+      "loss": 1.3801,
+      "step": 115065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.324658391123449e-06,
+      "loss": 1.0313,
+      "step": 115070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.32391339549027e-06,
+      "loss": 1.1254,
+      "step": 115075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.323168409428446e-06,
+      "loss": 1.0043,
+      "step": 115080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.322423432942237e-06,
+      "loss": 1.2046,
+      "step": 115085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.32167846603589e-06,
+      "loss": 1.0538,
+      "step": 115090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.320933508713659e-06,
+      "loss": 1.365,
+      "step": 115095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.320188560979803e-06,
+      "loss": 1.5185,
+      "step": 115100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.319443622838576e-06,
+      "loss": 1.2867,
+      "step": 115105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.318698694294226e-06,
+      "loss": 1.2083,
+      "step": 115110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.31795377535101e-06,
+      "loss": 1.1186,
+      "step": 115115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.317208866013186e-06,
+      "loss": 1.3177,
+      "step": 115120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.316463966285006e-06,
+      "loss": 1.1632,
+      "step": 115125
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.315719076170718e-06,
+      "loss": 1.3021,
+      "step": 115130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.314974195674582e-06,
+      "loss": 1.2695,
+      "step": 115135
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.31422932480085e-06,
+      "loss": 1.5302,
+      "step": 115140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.313484463553773e-06,
+      "loss": 1.4029,
+      "step": 115145
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.312739611937609e-06,
+      "loss": 1.0969,
+      "step": 115150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.311994769956608e-06,
+      "loss": 1.0236,
+      "step": 115155
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.311249937615023e-06,
+      "loss": 1.0795,
+      "step": 115160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.310505114917114e-06,
+      "loss": 1.4476,
+      "step": 115165
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.309760301867127e-06,
+      "loss": 1.454,
+      "step": 115170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.309015498469319e-06,
+      "loss": 1.4257,
+      "step": 115175
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.308270704727937e-06,
+      "loss": 1.2452,
+      "step": 115180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.307525920647245e-06,
+      "loss": 1.169,
+      "step": 115185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.30678114623149e-06,
+      "loss": 1.2588,
+      "step": 115190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.306036381484922e-06,
+      "loss": 1.3191,
+      "step": 115195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.3052916264118e-06,
+      "loss": 1.2563,
+      "step": 115200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.304546881016374e-06,
+      "loss": 1.3616,
+      "step": 115205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.303802145302894e-06,
+      "loss": 1.3945,
+      "step": 115210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.30305741927562e-06,
+      "loss": 1.0867,
+      "step": 115215
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.3023127029388e-06,
+      "loss": 1.2877,
+      "step": 115220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.301567996296685e-06,
+      "loss": 0.9531,
+      "step": 115225
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.300823299353532e-06,
+      "loss": 0.9183,
+      "step": 115230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.300078612113594e-06,
+      "loss": 1.2803,
+      "step": 115235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.299333934581115e-06,
+      "loss": 1.276,
+      "step": 115240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.298589266760362e-06,
+      "loss": 1.2436,
+      "step": 115245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.297844608655576e-06,
+      "loss": 1.2746,
+      "step": 115250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.297099960271015e-06,
+      "loss": 1.0782,
+      "step": 115255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.296355321610923e-06,
+      "loss": 1.1087,
+      "step": 115260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.295610692679563e-06,
+      "loss": 1.1182,
+      "step": 115265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.294866073481185e-06,
+      "loss": 1.4645,
+      "step": 115270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.29412146402003e-06,
+      "loss": 1.2853,
+      "step": 115275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.293376864300368e-06,
+      "loss": 1.152,
+      "step": 115280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.292632274326438e-06,
+      "loss": 1.2772,
+      "step": 115285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.291887694102496e-06,
+      "loss": 1.0167,
+      "step": 115290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.291143123632797e-06,
+      "loss": 0.9735,
+      "step": 115295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.290398562921587e-06,
+      "loss": 1.2704,
+      "step": 115300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.289654011973118e-06,
+      "loss": 0.9711,
+      "step": 115305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.288909470791648e-06,
+      "loss": 1.1101,
+      "step": 115310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.288164939381427e-06,
+      "loss": 1.4172,
+      "step": 115315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.287420417746705e-06,
+      "loss": 1.0476,
+      "step": 115320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.286675905891728e-06,
+      "loss": 1.5379,
+      "step": 115325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.28593140382076e-06,
+      "loss": 1.205,
+      "step": 115330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.285186911538042e-06,
+      "loss": 1.102,
+      "step": 115335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.284442429047827e-06,
+      "loss": 1.3266,
+      "step": 115340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.28369795635437e-06,
+      "loss": 1.2217,
+      "step": 115345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.282953493461923e-06,
+      "loss": 1.4879,
+      "step": 115350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.282209040374733e-06,
+      "loss": 1.149,
+      "step": 115355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.281464597097054e-06,
+      "loss": 1.2451,
+      "step": 115360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.280720163633138e-06,
+      "loss": 1.7461,
+      "step": 115365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.279975739987228e-06,
+      "loss": 1.1136,
+      "step": 115370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.279231326163589e-06,
+      "loss": 1.351,
+      "step": 115375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.278486922166463e-06,
+      "loss": 1.1652,
+      "step": 115380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.277742528000099e-06,
+      "loss": 1.5131,
+      "step": 115385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.276998143668756e-06,
+      "loss": 1.0289,
+      "step": 115390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.27625376917668e-06,
+      "loss": 1.6259,
+      "step": 115395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.27550940452812e-06,
+      "loss": 1.1103,
+      "step": 115400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.274765049727327e-06,
+      "loss": 0.9968,
+      "step": 115405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.274020704778558e-06,
+      "loss": 1.0725,
+      "step": 115410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.273276369686058e-06,
+      "loss": 1.2248,
+      "step": 115415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.272532044454075e-06,
+      "loss": 1.1816,
+      "step": 115420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.271787729086867e-06,
+      "loss": 1.3875,
+      "step": 115425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.27104342358868e-06,
+      "loss": 1.1822,
+      "step": 115430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.270299127963763e-06,
+      "loss": 1.1332,
+      "step": 115435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.26955484221637e-06,
+      "loss": 1.3819,
+      "step": 115440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.268810566350752e-06,
+      "loss": 1.3575,
+      "step": 115445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.268066300371151e-06,
+      "loss": 1.4149,
+      "step": 115450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.267322044281825e-06,
+      "loss": 1.5696,
+      "step": 115455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.266577798087025e-06,
+      "loss": 1.415,
+      "step": 115460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.265833561790997e-06,
+      "loss": 1.2033,
+      "step": 115465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.26508933539799e-06,
+      "loss": 1.2141,
+      "step": 115470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.264345118912256e-06,
+      "loss": 1.2999,
+      "step": 115475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.26360091233805e-06,
+      "loss": 1.2463,
+      "step": 115480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.262856715679611e-06,
+      "loss": 1.3273,
+      "step": 115485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.262112528941196e-06,
+      "loss": 0.9409,
+      "step": 115490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.261368352127054e-06,
+      "loss": 1.4311,
+      "step": 115495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.26062418524143e-06,
+      "loss": 1.3898,
+      "step": 115500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.259880028288584e-06,
+      "loss": 1.2327,
+      "step": 115505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.259135881272756e-06,
+      "loss": 1.2213,
+      "step": 115510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.258391744198195e-06,
+      "loss": 1.1617,
+      "step": 115515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.257647617069157e-06,
+      "loss": 1.2629,
+      "step": 115520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.256903499889893e-06,
+      "loss": 1.3773,
+      "step": 115525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.25615939266464e-06,
+      "loss": 1.0743,
+      "step": 115530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.255415295397658e-06,
+      "loss": 1.3263,
+      "step": 115535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.254671208093194e-06,
+      "loss": 1.3112,
+      "step": 115540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.253927130755495e-06,
+      "loss": 1.3994,
+      "step": 115545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.25318306338881e-06,
+      "loss": 1.0957,
+      "step": 115550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.252439005997389e-06,
+      "loss": 1.2914,
+      "step": 115555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.251694958585485e-06,
+      "loss": 1.614,
+      "step": 115560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.250950921157339e-06,
+      "loss": 1.374,
+      "step": 115565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.250206893717205e-06,
+      "loss": 1.3421,
+      "step": 115570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.249462876269334e-06,
+      "loss": 1.2685,
+      "step": 115575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.248718868817964e-06,
+      "loss": 1.1891,
+      "step": 115580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.247974871367359e-06,
+      "loss": 1.1977,
+      "step": 115585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.247230883921756e-06,
+      "loss": 1.1433,
+      "step": 115590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.246486906485407e-06,
+      "loss": 0.9389,
+      "step": 115595
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.245742939062565e-06,
+      "loss": 1.024,
+      "step": 115600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.24499898165747e-06,
+      "loss": 1.2171,
+      "step": 115605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.244255034274379e-06,
+      "loss": 1.6382,
+      "step": 115610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.243511096917528e-06,
+      "loss": 1.408,
+      "step": 115615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.242767169591181e-06,
+      "loss": 1.2881,
+      "step": 115620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.242023252299576e-06,
+      "loss": 1.2144,
+      "step": 115625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.241279345046963e-06,
+      "loss": 1.5951,
+      "step": 115630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.240535447837594e-06,
+      "loss": 1.1458,
+      "step": 115635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.239791560675711e-06,
+      "loss": 1.2967,
+      "step": 115640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.239047683565564e-06,
+      "loss": 1.4332,
+      "step": 115645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.238303816511404e-06,
+      "loss": 1.3004,
+      "step": 115650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.237559959517478e-06,
+      "loss": 1.386,
+      "step": 115655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.23681611258803e-06,
+      "loss": 1.2015,
+      "step": 115660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.236072275727311e-06,
+      "loss": 1.1222,
+      "step": 115665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.23532844893957e-06,
+      "loss": 1.7166,
+      "step": 115670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.234584632229046e-06,
+      "loss": 1.0788,
+      "step": 115675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.233840825600002e-06,
+      "loss": 1.0814,
+      "step": 115680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.233097029056672e-06,
+      "loss": 1.0807,
+      "step": 115685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.232353242603313e-06,
+      "loss": 1.3378,
+      "step": 115690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.231609466244162e-06,
+      "loss": 1.1695,
+      "step": 115695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.230865699983474e-06,
+      "loss": 1.2763,
+      "step": 115700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.230121943825494e-06,
+      "loss": 1.3176,
+      "step": 115705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.229378197774468e-06,
+      "loss": 1.145,
+      "step": 115710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.22863446183465e-06,
+      "loss": 1.0361,
+      "step": 115715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.227890736010278e-06,
+      "loss": 1.1119,
+      "step": 115720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.227147020305602e-06,
+      "loss": 1.4972,
+      "step": 115725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.226403314724874e-06,
+      "loss": 1.1329,
+      "step": 115730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.225659619272336e-06,
+      "loss": 1.1135,
+      "step": 115735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.224915933952232e-06,
+      "loss": 1.4549,
+      "step": 115740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.224172258768816e-06,
+      "loss": 1.1474,
+      "step": 115745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.223428593726333e-06,
+      "loss": 1.1729,
+      "step": 115750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.222684938829028e-06,
+      "loss": 1.3216,
+      "step": 115755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.221941294081144e-06,
+      "loss": 1.1865,
+      "step": 115760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.221197659486934e-06,
+      "loss": 1.1074,
+      "step": 115765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.220454035050645e-06,
+      "loss": 1.1106,
+      "step": 115770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.219710420776514e-06,
+      "loss": 1.2145,
+      "step": 115775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.218966816668799e-06,
+      "loss": 1.1623,
+      "step": 115780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.218223222731742e-06,
+      "loss": 1.1135,
+      "step": 115785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.217479638969585e-06,
+      "loss": 1.4541,
+      "step": 115790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.216736065386579e-06,
+      "loss": 1.1622,
+      "step": 115795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.21599250198697e-06,
+      "loss": 1.2332,
+      "step": 115800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.215248948775001e-06,
+      "loss": 1.4266,
+      "step": 115805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.214505405754926e-06,
+      "loss": 1.0756,
+      "step": 115810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.213761872930982e-06,
+      "loss": 1.3587,
+      "step": 115815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.213018350307416e-06,
+      "loss": 1.2732,
+      "step": 115820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.21227483788848e-06,
+      "loss": 1.7225,
+      "step": 115825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.211531335678417e-06,
+      "loss": 1.1858,
+      "step": 115830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.21078784368147e-06,
+      "loss": 1.4178,
+      "step": 115835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.210044361901886e-06,
+      "loss": 1.1748,
+      "step": 115840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.209300890343915e-06,
+      "loss": 1.0113,
+      "step": 115845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.208557429011795e-06,
+      "loss": 1.0286,
+      "step": 115850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.207813977909775e-06,
+      "loss": 1.0564,
+      "step": 115855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.207070537042103e-06,
+      "loss": 0.9972,
+      "step": 115860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.206327106413026e-06,
+      "loss": 1.3522,
+      "step": 115865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.20558368602678e-06,
+      "loss": 1.2303,
+      "step": 115870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.20484027588762e-06,
+      "loss": 1.9527,
+      "step": 115875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.204096875999789e-06,
+      "loss": 1.4911,
+      "step": 115880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.203353486367523e-06,
+      "loss": 1.0406,
+      "step": 115885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.202610106995083e-06,
+      "loss": 1.3589,
+      "step": 115890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.201866737886703e-06,
+      "loss": 1.4266,
+      "step": 115895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.201123379046634e-06,
+      "loss": 1.1831,
+      "step": 115900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.200380030479114e-06,
+      "loss": 1.1329,
+      "step": 115905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.199636692188393e-06,
+      "loss": 1.2229,
+      "step": 115910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.198893364178717e-06,
+      "loss": 1.3046,
+      "step": 115915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.198150046454323e-06,
+      "loss": 1.1912,
+      "step": 115920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.197406739019467e-06,
+      "loss": 1.1319,
+      "step": 115925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.196663441878387e-06,
+      "loss": 1.1924,
+      "step": 115930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.195920155035326e-06,
+      "loss": 1.5931,
+      "step": 115935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.195176878494534e-06,
+      "loss": 0.9729,
+      "step": 115940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.194433612260252e-06,
+      "loss": 0.8435,
+      "step": 115945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.193690356336724e-06,
+      "loss": 1.4452,
+      "step": 115950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.192947110728195e-06,
+      "loss": 1.0486,
+      "step": 115955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.192203875438915e-06,
+      "loss": 1.411,
+      "step": 115960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.191460650473117e-06,
+      "loss": 1.0684,
+      "step": 115965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.190717435835054e-06,
+      "loss": 1.1452,
+      "step": 115970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.18997423152897e-06,
+      "loss": 1.1252,
+      "step": 115975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.189231037559104e-06,
+      "loss": 1.0429,
+      "step": 115980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.188487853929702e-06,
+      "loss": 1.464,
+      "step": 115985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.18774468064501e-06,
+      "loss": 1.3763,
+      "step": 115990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.187001517709272e-06,
+      "loss": 1.4192,
+      "step": 115995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.186258365126727e-06,
+      "loss": 1.0935,
+      "step": 116000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.185515222901624e-06,
+      "loss": 1.768,
+      "step": 116005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.184772091038208e-06,
+      "loss": 1.4487,
+      "step": 116010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.184028969540711e-06,
+      "loss": 0.955,
+      "step": 116015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.183285858413395e-06,
+      "loss": 1.131,
+      "step": 116020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.18254275766049e-06,
+      "loss": 1.0544,
+      "step": 116025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.181799667286241e-06,
+      "loss": 0.94,
+      "step": 116030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.181056587294898e-06,
+      "loss": 1.241,
+      "step": 116035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.180313517690699e-06,
+      "loss": 1.2674,
+      "step": 116040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.179570458477887e-06,
+      "loss": 1.5863,
+      "step": 116045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.178827409660706e-06,
+      "loss": 1.416,
+      "step": 116050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.178084371243404e-06,
+      "loss": 1.5352,
+      "step": 116055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.177341343230218e-06,
+      "loss": 1.23,
+      "step": 116060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.176598325625391e-06,
+      "loss": 1.1466,
+      "step": 116065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.175855318433174e-06,
+      "loss": 1.701,
+      "step": 116070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.1751123216578e-06,
+      "loss": 1.1342,
+      "step": 116075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.174369335303514e-06,
+      "loss": 1.5817,
+      "step": 116080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.173626359374565e-06,
+      "loss": 0.9438,
+      "step": 116085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.172883393875192e-06,
+      "loss": 1.1977,
+      "step": 116090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.172140438809635e-06,
+      "loss": 1.2757,
+      "step": 116095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.17139749418214e-06,
+      "loss": 1.2738,
+      "step": 116100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.17065455999695e-06,
+      "loss": 1.2609,
+      "step": 116105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.169911636258303e-06,
+      "loss": 1.3705,
+      "step": 116110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.16916872297045e-06,
+      "loss": 1.1352,
+      "step": 116115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.168425820137625e-06,
+      "loss": 1.0462,
+      "step": 116120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.167682927764078e-06,
+      "loss": 1.3663,
+      "step": 116125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.16694004585404e-06,
+      "loss": 1.406,
+      "step": 116130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.166197174411765e-06,
+      "loss": 1.1017,
+      "step": 116135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.165454313441489e-06,
+      "loss": 1.0522,
+      "step": 116140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.164711462947453e-06,
+      "loss": 1.189,
+      "step": 116145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.163968622933907e-06,
+      "loss": 1.4459,
+      "step": 116150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.163225793405086e-06,
+      "loss": 1.1481,
+      "step": 116155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.16248297436523e-06,
+      "loss": 1.0282,
+      "step": 116160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.161740165818588e-06,
+      "loss": 1.3889,
+      "step": 116165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.160997367769399e-06,
+      "loss": 1.0153,
+      "step": 116170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.160254580221902e-06,
+      "loss": 1.3504,
+      "step": 116175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.159511803180341e-06,
+      "loss": 1.2503,
+      "step": 116180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.15876903664896e-06,
+      "loss": 1.2265,
+      "step": 116185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.158026280631997e-06,
+      "loss": 1.3009,
+      "step": 116190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.15728353513369e-06,
+      "loss": 1.2856,
+      "step": 116195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.15654080015829e-06,
+      "loss": 1.115,
+      "step": 116200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.155798075710034e-06,
+      "loss": 1.3639,
+      "step": 116205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.15505536179316e-06,
+      "loss": 1.3786,
+      "step": 116210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.154312658411914e-06,
+      "loss": 1.0549,
+      "step": 116215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.153569965570537e-06,
+      "loss": 1.3085,
+      "step": 116220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.152827283273264e-06,
+      "loss": 1.0875,
+      "step": 116225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.152084611524344e-06,
+      "loss": 1.5018,
+      "step": 116230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.151341950328015e-06,
+      "loss": 1.2617,
+      "step": 116235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.150599299688515e-06,
+      "loss": 1.3502,
+      "step": 116240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.149856659610092e-06,
+      "loss": 1.4548,
+      "step": 116245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.149114030096981e-06,
+      "loss": 1.194,
+      "step": 116250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.14837141115342e-06,
+      "loss": 1.3629,
+      "step": 116255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.147628802783658e-06,
+      "loss": 1.4797,
+      "step": 116260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.146886204991934e-06,
+      "loss": 1.1431,
+      "step": 116265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.146143617782485e-06,
+      "loss": 1.2741,
+      "step": 116270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.14540104115955e-06,
+      "loss": 1.1707,
+      "step": 116275
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.144658475127375e-06,
+      "loss": 1.4765,
+      "step": 116280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.1439159196902e-06,
+      "loss": 1.6695,
+      "step": 116285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.143173374852259e-06,
+      "loss": 1.0837,
+      "step": 116290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.142430840617798e-06,
+      "loss": 1.0061,
+      "step": 116295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.141688316991058e-06,
+      "loss": 0.8599,
+      "step": 116300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.140945803976274e-06,
+      "loss": 0.9995,
+      "step": 116305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.140203301577692e-06,
+      "loss": 1.0922,
+      "step": 116310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.139460809799552e-06,
+      "loss": 1.2012,
+      "step": 116315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.138718328646083e-06,
+      "loss": 1.3629,
+      "step": 116320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.137975858121543e-06,
+      "loss": 1.1386,
+      "step": 116325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.137233398230156e-06,
+      "loss": 1.173,
+      "step": 116330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.136490948976173e-06,
+      "loss": 1.0662,
+      "step": 116335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.135748510363824e-06,
+      "loss": 1.0645,
+      "step": 116340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.135006082397357e-06,
+      "loss": 1.2622,
+      "step": 116345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.134263665081007e-06,
+      "loss": 1.1962,
+      "step": 116350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.133521258419012e-06,
+      "loss": 1.0483,
+      "step": 116355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.13277886241562e-06,
+      "loss": 1.1715,
+      "step": 116360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.132036477075064e-06,
+      "loss": 1.1258,
+      "step": 116365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.13129410240158e-06,
+      "loss": 1.1561,
+      "step": 116370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.130551738399417e-06,
+      "loss": 1.0655,
+      "step": 116375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.129809385072809e-06,
+      "loss": 1.7142,
+      "step": 116380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.129067042425991e-06,
+      "loss": 1.6418,
+      "step": 116385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.128324710463208e-06,
+      "loss": 1.2627,
+      "step": 116390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.1275823891887e-06,
+      "loss": 1.2785,
+      "step": 116395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.1268400786067e-06,
+      "loss": 0.9999,
+      "step": 116400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.126097778721453e-06,
+      "loss": 1.1426,
+      "step": 116405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.125355489537195e-06,
+      "loss": 1.3779,
+      "step": 116410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.124613211058168e-06,
+      "loss": 1.1645,
+      "step": 116415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.123870943288604e-06,
+      "loss": 1.306,
+      "step": 116420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.123128686232748e-06,
+      "loss": 1.0077,
+      "step": 116425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.122386439894839e-06,
+      "loss": 1.3585,
+      "step": 116430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.12164420427911e-06,
+      "loss": 1.1126,
+      "step": 116435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.120901979389802e-06,
+      "loss": 1.2028,
+      "step": 116440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.120159765231157e-06,
+      "loss": 1.5184,
+      "step": 116445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.119417561807408e-06,
+      "loss": 1.6391,
+      "step": 116450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.1186753691228e-06,
+      "loss": 1.3602,
+      "step": 116455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.117933187181567e-06,
+      "loss": 1.3173,
+      "step": 116460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.117191015987943e-06,
+      "loss": 1.0365,
+      "step": 116465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.116448855546176e-06,
+      "loss": 1.1608,
+      "step": 116470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.115706705860498e-06,
+      "loss": 1.5313,
+      "step": 116475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.114964566935148e-06,
+      "loss": 1.2315,
+      "step": 116480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.114222438774362e-06,
+      "loss": 1.2059,
+      "step": 116485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.113480321382382e-06,
+      "loss": 1.0963,
+      "step": 116490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.112738214763445e-06,
+      "loss": 1.1828,
+      "step": 116495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.111996118921786e-06,
+      "loss": 1.2567,
+      "step": 116500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.111254033861645e-06,
+      "loss": 1.0852,
+      "step": 116505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.110511959587261e-06,
+      "loss": 1.0551,
+      "step": 116510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.109769896102869e-06,
+      "loss": 1.2901,
+      "step": 116515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.109027843412707e-06,
+      "loss": 1.1937,
+      "step": 116520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.108285801521015e-06,
+      "loss": 1.3001,
+      "step": 116525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.107543770432025e-06,
+      "loss": 1.0801,
+      "step": 116530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.106801750149979e-06,
+      "loss": 1.4511,
+      "step": 116535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.106059740679116e-06,
+      "loss": 1.2384,
+      "step": 116540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.105317742023666e-06,
+      "loss": 1.1449,
+      "step": 116545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.104575754187876e-06,
+      "loss": 1.1032,
+      "step": 116550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.103833777175976e-06,
+      "loss": 1.0974,
+      "step": 116555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.103091810992206e-06,
+      "loss": 1.2445,
+      "step": 116560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.102349855640797e-06,
+      "loss": 0.8871,
+      "step": 116565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.101607911125995e-06,
+      "loss": 1.2552,
+      "step": 116570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.100865977452033e-06,
+      "loss": 1.2088,
+      "step": 116575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.100124054623145e-06,
+      "loss": 1.4653,
+      "step": 116580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.099382142643575e-06,
+      "loss": 1.508,
+      "step": 116585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.098640241517552e-06,
+      "loss": 1.455,
+      "step": 116590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.097898351249315e-06,
+      "loss": 1.2156,
+      "step": 116595
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.097156471843102e-06,
+      "loss": 1.4107,
+      "step": 116600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.096414603303153e-06,
+      "loss": 1.7275,
+      "step": 116605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.095672745633696e-06,
+      "loss": 1.1647,
+      "step": 116610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.094930898838973e-06,
+      "loss": 0.9824,
+      "step": 116615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.094189062923223e-06,
+      "loss": 1.5703,
+      "step": 116620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.093447237890675e-06,
+      "loss": 1.3805,
+      "step": 116625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.092705423745566e-06,
+      "loss": 1.3419,
+      "step": 116630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.091963620492138e-06,
+      "loss": 1.1332,
+      "step": 116635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.091221828134626e-06,
+      "loss": 1.3834,
+      "step": 116640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.090480046677258e-06,
+      "loss": 1.1395,
+      "step": 116645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.089738276124281e-06,
+      "loss": 1.2732,
+      "step": 116650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.088996516479926e-06,
+      "loss": 1.2428,
+      "step": 116655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.088254767748423e-06,
+      "loss": 1.6735,
+      "step": 116660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.08751302993402e-06,
+      "loss": 1.4966,
+      "step": 116665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.086771303040944e-06,
+      "loss": 1.2106,
+      "step": 116670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.086029587073431e-06,
+      "loss": 1.3153,
+      "step": 116675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.085287882035723e-06,
+      "loss": 1.1763,
+      "step": 116680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.084546187932048e-06,
+      "loss": 1.0019,
+      "step": 116685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.083804504766642e-06,
+      "loss": 1.0975,
+      "step": 116690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.083062832543747e-06,
+      "loss": 1.228,
+      "step": 116695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.082321171267595e-06,
+      "loss": 1.329,
+      "step": 116700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.081579520942418e-06,
+      "loss": 1.2423,
+      "step": 116705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.080837881572452e-06,
+      "loss": 1.2057,
+      "step": 116710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.08009625316194e-06,
+      "loss": 1.1204,
+      "step": 116715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.079354635715107e-06,
+      "loss": 1.0731,
+      "step": 116720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.078613029236189e-06,
+      "loss": 0.9402,
+      "step": 116725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.077871433729428e-06,
+      "loss": 1.4745,
+      "step": 116730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.077129849199056e-06,
+      "loss": 1.5104,
+      "step": 116735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.076388275649303e-06,
+      "loss": 1.027,
+      "step": 116740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.07564671308441e-06,
+      "loss": 1.2184,
+      "step": 116745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.07490516150861e-06,
+      "loss": 1.083,
+      "step": 116750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.074163620926133e-06,
+      "loss": 1.3012,
+      "step": 116755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.073422091341222e-06,
+      "loss": 1.1361,
+      "step": 116760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.072680572758107e-06,
+      "loss": 0.936,
+      "step": 116765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.071939065181023e-06,
+      "loss": 1.178,
+      "step": 116770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.0711975686142e-06,
+      "loss": 1.1523,
+      "step": 116775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.070456083061878e-06,
+      "loss": 1.1482,
+      "step": 116780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.06971460852829e-06,
+      "loss": 1.2159,
+      "step": 116785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.068973145017668e-06,
+      "loss": 1.0498,
+      "step": 116790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.068231692534251e-06,
+      "loss": 1.0307,
+      "step": 116795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.06749025108227e-06,
+      "loss": 1.1968,
+      "step": 116800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.066748820665955e-06,
+      "loss": 1.1968,
+      "step": 116805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.066007401289546e-06,
+      "loss": 1.1665,
+      "step": 116810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.06526599295728e-06,
+      "loss": 1.0168,
+      "step": 116815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.064524595673377e-06,
+      "loss": 1.0024,
+      "step": 116820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.063783209442084e-06,
+      "loss": 1.294,
+      "step": 116825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.063041834267633e-06,
+      "loss": 1.3615,
+      "step": 116830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.06230047015425e-06,
+      "loss": 1.48,
+      "step": 116835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.061559117106176e-06,
+      "loss": 1.2061,
+      "step": 116840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.060817775127643e-06,
+      "loss": 1.2397,
+      "step": 116845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.060076444222884e-06,
+      "loss": 1.2957,
+      "step": 116850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.059335124396128e-06,
+      "loss": 1.2133,
+      "step": 116855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.058593815651613e-06,
+      "loss": 1.2069,
+      "step": 116860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.057852517993576e-06,
+      "loss": 1.3441,
+      "step": 116865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.05711123142624e-06,
+      "loss": 1.3246,
+      "step": 116870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.056369955953847e-06,
+      "loss": 1.1068,
+      "step": 116875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.055628691580627e-06,
+      "loss": 1.1893,
+      "step": 116880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.05488743831081e-06,
+      "loss": 1.1008,
+      "step": 116885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.054146196148636e-06,
+      "loss": 1.1722,
+      "step": 116890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.053404965098333e-06,
+      "loss": 0.9629,
+      "step": 116895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.05266374516413e-06,
+      "loss": 0.9315,
+      "step": 116900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.05192253635027e-06,
+      "loss": 1.0414,
+      "step": 116905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.05118133866098e-06,
+      "loss": 1.1741,
+      "step": 116910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.050440152100491e-06,
+      "loss": 1.093,
+      "step": 116915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.049698976673035e-06,
+      "loss": 1.2326,
+      "step": 116920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.048957812382851e-06,
+      "loss": 1.3882,
+      "step": 116925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.048216659234166e-06,
+      "loss": 1.3764,
+      "step": 116930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.04747551723121e-06,
+      "loss": 1.0599,
+      "step": 116935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.046734386378223e-06,
+      "loss": 1.5307,
+      "step": 116940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.045993266679434e-06,
+      "loss": 1.1758,
+      "step": 116945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.04525215813907e-06,
+      "loss": 1.3004,
+      "step": 116950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.04451106076137e-06,
+      "loss": 1.1842,
+      "step": 116955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.043769974550565e-06,
+      "loss": 1.1113,
+      "step": 116960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.043028899510883e-06,
+      "loss": 1.0307,
+      "step": 116965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.04228783564656e-06,
+      "loss": 1.3482,
+      "step": 116970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.041546782961827e-06,
+      "loss": 1.1577,
+      "step": 116975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.040805741460914e-06,
+      "loss": 1.2251,
+      "step": 116980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.040064711148055e-06,
+      "loss": 1.1211,
+      "step": 116985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.03932369202748e-06,
+      "loss": 1.1092,
+      "step": 116990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.038582684103424e-06,
+      "loss": 1.4382,
+      "step": 116995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.037841687380108e-06,
+      "loss": 1.2174,
+      "step": 117000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.03710070186178e-06,
+      "loss": 1.1538,
+      "step": 117005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.03635972755266e-06,
+      "loss": 1.4655,
+      "step": 117010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.035618764456978e-06,
+      "loss": 1.1546,
+      "step": 117015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.034877812578974e-06,
+      "loss": 1.6168,
+      "step": 117020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.034136871922875e-06,
+      "loss": 1.1522,
+      "step": 117025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.033395942492908e-06,
+      "loss": 1.2348,
+      "step": 117030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.03265502429331e-06,
+      "loss": 1.3252,
+      "step": 117035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.031914117328312e-06,
+      "loss": 1.4735,
+      "step": 117040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.03117322160214e-06,
+      "loss": 1.1391,
+      "step": 117045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.030432337119028e-06,
+      "loss": 1.0032,
+      "step": 117050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.02969146388321e-06,
+      "loss": 1.4112,
+      "step": 117055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.028950601898907e-06,
+      "loss": 1.0826,
+      "step": 117060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.028209751170363e-06,
+      "loss": 1.1507,
+      "step": 117065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.027468911701798e-06,
+      "loss": 1.2999,
+      "step": 117070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.02672808349745e-06,
+      "loss": 1.4851,
+      "step": 117075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.025987266561541e-06,
+      "loss": 1.3174,
+      "step": 117080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.025246460898311e-06,
+      "loss": 1.3686,
+      "step": 117085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.024505666511982e-06,
+      "loss": 1.701,
+      "step": 117090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.023764883406788e-06,
+      "loss": 1.5601,
+      "step": 117095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.023024111586965e-06,
+      "loss": 1.4796,
+      "step": 117100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.022283351056733e-06,
+      "loss": 1.3289,
+      "step": 117105
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.021542601820325e-06,
+      "loss": 1.2764,
+      "step": 117110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.02080186388198e-06,
+      "loss": 1.1014,
+      "step": 117115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.020061137245915e-06,
+      "loss": 1.1065,
+      "step": 117120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.019320421916365e-06,
+      "loss": 1.1185,
+      "step": 117125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.018579717897562e-06,
+      "loss": 1.2035,
+      "step": 117130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.01783902519374e-06,
+      "loss": 1.2374,
+      "step": 117135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.01709834380912e-06,
+      "loss": 1.121,
+      "step": 117140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.01635767374793e-06,
+      "loss": 1.2843,
+      "step": 117145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.01561701501441e-06,
+      "loss": 1.1814,
+      "step": 117150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.014876367612783e-06,
+      "loss": 1.4351,
+      "step": 117155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.014135731547278e-06,
+      "loss": 1.2245,
+      "step": 117160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.013395106822127e-06,
+      "loss": 1.8051,
+      "step": 117165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.01265449344156e-06,
+      "loss": 0.9692,
+      "step": 117170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.011913891409802e-06,
+      "loss": 1.1985,
+      "step": 117175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.011173300731087e-06,
+      "loss": 1.1047,
+      "step": 117180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.010432721409643e-06,
+      "loss": 1.3127,
+      "step": 117185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.009692153449694e-06,
+      "loss": 1.1279,
+      "step": 117190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.008951596855478e-06,
+      "loss": 1.1254,
+      "step": 117195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.008211051631218e-06,
+      "loss": 1.2308,
+      "step": 117200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.007470517781141e-06,
+      "loss": 1.1185,
+      "step": 117205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.006729995309482e-06,
+      "loss": 1.0126,
+      "step": 117210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.00598948422047e-06,
+      "loss": 1.7401,
+      "step": 117215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.005248984518326e-06,
+      "loss": 1.2666,
+      "step": 117220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.004508496207282e-06,
+      "loss": 1.5883,
+      "step": 117225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.00376801929157e-06,
+      "loss": 1.0393,
+      "step": 117230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.003027553775418e-06,
+      "loss": 1.3879,
+      "step": 117235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.00228709966305e-06,
+      "loss": 1.3865,
+      "step": 117240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.001546656958697e-06,
+      "loss": 1.3088,
+      "step": 117245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.00080622566659e-06,
+      "loss": 1.285,
+      "step": 117250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.000065805790951e-06,
+      "loss": 1.1929,
+      "step": 117255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.999325397336015e-06,
+      "loss": 1.1631,
+      "step": 117260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.998585000306008e-06,
+      "loss": 1.1476,
+      "step": 117265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.997844614705151e-06,
+      "loss": 1.316,
+      "step": 117270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.997104240537685e-06,
+      "loss": 1.3092,
+      "step": 117275
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.996363877807828e-06,
+      "loss": 1.5657,
+      "step": 117280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.995623526519815e-06,
+      "loss": 1.3323,
+      "step": 117285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.994883186677864e-06,
+      "loss": 1.015,
+      "step": 117290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.99414285828621e-06,
+      "loss": 1.303,
+      "step": 117295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.993402541349083e-06,
+      "loss": 1.233,
+      "step": 117300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.992662235870698e-06,
+      "loss": 1.1861,
+      "step": 117305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.9919219418553e-06,
+      "loss": 1.369,
+      "step": 117310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.991181659307104e-06,
+      "loss": 1.1135,
+      "step": 117315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.990441388230339e-06,
+      "loss": 1.3154,
+      "step": 117320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.989701128629241e-06,
+      "loss": 1.3239,
+      "step": 117325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.988960880508026e-06,
+      "loss": 1.312,
+      "step": 117330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.988220643870925e-06,
+      "loss": 1.7403,
+      "step": 117335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.987480418722169e-06,
+      "loss": 1.1404,
+      "step": 117340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.986740205065984e-06,
+      "loss": 1.0759,
+      "step": 117345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.98600000290659e-06,
+      "loss": 1.1396,
+      "step": 117350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.985259812248222e-06,
+      "loss": 0.9555,
+      "step": 117355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.984519633095106e-06,
+      "loss": 1.2738,
+      "step": 117360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.983779465451464e-06,
+      "loss": 1.204,
+      "step": 117365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.983039309321526e-06,
+      "loss": 1.3468,
+      "step": 117370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.98229916470952e-06,
+      "loss": 1.2467,
+      "step": 117375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.981559031619672e-06,
+      "loss": 0.9417,
+      "step": 117380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.980818910056202e-06,
+      "loss": 1.3133,
+      "step": 117385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.980078800023347e-06,
+      "loss": 0.9908,
+      "step": 117390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.979338701525327e-06,
+      "loss": 1.2078,
+      "step": 117395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.978598614566367e-06,
+      "loss": 1.3874,
+      "step": 117400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.977858539150701e-06,
+      "loss": 1.2761,
+      "step": 117405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.977118475282548e-06,
+      "loss": 1.6293,
+      "step": 117410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.976378422966136e-06,
+      "loss": 0.98,
+      "step": 117415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.975638382205695e-06,
+      "loss": 1.4322,
+      "step": 117420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.974898353005445e-06,
+      "loss": 0.9603,
+      "step": 117425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.974158335369615e-06,
+      "loss": 1.1143,
+      "step": 117430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.973418329302427e-06,
+      "loss": 1.1578,
+      "step": 117435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.972678334808116e-06,
+      "loss": 0.9678,
+      "step": 117440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.9719383518909e-06,
+      "loss": 1.062,
+      "step": 117445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.971198380555005e-06,
+      "loss": 1.211,
+      "step": 117450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.97045842080466e-06,
+      "loss": 1.2819,
+      "step": 117455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.969718472644092e-06,
+      "loss": 1.1784,
+      "step": 117460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.968978536077517e-06,
+      "loss": 1.0963,
+      "step": 117465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.968238611109172e-06,
+      "loss": 1.178,
+      "step": 117470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.967498697743276e-06,
+      "loss": 1.0126,
+      "step": 117475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.966758795984053e-06,
+      "loss": 1.3323,
+      "step": 117480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.966018905835735e-06,
+      "loss": 1.1453,
+      "step": 117485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.96527902730254e-06,
+      "loss": 1.5255,
+      "step": 117490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.964539160388695e-06,
+      "loss": 1.4113,
+      "step": 117495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.963799305098432e-06,
+      "loss": 1.1596,
+      "step": 117500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.963059461435966e-06,
+      "loss": 1.4662,
+      "step": 117505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.96231962940553e-06,
+      "loss": 1.198,
+      "step": 117510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.961579809011337e-06,
+      "loss": 0.9865,
+      "step": 117515
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.960840000257626e-06,
+      "loss": 1.3368,
+      "step": 117520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.960100203148615e-06,
+      "loss": 1.2163,
+      "step": 117525
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.959360417688525e-06,
+      "loss": 1.2476,
+      "step": 117530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.95862064388159e-06,
+      "loss": 1.0539,
+      "step": 117535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.957880881732027e-06,
+      "loss": 1.0045,
+      "step": 117540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.95714113124406e-06,
+      "loss": 1.1929,
+      "step": 117545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.956401392421918e-06,
+      "loss": 1.5361,
+      "step": 117550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.955661665269823e-06,
+      "loss": 1.0531,
+      "step": 117555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.954921949791998e-06,
+      "loss": 1.251,
+      "step": 117560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.954182245992669e-06,
+      "loss": 1.0188,
+      "step": 117565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.953442553876065e-06,
+      "loss": 1.6118,
+      "step": 117570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.9527028734464e-06,
+      "loss": 1.2551,
+      "step": 117575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.9519632047079e-06,
+      "loss": 1.6028,
+      "step": 117580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.951223547664796e-06,
+      "loss": 1.1298,
+      "step": 117585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.950483902321308e-06,
+      "loss": 1.2898,
+      "step": 117590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.949744268681656e-06,
+      "loss": 1.1808,
+      "step": 117595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.94900464675007e-06,
+      "loss": 1.1056,
+      "step": 117600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.948265036530772e-06,
+      "loss": 1.395,
+      "step": 117605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.947525438027976e-06,
+      "loss": 1.1199,
+      "step": 117610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.94678585124592e-06,
+      "loss": 1.2706,
+      "step": 117615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.946046276188822e-06,
+      "loss": 1.2278,
+      "step": 117620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.9453067128609e-06,
+      "loss": 1.2251,
+      "step": 117625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.944567161266386e-06,
+      "loss": 1.2333,
+      "step": 117630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.943827621409498e-06,
+      "loss": 1.0835,
+      "step": 117635
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.943088093294458e-06,
+      "loss": 1.0351,
+      "step": 117640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.942348576925492e-06,
+      "loss": 1.1146,
+      "step": 117645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.941609072306827e-06,
+      "loss": 1.1547,
+      "step": 117650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.940869579442677e-06,
+      "loss": 1.3198,
+      "step": 117655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.940130098337268e-06,
+      "loss": 1.3242,
+      "step": 117660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.939390628994827e-06,
+      "loss": 1.176,
+      "step": 117665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.938651171419573e-06,
+      "loss": 1.2228,
+      "step": 117670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.937911725615728e-06,
+      "loss": 1.1103,
+      "step": 117675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.937172291587517e-06,
+      "loss": 1.1738,
+      "step": 117680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.936432869339165e-06,
+      "loss": 1.4495,
+      "step": 117685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.935693458874885e-06,
+      "loss": 1.0224,
+      "step": 117690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.93495406019891e-06,
+      "loss": 1.3552,
+      "step": 117695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.93421467331546e-06,
+      "loss": 0.9847,
+      "step": 117700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.933475298228747e-06,
+      "loss": 1.3384,
+      "step": 117705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.932735934943009e-06,
+      "loss": 1.1302,
+      "step": 117710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.931996583462458e-06,
+      "loss": 1.0229,
+      "step": 117715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.93125724379132e-06,
+      "loss": 1.041,
+      "step": 117720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.930517915933814e-06,
+      "loss": 1.0618,
+      "step": 117725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.929778599894166e-06,
+      "loss": 1.2271,
+      "step": 117730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.929039295676594e-06,
+      "loss": 1.3053,
+      "step": 117735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.928300003285319e-06,
+      "loss": 1.2163,
+      "step": 117740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.927560722724571e-06,
+      "loss": 1.4214,
+      "step": 117745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.926821453998563e-06,
+      "loss": 1.1804,
+      "step": 117750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.926082197111517e-06,
+      "loss": 1.3085,
+      "step": 117755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.925342952067662e-06,
+      "loss": 1.4439,
+      "step": 117760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.924603718871213e-06,
+      "loss": 1.2017,
+      "step": 117765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.923864497526391e-06,
+      "loss": 1.3982,
+      "step": 117770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.923125288037422e-06,
+      "loss": 1.3107,
+      "step": 117775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.922386090408525e-06,
+      "loss": 1.0003,
+      "step": 117780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.921646904643917e-06,
+      "loss": 1.2254,
+      "step": 117785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.920907730747826e-06,
+      "loss": 1.1787,
+      "step": 117790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.920168568724468e-06,
+      "loss": 1.3663,
+      "step": 117795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.91942941857807e-06,
+      "loss": 1.0153,
+      "step": 117800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.918690280312847e-06,
+      "loss": 1.2438,
+      "step": 117805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.91795115393302e-06,
+      "loss": 1.0921,
+      "step": 117810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.917212039442816e-06,
+      "loss": 1.1969,
+      "step": 117815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.916472936846447e-06,
+      "loss": 1.7041,
+      "step": 117820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.91573384614814e-06,
+      "loss": 0.9816,
+      "step": 117825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.914994767352114e-06,
+      "loss": 1.3588,
+      "step": 117830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.914255700462586e-06,
+      "loss": 0.9216,
+      "step": 117835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.913516645483785e-06,
+      "loss": 0.9472,
+      "step": 117840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.912777602419924e-06,
+      "loss": 1.5496,
+      "step": 117845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.912038571275224e-06,
+      "loss": 1.1871,
+      "step": 117850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.911299552053909e-06,
+      "loss": 1.2139,
+      "step": 117855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.910560544760198e-06,
+      "loss": 1.0956,
+      "step": 117860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.909821549398308e-06,
+      "loss": 1.5676,
+      "step": 117865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.90908256597246e-06,
+      "loss": 1.1279,
+      "step": 117870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.908343594486878e-06,
+      "loss": 1.0138,
+      "step": 117875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.907604634945778e-06,
+      "loss": 1.2813,
+      "step": 117880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.906865687353379e-06,
+      "loss": 0.9924,
+      "step": 117885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.906126751713905e-06,
+      "loss": 1.0766,
+      "step": 117890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.905387828031574e-06,
+      "loss": 1.4832,
+      "step": 117895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.904648916310602e-06,
+      "loss": 0.9475,
+      "step": 117900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.903910016555215e-06,
+      "loss": 1.5259,
+      "step": 117905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.90317112876963e-06,
+      "loss": 1.1903,
+      "step": 117910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.902432252958058e-06,
+      "loss": 1.3032,
+      "step": 117915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.901693389124734e-06,
+      "loss": 1.6562,
+      "step": 117920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.900954537273868e-06,
+      "loss": 1.3298,
+      "step": 117925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.900215697409677e-06,
+      "loss": 1.0309,
+      "step": 117930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.899476869536387e-06,
+      "loss": 1.1614,
+      "step": 117935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.898738053658214e-06,
+      "loss": 1.4292,
+      "step": 117940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.897999249779378e-06,
+      "loss": 1.281,
+      "step": 117945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.89726045790409e-06,
+      "loss": 1.1816,
+      "step": 117950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.896521678036583e-06,
+      "loss": 1.564,
+      "step": 117955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.895782910181067e-06,
+      "loss": 1.3261,
+      "step": 117960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.89504415434176e-06,
+      "loss": 1.4389,
+      "step": 117965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.894305410522885e-06,
+      "loss": 1.3605,
+      "step": 117970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.893566678728658e-06,
+      "loss": 1.0512,
+      "step": 117975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.892827958963297e-06,
+      "loss": 1.106,
+      "step": 117980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.892089251231023e-06,
+      "loss": 1.3813,
+      "step": 117985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.891350555536054e-06,
+      "loss": 1.3046,
+      "step": 117990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.890611871882606e-06,
+      "loss": 1.0002,
+      "step": 117995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.8898732002749e-06,
+      "loss": 1.1338,
+      "step": 118000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.889134540717153e-06,
+      "loss": 1.2603,
+      "step": 118005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.888395893213584e-06,
+      "loss": 1.2464,
+      "step": 118010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.887657257768406e-06,
+      "loss": 1.4348,
+      "step": 118015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.886918634385842e-06,
+      "loss": 1.315,
+      "step": 118020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.886180023070114e-06,
+      "loss": 1.7449,
+      "step": 118025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.88544142382543e-06,
+      "loss": 1.5022,
+      "step": 118030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.884702836656014e-06,
+      "loss": 1.3505,
+      "step": 118035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.883964261566084e-06,
+      "loss": 1.1984,
+      "step": 118040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.883225698559853e-06,
+      "loss": 1.2617,
+      "step": 118045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.882487147641547e-06,
+      "loss": 1.3362,
+      "step": 118050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.881748608815376e-06,
+      "loss": 1.3752,
+      "step": 118055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.881010082085557e-06,
+      "loss": 1.075,
+      "step": 118060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.880271567456313e-06,
+      "loss": 1.525,
+      "step": 118065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.879533064931858e-06,
+      "loss": 1.1612,
+      "step": 118070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.878794574516407e-06,
+      "loss": 1.2291,
+      "step": 118075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.878056096214182e-06,
+      "loss": 1.4882,
+      "step": 118080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.877317630029401e-06,
+      "loss": 1.0425,
+      "step": 118085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.876579175966276e-06,
+      "loss": 1.3156,
+      "step": 118090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.875840734029023e-06,
+      "loss": 1.0322,
+      "step": 118095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.875102304221865e-06,
+      "loss": 1.3901,
+      "step": 118100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.874363886549017e-06,
+      "loss": 1.1301,
+      "step": 118105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.87362548101469e-06,
+      "loss": 1.3513,
+      "step": 118110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.872887087623109e-06,
+      "loss": 1.2396,
+      "step": 118115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.872148706378486e-06,
+      "loss": 1.0567,
+      "step": 118120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.871410337285036e-06,
+      "loss": 1.0704,
+      "step": 118125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.87067198034698e-06,
+      "loss": 0.9915,
+      "step": 118130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.869933635568532e-06,
+      "loss": 1.202,
+      "step": 118135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.869195302953906e-06,
+      "loss": 1.4219,
+      "step": 118140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.868456982507325e-06,
+      "loss": 1.2804,
+      "step": 118145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.867718674233e-06,
+      "loss": 1.2222,
+      "step": 118150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.86698037813515e-06,
+      "loss": 1.2676,
+      "step": 118155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.866242094217985e-06,
+      "loss": 1.0529,
+      "step": 118160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.865503822485727e-06,
+      "loss": 1.3667,
+      "step": 118165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.864765562942592e-06,
+      "loss": 1.4459,
+      "step": 118170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.864027315592789e-06,
+      "loss": 1.3742,
+      "step": 118175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.863289080440545e-06,
+      "loss": 1.3064,
+      "step": 118180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.862550857490067e-06,
+      "loss": 1.3356,
+      "step": 118185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.861812646745572e-06,
+      "loss": 1.3299,
+      "step": 118190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.861074448211276e-06,
+      "loss": 1.2749,
+      "step": 118195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.8603362618914e-06,
+      "loss": 1.2827,
+      "step": 118200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.85959808779015e-06,
+      "loss": 1.2951,
+      "step": 118205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.85885992591175e-06,
+      "loss": 1.3902,
+      "step": 118210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.858121776260411e-06,
+      "loss": 1.9241,
+      "step": 118215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.857383638840345e-06,
+      "loss": 1.359,
+      "step": 118220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.856645513655773e-06,
+      "loss": 1.2437,
+      "step": 118225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.855907400710909e-06,
+      "loss": 1.1104,
+      "step": 118230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.855169300009969e-06,
+      "loss": 1.5166,
+      "step": 118235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.85443121155716e-06,
+      "loss": 1.1957,
+      "step": 118240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.853693135356705e-06,
+      "loss": 1.1116,
+      "step": 118245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.85295507141282e-06,
+      "loss": 1.0875,
+      "step": 118250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.85221701972971e-06,
+      "loss": 1.1831,
+      "step": 118255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.851478980311602e-06,
+      "loss": 1.2592,
+      "step": 118260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.850740953162701e-06,
+      "loss": 1.2682,
+      "step": 118265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.850002938287225e-06,
+      "loss": 1.1442,
+      "step": 118270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.849264935689392e-06,
+      "loss": 1.2938,
+      "step": 118275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.848526945373412e-06,
+      "loss": 1.4699,
+      "step": 118280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.847788967343496e-06,
+      "loss": 1.0428,
+      "step": 118285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.847051001603868e-06,
+      "loss": 1.2039,
+      "step": 118290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.846313048158736e-06,
+      "loss": 1.5371,
+      "step": 118295
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.845575107012313e-06,
+      "loss": 1.1285,
+      "step": 118300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.844837178168814e-06,
+      "loss": 1.0294,
+      "step": 118305
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.844099261632457e-06,
+      "loss": 1.0926,
+      "step": 118310
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.843361357407451e-06,
+      "loss": 1.2352,
+      "step": 118315
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.84262346549801e-06,
+      "loss": 1.1383,
+      "step": 118320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.84188558590835e-06,
+      "loss": 1.4263,
+      "step": 118325
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.841147718642689e-06,
+      "loss": 1.3253,
+      "step": 118330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.840409863705229e-06,
+      "loss": 1.373,
+      "step": 118335
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.839672021100193e-06,
+      "loss": 1.4444,
+      "step": 118340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.838934190831793e-06,
+      "loss": 1.4634,
+      "step": 118345
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.838196372904237e-06,
+      "loss": 1.1669,
+      "step": 118350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.837458567321747e-06,
+      "loss": 1.2181,
+      "step": 118355
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.83672077408853e-06,
+      "loss": 1.1108,
+      "step": 118360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.8359829932088e-06,
+      "loss": 1.0651,
+      "step": 118365
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.835245224686774e-06,
+      "loss": 1.2833,
+      "step": 118370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.83450746852666e-06,
+      "loss": 1.41,
+      "step": 118375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.833769724732673e-06,
+      "loss": 1.3794,
+      "step": 118380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.833031993309024e-06,
+      "loss": 1.1213,
+      "step": 118385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.832294274259932e-06,
+      "loss": 1.302,
+      "step": 118390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.831556567589603e-06,
+      "loss": 1.0986,
+      "step": 118395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.83081887330225e-06,
+      "loss": 1.3775,
+      "step": 118400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.830081191402093e-06,
+      "loss": 1.0792,
+      "step": 118405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.829343521893337e-06,
+      "loss": 1.0357,
+      "step": 118410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.828605864780195e-06,
+      "loss": 1.2123,
+      "step": 118415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.827868220066884e-06,
+      "loss": 1.5212,
+      "step": 118420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.827130587757616e-06,
+      "loss": 1.6032,
+      "step": 118425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.826392967856597e-06,
+      "loss": 1.3086,
+      "step": 118430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.825655360368044e-06,
+      "loss": 1.0566,
+      "step": 118435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.824917765296169e-06,
+      "loss": 1.4021,
+      "step": 118440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.824180182645185e-06,
+      "loss": 1.402,
+      "step": 118445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.823442612419299e-06,
+      "loss": 1.3719,
+      "step": 118450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.822705054622728e-06,
+      "loss": 0.9211,
+      "step": 118455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.821967509259685e-06,
+      "loss": 1.2405,
+      "step": 118460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.821229976334375e-06,
+      "loss": 1.3224,
+      "step": 118465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.820492455851015e-06,
+      "loss": 1.446,
+      "step": 118470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.819754947813815e-06,
+      "loss": 1.2926,
+      "step": 118475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.819017452226985e-06,
+      "loss": 1.1692,
+      "step": 118480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.818279969094742e-06,
+      "loss": 1.2844,
+      "step": 118485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.817542498421294e-06,
+      "loss": 1.2306,
+      "step": 118490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.816805040210848e-06,
+      "loss": 1.3882,
+      "step": 118495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.816067594467622e-06,
+      "loss": 1.2034,
+      "step": 118500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.815330161195826e-06,
+      "loss": 0.9956,
+      "step": 118505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.814592740399667e-06,
+      "loss": 1.2239,
+      "step": 118510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.813855332083362e-06,
+      "loss": 1.2296,
+      "step": 118515
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.813117936251119e-06,
+      "loss": 1.0453,
+      "step": 118520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.812380552907147e-06,
+      "loss": 1.0379,
+      "step": 118525
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.811643182055655e-06,
+      "loss": 1.0714,
+      "step": 118530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.810905823700863e-06,
+      "loss": 1.4644,
+      "step": 118535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.810168477846975e-06,
+      "loss": 1.3196,
+      "step": 118540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.809431144498202e-06,
+      "loss": 1.3355,
+      "step": 118545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.808693823658756e-06,
+      "loss": 1.3466,
+      "step": 118550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.807956515332848e-06,
+      "loss": 1.0674,
+      "step": 118555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.807219219524681e-06,
+      "loss": 1.2877,
+      "step": 118560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.806481936238478e-06,
+      "loss": 1.3943,
+      "step": 118565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.805744665478443e-06,
+      "loss": 1.4779,
+      "step": 118570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.805007407248782e-06,
+      "loss": 1.1791,
+      "step": 118575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.804270161553714e-06,
+      "loss": 1.025,
+      "step": 118580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.803532928397441e-06,
+      "loss": 1.1576,
+      "step": 118585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.80279570778418e-06,
+      "loss": 1.03,
+      "step": 118590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.80205849971813e-06,
+      "loss": 1.0874,
+      "step": 118595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.801321304203514e-06,
+      "loss": 1.4483,
+      "step": 118600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.800584121244537e-06,
+      "loss": 1.1415,
+      "step": 118605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.799846950845402e-06,
+      "loss": 1.1935,
+      "step": 118610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.79910979301033e-06,
+      "loss": 1.2146,
+      "step": 118615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.798372647743523e-06,
+      "loss": 0.9294,
+      "step": 118620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.79763551504919e-06,
+      "loss": 1.482,
+      "step": 118625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.796898394931545e-06,
+      "loss": 1.165,
+      "step": 118630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.796161287394796e-06,
+      "loss": 1.1854,
+      "step": 118635
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.795424192443149e-06,
+      "loss": 1.4607,
+      "step": 118640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.794687110080816e-06,
+      "loss": 1.163,
+      "step": 118645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.793950040312009e-06,
+      "loss": 1.1159,
+      "step": 118650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.793212983140926e-06,
+      "loss": 1.3817,
+      "step": 118655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.792475938571791e-06,
+      "loss": 1.3419,
+      "step": 118660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.791738906608804e-06,
+      "loss": 1.3479,
+      "step": 118665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.791001887256177e-06,
+      "loss": 1.3723,
+      "step": 118670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.790264880518115e-06,
+      "loss": 1.4898,
+      "step": 118675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.789527886398828e-06,
+      "loss": 1.2355,
+      "step": 118680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.788790904902531e-06,
+      "loss": 1.5998,
+      "step": 118685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.78805393603342e-06,
+      "loss": 1.3457,
+      "step": 118690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.787316979795715e-06,
+      "loss": 1.2827,
+      "step": 118695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.78658003619362e-06,
+      "loss": 1.1688,
+      "step": 118700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.785843105231342e-06,
+      "loss": 1.2584,
+      "step": 118705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.785106186913093e-06,
+      "loss": 1.2845,
+      "step": 118710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.784369281243078e-06,
+      "loss": 1.0023,
+      "step": 118715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.783632388225504e-06,
+      "loss": 1.2212,
+      "step": 118720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.782895507864581e-06,
+      "loss": 0.9971,
+      "step": 118725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.782158640164521e-06,
+      "loss": 1.1182,
+      "step": 118730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.781421785129529e-06,
+      "loss": 1.1601,
+      "step": 118735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.780684942763806e-06,
+      "loss": 1.0295,
+      "step": 118740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.779948113071571e-06,
+      "loss": 1.3324,
+      "step": 118745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.779211296057023e-06,
+      "loss": 1.2676,
+      "step": 118750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.778474491724372e-06,
+      "loss": 1.2491,
+      "step": 118755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.777737700077828e-06,
+      "loss": 1.4576,
+      "step": 118760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.7770009211216e-06,
+      "loss": 1.2183,
+      "step": 118765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.776264154859886e-06,
+      "loss": 1.1442,
+      "step": 118770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.775527401296905e-06,
+      "loss": 1.5733,
+      "step": 118775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 1.2628,
+      "step": 118780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.77405393228395e-06,
+      "loss": 1.5733,
+      "step": 118785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.773317216842395e-06,
+      "loss": 1.1939,
+      "step": 118790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.772580514116396e-06,
+      "loss": 0.9982,
+      "step": 118795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.771843824110157e-06,
+      "loss": 1.2852,
+      "step": 118800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.771107146827893e-06,
+      "loss": 1.2316,
+      "step": 118805
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.770370482273803e-06,
+      "loss": 1.1986,
+      "step": 118810
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.769633830452099e-06,
+      "loss": 1.0317,
+      "step": 118815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.76889719136698e-06,
+      "loss": 1.5525,
+      "step": 118820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.768160565022665e-06,
+      "loss": 1.2394,
+      "step": 118825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.76742395142335e-06,
+      "loss": 1.4423,
+      "step": 118830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.766687350573245e-06,
+      "loss": 1.2252,
+      "step": 118835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.765950762476557e-06,
+      "loss": 1.1363,
+      "step": 118840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.765214187137493e-06,
+      "loss": 1.2626,
+      "step": 118845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.764477624560254e-06,
+      "loss": 1.6023,
+      "step": 118850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.763741074749054e-06,
+      "loss": 1.1673,
+      "step": 118855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.763004537708095e-06,
+      "loss": 1.2744,
+      "step": 118860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.76226801344158e-06,
+      "loss": 1.6782,
+      "step": 118865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.76153150195372e-06,
+      "loss": 1.4155,
+      "step": 118870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.760795003248722e-06,
+      "loss": 1.0531,
+      "step": 118875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.760058517330787e-06,
+      "loss": 1.5975,
+      "step": 118880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.759322044204121e-06,
+      "loss": 1.3817,
+      "step": 118885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.758585583872932e-06,
+      "loss": 1.3478,
+      "step": 118890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.757849136341428e-06,
+      "loss": 1.5237,
+      "step": 118895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.757112701613804e-06,
+      "loss": 1.0957,
+      "step": 118900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.756376279694279e-06,
+      "loss": 1.1745,
+      "step": 118905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.75563987058705e-06,
+      "loss": 1.1907,
+      "step": 118910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.754903474296323e-06,
+      "loss": 1.283,
+      "step": 118915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.75416709082631e-06,
+      "loss": 1.2414,
+      "step": 118920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.753430720181208e-06,
+      "loss": 1.0968,
+      "step": 118925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.752694362365221e-06,
+      "loss": 1.6466,
+      "step": 118930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.751958017382563e-06,
+      "loss": 1.2613,
+      "step": 118935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.751221685237434e-06,
+      "loss": 1.7032,
+      "step": 118940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.750485365934036e-06,
+      "loss": 0.9792,
+      "step": 118945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.749749059476577e-06,
+      "loss": 1.2353,
+      "step": 118950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.749012765869262e-06,
+      "loss": 1.2145,
+      "step": 118955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.748276485116295e-06,
+      "loss": 1.1578,
+      "step": 118960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.747540217221878e-06,
+      "loss": 1.2907,
+      "step": 118965
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.746803962190219e-06,
+      "loss": 1.5373,
+      "step": 118970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.746067720025525e-06,
+      "loss": 1.6101,
+      "step": 118975
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.74533149073199e-06,
+      "loss": 1.1513,
+      "step": 118980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.744595274313829e-06,
+      "loss": 1.3782,
+      "step": 118985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.743859070775244e-06,
+      "loss": 1.4229,
+      "step": 118990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.74312288012043e-06,
+      "loss": 1.3053,
+      "step": 118995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.742386702353607e-06,
+      "loss": 1.3576,
+      "step": 119000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.741650537478964e-06,
+      "loss": 1.2313,
+      "step": 119005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.740914385500712e-06,
+      "loss": 1.2638,
+      "step": 119010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.740178246423057e-06,
+      "loss": 1.2177,
+      "step": 119015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.739442120250199e-06,
+      "loss": 1.1811,
+      "step": 119020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.73870600698634e-06,
+      "loss": 1.1787,
+      "step": 119025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.737969906635686e-06,
+      "loss": 1.4804,
+      "step": 119030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.737233819202443e-06,
+      "loss": 1.266,
+      "step": 119035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.73649774469081e-06,
+      "loss": 1.1238,
+      "step": 119040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.73576168310499e-06,
+      "loss": 1.248,
+      "step": 119045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.735025634449194e-06,
+      "loss": 0.9234,
+      "step": 119050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.734289598727617e-06,
+      "loss": 1.0932,
+      "step": 119055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.733553575944462e-06,
+      "loss": 1.1912,
+      "step": 119060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.732817566103938e-06,
+      "loss": 1.6303,
+      "step": 119065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.732081569210248e-06,
+      "loss": 1.3017,
+      "step": 119070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.731345585267588e-06,
+      "loss": 1.0794,
+      "step": 119075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.730609614280165e-06,
+      "loss": 1.1833,
+      "step": 119080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.729873656252184e-06,
+      "loss": 1.0343,
+      "step": 119085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.729137711187843e-06,
+      "loss": 1.3549,
+      "step": 119090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.72840177909135e-06,
+      "loss": 1.0716,
+      "step": 119095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.727665859966903e-06,
+      "loss": 1.4754,
+      "step": 119100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.726929953818708e-06,
+      "loss": 1.1906,
+      "step": 119105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.726194060650962e-06,
+      "loss": 1.1777,
+      "step": 119110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.725458180467873e-06,
+      "loss": 1.0869,
+      "step": 119115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.72472231327364e-06,
+      "loss": 1.2406,
+      "step": 119120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.723986459072466e-06,
+      "loss": 1.2301,
+      "step": 119125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.723250617868557e-06,
+      "loss": 1.0442,
+      "step": 119130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.722514789666109e-06,
+      "loss": 1.2651,
+      "step": 119135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.721778974469325e-06,
+      "loss": 1.2838,
+      "step": 119140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.721043172282409e-06,
+      "loss": 1.1854,
+      "step": 119145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.720307383109565e-06,
+      "loss": 1.3978,
+      "step": 119150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.71957160695499e-06,
+      "loss": 1.4358,
+      "step": 119155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.718835843822888e-06,
+      "loss": 1.1934,
+      "step": 119160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.718100093717462e-06,
+      "loss": 1.2929,
+      "step": 119165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.717364356642909e-06,
+      "loss": 1.3139,
+      "step": 119170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.716628632603433e-06,
+      "loss": 1.1515,
+      "step": 119175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.715892921603236e-06,
+      "loss": 1.0379,
+      "step": 119180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.715157223646521e-06,
+      "loss": 1.2291,
+      "step": 119185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.714421538737484e-06,
+      "loss": 1.4387,
+      "step": 119190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.713685866880331e-06,
+      "loss": 1.0605,
+      "step": 119195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.712950208079265e-06,
+      "loss": 1.3346,
+      "step": 119200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.712214562338476e-06,
+      "loss": 1.0374,
+      "step": 119205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.711478929662179e-06,
+      "loss": 1.3134,
+      "step": 119210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.710743310054565e-06,
+      "loss": 1.0456,
+      "step": 119215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.710007703519836e-06,
+      "loss": 1.1315,
+      "step": 119220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.7092721100622e-06,
+      "loss": 1.1194,
+      "step": 119225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.70853652968585e-06,
+      "loss": 1.1173,
+      "step": 119230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.707800962394985e-06,
+      "loss": 1.1279,
+      "step": 119235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.707065408193814e-06,
+      "loss": 1.1452,
+      "step": 119240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.706329867086534e-06,
+      "loss": 1.2413,
+      "step": 119245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.705594339077341e-06,
+      "loss": 1.0573,
+      "step": 119250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.704858824170437e-06,
+      "loss": 1.0961,
+      "step": 119255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.704123322370029e-06,
+      "loss": 1.1832,
+      "step": 119260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.703387833680307e-06,
+      "loss": 1.301,
+      "step": 119265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.702652358105476e-06,
+      "loss": 1.2376,
+      "step": 119270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.701916895649735e-06,
+      "loss": 1.1724,
+      "step": 119275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.701181446317289e-06,
+      "loss": 1.2348,
+      "step": 119280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.700446010112329e-06,
+      "loss": 1.1467,
+      "step": 119285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.699710587039061e-06,
+      "loss": 1.2368,
+      "step": 119290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.698975177101684e-06,
+      "loss": 1.2086,
+      "step": 119295
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.69823978030439e-06,
+      "loss": 1.0272,
+      "step": 119300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.697504396651394e-06,
+      "loss": 1.1689,
+      "step": 119305
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.696769026146881e-06,
+      "loss": 1.1956,
+      "step": 119310
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.696033668795054e-06,
+      "loss": 1.2212,
+      "step": 119315
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.69529832460012e-06,
+      "loss": 1.4949,
+      "step": 119320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.694562993566268e-06,
+      "loss": 1.1593,
+      "step": 119325
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.693827675697704e-06,
+      "loss": 1.2783,
+      "step": 119330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.693092370998619e-06,
+      "loss": 1.4343,
+      "step": 119335
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.692357079473222e-06,
+      "loss": 0.9981,
+      "step": 119340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.691621801125706e-06,
+      "loss": 1.2268,
+      "step": 119345
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.69088653596027e-06,
+      "loss": 1.1135,
+      "step": 119350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.690151283981117e-06,
+      "loss": 1.1556,
+      "step": 119355
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.68941604519244e-06,
+      "loss": 1.1426,
+      "step": 119360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.688680819598438e-06,
+      "loss": 1.124,
+      "step": 119365
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.687945607203315e-06,
+      "loss": 1.1726,
+      "step": 119370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.687210408011267e-06,
+      "loss": 1.08,
+      "step": 119375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.686475222026489e-06,
+      "loss": 1.1149,
+      "step": 119380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.685740049253181e-06,
+      "loss": 1.2626,
+      "step": 119385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.685004889695547e-06,
+      "loss": 1.3998,
+      "step": 119390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.684269743357777e-06,
+      "loss": 1.1545,
+      "step": 119395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.68353461024407e-06,
+      "loss": 1.1999,
+      "step": 119400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.68279949035863e-06,
+      "loss": 1.2121,
+      "step": 119405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.682064383705651e-06,
+      "loss": 1.1553,
+      "step": 119410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.681329290289328e-06,
+      "loss": 1.0291,
+      "step": 119415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.680594210113864e-06,
+      "loss": 1.5994,
+      "step": 119420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.679859143183454e-06,
+      "loss": 1.5439,
+      "step": 119425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.679124089502293e-06,
+      "loss": 1.3927,
+      "step": 119430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.678389049074588e-06,
+      "loss": 1.198,
+      "step": 119435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.677654021904528e-06,
+      "loss": 1.727,
+      "step": 119440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.676919007996309e-06,
+      "loss": 1.3596,
+      "step": 119445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.676184007354137e-06,
+      "loss": 1.1853,
+      "step": 119450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.675449019982204e-06,
+      "loss": 1.1244,
+      "step": 119455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.674714045884703e-06,
+      "loss": 1.7686,
+      "step": 119460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.673979085065837e-06,
+      "loss": 1.1282,
+      "step": 119465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.673244137529806e-06,
+      "loss": 1.2261,
+      "step": 119470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.672509203280798e-06,
+      "loss": 1.3983,
+      "step": 119475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.671774282323012e-06,
+      "loss": 1.2299,
+      "step": 119480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.671039374660652e-06,
+      "loss": 1.3461,
+      "step": 119485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.67030448029791e-06,
+      "loss": 1.491,
+      "step": 119490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.669569599238977e-06,
+      "loss": 1.3392,
+      "step": 119495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.66883473148806e-06,
+      "loss": 1.0897,
+      "step": 119500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.668099877049351e-06,
+      "loss": 1.2662,
+      "step": 119505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.667365035927041e-06,
+      "loss": 1.3321,
+      "step": 119510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.666630208125334e-06,
+      "loss": 1.8707,
+      "step": 119515
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.665895393648423e-06,
+      "loss": 1.2227,
+      "step": 119520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.665160592500505e-06,
+      "loss": 1.447,
+      "step": 119525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.664425804685778e-06,
+      "loss": 1.6728,
+      "step": 119530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.663691030208433e-06,
+      "loss": 1.1699,
+      "step": 119535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.662956269072673e-06,
+      "loss": 1.1692,
+      "step": 119540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.662221521282682e-06,
+      "loss": 1.2275,
+      "step": 119545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.66148678684267e-06,
+      "loss": 1.1526,
+      "step": 119550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.660752065756824e-06,
+      "loss": 1.0712,
+      "step": 119555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.660017358029342e-06,
+      "loss": 1.1898,
+      "step": 119560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.65928266366442e-06,
+      "loss": 1.4551,
+      "step": 119565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.658547982666253e-06,
+      "loss": 0.9954,
+      "step": 119570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.657813315039035e-06,
+      "loss": 1.0563,
+      "step": 119575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.657078660786963e-06,
+      "loss": 1.3933,
+      "step": 119580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.656344019914234e-06,
+      "loss": 1.2758,
+      "step": 119585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.655609392425038e-06,
+      "loss": 1.4548,
+      "step": 119590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.654874778323575e-06,
+      "loss": 1.0158,
+      "step": 119595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.654140177614042e-06,
+      "loss": 1.2714,
+      "step": 119600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.653405590300623e-06,
+      "loss": 1.3025,
+      "step": 119605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.652671016387526e-06,
+      "loss": 1.2886,
+      "step": 119610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.651936455878937e-06,
+      "loss": 1.2145,
+      "step": 119615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.651201908779057e-06,
+      "loss": 1.3959,
+      "step": 119620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.650467375092073e-06,
+      "loss": 1.0936,
+      "step": 119625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.649732854822186e-06,
+      "loss": 1.2253,
+      "step": 119630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.648998347973592e-06,
+      "loss": 1.1464,
+      "step": 119635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.648263854550473e-06,
+      "loss": 0.9853,
+      "step": 119640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.64752937455704e-06,
+      "loss": 1.1068,
+      "step": 119645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.646794907997477e-06,
+      "loss": 1.1979,
+      "step": 119650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.64606045487598e-06,
+      "loss": 1.0313,
+      "step": 119655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.645326015196747e-06,
+      "loss": 1.3954,
+      "step": 119660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.644591588963968e-06,
+      "loss": 1.0966,
+      "step": 119665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.643857176181834e-06,
+      "loss": 1.5225,
+      "step": 119670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.643122776854546e-06,
+      "loss": 1.1684,
+      "step": 119675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.642388390986296e-06,
+      "loss": 1.0413,
+      "step": 119680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.641654018581275e-06,
+      "loss": 1.2125,
+      "step": 119685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.640919659643676e-06,
+      "loss": 1.0598,
+      "step": 119690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.640185314177697e-06,
+      "loss": 1.1827,
+      "step": 119695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.639450982187529e-06,
+      "loss": 1.1295,
+      "step": 119700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.638716663677362e-06,
+      "loss": 0.8737,
+      "step": 119705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.637982358651396e-06,
+      "loss": 1.1595,
+      "step": 119710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.637248067113822e-06,
+      "loss": 1.1251,
+      "step": 119715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.63651378906883e-06,
+      "loss": 1.257,
+      "step": 119720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.635779524520617e-06,
+      "loss": 1.1798,
+      "step": 119725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.635045273473374e-06,
+      "loss": 1.2802,
+      "step": 119730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.634311035931293e-06,
+      "loss": 1.3566,
+      "step": 119735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.633576811898571e-06,
+      "loss": 1.2472,
+      "step": 119740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.632842601379397e-06,
+      "loss": 1.1674,
+      "step": 119745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.632108404377964e-06,
+      "loss": 1.3465,
+      "step": 119750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.63137422089847e-06,
+      "loss": 1.1802,
+      "step": 119755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.630640050945099e-06,
+      "loss": 1.4414,
+      "step": 119760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.629905894522048e-06,
+      "loss": 1.1492,
+      "step": 119765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.629171751633507e-06,
+      "loss": 0.8163,
+      "step": 119770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.6284376222836756e-06,
+      "loss": 1.2337,
+      "step": 119775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.627703506476738e-06,
+      "loss": 0.8425,
+      "step": 119780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.626969404216887e-06,
+      "loss": 1.2206,
+      "step": 119785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.62623531550832e-06,
+      "loss": 1.2908,
+      "step": 119790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.625501240355229e-06,
+      "loss": 1.5201,
+      "step": 119795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.624767178761797e-06,
+      "loss": 0.9919,
+      "step": 119800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.624033130732224e-06,
+      "loss": 1.3325,
+      "step": 119805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.6232990962707046e-06,
+      "loss": 1.1036,
+      "step": 119810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.622565075381418e-06,
+      "loss": 1.127,
+      "step": 119815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.621831068068568e-06,
+      "loss": 1.2077,
+      "step": 119820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.621097074336341e-06,
+      "loss": 1.4655,
+      "step": 119825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.620363094188932e-06,
+      "loss": 1.2108,
+      "step": 119830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.619629127630523e-06,
+      "loss": 1.1304,
+      "step": 119835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.618895174665316e-06,
+      "loss": 1.1778,
+      "step": 119840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.618161235297501e-06,
+      "loss": 1.1316,
+      "step": 119845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.617427309531259e-06,
+      "loss": 1.2745,
+      "step": 119850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.6166933973707945e-06,
+      "loss": 1.5241,
+      "step": 119855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.615959498820291e-06,
+      "loss": 1.1648,
+      "step": 119860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.615225613883939e-06,
+      "loss": 1.3877,
+      "step": 119865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.614491742565936e-06,
+      "loss": 1.0628,
+      "step": 119870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.613757884870465e-06,
+      "loss": 1.3569,
+      "step": 119875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.613024040801719e-06,
+      "loss": 1.122,
+      "step": 119880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.612290210363891e-06,
+      "loss": 1.1574,
+      "step": 119885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.611556393561173e-06,
+      "loss": 1.3396,
+      "step": 119890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.610822590397746e-06,
+      "loss": 1.2027,
+      "step": 119895
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.610088800877812e-06,
+      "loss": 1.6561,
+      "step": 119900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.609355025005557e-06,
+      "loss": 1.2628,
+      "step": 119905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.608621262785169e-06,
+      "loss": 1.062,
+      "step": 119910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.6078875142208374e-06,
+      "loss": 1.3073,
+      "step": 119915
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.607153779316757e-06,
+      "loss": 1.3011,
+      "step": 119920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.6064200580771175e-06,
+      "loss": 1.0582,
+      "step": 119925
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.605686350506103e-06,
+      "loss": 1.2636,
+      "step": 119930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.604952656607909e-06,
+      "loss": 1.3859,
+      "step": 119935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.604218976386726e-06,
+      "loss": 1.273,
+      "step": 119940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.603485309846734e-06,
+      "loss": 1.1543,
+      "step": 119945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.602751656992137e-06,
+      "loss": 1.7812,
+      "step": 119950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.602018017827116e-06,
+      "loss": 1.2399,
+      "step": 119955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.601284392355858e-06,
+      "loss": 1.0936,
+      "step": 119960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.600550780582561e-06,
+      "loss": 1.3658,
+      "step": 119965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5998171825114075e-06,
+      "loss": 1.1667,
+      "step": 119970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5990835981465916e-06,
+      "loss": 0.9252,
+      "step": 119975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.598350027492293e-06,
+      "loss": 1.2025,
+      "step": 119980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.597616470552713e-06,
+      "loss": 1.0529,
+      "step": 119985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.596882927332034e-06,
+      "loss": 1.3743,
+      "step": 119990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.596149397834445e-06,
+      "loss": 1.1696,
+      "step": 119995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.595415882064137e-06,
+      "loss": 1.1102,
+      "step": 120000
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 1.2228082418441772,
+      "eval_runtime": 6476.2318,
+      "eval_samples_per_second": 3.568,
+      "eval_steps_per_second": 1.784,
+      "step": 120000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.594682380025298e-06,
+      "loss": 1.1652,
+      "step": 120005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.593948891722114e-06,
+      "loss": 1.1971,
+      "step": 120010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.593215417158777e-06,
+      "loss": 1.3364,
+      "step": 120015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.592481956339477e-06,
+      "loss": 1.2565,
+      "step": 120020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.591748509268396e-06,
+      "loss": 1.2136,
+      "step": 120025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5910150759497285e-06,
+      "loss": 1.1805,
+      "step": 120030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.590281656387661e-06,
+      "loss": 1.0492,
+      "step": 120035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5895482505863774e-06,
+      "loss": 1.4761,
+      "step": 120040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5888148585500755e-06,
+      "loss": 1.395,
+      "step": 120045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5880814802829335e-06,
+      "loss": 1.4677,
+      "step": 120050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.587348115789148e-06,
+      "loss": 1.3248,
+      "step": 120055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.586614765072897e-06,
+      "loss": 1.0842,
+      "step": 120060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.585881428138378e-06,
+      "loss": 1.2065,
+      "step": 120065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.585148104989771e-06,
+      "loss": 1.4283,
+      "step": 120070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5844147956312676e-06,
+      "loss": 1.3048,
+      "step": 120075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.583681500067058e-06,
+      "loss": 1.1353,
+      "step": 120080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.582948218301324e-06,
+      "loss": 1.376,
+      "step": 120085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.582214950338256e-06,
+      "loss": 1.4013,
+      "step": 120090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5814816961820425e-06,
+      "loss": 1.0081,
+      "step": 120095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.580748455836868e-06,
+      "loss": 1.2976,
+      "step": 120100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.580015229306921e-06,
+      "loss": 1.0091,
+      "step": 120105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.579282016596388e-06,
+      "loss": 1.4956,
+      "step": 120110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.57854881770946e-06,
+      "loss": 1.1943,
+      "step": 120115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5778156326503185e-06,
+      "loss": 1.1543,
+      "step": 120120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.577082461423151e-06,
+      "loss": 1.1832,
+      "step": 120125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5763493040321465e-06,
+      "loss": 1.1756,
+      "step": 120130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5756161604814935e-06,
+      "loss": 1.0222,
+      "step": 120135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.574883030775373e-06,
+      "loss": 0.9667,
+      "step": 120140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.574149914917975e-06,
+      "loss": 1.1526,
+      "step": 120145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.57341681291349e-06,
+      "loss": 1.2011,
+      "step": 120150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.572683724766095e-06,
+      "loss": 1.1204,
+      "step": 120155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.571950650479983e-06,
+      "loss": 1.0101,
+      "step": 120160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.571217590059339e-06,
+      "loss": 1.3025,
+      "step": 120165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.570484543508348e-06,
+      "loss": 1.2851,
+      "step": 120170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.569751510831201e-06,
+      "loss": 0.929,
+      "step": 120175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.569018492032077e-06,
+      "loss": 1.5099,
+      "step": 120180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.568285487115161e-06,
+      "loss": 1.2366,
+      "step": 120185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.567552496084647e-06,
+      "loss": 1.3716,
+      "step": 120190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.566819518944719e-06,
+      "loss": 1.1205,
+      "step": 120195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.566086555699558e-06,
+      "loss": 1.6169,
+      "step": 120200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.565353606353348e-06,
+      "loss": 1.1245,
+      "step": 120205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5646206709102845e-06,
+      "loss": 1.133,
+      "step": 120210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5638877493745445e-06,
+      "loss": 1.1475,
+      "step": 120215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5631548417503135e-06,
+      "loss": 1.2081,
+      "step": 120220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.56242194804178e-06,
+      "loss": 1.2481,
+      "step": 120225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.561689068253132e-06,
+      "loss": 1.0833,
+      "step": 120230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.560956202388545e-06,
+      "loss": 1.1905,
+      "step": 120235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.560223350452213e-06,
+      "loss": 1.4905,
+      "step": 120240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.55949051244832e-06,
+      "loss": 1.0107,
+      "step": 120245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.558757688381042e-06,
+      "loss": 1.0358,
+      "step": 120250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5580248782545784e-06,
+      "loss": 1.6471,
+      "step": 120255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.557292082073102e-06,
+      "loss": 1.4412,
+      "step": 120260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.556559299840805e-06,
+      "loss": 1.2945,
+      "step": 120265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.555826531561865e-06,
+      "loss": 1.3724,
+      "step": 120270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.555093777240471e-06,
+      "loss": 1.1632,
+      "step": 120275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.55436103688081e-06,
+      "loss": 1.5656,
+      "step": 120280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5536283104870545e-06,
+      "loss": 1.345,
+      "step": 120285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.552895598063404e-06,
+      "loss": 1.1754,
+      "step": 120290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.552162899614035e-06,
+      "loss": 1.54,
+      "step": 120295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.551430215143129e-06,
+      "loss": 1.1675,
+      "step": 120300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.550697544654879e-06,
+      "loss": 1.36,
+      "step": 120305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.549964888153461e-06,
+      "loss": 1.1729,
+      "step": 120310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.549232245643058e-06,
+      "loss": 1.0958,
+      "step": 120315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.548499617127859e-06,
+      "loss": 1.2112,
+      "step": 120320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5477670026120486e-06,
+      "loss": 1.1882,
+      "step": 120325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.547034402099804e-06,
+      "loss": 1.5054,
+      "step": 120330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.546301815595314e-06,
+      "loss": 1.0913,
+      "step": 120335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.545569243102762e-06,
+      "loss": 1.2031,
+      "step": 120340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.544836684626327e-06,
+      "loss": 1.4301,
+      "step": 120345
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.544104140170194e-06,
+      "loss": 1.2999,
+      "step": 120350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.543371609738549e-06,
+      "loss": 1.1601,
+      "step": 120355
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5426390933355765e-06,
+      "loss": 1.198,
+      "step": 120360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.541906590965451e-06,
+      "loss": 1.1538,
+      "step": 120365
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.541174102632364e-06,
+      "loss": 1.1115,
+      "step": 120370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.540441628340495e-06,
+      "loss": 1.1922,
+      "step": 120375
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.539709168094023e-06,
+      "loss": 1.1022,
+      "step": 120380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5389767218971425e-06,
+      "loss": 1.0801,
+      "step": 120385
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.538244289754025e-06,
+      "loss": 1.318,
+      "step": 120390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.537511871668854e-06,
+      "loss": 1.0287,
+      "step": 120395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.536779467645819e-06,
+      "loss": 1.1599,
+      "step": 120400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.536047077689097e-06,
+      "loss": 1.183,
+      "step": 120405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5353147018028716e-06,
+      "loss": 1.39,
+      "step": 120410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.534582339991321e-06,
+      "loss": 1.1452,
+      "step": 120415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.533849992258637e-06,
+      "loss": 1.0638,
+      "step": 120420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.533117658608994e-06,
+      "loss": 1.3705,
+      "step": 120425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.532385339046573e-06,
+      "loss": 1.1057,
+      "step": 120430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5316530335755636e-06,
+      "loss": 1.2782,
+      "step": 120435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.530920742200141e-06,
+      "loss": 1.0967,
+      "step": 120440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5301884649244865e-06,
+      "loss": 1.0403,
+      "step": 120445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5294562017527865e-06,
+      "loss": 0.9883,
+      "step": 120450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.528723952689222e-06,
+      "loss": 1.0763,
+      "step": 120455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.527991717737969e-06,
+      "loss": 1.2485,
+      "step": 120460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.527259496903215e-06,
+      "loss": 1.1121,
+      "step": 120465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5265272901891394e-06,
+      "loss": 1.5148,
+      "step": 120470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.525795097599922e-06,
+      "loss": 1.2992,
+      "step": 120475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.525062919139747e-06,
+      "loss": 1.0189,
+      "step": 120480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.524330754812794e-06,
+      "loss": 1.2182,
+      "step": 120485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.523598604623246e-06,
+      "loss": 1.232,
+      "step": 120490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.522866468575277e-06,
+      "loss": 1.23,
+      "step": 120495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5221343466730755e-06,
+      "loss": 1.2726,
+      "step": 120500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.521402238920818e-06,
+      "loss": 1.2706,
+      "step": 120505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.520670145322687e-06,
+      "loss": 1.398,
+      "step": 120510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.519938065882866e-06,
+      "loss": 1.1851,
+      "step": 120515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.51920600060553e-06,
+      "loss": 1.1698,
+      "step": 120520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.518473949494861e-06,
+      "loss": 1.1629,
+      "step": 120525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5177419125550425e-06,
+      "loss": 1.1578,
+      "step": 120530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.517009889790255e-06,
+      "loss": 1.3977,
+      "step": 120535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.516277881204671e-06,
+      "loss": 1.32,
+      "step": 120540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.515545886802479e-06,
+      "loss": 1.3221,
+      "step": 120545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5148139065878585e-06,
+      "loss": 1.2231,
+      "step": 120550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.514081940564985e-06,
+      "loss": 1.2674,
+      "step": 120555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.513349988738039e-06,
+      "loss": 1.2935,
+      "step": 120560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.512618051111204e-06,
+      "loss": 1.2914,
+      "step": 120565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.511886127688659e-06,
+      "loss": 1.4298,
+      "step": 120570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.511154218474579e-06,
+      "loss": 1.1762,
+      "step": 120575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.510422323473149e-06,
+      "loss": 1.1499,
+      "step": 120580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.509690442688549e-06,
+      "loss": 1.2605,
+      "step": 120585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5089585761249485e-06,
+      "loss": 0.7707,
+      "step": 120590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.508226723786541e-06,
+      "loss": 1.4573,
+      "step": 120595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.507494885677497e-06,
+      "loss": 1.0924,
+      "step": 120600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5067630618019975e-06,
+      "loss": 1.0525,
+      "step": 120605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.506031252164225e-06,
+      "loss": 1.2643,
+      "step": 120610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5052994567683534e-06,
+      "loss": 1.1704,
+      "step": 120615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.504567675618562e-06,
+      "loss": 1.0678,
+      "step": 120620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.503835908719031e-06,
+      "loss": 1.3986,
+      "step": 120625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.503104156073944e-06,
+      "loss": 1.1131,
+      "step": 120630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.502372417687472e-06,
+      "loss": 1.4352,
+      "step": 120635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.5016406935637966e-06,
+      "loss": 1.2506,
+      "step": 120640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.500908983707099e-06,
+      "loss": 1.1732,
+      "step": 120645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.500177288121553e-06,
+      "loss": 1.0788,
+      "step": 120650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4994456068113375e-06,
+      "loss": 1.011,
+      "step": 120655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4987139397806355e-06,
+      "loss": 1.8019,
+      "step": 120660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.497982287033624e-06,
+      "loss": 1.2174,
+      "step": 120665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.497250648574475e-06,
+      "loss": 1.1167,
+      "step": 120670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4965190244073726e-06,
+      "loss": 1.3214,
+      "step": 120675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.495787414536495e-06,
+      "loss": 1.016,
+      "step": 120680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.495055818966012e-06,
+      "loss": 1.2394,
+      "step": 120685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.494324237700114e-06,
+      "loss": 1.0688,
+      "step": 120690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.49359267074297e-06,
+      "loss": 1.2322,
+      "step": 120695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4928611180987645e-06,
+      "loss": 1.1124,
+      "step": 120700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.492129579771663e-06,
+      "loss": 1.3393,
+      "step": 120705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4913980557658535e-06,
+      "loss": 1.1808,
+      "step": 120710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.490666546085511e-06,
+      "loss": 1.1916,
+      "step": 120715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.489935050734809e-06,
+      "loss": 1.2676,
+      "step": 120720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.489203569717933e-06,
+      "loss": 1.0883,
+      "step": 120725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.488472103039053e-06,
+      "loss": 1.038,
+      "step": 120730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.487740650702346e-06,
+      "loss": 1.2079,
+      "step": 120735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.487009212711995e-06,
+      "loss": 0.8344,
+      "step": 120740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.48627778907217e-06,
+      "loss": 1.2812,
+      "step": 120745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.48554637978705e-06,
+      "loss": 1.212,
+      "step": 120750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.484814984860815e-06,
+      "loss": 1.384,
+      "step": 120755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.484083604297641e-06,
+      "loss": 1.1244,
+      "step": 120760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.483352238101697e-06,
+      "loss": 1.1547,
+      "step": 120765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.48262088627717e-06,
+      "loss": 1.1999,
+      "step": 120770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.48188954882823e-06,
+      "loss": 1.0551,
+      "step": 120775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.481158225759058e-06,
+      "loss": 1.2957,
+      "step": 120780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.480426917073822e-06,
+      "loss": 1.2971,
+      "step": 120785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.479695622776706e-06,
+      "loss": 1.2228,
+      "step": 120790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.478964342871886e-06,
+      "loss": 1.235,
+      "step": 120795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4782330773635316e-06,
+      "loss": 1.2424,
+      "step": 120800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.477501826255824e-06,
+      "loss": 1.3523,
+      "step": 120805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.476770589552937e-06,
+      "loss": 1.1682,
+      "step": 120810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.476039367259047e-06,
+      "loss": 1.3398,
+      "step": 120815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.475308159378334e-06,
+      "loss": 1.1504,
+      "step": 120820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.474576965914966e-06,
+      "loss": 1.1355,
+      "step": 120825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.473845786873119e-06,
+      "loss": 0.9607,
+      "step": 120830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.473114622256974e-06,
+      "loss": 1.1831,
+      "step": 120835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.472383472070707e-06,
+      "loss": 1.0628,
+      "step": 120840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.471652336318487e-06,
+      "loss": 1.2318,
+      "step": 120845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.47092121500449e-06,
+      "loss": 1.1604,
+      "step": 120850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.470190108132895e-06,
+      "loss": 1.1614,
+      "step": 120855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.469459015707876e-06,
+      "loss": 1.3231,
+      "step": 120860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.468727937733604e-06,
+      "loss": 0.9835,
+      "step": 120865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.467996874214259e-06,
+      "loss": 1.4074,
+      "step": 120870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4672658251540165e-06,
+      "loss": 1.3531,
+      "step": 120875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4665347905570425e-06,
+      "loss": 1.2205,
+      "step": 120880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.465803770427521e-06,
+      "loss": 0.8706,
+      "step": 120885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.465072764769625e-06,
+      "loss": 1.107,
+      "step": 120890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.46434177358752e-06,
+      "loss": 1.2898,
+      "step": 120895
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.463610796885394e-06,
+      "loss": 1.068,
+      "step": 120900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4628798346674125e-06,
+      "loss": 1.5096,
+      "step": 120905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.46214888693775e-06,
+      "loss": 1.5395,
+      "step": 120910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.461417953700585e-06,
+      "loss": 1.252,
+      "step": 120915
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.460687034960088e-06,
+      "loss": 0.9389,
+      "step": 120920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.459956130720436e-06,
+      "loss": 1.2931,
+      "step": 120925
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4592252409857945e-06,
+      "loss": 1.01,
+      "step": 120930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.458494365760349e-06,
+      "loss": 1.3286,
+      "step": 120935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.457763505048266e-06,
+      "loss": 1.2973,
+      "step": 120940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.457032658853718e-06,
+      "loss": 1.1939,
+      "step": 120945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4563018271808865e-06,
+      "loss": 1.1876,
+      "step": 120950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.455571010033937e-06,
+      "loss": 1.2354,
+      "step": 120955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4548402074170435e-06,
+      "loss": 0.9798,
+      "step": 120960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.454109419334383e-06,
+      "loss": 1.1743,
+      "step": 120965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.45337864579013e-06,
+      "loss": 1.368,
+      "step": 120970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.45264788678845e-06,
+      "loss": 0.9759,
+      "step": 120975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.451917142333522e-06,
+      "loss": 1.1431,
+      "step": 120980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.45118641242952e-06,
+      "loss": 1.2386,
+      "step": 120985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4504556970806144e-06,
+      "loss": 1.2854,
+      "step": 120990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.449724996290975e-06,
+      "loss": 1.2437,
+      "step": 120995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.448994310064779e-06,
+      "loss": 1.0974,
+      "step": 121000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.448263638406199e-06,
+      "loss": 1.1847,
+      "step": 121005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.447532981319403e-06,
+      "loss": 1.4112,
+      "step": 121010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.446802338808568e-06,
+      "loss": 1.1478,
+      "step": 121015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4460717108778665e-06,
+      "loss": 1.2981,
+      "step": 121020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.445341097531464e-06,
+      "loss": 1.1736,
+      "step": 121025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.444610498773543e-06,
+      "loss": 1.0536,
+      "step": 121030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.443879914608269e-06,
+      "loss": 1.0912,
+      "step": 121035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.443149345039814e-06,
+      "loss": 1.1532,
+      "step": 121040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.442418790072354e-06,
+      "loss": 1.4677,
+      "step": 121045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.441688249710059e-06,
+      "loss": 1.0343,
+      "step": 121050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.440957723957096e-06,
+      "loss": 1.2516,
+      "step": 121055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.440227212817643e-06,
+      "loss": 1.2291,
+      "step": 121060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.439496716295872e-06,
+      "loss": 1.1784,
+      "step": 121065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.438766234395949e-06,
+      "loss": 1.1858,
+      "step": 121070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.438035767122047e-06,
+      "loss": 1.1543,
+      "step": 121075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.437305314478343e-06,
+      "loss": 1.0296,
+      "step": 121080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.436574876469001e-06,
+      "loss": 1.1901,
+      "step": 121085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.435844453098194e-06,
+      "loss": 1.4924,
+      "step": 121090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4351140443700965e-06,
+      "loss": 1.6254,
+      "step": 121095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.434383650288879e-06,
+      "loss": 1.0987,
+      "step": 121100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.433653270858707e-06,
+      "loss": 1.2859,
+      "step": 121105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4329229060837575e-06,
+      "loss": 0.8765,
+      "step": 121110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.432192555968198e-06,
+      "loss": 1.1404,
+      "step": 121115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.431462220516198e-06,
+      "loss": 1.2571,
+      "step": 121120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.430731899731934e-06,
+      "loss": 1.0041,
+      "step": 121125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.430001593619571e-06,
+      "loss": 1.0553,
+      "step": 121130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.429271302183285e-06,
+      "loss": 1.2376,
+      "step": 121135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.428541025427236e-06,
+      "loss": 1.4563,
+      "step": 121140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.427810763355605e-06,
+      "loss": 0.9896,
+      "step": 121145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.427080515972556e-06,
+      "loss": 1.2818,
+      "step": 121150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.42635028328226e-06,
+      "loss": 1.2711,
+      "step": 121155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.425620065288892e-06,
+      "loss": 1.2539,
+      "step": 121160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.424889861996616e-06,
+      "loss": 1.2791,
+      "step": 121165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.424159673409604e-06,
+      "loss": 1.2434,
+      "step": 121170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.423429499532024e-06,
+      "loss": 1.1483,
+      "step": 121175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.422699340368052e-06,
+      "loss": 1.2097,
+      "step": 121180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4219691959218476e-06,
+      "loss": 1.365,
+      "step": 121185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4212390661975885e-06,
+      "loss": 1.068,
+      "step": 121190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4205089511994435e-06,
+      "loss": 1.0925,
+      "step": 121195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.419778850931575e-06,
+      "loss": 1.1883,
+      "step": 121200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.419048765398159e-06,
+      "loss": 0.9194,
+      "step": 121205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.418318694603364e-06,
+      "loss": 1.5012,
+      "step": 121210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.417588638551359e-06,
+      "loss": 0.9353,
+      "step": 121215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.416858597246307e-06,
+      "loss": 1.0262,
+      "step": 121220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.416128570692384e-06,
+      "loss": 1.1902,
+      "step": 121225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.415398558893761e-06,
+      "loss": 0.9904,
+      "step": 121230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.414668561854592e-06,
+      "loss": 1.1895,
+      "step": 121235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.413938579579065e-06,
+      "loss": 1.1119,
+      "step": 121240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.413208612071338e-06,
+      "loss": 1.5476,
+      "step": 121245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.412478659335578e-06,
+      "loss": 1.3101,
+      "step": 121250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.411748721375961e-06,
+      "loss": 0.9713,
+      "step": 121255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4110187981966485e-06,
+      "loss": 1.1055,
+      "step": 121260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.410288889801809e-06,
+      "loss": 1.1083,
+      "step": 121265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.409558996195616e-06,
+      "loss": 1.4267,
+      "step": 121270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.408829117382234e-06,
+      "loss": 1.1416,
+      "step": 121275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.408099253365833e-06,
+      "loss": 1.1227,
+      "step": 121280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.407369404150575e-06,
+      "loss": 1.0536,
+      "step": 121285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4066395697406365e-06,
+      "loss": 1.3715,
+      "step": 121290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4059097501401786e-06,
+      "loss": 1.0436,
+      "step": 121295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.405179945353369e-06,
+      "loss": 1.2687,
+      "step": 121300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.40445015538438e-06,
+      "loss": 1.0004,
+      "step": 121305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.403720380237379e-06,
+      "loss": 1.2498,
+      "step": 121310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.402990619916527e-06,
+      "loss": 1.3242,
+      "step": 121315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.402260874425997e-06,
+      "loss": 1.5354,
+      "step": 121320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4015311437699565e-06,
+      "loss": 1.0583,
+      "step": 121325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.400801427952565e-06,
+      "loss": 1.4554,
+      "step": 121330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.400071726978002e-06,
+      "loss": 1.1616,
+      "step": 121335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.399342040850425e-06,
+      "loss": 1.2824,
+      "step": 121340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.398612369574001e-06,
+      "loss": 1.2481,
+      "step": 121345
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.397882713152905e-06,
+      "loss": 1.1395,
+      "step": 121350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.3971530715912966e-06,
+      "loss": 1.1839,
+      "step": 121355
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.396423444893343e-06,
+      "loss": 1.5819,
+      "step": 121360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.395693833063209e-06,
+      "loss": 0.998,
+      "step": 121365
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.394964236105069e-06,
+      "loss": 1.4262,
+      "step": 121370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.3942346540230825e-06,
+      "loss": 1.1181,
+      "step": 121375
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.393505086821415e-06,
+      "loss": 1.1937,
+      "step": 121380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.3927755345042406e-06,
+      "loss": 1.0523,
+      "step": 121385
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.3920459970757176e-06,
+      "loss": 0.934,
+      "step": 121390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.391316474540012e-06,
+      "loss": 1.1365,
+      "step": 121395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.390586966901294e-06,
+      "loss": 1.2038,
+      "step": 121400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.389857474163732e-06,
+      "loss": 1.2521,
+      "step": 121405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.3891279963314826e-06,
+      "loss": 1.6181,
+      "step": 121410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.388398533408718e-06,
+      "loss": 1.5793,
+      "step": 121415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.387669085399602e-06,
+      "loss": 1.1139,
+      "step": 121420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.3869396523083045e-06,
+      "loss": 1.3589,
+      "step": 121425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.386210234138982e-06,
+      "loss": 1.3739,
+      "step": 121430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.385480830895807e-06,
+      "loss": 1.6433,
+      "step": 121435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.384751442582945e-06,
+      "loss": 1.5938,
+      "step": 121440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.384022069204554e-06,
+      "loss": 0.8514,
+      "step": 121445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.3832927107648065e-06,
+      "loss": 1.1481,
+      "step": 121450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.382563367267866e-06,
+      "loss": 1.2509,
+      "step": 121455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.381834038717892e-06,
+      "loss": 1.024,
+      "step": 121460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.381104725119059e-06,
+      "loss": 1.1011,
+      "step": 121465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.380375426475525e-06,
+      "loss": 0.9199,
+      "step": 121470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.379646142791454e-06,
+      "loss": 1.4547,
+      "step": 121475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.378916874071016e-06,
+      "loss": 1.369,
+      "step": 121480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.378187620318373e-06,
+      "loss": 1.0591,
+      "step": 121485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.377458381537685e-06,
+      "loss": 1.4036,
+      "step": 121490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.376729157733122e-06,
+      "loss": 1.1111,
+      "step": 121495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.37599994890885e-06,
+      "loss": 1.2734,
+      "step": 121500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.375270755069026e-06,
+      "loss": 1.1593,
+      "step": 121505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.374541576217817e-06,
+      "loss": 1.1074,
+      "step": 121510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.3738124123593895e-06,
+      "loss": 0.9945,
+      "step": 121515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.373083263497907e-06,
+      "loss": 1.2479,
+      "step": 121520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.372354129637527e-06,
+      "loss": 1.1176,
+      "step": 121525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.371625010782422e-06,
+      "loss": 1.1746,
+      "step": 121530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.3708959069367526e-06,
+      "loss": 1.2796,
+      "step": 121535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.370166818104676e-06,
+      "loss": 0.9261,
+      "step": 121540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.369437744290369e-06,
+      "loss": 1.2371,
+      "step": 121545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.368708685497983e-06,
+      "loss": 1.063,
+      "step": 121550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.367979641731684e-06,
+      "loss": 1.4721,
+      "step": 121555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.367250612995642e-06,
+      "loss": 1.1803,
+      "step": 121560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.366521599294012e-06,
+      "loss": 1.2587,
+      "step": 121565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.365792600630963e-06,
+      "loss": 1.5708,
+      "step": 121570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.3650636170106485e-06,
+      "loss": 1.5916,
+      "step": 121575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.364334648437244e-06,
+      "loss": 1.1537,
+      "step": 121580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.363605694914904e-06,
+      "loss": 1.0108,
+      "step": 121585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.362876756447792e-06,
+      "loss": 1.1582,
+      "step": 121590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.362147833040075e-06,
+      "loss": 1.1256,
+      "step": 121595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.361418924695911e-06,
+      "loss": 1.4465,
+      "step": 121600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.360690031419463e-06,
+      "loss": 1.1724,
+      "step": 121605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.359961153214896e-06,
+      "loss": 1.4101,
+      "step": 121610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3592322900863734e-06,
+      "loss": 1.3983,
+      "step": 121615
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.35850344203805e-06,
+      "loss": 1.1763,
+      "step": 121620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.357774609074096e-06,
+      "loss": 1.2166,
+      "step": 121625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.357045791198669e-06,
+      "loss": 1.3318,
+      "step": 121630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.356316988415928e-06,
+      "loss": 1.1608,
+      "step": 121635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.355588200730046e-06,
+      "loss": 1.2114,
+      "step": 121640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.354859428145175e-06,
+      "loss": 1.1668,
+      "step": 121645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.35413067066548e-06,
+      "loss": 1.5549,
+      "step": 121650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.353401928295119e-06,
+      "loss": 1.3922,
+      "step": 121655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.352673201038258e-06,
+      "loss": 1.5487,
+      "step": 121660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.35194448889906e-06,
+      "loss": 1.91,
+      "step": 121665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.351215791881676e-06,
+      "loss": 0.9952,
+      "step": 121670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3504871099902805e-06,
+      "loss": 1.4138,
+      "step": 121675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.349758443229028e-06,
+      "loss": 1.3437,
+      "step": 121680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.349029791602078e-06,
+      "loss": 1.5106,
+      "step": 121685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.348301155113598e-06,
+      "loss": 1.7123,
+      "step": 121690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3475725337677405e-06,
+      "loss": 1.4657,
+      "step": 121695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3468439275686694e-06,
+      "loss": 1.45,
+      "step": 121700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.34611533652055e-06,
+      "loss": 0.9997,
+      "step": 121705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.34538676062754e-06,
+      "loss": 1.1503,
+      "step": 121710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.344658199893795e-06,
+      "loss": 1.266,
+      "step": 121715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.343929654323483e-06,
+      "loss": 1.4372,
+      "step": 121720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.343201123920762e-06,
+      "loss": 1.3062,
+      "step": 121725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.34247260868979e-06,
+      "loss": 1.2725,
+      "step": 121730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.341744108634727e-06,
+      "loss": 1.4948,
+      "step": 121735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.341015623759737e-06,
+      "loss": 1.4123,
+      "step": 121740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3402871540689794e-06,
+      "loss": 1.0677,
+      "step": 121745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.339558699566609e-06,
+      "loss": 1.5896,
+      "step": 121750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.338830260256792e-06,
+      "loss": 1.2635,
+      "step": 121755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.338101836143685e-06,
+      "loss": 0.9375,
+      "step": 121760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3373734272314455e-06,
+      "loss": 1.2634,
+      "step": 121765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.336645033524241e-06,
+      "loss": 1.2248,
+      "step": 121770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.335916655026223e-06,
+      "loss": 1.0739,
+      "step": 121775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.335188291741553e-06,
+      "loss": 1.6555,
+      "step": 121780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.334459943674393e-06,
+      "loss": 1.2405,
+      "step": 121785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3337316108289e-06,
+      "loss": 1.5386,
+      "step": 121790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.333003293209233e-06,
+      "loss": 1.4291,
+      "step": 121795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.33227499081955e-06,
+      "loss": 1.428,
+      "step": 121800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.331546703664017e-06,
+      "loss": 1.413,
+      "step": 121805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.330818431746783e-06,
+      "loss": 0.9999,
+      "step": 121810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.330090175072011e-06,
+      "loss": 1.0737,
+      "step": 121815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3293619336438615e-06,
+      "loss": 1.2942,
+      "step": 121820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.328633707466494e-06,
+      "loss": 1.4828,
+      "step": 121825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.32790549654406e-06,
+      "loss": 1.3239,
+      "step": 121830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3271773008807255e-06,
+      "loss": 1.4271,
+      "step": 121835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.326449120480649e-06,
+      "loss": 0.9934,
+      "step": 121840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3257209553479805e-06,
+      "loss": 1.5404,
+      "step": 121845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.324992805486886e-06,
+      "loss": 1.2721,
+      "step": 121850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.324264670901522e-06,
+      "loss": 1.0046,
+      "step": 121855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.323536551596045e-06,
+      "loss": 1.2079,
+      "step": 121860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3228084475746156e-06,
+      "loss": 1.275,
+      "step": 121865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.322080358841388e-06,
+      "loss": 1.6197,
+      "step": 121870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.321352285400525e-06,
+      "loss": 1.1279,
+      "step": 121875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.320624227256175e-06,
+      "loss": 1.1506,
+      "step": 121880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.319896184412507e-06,
+      "loss": 1.3882,
+      "step": 121885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.319168156873672e-06,
+      "loss": 1.3301,
+      "step": 121890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3184401446438255e-06,
+      "loss": 0.9832,
+      "step": 121895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.317712147727134e-06,
+      "loss": 1.4829,
+      "step": 121900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.316984166127746e-06,
+      "loss": 1.2139,
+      "step": 121905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3162561998498195e-06,
+      "loss": 1.5667,
+      "step": 121910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.315528248897516e-06,
+      "loss": 1.1603,
+      "step": 121915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.314800313274991e-06,
+      "loss": 1.0587,
+      "step": 121920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.314072392986398e-06,
+      "loss": 1.3519,
+      "step": 121925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.313344488035896e-06,
+      "loss": 1.5758,
+      "step": 121930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.312616598427645e-06,
+      "loss": 1.1838,
+      "step": 121935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.311888724165798e-06,
+      "loss": 1.1331,
+      "step": 121940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.311160865254508e-06,
+      "loss": 1.3823,
+      "step": 121945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.310433021697939e-06,
+      "loss": 2.088,
+      "step": 121950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.309705193500246e-06,
+      "loss": 1.3376,
+      "step": 121955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3089773806655785e-06,
+      "loss": 1.449,
+      "step": 121960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3082495831981e-06,
+      "loss": 1.1704,
+      "step": 121965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.307521801101966e-06,
+      "loss": 1.2157,
+      "step": 121970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.306794034381326e-06,
+      "loss": 1.2663,
+      "step": 121975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.306066283040345e-06,
+      "loss": 1.0357,
+      "step": 121980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.305338547083174e-06,
+      "loss": 1.2304,
+      "step": 121985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.304610826513965e-06,
+      "loss": 1.3452,
+      "step": 121990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.303883121336883e-06,
+      "loss": 1.3904,
+      "step": 121995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.303155431556077e-06,
+      "loss": 1.1526,
+      "step": 122000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3024277571757e-06,
+      "loss": 1.3419,
+      "step": 122005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.301700098199917e-06,
+      "loss": 1.5181,
+      "step": 122010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.300972454632878e-06,
+      "loss": 1.1462,
+      "step": 122015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.300244826478735e-06,
+      "loss": 1.1111,
+      "step": 122020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.299517213741645e-06,
+      "loss": 1.2258,
+      "step": 122025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.298789616425768e-06,
+      "loss": 1.2879,
+      "step": 122030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2980620345352534e-06,
+      "loss": 1.0313,
+      "step": 122035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.297334468074256e-06,
+      "loss": 1.3714,
+      "step": 122040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.296606917046933e-06,
+      "loss": 1.1543,
+      "step": 122045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.295879381457441e-06,
+      "loss": 1.6037,
+      "step": 122050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.295151861309929e-06,
+      "loss": 1.245,
+      "step": 122055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.294424356608556e-06,
+      "loss": 1.3398,
+      "step": 122060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.293696867357476e-06,
+      "loss": 1.1936,
+      "step": 122065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2929693935608406e-06,
+      "loss": 1.2484,
+      "step": 122070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.292241935222808e-06,
+      "loss": 1.6252,
+      "step": 122075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.29151449234753e-06,
+      "loss": 1.6276,
+      "step": 122080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.290787064939162e-06,
+      "loss": 1.1759,
+      "step": 122085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.290059653001854e-06,
+      "loss": 1.1414,
+      "step": 122090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.289332256539765e-06,
+      "loss": 1.4472,
+      "step": 122095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.288604875557046e-06,
+      "loss": 1.484,
+      "step": 122100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.28787751005785e-06,
+      "loss": 1.228,
+      "step": 122105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.287150160046336e-06,
+      "loss": 1.2784,
+      "step": 122110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2864228255266515e-06,
+      "loss": 1.0551,
+      "step": 122115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.28569550650295e-06,
+      "loss": 1.3757,
+      "step": 122120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.284968202979392e-06,
+      "loss": 1.1721,
+      "step": 122125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.284240914960123e-06,
+      "loss": 1.3618,
+      "step": 122130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.283513642449297e-06,
+      "loss": 1.1971,
+      "step": 122135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.282786385451071e-06,
+      "loss": 1.1076,
+      "step": 122140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2820591439696e-06,
+      "loss": 1.1268,
+      "step": 122145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.281331918009027e-06,
+      "loss": 1.7685,
+      "step": 122150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.280604707573513e-06,
+      "loss": 1.2474,
+      "step": 122155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2798775126672105e-06,
+      "loss": 1.314,
+      "step": 122160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.279150333294272e-06,
+      "loss": 1.0409,
+      "step": 122165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.278423169458844e-06,
+      "loss": 0.8535,
+      "step": 122170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.277696021165084e-06,
+      "loss": 1.1314,
+      "step": 122175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.276968888417148e-06,
+      "loss": 1.1454,
+      "step": 122180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.276241771219179e-06,
+      "loss": 1.0296,
+      "step": 122185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.275514669575338e-06,
+      "loss": 1.1999,
+      "step": 122190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.274787583489772e-06,
+      "loss": 1.4515,
+      "step": 122195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.274060512966634e-06,
+      "loss": 1.187,
+      "step": 122200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.273333458010079e-06,
+      "loss": 1.7401,
+      "step": 122205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2726064186242555e-06,
+      "loss": 1.4458,
+      "step": 122210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2718793948133145e-06,
+      "loss": 1.5815,
+      "step": 122215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.271152386581412e-06,
+      "loss": 1.1876,
+      "step": 122220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2704253939326985e-06,
+      "loss": 1.0601,
+      "step": 122225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.269698416871321e-06,
+      "loss": 1.0646,
+      "step": 122230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.268971455401432e-06,
+      "loss": 1.1497,
+      "step": 122235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.268244509527191e-06,
+      "loss": 1.5846,
+      "step": 122240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.26751757925274e-06,
+      "loss": 0.9176,
+      "step": 122245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.266790664582231e-06,
+      "loss": 1.6643,
+      "step": 122250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.266063765519819e-06,
+      "loss": 1.0031,
+      "step": 122255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.265336882069657e-06,
+      "loss": 1.2076,
+      "step": 122260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.264610014235887e-06,
+      "loss": 1.6617,
+      "step": 122265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.263883162022667e-06,
+      "loss": 1.5322,
+      "step": 122270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2631563254341485e-06,
+      "loss": 1.3243,
+      "step": 122275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.262429504474473e-06,
+      "loss": 1.0563,
+      "step": 122280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.261702699147804e-06,
+      "loss": 1.08,
+      "step": 122285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.260975909458282e-06,
+      "loss": 1.2524,
+      "step": 122290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.26024913541006e-06,
+      "loss": 1.5661,
+      "step": 122295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.259522377007292e-06,
+      "loss": 1.1945,
+      "step": 122300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.258795634254123e-06,
+      "loss": 1.2283,
+      "step": 122305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.25806890715471e-06,
+      "loss": 1.659,
+      "step": 122310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.257342195713189e-06,
+      "loss": 1.4162,
+      "step": 122315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2566154999337265e-06,
+      "loss": 1.1594,
+      "step": 122320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.255888819820464e-06,
+      "loss": 1.3167,
+      "step": 122325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2551621553775485e-06,
+      "loss": 0.9892,
+      "step": 122330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.254435506609139e-06,
+      "loss": 1.0584,
+      "step": 122335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.253708873519376e-06,
+      "loss": 1.3634,
+      "step": 122340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2529822561124104e-06,
+      "loss": 1.1885,
+      "step": 122345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.252255654392398e-06,
+      "loss": 1.7512,
+      "step": 122350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.251529068363483e-06,
+      "loss": 1.1616,
+      "step": 122355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.250802498029812e-06,
+      "loss": 0.9889,
+      "step": 122360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.250075943395541e-06,
+      "loss": 1.3781,
+      "step": 122365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.249349404464816e-06,
+      "loss": 1.8635,
+      "step": 122370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.248622881241783e-06,
+      "loss": 1.4026,
+      "step": 122375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.247896373730592e-06,
+      "loss": 1.221,
+      "step": 122380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.247169881935394e-06,
+      "loss": 1.2474,
+      "step": 122385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.246443405860339e-06,
+      "loss": 1.3793,
+      "step": 122390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.24571694550957e-06,
+      "loss": 1.595,
+      "step": 122395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.244990500887239e-06,
+      "loss": 1.3419,
+      "step": 122400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.244264071997496e-06,
+      "loss": 1.4558,
+      "step": 122405
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.243537658844484e-06,
+      "loss": 1.1618,
+      "step": 122410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.242811261432359e-06,
+      "loss": 1.4111,
+      "step": 122415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.242084879765263e-06,
+      "loss": 1.4162,
+      "step": 122420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.241358513847343e-06,
+      "loss": 1.2313,
+      "step": 122425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.240632163682754e-06,
+      "loss": 1.1896,
+      "step": 122430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.239905829275638e-06,
+      "loss": 1.1867,
+      "step": 122435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.239179510630141e-06,
+      "loss": 1.2266,
+      "step": 122440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.238453207750416e-06,
+      "loss": 1.3919,
+      "step": 122445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.237726920640611e-06,
+      "loss": 1.4638,
+      "step": 122450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.237000649304868e-06,
+      "loss": 1.3894,
+      "step": 122455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.236274393747338e-06,
+      "loss": 0.9303,
+      "step": 122460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.235548153972167e-06,
+      "loss": 1.2153,
+      "step": 122465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.234821929983507e-06,
+      "loss": 1.0067,
+      "step": 122470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.234095721785494e-06,
+      "loss": 1.2713,
+      "step": 122475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.233369529382287e-06,
+      "loss": 1.1866,
+      "step": 122480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.232643352778029e-06,
+      "loss": 1.1823,
+      "step": 122485
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.231917191976862e-06,
+      "loss": 1.2518,
+      "step": 122490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.23119104698294e-06,
+      "loss": 1.3918,
+      "step": 122495
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.230464917800404e-06,
+      "loss": 1.0289,
+      "step": 122500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2297388044334015e-06,
+      "loss": 1.6022,
+      "step": 122505
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.229012706886085e-06,
+      "loss": 1.193,
+      "step": 122510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.228286625162594e-06,
+      "loss": 1.418,
+      "step": 122515
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2275605592670795e-06,
+      "loss": 1.2616,
+      "step": 122520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.22683450920368e-06,
+      "loss": 1.0436,
+      "step": 122525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.226108474976551e-06,
+      "loss": 1.3728,
+      "step": 122530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.225382456589833e-06,
+      "loss": 1.3137,
+      "step": 122535
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.224656454047673e-06,
+      "loss": 1.3024,
+      "step": 122540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.223930467354221e-06,
+      "loss": 1.4632,
+      "step": 122545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.223204496513616e-06,
+      "loss": 1.3537,
+      "step": 122550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.222478541530006e-06,
+      "loss": 1.238,
+      "step": 122555
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.221752602407539e-06,
+      "loss": 1.3521,
+      "step": 122560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.22102667915036e-06,
+      "loss": 1.154,
+      "step": 122565
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2203007717626105e-06,
+      "loss": 1.3691,
+      "step": 122570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.219574880248441e-06,
+      "loss": 1.2416,
+      "step": 122575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.218849004611995e-06,
+      "loss": 1.2492,
+      "step": 122580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.218123144857412e-06,
+      "loss": 1.1418,
+      "step": 122585
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.217397300988849e-06,
+      "loss": 1.1041,
+      "step": 122590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.216671473010439e-06,
+      "loss": 1.1706,
+      "step": 122595
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.215945660926337e-06,
+      "loss": 1.1553,
+      "step": 122600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.215219864740677e-06,
+      "loss": 0.9748,
+      "step": 122605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.214494084457613e-06,
+      "loss": 1.447,
+      "step": 122610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.213768320081287e-06,
+      "loss": 1.26,
+      "step": 122615
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.213042571615836e-06,
+      "loss": 1.4673,
+      "step": 122620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2123168390654165e-06,
+      "loss": 1.4417,
+      "step": 122625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2115911224341665e-06,
+      "loss": 1.118,
+      "step": 122630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.210865421726227e-06,
+      "loss": 1.0003,
+      "step": 122635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.210139736945751e-06,
+      "loss": 1.3984,
+      "step": 122640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.209414068096878e-06,
+      "loss": 1.8997,
+      "step": 122645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.208688415183747e-06,
+      "loss": 1.2153,
+      "step": 122650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.207962778210509e-06,
+      "loss": 1.1098,
+      "step": 122655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2072371571813085e-06,
+      "loss": 1.3157,
+      "step": 122660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.206511552100282e-06,
+      "loss": 1.0493,
+      "step": 122665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.205785962971577e-06,
+      "loss": 1.1611,
+      "step": 122670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.205060389799339e-06,
+      "loss": 1.0418,
+      "step": 122675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.204334832587709e-06,
+      "loss": 1.3604,
+      "step": 122680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.203609291340827e-06,
+      "loss": 0.9656,
+      "step": 122685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.202883766062843e-06,
+      "loss": 1.1573,
+      "step": 122690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.202158256757899e-06,
+      "loss": 1.0686,
+      "step": 122695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.201432763430133e-06,
+      "loss": 1.507,
+      "step": 122700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.200707286083693e-06,
+      "loss": 1.0703,
+      "step": 122705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.199981824722721e-06,
+      "loss": 1.2437,
+      "step": 122710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.199256379351353e-06,
+      "loss": 0.9573,
+      "step": 122715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.198530949973743e-06,
+      "loss": 1.1073,
+      "step": 122720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.197805536594027e-06,
+      "loss": 1.3843,
+      "step": 122725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.197080139216346e-06,
+      "loss": 1.0583,
+      "step": 122730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.19635475784485e-06,
+      "loss": 1.3966,
+      "step": 122735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.195629392483674e-06,
+      "loss": 1.3692,
+      "step": 122740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.194904043136961e-06,
+      "loss": 1.1621,
+      "step": 122745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.194178709808855e-06,
+      "loss": 1.1919,
+      "step": 122750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1934533925035e-06,
+      "loss": 1.5635,
+      "step": 122755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.192728091225034e-06,
+      "loss": 1.0374,
+      "step": 122760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.192002805977598e-06,
+      "loss": 0.984,
+      "step": 122765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.191277536765342e-06,
+      "loss": 1.2062,
+      "step": 122770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.190552283592398e-06,
+      "loss": 1.3491,
+      "step": 122775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1898270464629105e-06,
+      "loss": 1.0536,
+      "step": 122780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.189101825381024e-06,
+      "loss": 1.5267,
+      "step": 122785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.18837662035088e-06,
+      "loss": 1.1834,
+      "step": 122790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.187651431376612e-06,
+      "loss": 1.6434,
+      "step": 122795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1869262584623715e-06,
+      "loss": 1.1832,
+      "step": 122800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.186201101612293e-06,
+      "loss": 1.119,
+      "step": 122805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1854759608305235e-06,
+      "loss": 1.126,
+      "step": 122810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1847508361211945e-06,
+      "loss": 1.1011,
+      "step": 122815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.184025727488455e-06,
+      "loss": 1.1329,
+      "step": 122820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.183300634936446e-06,
+      "loss": 1.2012,
+      "step": 122825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1825755584693e-06,
+      "loss": 1.1605,
+      "step": 122830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.181850498091167e-06,
+      "loss": 1.2026,
+      "step": 122835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.181125453806181e-06,
+      "loss": 1.2403,
+      "step": 122840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.180400425618483e-06,
+      "loss": 1.2839,
+      "step": 122845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.179675413532221e-06,
+      "loss": 1.4924,
+      "step": 122850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.178950417551525e-06,
+      "loss": 1.1334,
+      "step": 122855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.178225437680539e-06,
+      "loss": 1.165,
+      "step": 122860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1775004739234055e-06,
+      "loss": 1.3069,
+      "step": 122865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.176775526284263e-06,
+      "loss": 1.2293,
+      "step": 122870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1760505947672474e-06,
+      "loss": 1.3067,
+      "step": 122875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1753256793765055e-06,
+      "loss": 0.895,
+      "step": 122880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.174600780116173e-06,
+      "loss": 1.4464,
+      "step": 122885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.173875896990389e-06,
+      "loss": 1.2296,
+      "step": 122890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.173151030003291e-06,
+      "loss": 1.0173,
+      "step": 122895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.172426179159022e-06,
+      "loss": 1.259,
+      "step": 122900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1717013444617245e-06,
+      "loss": 1.138,
+      "step": 122905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.170976525915528e-06,
+      "loss": 1.2655,
+      "step": 122910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1702517235245795e-06,
+      "loss": 1.0517,
+      "step": 122915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.169526937293019e-06,
+      "loss": 1.2455,
+      "step": 122920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.168802167224974e-06,
+      "loss": 1.2099,
+      "step": 122925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.168077413324599e-06,
+      "loss": 1.2819,
+      "step": 122930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.167352675596022e-06,
+      "loss": 2.0337,
+      "step": 122935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.166627954043382e-06,
+      "loss": 1.1408,
+      "step": 122940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.165903248670827e-06,
+      "loss": 1.321,
+      "step": 122945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.165178559482484e-06,
+      "loss": 1.541,
+      "step": 122950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1644538864825e-06,
+      "loss": 1.1541,
+      "step": 122955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.163729229675003e-06,
+      "loss": 1.3715,
+      "step": 122960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.163004589064143e-06,
+      "loss": 1.0734,
+      "step": 122965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.162279964654051e-06,
+      "loss": 1.4327,
+      "step": 122970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1615553564488635e-06,
+      "loss": 1.5387,
+      "step": 122975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1608307644527284e-06,
+      "loss": 1.0619,
+      "step": 122980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.160106188669773e-06,
+      "loss": 1.1734,
+      "step": 122985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1593816291041365e-06,
+      "loss": 1.1271,
+      "step": 122990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.158657085759961e-06,
+      "loss": 1.6359,
+      "step": 122995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.157932558641384e-06,
+      "loss": 1.3956,
+      "step": 123000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.157208047752536e-06,
+      "loss": 0.9269,
+      "step": 123005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.156483553097562e-06,
+      "loss": 1.2601,
+      "step": 123010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1557590746805975e-06,
+      "loss": 1.3155,
+      "step": 123015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.155034612505773e-06,
+      "loss": 1.3298,
+      "step": 123020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.154310166577236e-06,
+      "loss": 1.1619,
+      "step": 123025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.153585736899118e-06,
+      "loss": 1.1874,
+      "step": 123030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1528613234755575e-06,
+      "loss": 1.1687,
+      "step": 123035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.152136926310687e-06,
+      "loss": 1.4765,
+      "step": 123040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.151412545408647e-06,
+      "loss": 1.3486,
+      "step": 123045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.150688180773575e-06,
+      "loss": 1.3155,
+      "step": 123050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.149963832409603e-06,
+      "loss": 1.2628,
+      "step": 123055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1492395003208745e-06,
+      "loss": 1.321,
+      "step": 123060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.14851518451152e-06,
+      "loss": 1.1292,
+      "step": 123065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.147790884985674e-06,
+      "loss": 1.2365,
+      "step": 123070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.147066601747483e-06,
+      "loss": 1.3328,
+      "step": 123075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.146342334801073e-06,
+      "loss": 1.323,
+      "step": 123080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.145618084150581e-06,
+      "loss": 1.3421,
+      "step": 123085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.144893849800146e-06,
+      "loss": 1.2239,
+      "step": 123090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.144169631753906e-06,
+      "loss": 1.1323,
+      "step": 123095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.143445430015992e-06,
+      "loss": 1.2737,
+      "step": 123100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.142721244590537e-06,
+      "loss": 1.3462,
+      "step": 123105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.141997075481684e-06,
+      "loss": 1.3409,
+      "step": 123110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.141272922693565e-06,
+      "loss": 1.1056,
+      "step": 123115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.140548786230313e-06,
+      "loss": 1.5238,
+      "step": 123120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.139824666096067e-06,
+      "loss": 1.4516,
+      "step": 123125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.139100562294962e-06,
+      "loss": 1.0398,
+      "step": 123130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.138376474831128e-06,
+      "loss": 1.1635,
+      "step": 123135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.137652403708706e-06,
+      "loss": 1.2472,
+      "step": 123140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.136928348931828e-06,
+      "loss": 0.8585,
+      "step": 123145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.136204310504626e-06,
+      "loss": 1.1315,
+      "step": 123150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1354802884312425e-06,
+      "loss": 1.4993,
+      "step": 123155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.134756282715805e-06,
+      "loss": 1.118,
+      "step": 123160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.134032293362449e-06,
+      "loss": 1.377,
+      "step": 123165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.133308320375311e-06,
+      "loss": 1.5133,
+      "step": 123170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.132584363758527e-06,
+      "loss": 1.3356,
+      "step": 123175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1318604235162255e-06,
+      "loss": 0.9891,
+      "step": 123180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.131136499652542e-06,
+      "loss": 0.9593,
+      "step": 123185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.130412592171617e-06,
+      "loss": 1.2163,
+      "step": 123190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.129688701077577e-06,
+      "loss": 1.0983,
+      "step": 123195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.128964826374557e-06,
+      "loss": 1.2085,
+      "step": 123200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.128240968066693e-06,
+      "loss": 1.1379,
+      "step": 123205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.127517126158121e-06,
+      "loss": 1.1443,
+      "step": 123210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.126793300652966e-06,
+      "loss": 1.2197,
+      "step": 123215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.126069491555369e-06,
+      "loss": 1.2337,
+      "step": 123220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.125345698869461e-06,
+      "loss": 0.9998,
+      "step": 123225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1246219225993716e-06,
+      "loss": 1.5205,
+      "step": 123230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.123898162749243e-06,
+      "loss": 1.2241,
+      "step": 123235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.123174419323201e-06,
+      "loss": 1.3826,
+      "step": 123240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.122450692325382e-06,
+      "loss": 1.1779,
+      "step": 123245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.121726981759913e-06,
+      "loss": 1.1096,
+      "step": 123250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.121003287630934e-06,
+      "loss": 1.1068,
+      "step": 123255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.120279609942576e-06,
+      "loss": 1.2319,
+      "step": 123260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.119555948698964e-06,
+      "loss": 1.0971,
+      "step": 123265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1188323039042436e-06,
+      "loss": 1.2524,
+      "step": 123270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.118108675562538e-06,
+      "loss": 1.4592,
+      "step": 123275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.11738506367798e-06,
+      "loss": 1.1686,
+      "step": 123280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1166614682547074e-06,
+      "loss": 1.287,
+      "step": 123285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1159378892968465e-06,
+      "loss": 1.0456,
+      "step": 123290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.115214326808531e-06,
+      "loss": 1.1809,
+      "step": 123295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.114490780793892e-06,
+      "loss": 1.1366,
+      "step": 123300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.113767251257068e-06,
+      "loss": 1.0866,
+      "step": 123305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.11304373820218e-06,
+      "loss": 1.6322,
+      "step": 123310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.112320241633367e-06,
+      "loss": 1.7242,
+      "step": 123315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.11159676155476e-06,
+      "loss": 1.1796,
+      "step": 123320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.110873297970489e-06,
+      "loss": 0.983,
+      "step": 123325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.110149850884681e-06,
+      "loss": 1.4894,
+      "step": 123330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.109426420301475e-06,
+      "loss": 1.4846,
+      "step": 123335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.108703006224999e-06,
+      "loss": 1.1839,
+      "step": 123340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.107979608659383e-06,
+      "loss": 1.2804,
+      "step": 123345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.107256227608759e-06,
+      "loss": 1.1686,
+      "step": 123350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.10653286307726e-06,
+      "loss": 1.3811,
+      "step": 123355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.105809515069008e-06,
+      "loss": 1.3132,
+      "step": 123360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.105086183588148e-06,
+      "loss": 1.3563,
+      "step": 123365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1043628686388e-06,
+      "loss": 1.0507,
+      "step": 123370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.103639570225096e-06,
+      "loss": 1.0102,
+      "step": 123375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1029162883511705e-06,
+      "loss": 1.0244,
+      "step": 123380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.10219302302115e-06,
+      "loss": 1.3042,
+      "step": 123385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.101469774239167e-06,
+      "loss": 1.2601,
+      "step": 123390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.100746542009347e-06,
+      "loss": 1.1291,
+      "step": 123395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.100023326335827e-06,
+      "loss": 0.989,
+      "step": 123400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.099300127222733e-06,
+      "loss": 1.0874,
+      "step": 123405
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.098576944674194e-06,
+      "loss": 0.8715,
+      "step": 123410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.0978537786943435e-06,
+      "loss": 1.1993,
+      "step": 123415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.097130629287308e-06,
+      "loss": 1.2084,
+      "step": 123420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.096407496457216e-06,
+      "loss": 1.5662,
+      "step": 123425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.0956843802082e-06,
+      "loss": 1.5167,
+      "step": 123430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.09496128054439e-06,
+      "loss": 0.9562,
+      "step": 123435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.094238197469911e-06,
+      "loss": 1.3251,
+      "step": 123440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.093515130988896e-06,
+      "loss": 1.3059,
+      "step": 123445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.092792081105473e-06,
+      "loss": 1.1205,
+      "step": 123450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.092069047823769e-06,
+      "loss": 1.1331,
+      "step": 123455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.091346031147919e-06,
+      "loss": 1.3273,
+      "step": 123460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.090623031082044e-06,
+      "loss": 0.9862,
+      "step": 123465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.08990004763028e-06,
+      "loss": 1.3221,
+      "step": 123470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.089177080796748e-06,
+      "loss": 1.0317,
+      "step": 123475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.088454130585581e-06,
+      "loss": 1.4222,
+      "step": 123480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.087731197000909e-06,
+      "loss": 1.307,
+      "step": 123485
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.087008280046854e-06,
+      "loss": 1.17,
+      "step": 123490
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.086285379727553e-06,
+      "loss": 1.3688,
+      "step": 123495
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.085562496047129e-06,
+      "loss": 1.604,
+      "step": 123500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.084839629009708e-06,
+      "loss": 1.2777,
+      "step": 123505
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.084116778619422e-06,
+      "loss": 1.7111,
+      "step": 123510
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.0833939448804e-06,
+      "loss": 1.293,
+      "step": 123515
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.082671127796763e-06,
+      "loss": 1.557,
+      "step": 123520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.081948327372645e-06,
+      "loss": 1.379,
+      "step": 123525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.081225543612174e-06,
+      "loss": 1.3688,
+      "step": 123530
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.080502776519474e-06,
+      "loss": 0.9879,
+      "step": 123535
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.079780026098668e-06,
+      "loss": 0.9962,
+      "step": 123540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.079057292353894e-06,
+      "loss": 1.3661,
+      "step": 123545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.078334575289274e-06,
+      "loss": 1.394,
+      "step": 123550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.077611874908932e-06,
+      "loss": 1.3272,
+      "step": 123555
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.076889191217e-06,
+      "loss": 1.3832,
+      "step": 123560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.076166524217605e-06,
+      "loss": 1.1912,
+      "step": 123565
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.0754438739148645e-06,
+      "loss": 1.3473,
+      "step": 123570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.07472124031292e-06,
+      "loss": 1.4782,
+      "step": 123575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.073998623415887e-06,
+      "loss": 1.198,
+      "step": 123580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.0732760232278955e-06,
+      "loss": 1.1528,
+      "step": 123585
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.072553439753075e-06,
+      "loss": 1.5212,
+      "step": 123590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.071830872995547e-06,
+      "loss": 1.44,
+      "step": 123595
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.071108322959437e-06,
+      "loss": 1.0436,
+      "step": 123600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.070385789648877e-06,
+      "loss": 1.1689,
+      "step": 123605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.069663273067992e-06,
+      "loss": 1.454,
+      "step": 123610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.0689407732209026e-06,
+      "loss": 1.4113,
+      "step": 123615
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.068218290111737e-06,
+      "loss": 1.3099,
+      "step": 123620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.067495823744624e-06,
+      "loss": 0.97,
+      "step": 123625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.066773374123686e-06,
+      "loss": 1.355,
+      "step": 123630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.066050941253048e-06,
+      "loss": 1.1659,
+      "step": 123635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.06532852513684e-06,
+      "loss": 1.6451,
+      "step": 123640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.064606125779186e-06,
+      "loss": 1.0456,
+      "step": 123645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.0638837431842045e-06,
+      "loss": 1.1463,
+      "step": 123650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.0631613773560295e-06,
+      "loss": 1.2106,
+      "step": 123655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.062439028298784e-06,
+      "loss": 1.0111,
+      "step": 123660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.061716696016585e-06,
+      "loss": 1.1791,
+      "step": 123665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.060994380513571e-06,
+      "loss": 1.2081,
+      "step": 123670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.060272081793858e-06,
+      "loss": 1.1246,
+      "step": 123675
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.059549799861574e-06,
+      "loss": 0.9357,
+      "step": 123680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.058827534720838e-06,
+      "loss": 1.3064,
+      "step": 123685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.058105286375781e-06,
+      "loss": 1.1949,
+      "step": 123690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.057383054830525e-06,
+      "loss": 1.0786,
+      "step": 123695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.056660840089193e-06,
+      "loss": 1.5741,
+      "step": 123700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.055938642155915e-06,
+      "loss": 1.3213,
+      "step": 123705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.055216461034807e-06,
+      "loss": 1.3657,
+      "step": 123710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.054494296729997e-06,
+      "loss": 1.5329,
+      "step": 123715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.053772149245611e-06,
+      "loss": 1.2719,
+      "step": 123720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.05305001858577e-06,
+      "loss": 1.2845,
+      "step": 123725
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.052327904754594e-06,
+      "loss": 1.1664,
+      "step": 123730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0516058077562166e-06,
+      "loss": 1.0517,
+      "step": 123735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0508837275947565e-06,
+      "loss": 1.2373,
+      "step": 123740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.050161664274331e-06,
+      "loss": 1.2809,
+      "step": 123745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.049439617799072e-06,
+      "loss": 1.043,
+      "step": 123750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0487175881731e-06,
+      "loss": 1.2098,
+      "step": 123755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0479955754005405e-06,
+      "loss": 1.3184,
+      "step": 123760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.047273579485509e-06,
+      "loss": 1.2697,
+      "step": 123765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.046551600432137e-06,
+      "loss": 1.7728,
+      "step": 123770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.045829638244543e-06,
+      "loss": 1.2948,
+      "step": 123775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.045107692926849e-06,
+      "loss": 0.995,
+      "step": 123780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.044385764483182e-06,
+      "loss": 1.3294,
+      "step": 123785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.043663852917659e-06,
+      "loss": 1.1514,
+      "step": 123790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.042941958234406e-06,
+      "loss": 0.9633,
+      "step": 123795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.042220080437548e-06,
+      "loss": 1.6292,
+      "step": 123800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.041498219531202e-06,
+      "loss": 1.2867,
+      "step": 123805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.04077637551949e-06,
+      "loss": 1.2812,
+      "step": 123810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.04005454840654e-06,
+      "loss": 1.1646,
+      "step": 123815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0393327381964715e-06,
+      "loss": 1.1527,
+      "step": 123820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0386109448934035e-06,
+      "loss": 1.2613,
+      "step": 123825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0378891685014576e-06,
+      "loss": 1.352,
+      "step": 123830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.037167409024763e-06,
+      "loss": 1.367,
+      "step": 123835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.036445666467433e-06,
+      "loss": 1.2784,
+      "step": 123840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.03572394083359e-06,
+      "loss": 1.1283,
+      "step": 123845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.035002232127359e-06,
+      "loss": 1.4899,
+      "step": 123850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0342805403528645e-06,
+      "loss": 1.0376,
+      "step": 123855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.033558865514218e-06,
+      "loss": 1.3819,
+      "step": 123860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.032837207615547e-06,
+      "loss": 1.381,
+      "step": 123865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0321155666609744e-06,
+      "loss": 1.0949,
+      "step": 123870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.031393942654614e-06,
+      "loss": 1.4152,
+      "step": 123875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0306723356005925e-06,
+      "loss": 1.1369,
+      "step": 123880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.029950745503029e-06,
+      "loss": 1.2086,
+      "step": 123885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.029229172366042e-06,
+      "loss": 1.3075,
+      "step": 123890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.028507616193758e-06,
+      "loss": 1.1724,
+      "step": 123895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.027786076990293e-06,
+      "loss": 1.0845,
+      "step": 123900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.027064554759771e-06,
+      "loss": 1.2669,
+      "step": 123905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.026343049506301e-06,
+      "loss": 1.1098,
+      "step": 123910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.025621561234018e-06,
+      "loss": 1.3092,
+      "step": 123915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.024900089947035e-06,
+      "loss": 1.3876,
+      "step": 123920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.024178635649469e-06,
+      "loss": 1.4589,
+      "step": 123925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.023457198345449e-06,
+      "loss": 0.9296,
+      "step": 123930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.022735778039086e-06,
+      "loss": 1.1481,
+      "step": 123935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.022014374734502e-06,
+      "loss": 1.5082,
+      "step": 123940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.021292988435819e-06,
+      "loss": 1.5038,
+      "step": 123945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.020571619147158e-06,
+      "loss": 1.5566,
+      "step": 123950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.019850266872631e-06,
+      "loss": 1.4035,
+      "step": 123955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.019128931616363e-06,
+      "loss": 1.056,
+      "step": 123960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.018407613382476e-06,
+      "loss": 1.1402,
+      "step": 123965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.017686312175082e-06,
+      "loss": 1.2341,
+      "step": 123970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.016965027998301e-06,
+      "loss": 1.0957,
+      "step": 123975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0162437608562565e-06,
+      "loss": 1.1483,
+      "step": 123980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.015522510753068e-06,
+      "loss": 1.4063,
+      "step": 123985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.014801277692845e-06,
+      "loss": 1.3096,
+      "step": 123990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.014080061679716e-06,
+      "loss": 1.2533,
+      "step": 123995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.013358862717797e-06,
+      "loss": 1.1671,
+      "step": 124000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.012637680811198e-06,
+      "loss": 1.3958,
+      "step": 124005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0119165159640535e-06,
+      "loss": 1.2412,
+      "step": 124010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.011195368180469e-06,
+      "loss": 1.1624,
+      "step": 124015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.010474237464565e-06,
+      "loss": 1.2561,
+      "step": 124020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0097531238204654e-06,
+      "loss": 1.1128,
+      "step": 124025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.009032027252281e-06,
+      "loss": 1.1105,
+      "step": 124030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.00831094776413e-06,
+      "loss": 1.2664,
+      "step": 124035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.007589885360136e-06,
+      "loss": 1.1953,
+      "step": 124040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.006868840044414e-06,
+      "loss": 1.4489,
+      "step": 124045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.00614781182108e-06,
+      "loss": 1.0731,
+      "step": 124050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.005426800694249e-06,
+      "loss": 1.5306,
+      "step": 124055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0047058066680465e-06,
+      "loss": 1.2736,
+      "step": 124060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.003984829746583e-06,
+      "loss": 1.0102,
+      "step": 124065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0032638699339754e-06,
+      "loss": 1.0643,
+      "step": 124070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.002542927234343e-06,
+      "loss": 1.554,
+      "step": 124075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.001822001651808e-06,
+      "loss": 1.2247,
+      "step": 124080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.001101093190475e-06,
+      "loss": 0.9605,
+      "step": 124085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.000380201854471e-06,
+      "loss": 1.1878,
+      "step": 124090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9996593276479095e-06,
+      "loss": 1.4025,
+      "step": 124095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.998938470574905e-06,
+      "loss": 1.1869,
+      "step": 124100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.99821763063958e-06,
+      "loss": 0.9808,
+      "step": 124105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.997496807846045e-06,
+      "loss": 1.3042,
+      "step": 124110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.996776002198415e-06,
+      "loss": 1.3203,
+      "step": 124115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.996055213700815e-06,
+      "loss": 1.4324,
+      "step": 124120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.995334442357351e-06,
+      "loss": 1.3191,
+      "step": 124125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.994613688172145e-06,
+      "loss": 1.3536,
+      "step": 124130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.99389295114931e-06,
+      "loss": 1.045,
+      "step": 124135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.993172231292966e-06,
+      "loss": 1.0717,
+      "step": 124140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.992451528607225e-06,
+      "loss": 1.1303,
+      "step": 124145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9917308430962e-06,
+      "loss": 1.5988,
+      "step": 124150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.991010174764012e-06,
+      "loss": 1.6773,
+      "step": 124155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.990289523614777e-06,
+      "loss": 1.3925,
+      "step": 124160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9895688896526025e-06,
+      "loss": 1.546,
+      "step": 124165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.988848272881612e-06,
+      "loss": 1.3369,
+      "step": 124170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.988127673305919e-06,
+      "loss": 1.6512,
+      "step": 124175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.987407090929632e-06,
+      "loss": 1.552,
+      "step": 124180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.986686525756874e-06,
+      "loss": 1.4304,
+      "step": 124185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.985965977791757e-06,
+      "loss": 1.2591,
+      "step": 124190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.985245447038396e-06,
+      "loss": 1.4528,
+      "step": 124195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.984524933500903e-06,
+      "loss": 1.2414,
+      "step": 124200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.983804437183394e-06,
+      "loss": 1.3746,
+      "step": 124205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.983083958089988e-06,
+      "loss": 1.2649,
+      "step": 124210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.982363496224788e-06,
+      "loss": 0.9988,
+      "step": 124215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.981643051591922e-06,
+      "loss": 1.471,
+      "step": 124220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9809226241954955e-06,
+      "loss": 1.2929,
+      "step": 124225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.980202214039622e-06,
+      "loss": 1.2306,
+      "step": 124230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.979481821128424e-06,
+      "loss": 1.1445,
+      "step": 124235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.978761445466004e-06,
+      "loss": 1.2569,
+      "step": 124240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.978041087056482e-06,
+      "loss": 1.372,
+      "step": 124245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9773207459039725e-06,
+      "loss": 1.1583,
+      "step": 124250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.97660042201259e-06,
+      "loss": 1.3574,
+      "step": 124255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.975880115386439e-06,
+      "loss": 1.4198,
+      "step": 124260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.975159826029641e-06,
+      "loss": 1.7215,
+      "step": 124265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.974439553946311e-06,
+      "loss": 1.0234,
+      "step": 124270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9737192991405554e-06,
+      "loss": 1.1394,
+      "step": 124275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.97299906161649e-06,
+      "loss": 1.0415,
+      "step": 124280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.972278841378228e-06,
+      "loss": 1.3582,
+      "step": 124285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.971558638429885e-06,
+      "loss": 1.3552,
+      "step": 124290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.970838452775568e-06,
+      "loss": 1.266,
+      "step": 124295
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.970118284419393e-06,
+      "loss": 1.4673,
+      "step": 124300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.969398133365475e-06,
+      "loss": 1.2461,
+      "step": 124305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9686779996179175e-06,
+      "loss": 1.2548,
+      "step": 124310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.967957883180845e-06,
+      "loss": 1.2384,
+      "step": 124315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.967237784058363e-06,
+      "loss": 1.3977,
+      "step": 124320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.96651770225458e-06,
+      "loss": 1.2826,
+      "step": 124325
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.965797637773619e-06,
+      "loss": 1.421,
+      "step": 124330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.965077590619582e-06,
+      "loss": 1.2935,
+      "step": 124335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9643575607965845e-06,
+      "loss": 1.1803,
+      "step": 124340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.963637548308737e-06,
+      "loss": 0.9673,
+      "step": 124345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.962917553160156e-06,
+      "loss": 1.2486,
+      "step": 124350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.962197575354945e-06,
+      "loss": 1.2826,
+      "step": 124355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.961477614897219e-06,
+      "loss": 1.1496,
+      "step": 124360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.960757671791096e-06,
+      "loss": 1.1929,
+      "step": 124365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.960037746040678e-06,
+      "loss": 1.4661,
+      "step": 124370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.959317837650077e-06,
+      "loss": 0.9787,
+      "step": 124375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.958597946623409e-06,
+      "loss": 1.1955,
+      "step": 124380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.957878072964784e-06,
+      "loss": 1.3834,
+      "step": 124385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.957158216678308e-06,
+      "loss": 1.4926,
+      "step": 124390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.956438377768098e-06,
+      "loss": 1.2284,
+      "step": 124395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.955718556238262e-06,
+      "loss": 1.0588,
+      "step": 124400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.954998752092905e-06,
+      "loss": 1.4477,
+      "step": 124405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9542789653361496e-06,
+      "loss": 0.9626,
+      "step": 124410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9535591959720975e-06,
+      "loss": 1.2354,
+      "step": 124415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.952839444004862e-06,
+      "loss": 1.115,
+      "step": 124420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.95211970943855e-06,
+      "loss": 1.2051,
+      "step": 124425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.951399992277274e-06,
+      "loss": 1.0765,
+      "step": 124430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.950680292525144e-06,
+      "loss": 0.9597,
+      "step": 124435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.949960610186268e-06,
+      "loss": 1.1176,
+      "step": 124440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.949240945264761e-06,
+      "loss": 1.2372,
+      "step": 124445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.948521297764728e-06,
+      "loss": 1.279,
+      "step": 124450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.947801667690276e-06,
+      "loss": 1.4034,
+      "step": 124455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9470820550455235e-06,
+      "loss": 1.119,
+      "step": 124460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.946362459834573e-06,
+      "loss": 1.1735,
+      "step": 124465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.945642882061531e-06,
+      "loss": 1.4501,
+      "step": 124470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.944923321730515e-06,
+      "loss": 1.2265,
+      "step": 124475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9442037788456306e-06,
+      "loss": 1.3297,
+      "step": 124480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.943484253410985e-06,
+      "loss": 1.2758,
+      "step": 124485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.942764745430686e-06,
+      "loss": 1.1413,
+      "step": 124490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.942045254908847e-06,
+      "loss": 1.4853,
+      "step": 124495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.941325781849576e-06,
+      "loss": 1.1287,
+      "step": 124500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.940606326256976e-06,
+      "loss": 1.2605,
+      "step": 124505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.939886888135161e-06,
+      "loss": 1.1117,
+      "step": 124510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.93916746748824e-06,
+      "loss": 1.1332,
+      "step": 124515
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.938448064320316e-06,
+      "loss": 1.1112,
+      "step": 124520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.937728678635501e-06,
+      "loss": 1.4121,
+      "step": 124525
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.937009310437902e-06,
+      "loss": 1.2913,
+      "step": 124530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.936289959731626e-06,
+      "loss": 1.1273,
+      "step": 124535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9355706265207864e-06,
+      "loss": 1.4199,
+      "step": 124540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.934851310809485e-06,
+      "loss": 1.0566,
+      "step": 124545
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.934132012601829e-06,
+      "loss": 1.4482,
+      "step": 124550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.933412731901928e-06,
+      "loss": 1.1042,
+      "step": 124555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.932693468713894e-06,
+      "loss": 1.3428,
+      "step": 124560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.931974223041828e-06,
+      "loss": 1.3241,
+      "step": 124565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.931254994889837e-06,
+      "loss": 1.031,
+      "step": 124570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.930535784262033e-06,
+      "loss": 1.337,
+      "step": 124575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.929816591162522e-06,
+      "loss": 1.0341,
+      "step": 124580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.929097415595405e-06,
+      "loss": 1.2784,
+      "step": 124585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.928378257564796e-06,
+      "loss": 1.1751,
+      "step": 124590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.927659117074801e-06,
+      "loss": 1.5234,
+      "step": 124595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.92693999412952e-06,
+      "loss": 1.3101,
+      "step": 124600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.926220888733068e-06,
+      "loss": 1.1152,
+      "step": 124605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.92550180088955e-06,
+      "loss": 1.3158,
+      "step": 124610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.924782730603064e-06,
+      "loss": 1.3427,
+      "step": 124615
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.924063677877728e-06,
+      "loss": 1.4326,
+      "step": 124620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.923344642717642e-06,
+      "loss": 1.2735,
+      "step": 124625
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.922625625126914e-06,
+      "loss": 1.3785,
+      "step": 124630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9219066251096446e-06,
+      "loss": 1.126,
+      "step": 124635
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.921187642669946e-06,
+      "loss": 1.0319,
+      "step": 124640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.920468677811926e-06,
+      "loss": 1.1326,
+      "step": 124645
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9197497305396775e-06,
+      "loss": 1.5071,
+      "step": 124650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.919030800857323e-06,
+      "loss": 1.5198,
+      "step": 124655
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.918311888768956e-06,
+      "loss": 1.2841,
+      "step": 124660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.917592994278683e-06,
+      "loss": 1.634,
+      "step": 124665
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9168741173906175e-06,
+      "loss": 1.108,
+      "step": 124670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9161552581088565e-06,
+      "loss": 1.1687,
+      "step": 124675
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.915436416437507e-06,
+      "loss": 1.2865,
+      "step": 124680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.914717592380674e-06,
+      "loss": 1.3341,
+      "step": 124685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.913998785942466e-06,
+      "loss": 1.3387,
+      "step": 124690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.913279997126982e-06,
+      "loss": 1.4894,
+      "step": 124695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.91256122593833e-06,
+      "loss": 1.4613,
+      "step": 124700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.911842472380616e-06,
+      "loss": 1.2524,
+      "step": 124705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.911123736457941e-06,
+      "loss": 1.2748,
+      "step": 124710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9104050181744085e-06,
+      "loss": 1.3437,
+      "step": 124715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9096863175341275e-06,
+      "loss": 0.9597,
+      "step": 124720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.908967634541202e-06,
+      "loss": 1.3835,
+      "step": 124725
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.908248969199729e-06,
+      "loss": 0.952,
+      "step": 124730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.90753032151382e-06,
+      "loss": 1.0611,
+      "step": 124735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.906811691487575e-06,
+      "loss": 1.6905,
+      "step": 124740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.906093079125098e-06,
+      "loss": 1.0892,
+      "step": 124745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.905374484430497e-06,
+      "loss": 1.2571,
+      "step": 124750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.904655907407871e-06,
+      "loss": 1.1189,
+      "step": 124755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.903937348061321e-06,
+      "loss": 1.4158,
+      "step": 124760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9032188063949605e-06,
+      "loss": 1.2185,
+      "step": 124765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9025002824128815e-06,
+      "loss": 1.2187,
+      "step": 124770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.901781776119194e-06,
+      "loss": 1.2317,
+      "step": 124775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.901063287517995e-06,
+      "loss": 1.1135,
+      "step": 124780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.9003448166133965e-06,
+      "loss": 1.4059,
+      "step": 124785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.899626363409495e-06,
+      "loss": 1.3338,
+      "step": 124790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.898907927910391e-06,
+      "loss": 1.0795,
+      "step": 124795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.898189510120194e-06,
+      "loss": 0.9399,
+      "step": 124800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.897471110043004e-06,
+      "loss": 1.4617,
+      "step": 124805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.896752727682919e-06,
+      "loss": 1.1306,
+      "step": 124810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.896034363044048e-06,
+      "loss": 1.1697,
+      "step": 124815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.895316016130491e-06,
+      "loss": 1.2382,
+      "step": 124820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.894597686946346e-06,
+      "loss": 1.275,
+      "step": 124825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.89387937549572e-06,
+      "loss": 1.3694,
+      "step": 124830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8931610817827144e-06,
+      "loss": 1.0152,
+      "step": 124835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8924428058114275e-06,
+      "loss": 1.1221,
+      "step": 124840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.891724547585967e-06,
+      "loss": 1.289,
+      "step": 124845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.89100630711043e-06,
+      "loss": 1.1491,
+      "step": 124850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8902880843889206e-06,
+      "loss": 1.1718,
+      "step": 124855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.889569879425533e-06,
+      "loss": 1.2279,
+      "step": 124860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.888851692224381e-06,
+      "loss": 1.1679,
+      "step": 124865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.888133522789556e-06,
+      "loss": 1.2582,
+      "step": 124870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.887415371125161e-06,
+      "loss": 1.3143,
+      "step": 124875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.886697237235303e-06,
+      "loss": 0.9483,
+      "step": 124880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.885979121124076e-06,
+      "loss": 1.0774,
+      "step": 124885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.885261022795579e-06,
+      "loss": 1.0912,
+      "step": 124890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.88454294225392e-06,
+      "loss": 1.2148,
+      "step": 124895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8838248795032005e-06,
+      "loss": 1.2193,
+      "step": 124900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8831068345475105e-06,
+      "loss": 1.2158,
+      "step": 124905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.882388807390959e-06,
+      "loss": 1.2545,
+      "step": 124910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.881670798037647e-06,
+      "loss": 1.141,
+      "step": 124915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.880952806491671e-06,
+      "loss": 1.6744,
+      "step": 124920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.880234832757129e-06,
+      "loss": 1.1752,
+      "step": 124925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.879516876838126e-06,
+      "loss": 1.3472,
+      "step": 124930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.878798938738762e-06,
+      "loss": 0.9711,
+      "step": 124935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8780810184631295e-06,
+      "loss": 1.3772,
+      "step": 124940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.877363116015337e-06,
+      "loss": 1.0203,
+      "step": 124945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.876645231399483e-06,
+      "loss": 1.8119,
+      "step": 124950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.875927364619659e-06,
+      "loss": 0.8944,
+      "step": 124955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.875209515679975e-06,
+      "loss": 1.1653,
+      "step": 124960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.874491684584523e-06,
+      "loss": 1.1866,
+      "step": 124965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.873773871337404e-06,
+      "loss": 1.3464,
+      "step": 124970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.873056075942719e-06,
+      "loss": 1.4483,
+      "step": 124975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.872338298404566e-06,
+      "loss": 1.151,
+      "step": 124980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.871620538727041e-06,
+      "loss": 1.3211,
+      "step": 124985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.870902796914247e-06,
+      "loss": 1.2116,
+      "step": 124990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.870185072970284e-06,
+      "loss": 0.8922,
+      "step": 124995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.869467366899244e-06,
+      "loss": 1.2287,
+      "step": 125000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.868749678705228e-06,
+      "loss": 1.1802,
+      "step": 125005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.868032008392339e-06,
+      "loss": 1.0219,
+      "step": 125010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8673143559646705e-06,
+      "loss": 1.2868,
+      "step": 125015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.866596721426319e-06,
+      "loss": 1.6024,
+      "step": 125020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.865879104781389e-06,
+      "loss": 1.2213,
+      "step": 125025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.865161506033976e-06,
+      "loss": 1.335,
+      "step": 125030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.864443925188173e-06,
+      "loss": 1.4551,
+      "step": 125035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.863726362248085e-06,
+      "loss": 1.5028,
+      "step": 125040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8630088172178055e-06,
+      "loss": 1.1965,
+      "step": 125045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.862291290101428e-06,
+      "loss": 1.0962,
+      "step": 125050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8615737809030615e-06,
+      "loss": 1.2461,
+      "step": 125055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.860856289626794e-06,
+      "loss": 1.2804,
+      "step": 125060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.860138816276729e-06,
+      "loss": 1.3927,
+      "step": 125065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8594213608569546e-06,
+      "loss": 0.8533,
+      "step": 125070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.858703923371576e-06,
+      "loss": 1.5148,
+      "step": 125075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8579865038246875e-06,
+      "loss": 1.2045,
+      "step": 125080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.857269102220383e-06,
+      "loss": 1.0527,
+      "step": 125085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.856551718562768e-06,
+      "loss": 1.0806,
+      "step": 125090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.85583435285593e-06,
+      "loss": 1.2585,
+      "step": 125095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.855117005103969e-06,
+      "loss": 1.343,
+      "step": 125100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.854399675310983e-06,
+      "loss": 1.3238,
+      "step": 125105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.853682363481065e-06,
+      "loss": 1.2513,
+      "step": 125110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8529650696183126e-06,
+      "loss": 1.1625,
+      "step": 125115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.852247793726823e-06,
+      "loss": 1.0279,
+      "step": 125120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.851530535810693e-06,
+      "loss": 1.1662,
+      "step": 125125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8508132958740145e-06,
+      "loss": 1.0701,
+      "step": 125130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.850096073920888e-06,
+      "loss": 1.2514,
+      "step": 125135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.849378869955405e-06,
+      "loss": 1.0918,
+      "step": 125140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.848661683981667e-06,
+      "loss": 1.1914,
+      "step": 125145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.847944516003762e-06,
+      "loss": 1.3792,
+      "step": 125150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.84722736602579e-06,
+      "loss": 1.7523,
+      "step": 125155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.846510234051849e-06,
+      "loss": 1.1196,
+      "step": 125160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.845793120086024e-06,
+      "loss": 1.3827,
+      "step": 125165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.845076024132421e-06,
+      "loss": 1.1755,
+      "step": 125170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.844358946195131e-06,
+      "loss": 1.1606,
+      "step": 125175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.843641886278245e-06,
+      "loss": 1.3387,
+      "step": 125180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.842924844385867e-06,
+      "loss": 1.3681,
+      "step": 125185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8422078205220835e-06,
+      "loss": 1.1151,
+      "step": 125190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8414908146909896e-06,
+      "loss": 1.3319,
+      "step": 125195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.840773826896684e-06,
+      "loss": 1.2158,
+      "step": 125200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.840056857143262e-06,
+      "loss": 1.2263,
+      "step": 125205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.839339905434812e-06,
+      "loss": 1.0856,
+      "step": 125210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.838622971775428e-06,
+      "loss": 1.0725,
+      "step": 125215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.837906056169212e-06,
+      "loss": 1.1033,
+      "step": 125220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.837189158620252e-06,
+      "loss": 1.1116,
+      "step": 125225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.836472279132641e-06,
+      "loss": 2.0662,
+      "step": 125230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.835755417710477e-06,
+      "loss": 1.4793,
+      "step": 125235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.835038574357853e-06,
+      "loss": 0.9291,
+      "step": 125240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.834321749078856e-06,
+      "loss": 1.2703,
+      "step": 125245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.833604941877587e-06,
+      "loss": 1.2924,
+      "step": 125250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.832888152758139e-06,
+      "loss": 1.2098,
+      "step": 125255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8321713817245976e-06,
+      "loss": 1.0939,
+      "step": 125260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.831454628781067e-06,
+      "loss": 1.1451,
+      "step": 125265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8307378939316336e-06,
+      "loss": 1.5476,
+      "step": 125270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.830021177180388e-06,
+      "loss": 1.4162,
+      "step": 125275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.829304478531431e-06,
+      "loss": 1.1373,
+      "step": 125280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.828587797988849e-06,
+      "loss": 1.8029,
+      "step": 125285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.827871135556738e-06,
+      "loss": 1.4147,
+      "step": 125290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.827154491239183e-06,
+      "loss": 1.316,
+      "step": 125295
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.82643786504029e-06,
+      "loss": 1.0368,
+      "step": 125300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.825721256964139e-06,
+      "loss": 0.9584,
+      "step": 125305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.825004667014827e-06,
+      "loss": 1.2832,
+      "step": 125310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8242880951964495e-06,
+      "loss": 1.3278,
+      "step": 125315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.823571541513092e-06,
+      "loss": 1.2174,
+      "step": 125320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8228550059688495e-06,
+      "loss": 1.3786,
+      "step": 125325
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.822138488567813e-06,
+      "loss": 1.1514,
+      "step": 125330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.821421989314079e-06,
+      "loss": 1.4345,
+      "step": 125335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.820705508211731e-06,
+      "loss": 1.6632,
+      "step": 125340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.819989045264865e-06,
+      "loss": 1.3685,
+      "step": 125345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.819272600477576e-06,
+      "loss": 1.3069,
+      "step": 125350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.818556173853948e-06,
+      "loss": 1.1196,
+      "step": 125355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.817839765398074e-06,
+      "loss": 1.4821,
+      "step": 125360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.817123375114047e-06,
+      "loss": 1.2713,
+      "step": 125365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.81640700300596e-06,
+      "loss": 1.173,
+      "step": 125370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.815690649077899e-06,
+      "loss": 1.3204,
+      "step": 125375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.814974313333957e-06,
+      "loss": 1.3567,
+      "step": 125380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.814257995778226e-06,
+      "loss": 1.0142,
+      "step": 125385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8135416964147934e-06,
+      "loss": 1.1632,
+      "step": 125390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.812825415247754e-06,
+      "loss": 1.2672,
+      "step": 125395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.812109152281197e-06,
+      "loss": 1.3983,
+      "step": 125400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.811392907519205e-06,
+      "loss": 1.1167,
+      "step": 125405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.810676680965882e-06,
+      "loss": 1.0417,
+      "step": 125410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8099604726253075e-06,
+      "loss": 1.1616,
+      "step": 125415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.809244282501572e-06,
+      "loss": 1.3472,
+      "step": 125420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.808528110598769e-06,
+      "loss": 1.259,
+      "step": 125425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.807811956920991e-06,
+      "loss": 1.148,
+      "step": 125430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.80709582147232e-06,
+      "loss": 1.1037,
+      "step": 125435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.806379704256848e-06,
+      "loss": 1.0706,
+      "step": 125440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.805663605278667e-06,
+      "loss": 1.0062,
+      "step": 125445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.804947524541867e-06,
+      "loss": 1.0397,
+      "step": 125450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.804231462050532e-06,
+      "loss": 1.3442,
+      "step": 125455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.803515417808755e-06,
+      "loss": 1.3603,
+      "step": 125460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.802799391820627e-06,
+      "loss": 1.1707,
+      "step": 125465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.80208338409023e-06,
+      "loss": 1.2041,
+      "step": 125470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8013673946216585e-06,
+      "loss": 1.169,
+      "step": 125475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8006514234190006e-06,
+      "loss": 1.1223,
+      "step": 125480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.799935470486341e-06,
+      "loss": 1.3371,
+      "step": 125485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.799219535827774e-06,
+      "loss": 1.2418,
+      "step": 125490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.798503619447385e-06,
+      "loss": 1.1249,
+      "step": 125495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.797787721349264e-06,
+      "loss": 1.1274,
+      "step": 125500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.79707184153749e-06,
+      "loss": 1.13,
+      "step": 125505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.796355980016165e-06,
+      "loss": 1.5362,
+      "step": 125510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.79564013678937e-06,
+      "loss": 1.1947,
+      "step": 125515
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.79492431186119e-06,
+      "loss": 1.2684,
+      "step": 125520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.79420850523572e-06,
+      "loss": 1.3618,
+      "step": 125525
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.793492716917041e-06,
+      "loss": 1.0663,
+      "step": 125530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.792776946909242e-06,
+      "loss": 1.2376,
+      "step": 125535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.792061195216414e-06,
+      "loss": 0.9584,
+      "step": 125540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.791345461842643e-06,
+      "loss": 1.392,
+      "step": 125545
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.790629746792012e-06,
+      "loss": 1.1756,
+      "step": 125550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.789914050068613e-06,
+      "loss": 1.3508,
+      "step": 125555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.789198371676533e-06,
+      "loss": 1.6829,
+      "step": 125560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.7884827116198544e-06,
+      "loss": 1.2885,
+      "step": 125565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.787767069902669e-06,
+      "loss": 1.268,
+      "step": 125570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.787051446529059e-06,
+      "loss": 1.1208,
+      "step": 125575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.786335841503118e-06,
+      "loss": 1.2173,
+      "step": 125580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.7856202548289216e-06,
+      "loss": 1.2761,
+      "step": 125585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.784904686510565e-06,
+      "loss": 1.0306,
+      "step": 125590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.784189136552135e-06,
+      "loss": 1.2046,
+      "step": 125595
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.783473604957708e-06,
+      "loss": 1.8384,
+      "step": 125600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.782758091731383e-06,
+      "loss": 1.1167,
+      "step": 125605
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.782042596877236e-06,
+      "loss": 1.2002,
+      "step": 125610
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.781327120399357e-06,
+      "loss": 1.3009,
+      "step": 125615
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.780611662301834e-06,
+      "loss": 1.3664,
+      "step": 125620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.779896222588749e-06,
+      "loss": 1.5066,
+      "step": 125625
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.779180801264185e-06,
+      "loss": 1.2323,
+      "step": 125630
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.778465398332234e-06,
+      "loss": 1.0311,
+      "step": 125635
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.77775001379698e-06,
+      "loss": 1.3044,
+      "step": 125640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.777034647662504e-06,
+      "loss": 1.2721,
+      "step": 125645
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.776319299932892e-06,
+      "loss": 1.2007,
+      "step": 125650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.775603970612234e-06,
+      "loss": 1.0715,
+      "step": 125655
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.77488865970461e-06,
+      "loss": 1.0182,
+      "step": 125660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.774173367214105e-06,
+      "loss": 1.2059,
+      "step": 125665
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.773458093144807e-06,
+      "loss": 1.0954,
+      "step": 125670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.772742837500801e-06,
+      "loss": 1.1141,
+      "step": 125675
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.772027600286164e-06,
+      "loss": 1.0202,
+      "step": 125680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.771312381504989e-06,
+      "loss": 1.554,
+      "step": 125685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.770597181161358e-06,
+      "loss": 1.2695,
+      "step": 125690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.7698819992593486e-06,
+      "loss": 1.142,
+      "step": 125695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.769166835803056e-06,
+      "loss": 1.6372,
+      "step": 125700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.768451690796556e-06,
+      "loss": 1.0415,
+      "step": 125705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.767736564243933e-06,
+      "loss": 1.0577,
+      "step": 125710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.7670214561492765e-06,
+      "loss": 1.3855,
+      "step": 125715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.766306366516666e-06,
+      "loss": 1.2824,
+      "step": 125720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.765591295350185e-06,
+      "loss": 1.1654,
+      "step": 125725
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.764876242653913e-06,
+      "loss": 1.2567,
+      "step": 125730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.764161208431943e-06,
+      "loss": 1.188,
+      "step": 125735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.7634461926883525e-06,
+      "loss": 1.1781,
+      "step": 125740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.762731195427222e-06,
+      "loss": 1.3854,
+      "step": 125745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.762016216652641e-06,
+      "loss": 1.2404,
+      "step": 125750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.761301256368687e-06,
+      "loss": 1.5371,
+      "step": 125755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.760586314579443e-06,
+      "loss": 1.3791,
+      "step": 125760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7598713912889945e-06,
+      "loss": 1.4617,
+      "step": 125765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.759156486501426e-06,
+      "loss": 1.6359,
+      "step": 125770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.758441600220813e-06,
+      "loss": 1.1589,
+      "step": 125775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.757726732451244e-06,
+      "loss": 1.2431,
+      "step": 125780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7570118831968e-06,
+      "loss": 1.6092,
+      "step": 125785
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7562970524615626e-06,
+      "loss": 1.1992,
+      "step": 125790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.755582240249612e-06,
+      "loss": 1.093,
+      "step": 125795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.754867446565032e-06,
+      "loss": 1.2341,
+      "step": 125800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.754152671411907e-06,
+      "loss": 1.3809,
+      "step": 125805
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.753437914794312e-06,
+      "loss": 1.2445,
+      "step": 125810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.752723176716334e-06,
+      "loss": 1.8023,
+      "step": 125815
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.752008457182053e-06,
+      "loss": 1.1335,
+      "step": 125820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.751293756195549e-06,
+      "loss": 1.2649,
+      "step": 125825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.75057907376091e-06,
+      "loss": 1.4736,
+      "step": 125830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.74986440988221e-06,
+      "loss": 1.0347,
+      "step": 125835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.749149764563529e-06,
+      "loss": 1.039,
+      "step": 125840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.748435137808955e-06,
+      "loss": 1.0965,
+      "step": 125845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.747720529622566e-06,
+      "loss": 1.4353,
+      "step": 125850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.747005940008437e-06,
+      "loss": 1.2322,
+      "step": 125855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.746291368970659e-06,
+      "loss": 1.055,
+      "step": 125860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.745576816513306e-06,
+      "loss": 1.2185,
+      "step": 125865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.744862282640461e-06,
+      "loss": 1.2697,
+      "step": 125870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.744147767356198e-06,
+      "loss": 1.4864,
+      "step": 125875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.743433270664605e-06,
+      "loss": 1.3346,
+      "step": 125880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.742718792569764e-06,
+      "loss": 1.1919,
+      "step": 125885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.742004333075745e-06,
+      "loss": 0.966,
+      "step": 125890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7412898921866356e-06,
+      "loss": 1.5028,
+      "step": 125895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.740575469906515e-06,
+      "loss": 1.1749,
+      "step": 125900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.739861066239456e-06,
+      "loss": 1.1696,
+      "step": 125905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.73914668118955e-06,
+      "loss": 0.7892,
+      "step": 125910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7384323147608675e-06,
+      "loss": 1.2813,
+      "step": 125915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.73771796695749e-06,
+      "loss": 1.0767,
+      "step": 125920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.737003637783502e-06,
+      "loss": 0.8327,
+      "step": 125925
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.736289327242974e-06,
+      "loss": 1.1769,
+      "step": 125930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.735575035339993e-06,
+      "loss": 1.3596,
+      "step": 125935
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.734860762078628e-06,
+      "loss": 1.3528,
+      "step": 125940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7341465074629694e-06,
+      "loss": 1.3085,
+      "step": 125945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.733432271497089e-06,
+      "loss": 1.1591,
+      "step": 125950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.732718054185065e-06,
+      "loss": 1.578,
+      "step": 125955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.732003855530984e-06,
+      "loss": 1.0917,
+      "step": 125960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.731289675538916e-06,
+      "loss": 1.1031,
+      "step": 125965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.73057551421294e-06,
+      "loss": 1.1029,
+      "step": 125970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.729861371557138e-06,
+      "loss": 1.2888,
+      "step": 125975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.729147247575588e-06,
+      "loss": 1.2532,
+      "step": 125980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.728433142272363e-06,
+      "loss": 1.0908,
+      "step": 125985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7277190556515455e-06,
+      "loss": 1.3028,
+      "step": 125990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.727004987717216e-06,
+      "loss": 1.169,
+      "step": 125995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.726290938473441e-06,
+      "loss": 1.0876,
+      "step": 126000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.725576907924312e-06,
+      "loss": 1.2227,
+      "step": 126005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.724862896073898e-06,
+      "loss": 1.0641,
+      "step": 126010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.72414890292628e-06,
+      "loss": 1.1202,
+      "step": 126015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.723434928485531e-06,
+      "loss": 1.1708,
+      "step": 126020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.72272097275573e-06,
+      "loss": 1.0615,
+      "step": 126025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.722007035740957e-06,
+      "loss": 1.108,
+      "step": 126030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.721293117445284e-06,
+      "loss": 1.0771,
+      "step": 126035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7205792178727954e-06,
+      "loss": 1.084,
+      "step": 126040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.719865337027562e-06,
+      "loss": 1.0152,
+      "step": 126045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.719151474913659e-06,
+      "loss": 1.2937,
+      "step": 126050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.71843763153517e-06,
+      "loss": 1.2292,
+      "step": 126055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7177238068961645e-06,
+      "loss": 1.4753,
+      "step": 126060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.71701000100072e-06,
+      "loss": 1.0322,
+      "step": 126065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.716296213852916e-06,
+      "loss": 1.4649,
+      "step": 126070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.715582445456829e-06,
+      "loss": 1.0661,
+      "step": 126075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.71486869581653e-06,
+      "loss": 1.5857,
+      "step": 126080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.714154964936097e-06,
+      "loss": 1.0671,
+      "step": 126085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7134412528196095e-06,
+      "loss": 1.6687,
+      "step": 126090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.712727559471137e-06,
+      "loss": 0.9704,
+      "step": 126095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.712013884894758e-06,
+      "loss": 1.1284,
+      "step": 126100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.71130022909455e-06,
+      "loss": 1.1378,
+      "step": 126105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.710586592074588e-06,
+      "loss": 1.2381,
+      "step": 126110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.709872973838942e-06,
+      "loss": 1.4168,
+      "step": 126115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.709159374391693e-06,
+      "loss": 0.9127,
+      "step": 126120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.708445793736913e-06,
+      "loss": 1.1499,
+      "step": 126125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.707732231878676e-06,
+      "loss": 1.0527,
+      "step": 126130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.707018688821065e-06,
+      "loss": 1.1939,
+      "step": 126135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.706305164568144e-06,
+      "loss": 1.219,
+      "step": 126140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.705591659123991e-06,
+      "loss": 1.1533,
+      "step": 126145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.704878172492686e-06,
+      "loss": 1.4784,
+      "step": 126150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7041647046782955e-06,
+      "loss": 1.0575,
+      "step": 126155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.703451255684898e-06,
+      "loss": 1.3849,
+      "step": 126160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.702737825516565e-06,
+      "loss": 1.3325,
+      "step": 126165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.702024414177377e-06,
+      "loss": 1.0939,
+      "step": 126170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.7013110216714e-06,
+      "loss": 1.4855,
+      "step": 126175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.700597648002711e-06,
+      "loss": 1.1411,
+      "step": 126180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.699884293175385e-06,
+      "loss": 1.2881,
+      "step": 126185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.699170957193497e-06,
+      "loss": 1.0341,
+      "step": 126190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.698457640061113e-06,
+      "loss": 0.9684,
+      "step": 126195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6977443417823155e-06,
+      "loss": 1.318,
+      "step": 126200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.697031062361175e-06,
+      "loss": 1.4583,
+      "step": 126205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.696317801801759e-06,
+      "loss": 1.0512,
+      "step": 126210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.695604560108148e-06,
+      "loss": 1.1003,
+      "step": 126215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.694891337284413e-06,
+      "loss": 1.0259,
+      "step": 126220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.694178133334626e-06,
+      "loss": 1.2056,
+      "step": 126225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.693464948262858e-06,
+      "loss": 1.272,
+      "step": 126230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.692751782073183e-06,
+      "loss": 1.3702,
+      "step": 126235
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.692038634769678e-06,
+      "loss": 1.0791,
+      "step": 126240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.691325506356405e-06,
+      "loss": 1.1758,
+      "step": 126245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.69061239683745e-06,
+      "loss": 1.3653,
+      "step": 126250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.689899306216876e-06,
+      "loss": 1.284,
+      "step": 126255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.689186234498754e-06,
+      "loss": 1.4633,
+      "step": 126260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.688473181687165e-06,
+      "loss": 1.152,
+      "step": 126265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.687760147786173e-06,
+      "loss": 1.3398,
+      "step": 126270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.687047132799849e-06,
+      "loss": 1.2891,
+      "step": 126275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.686334136732271e-06,
+      "loss": 1.1545,
+      "step": 126280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.685621159587507e-06,
+      "loss": 1.0691,
+      "step": 126285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.684908201369626e-06,
+      "loss": 1.1543,
+      "step": 126290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.684195262082705e-06,
+      "loss": 1.209,
+      "step": 126295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.683482341730815e-06,
+      "loss": 1.1864,
+      "step": 126300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.682769440318022e-06,
+      "loss": 1.4802,
+      "step": 126305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.682056557848397e-06,
+      "loss": 0.9846,
+      "step": 126310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6813436943260156e-06,
+      "loss": 1.231,
+      "step": 126315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.68063084975495e-06,
+      "loss": 1.5437,
+      "step": 126320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.679918024139263e-06,
+      "loss": 1.362,
+      "step": 126325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.67920521748303e-06,
+      "loss": 1.2999,
+      "step": 126330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.678492429790325e-06,
+      "loss": 1.0408,
+      "step": 126335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.677779661065209e-06,
+      "loss": 1.1273,
+      "step": 126340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.677066911311762e-06,
+      "loss": 1.0331,
+      "step": 126345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.67635418053405e-06,
+      "loss": 1.1341,
+      "step": 126350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.67564146873614e-06,
+      "loss": 1.2359,
+      "step": 126355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6749287759221115e-06,
+      "loss": 0.9582,
+      "step": 126360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.674216102096025e-06,
+      "loss": 0.9755,
+      "step": 126365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6735034472619535e-06,
+      "loss": 1.1546,
+      "step": 126370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.672790811423963e-06,
+      "loss": 1.1423,
+      "step": 126375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.672078194586131e-06,
+      "loss": 1.357,
+      "step": 126380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.671365596752521e-06,
+      "loss": 1.0917,
+      "step": 126385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6706530179272026e-06,
+      "loss": 1.3691,
+      "step": 126390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.669940458114249e-06,
+      "loss": 1.1729,
+      "step": 126395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6692279173177246e-06,
+      "loss": 1.1963,
+      "step": 126400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.668515395541698e-06,
+      "loss": 1.0519,
+      "step": 126405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.667802892790243e-06,
+      "loss": 0.9963,
+      "step": 126410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.667090409067427e-06,
+      "loss": 1.2558,
+      "step": 126415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.666377944377313e-06,
+      "loss": 1.1673,
+      "step": 126420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.665665498723978e-06,
+      "loss": 1.2777,
+      "step": 126425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.664953072111485e-06,
+      "loss": 1.1153,
+      "step": 126430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.664240664543902e-06,
+      "loss": 1.2006,
+      "step": 126435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.663528276025303e-06,
+      "loss": 1.4173,
+      "step": 126440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.66281590655975e-06,
+      "loss": 1.3132,
+      "step": 126445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.662103556151315e-06,
+      "loss": 1.1799,
+      "step": 126450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.661391224804059e-06,
+      "loss": 1.4211,
+      "step": 126455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.660678912522058e-06,
+      "loss": 1.1492,
+      "step": 126460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.659966619309377e-06,
+      "loss": 1.1679,
+      "step": 126465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.65925434517008e-06,
+      "loss": 1.3077,
+      "step": 126470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.658542090108242e-06,
+      "loss": 1.3266,
+      "step": 126475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.657829854127923e-06,
+      "loss": 1.1374,
+      "step": 126480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.657117637233192e-06,
+      "loss": 1.0152,
+      "step": 126485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.656405439428119e-06,
+      "loss": 1.1364,
+      "step": 126490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.655693260716771e-06,
+      "loss": 1.349,
+      "step": 126495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.65498110110321e-06,
+      "loss": 1.0837,
+      "step": 126500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.654268960591508e-06,
+      "loss": 1.1318,
+      "step": 126505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6535568391857306e-06,
+      "loss": 1.57,
+      "step": 126510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.652844736889941e-06,
+      "loss": 1.2254,
+      "step": 126515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6521326537082085e-06,
+      "loss": 1.2216,
+      "step": 126520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6514205896446e-06,
+      "loss": 1.048,
+      "step": 126525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.650708544703184e-06,
+      "loss": 1.503,
+      "step": 126530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6499965188880185e-06,
+      "loss": 1.3521,
+      "step": 126535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6492845122031775e-06,
+      "loss": 1.2802,
+      "step": 126540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.648572524652725e-06,
+      "loss": 1.4447,
+      "step": 126545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.647860556240721e-06,
+      "loss": 1.1275,
+      "step": 126550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.647148606971241e-06,
+      "loss": 1.3789,
+      "step": 126555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.646436676848344e-06,
+      "loss": 1.1542,
+      "step": 126560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.645724765876097e-06,
+      "loss": 1.2201,
+      "step": 126565
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.645012874058569e-06,
+      "loss": 1.3173,
+      "step": 126570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.64430100139982e-06,
+      "loss": 1.2637,
+      "step": 126575
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.643589147903914e-06,
+      "loss": 1.38,
+      "step": 126580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.642877313574922e-06,
+      "loss": 1.2091,
+      "step": 126585
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6421654984169085e-06,
+      "loss": 1.0023,
+      "step": 126590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.641453702433933e-06,
+      "loss": 1.1793,
+      "step": 126595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6407419256300605e-06,
+      "loss": 1.0844,
+      "step": 126600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.640030168009365e-06,
+      "loss": 1.1642,
+      "step": 126605
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6393184295759e-06,
+      "loss": 1.255,
+      "step": 126610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.638606710333735e-06,
+      "loss": 1.2975,
+      "step": 126615
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.637895010286934e-06,
+      "loss": 1.2219,
+      "step": 126620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.637183329439562e-06,
+      "loss": 1.7695,
+      "step": 126625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.636471667795679e-06,
+      "loss": 1.0157,
+      "step": 126630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.635760025359355e-06,
+      "loss": 1.3238,
+      "step": 126635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.63504840213465e-06,
+      "loss": 1.3214,
+      "step": 126640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.634336798125624e-06,
+      "loss": 1.1717,
+      "step": 126645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.633625213336351e-06,
+      "loss": 1.1218,
+      "step": 126650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.632913647770887e-06,
+      "loss": 1.2272,
+      "step": 126655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6322021014332935e-06,
+      "loss": 1.1035,
+      "step": 126660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.631490574327644e-06,
+      "loss": 1.1143,
+      "step": 126665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.630779066457991e-06,
+      "loss": 1.2285,
+      "step": 126670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.630067577828405e-06,
+      "loss": 1.1739,
+      "step": 126675
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.629356108442938e-06,
+      "loss": 1.0235,
+      "step": 126680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6286446583056675e-06,
+      "loss": 1.3009,
+      "step": 126685
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.627933227420648e-06,
+      "loss": 1.3034,
+      "step": 126690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.62722181579194e-06,
+      "loss": 1.2783,
+      "step": 126695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.626510423423614e-06,
+      "loss": 1.1649,
+      "step": 126700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.625799050319727e-06,
+      "loss": 1.1122,
+      "step": 126705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.625087696484339e-06,
+      "loss": 1.0276,
+      "step": 126710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.624376361921518e-06,
+      "loss": 1.289,
+      "step": 126715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.623665046635326e-06,
+      "loss": 1.146,
+      "step": 126720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6229537506298165e-06,
+      "loss": 1.4053,
+      "step": 126725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.622242473909062e-06,
+      "loss": 1.4384,
+      "step": 126730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.62153121647712e-06,
+      "loss": 1.1646,
+      "step": 126735
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.620819978338051e-06,
+      "loss": 1.4825,
+      "step": 126740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6201087594959146e-06,
+      "loss": 1.3335,
+      "step": 126745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.619397559954776e-06,
+      "loss": 1.2731,
+      "step": 126750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6186863797186995e-06,
+      "loss": 1.3746,
+      "step": 126755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.617975218791739e-06,
+      "loss": 1.3569,
+      "step": 126760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.617264077177959e-06,
+      "loss": 1.6723,
+      "step": 126765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.616552954881421e-06,
+      "loss": 1.1901,
+      "step": 126770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.615841851906184e-06,
+      "loss": 1.3296,
+      "step": 126775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.615130768256315e-06,
+      "loss": 1.2419,
+      "step": 126780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.614419703935866e-06,
+      "loss": 1.3046,
+      "step": 126785
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.613708658948901e-06,
+      "loss": 1.0385,
+      "step": 126790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.612997633299485e-06,
+      "loss": 1.369,
+      "step": 126795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.612286626991671e-06,
+      "loss": 1.374,
+      "step": 126800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.611575640029521e-06,
+      "loss": 1.1122,
+      "step": 126805
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.610864672417098e-06,
+      "loss": 1.274,
+      "step": 126810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.610153724158464e-06,
+      "loss": 1.5108,
+      "step": 126815
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.609442795257673e-06,
+      "loss": 1.2149,
+      "step": 126820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.608731885718783e-06,
+      "loss": 1.4934,
+      "step": 126825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.608020995545862e-06,
+      "loss": 1.1626,
+      "step": 126830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6073101247429675e-06,
+      "loss": 1.2344,
+      "step": 126835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.606599273314152e-06,
+      "loss": 1.1892,
+      "step": 126840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.605888441263483e-06,
+      "loss": 1.3311,
+      "step": 126845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.605177628595018e-06,
+      "loss": 1.0913,
+      "step": 126850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.60446683531281e-06,
+      "loss": 1.2349,
+      "step": 126855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.603756061420925e-06,
+      "loss": 0.9817,
+      "step": 126860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.603045306923421e-06,
+      "loss": 1.5638,
+      "step": 126865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6023345718243515e-06,
+      "loss": 1.3068,
+      "step": 126870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.601623856127783e-06,
+      "loss": 1.2812,
+      "step": 126875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6009131598377695e-06,
+      "loss": 1.2776,
+      "step": 126880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.600202482958372e-06,
+      "loss": 1.0031,
+      "step": 126885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.59949182549364e-06,
+      "loss": 1.3235,
+      "step": 126890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.598781187447645e-06,
+      "loss": 1.0549,
+      "step": 126895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.598070568824437e-06,
+      "loss": 1.5225,
+      "step": 126900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.597359969628075e-06,
+      "loss": 1.1324,
+      "step": 126905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.596649389862619e-06,
+      "loss": 1.5442,
+      "step": 126910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.595938829532125e-06,
+      "loss": 1.2031,
+      "step": 126915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.595228288640648e-06,
+      "loss": 1.0626,
+      "step": 126920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.594517767192252e-06,
+      "loss": 1.3964,
+      "step": 126925
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.593807265190992e-06,
+      "loss": 1.5178,
+      "step": 126930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.593096782640922e-06,
+      "loss": 1.0268,
+      "step": 126935
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.592386319546103e-06,
+      "loss": 1.1176,
+      "step": 126940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.591675875910594e-06,
+      "loss": 1.1483,
+      "step": 126945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.590965451738441e-06,
+      "loss": 1.2858,
+      "step": 126950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.590255047033716e-06,
+      "loss": 1.2741,
+      "step": 126955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.589544661800467e-06,
+      "loss": 1.0928,
+      "step": 126960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.588834296042754e-06,
+      "loss": 1.6363,
+      "step": 126965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.588123949764627e-06,
+      "loss": 1.3168,
+      "step": 126970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.58741362297015e-06,
+      "loss": 1.1596,
+      "step": 126975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.586703315663379e-06,
+      "loss": 1.4375,
+      "step": 126980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.585993027848361e-06,
+      "loss": 1.2146,
+      "step": 126985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.585282759529166e-06,
+      "loss": 1.2747,
+      "step": 126990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.584572510709842e-06,
+      "loss": 0.9964,
+      "step": 126995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.583862281394441e-06,
+      "loss": 1.0857,
+      "step": 127000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.583152071587029e-06,
+      "loss": 1.014,
+      "step": 127005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.582441881291654e-06,
+      "loss": 1.2742,
+      "step": 127010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.581731710512374e-06,
+      "loss": 1.4722,
+      "step": 127015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.581021559253244e-06,
+      "loss": 1.4359,
+      "step": 127020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.580311427518324e-06,
+      "loss": 1.4758,
+      "step": 127025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.579601315311662e-06,
+      "loss": 0.9644,
+      "step": 127030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.578891222637315e-06,
+      "loss": 1.4272,
+      "step": 127035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.578181149499342e-06,
+      "loss": 1.612,
+      "step": 127040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.577471095901793e-06,
+      "loss": 1.321,
+      "step": 127045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.576761061848724e-06,
+      "loss": 1.1492,
+      "step": 127050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.576051047344192e-06,
+      "loss": 1.1818,
+      "step": 127055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.575341052392252e-06,
+      "loss": 1.2078,
+      "step": 127060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.574631076996952e-06,
+      "loss": 1.3235,
+      "step": 127065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.573921121162353e-06,
+      "loss": 1.0675,
+      "step": 127070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5732111848925075e-06,
+      "loss": 1.1874,
+      "step": 127075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.572501268191468e-06,
+      "loss": 1.0654,
+      "step": 127080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.571791371063293e-06,
+      "loss": 1.174,
+      "step": 127085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.571081493512031e-06,
+      "loss": 1.3801,
+      "step": 127090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5703716355417355e-06,
+      "loss": 1.1625,
+      "step": 127095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.569661797156468e-06,
+      "loss": 1.0006,
+      "step": 127100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.568951978360274e-06,
+      "loss": 1.1769,
+      "step": 127105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.568242179157209e-06,
+      "loss": 0.8687,
+      "step": 127110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.567532399551325e-06,
+      "loss": 0.9685,
+      "step": 127115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.566822639546682e-06,
+      "loss": 0.984,
+      "step": 127120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.566112899147325e-06,
+      "loss": 1.401,
+      "step": 127125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.56540317835731e-06,
+      "loss": 1.3576,
+      "step": 127130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.56469347718069e-06,
+      "loss": 1.0924,
+      "step": 127135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.563983795621522e-06,
+      "loss": 1.1146,
+      "step": 127140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.56327413368385e-06,
+      "loss": 1.2559,
+      "step": 127145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.562564491371732e-06,
+      "loss": 1.4049,
+      "step": 127150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.561854868689224e-06,
+      "loss": 1.435,
+      "step": 127155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.561145265640367e-06,
+      "loss": 1.4254,
+      "step": 127160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5604356822292244e-06,
+      "loss": 1.2704,
+      "step": 127165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.559726118459844e-06,
+      "loss": 1.2662,
+      "step": 127170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5590165743362784e-06,
+      "loss": 1.2281,
+      "step": 127175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.558307049862576e-06,
+      "loss": 1.5925,
+      "step": 127180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.557597545042792e-06,
+      "loss": 1.4442,
+      "step": 127185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.55688805988098e-06,
+      "loss": 1.0546,
+      "step": 127190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.556178594381184e-06,
+      "loss": 1.5365,
+      "step": 127195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.555469148547467e-06,
+      "loss": 1.3232,
+      "step": 127200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.55475972238387e-06,
+      "loss": 1.3466,
+      "step": 127205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.554050315894447e-06,
+      "loss": 1.4689,
+      "step": 127210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.553340929083254e-06,
+      "loss": 1.2414,
+      "step": 127215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.552631561954335e-06,
+      "loss": 1.3357,
+      "step": 127220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.551922214511744e-06,
+      "loss": 1.1675,
+      "step": 127225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.551212886759532e-06,
+      "loss": 1.2032,
+      "step": 127230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.550503578701753e-06,
+      "loss": 1.4799,
+      "step": 127235
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5497942903424484e-06,
+      "loss": 1.4346,
+      "step": 127240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.549085021685677e-06,
+      "loss": 1.1711,
+      "step": 127245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.548375772735488e-06,
+      "loss": 1.0879,
+      "step": 127250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.547666543495927e-06,
+      "loss": 1.21,
+      "step": 127255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.546957333971046e-06,
+      "loss": 1.0558,
+      "step": 127260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.546248144164897e-06,
+      "loss": 1.1557,
+      "step": 127265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5455389740815316e-06,
+      "loss": 1.1867,
+      "step": 127270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5448298237249915e-06,
+      "loss": 1.3838,
+      "step": 127275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.544120693099334e-06,
+      "loss": 1.1517,
+      "step": 127280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.54341158220861e-06,
+      "loss": 1.1451,
+      "step": 127285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.542702491056856e-06,
+      "loss": 1.1468,
+      "step": 127290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.541993419648138e-06,
+      "loss": 1.3152,
+      "step": 127295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5412843679864945e-06,
+      "loss": 0.9884,
+      "step": 127300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.540575336075976e-06,
+      "loss": 1.3292,
+      "step": 127305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.539866323920637e-06,
+      "loss": 1.13,
+      "step": 127310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.53915733152452e-06,
+      "loss": 1.2236,
+      "step": 127315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5384483588916794e-06,
+      "loss": 1.2472,
+      "step": 127320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.537739406026153e-06,
+      "loss": 1.4278,
+      "step": 127325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.537030472932004e-06,
+      "loss": 1.6901,
+      "step": 127330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.53632155961327e-06,
+      "loss": 1.1376,
+      "step": 127335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.535612666074002e-06,
+      "loss": 1.3623,
+      "step": 127340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.534903792318252e-06,
+      "loss": 1.3647,
+      "step": 127345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5341949383500644e-06,
+      "loss": 1.2065,
+      "step": 127350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.533486104173486e-06,
+      "loss": 1.2346,
+      "step": 127355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.532777289792567e-06,
+      "loss": 1.7029,
+      "step": 127360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.532068495211357e-06,
+      "loss": 1.3292,
+      "step": 127365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.531359720433896e-06,
+      "loss": 1.0799,
+      "step": 127370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.53065096546424e-06,
+      "loss": 1.0341,
+      "step": 127375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.529942230306435e-06,
+      "loss": 1.3971,
+      "step": 127380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.52923351496452e-06,
+      "loss": 1.0809,
+      "step": 127385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.528524819442554e-06,
+      "loss": 1.2693,
+      "step": 127390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.527816143744578e-06,
+      "loss": 1.3052,
+      "step": 127395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.52710748787464e-06,
+      "loss": 1.1891,
+      "step": 127400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.526398851836781e-06,
+      "loss": 1.3372,
+      "step": 127405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.525690235635057e-06,
+      "loss": 1.3205,
+      "step": 127410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.52498163927351e-06,
+      "loss": 1.1522,
+      "step": 127415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.524273062756185e-06,
+      "loss": 1.5235,
+      "step": 127420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.523564506087134e-06,
+      "loss": 1.1178,
+      "step": 127425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.522855969270396e-06,
+      "loss": 1.0351,
+      "step": 127430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5221474523100195e-06,
+      "loss": 1.252,
+      "step": 127435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5214389552100564e-06,
+      "loss": 1.3255,
+      "step": 127440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5207304779745466e-06,
+      "loss": 1.2338,
+      "step": 127445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.520022020607533e-06,
+      "loss": 1.1294,
+      "step": 127450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.519313583113067e-06,
+      "loss": 1.1437,
+      "step": 127455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.518605165495196e-06,
+      "loss": 1.0476,
+      "step": 127460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.51789676775796e-06,
+      "loss": 1.4945,
+      "step": 127465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.517188389905405e-06,
+      "loss": 1.2933,
+      "step": 127470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.516480031941579e-06,
+      "loss": 1.3469,
+      "step": 127475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.515771693870526e-06,
+      "loss": 1.2838,
+      "step": 127480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5150633756962885e-06,
+      "loss": 1.3629,
+      "step": 127485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5143550774229145e-06,
+      "loss": 1.3622,
+      "step": 127490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.51364679905445e-06,
+      "loss": 0.9578,
+      "step": 127495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.512938540594933e-06,
+      "loss": 1.4201,
+      "step": 127500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.512230302048416e-06,
+      "loss": 1.3377,
+      "step": 127505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.511522083418938e-06,
+      "loss": 1.4541,
+      "step": 127510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.510813884710545e-06,
+      "loss": 1.2133,
+      "step": 127515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.510105705927285e-06,
+      "loss": 1.1085,
+      "step": 127520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.509397547073195e-06,
+      "loss": 1.1655,
+      "step": 127525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.508689408152321e-06,
+      "loss": 1.2485,
+      "step": 127530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.507981289168711e-06,
+      "loss": 0.9696,
+      "step": 127535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.507273190126408e-06,
+      "loss": 1.1544,
+      "step": 127540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.50656511102945e-06,
+      "loss": 1.3288,
+      "step": 127545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.505857051881884e-06,
+      "loss": 1.0899,
+      "step": 127550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.505149012687757e-06,
+      "loss": 1.3887,
+      "step": 127555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.504440993451108e-06,
+      "loss": 1.1795,
+      "step": 127560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.503732994175977e-06,
+      "loss": 1.2029,
+      "step": 127565
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5030250148664145e-06,
+      "loss": 1.1345,
+      "step": 127570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.502317055526462e-06,
+      "loss": 1.2654,
+      "step": 127575
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.501609116160155e-06,
+      "loss": 0.9861,
+      "step": 127580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.500901196771545e-06,
+      "loss": 1.2728,
+      "step": 127585
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.500193297364672e-06,
+      "loss": 1.103,
+      "step": 127590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.499485417943573e-06,
+      "loss": 1.4094,
+      "step": 127595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.4987775585123e-06,
+      "loss": 1.4696,
+      "step": 127600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.498069719074888e-06,
+      "loss": 1.2327,
+      "step": 127605
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.497361899635385e-06,
+      "loss": 1.2894,
+      "step": 127610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.496654100197824e-06,
+      "loss": 1.4836,
+      "step": 127615
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.495946320766255e-06,
+      "loss": 1.0896,
+      "step": 127620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.495238561344719e-06,
+      "loss": 1.3233,
+      "step": 127625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.4945308219372506e-06,
+      "loss": 1.4305,
+      "step": 127630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.4938231025479004e-06,
+      "loss": 1.3605,
+      "step": 127635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.493115403180707e-06,
+      "loss": 1.5636,
+      "step": 127640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.492407723839706e-06,
+      "loss": 1.3889,
+      "step": 127645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.4917000645289495e-06,
+      "loss": 1.4677,
+      "step": 127650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.490992425252469e-06,
+      "loss": 1.1509,
+      "step": 127655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.490284806014309e-06,
+      "loss": 1.2453,
+      "step": 127660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.489577206818512e-06,
+      "loss": 1.3839,
+      "step": 127665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.488869627669119e-06,
+      "loss": 1.1533,
+      "step": 127670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.488162068570164e-06,
+      "loss": 1.5069,
+      "step": 127675
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.487454529525694e-06,
+      "loss": 1.3661,
+      "step": 127680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.4867470105397516e-06,
+      "loss": 1.1048,
+      "step": 127685
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.486039511616371e-06,
+      "loss": 1.1325,
+      "step": 127690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.485332032759593e-06,
+      "loss": 0.9858,
+      "step": 127695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.48462457397346e-06,
+      "loss": 1.0043,
+      "step": 127700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.483917135262014e-06,
+      "loss": 1.4524,
+      "step": 127705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.48320971662929e-06,
+      "loss": 1.3073,
+      "step": 127710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.48250231807933e-06,
+      "loss": 0.9672,
+      "step": 127715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.481794939616175e-06,
+      "loss": 1.3331,
+      "step": 127720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.48108758124386e-06,
+      "loss": 1.2225,
+      "step": 127725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.480380242966433e-06,
+      "loss": 1.1512,
+      "step": 127730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.479672924787924e-06,
+      "loss": 1.2431,
+      "step": 127735
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.478965626712375e-06,
+      "loss": 1.0175,
+      "step": 127740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.478258348743831e-06,
+      "loss": 1.1435,
+      "step": 127745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.4775510908863224e-06,
+      "loss": 1.0799,
+      "step": 127750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.476843853143892e-06,
+      "loss": 0.8416,
+      "step": 127755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.476136635520577e-06,
+      "loss": 1.0496,
+      "step": 127760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.4754294380204196e-06,
+      "loss": 1.3825,
+      "step": 127765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.474722260647455e-06,
+      "loss": 1.2367,
+      "step": 127770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.47401510340572e-06,
+      "loss": 1.6881,
+      "step": 127775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.473307966299258e-06,
+      "loss": 1.2161,
+      "step": 127780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.472600849332106e-06,
+      "loss": 1.3578,
+      "step": 127785
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.471893752508296e-06,
+      "loss": 1.3226,
+      "step": 127790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.471186675831872e-06,
+      "loss": 1.2236,
+      "step": 127795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.470479619306873e-06,
+      "loss": 1.1599,
+      "step": 127800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.4697725829373285e-06,
+      "loss": 1.2783,
+      "step": 127805
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.469065566727284e-06,
+      "loss": 0.8912,
+      "step": 127810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.468358570680774e-06,
+      "loss": 1.2788,
+      "step": 127815
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.467651594801834e-06,
+      "loss": 1.3987,
+      "step": 127820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.466944639094508e-06,
+      "loss": 1.1889,
+      "step": 127825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.4662377035628255e-06,
+      "loss": 0.9983,
+      "step": 127830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.465530788210829e-06,
+      "loss": 1.073,
+      "step": 127835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.464823893042549e-06,
+      "loss": 1.4872,
+      "step": 127840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.464117018062026e-06,
+      "loss": 1.0358,
+      "step": 127845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.463410163273298e-06,
+      "loss": 1.1319,
+      "step": 127850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.462703328680399e-06,
+      "loss": 1.3487,
+      "step": 127855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4619965142873695e-06,
+      "loss": 0.9228,
+      "step": 127860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.461289720098241e-06,
+      "loss": 1.3972,
+      "step": 127865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.460582946117049e-06,
+      "loss": 1.2308,
+      "step": 127870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4598761923478356e-06,
+      "loss": 1.3602,
+      "step": 127875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.459169458794634e-06,
+      "loss": 1.1493,
+      "step": 127880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.458462745461475e-06,
+      "loss": 1.2741,
+      "step": 127885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.457756052352402e-06,
+      "loss": 1.3339,
+      "step": 127890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.457049379471447e-06,
+      "loss": 1.1801,
+      "step": 127895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.456342726822646e-06,
+      "loss": 1.2261,
+      "step": 127900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4556360944100295e-06,
+      "loss": 0.9624,
+      "step": 127905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.454929482237641e-06,
+      "loss": 1.4268,
+      "step": 127910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.454222890309514e-06,
+      "loss": 1.193,
+      "step": 127915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.453516318629676e-06,
+      "loss": 1.4562,
+      "step": 127920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4528097672021705e-06,
+      "loss": 1.3921,
+      "step": 127925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.452103236031031e-06,
+      "loss": 1.333,
+      "step": 127930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4513967251202825e-06,
+      "loss": 1.2948,
+      "step": 127935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.450690234473975e-06,
+      "loss": 1.2945,
+      "step": 127940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.449983764096132e-06,
+      "loss": 1.4395,
+      "step": 127945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.449277313990792e-06,
+      "loss": 1.1197,
+      "step": 127950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.44857088416199e-06,
+      "loss": 1.0288,
+      "step": 127955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.447864474613756e-06,
+      "loss": 1.3675,
+      "step": 127960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4471580853501255e-06,
+      "loss": 1.1449,
+      "step": 127965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.446451716375135e-06,
+      "loss": 1.2305,
+      "step": 127970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4457453676928195e-06,
+      "loss": 1.2121,
+      "step": 127975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.445039039307206e-06,
+      "loss": 1.4231,
+      "step": 127980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.444332731222329e-06,
+      "loss": 1.1988,
+      "step": 127985
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.44362644344223e-06,
+      "loss": 1.1832,
+      "step": 127990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.442920175970935e-06,
+      "loss": 1.1933,
+      "step": 127995
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.442213928812476e-06,
+      "loss": 1.166,
+      "step": 128000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.441507701970892e-06,
+      "loss": 1.954,
+      "step": 128005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.440801495450215e-06,
+      "loss": 1.5111,
+      "step": 128010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.44009530925447e-06,
+      "loss": 1.5228,
+      "step": 128015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4393891433876995e-06,
+      "loss": 1.4562,
+      "step": 128020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.438682997853934e-06,
+      "loss": 1.262,
+      "step": 128025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.437976872657197e-06,
+      "loss": 1.1984,
+      "step": 128030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4372707678015334e-06,
+      "loss": 1.3107,
+      "step": 128035
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.436564683290969e-06,
+      "loss": 1.3467,
+      "step": 128040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.435858619129538e-06,
+      "loss": 1.1383,
+      "step": 128045
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.435152575321269e-06,
+      "loss": 1.2661,
+      "step": 128050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.434446551870197e-06,
+      "loss": 1.0566,
+      "step": 128055
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.433740548780353e-06,
+      "loss": 1.4682,
+      "step": 128060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.433034566055767e-06,
+      "loss": 1.3293,
+      "step": 128065
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.432328603700475e-06,
+      "loss": 1.0313,
+      "step": 128070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.431622661718504e-06,
+      "loss": 1.6553,
+      "step": 128075
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.430916740113886e-06,
+      "loss": 1.2377,
+      "step": 128080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4302108388906556e-06,
+      "loss": 1.1117,
+      "step": 128085
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.42950495805284e-06,
+      "loss": 0.9648,
+      "step": 128090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4287990976044704e-06,
+      "loss": 1.4013,
+      "step": 128095
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4280932575495795e-06,
+      "loss": 1.2222,
+      "step": 128100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.427387437892198e-06,
+      "loss": 1.4604,
+      "step": 128105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.426681638636354e-06,
+      "loss": 1.3273,
+      "step": 128110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.425975859786081e-06,
+      "loss": 0.9758,
+      "step": 128115
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.425270101345409e-06,
+      "loss": 1.0914,
+      "step": 128120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.424564363318368e-06,
+      "loss": 1.2201,
+      "step": 128125
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.423858645708984e-06,
+      "loss": 1.1316,
+      "step": 128130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.423152948521292e-06,
+      "loss": 1.2529,
+      "step": 128135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.422447271759323e-06,
+      "loss": 1.0931,
+      "step": 128140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.421741615427102e-06,
+      "loss": 1.0864,
+      "step": 128145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4210359795286606e-06,
+      "loss": 1.1972,
+      "step": 128150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.420330364068029e-06,
+      "loss": 1.0694,
+      "step": 128155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.419624769049235e-06,
+      "loss": 0.9518,
+      "step": 128160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.418919194476313e-06,
+      "loss": 0.9616,
+      "step": 128165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.418213640353287e-06,
+      "loss": 1.4598,
+      "step": 128170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.417508106684185e-06,
+      "loss": 1.5565,
+      "step": 128175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4168025934730405e-06,
+      "loss": 1.499,
+      "step": 128180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.416097100723884e-06,
+      "loss": 1.3196,
+      "step": 128185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.415391628440737e-06,
+      "loss": 1.2972,
+      "step": 128190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.414686176627629e-06,
+      "loss": 1.1614,
+      "step": 128195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4139807452885975e-06,
+      "loss": 1.281,
+      "step": 128200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.413275334427661e-06,
+      "loss": 1.3484,
+      "step": 128205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.41256994404885e-06,
+      "loss": 0.8751,
+      "step": 128210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.411864574156197e-06,
+      "loss": 1.198,
+      "step": 128215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.411159224753727e-06,
+      "loss": 1.165,
+      "step": 128220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.410453895845465e-06,
+      "loss": 1.2527,
+      "step": 128225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.409748587435445e-06,
+      "loss": 1.4699,
+      "step": 128230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.409043299527693e-06,
+      "loss": 1.0508,
+      "step": 128235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.408338032126229e-06,
+      "loss": 1.1611,
+      "step": 128240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.407632785235094e-06,
+      "loss": 1.0582,
+      "step": 128245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.406927558858304e-06,
+      "loss": 0.9984,
+      "step": 128250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.406222352999889e-06,
+      "loss": 1.0914,
+      "step": 128255
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.405517167663882e-06,
+      "loss": 1.0884,
+      "step": 128260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.4048120028543015e-06,
+      "loss": 1.1606,
+      "step": 128265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.404106858575181e-06,
+      "loss": 1.2517,
+      "step": 128270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.403401734830539e-06,
+      "loss": 1.1114,
+      "step": 128275
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.402696631624412e-06,
+      "loss": 1.1042,
+      "step": 128280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.401991548960822e-06,
+      "loss": 1.2832,
+      "step": 128285
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.401286486843792e-06,
+      "loss": 0.9883,
+      "step": 128290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.400581445277356e-06,
+      "loss": 1.2582,
+      "step": 128295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.399876424265534e-06,
+      "loss": 1.2039,
+      "step": 128300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.399171423812354e-06,
+      "loss": 1.019,
+      "step": 128305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3984664439218405e-06,
+      "loss": 1.0585,
+      "step": 128310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.397761484598024e-06,
+      "loss": 0.9627,
+      "step": 128315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.397056545844924e-06,
+      "loss": 1.3075,
+      "step": 128320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.39635162766657e-06,
+      "loss": 1.456,
+      "step": 128325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3956467300669886e-06,
+      "loss": 1.1793,
+      "step": 128330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.394941853050201e-06,
+      "loss": 1.2217,
+      "step": 128335
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.394236996620231e-06,
+      "loss": 1.1846,
+      "step": 128340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3935321607811105e-06,
+      "loss": 1.3552,
+      "step": 128345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.392827345536864e-06,
+      "loss": 1.2083,
+      "step": 128350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.392122550891509e-06,
+      "loss": 1.1442,
+      "step": 128355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.391417776849076e-06,
+      "loss": 1.2603,
+      "step": 128360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.390713023413591e-06,
+      "loss": 1.1573,
+      "step": 128365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.390008290589069e-06,
+      "loss": 1.142,
+      "step": 128370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.389303578379548e-06,
+      "loss": 1.1344,
+      "step": 128375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.388598886789044e-06,
+      "loss": 1.0055,
+      "step": 128380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.38789421582158e-06,
+      "loss": 1.2221,
+      "step": 128385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3871895654811865e-06,
+      "loss": 0.9166,
+      "step": 128390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3864849357718815e-06,
+      "loss": 1.1151,
+      "step": 128395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.38578032669769e-06,
+      "loss": 1.3297,
+      "step": 128400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.385075738262639e-06,
+      "loss": 1.1119,
+      "step": 128405
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.384371170470751e-06,
+      "loss": 1.5207,
+      "step": 128410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.383666623326047e-06,
+      "loss": 1.1531,
+      "step": 128415
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.382962096832549e-06,
+      "loss": 1.0959,
+      "step": 128420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3822575909942875e-06,
+      "loss": 1.1574,
+      "step": 128425
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.381553105815279e-06,
+      "loss": 1.2983,
+      "step": 128430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.380848641299545e-06,
+      "loss": 1.3415,
+      "step": 128435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.380144197451115e-06,
+      "loss": 1.1261,
+      "step": 128440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.37943977427401e-06,
+      "loss": 1.2506,
+      "step": 128445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3787353717722485e-06,
+      "loss": 1.0755,
+      "step": 128450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.378030989949857e-06,
+      "loss": 1.106,
+      "step": 128455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.377326628810856e-06,
+      "loss": 1.2573,
+      "step": 128460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.376622288359267e-06,
+      "loss": 1.3469,
+      "step": 128465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.375917968599118e-06,
+      "loss": 1.2777,
+      "step": 128470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3752136695344235e-06,
+      "loss": 1.1578,
+      "step": 128475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.374509391169212e-06,
+      "loss": 1.2268,
+      "step": 128480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.373805133507497e-06,
+      "loss": 1.0364,
+      "step": 128485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.373100896553307e-06,
+      "loss": 1.4438,
+      "step": 128490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.372396680310662e-06,
+      "loss": 1.4428,
+      "step": 128495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.371692484783581e-06,
+      "loss": 1.4029,
+      "step": 128500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.370988309976091e-06,
+      "loss": 1.4603,
+      "step": 128505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.370284155892208e-06,
+      "loss": 1.0895,
+      "step": 128510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.369580022535953e-06,
+      "loss": 1.0919,
+      "step": 128515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.36887590991135e-06,
+      "loss": 1.1239,
+      "step": 128520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.36817181802242e-06,
+      "loss": 1.0516,
+      "step": 128525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.367467746873179e-06,
+      "loss": 1.1251,
+      "step": 128530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.366763696467653e-06,
+      "loss": 1.2107,
+      "step": 128535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3660596668098625e-06,
+      "loss": 1.1542,
+      "step": 128540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.365355657903822e-06,
+      "loss": 1.3117,
+      "step": 128545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.364651669753556e-06,
+      "loss": 1.2377,
+      "step": 128550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.363947702363087e-06,
+      "loss": 1.5393,
+      "step": 128555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.363243755736432e-06,
+      "loss": 1.2034,
+      "step": 128560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.36253982987761e-06,
+      "loss": 1.2149,
+      "step": 128565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.361835924790641e-06,
+      "loss": 1.0274,
+      "step": 128570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3611320404795495e-06,
+      "loss": 1.0346,
+      "step": 128575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.360428176948343e-06,
+      "loss": 1.348,
+      "step": 128580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.359724334201058e-06,
+      "loss": 1.1912,
+      "step": 128585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.359020512241701e-06,
+      "loss": 1.1626,
+      "step": 128590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.358316711074294e-06,
+      "loss": 1.2851,
+      "step": 128595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.357612930702861e-06,
+      "loss": 1.2217,
+      "step": 128600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.356909171131416e-06,
+      "loss": 1.2487,
+      "step": 128605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.356205432363976e-06,
+      "loss": 1.2578,
+      "step": 128610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.355501714404565e-06,
+      "loss": 1.2588,
+      "step": 128615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.354798017257202e-06,
+      "loss": 1.0393,
+      "step": 128620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3540943409259e-06,
+      "loss": 1.3249,
+      "step": 128625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3533906854146785e-06,
+      "loss": 1.3209,
+      "step": 128630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.352687050727562e-06,
+      "loss": 1.0602,
+      "step": 128635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.351983436868563e-06,
+      "loss": 1.1399,
+      "step": 128640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.351279843841697e-06,
+      "loss": 1.3274,
+      "step": 128645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.350576271650988e-06,
+      "loss": 1.2809,
+      "step": 128650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.349872720300454e-06,
+      "loss": 1.2534,
+      "step": 128655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3491691897941045e-06,
+      "loss": 1.307,
+      "step": 128660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.348465680135967e-06,
+      "loss": 1.0941,
+      "step": 128665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.347762191330053e-06,
+      "loss": 1.2183,
+      "step": 128670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.347058723380378e-06,
+      "loss": 1.4151,
+      "step": 128675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3463552762909676e-06,
+      "loss": 1.2229,
+      "step": 128680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.345651850065833e-06,
+      "loss": 1.0351,
+      "step": 128685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.344948444708989e-06,
+      "loss": 1.0464,
+      "step": 128690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.344245060224458e-06,
+      "loss": 1.0106,
+      "step": 128695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.343541696616254e-06,
+      "loss": 1.1424,
+      "step": 128700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.342838353888392e-06,
+      "loss": 0.9892,
+      "step": 128705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3421350320448875e-06,
+      "loss": 1.0926,
+      "step": 128710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.341431731089764e-06,
+      "loss": 1.4351,
+      "step": 128715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3407284510270315e-06,
+      "loss": 1.1967,
+      "step": 128720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.340025191860706e-06,
+      "loss": 1.405,
+      "step": 128725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3393219535948085e-06,
+      "loss": 1.1832,
+      "step": 128730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.338618736233351e-06,
+      "loss": 1.1132,
+      "step": 128735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.337915539780346e-06,
+      "loss": 0.9788,
+      "step": 128740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.337212364239815e-06,
+      "loss": 1.1154,
+      "step": 128745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3365092096157735e-06,
+      "loss": 1.4432,
+      "step": 128750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.335806075912232e-06,
+      "loss": 1.3689,
+      "step": 128755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.33510296313321e-06,
+      "loss": 1.2657,
+      "step": 128760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.33439987128272e-06,
+      "loss": 1.3795,
+      "step": 128765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.333696800364778e-06,
+      "loss": 1.2723,
+      "step": 128770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3329937503834024e-06,
+      "loss": 1.2261,
+      "step": 128775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.332290721342603e-06,
+      "loss": 1.2132,
+      "step": 128780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.331587713246399e-06,
+      "loss": 1.0038,
+      "step": 128785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.330884726098797e-06,
+      "loss": 1.3553,
+      "step": 128790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.330181759903819e-06,
+      "loss": 1.3523,
+      "step": 128795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.329478814665477e-06,
+      "loss": 1.6073,
+      "step": 128800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.328775890387784e-06,
+      "loss": 1.1608,
+      "step": 128805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.328072987074758e-06,
+      "loss": 1.1787,
+      "step": 128810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.327370104730408e-06,
+      "loss": 1.2741,
+      "step": 128815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.32666724335875e-06,
+      "loss": 1.1027,
+      "step": 128820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.325964402963798e-06,
+      "loss": 1.0226,
+      "step": 128825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.325261583549568e-06,
+      "loss": 1.7101,
+      "step": 128830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.324558785120067e-06,
+      "loss": 1.4597,
+      "step": 128835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.323856007679314e-06,
+      "loss": 1.3995,
+      "step": 128840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.323153251231322e-06,
+      "loss": 0.9531,
+      "step": 128845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.322450515780101e-06,
+      "loss": 1.0391,
+      "step": 128850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.321747801329663e-06,
+      "loss": 0.9785,
+      "step": 128855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.321045107884025e-06,
+      "loss": 1.1109,
+      "step": 128860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.320342435447202e-06,
+      "loss": 1.139,
+      "step": 128865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.319639784023197e-06,
+      "loss": 1.147,
+      "step": 128870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.318937153616031e-06,
+      "loss": 1.1762,
+      "step": 128875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.318234544229717e-06,
+      "loss": 1.2614,
+      "step": 128880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.317531955868257e-06,
+      "loss": 0.9295,
+      "step": 128885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.316829388535674e-06,
+      "loss": 1.1124,
+      "step": 128890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.316126842235977e-06,
+      "loss": 1.0046,
+      "step": 128895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.315424316973174e-06,
+      "loss": 1.2324,
+      "step": 128900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.314721812751284e-06,
+      "loss": 1.1781,
+      "step": 128905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.314019329574314e-06,
+      "loss": 1.2963,
+      "step": 128910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.313316867446272e-06,
+      "loss": 1.0669,
+      "step": 128915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3126144263711766e-06,
+      "loss": 1.0525,
+      "step": 128920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3119120063530385e-06,
+      "loss": 0.9217,
+      "step": 128925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.311209607395863e-06,
+      "loss": 1.238,
+      "step": 128930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.310507229503664e-06,
+      "loss": 0.9674,
+      "step": 128935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.309804872680456e-06,
+      "loss": 1.1168,
+      "step": 128940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.309102536930247e-06,
+      "loss": 1.1444,
+      "step": 128945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.308400222257044e-06,
+      "loss": 1.2329,
+      "step": 128950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.307697928664863e-06,
+      "loss": 1.1816,
+      "step": 128955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.306995656157716e-06,
+      "loss": 1.2618,
+      "step": 128960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.306293404739605e-06,
+      "loss": 1.2955,
+      "step": 128965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.305591174414548e-06,
+      "loss": 1.3979,
+      "step": 128970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.304888965186554e-06,
+      "loss": 1.5247,
+      "step": 128975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.304186777059626e-06,
+      "loss": 1.0099,
+      "step": 128980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.303484610037784e-06,
+      "loss": 1.3069,
+      "step": 128985
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.302782464125032e-06,
+      "loss": 1.3259,
+      "step": 128990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.302080339325384e-06,
+      "loss": 1.2056,
+      "step": 128995
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.301378235642841e-06,
+      "loss": 1.0961,
+      "step": 129000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.30067615308142e-06,
+      "loss": 1.1653,
+      "step": 129005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.299974091645129e-06,
+      "loss": 1.3347,
+      "step": 129010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.29927205133797e-06,
+      "loss": 1.1892,
+      "step": 129015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.298570032163964e-06,
+      "loss": 1.0897,
+      "step": 129020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.297868034127113e-06,
+      "loss": 1.2181,
+      "step": 129025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.297166057231423e-06,
+      "loss": 1.3364,
+      "step": 129030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.296464101480912e-06,
+      "loss": 1.0879,
+      "step": 129035
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2957621668795795e-06,
+      "loss": 1.3066,
+      "step": 129040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.295060253431438e-06,
+      "loss": 1.5719,
+      "step": 129045
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.294358361140494e-06,
+      "loss": 0.9903,
+      "step": 129050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.29365649001076e-06,
+      "loss": 1.1279,
+      "step": 129055
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.292954640046238e-06,
+      "loss": 1.4504,
+      "step": 129060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.292252811250939e-06,
+      "loss": 1.389,
+      "step": 129065
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.291551003628875e-06,
+      "loss": 1.5176,
+      "step": 129070
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.290849217184045e-06,
+      "loss": 1.1331,
+      "step": 129075
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.290147451920461e-06,
+      "loss": 1.3005,
+      "step": 129080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.28944570784213e-06,
+      "loss": 1.571,
+      "step": 129085
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2887439849530644e-06,
+      "loss": 1.4338,
+      "step": 129090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.288042283257261e-06,
+      "loss": 1.1389,
+      "step": 129095
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.287340602758736e-06,
+      "loss": 1.1931,
+      "step": 129100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.286638943461492e-06,
+      "loss": 1.2932,
+      "step": 129105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.285937305369534e-06,
+      "loss": 1.5662,
+      "step": 129110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.285235688486877e-06,
+      "loss": 1.1332,
+      "step": 129115
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.28453409281752e-06,
+      "loss": 1.2692,
+      "step": 129120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.283832518365469e-06,
+      "loss": 1.1782,
+      "step": 129125
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.283130965134738e-06,
+      "loss": 1.2028,
+      "step": 129130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2824294331293255e-06,
+      "loss": 1.3182,
+      "step": 129135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.281727922353241e-06,
+      "loss": 1.2634,
+      "step": 129140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.281026432810488e-06,
+      "loss": 0.9852,
+      "step": 129145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.280324964505078e-06,
+      "loss": 1.3625,
+      "step": 129150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2796235174410116e-06,
+      "loss": 0.9429,
+      "step": 129155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.278922091622293e-06,
+      "loss": 1.1808,
+      "step": 129160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.278220687052934e-06,
+      "loss": 1.1394,
+      "step": 129165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.277519303736937e-06,
+      "loss": 1.3703,
+      "step": 129170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2768179416783035e-06,
+      "loss": 1.2325,
+      "step": 129175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.276116600881045e-06,
+      "loss": 1.0763,
+      "step": 129180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.275415281349164e-06,
+      "loss": 1.8636,
+      "step": 129185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.274713983086662e-06,
+      "loss": 1.3196,
+      "step": 129190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.27401270609755e-06,
+      "loss": 1.0349,
+      "step": 129195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.273311450385828e-06,
+      "loss": 1.6021,
+      "step": 129200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.272610215955499e-06,
+      "loss": 1.3989,
+      "step": 129205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.271909002810577e-06,
+      "loss": 1.1108,
+      "step": 129210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2712078109550554e-06,
+      "loss": 1.1519,
+      "step": 129215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.270506640392946e-06,
+      "loss": 1.1726,
+      "step": 129220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.269805491128244e-06,
+      "loss": 0.9842,
+      "step": 129225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2691043631649636e-06,
+      "loss": 0.9836,
+      "step": 129230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.268403256507103e-06,
+      "loss": 1.0856,
+      "step": 129235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.267702171158665e-06,
+      "loss": 1.0724,
+      "step": 129240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.267001107123658e-06,
+      "loss": 1.2037,
+      "step": 129245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.266300064406081e-06,
+      "loss": 0.9541,
+      "step": 129250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.265599043009938e-06,
+      "loss": 1.2856,
+      "step": 129255
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2648980429392336e-06,
+      "loss": 1.063,
+      "step": 129260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.264197064197972e-06,
+      "loss": 1.2687,
+      "step": 129265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2634961067901525e-06,
+      "loss": 1.6732,
+      "step": 129270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.262795170719779e-06,
+      "loss": 1.027,
+      "step": 129275
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.262094255990859e-06,
+      "loss": 1.0942,
+      "step": 129280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.261393362607389e-06,
+      "loss": 1.1852,
+      "step": 129285
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.260692490573372e-06,
+      "loss": 1.23,
+      "step": 129290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.259991639892813e-06,
+      "loss": 1.4035,
+      "step": 129295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.259290810569716e-06,
+      "loss": 1.3317,
+      "step": 129300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.258590002608077e-06,
+      "loss": 1.0337,
+      "step": 129305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.257889216011903e-06,
+      "loss": 1.3147,
+      "step": 129310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.257188450785196e-06,
+      "loss": 1.1284,
+      "step": 129315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.25648770693195e-06,
+      "loss": 1.2748,
+      "step": 129320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.255786984456178e-06,
+      "loss": 1.3608,
+      "step": 129325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.255086283361875e-06,
+      "loss": 1.2129,
+      "step": 129330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.254385603653041e-06,
+      "loss": 1.3988,
+      "step": 129335
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.253684945333684e-06,
+      "loss": 0.9698,
+      "step": 129340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.252984308407801e-06,
+      "loss": 1.326,
+      "step": 129345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.252283692879389e-06,
+      "loss": 1.3189,
+      "step": 129350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.251583098752454e-06,
+      "loss": 1.0354,
+      "step": 129355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.250882526030999e-06,
+      "loss": 1.3525,
+      "step": 129360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.250181974719018e-06,
+      "loss": 1.2125,
+      "step": 129365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.249481444820514e-06,
+      "loss": 0.9939,
+      "step": 129370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.248780936339492e-06,
+      "loss": 1.1449,
+      "step": 129375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.248080449279946e-06,
+      "loss": 1.1218,
+      "step": 129380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2473799836458775e-06,
+      "loss": 1.0634,
+      "step": 129385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.246679539441289e-06,
+      "loss": 1.2337,
+      "step": 129390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.24597911667018e-06,
+      "loss": 0.9232,
+      "step": 129395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.245278715336547e-06,
+      "loss": 0.9348,
+      "step": 129400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2445783354443935e-06,
+      "loss": 1.085,
+      "step": 129405
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.243877976997718e-06,
+      "loss": 1.1595,
+      "step": 129410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.243177640000516e-06,
+      "loss": 1.16,
+      "step": 129415
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.242477324456795e-06,
+      "loss": 1.1662,
+      "step": 129420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.241777030370548e-06,
+      "loss": 1.2243,
+      "step": 129425
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2410767577457775e-06,
+      "loss": 1.0592,
+      "step": 129430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.240376506586476e-06,
+      "loss": 1.0968,
+      "step": 129435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.239676276896649e-06,
+      "loss": 0.983,
+      "step": 129440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2389760686802936e-06,
+      "loss": 0.8648,
+      "step": 129445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.238275881941404e-06,
+      "loss": 1.1537,
+      "step": 129450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.237575716683988e-06,
+      "loss": 1.1344,
+      "step": 129455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.236875572912035e-06,
+      "loss": 1.0982,
+      "step": 129460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.236175450629546e-06,
+      "loss": 1.1575,
+      "step": 129465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.235475349840521e-06,
+      "loss": 1.265,
+      "step": 129470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.234775270548958e-06,
+      "loss": 1.2881,
+      "step": 129475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.23407521275885e-06,
+      "loss": 1.3714,
+      "step": 129480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.233375176474199e-06,
+      "loss": 1.358,
+      "step": 129485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.232675161699004e-06,
+      "loss": 1.3803,
+      "step": 129490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2319751684372565e-06,
+      "loss": 1.1659,
+      "step": 129495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2312751966929595e-06,
+      "loss": 1.1294,
+      "step": 129500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.230575246470108e-06,
+      "loss": 1.3645,
+      "step": 129505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.229875317772701e-06,
+      "loss": 1.5246,
+      "step": 129510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.22917541060473e-06,
+      "loss": 1.3283,
+      "step": 129515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.228475524970197e-06,
+      "loss": 1.1544,
+      "step": 129520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2277756608731e-06,
+      "loss": 1.1244,
+      "step": 129525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.227075818317428e-06,
+      "loss": 1.1673,
+      "step": 129530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.226375997307186e-06,
+      "loss": 1.416,
+      "step": 129535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.225676197846365e-06,
+      "loss": 1.77,
+      "step": 129540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.224976419938962e-06,
+      "loss": 1.3065,
+      "step": 129545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.224276663588977e-06,
+      "loss": 1.3588,
+      "step": 129550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.223576928800402e-06,
+      "loss": 1.1546,
+      "step": 129555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.222877215577232e-06,
+      "loss": 1.1173,
+      "step": 129560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.222177523923465e-06,
+      "loss": 1.1338,
+      "step": 129565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.221477853843101e-06,
+      "loss": 1.2137,
+      "step": 129570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.220778205340126e-06,
+      "loss": 1.0869,
+      "step": 129575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.22007857841854e-06,
+      "loss": 1.1832,
+      "step": 129580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.219378973082342e-06,
+      "loss": 1.0927,
+      "step": 129585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.21867938933552e-06,
+      "loss": 1.2174,
+      "step": 129590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2179798271820725e-06,
+      "loss": 1.3224,
+      "step": 129595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.217280286625995e-06,
+      "loss": 1.4763,
+      "step": 129600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.216580767671284e-06,
+      "loss": 1.0722,
+      "step": 129605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2158812703219285e-06,
+      "loss": 1.2305,
+      "step": 129610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.215181794581927e-06,
+      "loss": 1.304,
+      "step": 129615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.214482340455275e-06,
+      "loss": 1.0845,
+      "step": 129620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.21378290794596e-06,
+      "loss": 1.1089,
+      "step": 129625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.213083497057987e-06,
+      "loss": 0.9929,
+      "step": 129630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.212384107795342e-06,
+      "loss": 1.2889,
+      "step": 129635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.211684740162018e-06,
+      "loss": 1.5135,
+      "step": 129640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.210985394162017e-06,
+      "loss": 0.9472,
+      "step": 129645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.210286069799323e-06,
+      "loss": 1.3864,
+      "step": 129650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.209586767077938e-06,
+      "loss": 1.6017,
+      "step": 129655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2088874860018445e-06,
+      "loss": 1.1164,
+      "step": 129660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.208188226575049e-06,
+      "loss": 1.5019,
+      "step": 129665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.207488988801535e-06,
+      "loss": 0.9733,
+      "step": 129670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.206789772685298e-06,
+      "loss": 1.8578,
+      "step": 129675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.206090578230335e-06,
+      "loss": 1.1161,
+      "step": 129680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.205391405440634e-06,
+      "loss": 1.3064,
+      "step": 129685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.204692254320186e-06,
+      "loss": 1.1759,
+      "step": 129690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.203993124872989e-06,
+      "loss": 1.0358,
+      "step": 129695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.203294017103034e-06,
+      "loss": 1.3588,
+      "step": 129700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.20259493101431e-06,
+      "loss": 1.1574,
+      "step": 129705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.201895866610812e-06,
+      "loss": 1.2081,
+      "step": 129710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.2011968238965344e-06,
+      "loss": 1.3273,
+      "step": 129715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.200497802875465e-06,
+      "loss": 1.0433,
+      "step": 129720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.199798803551593e-06,
+      "loss": 1.3612,
+      "step": 129725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.199099825928916e-06,
+      "loss": 1.0834,
+      "step": 129730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.198400870011425e-06,
+      "loss": 1.2568,
+      "step": 129735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.1977019358031066e-06,
+      "loss": 1.1317,
+      "step": 129740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.197003023307957e-06,
+      "loss": 1.0892,
+      "step": 129745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.196304132529964e-06,
+      "loss": 1.249,
+      "step": 129750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.195605263473121e-06,
+      "loss": 1.2655,
+      "step": 129755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.194906416141421e-06,
+      "loss": 1.1809,
+      "step": 129760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.19420759053885e-06,
+      "loss": 1.1253,
+      "step": 129765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.193508786669398e-06,
+      "loss": 1.4196,
+      "step": 129770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.192810004537062e-06,
+      "loss": 1.3245,
+      "step": 129775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.192111244145828e-06,
+      "loss": 1.5725,
+      "step": 129780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.191412505499683e-06,
+      "loss": 1.0323,
+      "step": 129785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.1907137886026245e-06,
+      "loss": 1.0712,
+      "step": 129790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.190015093458641e-06,
+      "loss": 1.0823,
+      "step": 129795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.189316420071718e-06,
+      "loss": 1.451,
+      "step": 129800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.1886177684458455e-06,
+      "loss": 1.1668,
+      "step": 129805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.187919138585019e-06,
+      "loss": 1.4478,
+      "step": 129810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.187220530493225e-06,
+      "loss": 1.1653,
+      "step": 129815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.186521944174448e-06,
+      "loss": 1.0267,
+      "step": 129820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.185823379632686e-06,
+      "loss": 1.3009,
+      "step": 129825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.185124836871925e-06,
+      "loss": 0.8493,
+      "step": 129830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.1844263158961485e-06,
+      "loss": 1.3912,
+      "step": 129835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.183727816709353e-06,
+      "loss": 1.1621,
+      "step": 129840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.183029339315523e-06,
+      "loss": 1.1078,
+      "step": 129845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.182330883718647e-06,
+      "loss": 1.1386,
+      "step": 129850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.181632449922718e-06,
+      "loss": 1.3256,
+      "step": 129855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.18093403793172e-06,
+      "loss": 1.2371,
+      "step": 129860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.180235647749646e-06,
+      "loss": 1.1259,
+      "step": 129865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.179537279380475e-06,
+      "loss": 1.403,
+      "step": 129870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.178838932828206e-06,
+      "loss": 1.2294,
+      "step": 129875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.17814060809682e-06,
+      "loss": 1.01,
+      "step": 129880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.1774423051903046e-06,
+      "loss": 1.1596,
+      "step": 129885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.176744024112653e-06,
+      "loss": 1.4196,
+      "step": 129890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.176045764867848e-06,
+      "loss": 1.0381,
+      "step": 129895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.175347527459876e-06,
+      "loss": 1.3605,
+      "step": 129900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.1746493118927295e-06,
+      "loss": 1.1973,
+      "step": 129905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.1739511181703955e-06,
+      "loss": 1.3919,
+      "step": 129910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.173252946296853e-06,
+      "loss": 0.9709,
+      "step": 129915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.172554796276098e-06,
+      "loss": 1.3453,
+      "step": 129920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.171856668112114e-06,
+      "loss": 1.5752,
+      "step": 129925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.171158561808881e-06,
+      "loss": 1.2888,
+      "step": 129930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.170460477370399e-06,
+      "loss": 1.2081,
+      "step": 129935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.169762414800646e-06,
+      "loss": 1.3168,
+      "step": 129940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.169064374103611e-06,
+      "loss": 1.4502,
+      "step": 129945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.168366355283275e-06,
+      "loss": 1.2925,
+      "step": 129950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.16766835834363e-06,
+      "loss": 1.2867,
+      "step": 129955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1669703832886614e-06,
+      "loss": 1.2811,
+      "step": 129960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.166272430122349e-06,
+      "loss": 1.0438,
+      "step": 129965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1655744988486875e-06,
+      "loss": 1.1801,
+      "step": 129970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1648765894716565e-06,
+      "loss": 1.091,
+      "step": 129975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.164178701995242e-06,
+      "loss": 1.0724,
+      "step": 129980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1634808364234325e-06,
+      "loss": 1.1273,
+      "step": 129985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.16278299276021e-06,
+      "loss": 1.5067,
+      "step": 129990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1620851710095574e-06,
+      "loss": 1.3151,
+      "step": 129995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.161387371175465e-06,
+      "loss": 1.3166,
+      "step": 130000
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 1.2218666076660156,
+      "eval_runtime": 6474.0806,
+      "eval_samples_per_second": 3.57,
+      "eval_steps_per_second": 1.785,
+      "step": 130000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.160689593261918e-06,
+      "loss": 1.3421,
+      "step": 130005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1599918372728965e-06,
+      "loss": 1.356,
+      "step": 130010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.159294103212383e-06,
+      "loss": 1.1014,
+      "step": 130015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.158596391084373e-06,
+      "loss": 1.0564,
+      "step": 130020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1578987008928395e-06,
+      "loss": 1.37,
+      "step": 130025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1572010326417685e-06,
+      "loss": 1.3231,
+      "step": 130030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.156503386335151e-06,
+      "loss": 1.1617,
+      "step": 130035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1558057619769666e-06,
+      "loss": 1.0639,
+      "step": 130040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.155108159571195e-06,
+      "loss": 1.0897,
+      "step": 130045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.154410579121826e-06,
+      "loss": 1.0745,
+      "step": 130050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.153713020632843e-06,
+      "loss": 1.5273,
+      "step": 130055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.15301548410822e-06,
+      "loss": 1.2209,
+      "step": 130060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.152317969551955e-06,
+      "loss": 1.1174,
+      "step": 130065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.151620476968023e-06,
+      "loss": 1.4885,
+      "step": 130070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1509230063604045e-06,
+      "loss": 1.2733,
+      "step": 130075
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1502255577330895e-06,
+      "loss": 1.3086,
+      "step": 130080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.149528131090056e-06,
+      "loss": 1.1802,
+      "step": 130085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.148830726435289e-06,
+      "loss": 0.9791,
+      "step": 130090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.148133343772767e-06,
+      "loss": 1.2518,
+      "step": 130095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1474359831064775e-06,
+      "loss": 1.3134,
+      "step": 130100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1467386444403995e-06,
+      "loss": 1.3324,
+      "step": 130105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.146041327778516e-06,
+      "loss": 1.3481,
+      "step": 130110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.145344033124811e-06,
+      "loss": 1.1009,
+      "step": 130115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.144646760483266e-06,
+      "loss": 1.4033,
+      "step": 130120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.143949509857856e-06,
+      "loss": 1.3662,
+      "step": 130125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.143252281252573e-06,
+      "loss": 1.306,
+      "step": 130130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1425550746713945e-06,
+      "loss": 1.7294,
+      "step": 130135
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.141857890118298e-06,
+      "loss": 1.0666,
+      "step": 130140
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.141160727597269e-06,
+      "loss": 1.2011,
+      "step": 130145
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.140463587112289e-06,
+      "loss": 1.1336,
+      "step": 130150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.139766468667338e-06,
+      "loss": 1.1428,
+      "step": 130155
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.139069372266395e-06,
+      "loss": 1.1666,
+      "step": 130160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.138372297913443e-06,
+      "loss": 1.4667,
+      "step": 130165
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1376752456124645e-06,
+      "loss": 1.0045,
+      "step": 130170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.136978215367434e-06,
+      "loss": 1.1526,
+      "step": 130175
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1362812071823375e-06,
+      "loss": 0.9941,
+      "step": 130180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.135584221061153e-06,
+      "loss": 1.2812,
+      "step": 130185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.13488725700786e-06,
+      "loss": 1.04,
+      "step": 130190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.134190315026443e-06,
+      "loss": 1.0353,
+      "step": 130195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1334933951208755e-06,
+      "loss": 1.199,
+      "step": 130200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.13279649729514e-06,
+      "loss": 1.0093,
+      "step": 130205
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.132099621553218e-06,
+      "loss": 1.32,
+      "step": 130210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.131402767899088e-06,
+      "loss": 1.1256,
+      "step": 130215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.130705936336727e-06,
+      "loss": 1.0793,
+      "step": 130220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.130009126870118e-06,
+      "loss": 1.3877,
+      "step": 130225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.129312339503239e-06,
+      "loss": 1.4941,
+      "step": 130230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.128615574240068e-06,
+      "loss": 1.5498,
+      "step": 130235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.127918831084582e-06,
+      "loss": 1.1304,
+      "step": 130240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.127222110040762e-06,
+      "loss": 1.3621,
+      "step": 130245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1265254111125915e-06,
+      "loss": 1.0009,
+      "step": 130250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.125828734304039e-06,
+      "loss": 1.2461,
+      "step": 130255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1251320796190905e-06,
+      "loss": 1.3889,
+      "step": 130260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.124435447061723e-06,
+      "loss": 1.831,
+      "step": 130265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1237388366359085e-06,
+      "loss": 1.2879,
+      "step": 130270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1230422483456365e-06,
+      "loss": 1.4148,
+      "step": 130275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.122345682194875e-06,
+      "loss": 1.0629,
+      "step": 130280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.121649138187604e-06,
+      "loss": 1.1065,
+      "step": 130285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.120952616327806e-06,
+      "loss": 1.1229,
+      "step": 130290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.120256116619454e-06,
+      "loss": 1.036,
+      "step": 130295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.119559639066528e-06,
+      "loss": 1.5039,
+      "step": 130300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.118863183672998e-06,
+      "loss": 0.9772,
+      "step": 130305
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.118166750442852e-06,
+      "loss": 1.4379,
+      "step": 130310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.117470339380061e-06,
+      "loss": 1.1701,
+      "step": 130315
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1167739504885984e-06,
+      "loss": 1.1469,
+      "step": 130320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1160775837724495e-06,
+      "loss": 1.4486,
+      "step": 130325
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.115381239235585e-06,
+      "loss": 1.2334,
+      "step": 130330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.114684916881981e-06,
+      "loss": 1.1017,
+      "step": 130335
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.113988616715618e-06,
+      "loss": 1.1359,
+      "step": 130340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.113292338740473e-06,
+      "loss": 1.2597,
+      "step": 130345
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.112596082960514e-06,
+      "loss": 1.2348,
+      "step": 130350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1118998493797245e-06,
+      "loss": 1.1889,
+      "step": 130355
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.11120363800208e-06,
+      "loss": 1.1322,
+      "step": 130360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1105074488315484e-06,
+      "loss": 1.2557,
+      "step": 130365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.109811281872118e-06,
+      "loss": 1.3959,
+      "step": 130370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1091151371277544e-06,
+      "loss": 1.4445,
+      "step": 130375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.108419014602438e-06,
+      "loss": 1.1687,
+      "step": 130380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.107722914300138e-06,
+      "loss": 1.3403,
+      "step": 130385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.107026836224836e-06,
+      "loss": 1.3256,
+      "step": 130390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.106330780380505e-06,
+      "loss": 1.1811,
+      "step": 130395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1056347467711185e-06,
+      "loss": 1.0768,
+      "step": 130400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1049387354006536e-06,
+      "loss": 1.4349,
+      "step": 130405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.104242746273083e-06,
+      "loss": 1.2174,
+      "step": 130410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.10354677939238e-06,
+      "loss": 1.074,
+      "step": 130415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.102850834762523e-06,
+      "loss": 1.0383,
+      "step": 130420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.102154912387484e-06,
+      "loss": 1.1961,
+      "step": 130425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.101459012271233e-06,
+      "loss": 1.3729,
+      "step": 130430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.100763134417752e-06,
+      "loss": 0.9841,
+      "step": 130435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.1000672788310125e-06,
+      "loss": 1.2874,
+      "step": 130440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.099371445514983e-06,
+      "loss": 1.145,
+      "step": 130445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.098675634473639e-06,
+      "loss": 1.2134,
+      "step": 130450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0979798457109575e-06,
+      "loss": 1.0516,
+      "step": 130455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.097284079230911e-06,
+      "loss": 1.1193,
+      "step": 130460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.096588335037468e-06,
+      "loss": 1.0913,
+      "step": 130465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.095892613134608e-06,
+      "loss": 1.1369,
+      "step": 130470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.095196913526302e-06,
+      "loss": 0.9899,
+      "step": 130475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0945012362165165e-06,
+      "loss": 0.9954,
+      "step": 130480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.093805581209235e-06,
+      "loss": 1.3939,
+      "step": 130485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0931099485084225e-06,
+      "loss": 1.5247,
+      "step": 130490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0924143381180514e-06,
+      "loss": 1.2416,
+      "step": 130495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.091718750042099e-06,
+      "loss": 1.0475,
+      "step": 130500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0910231842845345e-06,
+      "loss": 1.0787,
+      "step": 130505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.090327640849329e-06,
+      "loss": 1.3115,
+      "step": 130510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.089632119740455e-06,
+      "loss": 1.4946,
+      "step": 130515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.088936620961889e-06,
+      "loss": 1.2249,
+      "step": 130520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0882411445175945e-06,
+      "loss": 1.3894,
+      "step": 130525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.087545690411546e-06,
+      "loss": 1.1945,
+      "step": 130530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.086850258647721e-06,
+      "loss": 1.2797,
+      "step": 130535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0861548492300815e-06,
+      "loss": 1.2237,
+      "step": 130540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.085459462162602e-06,
+      "loss": 1.2577,
+      "step": 130545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.084764097449256e-06,
+      "loss": 1.2699,
+      "step": 130550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.084068755094015e-06,
+      "loss": 1.2711,
+      "step": 130555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.083373435100843e-06,
+      "loss": 0.992,
+      "step": 130560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0826781374737184e-06,
+      "loss": 1.3224,
+      "step": 130565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.081982862216611e-06,
+      "loss": 1.4141,
+      "step": 130570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.081287609333482e-06,
+      "loss": 1.254,
+      "step": 130575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0805923788283135e-06,
+      "loss": 1.6012,
+      "step": 130580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.07989717070507e-06,
+      "loss": 1.0367,
+      "step": 130585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.079201984967724e-06,
+      "loss": 1.2291,
+      "step": 130590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.078506821620238e-06,
+      "loss": 0.8803,
+      "step": 130595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.077811680666591e-06,
+      "loss": 1.0129,
+      "step": 130600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.07711656211075e-06,
+      "loss": 1.4216,
+      "step": 130605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.076421465956678e-06,
+      "loss": 1.024,
+      "step": 130610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.075726392208354e-06,
+      "loss": 1.3085,
+      "step": 130615
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.075031340869744e-06,
+      "loss": 1.3932,
+      "step": 130620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.074336311944812e-06,
+      "loss": 1.0831,
+      "step": 130625
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.073641305437534e-06,
+      "loss": 1.147,
+      "step": 130630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.072946321351876e-06,
+      "loss": 1.2251,
+      "step": 130635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0722513596918046e-06,
+      "loss": 1.0215,
+      "step": 130640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.071556420461291e-06,
+      "loss": 1.0655,
+      "step": 130645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.070861503664307e-06,
+      "loss": 1.0502,
+      "step": 130650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.070166609304812e-06,
+      "loss": 1.245,
+      "step": 130655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.069471737386781e-06,
+      "loss": 1.262,
+      "step": 130660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.068776887914183e-06,
+      "loss": 1.5983,
+      "step": 130665
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.068082060890982e-06,
+      "loss": 1.0679,
+      "step": 130670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.067387256321144e-06,
+      "loss": 1.2803,
+      "step": 130675
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.066692474208642e-06,
+      "loss": 1.4405,
+      "step": 130680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.065997714557446e-06,
+      "loss": 1.5543,
+      "step": 130685
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.065302977371512e-06,
+      "loss": 1.0968,
+      "step": 130690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.064608262654817e-06,
+      "loss": 0.9025,
+      "step": 130695
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.063913570411328e-06,
+      "loss": 1.6191,
+      "step": 130700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.063218900645002e-06,
+      "loss": 1.0679,
+      "step": 130705
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.062524253359822e-06,
+      "loss": 1.2623,
+      "step": 130710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0618296285597436e-06,
+      "loss": 1.4854,
+      "step": 130715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.061135026248732e-06,
+      "loss": 1.1884,
+      "step": 130720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0604404464307645e-06,
+      "loss": 1.426,
+      "step": 130725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.059745889109798e-06,
+      "loss": 1.3666,
+      "step": 130730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.059051354289802e-06,
+      "loss": 1.3724,
+      "step": 130735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.05835684197474e-06,
+      "loss": 1.2212,
+      "step": 130740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.057662352168585e-06,
+      "loss": 1.255,
+      "step": 130745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.056967884875296e-06,
+      "loss": 1.2314,
+      "step": 130750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.056273440098839e-06,
+      "loss": 1.1549,
+      "step": 130755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.055579017843185e-06,
+      "loss": 1.3299,
+      "step": 130760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0548846181122955e-06,
+      "loss": 1.2981,
+      "step": 130765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.054190240910134e-06,
+      "loss": 1.1802,
+      "step": 130770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.053495886240671e-06,
+      "loss": 1.2568,
+      "step": 130775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.052801554107872e-06,
+      "loss": 1.115,
+      "step": 130780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.052107244515694e-06,
+      "loss": 1.3018,
+      "step": 130785
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.051412957468109e-06,
+      "loss": 1.4908,
+      "step": 130790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.050718692969079e-06,
+      "loss": 1.1155,
+      "step": 130795
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0500244510225696e-06,
+      "loss": 1.4758,
+      "step": 130800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.049330231632547e-06,
+      "loss": 1.0621,
+      "step": 130805
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.048636034802973e-06,
+      "loss": 1.3068,
+      "step": 130810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.047941860537813e-06,
+      "loss": 1.384,
+      "step": 130815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0472477088410285e-06,
+      "loss": 1.5815,
+      "step": 130820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0465535797165874e-06,
+      "loss": 1.0635,
+      "step": 130825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.045859473168451e-06,
+      "loss": 1.0949,
+      "step": 130830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0451653892005815e-06,
+      "loss": 1.2071,
+      "step": 130835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.04447132781695e-06,
+      "loss": 1.864,
+      "step": 130840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.043777289021511e-06,
+      "loss": 0.999,
+      "step": 130845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.04308327281823e-06,
+      "loss": 1.5411,
+      "step": 130850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0423892792110735e-06,
+      "loss": 1.0437,
+      "step": 130855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.041695308204005e-06,
+      "loss": 1.0706,
+      "step": 130860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.041001359800982e-06,
+      "loss": 1.2482,
+      "step": 130865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.040307434005973e-06,
+      "loss": 1.3177,
+      "step": 130870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0396135308229386e-06,
+      "loss": 1.5022,
+      "step": 130875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.038919650255841e-06,
+      "loss": 1.1219,
+      "step": 130880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.038225792308638e-06,
+      "loss": 1.3656,
+      "step": 130885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.037531956985301e-06,
+      "loss": 1.2673,
+      "step": 130890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.036838144289789e-06,
+      "loss": 1.1198,
+      "step": 130895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.036144354226059e-06,
+      "loss": 1.2191,
+      "step": 130900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.035450586798078e-06,
+      "loss": 0.9469,
+      "step": 130905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0347568420098105e-06,
+      "loss": 1.2793,
+      "step": 130910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.034063119865205e-06,
+      "loss": 1.113,
+      "step": 130915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.03336942036824e-06,
+      "loss": 1.2578,
+      "step": 130920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.032675743522866e-06,
+      "loss": 1.3023,
+      "step": 130925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.031982089333047e-06,
+      "loss": 1.3349,
+      "step": 130930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0312884578027485e-06,
+      "loss": 1.2588,
+      "step": 130935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.030594848935924e-06,
+      "loss": 1.1258,
+      "step": 130940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.029901262736537e-06,
+      "loss": 1.1512,
+      "step": 130945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.02920769920855e-06,
+      "loss": 1.2559,
+      "step": 130950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.028514158355926e-06,
+      "loss": 1.189,
+      "step": 130955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.02782064018262e-06,
+      "loss": 1.2128,
+      "step": 130960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0271271446925905e-06,
+      "loss": 1.2991,
+      "step": 130965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.026433671889808e-06,
+      "loss": 1.1236,
+      "step": 130970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.025740221778225e-06,
+      "loss": 1.0324,
+      "step": 130975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0250467943618e-06,
+      "loss": 1.3176,
+      "step": 130980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.024353389644497e-06,
+      "loss": 1.2756,
+      "step": 130985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.023660007630276e-06,
+      "loss": 1.0542,
+      "step": 130990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.022966648323092e-06,
+      "loss": 1.1125,
+      "step": 130995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0222733117269095e-06,
+      "loss": 1.2447,
+      "step": 131000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.021579997845688e-06,
+      "loss": 1.3955,
+      "step": 131005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.020886706683379e-06,
+      "loss": 1.1934,
+      "step": 131010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.020193438243953e-06,
+      "loss": 1.4967,
+      "step": 131015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.01950019253136e-06,
+      "loss": 1.3186,
+      "step": 131020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.018806969549564e-06,
+      "loss": 1.3368,
+      "step": 131025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.018113769302518e-06,
+      "loss": 1.3219,
+      "step": 131030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.017420591794185e-06,
+      "loss": 1.0411,
+      "step": 131035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0167274370285245e-06,
+      "loss": 1.1531,
+      "step": 131040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.016034305009489e-06,
+      "loss": 1.0332,
+      "step": 131045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.015341195741045e-06,
+      "loss": 1.1238,
+      "step": 131050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.014648109227144e-06,
+      "loss": 1.6201,
+      "step": 131055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.013955045471745e-06,
+      "loss": 1.629,
+      "step": 131060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.013262004478808e-06,
+      "loss": 1.1948,
+      "step": 131065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0125689862522885e-06,
+      "loss": 1.6588,
+      "step": 131070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0118759907961436e-06,
+      "loss": 1.1574,
+      "step": 131075
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.011183018114333e-06,
+      "loss": 1.2872,
+      "step": 131080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.010490068210815e-06,
+      "loss": 1.0638,
+      "step": 131085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.009797141089539e-06,
+      "loss": 1.3018,
+      "step": 131090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.00910423675447e-06,
+      "loss": 1.0881,
+      "step": 131095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.008411355209565e-06,
+      "loss": 1.1821,
+      "step": 131100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.007718496458776e-06,
+      "loss": 1.3593,
+      "step": 131105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.00702566050606e-06,
+      "loss": 1.0098,
+      "step": 131110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0063328473553776e-06,
+      "loss": 1.2469,
+      "step": 131115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.005640057010684e-06,
+      "loss": 1.5932,
+      "step": 131120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.00494728947593e-06,
+      "loss": 1.0893,
+      "step": 131125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.004254544755077e-06,
+      "loss": 1.2251,
+      "step": 131130
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.00356182285208e-06,
+      "loss": 1.2146,
+      "step": 131135
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.002869123770894e-06,
+      "loss": 1.1775,
+      "step": 131140
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.002176447515478e-06,
+      "loss": 1.2903,
+      "step": 131145
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.001483794089782e-06,
+      "loss": 1.3193,
+      "step": 131150
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.000791163497764e-06,
+      "loss": 1.2639,
+      "step": 131155
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.0000985557433785e-06,
+      "loss": 1.1084,
+      "step": 131160
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.999405970830585e-06,
+      "loss": 1.2865,
+      "step": 131165
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.998713408763333e-06,
+      "loss": 1.3684,
+      "step": 131170
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.99802086954558e-06,
+      "loss": 1.3208,
+      "step": 131175
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.997328353181282e-06,
+      "loss": 1.2756,
+      "step": 131180
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.99663585967439e-06,
+      "loss": 1.43,
+      "step": 131185
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.995943389028859e-06,
+      "loss": 1.4007,
+      "step": 131190
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.995250941248647e-06,
+      "loss": 1.2057,
+      "step": 131195
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.994558516337707e-06,
+      "loss": 1.3078,
+      "step": 131200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.993866114299989e-06,
+      "loss": 1.3295,
+      "step": 131205
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.993173735139453e-06,
+      "loss": 1.5601,
+      "step": 131210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.99248137886005e-06,
+      "loss": 1.3435,
+      "step": 131215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.991789045465729e-06,
+      "loss": 1.2245,
+      "step": 131220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.991096734960454e-06,
+      "loss": 1.1628,
+      "step": 131225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.990404447348171e-06,
+      "loss": 1.1394,
+      "step": 131230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.989712182632834e-06,
+      "loss": 1.2837,
+      "step": 131235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9890199408184e-06,
+      "loss": 1.6424,
+      "step": 131240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.988327721908819e-06,
+      "loss": 1.4226,
+      "step": 131245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.987635525908044e-06,
+      "loss": 1.2264,
+      "step": 131250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.986943352820025e-06,
+      "loss": 1.195,
+      "step": 131255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.986251202648723e-06,
+      "loss": 1.2576,
+      "step": 131260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9855590753980816e-06,
+      "loss": 1.2615,
+      "step": 131265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.984866971072058e-06,
+      "loss": 1.0926,
+      "step": 131270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.984174889674605e-06,
+      "loss": 1.3234,
+      "step": 131275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9834828312096725e-06,
+      "loss": 1.1222,
+      "step": 131280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.982790795681211e-06,
+      "loss": 1.3473,
+      "step": 131285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.982098783093177e-06,
+      "loss": 1.4863,
+      "step": 131290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9814067934495225e-06,
+      "loss": 1.025,
+      "step": 131295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.980714826754193e-06,
+      "loss": 1.2253,
+      "step": 131300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.980022883011143e-06,
+      "loss": 1.379,
+      "step": 131305
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.979330962224329e-06,
+      "loss": 1.0293,
+      "step": 131310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.978639064397691e-06,
+      "loss": 1.6116,
+      "step": 131315
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.977947189535192e-06,
+      "loss": 1.2135,
+      "step": 131320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9772553376407775e-06,
+      "loss": 1.3884,
+      "step": 131325
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.976563508718398e-06,
+      "loss": 1.1004,
+      "step": 131330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9758717027720035e-06,
+      "loss": 1.1935,
+      "step": 131335
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9751799198055475e-06,
+      "loss": 1.3996,
+      "step": 131340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.97448815982298e-06,
+      "loss": 1.3243,
+      "step": 131345
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.973796422828244e-06,
+      "loss": 1.3731,
+      "step": 131350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.973104708825303e-06,
+      "loss": 1.7738,
+      "step": 131355
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.972413017818097e-06,
+      "loss": 1.4968,
+      "step": 131360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9717213498105785e-06,
+      "loss": 1.1881,
+      "step": 131365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.971029704806701e-06,
+      "loss": 1.1867,
+      "step": 131370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.970338082810408e-06,
+      "loss": 1.1963,
+      "step": 131375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.969646483825651e-06,
+      "loss": 1.3077,
+      "step": 131380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.968954907856382e-06,
+      "loss": 1.3766,
+      "step": 131385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.96826335490655e-06,
+      "loss": 1.0121,
+      "step": 131390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.967571824980101e-06,
+      "loss": 1.1361,
+      "step": 131395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9668803180809835e-06,
+      "loss": 1.514,
+      "step": 131400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.966188834213154e-06,
+      "loss": 1.2305,
+      "step": 131405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.965497373380552e-06,
+      "loss": 1.3193,
+      "step": 131410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9648059355871294e-06,
+      "loss": 1.2443,
+      "step": 131415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.964114520836837e-06,
+      "loss": 1.3432,
+      "step": 131420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.963423129133623e-06,
+      "loss": 1.0659,
+      "step": 131425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9627317604814295e-06,
+      "loss": 1.309,
+      "step": 131430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.962040414884212e-06,
+      "loss": 1.1354,
+      "step": 131435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.961349092345914e-06,
+      "loss": 1.109,
+      "step": 131440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9606577928704845e-06,
+      "loss": 1.2501,
+      "step": 131445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.959966516461875e-06,
+      "loss": 1.17,
+      "step": 131450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.959275263124028e-06,
+      "loss": 1.0732,
+      "step": 131455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.95858403286089e-06,
+      "loss": 1.0075,
+      "step": 131460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.957892825676416e-06,
+      "loss": 1.2288,
+      "step": 131465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.957201641574545e-06,
+      "loss": 1.1117,
+      "step": 131470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.956510480559228e-06,
+      "loss": 1.0124,
+      "step": 131475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.955819342634408e-06,
+      "loss": 1.524,
+      "step": 131480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.955128227804039e-06,
+      "loss": 1.4301,
+      "step": 131485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.954437136072061e-06,
+      "loss": 1.1713,
+      "step": 131490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.953746067442421e-06,
+      "loss": 1.2756,
+      "step": 131495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.95305502191907e-06,
+      "loss": 1.0381,
+      "step": 131500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.952363999505953e-06,
+      "loss": 1.0638,
+      "step": 131505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.95167300020701e-06,
+      "loss": 1.0838,
+      "step": 131510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.950982024026194e-06,
+      "loss": 1.1691,
+      "step": 131515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.950291070967451e-06,
+      "loss": 1.1867,
+      "step": 131520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.949600141034719e-06,
+      "loss": 1.2655,
+      "step": 131525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9489092342319525e-06,
+      "loss": 1.9425,
+      "step": 131530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9482183505630905e-06,
+      "loss": 1.3265,
+      "step": 131535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9475274900320846e-06,
+      "loss": 1.0418,
+      "step": 131540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9468366526428725e-06,
+      "loss": 1.1172,
+      "step": 131545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.946145838399404e-06,
+      "loss": 1.3157,
+      "step": 131550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.945455047305626e-06,
+      "loss": 1.3124,
+      "step": 131555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.944764279365476e-06,
+      "loss": 1.2969,
+      "step": 131560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9440735345829055e-06,
+      "loss": 1.0779,
+      "step": 131565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.943382812961857e-06,
+      "loss": 1.0528,
+      "step": 131570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.942692114506272e-06,
+      "loss": 1.2461,
+      "step": 131575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.942001439220102e-06,
+      "loss": 0.9701,
+      "step": 131580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.941310787107284e-06,
+      "loss": 1.1155,
+      "step": 131585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.940620158171761e-06,
+      "loss": 1.4344,
+      "step": 131590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.939929552417485e-06,
+      "loss": 1.1705,
+      "step": 131595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.939238969848395e-06,
+      "loss": 1.4404,
+      "step": 131600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.938548410468431e-06,
+      "loss": 1.0754,
+      "step": 131605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.937857874281542e-06,
+      "loss": 1.2602,
+      "step": 131610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.937167361291672e-06,
+      "loss": 1.2472,
+      "step": 131615
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.936476871502758e-06,
+      "loss": 0.9614,
+      "step": 131620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9357864049187455e-06,
+      "loss": 1.1417,
+      "step": 131625
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.93509596154358e-06,
+      "loss": 1.1846,
+      "step": 131630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.934405541381206e-06,
+      "loss": 1.0628,
+      "step": 131635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.933715144435559e-06,
+      "loss": 1.2292,
+      "step": 131640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9330247707105874e-06,
+      "loss": 1.3109,
+      "step": 131645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.932334420210233e-06,
+      "loss": 1.3401,
+      "step": 131650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.931644092938431e-06,
+      "loss": 0.8481,
+      "step": 131655
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.930953788899137e-06,
+      "loss": 1.3539,
+      "step": 131660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.930263508096281e-06,
+      "loss": 1.013,
+      "step": 131665
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.929573250533809e-06,
+      "loss": 0.8617,
+      "step": 131670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.928883016215666e-06,
+      "loss": 1.3196,
+      "step": 131675
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.928192805145789e-06,
+      "loss": 1.3289,
+      "step": 131680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.927502617328121e-06,
+      "loss": 1.5452,
+      "step": 131685
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.926812452766602e-06,
+      "loss": 1.4276,
+      "step": 131690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.926122311465179e-06,
+      "loss": 1.3368,
+      "step": 131695
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.925432193427786e-06,
+      "loss": 1.0812,
+      "step": 131700
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.924742098658365e-06,
+      "loss": 0.9214,
+      "step": 131705
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.924052027160863e-06,
+      "loss": 1.2513,
+      "step": 131710
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.923361978939213e-06,
+      "loss": 1.5789,
+      "step": 131715
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.922671953997356e-06,
+      "loss": 1.1042,
+      "step": 131720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.92198195233924e-06,
+      "loss": 1.2286,
+      "step": 131725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9212919739688e-06,
+      "loss": 1.2357,
+      "step": 131730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.920602018889974e-06,
+      "loss": 1.2111,
+      "step": 131735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.919912087106705e-06,
+      "loss": 0.9404,
+      "step": 131740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.919222178622934e-06,
+      "loss": 1.0839,
+      "step": 131745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.918532293442596e-06,
+      "loss": 1.3383,
+      "step": 131750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.917842431569638e-06,
+      "loss": 1.1855,
+      "step": 131755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.917152593007995e-06,
+      "loss": 1.0573,
+      "step": 131760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.916462777761607e-06,
+      "loss": 1.3119,
+      "step": 131765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9157729858344095e-06,
+      "loss": 1.1975,
+      "step": 131770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.915083217230346e-06,
+      "loss": 1.2739,
+      "step": 131775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.914393471953356e-06,
+      "loss": 1.3001,
+      "step": 131780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9137037500073735e-06,
+      "loss": 1.3913,
+      "step": 131785
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9130140513963456e-06,
+      "loss": 1.2291,
+      "step": 131790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.912324376124203e-06,
+      "loss": 1.0861,
+      "step": 131795
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.911634724194883e-06,
+      "loss": 1.4051,
+      "step": 131800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.910945095612335e-06,
+      "loss": 1.2348,
+      "step": 131805
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.910255490380486e-06,
+      "loss": 1.2351,
+      "step": 131810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.909565908503276e-06,
+      "loss": 1.0979,
+      "step": 131815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.908876349984647e-06,
+      "loss": 1.0695,
+      "step": 131820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.908186814828536e-06,
+      "loss": 1.1373,
+      "step": 131825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.907497303038876e-06,
+      "loss": 0.8461,
+      "step": 131830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.906807814619608e-06,
+      "loss": 0.9703,
+      "step": 131835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.90611834957467e-06,
+      "loss": 0.9856,
+      "step": 131840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.905428907908e-06,
+      "loss": 1.1263,
+      "step": 131845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.904739489623528e-06,
+      "loss": 1.3437,
+      "step": 131850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9040500947252e-06,
+      "loss": 1.5754,
+      "step": 131855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.903360723216949e-06,
+      "loss": 1.315,
+      "step": 131860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.902671375102709e-06,
+      "loss": 1.3599,
+      "step": 131865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.901982050386421e-06,
+      "loss": 1.301,
+      "step": 131870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.90129274907202e-06,
+      "loss": 1.42,
+      "step": 131875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9006034711634395e-06,
+      "loss": 1.4947,
+      "step": 131880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.899914216664622e-06,
+      "loss": 1.216,
+      "step": 131885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.899224985579497e-06,
+      "loss": 1.1167,
+      "step": 131890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.898535777912002e-06,
+      "loss": 1.2148,
+      "step": 131895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.897846593666076e-06,
+      "loss": 1.2782,
+      "step": 131900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.897157432845653e-06,
+      "loss": 1.7229,
+      "step": 131905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.896468295454667e-06,
+      "loss": 1.1473,
+      "step": 131910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.895779181497051e-06,
+      "loss": 0.8461,
+      "step": 131915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.895090090976748e-06,
+      "loss": 1.4761,
+      "step": 131920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.894401023897687e-06,
+      "loss": 1.4725,
+      "step": 131925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.893711980263801e-06,
+      "loss": 1.2458,
+      "step": 131930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.893022960079031e-06,
+      "loss": 1.0101,
+      "step": 131935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.89233396334731e-06,
+      "loss": 1.289,
+      "step": 131940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.891644990072569e-06,
+      "loss": 1.2389,
+      "step": 131945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.8909560402587455e-06,
+      "loss": 1.2566,
+      "step": 131950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.890267113909776e-06,
+      "loss": 1.291,
+      "step": 131955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.889578211029584e-06,
+      "loss": 1.2508,
+      "step": 131960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.888889331622119e-06,
+      "loss": 1.0704,
+      "step": 131965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.888200475691303e-06,
+      "loss": 1.3865,
+      "step": 131970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.887511643241078e-06,
+      "loss": 1.1261,
+      "step": 131975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.886822834275369e-06,
+      "loss": 1.3428,
+      "step": 131980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.886134048798115e-06,
+      "loss": 1.1577,
+      "step": 131985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.885445286813251e-06,
+      "loss": 0.9718,
+      "step": 131990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.884756548324702e-06,
+      "loss": 1.3711,
+      "step": 131995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.884067833336411e-06,
+      "loss": 1.1675,
+      "step": 132000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.883379141852306e-06,
+      "loss": 1.4158,
+      "step": 132005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.882690473876317e-06,
+      "loss": 1.1816,
+      "step": 132010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.882001829412385e-06,
+      "loss": 1.2039,
+      "step": 132015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8813132084644345e-06,
+      "loss": 1.1801,
+      "step": 132020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.880624611036398e-06,
+      "loss": 1.1267,
+      "step": 132025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.879936037132215e-06,
+      "loss": 1.4954,
+      "step": 132030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.879247486755813e-06,
+      "loss": 1.2561,
+      "step": 132035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.87855895991112e-06,
+      "loss": 1.1934,
+      "step": 132040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.877870456602076e-06,
+      "loss": 1.4325,
+      "step": 132045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.87718197683261e-06,
+      "loss": 1.6458,
+      "step": 132050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.87649352060665e-06,
+      "loss": 1.4208,
+      "step": 132055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.875805087928128e-06,
+      "loss": 1.481,
+      "step": 132060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.875116678800979e-06,
+      "loss": 0.8761,
+      "step": 132065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.874428293229135e-06,
+      "loss": 1.2238,
+      "step": 132070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.87373993121652e-06,
+      "loss": 0.9203,
+      "step": 132075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.873051592767072e-06,
+      "loss": 1.5624,
+      "step": 132080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.872363277884719e-06,
+      "loss": 1.0387,
+      "step": 132085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.871674986573388e-06,
+      "loss": 0.9219,
+      "step": 132090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.870986718837018e-06,
+      "loss": 1.1169,
+      "step": 132095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.870298474679534e-06,
+      "loss": 1.1184,
+      "step": 132100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.869610254104864e-06,
+      "loss": 0.935,
+      "step": 132105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8689220571169444e-06,
+      "loss": 1.1808,
+      "step": 132110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8682338837197e-06,
+      "loss": 1.0029,
+      "step": 132115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.867545733917064e-06,
+      "loss": 1.2612,
+      "step": 132120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.86685760771296e-06,
+      "loss": 1.1707,
+      "step": 132125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.866169505111328e-06,
+      "loss": 0.9529,
+      "step": 132130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.865481426116089e-06,
+      "loss": 1.1561,
+      "step": 132135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.864793370731172e-06,
+      "loss": 0.9741,
+      "step": 132140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.864105338960512e-06,
+      "loss": 1.1189,
+      "step": 132145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.863417330808037e-06,
+      "loss": 1.0861,
+      "step": 132150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.862729346277669e-06,
+      "loss": 1.191,
+      "step": 132155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.862041385373344e-06,
+      "loss": 1.1473,
+      "step": 132160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.861353448098989e-06,
+      "loss": 1.1706,
+      "step": 132165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8606655344585305e-06,
+      "loss": 1.216,
+      "step": 132170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.859977644455899e-06,
+      "loss": 1.0733,
+      "step": 132175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.859289778095022e-06,
+      "loss": 0.9711,
+      "step": 132180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.858601935379825e-06,
+      "loss": 1.407,
+      "step": 132185
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.857914116314243e-06,
+      "loss": 1.3357,
+      "step": 132190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.857226320902196e-06,
+      "loss": 1.447,
+      "step": 132195
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.856538549147619e-06,
+      "loss": 1.4328,
+      "step": 132200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.855850801054428e-06,
+      "loss": 1.1882,
+      "step": 132205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.855163076626564e-06,
+      "loss": 1.2917,
+      "step": 132210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.854475375867946e-06,
+      "loss": 1.0018,
+      "step": 132215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.853787698782502e-06,
+      "loss": 1.4272,
+      "step": 132220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.853100045374164e-06,
+      "loss": 1.1731,
+      "step": 132225
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8524124156468536e-06,
+      "loss": 1.4644,
+      "step": 132230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.851724809604495e-06,
+      "loss": 1.191,
+      "step": 132235
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.851037227251023e-06,
+      "loss": 1.1508,
+      "step": 132240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8503496685903615e-06,
+      "loss": 1.4893,
+      "step": 132245
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.849662133626431e-06,
+      "loss": 1.5408,
+      "step": 132250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8489746223631636e-06,
+      "loss": 0.9882,
+      "step": 132255
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.848287134804487e-06,
+      "loss": 1.0365,
+      "step": 132260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.84759967095432e-06,
+      "loss": 1.3677,
+      "step": 132265
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.846912230816593e-06,
+      "loss": 1.0384,
+      "step": 132270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.84622481439523e-06,
+      "loss": 1.4372,
+      "step": 132275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.845537421694161e-06,
+      "loss": 1.1161,
+      "step": 132280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.844850052717303e-06,
+      "loss": 1.2722,
+      "step": 132285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.84416270746859e-06,
+      "loss": 1.0012,
+      "step": 132290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.843475385951943e-06,
+      "loss": 1.2022,
+      "step": 132295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.842788088171283e-06,
+      "loss": 1.4743,
+      "step": 132300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.842100814130542e-06,
+      "loss": 1.2668,
+      "step": 132305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.841413563833642e-06,
+      "loss": 1.2811,
+      "step": 132310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.840726337284506e-06,
+      "loss": 1.3026,
+      "step": 132315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8400391344870624e-06,
+      "loss": 1.4666,
+      "step": 132320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.839351955445231e-06,
+      "loss": 1.1427,
+      "step": 132325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.838664800162935e-06,
+      "loss": 0.8196,
+      "step": 132330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.837977668644106e-06,
+      "loss": 1.2605,
+      "step": 132335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.837290560892659e-06,
+      "loss": 1.1585,
+      "step": 132340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.836603476912526e-06,
+      "loss": 1.2404,
+      "step": 132345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.835916416707622e-06,
+      "loss": 1.2074,
+      "step": 132350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.83522938028188e-06,
+      "loss": 1.0556,
+      "step": 132355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.834542367639217e-06,
+      "loss": 1.2422,
+      "step": 132360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.833855378783549e-06,
+      "loss": 1.4093,
+      "step": 132365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8331684137188176e-06,
+      "loss": 1.1936,
+      "step": 132370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8324814724489344e-06,
+      "loss": 1.4231,
+      "step": 132375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.831794554977818e-06,
+      "loss": 1.5447,
+      "step": 132380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.831107661309403e-06,
+      "loss": 1.2426,
+      "step": 132385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.830420791447601e-06,
+      "loss": 1.117,
+      "step": 132390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.829733945396339e-06,
+      "loss": 1.206,
+      "step": 132395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8290471231595416e-06,
+      "loss": 1.5483,
+      "step": 132400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.828360324741129e-06,
+      "loss": 1.56,
+      "step": 132405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.827673550145021e-06,
+      "loss": 1.3889,
+      "step": 132410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8269867993751385e-06,
+      "loss": 1.2081,
+      "step": 132415
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.826300072435411e-06,
+      "loss": 1.1359,
+      "step": 132420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8256133693297544e-06,
+      "loss": 1.1928,
+      "step": 132425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8249266900620875e-06,
+      "loss": 1.0838,
+      "step": 132430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.824240034636334e-06,
+      "loss": 1.0692,
+      "step": 132435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.82355340305642e-06,
+      "loss": 1.0985,
+      "step": 132440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.822866795326258e-06,
+      "loss": 1.8523,
+      "step": 132445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.822180211449777e-06,
+      "loss": 1.5655,
+      "step": 132450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.821493651430892e-06,
+      "loss": 1.1045,
+      "step": 132455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.820807115273524e-06,
+      "loss": 1.0102,
+      "step": 132460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.820120602981601e-06,
+      "loss": 1.0151,
+      "step": 132465
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8194341145590346e-06,
+      "loss": 1.4302,
+      "step": 132470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8187476500097465e-06,
+      "loss": 1.2711,
+      "step": 132475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.81806120933766e-06,
+      "loss": 1.6617,
+      "step": 132480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.817374792546691e-06,
+      "loss": 1.4049,
+      "step": 132485
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.816688399640764e-06,
+      "loss": 1.2983,
+      "step": 132490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.816002030623793e-06,
+      "loss": 1.1978,
+      "step": 132495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.815315685499705e-06,
+      "loss": 1.415,
+      "step": 132500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.814629364272413e-06,
+      "loss": 1.5585,
+      "step": 132505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.813943066945836e-06,
+      "loss": 1.1631,
+      "step": 132510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8132567935239e-06,
+      "loss": 1.4607,
+      "step": 132515
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.812570544010519e-06,
+      "loss": 0.9803,
+      "step": 132520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.811884318409609e-06,
+      "loss": 1.3525,
+      "step": 132525
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.811198116725093e-06,
+      "loss": 1.2,
+      "step": 132530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.810511938960892e-06,
+      "loss": 1.2136,
+      "step": 132535
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.809825785120917e-06,
+      "loss": 0.9734,
+      "step": 132540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.809139655209095e-06,
+      "loss": 1.551,
+      "step": 132545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8084535492293345e-06,
+      "loss": 1.2336,
+      "step": 132550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.807767467185563e-06,
+      "loss": 1.523,
+      "step": 132555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.807081409081692e-06,
+      "loss": 1.1862,
+      "step": 132560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8063953749216425e-06,
+      "loss": 1.2698,
+      "step": 132565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.805709364709333e-06,
+      "loss": 1.3091,
+      "step": 132570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.80502337844867e-06,
+      "loss": 1.2326,
+      "step": 132575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.804337416143588e-06,
+      "loss": 1.6149,
+      "step": 132580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.803651477797996e-06,
+      "loss": 1.326,
+      "step": 132585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.802965563415807e-06,
+      "loss": 1.3026,
+      "step": 132590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.802279673000946e-06,
+      "loss": 1.274,
+      "step": 132595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.801593806557322e-06,
+      "loss": 1.212,
+      "step": 132600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.800907964088856e-06,
+      "loss": 1.26,
+      "step": 132605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.8002221455994665e-06,
+      "loss": 1.5684,
+      "step": 132610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.799536351093069e-06,
+      "loss": 1.1914,
+      "step": 132615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.798850580573573e-06,
+      "loss": 1.1111,
+      "step": 132620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7981648340448995e-06,
+      "loss": 1.4776,
+      "step": 132625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.797479111510971e-06,
+      "loss": 1.2393,
+      "step": 132630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.796793412975694e-06,
+      "loss": 1.1737,
+      "step": 132635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.796107738442986e-06,
+      "loss": 1.0653,
+      "step": 132640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.795422087916762e-06,
+      "loss": 1.3229,
+      "step": 132645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7947364614009445e-06,
+      "loss": 1.057,
+      "step": 132650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.794050858899439e-06,
+      "loss": 1.4344,
+      "step": 132655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.79336528041617e-06,
+      "loss": 1.2511,
+      "step": 132660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.792679725955047e-06,
+      "loss": 1.3575,
+      "step": 132665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.791994195519979e-06,
+      "loss": 1.145,
+      "step": 132670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.791308689114894e-06,
+      "loss": 1.1729,
+      "step": 132675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7906232067437e-06,
+      "loss": 1.0849,
+      "step": 132680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.789937748410308e-06,
+      "loss": 1.0711,
+      "step": 132685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.78925231411864e-06,
+      "loss": 1.1646,
+      "step": 132690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.788566903872601e-06,
+      "loss": 1.5488,
+      "step": 132695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.787881517676115e-06,
+      "loss": 1.4829,
+      "step": 132700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.787196155533088e-06,
+      "loss": 1.5246,
+      "step": 132705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.786510817447439e-06,
+      "loss": 1.3542,
+      "step": 132710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.785825503423077e-06,
+      "loss": 1.4637,
+      "step": 132715
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.785140213463918e-06,
+      "loss": 1.1542,
+      "step": 132720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.784454947573878e-06,
+      "loss": 1.0847,
+      "step": 132725
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.783769705756867e-06,
+      "loss": 1.3507,
+      "step": 132730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.783084488016797e-06,
+      "loss": 1.1777,
+      "step": 132735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.782399294357581e-06,
+      "loss": 1.1418,
+      "step": 132740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.781714124783139e-06,
+      "loss": 1.2736,
+      "step": 132745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7810289792973725e-06,
+      "loss": 1.3326,
+      "step": 132750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.780343857904205e-06,
+      "loss": 1.1877,
+      "step": 132755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.779658760607538e-06,
+      "loss": 1.0312,
+      "step": 132760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.778973687411291e-06,
+      "loss": 1.4561,
+      "step": 132765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.778288638319378e-06,
+      "loss": 1.0149,
+      "step": 132770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.777603613335707e-06,
+      "loss": 1.2363,
+      "step": 132775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.776918612464191e-06,
+      "loss": 1.1792,
+      "step": 132780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.776233635708737e-06,
+      "loss": 1.2087,
+      "step": 132785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.775548683073261e-06,
+      "loss": 1.665,
+      "step": 132790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7748637545616785e-06,
+      "loss": 1.1324,
+      "step": 132795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.774178850177892e-06,
+      "loss": 1.2016,
+      "step": 132800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.773493969925821e-06,
+      "loss": 1.1669,
+      "step": 132805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7728091138093675e-06,
+      "loss": 1.1322,
+      "step": 132810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.77212428183245e-06,
+      "loss": 1.2289,
+      "step": 132815
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.771439473998979e-06,
+      "loss": 1.4174,
+      "step": 132820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.770754690312863e-06,
+      "loss": 1.2249,
+      "step": 132825
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.770069930778008e-06,
+      "loss": 1.001,
+      "step": 132830
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.769385195398331e-06,
+      "loss": 1.6247,
+      "step": 132835
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.768700484177743e-06,
+      "loss": 1.0803,
+      "step": 132840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.76801579712015e-06,
+      "loss": 1.2522,
+      "step": 132845
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.767331134229459e-06,
+      "loss": 0.9745,
+      "step": 132850
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.766646495509585e-06,
+      "loss": 1.0439,
+      "step": 132855
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.76596188096444e-06,
+      "loss": 1.311,
+      "step": 132860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.765277290597926e-06,
+      "loss": 1.0837,
+      "step": 132865
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7645927244139595e-06,
+      "loss": 1.0331,
+      "step": 132870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.763908182416447e-06,
+      "loss": 1.1649,
+      "step": 132875
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.76322366460929e-06,
+      "loss": 1.9477,
+      "step": 132880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.762539170996412e-06,
+      "loss": 1.289,
+      "step": 132885
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.761854701581715e-06,
+      "loss": 1.3414,
+      "step": 132890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.761170256369102e-06,
+      "loss": 1.2504,
+      "step": 132895
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.760485835362492e-06,
+      "loss": 1.3304,
+      "step": 132900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.759801438565783e-06,
+      "loss": 1.2572,
+      "step": 132905
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.759117065982889e-06,
+      "loss": 1.2025,
+      "step": 132910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7584327176177214e-06,
+      "loss": 1.4698,
+      "step": 132915
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.757748393474184e-06,
+      "loss": 1.3506,
+      "step": 132920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.757064093556182e-06,
+      "loss": 1.662,
+      "step": 132925
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.756379817867625e-06,
+      "loss": 1.4528,
+      "step": 132930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.755695566412425e-06,
+      "loss": 1.0169,
+      "step": 132935
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.755011339194487e-06,
+      "loss": 1.5919,
+      "step": 132940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.754327136217714e-06,
+      "loss": 1.2173,
+      "step": 132945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.753642957486017e-06,
+      "loss": 1.0717,
+      "step": 132950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7529588030033055e-06,
+      "loss": 1.032,
+      "step": 132955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.752274672773479e-06,
+      "loss": 1.3217,
+      "step": 132960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.751590566800454e-06,
+      "loss": 1.0927,
+      "step": 132965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.750906485088132e-06,
+      "loss": 1.113,
+      "step": 132970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.750222427640411e-06,
+      "loss": 1.5018,
+      "step": 132975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.749538394461214e-06,
+      "loss": 1.2804,
+      "step": 132980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.748854385554437e-06,
+      "loss": 0.9094,
+      "step": 132985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.748170400923989e-06,
+      "loss": 1.3249,
+      "step": 132990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.747486440573772e-06,
+      "loss": 1.034,
+      "step": 132995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.746802504507693e-06,
+      "loss": 1.3551,
+      "step": 133000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.746118592729665e-06,
+      "loss": 1.203,
+      "step": 133005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.745434705243582e-06,
+      "loss": 1.0263,
+      "step": 133010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.744750842053361e-06,
+      "loss": 1.1661,
+      "step": 133015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.744067003162897e-06,
+      "loss": 1.399,
+      "step": 133020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.743383188576098e-06,
+      "loss": 1.0592,
+      "step": 133025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7426993982968755e-06,
+      "loss": 1.1005,
+      "step": 133030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.74201563232913e-06,
+      "loss": 1.1692,
+      "step": 133035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7413318906767605e-06,
+      "loss": 1.6391,
+      "step": 133040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.740648173343676e-06,
+      "loss": 1.4202,
+      "step": 133045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.739964480333786e-06,
+      "loss": 1.2834,
+      "step": 133050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.739280811650986e-06,
+      "loss": 1.6892,
+      "step": 133055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.738597167299188e-06,
+      "loss": 1.2745,
+      "step": 133060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.737913547282292e-06,
+      "loss": 1.6587,
+      "step": 133065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7372299516041986e-06,
+      "loss": 1.2418,
+      "step": 133070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7365463802688145e-06,
+      "loss": 1.2436,
+      "step": 133075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.735862833280048e-06,
+      "loss": 1.3633,
+      "step": 133080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.735179310641797e-06,
+      "loss": 1.0956,
+      "step": 133085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.734495812357963e-06,
+      "loss": 1.1975,
+      "step": 133090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.733812338432453e-06,
+      "loss": 1.0478,
+      "step": 133095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.733128888869172e-06,
+      "loss": 1.2615,
+      "step": 133100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.732445463672016e-06,
+      "loss": 1.0244,
+      "step": 133105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7317620628448964e-06,
+      "loss": 1.4681,
+      "step": 133110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.731078686391707e-06,
+      "loss": 1.2935,
+      "step": 133115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.730395334316353e-06,
+      "loss": 1.1028,
+      "step": 133120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.729712006622743e-06,
+      "loss": 1.177,
+      "step": 133125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.729028703314774e-06,
+      "loss": 0.9706,
+      "step": 133130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.728345424396344e-06,
+      "loss": 1.288,
+      "step": 133135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.727662169871361e-06,
+      "loss": 1.6178,
+      "step": 133140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.726978939743727e-06,
+      "loss": 1.066,
+      "step": 133145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.726295734017342e-06,
+      "loss": 1.135,
+      "step": 133150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.725612552696102e-06,
+      "loss": 1.0385,
+      "step": 133155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7249293957839155e-06,
+      "loss": 1.0572,
+      "step": 133160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.724246263284685e-06,
+      "loss": 1.5453,
+      "step": 133165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.723563155202303e-06,
+      "loss": 1.1936,
+      "step": 133170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.72288007154068e-06,
+      "loss": 1.5619,
+      "step": 133175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.722197012303712e-06,
+      "loss": 1.3843,
+      "step": 133180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.721513977495297e-06,
+      "loss": 1.1744,
+      "step": 133185
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7208309671193365e-06,
+      "loss": 0.9383,
+      "step": 133190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.720147981179738e-06,
+      "loss": 1.2137,
+      "step": 133195
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.719465019680392e-06,
+      "loss": 1.252,
+      "step": 133200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.718782082625207e-06,
+      "loss": 1.531,
+      "step": 133205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7180991700180745e-06,
+      "loss": 0.958,
+      "step": 133210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.717416281862904e-06,
+      "loss": 1.4018,
+      "step": 133215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.716733418163585e-06,
+      "loss": 1.1384,
+      "step": 133220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.716050578924026e-06,
+      "loss": 1.3633,
+      "step": 133225
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.715367764148119e-06,
+      "loss": 1.2163,
+      "step": 133230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7146849738397655e-06,
+      "loss": 1.3013,
+      "step": 133235
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.714002208002871e-06,
+      "loss": 1.3236,
+      "step": 133240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7133194666413286e-06,
+      "loss": 1.3861,
+      "step": 133245
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7126367497590325e-06,
+      "loss": 1.0596,
+      "step": 133250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.711954057359888e-06,
+      "loss": 1.2092,
+      "step": 133255
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.711271389447797e-06,
+      "loss": 1.4636,
+      "step": 133260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7105887460266464e-06,
+      "loss": 1.2431,
+      "step": 133265
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.709906127100346e-06,
+      "loss": 1.4305,
+      "step": 133270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7092235326727895e-06,
+      "loss": 1.3308,
+      "step": 133275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7085409627478714e-06,
+      "loss": 1.3445,
+      "step": 133280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.707858417329492e-06,
+      "loss": 1.3948,
+      "step": 133285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.707175896421553e-06,
+      "loss": 1.3121,
+      "step": 133290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.706493400027948e-06,
+      "loss": 1.2946,
+      "step": 133295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.705810928152573e-06,
+      "loss": 1.2557,
+      "step": 133300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7051284807993244e-06,
+      "loss": 1.6066,
+      "step": 133305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.704446057972108e-06,
+      "loss": 1.0867,
+      "step": 133310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.703763659674812e-06,
+      "loss": 0.9886,
+      "step": 133315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.703081285911338e-06,
+      "loss": 1.2136,
+      "step": 133320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7023989366855785e-06,
+      "loss": 1.2106,
+      "step": 133325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.701716612001431e-06,
+      "loss": 1.5291,
+      "step": 133330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.7010343118627985e-06,
+      "loss": 1.2519,
+      "step": 133335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.700352036273573e-06,
+      "loss": 1.2814,
+      "step": 133340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.699669785237645e-06,
+      "loss": 1.249,
+      "step": 133345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.698987558758915e-06,
+      "loss": 1.1899,
+      "step": 133350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.698305356841285e-06,
+      "loss": 1.2535,
+      "step": 133355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.697623179488644e-06,
+      "loss": 1.2452,
+      "step": 133360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6969410267048856e-06,
+      "loss": 1.1303,
+      "step": 133365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.696258898493911e-06,
+      "loss": 1.3483,
+      "step": 133370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.695576794859609e-06,
+      "loss": 1.2699,
+      "step": 133375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.694894715805879e-06,
+      "loss": 1.0151,
+      "step": 133380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.69421266133662e-06,
+      "loss": 1.0424,
+      "step": 133385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.693530631455721e-06,
+      "loss": 1.4457,
+      "step": 133390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.692848626167077e-06,
+      "loss": 1.4937,
+      "step": 133395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.692166645474582e-06,
+      "loss": 1.1734,
+      "step": 133400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6914846893821375e-06,
+      "loss": 0.9644,
+      "step": 133405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.690802757893628e-06,
+      "loss": 1.3405,
+      "step": 133410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6901208510129565e-06,
+      "loss": 1.4278,
+      "step": 133415
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.68943896874401e-06,
+      "loss": 1.2758,
+      "step": 133420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.688757111090688e-06,
+      "loss": 1.3857,
+      "step": 133425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.688075278056879e-06,
+      "loss": 1.2469,
+      "step": 133430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6873934696464825e-06,
+      "loss": 1.3604,
+      "step": 133435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.686711685863385e-06,
+      "loss": 1.1419,
+      "step": 133440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.686029926711484e-06,
+      "loss": 1.3028,
+      "step": 133445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.685348192194676e-06,
+      "loss": 0.9976,
+      "step": 133450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.684666482316851e-06,
+      "loss": 1.0884,
+      "step": 133455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.683984797081896e-06,
+      "loss": 1.3319,
+      "step": 133460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.683303136493712e-06,
+      "loss": 1.2922,
+      "step": 133465
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.68262150055619e-06,
+      "loss": 1.1973,
+      "step": 133470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.681939889273218e-06,
+      "loss": 1.5162,
+      "step": 133475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.681258302648697e-06,
+      "loss": 1.2615,
+      "step": 133480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.680576740686513e-06,
+      "loss": 1.1645,
+      "step": 133485
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.679895203390555e-06,
+      "loss": 1.5487,
+      "step": 133490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.679213690764719e-06,
+      "loss": 1.1583,
+      "step": 133495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6785322028129005e-06,
+      "loss": 1.4418,
+      "step": 133500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.677850739538988e-06,
+      "loss": 1.3506,
+      "step": 133505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.677169300946867e-06,
+      "loss": 1.261,
+      "step": 133510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.676487887040436e-06,
+      "loss": 1.1764,
+      "step": 133515
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.675806497823588e-06,
+      "loss": 1.1236,
+      "step": 133520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.675125133300208e-06,
+      "loss": 1.3703,
+      "step": 133525
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.674443793474191e-06,
+      "loss": 1.4274,
+      "step": 133530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.673762478349425e-06,
+      "loss": 1.1117,
+      "step": 133535
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.673081187929801e-06,
+      "loss": 1.1226,
+      "step": 133540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.672399922219215e-06,
+      "loss": 1.372,
+      "step": 133545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.671718681221552e-06,
+      "loss": 1.1305,
+      "step": 133550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6710374649407e-06,
+      "loss": 1.3849,
+      "step": 133555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6703562733805525e-06,
+      "loss": 0.9713,
+      "step": 133560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.669675106545004e-06,
+      "loss": 0.9755,
+      "step": 133565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.668993964437935e-06,
+      "loss": 1.4871,
+      "step": 133570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.668312847063244e-06,
+      "loss": 0.9923,
+      "step": 133575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.667631754424816e-06,
+      "loss": 1.5785,
+      "step": 133580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.666950686526538e-06,
+      "loss": 1.1058,
+      "step": 133585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.666269643372301e-06,
+      "loss": 1.6926,
+      "step": 133590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.665588624965999e-06,
+      "loss": 0.974,
+      "step": 133595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.664907631311518e-06,
+      "loss": 1.257,
+      "step": 133600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.664226662412742e-06,
+      "loss": 1.1093,
+      "step": 133605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.663545718273565e-06,
+      "loss": 1.1385,
+      "step": 133610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.662864798897876e-06,
+      "loss": 1.4443,
+      "step": 133615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.662183904289559e-06,
+      "loss": 1.1453,
+      "step": 133620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.661503034452509e-06,
+      "loss": 1.2507,
+      "step": 133625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.660822189390604e-06,
+      "loss": 1.1627,
+      "step": 133630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.660141369107741e-06,
+      "loss": 0.8473,
+      "step": 133635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.659460573607809e-06,
+      "loss": 1.2963,
+      "step": 133640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.65877980289469e-06,
+      "loss": 0.968,
+      "step": 133645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.658099056972274e-06,
+      "loss": 1.2777,
+      "step": 133650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.65741833584444e-06,
+      "loss": 1.508,
+      "step": 133655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.656737639515091e-06,
+      "loss": 1.1456,
+      "step": 133660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.656056967988105e-06,
+      "loss": 1.3434,
+      "step": 133665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.655376321267368e-06,
+      "loss": 1.2067,
+      "step": 133670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.654695699356771e-06,
+      "loss": 1.0517,
+      "step": 133675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.654015102260195e-06,
+      "loss": 1.3556,
+      "step": 133680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6533345299815316e-06,
+      "loss": 1.1019,
+      "step": 133685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6526539825246686e-06,
+      "loss": 1.0283,
+      "step": 133690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.651973459893489e-06,
+      "loss": 1.4007,
+      "step": 133695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.651292962091878e-06,
+      "loss": 1.2429,
+      "step": 133700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6506124891237215e-06,
+      "loss": 0.9788,
+      "step": 133705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.649932040992912e-06,
+      "loss": 1.1669,
+      "step": 133710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.649251617703324e-06,
+      "loss": 1.6543,
+      "step": 133715
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.648571219258856e-06,
+      "loss": 1.3306,
+      "step": 133720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.64789084566338e-06,
+      "loss": 1.0882,
+      "step": 133725
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.647210496920794e-06,
+      "loss": 1.5276,
+      "step": 133730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.646530173034974e-06,
+      "loss": 1.383,
+      "step": 133735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.64584987400981e-06,
+      "loss": 1.1267,
+      "step": 133740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.645169599849183e-06,
+      "loss": 1.1327,
+      "step": 133745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.644489350556979e-06,
+      "loss": 1.2347,
+      "step": 133750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6438091261370876e-06,
+      "loss": 1.1559,
+      "step": 133755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6431289265933884e-06,
+      "loss": 1.0987,
+      "step": 133760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.642448751929763e-06,
+      "loss": 0.7937,
+      "step": 133765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.641768602150104e-06,
+      "loss": 1.1793,
+      "step": 133770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.641088477258284e-06,
+      "loss": 1.0935,
+      "step": 133775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.640408377258196e-06,
+      "loss": 0.9779,
+      "step": 133780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.639728302153722e-06,
+      "loss": 1.2887,
+      "step": 133785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.639048251948747e-06,
+      "loss": 1.1996,
+      "step": 133790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6383682266471465e-06,
+      "loss": 1.1358,
+      "step": 133795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.63768822625281e-06,
+      "loss": 1.3566,
+      "step": 133800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.637008250769625e-06,
+      "loss": 1.3288,
+      "step": 133805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.636328300201468e-06,
+      "loss": 1.2076,
+      "step": 133810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.635648374552219e-06,
+      "loss": 1.5494,
+      "step": 133815
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.634968473825766e-06,
+      "loss": 1.2278,
+      "step": 133820
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6342885980259945e-06,
+      "loss": 1.2674,
+      "step": 133825
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.63360874715678e-06,
+      "loss": 1.1325,
+      "step": 133830
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.632928921222012e-06,
+      "loss": 1.257,
+      "step": 133835
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6322491202255635e-06,
+      "loss": 1.2266,
+      "step": 133840
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.631569344171323e-06,
+      "loss": 1.0751,
+      "step": 133845
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.630889593063175e-06,
+      "loss": 1.1791,
+      "step": 133850
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.630209866904996e-06,
+      "loss": 1.0544,
+      "step": 133855
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.629530165700667e-06,
+      "loss": 1.0586,
+      "step": 133860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.62885048945407e-06,
+      "loss": 1.9635,
+      "step": 133865
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6281708381690915e-06,
+      "loss": 1.0127,
+      "step": 133870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.627491211849609e-06,
+      "loss": 1.0294,
+      "step": 133875
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.626811610499501e-06,
+      "loss": 1.2866,
+      "step": 133880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.626132034122653e-06,
+      "loss": 1.1318,
+      "step": 133885
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.625452482722941e-06,
+      "loss": 1.0629,
+      "step": 133890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.624772956304247e-06,
+      "loss": 1.2445,
+      "step": 133895
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.624093454870457e-06,
+      "loss": 1.2368,
+      "step": 133900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.623413978425447e-06,
+      "loss": 1.2746,
+      "step": 133905
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.622734526973094e-06,
+      "loss": 1.2577,
+      "step": 133910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.622055100517281e-06,
+      "loss": 1.0187,
+      "step": 133915
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.62137569906189e-06,
+      "loss": 1.1698,
+      "step": 133920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6206963226107965e-06,
+      "loss": 1.1596,
+      "step": 133925
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.620016971167886e-06,
+      "loss": 1.1721,
+      "step": 133930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.619337644737031e-06,
+      "loss": 1.082,
+      "step": 133935
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6186583433221184e-06,
+      "loss": 1.5638,
+      "step": 133940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.617979066927019e-06,
+      "loss": 1.1907,
+      "step": 133945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.617299815555619e-06,
+      "loss": 1.4025,
+      "step": 133950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6166205892117944e-06,
+      "loss": 1.0503,
+      "step": 133955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.615941387899417e-06,
+      "loss": 1.0442,
+      "step": 133960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.615262211622379e-06,
+      "loss": 0.9387,
+      "step": 133965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.614583060384553e-06,
+      "loss": 1.2845,
+      "step": 133970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.613903934189813e-06,
+      "loss": 1.0682,
+      "step": 133975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.613224833042044e-06,
+      "loss": 0.8651,
+      "step": 133980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6125457569451156e-06,
+      "loss": 1.0791,
+      "step": 133985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.611866705902911e-06,
+      "loss": 1.1565,
+      "step": 133990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6111876799193124e-06,
+      "loss": 1.3545,
+      "step": 133995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.610508678998192e-06,
+      "loss": 1.261,
+      "step": 134000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.609829703143425e-06,
+      "loss": 1.1318,
+      "step": 134005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.609150752358891e-06,
+      "loss": 1.5776,
+      "step": 134010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.608471826648471e-06,
+      "loss": 1.1708,
+      "step": 134015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.60779292601604e-06,
+      "loss": 1.2298,
+      "step": 134020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.60711405046547e-06,
+      "loss": 1.5255,
+      "step": 134025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.606435200000642e-06,
+      "loss": 1.1661,
+      "step": 134030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.605756374625434e-06,
+      "loss": 1.2178,
+      "step": 134035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6050775743437184e-06,
+      "loss": 1.3253,
+      "step": 134040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6043987991593765e-06,
+      "loss": 1.4878,
+      "step": 134045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6037200490762825e-06,
+      "loss": 1.3012,
+      "step": 134050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6030413240983015e-06,
+      "loss": 1.0493,
+      "step": 134055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.60236262422933e-06,
+      "loss": 1.192,
+      "step": 134060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.6016839494732324e-06,
+      "loss": 1.2289,
+      "step": 134065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.601005299833881e-06,
+      "loss": 1.1454,
+      "step": 134070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.60032667531516e-06,
+      "loss": 1.3802,
+      "step": 134075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.599648075920935e-06,
+      "loss": 1.3136,
+      "step": 134080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5989695016550905e-06,
+      "loss": 1.1513,
+      "step": 134085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.598290952521492e-06,
+      "loss": 1.1586,
+      "step": 134090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5976124285240264e-06,
+      "loss": 1.5579,
+      "step": 134095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.596933929666556e-06,
+      "loss": 1.2152,
+      "step": 134100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.596255455952961e-06,
+      "loss": 1.4001,
+      "step": 134105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.595577007387121e-06,
+      "loss": 1.158,
+      "step": 134110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.594898583972904e-06,
+      "loss": 1.1714,
+      "step": 134115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.594220185714182e-06,
+      "loss": 1.31,
+      "step": 134120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.593541812614832e-06,
+      "loss": 1.4656,
+      "step": 134125
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5928634646787325e-06,
+      "loss": 1.2109,
+      "step": 134130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.59218514190975e-06,
+      "loss": 1.2211,
+      "step": 134135
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.591506844311763e-06,
+      "loss": 1.1921,
+      "step": 134140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.59082857188864e-06,
+      "loss": 1.4686,
+      "step": 134145
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5901503246442576e-06,
+      "loss": 1.3586,
+      "step": 134150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.589472102582493e-06,
+      "loss": 1.0599,
+      "step": 134155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.588793905707215e-06,
+      "loss": 1.21,
+      "step": 134160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.588115734022294e-06,
+      "loss": 1.4826,
+      "step": 134165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.587437587531603e-06,
+      "loss": 1.3413,
+      "step": 134170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.586759466239018e-06,
+      "loss": 1.2032,
+      "step": 134175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.586081370148413e-06,
+      "loss": 1.5825,
+      "step": 134180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.585403299263652e-06,
+      "loss": 1.0665,
+      "step": 134185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5847252535886185e-06,
+      "loss": 1.4879,
+      "step": 134190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5840472331271735e-06,
+      "loss": 1.1977,
+      "step": 134195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5833692378831935e-06,
+      "loss": 1.058,
+      "step": 134200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.582691267860556e-06,
+      "loss": 1.3542,
+      "step": 134205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.582013323063127e-06,
+      "loss": 1.1192,
+      "step": 134210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.581335403494774e-06,
+      "loss": 1.1207,
+      "step": 134215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5806575091593715e-06,
+      "loss": 1.1174,
+      "step": 134220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.579979640060796e-06,
+      "loss": 1.0737,
+      "step": 134225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.579301796202914e-06,
+      "loss": 1.386,
+      "step": 134230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.578623977589594e-06,
+      "loss": 1.4874,
+      "step": 134235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.577946184224707e-06,
+      "loss": 1.1053,
+      "step": 134240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.577268416112132e-06,
+      "loss": 1.0748,
+      "step": 134245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5765906732557265e-06,
+      "loss": 1.31,
+      "step": 134250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.575912955659373e-06,
+      "loss": 1.0944,
+      "step": 134255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5752352633269345e-06,
+      "loss": 1.2385,
+      "step": 134260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.574557596262277e-06,
+      "loss": 1.2132,
+      "step": 134265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.573879954469281e-06,
+      "loss": 1.1526,
+      "step": 134270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.573202337951813e-06,
+      "loss": 1.0936,
+      "step": 134275
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.572524746713736e-06,
+      "loss": 1.2501,
+      "step": 134280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.571847180758929e-06,
+      "loss": 1.2692,
+      "step": 134285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.571169640091252e-06,
+      "loss": 1.0089,
+      "step": 134290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.570492124714577e-06,
+      "loss": 1.0715,
+      "step": 134295
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.569814634632779e-06,
+      "loss": 1.1319,
+      "step": 134300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.569137169849723e-06,
+      "loss": 1.3007,
+      "step": 134305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.568459730369272e-06,
+      "loss": 1.4175,
+      "step": 134310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5677823161953005e-06,
+      "loss": 1.3592,
+      "step": 134315
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.567104927331679e-06,
+      "loss": 1.2911,
+      "step": 134320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.566427563782273e-06,
+      "loss": 1.0184,
+      "step": 134325
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.565750225550945e-06,
+      "loss": 1.2919,
+      "step": 134330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.56507291264157e-06,
+      "loss": 1.058,
+      "step": 134335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.564395625058017e-06,
+      "loss": 1.0123,
+      "step": 134340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.563718362804147e-06,
+      "loss": 1.0492,
+      "step": 134345
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.563041125883836e-06,
+      "loss": 1.2867,
+      "step": 134350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.562363914300945e-06,
+      "loss": 1.8045,
+      "step": 134355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.561686728059335e-06,
+      "loss": 1.5174,
+      "step": 134360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.561009567162891e-06,
+      "loss": 1.3906,
+      "step": 134365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.560332431615467e-06,
+      "loss": 1.2808,
+      "step": 134370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.559655321420935e-06,
+      "loss": 1.1959,
+      "step": 134375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.558978236583155e-06,
+      "loss": 1.2371,
+      "step": 134380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.558301177105997e-06,
+      "loss": 1.0902,
+      "step": 134385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.557624142993332e-06,
+      "loss": 1.169,
+      "step": 134390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.55694713424902e-06,
+      "loss": 1.2217,
+      "step": 134395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.556270150876932e-06,
+      "loss": 1.2935,
+      "step": 134400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.555593192880928e-06,
+      "loss": 1.1758,
+      "step": 134405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5549162602648774e-06,
+      "loss": 1.3031,
+      "step": 134410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.55423935303265e-06,
+      "loss": 1.278,
+      "step": 134415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.553562471188108e-06,
+      "loss": 1.357,
+      "step": 134420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.55288561473511e-06,
+      "loss": 0.974,
+      "step": 134425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5522087836775274e-06,
+      "loss": 1.1557,
+      "step": 134430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.551531978019229e-06,
+      "loss": 1.2263,
+      "step": 134435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5508551977640725e-06,
+      "loss": 1.2976,
+      "step": 134440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.550178442915929e-06,
+      "loss": 1.0967,
+      "step": 134445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5495017134786555e-06,
+      "loss": 1.1715,
+      "step": 134450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.548825009456126e-06,
+      "loss": 1.0431,
+      "step": 134455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.548148330852196e-06,
+      "loss": 1.3953,
+      "step": 134460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.547471677670736e-06,
+      "loss": 1.2904,
+      "step": 134465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.546795049915607e-06,
+      "loss": 1.2301,
+      "step": 134470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5461184475906715e-06,
+      "loss": 1.0405,
+      "step": 134475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.545441870699795e-06,
+      "loss": 1.096,
+      "step": 134480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.544765319246846e-06,
+      "loss": 1.1776,
+      "step": 134485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.544088793235677e-06,
+      "loss": 1.2731,
+      "step": 134490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5434122926701626e-06,
+      "loss": 1.379,
+      "step": 134495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.542735817554157e-06,
+      "loss": 1.1881,
+      "step": 134500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.542059367891527e-06,
+      "loss": 1.4736,
+      "step": 134505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5413829436861395e-06,
+      "loss": 1.3265,
+      "step": 134510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.540706544941854e-06,
+      "loss": 1.0143,
+      "step": 134515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5400301716625294e-06,
+      "loss": 1.2096,
+      "step": 134520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.539353823852029e-06,
+      "loss": 1.2564,
+      "step": 134525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.538677501514223e-06,
+      "loss": 1.3063,
+      "step": 134530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.538001204652968e-06,
+      "loss": 1.2571,
+      "step": 134535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.537324933272122e-06,
+      "loss": 1.1828,
+      "step": 134540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.536648687375552e-06,
+      "loss": 1.107,
+      "step": 134545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.535972466967121e-06,
+      "loss": 1.0574,
+      "step": 134550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.535296272050684e-06,
+      "loss": 1.0894,
+      "step": 134555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.534620102630112e-06,
+      "loss": 1.2628,
+      "step": 134560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5339439587092606e-06,
+      "loss": 1.1537,
+      "step": 134565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.533267840291983e-06,
+      "loss": 1.1208,
+      "step": 134570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.532591747382156e-06,
+      "loss": 1.2067,
+      "step": 134575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.531915679983632e-06,
+      "loss": 1.477,
+      "step": 134580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.53123963810027e-06,
+      "loss": 1.4376,
+      "step": 134585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.530563621735938e-06,
+      "loss": 1.1185,
+      "step": 134590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.529887630894486e-06,
+      "loss": 0.7203,
+      "step": 134595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5292116655797845e-06,
+      "loss": 1.095,
+      "step": 134600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.528535725795685e-06,
+      "loss": 1.3414,
+      "step": 134605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.527859811546056e-06,
+      "loss": 1.3482,
+      "step": 134610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.527183922834749e-06,
+      "loss": 1.1315,
+      "step": 134615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5265080596656265e-06,
+      "loss": 1.1681,
+      "step": 134620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.525832222042553e-06,
+      "loss": 1.0676,
+      "step": 134625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.525156409969386e-06,
+      "loss": 1.1358,
+      "step": 134630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.524480623449976e-06,
+      "loss": 1.2196,
+      "step": 134635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.52380486248819e-06,
+      "loss": 1.1658,
+      "step": 134640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.52312912708789e-06,
+      "loss": 1.3052,
+      "step": 134645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5224534172529265e-06,
+      "loss": 1.0304,
+      "step": 134650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5217777329871656e-06,
+      "loss": 1.1211,
+      "step": 134655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.521102074294462e-06,
+      "loss": 1.1843,
+      "step": 134660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5204264411786725e-06,
+      "loss": 0.9334,
+      "step": 134665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.519750833643656e-06,
+      "loss": 1.1804,
+      "step": 134670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.519075251693276e-06,
+      "loss": 0.9438,
+      "step": 134675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.518399695331387e-06,
+      "loss": 1.4048,
+      "step": 134680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.517724164561842e-06,
+      "loss": 1.438,
+      "step": 134685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.517048659388502e-06,
+      "loss": 1.3279,
+      "step": 134690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.516373179815231e-06,
+      "loss": 1.3309,
+      "step": 134695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.515697725845875e-06,
+      "loss": 1.0477,
+      "step": 134700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.515022297484302e-06,
+      "loss": 1.113,
+      "step": 134705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.514346894734359e-06,
+      "loss": 1.1615,
+      "step": 134710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.513671517599908e-06,
+      "loss": 1.238,
+      "step": 134715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5129961660848094e-06,
+      "loss": 1.0984,
+      "step": 134720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.512320840192918e-06,
+      "loss": 1.4947,
+      "step": 134725
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.511645539928082e-06,
+      "loss": 1.3604,
+      "step": 134730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5109702652941655e-06,
+      "loss": 1.3584,
+      "step": 134735
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.510295016295027e-06,
+      "loss": 1.0834,
+      "step": 134740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5096197929345174e-06,
+      "loss": 1.1881,
+      "step": 134745
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.508944595216492e-06,
+      "loss": 0.9413,
+      "step": 134750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.5082694231448104e-06,
+      "loss": 1.1221,
+      "step": 134755
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.507594276723323e-06,
+      "loss": 1.268,
+      "step": 134760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.50691915595589e-06,
+      "loss": 1.2057,
+      "step": 134765
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.506244060846369e-06,
+      "loss": 1.3018,
+      "step": 134770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.505568991398609e-06,
+      "loss": 1.1667,
+      "step": 134775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.504893947616465e-06,
+      "loss": 1.2638,
+      "step": 134780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.504218929503794e-06,
+      "loss": 1.0911,
+      "step": 134785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.503543937064455e-06,
+      "loss": 1.057,
+      "step": 134790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.502868970302295e-06,
+      "loss": 1.1943,
+      "step": 134795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.502194029221175e-06,
+      "loss": 1.1939,
+      "step": 134800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.501519113824943e-06,
+      "loss": 1.0129,
+      "step": 134805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.500844224117461e-06,
+      "loss": 0.9633,
+      "step": 134810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.500169360102574e-06,
+      "loss": 1.2875,
+      "step": 134815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.499494521784144e-06,
+      "loss": 1.3495,
+      "step": 134820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.498819709166016e-06,
+      "loss": 1.5161,
+      "step": 134825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.498144922252049e-06,
+      "loss": 1.2961,
+      "step": 134830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4974701610461e-06,
+      "loss": 1.3218,
+      "step": 134835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.496795425552018e-06,
+      "loss": 1.3401,
+      "step": 134840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.496120715773652e-06,
+      "loss": 1.1316,
+      "step": 134845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.495446031714859e-06,
+      "loss": 1.1486,
+      "step": 134850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4947713733794976e-06,
+      "loss": 1.4528,
+      "step": 134855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.494096740771408e-06,
+      "loss": 1.108,
+      "step": 134860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.493422133894455e-06,
+      "loss": 1.3916,
+      "step": 134865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.492747552752487e-06,
+      "loss": 1.077,
+      "step": 134870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.492072997349349e-06,
+      "loss": 0.9581,
+      "step": 134875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.491398467688898e-06,
+      "loss": 1.1397,
+      "step": 134880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.490723963774992e-06,
+      "loss": 1.4399,
+      "step": 134885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.490049485611477e-06,
+      "loss": 1.1732,
+      "step": 134890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.489375033202202e-06,
+      "loss": 1.0092,
+      "step": 134895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.488700606551021e-06,
+      "loss": 1.0381,
+      "step": 134900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.488026205661789e-06,
+      "loss": 1.1608,
+      "step": 134905
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.487351830538352e-06,
+      "loss": 1.1392,
+      "step": 134910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.486677481184565e-06,
+      "loss": 1.2604,
+      "step": 134915
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.486003157604275e-06,
+      "loss": 1.2601,
+      "step": 134920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.485328859801334e-06,
+      "loss": 1.3543,
+      "step": 134925
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.484654587779598e-06,
+      "loss": 1.212,
+      "step": 134930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.483980341542912e-06,
+      "loss": 1.1176,
+      "step": 134935
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4833061210951245e-06,
+      "loss": 1.0086,
+      "step": 134940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.482631926440088e-06,
+      "loss": 1.1505,
+      "step": 134945
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.481957757581655e-06,
+      "loss": 1.4262,
+      "step": 134950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.481283614523675e-06,
+      "loss": 1.1857,
+      "step": 134955
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.480609497269991e-06,
+      "loss": 1.2138,
+      "step": 134960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.479935405824462e-06,
+      "loss": 0.9862,
+      "step": 134965
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.479261340190929e-06,
+      "loss": 1.3927,
+      "step": 134970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4785873003732455e-06,
+      "loss": 1.2463,
+      "step": 134975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.477913286375265e-06,
+      "loss": 1.303,
+      "step": 134980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.477239298200831e-06,
+      "loss": 0.9957,
+      "step": 134985
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.476565335853789e-06,
+      "loss": 1.2854,
+      "step": 134990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.475891399337992e-06,
+      "loss": 1.0521,
+      "step": 134995
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.475217488657293e-06,
+      "loss": 1.3096,
+      "step": 135000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.474543603815531e-06,
+      "loss": 1.1171,
+      "step": 135005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.473869744816563e-06,
+      "loss": 0.9672,
+      "step": 135010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.473195911664229e-06,
+      "loss": 1.1517,
+      "step": 135015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.472522104362383e-06,
+      "loss": 1.0064,
+      "step": 135020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.471848322914872e-06,
+      "loss": 1.3963,
+      "step": 135025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.471174567325545e-06,
+      "loss": 1.1648,
+      "step": 135030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.470500837598242e-06,
+      "loss": 1.1511,
+      "step": 135035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.469827133736816e-06,
+      "loss": 1.279,
+      "step": 135040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4691534557451174e-06,
+      "loss": 1.3648,
+      "step": 135045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.468479803626989e-06,
+      "loss": 1.3702,
+      "step": 135050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.467806177386274e-06,
+      "loss": 1.0177,
+      "step": 135055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.467132577026828e-06,
+      "loss": 1.4357,
+      "step": 135060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4664590025524885e-06,
+      "loss": 1.1166,
+      "step": 135065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.465785453967107e-06,
+      "loss": 1.3589,
+      "step": 135070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.465111931274534e-06,
+      "loss": 1.2993,
+      "step": 135075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.464438434478609e-06,
+      "loss": 1.0842,
+      "step": 135080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.463764963583178e-06,
+      "loss": 1.2617,
+      "step": 135085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.463091518592087e-06,
+      "loss": 1.1457,
+      "step": 135090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.462418099509189e-06,
+      "loss": 1.3884,
+      "step": 135095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4617447063383235e-06,
+      "loss": 1.3295,
+      "step": 135100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.461071339083333e-06,
+      "loss": 0.9906,
+      "step": 135105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.460397997748067e-06,
+      "loss": 1.4629,
+      "step": 135110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.459724682336374e-06,
+      "loss": 1.5527,
+      "step": 135115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.45905139285209e-06,
+      "loss": 1.3688,
+      "step": 135120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4583781292990686e-06,
+      "loss": 1.1935,
+      "step": 135125
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.457704891681148e-06,
+      "loss": 1.0457,
+      "step": 135130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.457031680002176e-06,
+      "loss": 1.1692,
+      "step": 135135
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.456358494265998e-06,
+      "loss": 1.2368,
+      "step": 135140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.455685334476459e-06,
+      "loss": 1.0593,
+      "step": 135145
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.455012200637397e-06,
+      "loss": 1.3264,
+      "step": 135150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.454339092752659e-06,
+      "loss": 0.9711,
+      "step": 135155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4536660108260944e-06,
+      "loss": 1.2017,
+      "step": 135160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.452992954861538e-06,
+      "loss": 1.1777,
+      "step": 135165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.452319924862841e-06,
+      "loss": 1.349,
+      "step": 135170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.451646920833843e-06,
+      "loss": 1.3069,
+      "step": 135175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.450973942778384e-06,
+      "loss": 1.2585,
+      "step": 135180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4503009907003105e-06,
+      "loss": 1.3768,
+      "step": 135185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.449628064603469e-06,
+      "loss": 1.7419,
+      "step": 135190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.448955164491699e-06,
+      "loss": 1.3946,
+      "step": 135195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.448282290368838e-06,
+      "loss": 1.1694,
+      "step": 135200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.447609442238735e-06,
+      "loss": 0.9951,
+      "step": 135205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.446936620105232e-06,
+      "loss": 1.4353,
+      "step": 135210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.446263823972168e-06,
+      "loss": 1.0419,
+      "step": 135215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4455910538433895e-06,
+      "loss": 0.986,
+      "step": 135220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.444918309722733e-06,
+      "loss": 0.9609,
+      "step": 135225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4442455916140415e-06,
+      "loss": 1.4704,
+      "step": 135230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.443572899521163e-06,
+      "loss": 0.8461,
+      "step": 135235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.442900233447934e-06,
+      "loss": 1.1417,
+      "step": 135240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.442227593398195e-06,
+      "loss": 1.3017,
+      "step": 135245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.44155497937578e-06,
+      "loss": 1.0591,
+      "step": 135250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.440882391384546e-06,
+      "loss": 1.2897,
+      "step": 135255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.440209829428327e-06,
+      "loss": 1.0731,
+      "step": 135260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.439537293510957e-06,
+      "loss": 1.2521,
+      "step": 135265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.438864783636285e-06,
+      "loss": 1.0992,
+      "step": 135270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.438192299808145e-06,
+      "loss": 1.2732,
+      "step": 135275
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.437519842030381e-06,
+      "loss": 1.4459,
+      "step": 135280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.436847410306837e-06,
+      "loss": 1.1361,
+      "step": 135285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.436175004641347e-06,
+      "loss": 1.3918,
+      "step": 135290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.435502625037749e-06,
+      "loss": 1.4596,
+      "step": 135295
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.434830271499888e-06,
+      "loss": 1.1725,
+      "step": 135300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.434157944031604e-06,
+      "loss": 1.1336,
+      "step": 135305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.433485642636729e-06,
+      "loss": 1.0052,
+      "step": 135310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.432813367319112e-06,
+      "loss": 1.2935,
+      "step": 135315
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4321411180825835e-06,
+      "loss": 1.1009,
+      "step": 135320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4314688949309905e-06,
+      "loss": 1.2964,
+      "step": 135325
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.430796697868161e-06,
+      "loss": 1.1144,
+      "step": 135330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4301245268979465e-06,
+      "loss": 1.0637,
+      "step": 135335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.429452382024177e-06,
+      "loss": 1.1223,
+      "step": 135340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.428780263250686e-06,
+      "loss": 1.1793,
+      "step": 135345
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.428108170581326e-06,
+      "loss": 1.2911,
+      "step": 135350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.427436104019928e-06,
+      "loss": 1.3533,
+      "step": 135355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4267640635703234e-06,
+      "loss": 1.4554,
+      "step": 135360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.42609204923636e-06,
+      "loss": 1.1617,
+      "step": 135365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.425420061021869e-06,
+      "loss": 1.4417,
+      "step": 135370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.424748098930688e-06,
+      "loss": 1.2212,
+      "step": 135375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.42407616296666e-06,
+      "loss": 1.4329,
+      "step": 135380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.423404253133619e-06,
+      "loss": 1.0848,
+      "step": 135385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.422732369435398e-06,
+      "loss": 1.1674,
+      "step": 135390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.422060511875838e-06,
+      "loss": 1.0486,
+      "step": 135395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.421388680458776e-06,
+      "loss": 1.3176,
+      "step": 135400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.420716875188048e-06,
+      "loss": 1.3083,
+      "step": 135405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4200450960674855e-06,
+      "loss": 0.9871,
+      "step": 135410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.41937334310093e-06,
+      "loss": 1.2326,
+      "step": 135415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.41870161629222e-06,
+      "loss": 1.2892,
+      "step": 135420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.418029915645184e-06,
+      "loss": 1.2523,
+      "step": 135425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.417358241163665e-06,
+      "loss": 1.3847,
+      "step": 135430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4166865928514915e-06,
+      "loss": 1.3428,
+      "step": 135435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.416014970712502e-06,
+      "loss": 1.135,
+      "step": 135440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.415343374750537e-06,
+      "loss": 1.0512,
+      "step": 135445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.414671804969428e-06,
+      "loss": 1.3833,
+      "step": 135450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.414000261373005e-06,
+      "loss": 1.4201,
+      "step": 135455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4133287439651094e-06,
+      "loss": 1.1452,
+      "step": 135460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.412657252749571e-06,
+      "loss": 1.3686,
+      "step": 135465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.411985787730232e-06,
+      "loss": 1.2646,
+      "step": 135470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.411314348910916e-06,
+      "loss": 1.2521,
+      "step": 135475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.410642936295468e-06,
+      "loss": 1.1689,
+      "step": 135480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.409971549887713e-06,
+      "loss": 1.2665,
+      "step": 135485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.409300189691489e-06,
+      "loss": 1.2259,
+      "step": 135490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.408628855710635e-06,
+      "loss": 1.2191,
+      "step": 135495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4079575479489784e-06,
+      "loss": 1.2924,
+      "step": 135500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.407286266410351e-06,
+      "loss": 1.4465,
+      "step": 135505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.406615011098588e-06,
+      "loss": 1.3702,
+      "step": 135510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.40594378201753e-06,
+      "loss": 1.3023,
+      "step": 135515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4052725791709985e-06,
+      "loss": 1.1116,
+      "step": 135520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.404601402562836e-06,
+      "loss": 1.2681,
+      "step": 135525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.4039302521968675e-06,
+      "loss": 1.2746,
+      "step": 135530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.403259128076932e-06,
+      "loss": 1.1106,
+      "step": 135535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.402588030206857e-06,
+      "loss": 1.2674,
+      "step": 135540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.40191695859048e-06,
+      "loss": 1.3346,
+      "step": 135545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.40124591323163e-06,
+      "loss": 1.183,
+      "step": 135550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.400574894134132e-06,
+      "loss": 1.2526,
+      "step": 135555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.399903901301833e-06,
+      "loss": 1.1851,
+      "step": 135560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.399232934738556e-06,
+      "loss": 1.3099,
+      "step": 135565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.398561994448131e-06,
+      "loss": 1.2182,
+      "step": 135570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.397891080434395e-06,
+      "loss": 0.9754,
+      "step": 135575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.397220192701172e-06,
+      "loss": 1.175,
+      "step": 135580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.396549331252298e-06,
+      "loss": 1.5824,
+      "step": 135585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.395878496091607e-06,
+      "loss": 1.5962,
+      "step": 135590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3952076872229264e-06,
+      "loss": 1.202,
+      "step": 135595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.394536904650083e-06,
+      "loss": 1.0301,
+      "step": 135600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.393866148376909e-06,
+      "loss": 1.0408,
+      "step": 135605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.393195418407242e-06,
+      "loss": 1.0124,
+      "step": 135610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.392524714744908e-06,
+      "loss": 1.2394,
+      "step": 135615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3918540373937314e-06,
+      "loss": 1.0618,
+      "step": 135620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.391183386357546e-06,
+      "loss": 1.6377,
+      "step": 135625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.390512761640186e-06,
+      "loss": 1.2807,
+      "step": 135630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.389842163245474e-06,
+      "loss": 1.3191,
+      "step": 135635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.389171591177247e-06,
+      "loss": 1.1554,
+      "step": 135640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.388501045439329e-06,
+      "loss": 1.4939,
+      "step": 135645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.387830526035544e-06,
+      "loss": 0.9396,
+      "step": 135650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.387160032969735e-06,
+      "loss": 1.2551,
+      "step": 135655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.386489566245723e-06,
+      "loss": 1.7136,
+      "step": 135660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.385819125867333e-06,
+      "loss": 1.7165,
+      "step": 135665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.385148711838401e-06,
+      "loss": 2.0574,
+      "step": 135670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3844783241627485e-06,
+      "loss": 1.1161,
+      "step": 135675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.383807962844212e-06,
+      "loss": 1.0683,
+      "step": 135680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.38313762788661e-06,
+      "loss": 1.0422,
+      "step": 135685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.382467319293778e-06,
+      "loss": 1.0459,
+      "step": 135690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.381797037069538e-06,
+      "loss": 1.6755,
+      "step": 135695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.381126781217723e-06,
+      "loss": 1.1247,
+      "step": 135700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.380456551742159e-06,
+      "loss": 1.2341,
+      "step": 135705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.379786348646675e-06,
+      "loss": 0.9963,
+      "step": 135710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3791161719350915e-06,
+      "loss": 1.0569,
+      "step": 135715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.37844602161124e-06,
+      "loss": 1.369,
+      "step": 135720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.377775897678952e-06,
+      "loss": 0.9844,
+      "step": 135725
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.377105800142044e-06,
+      "loss": 1.164,
+      "step": 135730
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.376435729004353e-06,
+      "loss": 1.2914,
+      "step": 135735
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.375765684269701e-06,
+      "loss": 1.2623,
+      "step": 135740
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.375095665941908e-06,
+      "loss": 1.2027,
+      "step": 135745
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.37442567402481e-06,
+      "loss": 1.3514,
+      "step": 135750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.373755708522232e-06,
+      "loss": 1.4144,
+      "step": 135755
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.373085769437998e-06,
+      "loss": 1.4255,
+      "step": 135760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3724158567759274e-06,
+      "loss": 1.3565,
+      "step": 135765
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.371745970539851e-06,
+      "loss": 1.1252,
+      "step": 135770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.371076110733599e-06,
+      "loss": 1.0785,
+      "step": 135775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.370406277360989e-06,
+      "loss": 1.0471,
+      "step": 135780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.369736470425852e-06,
+      "loss": 1.0045,
+      "step": 135785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.369066689932005e-06,
+      "loss": 0.9895,
+      "step": 135790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.36839693588328e-06,
+      "loss": 1.0795,
+      "step": 135795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.367727208283503e-06,
+      "loss": 1.3451,
+      "step": 135800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.367057507136495e-06,
+      "loss": 1.2024,
+      "step": 135805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.366387832446076e-06,
+      "loss": 1.2408,
+      "step": 135810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.365718184216076e-06,
+      "loss": 1.1627,
+      "step": 135815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.365048562450321e-06,
+      "loss": 1.1927,
+      "step": 135820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.364378967152627e-06,
+      "loss": 1.1796,
+      "step": 135825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.363709398326826e-06,
+      "loss": 1.4231,
+      "step": 135830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.363039855976736e-06,
+      "loss": 1.2556,
+      "step": 135835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.362370340106186e-06,
+      "loss": 1.0269,
+      "step": 135840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3617008507189905e-06,
+      "loss": 1.1683,
+      "step": 135845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.361031387818982e-06,
+      "loss": 1.1235,
+      "step": 135850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.36036195140998e-06,
+      "loss": 1.2841,
+      "step": 135855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.359692541495802e-06,
+      "loss": 1.3746,
+      "step": 135860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.359023158080277e-06,
+      "loss": 1.333,
+      "step": 135865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.358353801167229e-06,
+      "loss": 1.1882,
+      "step": 135870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3576844707604714e-06,
+      "loss": 1.1364,
+      "step": 135875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.35701516686384e-06,
+      "loss": 1.7849,
+      "step": 135880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3563458894811424e-06,
+      "loss": 1.379,
+      "step": 135885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3556766386162076e-06,
+      "loss": 1.1901,
+      "step": 135890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.355007414272861e-06,
+      "loss": 1.2632,
+      "step": 135895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.354338216454922e-06,
+      "loss": 1.2849,
+      "step": 135900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.353669045166205e-06,
+      "loss": 1.1451,
+      "step": 135905
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3529999004105385e-06,
+      "loss": 1.4541,
+      "step": 135910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.352330782191745e-06,
+      "loss": 1.1648,
+      "step": 135915
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.351661690513642e-06,
+      "loss": 1.1878,
+      "step": 135920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.350992625380047e-06,
+      "loss": 1.2452,
+      "step": 135925
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3503235867947855e-06,
+      "loss": 1.392,
+      "step": 135930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.34965457476168e-06,
+      "loss": 1.3414,
+      "step": 135935
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3489855892845456e-06,
+      "loss": 1.3348,
+      "step": 135940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.348316630367207e-06,
+      "loss": 1.315,
+      "step": 135945
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3476476980134825e-06,
+      "loss": 1.0879,
+      "step": 135950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.346978792227187e-06,
+      "loss": 1.1199,
+      "step": 135955
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.34630991301215e-06,
+      "loss": 1.4811,
+      "step": 135960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3456410603721875e-06,
+      "loss": 1.5896,
+      "step": 135965
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.344972234311115e-06,
+      "loss": 1.1819,
+      "step": 135970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.344303434832756e-06,
+      "loss": 1.4301,
+      "step": 135975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.343634661940926e-06,
+      "loss": 1.5793,
+      "step": 135980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.342965915639451e-06,
+      "loss": 0.8821,
+      "step": 135985
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.342297195932141e-06,
+      "loss": 1.1258,
+      "step": 135990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.341628502822823e-06,
+      "loss": 1.7177,
+      "step": 135995
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.340959836315308e-06,
+      "loss": 1.2034,
+      "step": 136000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.340291196413418e-06,
+      "loss": 1.1915,
+      "step": 136005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.339622583120976e-06,
+      "loss": 1.1711,
+      "step": 136010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.338953996441796e-06,
+      "loss": 1.7165,
+      "step": 136015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.338285436379691e-06,
+      "loss": 1.3595,
+      "step": 136020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.337616902938483e-06,
+      "loss": 1.4686,
+      "step": 136025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.336948396121995e-06,
+      "loss": 1.1737,
+      "step": 136030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.336279915934035e-06,
+      "loss": 0.997,
+      "step": 136035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.335611462378429e-06,
+      "loss": 1.248,
+      "step": 136040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.33494303545899e-06,
+      "loss": 1.0338,
+      "step": 136045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.334274635179532e-06,
+      "loss": 1.5194,
+      "step": 136050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.333606261543875e-06,
+      "loss": 1.0085,
+      "step": 136055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.33293791455584e-06,
+      "loss": 1.2212,
+      "step": 136060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.332269594219241e-06,
+      "loss": 1.4041,
+      "step": 136065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.331601300537889e-06,
+      "loss": 1.3512,
+      "step": 136070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.330933033515603e-06,
+      "loss": 1.3415,
+      "step": 136075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.330264793156206e-06,
+      "loss": 1.4116,
+      "step": 136080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.329596579463503e-06,
+      "loss": 1.0951,
+      "step": 136085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3289283924413215e-06,
+      "loss": 1.2044,
+      "step": 136090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.328260232093465e-06,
+      "loss": 1.3674,
+      "step": 136095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.327592098423757e-06,
+      "loss": 1.1835,
+      "step": 136100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3269239914360145e-06,
+      "loss": 0.8544,
+      "step": 136105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3262559111340506e-06,
+      "loss": 1.8527,
+      "step": 136110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.325587857521675e-06,
+      "loss": 1.1361,
+      "step": 136115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.324919830602705e-06,
+      "loss": 1.1501,
+      "step": 136120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.324251830380962e-06,
+      "loss": 1.0555,
+      "step": 136125
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.323583856860257e-06,
+      "loss": 1.1389,
+      "step": 136130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.3229159100443996e-06,
+      "loss": 1.3218,
+      "step": 136135
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.322247989937206e-06,
+      "loss": 1.2304,
+      "step": 136140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.321580096542498e-06,
+      "loss": 1.2407,
+      "step": 136145
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.320912229864079e-06,
+      "loss": 1.047,
+      "step": 136150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3202443899057724e-06,
+      "loss": 1.2881,
+      "step": 136155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.319576576671387e-06,
+      "loss": 1.1582,
+      "step": 136160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.318908790164733e-06,
+      "loss": 1.2727,
+      "step": 136165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.318241030389627e-06,
+      "loss": 1.0809,
+      "step": 136170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.317573297349887e-06,
+      "loss": 1.5507,
+      "step": 136175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.31690559104932e-06,
+      "loss": 1.2607,
+      "step": 136180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.316237911491742e-06,
+      "loss": 1.2442,
+      "step": 136185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.315570258680962e-06,
+      "loss": 1.0201,
+      "step": 136190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.314902632620799e-06,
+      "loss": 1.1992,
+      "step": 136195
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3142350333150575e-06,
+      "loss": 1.6804,
+      "step": 136200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.313567460767559e-06,
+      "loss": 1.1617,
+      "step": 136205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.312899914982106e-06,
+      "loss": 1.3985,
+      "step": 136210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.312232395962515e-06,
+      "loss": 1.2608,
+      "step": 136215
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.311564903712605e-06,
+      "loss": 1.4784,
+      "step": 136220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.310897438236179e-06,
+      "loss": 1.6523,
+      "step": 136225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3102299995370465e-06,
+      "loss": 1.0613,
+      "step": 136230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.309562587619023e-06,
+      "loss": 1.1703,
+      "step": 136235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.308895202485926e-06,
+      "loss": 1.1401,
+      "step": 136240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.308227844141556e-06,
+      "loss": 1.1906,
+      "step": 136245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.307560512589731e-06,
+      "loss": 1.3697,
+      "step": 136250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.306893207834259e-06,
+      "loss": 1.1551,
+      "step": 136255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3062259298789445e-06,
+      "loss": 1.1462,
+      "step": 136260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.305558678727611e-06,
+      "loss": 1.2909,
+      "step": 136265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.304891454384065e-06,
+      "loss": 1.2599,
+      "step": 136270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.304224256852111e-06,
+      "loss": 1.3576,
+      "step": 136275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.30355708613556e-06,
+      "loss": 1.2401,
+      "step": 136280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.302889942238223e-06,
+      "loss": 1.3188,
+      "step": 136285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.302222825163916e-06,
+      "loss": 1.2953,
+      "step": 136290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.301555734916438e-06,
+      "loss": 0.987,
+      "step": 136295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3008886714996084e-06,
+      "loss": 1.3462,
+      "step": 136300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.300221634917228e-06,
+      "loss": 1.5447,
+      "step": 136305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.29955462517311e-06,
+      "loss": 1.4367,
+      "step": 136310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.298887642271067e-06,
+      "loss": 1.5579,
+      "step": 136315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.298220686214904e-06,
+      "loss": 1.0241,
+      "step": 136320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.297553757008427e-06,
+      "loss": 1.5103,
+      "step": 136325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.296886854655445e-06,
+      "loss": 1.5796,
+      "step": 136330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.296219979159773e-06,
+      "loss": 1.302,
+      "step": 136335
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.295553130525217e-06,
+      "loss": 1.3148,
+      "step": 136340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.294886308755578e-06,
+      "loss": 1.4337,
+      "step": 136345
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.294219513854672e-06,
+      "loss": 1.1437,
+      "step": 136350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2935527458263006e-06,
+      "loss": 1.2037,
+      "step": 136355
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.292886004674273e-06,
+      "loss": 1.3422,
+      "step": 136360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.292219290402403e-06,
+      "loss": 1.3334,
+      "step": 136365
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.291552603014492e-06,
+      "loss": 1.1691,
+      "step": 136370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.290885942514345e-06,
+      "loss": 1.3646,
+      "step": 136375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.290219308905773e-06,
+      "loss": 1.2306,
+      "step": 136380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.289552702192584e-06,
+      "loss": 1.4298,
+      "step": 136385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.28888612237858e-06,
+      "loss": 1.6185,
+      "step": 136390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.288219569467574e-06,
+      "loss": 1.3348,
+      "step": 136395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2875530434633635e-06,
+      "loss": 1.4097,
+      "step": 136400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.28688654436976e-06,
+      "loss": 1.2622,
+      "step": 136405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.286220072190573e-06,
+      "loss": 1.056,
+      "step": 136410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.285553626929606e-06,
+      "loss": 1.314,
+      "step": 136415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.284887208590659e-06,
+      "loss": 1.2795,
+      "step": 136420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.284220817177541e-06,
+      "loss": 1.3562,
+      "step": 136425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.283554452694064e-06,
+      "loss": 1.1131,
+      "step": 136430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.282888115144027e-06,
+      "loss": 1.3536,
+      "step": 136435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.282221804531233e-06,
+      "loss": 1.3641,
+      "step": 136440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.281555520859493e-06,
+      "loss": 1.1227,
+      "step": 136445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.280889264132604e-06,
+      "loss": 1.348,
+      "step": 136450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.280223034354377e-06,
+      "loss": 1.5426,
+      "step": 136455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.279556831528618e-06,
+      "loss": 1.4215,
+      "step": 136460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.278890655659129e-06,
+      "loss": 1.1713,
+      "step": 136465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.278224506749709e-06,
+      "loss": 1.1305,
+      "step": 136470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.277558384804168e-06,
+      "loss": 1.1201,
+      "step": 136475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.276892289826312e-06,
+      "loss": 1.108,
+      "step": 136480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.27622622181994e-06,
+      "loss": 1.2353,
+      "step": 136485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.275560180788854e-06,
+      "loss": 1.2344,
+      "step": 136490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.274894166736861e-06,
+      "loss": 1.0668,
+      "step": 136495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.274228179667768e-06,
+      "loss": 1.0068,
+      "step": 136500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2735622195853684e-06,
+      "loss": 1.52,
+      "step": 136505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.272896286493475e-06,
+      "loss": 1.4939,
+      "step": 136510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.272230380395883e-06,
+      "loss": 1.1958,
+      "step": 136515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2715645012963975e-06,
+      "loss": 1.1674,
+      "step": 136520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.270898649198827e-06,
+      "loss": 1.0348,
+      "step": 136525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.270232824106968e-06,
+      "loss": 1.322,
+      "step": 136530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.269567026024619e-06,
+      "loss": 1.2778,
+      "step": 136535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.268901254955587e-06,
+      "loss": 1.1907,
+      "step": 136540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.268235510903678e-06,
+      "loss": 1.4833,
+      "step": 136545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.267569793872685e-06,
+      "loss": 1.1258,
+      "step": 136550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.266904103866417e-06,
+      "loss": 1.1402,
+      "step": 136555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.266238440888673e-06,
+      "loss": 1.2179,
+      "step": 136560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.265572804943251e-06,
+      "loss": 1.2387,
+      "step": 136565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.264907196033953e-06,
+      "loss": 1.2407,
+      "step": 136570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.264241614164584e-06,
+      "loss": 1.1328,
+      "step": 136575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2635760593389466e-06,
+      "loss": 1.4697,
+      "step": 136580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.26291053156083e-06,
+      "loss": 1.3306,
+      "step": 136585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.262245030834043e-06,
+      "loss": 0.9778,
+      "step": 136590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.261579557162391e-06,
+      "loss": 1.0216,
+      "step": 136595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.260914110549662e-06,
+      "loss": 1.3968,
+      "step": 136600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.260248690999668e-06,
+      "loss": 1.6427,
+      "step": 136605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.259583298516198e-06,
+      "loss": 1.0549,
+      "step": 136610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.258917933103056e-06,
+      "loss": 1.4482,
+      "step": 136615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.258252594764048e-06,
+      "loss": 1.1276,
+      "step": 136620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.257587283502968e-06,
+      "loss": 1.4196,
+      "step": 136625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.256921999323617e-06,
+      "loss": 1.2875,
+      "step": 136630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.256256742229781e-06,
+      "loss": 1.334,
+      "step": 136635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.255591512225281e-06,
+      "loss": 1.3235,
+      "step": 136640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.254926309313904e-06,
+      "loss": 1.1842,
+      "step": 136645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.254261133499447e-06,
+      "loss": 1.0285,
+      "step": 136650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.253595984785715e-06,
+      "loss": 1.2865,
+      "step": 136655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.252930863176499e-06,
+      "loss": 1.0461,
+      "step": 136660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.252265768675599e-06,
+      "loss": 0.9065,
+      "step": 136665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.251600701286821e-06,
+      "loss": 1.5118,
+      "step": 136670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.250935661013957e-06,
+      "loss": 1.2512,
+      "step": 136675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2502706478608e-06,
+      "loss": 1.3665,
+      "step": 136680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.249605661831153e-06,
+      "loss": 1.4238,
+      "step": 136685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.248940702928817e-06,
+      "loss": 1.1032,
+      "step": 136690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.248275771157579e-06,
+      "loss": 1.0864,
+      "step": 136695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.247610866521248e-06,
+      "loss": 1.1145,
+      "step": 136700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.246945989023611e-06,
+      "loss": 1.2851,
+      "step": 136705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.246281138668471e-06,
+      "loss": 1.2566,
+      "step": 136710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.24561631545962e-06,
+      "loss": 1.116,
+      "step": 136715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.244951519400861e-06,
+      "loss": 1.0514,
+      "step": 136720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.244286750495981e-06,
+      "loss": 1.0233,
+      "step": 136725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.243622008748783e-06,
+      "loss": 1.1728,
+      "step": 136730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.242957294163065e-06,
+      "loss": 1.6474,
+      "step": 136735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2422926067426195e-06,
+      "loss": 1.4543,
+      "step": 136740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.241627946491238e-06,
+      "loss": 1.1811,
+      "step": 136745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.240963313412723e-06,
+      "loss": 1.0035,
+      "step": 136750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.240298707510866e-06,
+      "loss": 1.1714,
+      "step": 136755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.239634128789462e-06,
+      "loss": 1.4298,
+      "step": 136760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.238969577252311e-06,
+      "loss": 1.2426,
+      "step": 136765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.238305052903205e-06,
+      "loss": 0.9451,
+      "step": 136770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.237640555745933e-06,
+      "loss": 1.3688,
+      "step": 136775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.236976085784297e-06,
+      "loss": 1.5149,
+      "step": 136780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.236311643022092e-06,
+      "loss": 1.1237,
+      "step": 136785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.23564722746311e-06,
+      "loss": 1.3209,
+      "step": 136790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.23498283911114e-06,
+      "loss": 1.3102,
+      "step": 136795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.234318477969983e-06,
+      "loss": 1.2276,
+      "step": 136800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.233654144043433e-06,
+      "loss": 1.3449,
+      "step": 136805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2329898373352785e-06,
+      "loss": 1.413,
+      "step": 136810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.232325557849318e-06,
+      "loss": 1.1633,
+      "step": 136815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.231661305589342e-06,
+      "loss": 1.2278,
+      "step": 136820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.230997080559142e-06,
+      "loss": 1.3127,
+      "step": 136825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.230332882762519e-06,
+      "loss": 1.4925,
+      "step": 136830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.22966871220326e-06,
+      "loss": 1.0575,
+      "step": 136835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.229004568885154e-06,
+      "loss": 1.2237,
+      "step": 136840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.228340452811998e-06,
+      "loss": 1.5892,
+      "step": 136845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2276763639875894e-06,
+      "loss": 1.2012,
+      "step": 136850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2270123024157145e-06,
+      "loss": 1.1809,
+      "step": 136855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.226348268100163e-06,
+      "loss": 1.0681,
+      "step": 136860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2256842610447345e-06,
+      "loss": 1.0169,
+      "step": 136865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.225020281253214e-06,
+      "loss": 1.3094,
+      "step": 136870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.224356328729394e-06,
+      "loss": 1.5041,
+      "step": 136875
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.223692403477072e-06,
+      "loss": 1.0657,
+      "step": 136880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.223028505500036e-06,
+      "loss": 1.1288,
+      "step": 136885
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.222364634802073e-06,
+      "loss": 1.2016,
+      "step": 136890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.221700791386977e-06,
+      "loss": 1.0644,
+      "step": 136895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2210369752585446e-06,
+      "loss": 1.2457,
+      "step": 136900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.220373186420555e-06,
+      "loss": 1.2021,
+      "step": 136905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.219709424876811e-06,
+      "loss": 1.2565,
+      "step": 136910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.219045690631094e-06,
+      "loss": 1.1897,
+      "step": 136915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2183819836872e-06,
+      "loss": 1.0228,
+      "step": 136920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.217718304048913e-06,
+      "loss": 1.1263,
+      "step": 136925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.217054651720031e-06,
+      "loss": 1.1594,
+      "step": 136930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.216391026704339e-06,
+      "loss": 1.4179,
+      "step": 136935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.215727429005619e-06,
+      "loss": 1.6715,
+      "step": 136940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.215063858627678e-06,
+      "loss": 1.218,
+      "step": 136945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.214400315574295e-06,
+      "loss": 1.0476,
+      "step": 136950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.213736799849256e-06,
+      "loss": 0.9992,
+      "step": 136955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.213073311456358e-06,
+      "loss": 1.3807,
+      "step": 136960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.212409850399382e-06,
+      "loss": 1.2662,
+      "step": 136965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.211746416682122e-06,
+      "loss": 1.3233,
+      "step": 136970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.211083010308369e-06,
+      "loss": 1.6293,
+      "step": 136975
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.210419631281907e-06,
+      "loss": 1.165,
+      "step": 136980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.209756279606521e-06,
+      "loss": 1.0805,
+      "step": 136985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.209092955286004e-06,
+      "loss": 1.2227,
+      "step": 136990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.208429658324146e-06,
+      "loss": 1.0978,
+      "step": 136995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.207766388724731e-06,
+      "loss": 1.1328,
+      "step": 137000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.207103146491546e-06,
+      "loss": 1.2117,
+      "step": 137005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.206439931628377e-06,
+      "loss": 1.417,
+      "step": 137010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.205776744139021e-06,
+      "loss": 1.4342,
+      "step": 137015
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.205113584027253e-06,
+      "loss": 1.1764,
+      "step": 137020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.204450451296869e-06,
+      "loss": 1.2001,
+      "step": 137025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.203787345951652e-06,
+      "loss": 1.5617,
+      "step": 137030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.203124267995382e-06,
+      "loss": 1.0041,
+      "step": 137035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.20246121743186e-06,
+      "loss": 1.2664,
+      "step": 137040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.2017981942648646e-06,
+      "loss": 1.2654,
+      "step": 137045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.201135198498179e-06,
+      "loss": 1.0984,
+      "step": 137050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.200472230135596e-06,
+      "loss": 1.14,
+      "step": 137055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.199809289180894e-06,
+      "loss": 0.987,
+      "step": 137060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.199146375637867e-06,
+      "loss": 1.2752,
+      "step": 137065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.198483489510291e-06,
+      "loss": 1.0443,
+      "step": 137070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.197820630801962e-06,
+      "loss": 1.4817,
+      "step": 137075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.197157799516657e-06,
+      "loss": 1.2485,
+      "step": 137080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.196494995658162e-06,
+      "loss": 1.1865,
+      "step": 137085
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1958322192302705e-06,
+      "loss": 1.1022,
+      "step": 137090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.195169470236759e-06,
+      "loss": 1.215,
+      "step": 137095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.19450674868141e-06,
+      "loss": 1.0571,
+      "step": 137100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.193844054568014e-06,
+      "loss": 1.2208,
+      "step": 137105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.193181387900355e-06,
+      "loss": 1.3222,
+      "step": 137110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1925187486822135e-06,
+      "loss": 1.1481,
+      "step": 137115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.191856136917379e-06,
+      "loss": 1.1859,
+      "step": 137120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.191193552609628e-06,
+      "loss": 1.1413,
+      "step": 137125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.190530995762749e-06,
+      "loss": 1.1255,
+      "step": 137130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.189868466380528e-06,
+      "loss": 0.9555,
+      "step": 137135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.189205964466746e-06,
+      "loss": 1.1746,
+      "step": 137140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.188543490025184e-06,
+      "loss": 1.4661,
+      "step": 137145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.187881043059624e-06,
+      "loss": 1.35,
+      "step": 137150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.187218623573851e-06,
+      "loss": 1.1826,
+      "step": 137155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.186556231571651e-06,
+      "loss": 1.469,
+      "step": 137160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.185893867056802e-06,
+      "loss": 1.2327,
+      "step": 137165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.18523153003309e-06,
+      "loss": 1.5663,
+      "step": 137170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.184569220504294e-06,
+      "loss": 1.2764,
+      "step": 137175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.183906938474197e-06,
+      "loss": 1.166,
+      "step": 137180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.183244683946584e-06,
+      "loss": 1.3543,
+      "step": 137185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.182582456925236e-06,
+      "loss": 0.9609,
+      "step": 137190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.181920257413929e-06,
+      "loss": 1.2297,
+      "step": 137195
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.18125808541645e-06,
+      "loss": 1.118,
+      "step": 137200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.180595940936581e-06,
+      "loss": 1.366,
+      "step": 137205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1799338239781025e-06,
+      "loss": 1.0123,
+      "step": 137210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1792717345447894e-06,
+      "loss": 1.1663,
+      "step": 137215
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.178609672640429e-06,
+      "loss": 1.2997,
+      "step": 137220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.177947638268804e-06,
+      "loss": 1.2118,
+      "step": 137225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.177285631433686e-06,
+      "loss": 1.1391,
+      "step": 137230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.176623652138867e-06,
+      "loss": 1.3226,
+      "step": 137235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.17596170038812e-06,
+      "loss": 1.1133,
+      "step": 137240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1752997761852184e-06,
+      "loss": 1.2484,
+      "step": 137245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.174637879533959e-06,
+      "loss": 0.9847,
+      "step": 137250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.173976010438112e-06,
+      "loss": 1.1695,
+      "step": 137255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.173314168901455e-06,
+      "loss": 1.2814,
+      "step": 137260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1726523549277715e-06,
+      "loss": 0.9822,
+      "step": 137265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1719905685208375e-06,
+      "loss": 1.2698,
+      "step": 137270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.171328809684433e-06,
+      "loss": 0.9743,
+      "step": 137275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.170667078422342e-06,
+      "loss": 1.2346,
+      "step": 137280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.170005374738341e-06,
+      "loss": 1.0069,
+      "step": 137285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.169343698636202e-06,
+      "loss": 1.3379,
+      "step": 137290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.168682050119709e-06,
+      "loss": 1.3313,
+      "step": 137295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.168020429192644e-06,
+      "loss": 1.2746,
+      "step": 137300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.16735883585878e-06,
+      "loss": 1.4194,
+      "step": 137305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1666972701218945e-06,
+      "loss": 0.8763,
+      "step": 137310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.166035731985765e-06,
+      "loss": 1.092,
+      "step": 137315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.165374221454177e-06,
+      "loss": 1.3809,
+      "step": 137320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.164712738530898e-06,
+      "loss": 1.4716,
+      "step": 137325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.164051283219713e-06,
+      "loss": 1.2801,
+      "step": 137330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.163389855524398e-06,
+      "loss": 1.1363,
+      "step": 137335
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.16272845544872e-06,
+      "loss": 1.0868,
+      "step": 137340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1620670829964715e-06,
+      "loss": 0.97,
+      "step": 137345
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.161405738171421e-06,
+      "loss": 1.2426,
+      "step": 137350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.160744420977348e-06,
+      "loss": 1.2017,
+      "step": 137355
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.160083131418023e-06,
+      "loss": 1.4715,
+      "step": 137360
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.159421869497225e-06,
+      "loss": 1.3592,
+      "step": 137365
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.158760635218738e-06,
+      "loss": 1.1899,
+      "step": 137370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.158099428586325e-06,
+      "loss": 1.0251,
+      "step": 137375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1574382496037745e-06,
+      "loss": 1.3808,
+      "step": 137380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.156777098274852e-06,
+      "loss": 1.2727,
+      "step": 137385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.156115974603336e-06,
+      "loss": 1.1315,
+      "step": 137390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.155454878593008e-06,
+      "loss": 1.3021,
+      "step": 137395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.154793810247638e-06,
+      "loss": 1.1138,
+      "step": 137400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.154132769570998e-06,
+      "loss": 1.0191,
+      "step": 137405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.153471756566866e-06,
+      "loss": 1.071,
+      "step": 137410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.152810771239021e-06,
+      "loss": 1.517,
+      "step": 137415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1521498135912285e-06,
+      "loss": 1.146,
+      "step": 137420
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.151488883627272e-06,
+      "loss": 1.0369,
+      "step": 137425
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.150827981350923e-06,
+      "loss": 1.0831,
+      "step": 137430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.15016710676595e-06,
+      "loss": 1.2215,
+      "step": 137435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1495062598761314e-06,
+      "loss": 1.341,
+      "step": 137440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.148845440685244e-06,
+      "loss": 1.2494,
+      "step": 137445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.148184649197059e-06,
+      "loss": 1.0623,
+      "step": 137450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1475238854153456e-06,
+      "loss": 1.2363,
+      "step": 137455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.14686314934388e-06,
+      "loss": 1.2036,
+      "step": 137460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1462024409864405e-06,
+      "loss": 1.6006,
+      "step": 137465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.145541760346791e-06,
+      "loss": 1.0968,
+      "step": 137470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1448811074287145e-06,
+      "loss": 1.1106,
+      "step": 137475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.144220482235973e-06,
+      "loss": 0.9932,
+      "step": 137480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.143559884772344e-06,
+      "loss": 0.9938,
+      "step": 137485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.142899315041605e-06,
+      "loss": 1.2379,
+      "step": 137490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.142238773047523e-06,
+      "loss": 1.2375,
+      "step": 137495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.141578258793867e-06,
+      "loss": 1.163,
+      "step": 137500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1409177722844115e-06,
+      "loss": 0.9677,
+      "step": 137505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.140257313522934e-06,
+      "loss": 1.4637,
+      "step": 137510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.139596882513199e-06,
+      "loss": 1.3278,
+      "step": 137515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.138936479258978e-06,
+      "loss": 1.0333,
+      "step": 137520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.138276103764044e-06,
+      "loss": 1.07,
+      "step": 137525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.13761575603217e-06,
+      "loss": 1.6296,
+      "step": 137530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.136955436067124e-06,
+      "loss": 1.3034,
+      "step": 137535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.13629514387268e-06,
+      "loss": 1.3605,
+      "step": 137540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.135634879452609e-06,
+      "loss": 1.3552,
+      "step": 137545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.134974642810674e-06,
+      "loss": 1.1661,
+      "step": 137550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.13431443395065e-06,
+      "loss": 1.5321,
+      "step": 137555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1336542528763125e-06,
+      "loss": 1.2017,
+      "step": 137560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.132994099591422e-06,
+      "loss": 1.2059,
+      "step": 137565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1323339740997565e-06,
+      "loss": 0.8978,
+      "step": 137570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.131673876405079e-06,
+      "loss": 1.0913,
+      "step": 137575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.131013806511165e-06,
+      "loss": 1.0201,
+      "step": 137580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.130353764421778e-06,
+      "loss": 1.073,
+      "step": 137585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.129693750140694e-06,
+      "loss": 1.1853,
+      "step": 137590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1290337636716735e-06,
+      "loss": 1.0814,
+      "step": 137595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.128373805018492e-06,
+      "loss": 1.6009,
+      "step": 137600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.127713874184918e-06,
+      "loss": 1.2177,
+      "step": 137605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.127053971174719e-06,
+      "loss": 1.3777,
+      "step": 137610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.126394095991659e-06,
+      "loss": 1.1373,
+      "step": 137615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.12573424863951e-06,
+      "loss": 0.9695,
+      "step": 137620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.125074429122044e-06,
+      "loss": 1.3961,
+      "step": 137625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.12441463744302e-06,
+      "loss": 1.1829,
+      "step": 137630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.123754873606217e-06,
+      "loss": 1.2369,
+      "step": 137635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.123095137615395e-06,
+      "loss": 1.0838,
+      "step": 137640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.12243542947432e-06,
+      "loss": 1.3727,
+      "step": 137645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.12177574918676e-06,
+      "loss": 1.2879,
+      "step": 137650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1211160967564884e-06,
+      "loss": 1.1055,
+      "step": 137655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.12045647218727e-06,
+      "loss": 1.0647,
+      "step": 137660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.119796875482864e-06,
+      "loss": 1.2238,
+      "step": 137665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.119137306647044e-06,
+      "loss": 1.3028,
+      "step": 137670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.118477765683578e-06,
+      "loss": 1.2116,
+      "step": 137675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.117818252596225e-06,
+      "loss": 0.9716,
+      "step": 137680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.117158767388761e-06,
+      "loss": 1.3388,
+      "step": 137685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1164993100649415e-06,
+      "loss": 1.0487,
+      "step": 137690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.115839880628539e-06,
+      "loss": 1.1196,
+      "step": 137695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1151804790833216e-06,
+      "loss": 1.3015,
+      "step": 137700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.11452110543305e-06,
+      "loss": 1.0195,
+      "step": 137705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1138617596814864e-06,
+      "loss": 1.3561,
+      "step": 137710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.113202441832401e-06,
+      "loss": 1.0497,
+      "step": 137715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.112543151889562e-06,
+      "loss": 1.3932,
+      "step": 137720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.11188388985673e-06,
+      "loss": 1.1861,
+      "step": 137725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.111224655737665e-06,
+      "loss": 1.4351,
+      "step": 137730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.110565449536142e-06,
+      "loss": 1.1805,
+      "step": 137735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.109906271255917e-06,
+      "loss": 1.0527,
+      "step": 137740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.109247120900755e-06,
+      "loss": 0.9088,
+      "step": 137745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.108587998474428e-06,
+      "loss": 1.421,
+      "step": 137750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.107928903980695e-06,
+      "loss": 1.1196,
+      "step": 137755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.107269837423314e-06,
+      "loss": 1.205,
+      "step": 137760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.106610798806054e-06,
+      "loss": 1.1647,
+      "step": 137765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1059517881326826e-06,
+      "loss": 1.1864,
+      "step": 137770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.105292805406954e-06,
+      "loss": 1.1616,
+      "step": 137775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.104633850632641e-06,
+      "loss": 1.1866,
+      "step": 137780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.103974923813498e-06,
+      "loss": 1.1742,
+      "step": 137785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.103316024953293e-06,
+      "loss": 1.0246,
+      "step": 137790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.102657154055785e-06,
+      "loss": 1.1543,
+      "step": 137795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.1019983111247425e-06,
+      "loss": 1.2092,
+      "step": 137800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.101339496163922e-06,
+      "loss": 1.4695,
+      "step": 137805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.100680709177085e-06,
+      "loss": 1.2956,
+      "step": 137810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.100021950168002e-06,
+      "loss": 1.221,
+      "step": 137815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.099363219140427e-06,
+      "loss": 1.163,
+      "step": 137820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.098704516098121e-06,
+      "loss": 1.351,
+      "step": 137825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0980458410448494e-06,
+      "loss": 1.3351,
+      "step": 137830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.097387193984375e-06,
+      "loss": 1.0444,
+      "step": 137835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.096728574920453e-06,
+      "loss": 1.256,
+      "step": 137840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.096069983856853e-06,
+      "loss": 1.1035,
+      "step": 137845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.095411420797331e-06,
+      "loss": 1.1169,
+      "step": 137850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.094752885745642e-06,
+      "loss": 1.2513,
+      "step": 137855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0940943787055545e-06,
+      "loss": 0.9897,
+      "step": 137860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.09343589968083e-06,
+      "loss": 1.1003,
+      "step": 137865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.092777448675224e-06,
+      "loss": 1.0514,
+      "step": 137870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.092119025692496e-06,
+      "loss": 1.3342,
+      "step": 137875
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.091460630736407e-06,
+      "loss": 1.1913,
+      "step": 137880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.090802263810722e-06,
+      "loss": 1.3057,
+      "step": 137885
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.090143924919193e-06,
+      "loss": 0.981,
+      "step": 137890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0894856140655864e-06,
+      "loss": 1.1989,
+      "step": 137895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0888273312536555e-06,
+      "loss": 1.2755,
+      "step": 137900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.088169076487161e-06,
+      "loss": 1.0478,
+      "step": 137905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0875108497698656e-06,
+      "loss": 1.1343,
+      "step": 137910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.086852651105527e-06,
+      "loss": 1.0274,
+      "step": 137915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0861944804978995e-06,
+      "loss": 1.4607,
+      "step": 137920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.085536337950742e-06,
+      "loss": 1.1723,
+      "step": 137925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.084878223467819e-06,
+      "loss": 1.1472,
+      "step": 137930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.084220137052886e-06,
+      "loss": 1.1088,
+      "step": 137935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.083562078709697e-06,
+      "loss": 1.1411,
+      "step": 137940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.082904048442017e-06,
+      "loss": 1.2574,
+      "step": 137945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.082246046253594e-06,
+      "loss": 1.29,
+      "step": 137950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0815880721481915e-06,
+      "loss": 1.31,
+      "step": 137955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.080930126129571e-06,
+      "loss": 1.0827,
+      "step": 137960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.080272208201484e-06,
+      "loss": 1.2446,
+      "step": 137965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.079614318367686e-06,
+      "loss": 1.0282,
+      "step": 137970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.078956456631936e-06,
+      "loss": 1.0878,
+      "step": 137975
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.078298622997996e-06,
+      "loss": 1.3384,
+      "step": 137980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.077640817469613e-06,
+      "loss": 1.4106,
+      "step": 137985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.076983040050554e-06,
+      "loss": 1.2062,
+      "step": 137990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.076325290744564e-06,
+      "loss": 1.3666,
+      "step": 137995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.075667569555407e-06,
+      "loss": 0.8853,
+      "step": 138000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0750098764868385e-06,
+      "loss": 1.2075,
+      "step": 138005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.074352211542614e-06,
+      "loss": 1.3641,
+      "step": 138010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.073694574726487e-06,
+      "loss": 1.3132,
+      "step": 138015
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.073036966042205e-06,
+      "loss": 1.0794,
+      "step": 138020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.07237938549354e-06,
+      "loss": 1.1083,
+      "step": 138025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.07172183308424e-06,
+      "loss": 1.0696,
+      "step": 138030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.071064308818054e-06,
+      "loss": 1.053,
+      "step": 138035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0704068126987475e-06,
+      "loss": 1.3606,
+      "step": 138040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.069749344730064e-06,
+      "loss": 1.1129,
+      "step": 138045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0690919049157635e-06,
+      "loss": 0.9727,
+      "step": 138050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0684344932596056e-06,
+      "loss": 1.2904,
+      "step": 138055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.067777109765338e-06,
+      "loss": 0.8816,
+      "step": 138060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.067119754436712e-06,
+      "loss": 1.494,
+      "step": 138065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.066462427277486e-06,
+      "loss": 1.3331,
+      "step": 138070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.065805128291416e-06,
+      "loss": 1.2362,
+      "step": 138075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.065147857482253e-06,
+      "loss": 1.3359,
+      "step": 138080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.064490614853748e-06,
+      "loss": 1.4905,
+      "step": 138085
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0638334004096545e-06,
+      "loss": 1.018,
+      "step": 138090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.063176214153731e-06,
+      "loss": 1.1086,
+      "step": 138095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.062519056089722e-06,
+      "loss": 1.2886,
+      "step": 138100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.06186192622139e-06,
+      "loss": 1.08,
+      "step": 138105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.061204824552477e-06,
+      "loss": 1.0659,
+      "step": 138110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0605477510867414e-06,
+      "loss": 1.1088,
+      "step": 138115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.05989070582794e-06,
+      "loss": 1.2318,
+      "step": 138120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.059233688779819e-06,
+      "loss": 1.27,
+      "step": 138125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.058576699946127e-06,
+      "loss": 1.2513,
+      "step": 138130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.057919739330623e-06,
+      "loss": 1.2815,
+      "step": 138135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.057262806937053e-06,
+      "loss": 1.1717,
+      "step": 138140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.056605902769169e-06,
+      "loss": 1.3929,
+      "step": 138145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.055949026830728e-06,
+      "loss": 1.018,
+      "step": 138150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.05529217912548e-06,
+      "loss": 1.0185,
+      "step": 138155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.054635359657167e-06,
+      "loss": 1.083,
+      "step": 138160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.053978568429547e-06,
+      "loss": 1.1983,
+      "step": 138165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0533218054463726e-06,
+      "loss": 1.1419,
+      "step": 138170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.052665070711391e-06,
+      "loss": 1.0592,
+      "step": 138175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.05200836422835e-06,
+      "loss": 1.295,
+      "step": 138180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.051351686001003e-06,
+      "loss": 1.1827,
+      "step": 138185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.050695036033102e-06,
+      "loss": 0.9639,
+      "step": 138190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.050038414328391e-06,
+      "loss": 1.1937,
+      "step": 138195
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.049381820890626e-06,
+      "loss": 1.2295,
+      "step": 138200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.048725255723551e-06,
+      "loss": 0.9087,
+      "step": 138205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0480687188309175e-06,
+      "loss": 1.2483,
+      "step": 138210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.047412210216478e-06,
+      "loss": 1.207,
+      "step": 138215
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.04675572988398e-06,
+      "loss": 1.0457,
+      "step": 138220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.046099277837166e-06,
+      "loss": 1.3425,
+      "step": 138225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.0454428540797895e-06,
+      "loss": 1.3991,
+      "step": 138230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.044786458615603e-06,
+      "loss": 1.6824,
+      "step": 138235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.044130091448351e-06,
+      "loss": 1.1004,
+      "step": 138240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.043473752581779e-06,
+      "loss": 1.5291,
+      "step": 138245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.042817442019641e-06,
+      "loss": 1.1626,
+      "step": 138250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.042161159765677e-06,
+      "loss": 1.2352,
+      "step": 138255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.04150490582364e-06,
+      "loss": 1.3988,
+      "step": 138260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.040848680197281e-06,
+      "loss": 1.2674,
+      "step": 138265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.040192482890343e-06,
+      "loss": 1.0791,
+      "step": 138270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.03953631390657e-06,
+      "loss": 1.1514,
+      "step": 138275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.038880173249711e-06,
+      "loss": 1.3096,
+      "step": 138280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.03822406092352e-06,
+      "loss": 1.2577,
+      "step": 138285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.037567976931734e-06,
+      "loss": 1.2181,
+      "step": 138290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.036911921278109e-06,
+      "loss": 1.3223,
+      "step": 138295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.036255893966382e-06,
+      "loss": 1.1075,
+      "step": 138300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.035599895000307e-06,
+      "loss": 0.9977,
+      "step": 138305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.034943924383622e-06,
+      "loss": 1.2225,
+      "step": 138310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.034287982120083e-06,
+      "loss": 1.324,
+      "step": 138315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.03363206821343e-06,
+      "loss": 1.064,
+      "step": 138320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0329761826674e-06,
+      "loss": 1.3419,
+      "step": 138325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0323203254857565e-06,
+      "loss": 1.2169,
+      "step": 138330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.031664496672236e-06,
+      "loss": 1.4284,
+      "step": 138335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.031008696230579e-06,
+      "loss": 1.0786,
+      "step": 138340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.030352924164539e-06,
+      "loss": 1.1155,
+      "step": 138345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.029697180477853e-06,
+      "loss": 1.01,
+      "step": 138350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.029041465174269e-06,
+      "loss": 1.1938,
+      "step": 138355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.028385778257536e-06,
+      "loss": 1.276,
+      "step": 138360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.027730119731396e-06,
+      "loss": 1.4573,
+      "step": 138365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.027074489599585e-06,
+      "loss": 1.2863,
+      "step": 138370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0264188878658535e-06,
+      "loss": 1.4215,
+      "step": 138375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.02576331453395e-06,
+      "loss": 0.9549,
+      "step": 138380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.025107769607613e-06,
+      "loss": 0.9757,
+      "step": 138385
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.024452253090583e-06,
+      "loss": 1.1593,
+      "step": 138390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.023796764986606e-06,
+      "loss": 1.132,
+      "step": 138395
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.023141305299431e-06,
+      "loss": 1.2661,
+      "step": 138400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0224858740327895e-06,
+      "loss": 1.2036,
+      "step": 138405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.021830471190438e-06,
+      "loss": 1.1498,
+      "step": 138410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0211750967761056e-06,
+      "loss": 1.0654,
+      "step": 138415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.020519750793542e-06,
+      "loss": 0.9488,
+      "step": 138420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0198644332464926e-06,
+      "loss": 1.1901,
+      "step": 138425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.019209144138697e-06,
+      "loss": 1.512,
+      "step": 138430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0185538834738914e-06,
+      "loss": 1.2227,
+      "step": 138435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.017898651255826e-06,
+      "loss": 0.851,
+      "step": 138440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0172434474882355e-06,
+      "loss": 1.041,
+      "step": 138445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.016588272174869e-06,
+      "loss": 1.1924,
+      "step": 138450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.01593312531946e-06,
+      "loss": 1.1242,
+      "step": 138455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.015278006925756e-06,
+      "loss": 1.2707,
+      "step": 138460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.014622916997494e-06,
+      "loss": 1.0196,
+      "step": 138465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.013967855538415e-06,
+      "loss": 0.9847,
+      "step": 138470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0133128225522655e-06,
+      "loss": 1.1797,
+      "step": 138475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.012657818042782e-06,
+      "loss": 1.2406,
+      "step": 138480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.012002842013701e-06,
+      "loss": 1.167,
+      "step": 138485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.011347894468765e-06,
+      "loss": 1.2202,
+      "step": 138490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.010692975411721e-06,
+      "loss": 1.3338,
+      "step": 138495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0100380848463e-06,
+      "loss": 1.25,
+      "step": 138500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.00938322277625e-06,
+      "loss": 1.5651,
+      "step": 138505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.008728389205302e-06,
+      "loss": 1.2147,
+      "step": 138510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.008073584137198e-06,
+      "loss": 1.2394,
+      "step": 138515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.007418807575683e-06,
+      "loss": 1.6433,
+      "step": 138520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0067640595244915e-06,
+      "loss": 0.9921,
+      "step": 138525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.006109339987364e-06,
+      "loss": 1.3582,
+      "step": 138530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.00545464896803e-06,
+      "loss": 1.0939,
+      "step": 138535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0047999864702436e-06,
+      "loss": 1.6186,
+      "step": 138540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0041453524977355e-06,
+      "loss": 1.4163,
+      "step": 138545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.003490747054242e-06,
+      "loss": 1.9673,
+      "step": 138550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.002836170143507e-06,
+      "loss": 1.3092,
+      "step": 138555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.002181621769262e-06,
+      "loss": 1.4477,
+      "step": 138560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.001527101935247e-06,
+      "loss": 1.4257,
+      "step": 138565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.000872610645205e-06,
+      "loss": 1.5168,
+      "step": 138570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.000218147902869e-06,
+      "loss": 1.053,
+      "step": 138575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.999563713711972e-06,
+      "loss": 1.211,
+      "step": 138580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.998909308076257e-06,
+      "loss": 1.4281,
+      "step": 138585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.998254930999462e-06,
+      "loss": 1.3208,
+      "step": 138590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.997600582485323e-06,
+      "loss": 1.3969,
+      "step": 138595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.996946262537571e-06,
+      "loss": 1.1602,
+      "step": 138600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.996291971159947e-06,
+      "loss": 1.2549,
+      "step": 138605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.99563770835619e-06,
+      "loss": 1.0543,
+      "step": 138610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.994983474130029e-06,
+      "loss": 1.1059,
+      "step": 138615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.99432926848521e-06,
+      "loss": 1.4714,
+      "step": 138620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.993675091425462e-06,
+      "loss": 1.0092,
+      "step": 138625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.993020942954515e-06,
+      "loss": 1.2474,
+      "step": 138630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.992366823076118e-06,
+      "loss": 1.1683,
+      "step": 138635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.991712731794001e-06,
+      "loss": 1.0893,
+      "step": 138640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.991058669111893e-06,
+      "loss": 1.0195,
+      "step": 138645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.990404635033539e-06,
+      "loss": 1.1619,
+      "step": 138650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.989750629562664e-06,
+      "loss": 1.2486,
+      "step": 138655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.98909665270301e-06,
+      "loss": 1.0928,
+      "step": 138660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.98844270445831e-06,
+      "loss": 1.2107,
+      "step": 138665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9877887848322995e-06,
+      "loss": 1.1935,
+      "step": 138670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.987134893828707e-06,
+      "loss": 1.264,
+      "step": 138675
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9864810314512694e-06,
+      "loss": 0.8644,
+      "step": 138680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.985827197703727e-06,
+      "loss": 1.7775,
+      "step": 138685
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.985173392589806e-06,
+      "loss": 1.3387,
+      "step": 138690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.98451961611324e-06,
+      "loss": 1.1644,
+      "step": 138695
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.983865868277764e-06,
+      "loss": 1.6116,
+      "step": 138700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.983212149087116e-06,
+      "loss": 0.9496,
+      "step": 138705
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.982558458545022e-06,
+      "loss": 1.2315,
+      "step": 138710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.98190479665522e-06,
+      "loss": 1.2214,
+      "step": 138715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.981251163421441e-06,
+      "loss": 1.0579,
+      "step": 138720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.98059755884741e-06,
+      "loss": 1.0852,
+      "step": 138725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9799439829368735e-06,
+      "loss": 1.2142,
+      "step": 138730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.979290435693558e-06,
+      "loss": 1.3042,
+      "step": 138735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9786369171211935e-06,
+      "loss": 1.2685,
+      "step": 138740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.97798342722351e-06,
+      "loss": 1.0243,
+      "step": 138745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.977329966004242e-06,
+      "loss": 1.2607,
+      "step": 138750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.976676533467126e-06,
+      "loss": 1.0105,
+      "step": 138755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9760231296158826e-06,
+      "loss": 1.1757,
+      "step": 138760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9753697544542545e-06,
+      "loss": 0.8926,
+      "step": 138765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.974716407985964e-06,
+      "loss": 1.3213,
+      "step": 138770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.974063090214746e-06,
+      "loss": 1.1438,
+      "step": 138775
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9734098011443335e-06,
+      "loss": 1.6657,
+      "step": 138780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.972756540778455e-06,
+      "loss": 1.183,
+      "step": 138785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.972103309120838e-06,
+      "loss": 1.0604,
+      "step": 138790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.971450106175214e-06,
+      "loss": 1.6379,
+      "step": 138795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.970796931945318e-06,
+      "loss": 1.471,
+      "step": 138800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.970143786434875e-06,
+      "loss": 1.1237,
+      "step": 138805
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.969490669647617e-06,
+      "loss": 1.3118,
+      "step": 138810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.968837581587271e-06,
+      "loss": 1.3959,
+      "step": 138815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.968184522257573e-06,
+      "loss": 1.452,
+      "step": 138820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.967531491662243e-06,
+      "loss": 1.5734,
+      "step": 138825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9668784898050185e-06,
+      "loss": 1.388,
+      "step": 138830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.966225516689624e-06,
+      "loss": 1.3415,
+      "step": 138835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.965572572319787e-06,
+      "loss": 1.2475,
+      "step": 138840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.964919656699237e-06,
+      "loss": 1.0654,
+      "step": 138845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.964266769831709e-06,
+      "loss": 1.204,
+      "step": 138850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.963613911720923e-06,
+      "loss": 0.9247,
+      "step": 138855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.962961082370612e-06,
+      "loss": 1.2507,
+      "step": 138860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9623082817845005e-06,
+      "loss": 1.0651,
+      "step": 138865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.961655509966318e-06,
+      "loss": 1.2088,
+      "step": 138870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9610027669197944e-06,
+      "loss": 1.3568,
+      "step": 138875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.960350052648657e-06,
+      "loss": 1.3693,
+      "step": 138880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.959697367156627e-06,
+      "loss": 1.009,
+      "step": 138885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.959044710447436e-06,
+      "loss": 1.1899,
+      "step": 138890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.958392082524815e-06,
+      "loss": 1.1873,
+      "step": 138895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.957739483392487e-06,
+      "loss": 1.3054,
+      "step": 138900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.957086913054174e-06,
+      "loss": 1.0985,
+      "step": 138905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.956434371513608e-06,
+      "loss": 1.0547,
+      "step": 138910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.955781858774518e-06,
+      "loss": 1.1738,
+      "step": 138915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.955129374840623e-06,
+      "loss": 1.3889,
+      "step": 138920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.954476919715655e-06,
+      "loss": 1.1022,
+      "step": 138925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.95382449340334e-06,
+      "loss": 1.3183,
+      "step": 138930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.953172095907391e-06,
+      "loss": 1.1845,
+      "step": 138935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.952519727231553e-06,
+      "loss": 1.3304,
+      "step": 138940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.951867387379542e-06,
+      "loss": 1.1897,
+      "step": 138945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.951215076355079e-06,
+      "loss": 1.0927,
+      "step": 138950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9505627941618985e-06,
+      "loss": 0.9875,
+      "step": 138955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.949910540803715e-06,
+      "loss": 1.0741,
+      "step": 138960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.949258316284262e-06,
+      "loss": 1.2995,
+      "step": 138965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.948606120607259e-06,
+      "loss": 1.3395,
+      "step": 138970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9479539537764345e-06,
+      "loss": 1.0614,
+      "step": 138975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.947301815795506e-06,
+      "loss": 1.3428,
+      "step": 138980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.946649706668202e-06,
+      "loss": 1.1319,
+      "step": 138985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.94599762639825e-06,
+      "loss": 1.5155,
+      "step": 138990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9453455749893695e-06,
+      "loss": 0.9892,
+      "step": 138995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.944693552445281e-06,
+      "loss": 1.1781,
+      "step": 139000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.944041558769711e-06,
+      "loss": 1.1374,
+      "step": 139005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9433895939663865e-06,
+      "loss": 1.0608,
+      "step": 139010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9427376580390255e-06,
+      "loss": 1.3037,
+      "step": 139015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9420857509913545e-06,
+      "loss": 1.1309,
+      "step": 139020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.941433872827094e-06,
+      "loss": 1.056,
+      "step": 139025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.940782023549964e-06,
+      "loss": 1.0391,
+      "step": 139030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.940130203163689e-06,
+      "loss": 1.3104,
+      "step": 139035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.939478411671997e-06,
+      "loss": 0.893,
+      "step": 139040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.938826649078603e-06,
+      "loss": 1.2182,
+      "step": 139045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.93817491538723e-06,
+      "loss": 0.9371,
+      "step": 139050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.937523210601598e-06,
+      "loss": 1.1565,
+      "step": 139055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.936871534725436e-06,
+      "loss": 0.9451,
+      "step": 139060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.936219887762458e-06,
+      "loss": 0.9845,
+      "step": 139065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.935568269716391e-06,
+      "loss": 1.2783,
+      "step": 139070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.934916680590949e-06,
+      "loss": 1.2154,
+      "step": 139075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.934265120389857e-06,
+      "loss": 1.2038,
+      "step": 139080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.93361358911684e-06,
+      "loss": 1.1445,
+      "step": 139085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9329620867756144e-06,
+      "loss": 1.3257,
+      "step": 139090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.932310613369897e-06,
+      "loss": 0.913,
+      "step": 139095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.931659168903411e-06,
+      "loss": 1.4135,
+      "step": 139100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.931007753379882e-06,
+      "loss": 1.0622,
+      "step": 139105
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.930356366803024e-06,
+      "loss": 1.1412,
+      "step": 139110
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.929705009176555e-06,
+      "loss": 1.1154,
+      "step": 139115
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.929053680504197e-06,
+      "loss": 1.2404,
+      "step": 139120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.928402380789674e-06,
+      "loss": 1.5076,
+      "step": 139125
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.927751110036697e-06,
+      "loss": 1.3653,
+      "step": 139130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9270998682489936e-06,
+      "loss": 1.3065,
+      "step": 139135
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9264486554302774e-06,
+      "loss": 1.1197,
+      "step": 139140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.925797471584265e-06,
+      "loss": 1.6277,
+      "step": 139145
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9251463167146785e-06,
+      "loss": 1.2925,
+      "step": 139150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.924495190825239e-06,
+      "loss": 1.1914,
+      "step": 139155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.923844093919658e-06,
+      "loss": 1.3432,
+      "step": 139160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.923193026001662e-06,
+      "loss": 1.107,
+      "step": 139165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.922541987074961e-06,
+      "loss": 1.6015,
+      "step": 139170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.92189097714328e-06,
+      "loss": 1.106,
+      "step": 139175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.921239996210328e-06,
+      "loss": 1.1675,
+      "step": 139180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.920589044279831e-06,
+      "loss": 0.9635,
+      "step": 139185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9199381213554995e-06,
+      "loss": 1.3297,
+      "step": 139190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9192872274410534e-06,
+      "loss": 1.2143,
+      "step": 139195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.918636362540213e-06,
+      "loss": 1.1837,
+      "step": 139200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.917985526656693e-06,
+      "loss": 1.4127,
+      "step": 139205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.917334719794204e-06,
+      "loss": 1.1645,
+      "step": 139210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9166839419564696e-06,
+      "loss": 1.0922,
+      "step": 139215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.916033193147206e-06,
+      "loss": 1.2753,
+      "step": 139220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.915382473370124e-06,
+      "loss": 1.3968,
+      "step": 139225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9147317826289465e-06,
+      "loss": 1.508,
+      "step": 139230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.914081120927386e-06,
+      "loss": 1.5806,
+      "step": 139235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.913430488269154e-06,
+      "loss": 1.2587,
+      "step": 139240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9127798846579685e-06,
+      "loss": 1.2448,
+      "step": 139245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.912129310097552e-06,
+      "loss": 1.2561,
+      "step": 139250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.911478764591612e-06,
+      "loss": 1.1271,
+      "step": 139255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.910828248143862e-06,
+      "loss": 1.4552,
+      "step": 139260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.910177760758019e-06,
+      "loss": 1.3451,
+      "step": 139265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.909527302437803e-06,
+      "loss": 1.5892,
+      "step": 139270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.908876873186921e-06,
+      "loss": 1.2465,
+      "step": 139275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.908226473009093e-06,
+      "loss": 1.5286,
+      "step": 139280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.907576101908027e-06,
+      "loss": 1.4215,
+      "step": 139285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.906925759887441e-06,
+      "loss": 0.9462,
+      "step": 139290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.906275446951051e-06,
+      "loss": 1.3264,
+      "step": 139295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.905625163102569e-06,
+      "loss": 1.4289,
+      "step": 139300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.9049749083457035e-06,
+      "loss": 1.2144,
+      "step": 139305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.904324682684171e-06,
+      "loss": 1.2085,
+      "step": 139310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.903674486121689e-06,
+      "loss": 1.2903,
+      "step": 139315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.903024318661966e-06,
+      "loss": 1.3268,
+      "step": 139320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.902374180308714e-06,
+      "loss": 1.2243,
+      "step": 139325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.901724071065649e-06,
+      "loss": 1.4098,
+      "step": 139330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.901073990936478e-06,
+      "loss": 1.1988,
+      "step": 139335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.900423939924918e-06,
+      "loss": 1.595,
+      "step": 139340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.899773918034684e-06,
+      "loss": 1.2555,
+      "step": 139345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.899123925269483e-06,
+      "loss": 1.1736,
+      "step": 139350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.898473961633025e-06,
+      "loss": 1.3025,
+      "step": 139355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.897824027129024e-06,
+      "loss": 1.0023,
+      "step": 139360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.897174121761195e-06,
+      "loss": 1.1685,
+      "step": 139365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.896524245533245e-06,
+      "loss": 1.2942,
+      "step": 139370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.895874398448889e-06,
+      "loss": 1.284,
+      "step": 139375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8952245805118315e-06,
+      "loss": 1.4673,
+      "step": 139380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.894574791725787e-06,
+      "loss": 1.2712,
+      "step": 139385
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.89392503209447e-06,
+      "loss": 1.5933,
+      "step": 139390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.893275301621589e-06,
+      "loss": 1.2539,
+      "step": 139395
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.892625600310846e-06,
+      "loss": 1.2082,
+      "step": 139400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.891975928165961e-06,
+      "loss": 1.1846,
+      "step": 139405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.891326285190643e-06,
+      "loss": 1.2353,
+      "step": 139410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.890676671388599e-06,
+      "loss": 0.9047,
+      "step": 139415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.890027086763537e-06,
+      "loss": 1.1072,
+      "step": 139420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.889377531319171e-06,
+      "loss": 1.1876,
+      "step": 139425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.888728005059206e-06,
+      "loss": 1.3893,
+      "step": 139430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8880785079873515e-06,
+      "loss": 1.1607,
+      "step": 139435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.887429040107324e-06,
+      "loss": 1.4993,
+      "step": 139440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.886779601422824e-06,
+      "loss": 1.1846,
+      "step": 139445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.88613019193756e-06,
+      "loss": 1.1144,
+      "step": 139450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8854808116552435e-06,
+      "loss": 1.2078,
+      "step": 139455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.884831460579585e-06,
+      "loss": 1.5624,
+      "step": 139460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.884182138714292e-06,
+      "loss": 1.1876,
+      "step": 139465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.883532846063066e-06,
+      "loss": 1.1838,
+      "step": 139470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8828835826296185e-06,
+      "loss": 1.3387,
+      "step": 139475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.882234348417663e-06,
+      "loss": 1.533,
+      "step": 139480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.881585143430897e-06,
+      "loss": 1.5836,
+      "step": 139485
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.880935967673037e-06,
+      "loss": 1.2207,
+      "step": 139490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.880286821147783e-06,
+      "loss": 1.4095,
+      "step": 139495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.879637703858844e-06,
+      "loss": 1.1543,
+      "step": 139500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8789886158099305e-06,
+      "loss": 1.2573,
+      "step": 139505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.878339557004749e-06,
+      "loss": 1.452,
+      "step": 139510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.877690527446998e-06,
+      "loss": 1.1719,
+      "step": 139515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8770415271403885e-06,
+      "loss": 1.3248,
+      "step": 139520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.876392556088632e-06,
+      "loss": 1.5463,
+      "step": 139525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.875743614295425e-06,
+      "loss": 1.3602,
+      "step": 139530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.875094701764483e-06,
+      "loss": 1.3151,
+      "step": 139535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.874445818499506e-06,
+      "loss": 1.1812,
+      "step": 139540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8737969645041975e-06,
+      "loss": 1.1168,
+      "step": 139545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.873148139782263e-06,
+      "loss": 1.0357,
+      "step": 139550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.872499344337416e-06,
+      "loss": 1.4136,
+      "step": 139555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.871850578173355e-06,
+      "loss": 1.0171,
+      "step": 139560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.871201841293783e-06,
+      "loss": 1.364,
+      "step": 139565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.870553133702405e-06,
+      "loss": 1.3033,
+      "step": 139570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.869904455402931e-06,
+      "loss": 1.3984,
+      "step": 139575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.869255806399059e-06,
+      "loss": 1.0151,
+      "step": 139580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.868607186694501e-06,
+      "loss": 1.2121,
+      "step": 139585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.86795859629295e-06,
+      "loss": 0.9431,
+      "step": 139590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.867310035198116e-06,
+      "loss": 0.9974,
+      "step": 139595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.866661503413708e-06,
+      "loss": 1.1434,
+      "step": 139600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.866013000943421e-06,
+      "loss": 1.2194,
+      "step": 139605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.865364527790963e-06,
+      "loss": 1.0082,
+      "step": 139610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.864716083960027e-06,
+      "loss": 1.6599,
+      "step": 139615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8640676694543325e-06,
+      "loss": 1.4243,
+      "step": 139620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.863419284277573e-06,
+      "loss": 1.4352,
+      "step": 139625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.862770928433449e-06,
+      "loss": 1.073,
+      "step": 139630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8621226019256695e-06,
+      "loss": 1.3032,
+      "step": 139635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861474304757929e-06,
+      "loss": 1.6142,
+      "step": 139640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860826036933934e-06,
+      "loss": 1.291,
+      "step": 139645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860177798457392e-06,
+      "loss": 1.4462,
+      "step": 139650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.859529589331998e-06,
+      "loss": 1.1902,
+      "step": 139655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8588814095614515e-06,
+      "loss": 1.1517,
+      "step": 139660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.858233259149456e-06,
+      "loss": 1.4301,
+      "step": 139665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8575851380997195e-06,
+      "loss": 1.1115,
+      "step": 139670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.856937046415933e-06,
+      "loss": 1.325,
+      "step": 139675
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.856288984101807e-06,
+      "loss": 1.2286,
+      "step": 139680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8556409511610325e-06,
+      "loss": 1.2281,
+      "step": 139685
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.854992947597319e-06,
+      "loss": 1.1232,
+      "step": 139690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.85434497341436e-06,
+      "loss": 1.3117,
+      "step": 139695
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.853697028615861e-06,
+      "loss": 1.1935,
+      "step": 139700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.853049113205521e-06,
+      "loss": 1.2355,
+      "step": 139705
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.852401227187032e-06,
+      "loss": 1.2002,
+      "step": 139710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.851753370564108e-06,
+      "loss": 1.1508,
+      "step": 139715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.851105543340441e-06,
+      "loss": 1.7801,
+      "step": 139720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.850457745519727e-06,
+      "loss": 1.1013,
+      "step": 139725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.849809977105671e-06,
+      "loss": 1.6605,
+      "step": 139730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.849162238101969e-06,
+      "loss": 0.8775,
+      "step": 139735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.848514528512319e-06,
+      "loss": 1.3353,
+      "step": 139740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.847866848340427e-06,
+      "loss": 1.0188,
+      "step": 139745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8472191975899856e-06,
+      "loss": 1.2245,
+      "step": 139750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.846571576264689e-06,
+      "loss": 1.2087,
+      "step": 139755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.845923984368243e-06,
+      "loss": 1.509,
+      "step": 139760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.845276421904345e-06,
+      "loss": 1.2465,
+      "step": 139765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.844628888876693e-06,
+      "loss": 1.1621,
+      "step": 139770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.843981385288977e-06,
+      "loss": 1.3186,
+      "step": 139775
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.843333911144902e-06,
+      "loss": 1.2467,
+      "step": 139780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.842686466448167e-06,
+      "loss": 1.2459,
+      "step": 139785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.842039051202462e-06,
+      "loss": 1.1501,
+      "step": 139790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8413916654114925e-06,
+      "loss": 1.375,
+      "step": 139795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8407443090789475e-06,
+      "loss": 1.4143,
+      "step": 139800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8400969822085275e-06,
+      "loss": 1.1712,
+      "step": 139805
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.839449684803932e-06,
+      "loss": 1.267,
+      "step": 139810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8388024168688555e-06,
+      "loss": 1.2642,
+      "step": 139815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8381551784069895e-06,
+      "loss": 1.3174,
+      "step": 139820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.837507969422036e-06,
+      "loss": 1.2494,
+      "step": 139825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.836860789917686e-06,
+      "loss": 1.2899,
+      "step": 139830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.836213639897641e-06,
+      "loss": 1.4284,
+      "step": 139835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.83556651936559e-06,
+      "loss": 1.1549,
+      "step": 139840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.834919428325237e-06,
+      "loss": 1.2291,
+      "step": 139845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.834272366780266e-06,
+      "loss": 1.2418,
+      "step": 139850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.83362533473438e-06,
+      "loss": 1.2237,
+      "step": 139855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.832978332191275e-06,
+      "loss": 1.2655,
+      "step": 139860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.832331359154643e-06,
+      "loss": 1.2181,
+      "step": 139865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.831684415628174e-06,
+      "loss": 1.2277,
+      "step": 139870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.831037501615567e-06,
+      "loss": 1.6862,
+      "step": 139875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.830390617120521e-06,
+      "loss": 1.0711,
+      "step": 139880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.82974376214672e-06,
+      "loss": 1.3623,
+      "step": 139885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.829096936697867e-06,
+      "loss": 1.2495,
+      "step": 139890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.828450140777649e-06,
+      "loss": 1.0486,
+      "step": 139895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.827803374389764e-06,
+      "loss": 1.2523,
+      "step": 139900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.827156637537901e-06,
+      "loss": 1.0086,
+      "step": 139905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.82650993022576e-06,
+      "loss": 1.1506,
+      "step": 139910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.825863252457029e-06,
+      "loss": 1.2959,
+      "step": 139915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.825216604235394e-06,
+      "loss": 1.079,
+      "step": 139920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.824569985564565e-06,
+      "loss": 1.3064,
+      "step": 139925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.823923396448222e-06,
+      "loss": 1.3944,
+      "step": 139930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.823276836890058e-06,
+      "loss": 1.1193,
+      "step": 139935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.822630306893772e-06,
+      "loss": 1.0712,
+      "step": 139940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.821983806463047e-06,
+      "loss": 0.9717,
+      "step": 139945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.82133733560158e-06,
+      "loss": 1.119,
+      "step": 139950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8206908943130656e-06,
+      "loss": 0.9365,
+      "step": 139955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8200444826011924e-06,
+      "loss": 1.256,
+      "step": 139960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.819398100469648e-06,
+      "loss": 1.2811,
+      "step": 139965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.818751747922126e-06,
+      "loss": 1.5385,
+      "step": 139970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.818105424962322e-06,
+      "loss": 1.0581,
+      "step": 139975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.817459131593923e-06,
+      "loss": 1.4142,
+      "step": 139980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.816812867820616e-06,
+      "loss": 0.7313,
+      "step": 139985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.816166633646096e-06,
+      "loss": 1.2832,
+      "step": 139990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8155204290740554e-06,
+      "loss": 1.2318,
+      "step": 139995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.814874254108178e-06,
+      "loss": 1.1452,
+      "step": 140000
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 1.2217296361923218,
+      "eval_runtime": 6468.955,
+      "eval_samples_per_second": 3.572,
+      "eval_steps_per_second": 1.786,
+      "step": 140000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.814228108752161e-06,
+      "loss": 1.1431,
+      "step": 140005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.813581993009692e-06,
+      "loss": 1.0735,
+      "step": 140010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.81293590688445e-06,
+      "loss": 1.4862,
+      "step": 140015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.812289850380143e-06,
+      "loss": 1.1652,
+      "step": 140020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.811643823500452e-06,
+      "loss": 1.2635,
+      "step": 140025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.81099782624906e-06,
+      "loss": 1.2333,
+      "step": 140030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.810351858629665e-06,
+      "loss": 1.162,
+      "step": 140035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.809705920645948e-06,
+      "loss": 1.3006,
+      "step": 140040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.809060012301606e-06,
+      "loss": 1.1921,
+      "step": 140045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.80841413360032e-06,
+      "loss": 1.1844,
+      "step": 140050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.807768284545785e-06,
+      "loss": 1.7055,
+      "step": 140055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.807122465141683e-06,
+      "loss": 1.1529,
+      "step": 140060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8064766753917034e-06,
+      "loss": 0.9577,
+      "step": 140065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.80583091529954e-06,
+      "loss": 1.0362,
+      "step": 140070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.805185184868874e-06,
+      "loss": 1.2972,
+      "step": 140075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.804539484103393e-06,
+      "loss": 1.0448,
+      "step": 140080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.803893813006785e-06,
+      "loss": 1.2192,
+      "step": 140085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.803248171582742e-06,
+      "loss": 1.2869,
+      "step": 140090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.802602559834943e-06,
+      "loss": 1.3197,
+      "step": 140095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.801956977767082e-06,
+      "loss": 1.1017,
+      "step": 140100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.801311425382841e-06,
+      "loss": 1.4222,
+      "step": 140105
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.800665902685906e-06,
+      "loss": 1.2705,
+      "step": 140110
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.80002040967997e-06,
+      "loss": 1.2683,
+      "step": 140115
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.799374946368712e-06,
+      "loss": 1.1158,
+      "step": 140120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.798729512755822e-06,
+      "loss": 1.426,
+      "step": 140125
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.79808410884498e-06,
+      "loss": 1.0583,
+      "step": 140130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.797438734639877e-06,
+      "loss": 0.9852,
+      "step": 140135
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.796793390144198e-06,
+      "loss": 1.5267,
+      "step": 140140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.796148075361625e-06,
+      "loss": 1.3337,
+      "step": 140145
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.795502790295849e-06,
+      "loss": 1.3462,
+      "step": 140150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.794857534950549e-06,
+      "loss": 1.4872,
+      "step": 140155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7942123093294105e-06,
+      "loss": 1.1769,
+      "step": 140160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.793567113436123e-06,
+      "loss": 1.3325,
+      "step": 140165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.792921947274368e-06,
+      "loss": 1.1494,
+      "step": 140170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.792276810847827e-06,
+      "loss": 1.4228,
+      "step": 140175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.791631704160185e-06,
+      "loss": 1.3526,
+      "step": 140180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7909866272151305e-06,
+      "loss": 1.1529,
+      "step": 140185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.790341580016345e-06,
+      "loss": 1.1277,
+      "step": 140190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.789696562567507e-06,
+      "loss": 1.164,
+      "step": 140195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7890515748723045e-06,
+      "loss": 1.2993,
+      "step": 140200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.788406616934423e-06,
+      "loss": 1.2057,
+      "step": 140205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.78776168875754e-06,
+      "loss": 1.2526,
+      "step": 140210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.787116790345344e-06,
+      "loss": 1.4163,
+      "step": 140215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.786471921701517e-06,
+      "loss": 1.4283,
+      "step": 140220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7858270828297295e-06,
+      "loss": 1.4703,
+      "step": 140225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.785182273733682e-06,
+      "loss": 1.0883,
+      "step": 140230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.784537494417049e-06,
+      "loss": 1.3768,
+      "step": 140235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.783892744883508e-06,
+      "loss": 1.0244,
+      "step": 140240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.783248025136749e-06,
+      "loss": 1.0049,
+      "step": 140245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.782603335180446e-06,
+      "loss": 1.07,
+      "step": 140250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7819586750182835e-06,
+      "loss": 1.2093,
+      "step": 140255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7813140446539475e-06,
+      "loss": 1.2658,
+      "step": 140260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.780669444091115e-06,
+      "loss": 1.185,
+      "step": 140265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.780024873333463e-06,
+      "loss": 1.2556,
+      "step": 140270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.779380332384678e-06,
+      "loss": 0.983,
+      "step": 140275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7787358212484414e-06,
+      "loss": 1.1256,
+      "step": 140280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.778091339928432e-06,
+      "loss": 1.2121,
+      "step": 140285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.777446888428327e-06,
+      "loss": 1.0146,
+      "step": 140290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.776802466751808e-06,
+      "loss": 1.0521,
+      "step": 140295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7761580749025605e-06,
+      "loss": 1.3351,
+      "step": 140300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.775513712884257e-06,
+      "loss": 1.2677,
+      "step": 140305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774869380700582e-06,
+      "loss": 1.5566,
+      "step": 140310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774225078355215e-06,
+      "loss": 1.19,
+      "step": 140315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7735808058518255e-06,
+      "loss": 0.9421,
+      "step": 140320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772936563194107e-06,
+      "loss": 1.3202,
+      "step": 140325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772292350385733e-06,
+      "loss": 1.2618,
+      "step": 140330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.771648167430382e-06,
+      "loss": 1.3679,
+      "step": 140335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.771004014331728e-06,
+      "loss": 1.4521,
+      "step": 140340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7703598910934535e-06,
+      "loss": 1.326,
+      "step": 140345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.76971579771924e-06,
+      "loss": 1.2816,
+      "step": 140350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.769071734212759e-06,
+      "loss": 1.3361,
+      "step": 140355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.768427700577696e-06,
+      "loss": 1.1846,
+      "step": 140360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.767783696817719e-06,
+      "loss": 0.9097,
+      "step": 140365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.767139722936514e-06,
+      "loss": 1.2598,
+      "step": 140370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7664957789377575e-06,
+      "loss": 1.3684,
+      "step": 140375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.765851864825126e-06,
+      "loss": 1.4956,
+      "step": 140380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.765207980602291e-06,
+      "loss": 1.1057,
+      "step": 140385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.764564126272935e-06,
+      "loss": 1.165,
+      "step": 140390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.763920301840736e-06,
+      "loss": 1.3836,
+      "step": 140395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7632765073093654e-06,
+      "loss": 1.2304,
+      "step": 140400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.762632742682507e-06,
+      "loss": 1.7085,
+      "step": 140405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.761989007963832e-06,
+      "loss": 1.1879,
+      "step": 140410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.761345303157013e-06,
+      "loss": 1.0733,
+      "step": 140415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.760701628265731e-06,
+      "loss": 1.1733,
+      "step": 140420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.760057983293663e-06,
+      "loss": 1.095,
+      "step": 140425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.759414368244483e-06,
+      "loss": 1.0141,
+      "step": 140430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7587707831218614e-06,
+      "loss": 0.9381,
+      "step": 140435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.758127227929478e-06,
+      "loss": 1.0224,
+      "step": 140440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.757483702671011e-06,
+      "loss": 1.3497,
+      "step": 140445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7568402073501285e-06,
+      "loss": 1.1587,
+      "step": 140450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.756196741970511e-06,
+      "loss": 1.1276,
+      "step": 140455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7555533065358285e-06,
+      "loss": 1.1333,
+      "step": 140460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.754909901049756e-06,
+      "loss": 1.3538,
+      "step": 140465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7542665255159735e-06,
+      "loss": 1.8814,
+      "step": 140470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.753623179938151e-06,
+      "loss": 1.5007,
+      "step": 140475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.752979864319956e-06,
+      "loss": 1.2103,
+      "step": 140480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.75233657866507e-06,
+      "loss": 1.3154,
+      "step": 140485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.751693322977168e-06,
+      "loss": 1.1662,
+      "step": 140490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.75105009725992e-06,
+      "loss": 1.3725,
+      "step": 140495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.750406901516995e-06,
+      "loss": 1.4244,
+      "step": 140500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.749763735752071e-06,
+      "loss": 0.9314,
+      "step": 140505
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.749120599968823e-06,
+      "loss": 1.2447,
+      "step": 140510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.748477494170916e-06,
+      "loss": 1.2516,
+      "step": 140515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.747834418362032e-06,
+      "loss": 1.6904,
+      "step": 140520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.747191372545839e-06,
+      "loss": 1.4134,
+      "step": 140525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.746548356726003e-06,
+      "loss": 1.1673,
+      "step": 140530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.745905370906202e-06,
+      "loss": 1.2477,
+      "step": 140535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.745262415090111e-06,
+      "loss": 1.5363,
+      "step": 140540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.744619489281395e-06,
+      "loss": 1.0817,
+      "step": 140545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.743976593483731e-06,
+      "loss": 1.6375,
+      "step": 140550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.743333727700784e-06,
+      "loss": 1.1681,
+      "step": 140555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.742690891936233e-06,
+      "loss": 1.3783,
+      "step": 140560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.74204808619374e-06,
+      "loss": 1.1892,
+      "step": 140565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7414053104769855e-06,
+      "loss": 1.3735,
+      "step": 140570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.74076256478963e-06,
+      "loss": 1.1801,
+      "step": 140575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.740119849135351e-06,
+      "loss": 1.0194,
+      "step": 140580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.739477163517818e-06,
+      "loss": 1.1864,
+      "step": 140585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.738834507940701e-06,
+      "loss": 1.6599,
+      "step": 140590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.738191882407665e-06,
+      "loss": 1.068,
+      "step": 140595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.737549286922383e-06,
+      "loss": 1.2227,
+      "step": 140600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.736906721488529e-06,
+      "loss": 1.5704,
+      "step": 140605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.736264186109765e-06,
+      "loss": 1.0581,
+      "step": 140610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.735621680789766e-06,
+      "loss": 1.8177,
+      "step": 140615
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7349792055321995e-06,
+      "loss": 1.7058,
+      "step": 140620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.734336760340727e-06,
+      "loss": 1.1262,
+      "step": 140625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.733694345219029e-06,
+      "loss": 1.2406,
+      "step": 140630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.733051960170771e-06,
+      "loss": 1.3035,
+      "step": 140635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.732409605199618e-06,
+      "loss": 1.3006,
+      "step": 140640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7317672803092365e-06,
+      "loss": 1.2483,
+      "step": 140645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.731124985503296e-06,
+      "loss": 1.2783,
+      "step": 140650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.730482720785471e-06,
+      "loss": 1.3222,
+      "step": 140655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.729840486159418e-06,
+      "loss": 1.164,
+      "step": 140660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7291982816288165e-06,
+      "loss": 1.2675,
+      "step": 140665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.728556107197323e-06,
+      "loss": 0.9905,
+      "step": 140670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.727913962868609e-06,
+      "loss": 1.1501,
+      "step": 140675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.727271848646346e-06,
+      "loss": 0.9133,
+      "step": 140680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.726629764534196e-06,
+      "loss": 1.0052,
+      "step": 140685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.725987710535824e-06,
+      "loss": 1.3099,
+      "step": 140690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.725345686654897e-06,
+      "loss": 1.0064,
+      "step": 140695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.724703692895089e-06,
+      "loss": 1.025,
+      "step": 140700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.724061729260058e-06,
+      "loss": 1.1252,
+      "step": 140705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.723419795753471e-06,
+      "loss": 1.1383,
+      "step": 140710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.722777892378997e-06,
+      "loss": 1.0288,
+      "step": 140715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.722136019140296e-06,
+      "loss": 1.322,
+      "step": 140720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.721494176041037e-06,
+      "loss": 1.3422,
+      "step": 140725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.72085236308489e-06,
+      "loss": 1.3161,
+      "step": 140730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.720210580275515e-06,
+      "loss": 1.2757,
+      "step": 140735
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.719568827616573e-06,
+      "loss": 1.2753,
+      "step": 140740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.718927105111733e-06,
+      "loss": 1.3275,
+      "step": 140745
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7182854127646635e-06,
+      "loss": 0.9694,
+      "step": 140750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.717643750579022e-06,
+      "loss": 1.2572,
+      "step": 140755
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7170021185584795e-06,
+      "loss": 1.3134,
+      "step": 140760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.716360516706692e-06,
+      "loss": 1.1823,
+      "step": 140765
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.715718945027328e-06,
+      "loss": 1.5705,
+      "step": 140770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.715077403524055e-06,
+      "loss": 1.1707,
+      "step": 140775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.714435892200533e-06,
+      "loss": 1.8109,
+      "step": 140780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7137944110604195e-06,
+      "loss": 1.1383,
+      "step": 140785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.713152960107385e-06,
+      "loss": 1.1119,
+      "step": 140790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.712511539345093e-06,
+      "loss": 1.0921,
+      "step": 140795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.711870148777205e-06,
+      "loss": 1.1969,
+      "step": 140800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7112287884073785e-06,
+      "loss": 0.9823,
+      "step": 140805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7105874582392795e-06,
+      "loss": 1.187,
+      "step": 140810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.709946158276576e-06,
+      "loss": 1.1963,
+      "step": 140815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.709304888522921e-06,
+      "loss": 1.1409,
+      "step": 140820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.708663648981984e-06,
+      "loss": 1.1579,
+      "step": 140825
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.708022439657424e-06,
+      "loss": 1.2866,
+      "step": 140830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.707381260552898e-06,
+      "loss": 1.262,
+      "step": 140835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.706740111672071e-06,
+      "loss": 1.4809,
+      "step": 140840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.706098993018611e-06,
+      "loss": 1.5753,
+      "step": 140845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.705457904596171e-06,
+      "loss": 1.2658,
+      "step": 140850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.704816846408411e-06,
+      "loss": 1.0734,
+      "step": 140855
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.704175818458994e-06,
+      "loss": 0.947,
+      "step": 140860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.703534820751587e-06,
+      "loss": 1.6149,
+      "step": 140865
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.70289385328984e-06,
+      "loss": 1.1096,
+      "step": 140870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.702252916077422e-06,
+      "loss": 1.2173,
+      "step": 140875
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7016120091179865e-06,
+      "loss": 1.1236,
+      "step": 140880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.700971132415194e-06,
+      "loss": 1.2617,
+      "step": 140885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.700330285972712e-06,
+      "loss": 0.9278,
+      "step": 140890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.699689469794194e-06,
+      "loss": 1.2285,
+      "step": 140895
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.699048683883296e-06,
+      "loss": 0.9794,
+      "step": 140900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.698407928243681e-06,
+      "loss": 1.4581,
+      "step": 140905
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.697767202879012e-06,
+      "loss": 1.2624,
+      "step": 140910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.697126507792941e-06,
+      "loss": 1.1305,
+      "step": 140915
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.696485842989134e-06,
+      "loss": 1.3225,
+      "step": 140920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.695845208471244e-06,
+      "loss": 1.2716,
+      "step": 140925
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.695204604242927e-06,
+      "loss": 1.2068,
+      "step": 140930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.694564030307846e-06,
+      "loss": 1.0829,
+      "step": 140935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.69392348666966e-06,
+      "loss": 1.1263,
+      "step": 140940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.693282973332026e-06,
+      "loss": 1.2845,
+      "step": 140945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.692642490298596e-06,
+      "loss": 1.2073,
+      "step": 140950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.692002037573032e-06,
+      "loss": 1.5167,
+      "step": 140955
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.691361615158997e-06,
+      "loss": 1.0372,
+      "step": 140960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.690721223060136e-06,
+      "loss": 0.9794,
+      "step": 140965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.690080861280117e-06,
+      "loss": 1.4543,
+      "step": 140970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.68944052982259e-06,
+      "loss": 1.0111,
+      "step": 140975
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.688800228691213e-06,
+      "loss": 1.543,
+      "step": 140980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.688159957889646e-06,
+      "loss": 1.5049,
+      "step": 140985
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.687519717421544e-06,
+      "loss": 0.9332,
+      "step": 140990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6868795072905605e-06,
+      "loss": 1.145,
+      "step": 140995
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.686239327500346e-06,
+      "loss": 1.3978,
+      "step": 141000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.685599178054571e-06,
+      "loss": 1.4364,
+      "step": 141005
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.684959058956882e-06,
+      "loss": 1.0527,
+      "step": 141010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.684318970210932e-06,
+      "loss": 1.1276,
+      "step": 141015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.683678911820383e-06,
+      "loss": 1.3341,
+      "step": 141020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.683038883788884e-06,
+      "loss": 1.2456,
+      "step": 141025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.682398886120092e-06,
+      "loss": 1.2124,
+      "step": 141030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.681758918817667e-06,
+      "loss": 1.2203,
+      "step": 141035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.681118981885259e-06,
+      "loss": 1.1249,
+      "step": 141040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6804790753265176e-06,
+      "loss": 1.0553,
+      "step": 141045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.679839199145102e-06,
+      "loss": 1.3293,
+      "step": 141050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.67919935334467e-06,
+      "loss": 1.4364,
+      "step": 141055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.678559537928869e-06,
+      "loss": 0.9623,
+      "step": 141060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.677919752901357e-06,
+      "loss": 1.1395,
+      "step": 141065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.677279998265783e-06,
+      "loss": 1.6289,
+      "step": 141070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.676640274025806e-06,
+      "loss": 1.3108,
+      "step": 141075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.676000580185073e-06,
+      "loss": 1.4877,
+      "step": 141080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.675360916747244e-06,
+      "loss": 1.3643,
+      "step": 141085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.674721283715965e-06,
+      "loss": 1.1138,
+      "step": 141090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.674081681094891e-06,
+      "loss": 1.1793,
+      "step": 141095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.673442108887678e-06,
+      "loss": 1.1129,
+      "step": 141100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.672802567097977e-06,
+      "loss": 1.2961,
+      "step": 141105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.672163055729434e-06,
+      "loss": 1.0687,
+      "step": 141110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.671523574785709e-06,
+      "loss": 1.2336,
+      "step": 141115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6708841242704475e-06,
+      "loss": 1.3077,
+      "step": 141120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.670244704187304e-06,
+      "loss": 1.2499,
+      "step": 141125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.669605314539932e-06,
+      "loss": 1.2015,
+      "step": 141130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.668965955331982e-06,
+      "loss": 1.4723,
+      "step": 141135
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6683266265671e-06,
+      "loss": 1.342,
+      "step": 141140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.66768732824894e-06,
+      "loss": 1.2579,
+      "step": 141145
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.667048060381158e-06,
+      "loss": 1.4155,
+      "step": 141150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.666408822967399e-06,
+      "loss": 1.1582,
+      "step": 141155
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.665769616011311e-06,
+      "loss": 1.3842,
+      "step": 141160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6651304395165465e-06,
+      "loss": 1.4774,
+      "step": 141165
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6644912934867615e-06,
+      "loss": 1.0121,
+      "step": 141170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.663852177925596e-06,
+      "loss": 1.1385,
+      "step": 141175
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.66321309283671e-06,
+      "loss": 1.1933,
+      "step": 141180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.662574038223743e-06,
+      "loss": 1.0424,
+      "step": 141185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.66193501409035e-06,
+      "loss": 1.0541,
+      "step": 141190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.661296020440181e-06,
+      "loss": 1.3775,
+      "step": 141195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.660657057276884e-06,
+      "loss": 1.8311,
+      "step": 141200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6600181246041034e-06,
+      "loss": 1.3767,
+      "step": 141205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.659379222425492e-06,
+      "loss": 1.0678,
+      "step": 141210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6587403507447e-06,
+      "loss": 1.5247,
+      "step": 141215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.658101509565375e-06,
+      "loss": 1.2944,
+      "step": 141220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.657462698891159e-06,
+      "loss": 1.1184,
+      "step": 141225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.656823918725708e-06,
+      "loss": 1.2423,
+      "step": 141230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.656185169072663e-06,
+      "loss": 1.3392,
+      "step": 141235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.655546449935675e-06,
+      "loss": 1.2253,
+      "step": 141240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.654907761318394e-06,
+      "loss": 1.0783,
+      "step": 141245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.654269103224466e-06,
+      "loss": 1.3806,
+      "step": 141250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6536304756575325e-06,
+      "loss": 0.9917,
+      "step": 141255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.652991878621243e-06,
+      "loss": 1.174,
+      "step": 141260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.652353312119251e-06,
+      "loss": 1.601,
+      "step": 141265
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.651714776155194e-06,
+      "loss": 1.0667,
+      "step": 141270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.651076270732726e-06,
+      "loss": 1.4942,
+      "step": 141275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.650437795855486e-06,
+      "loss": 1.0486,
+      "step": 141280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.649799351527125e-06,
+      "loss": 1.1063,
+      "step": 141285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6491609377512856e-06,
+      "loss": 1.6749,
+      "step": 141290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.648522554531619e-06,
+      "loss": 1.1408,
+      "step": 141295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.647884201871766e-06,
+      "loss": 1.2257,
+      "step": 141300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.647245879775366e-06,
+      "loss": 1.4724,
+      "step": 141305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.646607588246079e-06,
+      "loss": 1.3908,
+      "step": 141310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.645969327287542e-06,
+      "loss": 1.1489,
+      "step": 141315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.645331096903395e-06,
+      "loss": 1.3718,
+      "step": 141320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.644692897097292e-06,
+      "loss": 1.3288,
+      "step": 141325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.644054727872869e-06,
+      "loss": 1.5358,
+      "step": 141330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.643416589233774e-06,
+      "loss": 1.2534,
+      "step": 141335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.642778481183654e-06,
+      "loss": 1.3654,
+      "step": 141340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.642140403726152e-06,
+      "loss": 1.0509,
+      "step": 141345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.641502356864905e-06,
+      "loss": 1.2136,
+      "step": 141350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.640864340603562e-06,
+      "loss": 1.7586,
+      "step": 141355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6402263549457705e-06,
+      "loss": 1.3212,
+      "step": 141360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.639588399895169e-06,
+      "loss": 1.1801,
+      "step": 141365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.638950475455397e-06,
+      "loss": 1.3887,
+      "step": 141370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.638312581630101e-06,
+      "loss": 1.1044,
+      "step": 141375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.637674718422929e-06,
+      "loss": 1.1757,
+      "step": 141380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.637036885837513e-06,
+      "loss": 1.2882,
+      "step": 141385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.636399083877505e-06,
+      "loss": 1.1387,
+      "step": 141390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.635761312546544e-06,
+      "loss": 1.2724,
+      "step": 141395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.635123571848262e-06,
+      "loss": 1.0407,
+      "step": 141400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.634485861786319e-06,
+      "loss": 1.0708,
+      "step": 141405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.633848182364346e-06,
+      "loss": 1.1083,
+      "step": 141410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.633210533585983e-06,
+      "loss": 1.3493,
+      "step": 141415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.632572915454879e-06,
+      "loss": 1.5146,
+      "step": 141420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.631935327974666e-06,
+      "loss": 1.1773,
+      "step": 141425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.631297771148994e-06,
+      "loss": 0.9983,
+      "step": 141430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.630660244981495e-06,
+      "loss": 1.5703,
+      "step": 141435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.630022749475817e-06,
+      "loss": 1.1269,
+      "step": 141440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6293852846355945e-06,
+      "loss": 1.3564,
+      "step": 141445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.62874785046447e-06,
+      "loss": 0.9591,
+      "step": 141450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.628110446966088e-06,
+      "loss": 1.2237,
+      "step": 141455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.627473074144085e-06,
+      "loss": 1.0944,
+      "step": 141460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.626835732002097e-06,
+      "loss": 1.2743,
+      "step": 141465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.626198420543767e-06,
+      "loss": 1.1422,
+      "step": 141470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.625561139772737e-06,
+      "loss": 1.4192,
+      "step": 141475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.624923889692641e-06,
+      "loss": 1.2369,
+      "step": 141480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.624286670307123e-06,
+      "loss": 1.2303,
+      "step": 141485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.623649481619817e-06,
+      "loss": 1.592,
+      "step": 141490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.623012323634363e-06,
+      "loss": 1.4445,
+      "step": 141495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.622375196354405e-06,
+      "loss": 1.1238,
+      "step": 141500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.621738099783577e-06,
+      "loss": 1.4731,
+      "step": 141505
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.621101033925517e-06,
+      "loss": 1.1058,
+      "step": 141510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.62046399878386e-06,
+      "loss": 1.3903,
+      "step": 141515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.619826994362247e-06,
+      "loss": 1.3789,
+      "step": 141520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.619190020664318e-06,
+      "loss": 1.0585,
+      "step": 141525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.618553077693705e-06,
+      "loss": 1.6796,
+      "step": 141530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.617916165454052e-06,
+      "loss": 1.3778,
+      "step": 141535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.617279283948989e-06,
+      "loss": 1.0415,
+      "step": 141540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.616642433182157e-06,
+      "loss": 1.4089,
+      "step": 141545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.616005613157195e-06,
+      "loss": 0.9066,
+      "step": 141550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.615368823877737e-06,
+      "loss": 1.0124,
+      "step": 141555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.614732065347415e-06,
+      "loss": 1.5761,
+      "step": 141560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.61409533756987e-06,
+      "loss": 1.3186,
+      "step": 141565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.613458640548742e-06,
+      "loss": 1.2321,
+      "step": 141570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6128219742876604e-06,
+      "loss": 1.0579,
+      "step": 141575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.61218533879026e-06,
+      "loss": 1.399,
+      "step": 141580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.61154873406018e-06,
+      "loss": 1.4602,
+      "step": 141585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.610912160101056e-06,
+      "loss": 1.2441,
+      "step": 141590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.610275616916521e-06,
+      "loss": 1.1279,
+      "step": 141595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.609639104510214e-06,
+      "loss": 1.5324,
+      "step": 141600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.609002622885767e-06,
+      "loss": 1.0419,
+      "step": 141605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.608366172046807e-06,
+      "loss": 1.4342,
+      "step": 141610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.607729751996983e-06,
+      "loss": 1.3508,
+      "step": 141615
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.607093362739924e-06,
+      "loss": 1.0321,
+      "step": 141620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.606457004279258e-06,
+      "loss": 1.1307,
+      "step": 141625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.605820676618626e-06,
+      "loss": 1.7564,
+      "step": 141630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.605184379761657e-06,
+      "loss": 1.2464,
+      "step": 141635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6045481137119855e-06,
+      "loss": 1.3064,
+      "step": 141640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.60391187847325e-06,
+      "loss": 1.3728,
+      "step": 141645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6032756740490814e-06,
+      "loss": 1.2782,
+      "step": 141650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.602639500443107e-06,
+      "loss": 1.0693,
+      "step": 141655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.602003357658964e-06,
+      "loss": 1.0055,
+      "step": 141660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.601367245700288e-06,
+      "loss": 1.2701,
+      "step": 141665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6007311645707094e-06,
+      "loss": 1.0393,
+      "step": 141670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.600095114273856e-06,
+      "loss": 1.0254,
+      "step": 141675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.599459094813364e-06,
+      "loss": 1.3533,
+      "step": 141680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.598823106192869e-06,
+      "loss": 1.4132,
+      "step": 141685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5981871484159955e-06,
+      "loss": 1.2609,
+      "step": 141690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.597551221486383e-06,
+      "loss": 1.1843,
+      "step": 141695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.596915325407657e-06,
+      "loss": 1.1231,
+      "step": 141700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5962794601834444e-06,
+      "loss": 1.0603,
+      "step": 141705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.59564362581739e-06,
+      "loss": 1.1085,
+      "step": 141710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.595007822313117e-06,
+      "loss": 0.9375,
+      "step": 141715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.594372049674256e-06,
+      "loss": 1.1115,
+      "step": 141720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.593736307904435e-06,
+      "loss": 1.271,
+      "step": 141725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5931005970072875e-06,
+      "loss": 1.0154,
+      "step": 141730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.592464916986447e-06,
+      "loss": 1.2147,
+      "step": 141735
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5918292678455364e-06,
+      "loss": 1.1999,
+      "step": 141740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5911936495881936e-06,
+      "loss": 1.2073,
+      "step": 141745
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.59055806221804e-06,
+      "loss": 1.0672,
+      "step": 141750
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.58992250573871e-06,
+      "loss": 1.0239,
+      "step": 141755
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.589286980153836e-06,
+      "loss": 1.4925,
+      "step": 141760
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.588651485467043e-06,
+      "loss": 1.1102,
+      "step": 141765
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.588016021681957e-06,
+      "loss": 1.3677,
+      "step": 141770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5873805888022094e-06,
+      "loss": 1.3582,
+      "step": 141775
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.586745186831435e-06,
+      "loss": 1.1788,
+      "step": 141780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.586109815773253e-06,
+      "loss": 1.2631,
+      "step": 141785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5854744756313e-06,
+      "loss": 1.1471,
+      "step": 141790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.584839166409195e-06,
+      "loss": 0.9996,
+      "step": 141795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.584203888110576e-06,
+      "loss": 1.3101,
+      "step": 141800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.58356864073906e-06,
+      "loss": 1.0096,
+      "step": 141805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.582933424298286e-06,
+      "loss": 1.3201,
+      "step": 141810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.582298238791875e-06,
+      "loss": 1.2742,
+      "step": 141815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.581663084223451e-06,
+      "loss": 1.1419,
+      "step": 141820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.581027960596645e-06,
+      "loss": 1.3886,
+      "step": 141825
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5803928679150885e-06,
+      "loss": 1.7667,
+      "step": 141830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5797578061824e-06,
+      "loss": 1.5515,
+      "step": 141835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.579122775402213e-06,
+      "loss": 1.0519,
+      "step": 141840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.578487775578146e-06,
+      "loss": 1.1424,
+      "step": 141845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.57785280671383e-06,
+      "loss": 1.381,
+      "step": 141850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.577217868812896e-06,
+      "loss": 1.1584,
+      "step": 141855
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.576582961878963e-06,
+      "loss": 1.4036,
+      "step": 141860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.575948085915656e-06,
+      "loss": 1.2244,
+      "step": 141865
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.575313240926602e-06,
+      "loss": 1.4459,
+      "step": 141870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.574678426915431e-06,
+      "loss": 0.9313,
+      "step": 141875
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.574043643885764e-06,
+      "loss": 0.8389,
+      "step": 141880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.573408891841223e-06,
+      "loss": 1.0255,
+      "step": 141885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.572774170785435e-06,
+      "loss": 1.4463,
+      "step": 141890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5721394807220286e-06,
+      "loss": 1.6209,
+      "step": 141895
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.571504821654623e-06,
+      "loss": 1.1298,
+      "step": 141900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.570870193586847e-06,
+      "loss": 1.314,
+      "step": 141905
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5702355965223235e-06,
+      "loss": 1.1819,
+      "step": 141910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.569601030464668e-06,
+      "loss": 1.0605,
+      "step": 141915
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.568966495417517e-06,
+      "loss": 1.1864,
+      "step": 141920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.568331991384491e-06,
+      "loss": 1.2385,
+      "step": 141925
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.567697518369204e-06,
+      "loss": 1.214,
+      "step": 141930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5670630763752925e-06,
+      "loss": 1.1905,
+      "step": 141935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.566428665406368e-06,
+      "loss": 1.2512,
+      "step": 141940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.565794285466061e-06,
+      "loss": 1.2663,
+      "step": 141945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.565159936557989e-06,
+      "loss": 1.0608,
+      "step": 141950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.56452561868578e-06,
+      "loss": 1.1186,
+      "step": 141955
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.56389133185305e-06,
+      "loss": 1.0759,
+      "step": 141960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.563257076063423e-06,
+      "loss": 1.3213,
+      "step": 141965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.562622851320528e-06,
+      "loss": 0.9413,
+      "step": 141970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.56198865762798e-06,
+      "loss": 1.2225,
+      "step": 141975
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5613544949893975e-06,
+      "loss": 1.0379,
+      "step": 141980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.560720363408406e-06,
+      "loss": 1.0362,
+      "step": 141985
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.560086262888631e-06,
+      "loss": 1.6481,
+      "step": 141990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.559452193433687e-06,
+      "loss": 1.1562,
+      "step": 141995
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5588181550471986e-06,
+      "loss": 1.3141,
+      "step": 142000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.558184147732786e-06,
+      "loss": 1.0081,
+      "step": 142005
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.557550171494065e-06,
+      "loss": 1.1973,
+      "step": 142010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.55691622633466e-06,
+      "loss": 1.125,
+      "step": 142015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.556282312258194e-06,
+      "loss": 1.2108,
+      "step": 142020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5556484292682844e-06,
+      "loss": 1.1573,
+      "step": 142025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.555014577368547e-06,
+      "loss": 1.394,
+      "step": 142030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.554380756562604e-06,
+      "loss": 1.0698,
+      "step": 142035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.55374696685408e-06,
+      "loss": 1.1259,
+      "step": 142040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.553113208246587e-06,
+      "loss": 1.1766,
+      "step": 142045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.552479480743751e-06,
+      "loss": 1.1332,
+      "step": 142050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.551845784349183e-06,
+      "loss": 1.2141,
+      "step": 142055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.551212119066506e-06,
+      "loss": 1.1207,
+      "step": 142060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.550578484899342e-06,
+      "loss": 1.0601,
+      "step": 142065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5499448818513044e-06,
+      "loss": 1.1759,
+      "step": 142070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.549311309926011e-06,
+      "loss": 1.1652,
+      "step": 142075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5486777691270815e-06,
+      "loss": 1.0974,
+      "step": 142080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.548044259458137e-06,
+      "loss": 1.4907,
+      "step": 142085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.547410780922793e-06,
+      "loss": 1.0414,
+      "step": 142090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5467773335246624e-06,
+      "loss": 0.9036,
+      "step": 142095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.54614391726737e-06,
+      "loss": 1.1888,
+      "step": 142100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.545510532154525e-06,
+      "loss": 1.1397,
+      "step": 142105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5448771781897476e-06,
+      "loss": 1.0096,
+      "step": 142110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.54424385537666e-06,
+      "loss": 1.3924,
+      "step": 142115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.543610563718875e-06,
+      "loss": 1.1873,
+      "step": 142120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.542977303220004e-06,
+      "loss": 0.9995,
+      "step": 142125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.542344073883668e-06,
+      "loss": 1.3099,
+      "step": 142130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.541710875713486e-06,
+      "loss": 1.6906,
+      "step": 142135
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5410777087130675e-06,
+      "loss": 1.2126,
+      "step": 142140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5404445728860355e-06,
+      "loss": 1.0166,
+      "step": 142145
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.539811468235998e-06,
+      "loss": 1.1014,
+      "step": 142150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.539178394766576e-06,
+      "loss": 1.1948,
+      "step": 142155
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.538545352481379e-06,
+      "loss": 1.1349,
+      "step": 142160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5379123413840295e-06,
+      "loss": 0.9357,
+      "step": 142165
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5372793614781354e-06,
+      "loss": 1.1394,
+      "step": 142170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.536646412767314e-06,
+      "loss": 1.335,
+      "step": 142175
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.536013495255183e-06,
+      "loss": 1.0527,
+      "step": 142180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.535380608945354e-06,
+      "loss": 1.5499,
+      "step": 142185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.534747753841437e-06,
+      "loss": 1.1716,
+      "step": 142190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.53411492994705e-06,
+      "loss": 1.2387,
+      "step": 142195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.533482137265811e-06,
+      "loss": 1.2289,
+      "step": 142200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.532849375801325e-06,
+      "loss": 1.1773,
+      "step": 142205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.532216645557213e-06,
+      "loss": 1.1246,
+      "step": 142210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.531583946537085e-06,
+      "loss": 1.2125,
+      "step": 142215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5309512787445496e-06,
+      "loss": 1.0788,
+      "step": 142220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.530318642183226e-06,
+      "loss": 1.2128,
+      "step": 142225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.529686036856726e-06,
+      "loss": 1.039,
+      "step": 142230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.529053462768662e-06,
+      "loss": 1.4129,
+      "step": 142235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5284209199226415e-06,
+      "loss": 1.1379,
+      "step": 142240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.527788408322281e-06,
+      "loss": 1.2049,
+      "step": 142245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.527155927971195e-06,
+      "loss": 1.2652,
+      "step": 142250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.52652347887299e-06,
+      "loss": 1.2492,
+      "step": 142255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.525891061031282e-06,
+      "loss": 1.228,
+      "step": 142260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5252586744496766e-06,
+      "loss": 1.2224,
+      "step": 142265
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.524626319131788e-06,
+      "loss": 1.3616,
+      "step": 142270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.5239939950812336e-06,
+      "loss": 1.2904,
+      "step": 142275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.523361702301617e-06,
+      "loss": 1.2077,
+      "step": 142280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.522729440796548e-06,
+      "loss": 1.446,
+      "step": 142285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.52209721056964e-06,
+      "loss": 1.2638,
+      "step": 142290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.521465011624507e-06,
+      "loss": 1.1283,
+      "step": 142295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.520832843964755e-06,
+      "loss": 0.9901,
+      "step": 142300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.520200707593992e-06,
+      "loss": 1.2199,
+      "step": 142305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.519568602515832e-06,
+      "loss": 1.1537,
+      "step": 142310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.518936528733881e-06,
+      "loss": 1.2944,
+      "step": 142315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.518304486251749e-06,
+      "loss": 1.1796,
+      "step": 142320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.517672475073052e-06,
+      "loss": 1.1489,
+      "step": 142325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.517040495201392e-06,
+      "loss": 0.9518,
+      "step": 142330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.516408546640377e-06,
+      "loss": 1.5405,
+      "step": 142335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.515776629393619e-06,
+      "loss": 1.1292,
+      "step": 142340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.515144743464729e-06,
+      "loss": 1.3,
+      "step": 142345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.51451288885731e-06,
+      "loss": 1.4472,
+      "step": 142350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.513881065574976e-06,
+      "loss": 1.1571,
+      "step": 142355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.513249273621327e-06,
+      "loss": 1.2885,
+      "step": 142360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.512617512999976e-06,
+      "loss": 0.994,
+      "step": 142365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.511985783714535e-06,
+      "loss": 1.6971,
+      "step": 142370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.511354085768607e-06,
+      "loss": 1.215,
+      "step": 142375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.510722419165795e-06,
+      "loss": 1.1329,
+      "step": 142380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.51009078390971e-06,
+      "loss": 1.1246,
+      "step": 142385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5094591800039645e-06,
+      "loss": 1.2097,
+      "step": 142390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.508827607452159e-06,
+      "loss": 1.1536,
+      "step": 142395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.508196066257897e-06,
+      "loss": 1.0129,
+      "step": 142400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507564556424795e-06,
+      "loss": 1.1779,
+      "step": 142405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.506933077956448e-06,
+      "loss": 1.2463,
+      "step": 142410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.506301630856469e-06,
+      "loss": 1.4875,
+      "step": 142415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.505670215128465e-06,
+      "loss": 1.1197,
+      "step": 142420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.505038830776039e-06,
+      "loss": 1.2995,
+      "step": 142425
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5044074778027945e-06,
+      "loss": 1.1619,
+      "step": 142430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.503776156212338e-06,
+      "loss": 1.0973,
+      "step": 142435
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.50314486600828e-06,
+      "loss": 1.1247,
+      "step": 142440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.502513607194221e-06,
+      "loss": 1.2982,
+      "step": 142445
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.501882379773762e-06,
+      "loss": 1.0726,
+      "step": 142450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.501251183750513e-06,
+      "loss": 1.4209,
+      "step": 142455
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.50062001912808e-06,
+      "loss": 1.6672,
+      "step": 142460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.499988885910062e-06,
+      "loss": 1.1153,
+      "step": 142465
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.499357784100068e-06,
+      "loss": 0.9861,
+      "step": 142470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.498726713701697e-06,
+      "loss": 1.1282,
+      "step": 142475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.498095674718554e-06,
+      "loss": 1.27,
+      "step": 142480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497464667154248e-06,
+      "loss": 1.3065,
+      "step": 142485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.496833691012379e-06,
+      "loss": 1.3045,
+      "step": 142490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.496202746296545e-06,
+      "loss": 1.0727,
+      "step": 142495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.495571833010354e-06,
+      "loss": 1.4187,
+      "step": 142500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4949409511574125e-06,
+      "loss": 1.2117,
+      "step": 142505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.494310100741315e-06,
+      "loss": 1.0548,
+      "step": 142510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.493679281765672e-06,
+      "loss": 1.6178,
+      "step": 142515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.493048494234081e-06,
+      "loss": 1.2428,
+      "step": 142520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.492417738150142e-06,
+      "loss": 1.1401,
+      "step": 142525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.49178701351746e-06,
+      "loss": 1.5805,
+      "step": 142530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.49115632033964e-06,
+      "loss": 1.1759,
+      "step": 142535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4905256586202815e-06,
+      "loss": 1.117,
+      "step": 142540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.489895028362981e-06,
+      "loss": 1.1335,
+      "step": 142545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4892644295713415e-06,
+      "loss": 1.2406,
+      "step": 142550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.488633862248971e-06,
+      "loss": 0.9944,
+      "step": 142555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.488003326399463e-06,
+      "loss": 1.3056,
+      "step": 142560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.487372822026423e-06,
+      "loss": 1.2459,
+      "step": 142565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.486742349133445e-06,
+      "loss": 1.3303,
+      "step": 142570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.486111907724135e-06,
+      "loss": 1.3024,
+      "step": 142575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.485481497802094e-06,
+      "loss": 1.1484,
+      "step": 142580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.48485111937092e-06,
+      "loss": 1.2324,
+      "step": 142585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4842207724342125e-06,
+      "loss": 1.4476,
+      "step": 142590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.483590456995563e-06,
+      "loss": 1.2806,
+      "step": 142595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.482960173058588e-06,
+      "loss": 0.9677,
+      "step": 142600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.482329920626877e-06,
+      "loss": 1.4442,
+      "step": 142605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.481699699704026e-06,
+      "loss": 1.1125,
+      "step": 142610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.481069510293642e-06,
+      "loss": 1.5523,
+      "step": 142615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.480439352399315e-06,
+      "loss": 1.3921,
+      "step": 142620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.479809226024649e-06,
+      "loss": 1.2404,
+      "step": 142625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.479179131173244e-06,
+      "loss": 1.2643,
+      "step": 142630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4785490678486954e-06,
+      "loss": 1.3331,
+      "step": 142635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.477919036054599e-06,
+      "loss": 0.9769,
+      "step": 142640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.477289035794555e-06,
+      "loss": 1.3189,
+      "step": 142645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.476659067072162e-06,
+      "loss": 1.0275,
+      "step": 142650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.476029129891015e-06,
+      "loss": 1.2609,
+      "step": 142655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4753992242547165e-06,
+      "loss": 1.3376,
+      "step": 142660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.474769350166855e-06,
+      "loss": 0.9125,
+      "step": 142665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4741395076310365e-06,
+      "loss": 1.8303,
+      "step": 142670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.473509696650851e-06,
+      "loss": 1.026,
+      "step": 142675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4728799172299e-06,
+      "loss": 1.4614,
+      "step": 142680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4722501693717775e-06,
+      "loss": 1.2885,
+      "step": 142685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.471620453080072e-06,
+      "loss": 1.2913,
+      "step": 142690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.470990768358395e-06,
+      "loss": 1.1952,
+      "step": 142695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.470361115210335e-06,
+      "loss": 1.2188,
+      "step": 142700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.469731493639483e-06,
+      "loss": 1.1835,
+      "step": 142705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.469101903649442e-06,
+      "loss": 1.3188,
+      "step": 142710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4684723452438e-06,
+      "loss": 1.1478,
+      "step": 142715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.467842818426158e-06,
+      "loss": 1.0734,
+      "step": 142720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.467213323200111e-06,
+      "loss": 1.4554,
+      "step": 142725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4665838595692524e-06,
+      "loss": 1.2385,
+      "step": 142730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.465954427537171e-06,
+      "loss": 1.1289,
+      "step": 142735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.465325027107468e-06,
+      "loss": 1.2288,
+      "step": 142740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.464695658283739e-06,
+      "loss": 0.9837,
+      "step": 142745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.464066321069575e-06,
+      "loss": 1.0766,
+      "step": 142750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4634370154685655e-06,
+      "loss": 1.1618,
+      "step": 142755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.46280774148431e-06,
+      "loss": 1.3566,
+      "step": 142760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4621784991204035e-06,
+      "loss": 0.9853,
+      "step": 142765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.461549288380433e-06,
+      "loss": 1.3377,
+      "step": 142770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.460920109267997e-06,
+      "loss": 1.1006,
+      "step": 142775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.460290961786684e-06,
+      "loss": 2.0329,
+      "step": 142780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.459661845940089e-06,
+      "loss": 1.2395,
+      "step": 142785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.459032761731807e-06,
+      "loss": 1.1495,
+      "step": 142790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.45840370916543e-06,
+      "loss": 1.2229,
+      "step": 142795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.457774688244545e-06,
+      "loss": 1.1921,
+      "step": 142800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.45714569897275e-06,
+      "loss": 0.795,
+      "step": 142805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.45651674135363e-06,
+      "loss": 1.4822,
+      "step": 142810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.455887815390786e-06,
+      "loss": 0.9494,
+      "step": 142815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.455258921087801e-06,
+      "loss": 1.6481,
+      "step": 142820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.454630058448273e-06,
+      "loss": 1.1441,
+      "step": 142825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.454001227475785e-06,
+      "loss": 1.1163,
+      "step": 142830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.453372428173934e-06,
+      "loss": 1.3288,
+      "step": 142835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4527436605463135e-06,
+      "loss": 1.1949,
+      "step": 142840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.45211492459651e-06,
+      "loss": 1.0776,
+      "step": 142845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.451486220328112e-06,
+      "loss": 1.1231,
+      "step": 142850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.45085754774471e-06,
+      "loss": 1.298,
+      "step": 142855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.450228906849901e-06,
+      "loss": 0.9917,
+      "step": 142860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4496002976472656e-06,
+      "loss": 1.252,
+      "step": 142865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.448971720140402e-06,
+      "loss": 1.1952,
+      "step": 142870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.448343174332889e-06,
+      "loss": 1.3939,
+      "step": 142875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.44771466022833e-06,
+      "loss": 1.1141,
+      "step": 142880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.447086177830301e-06,
+      "loss": 1.3981,
+      "step": 142885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4464577271424e-06,
+      "loss": 1.6764,
+      "step": 142890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.445829308168214e-06,
+      "loss": 1.0799,
+      "step": 142895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.445200920911321e-06,
+      "loss": 1.0609,
+      "step": 142900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.444572565375327e-06,
+      "loss": 1.2146,
+      "step": 142905
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.443944241563811e-06,
+      "loss": 1.1929,
+      "step": 142910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.443315949480358e-06,
+      "loss": 1.195,
+      "step": 142915
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.442687689128563e-06,
+      "loss": 1.4379,
+      "step": 142920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.442059460512007e-06,
+      "loss": 1.1998,
+      "step": 142925
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.441431263634282e-06,
+      "loss": 1.1649,
+      "step": 142930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.440803098498976e-06,
+      "loss": 1.2691,
+      "step": 142935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.440174965109675e-06,
+      "loss": 1.2979,
+      "step": 142940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.439546863469961e-06,
+      "loss": 1.4103,
+      "step": 142945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.438918793583427e-06,
+      "loss": 0.7793,
+      "step": 142950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.438290755453659e-06,
+      "loss": 1.3329,
+      "step": 142955
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4376627490842435e-06,
+      "loss": 1.1615,
+      "step": 142960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.437034774478761e-06,
+      "loss": 1.2106,
+      "step": 142965
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.436406831640803e-06,
+      "loss": 1.1394,
+      "step": 142970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.435778920573958e-06,
+      "loss": 1.332,
+      "step": 142975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.435151041281804e-06,
+      "loss": 1.153,
+      "step": 142980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.434523193767935e-06,
+      "loss": 1.5059,
+      "step": 142985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.43389537803593e-06,
+      "loss": 1.4478,
+      "step": 142990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.433267594089371e-06,
+      "loss": 0.9223,
+      "step": 142995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.432639841931854e-06,
+      "loss": 1.1069,
+      "step": 143000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4320121215669584e-06,
+      "loss": 1.0302,
+      "step": 143005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.431384432998265e-06,
+      "loss": 1.1645,
+      "step": 143010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4307567762293646e-06,
+      "loss": 1.1864,
+      "step": 143015
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.430129151263835e-06,
+      "loss": 1.214,
+      "step": 143020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.429501558105265e-06,
+      "loss": 1.133,
+      "step": 143025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.428873996757239e-06,
+      "loss": 1.1034,
+      "step": 143030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.428246467223339e-06,
+      "loss": 1.1973,
+      "step": 143035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.427618969507146e-06,
+      "loss": 1.2728,
+      "step": 143040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.426991503612246e-06,
+      "loss": 1.098,
+      "step": 143045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4263640695422236e-06,
+      "loss": 1.1804,
+      "step": 143050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.425736667300662e-06,
+      "loss": 1.4115,
+      "step": 143055
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.425109296891138e-06,
+      "loss": 1.1773,
+      "step": 143060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.424481958317238e-06,
+      "loss": 1.3323,
+      "step": 143065
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.423854651582549e-06,
+      "loss": 0.9858,
+      "step": 143070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.423227376690645e-06,
+      "loss": 1.153,
+      "step": 143075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4226001336451166e-06,
+      "loss": 1.2898,
+      "step": 143080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4219729224495365e-06,
+      "loss": 1.232,
+      "step": 143085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.421345743107493e-06,
+      "loss": 1.0043,
+      "step": 143090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.420718595622567e-06,
+      "loss": 1.2905,
+      "step": 143095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.42009147999834e-06,
+      "loss": 1.1065,
+      "step": 143100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.419464396238392e-06,
+      "loss": 1.6129,
+      "step": 143105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.418837344346299e-06,
+      "loss": 1.2875,
+      "step": 143110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.418210324325648e-06,
+      "loss": 1.3615,
+      "step": 143115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.417583336180021e-06,
+      "loss": 1.3399,
+      "step": 143120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.416956379912992e-06,
+      "loss": 1.1556,
+      "step": 143125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.416329455528149e-06,
+      "loss": 1.1543,
+      "step": 143130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.415702563029064e-06,
+      "loss": 1.225,
+      "step": 143135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.415075702419322e-06,
+      "loss": 0.9846,
+      "step": 143140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.414448873702504e-06,
+      "loss": 1.3499,
+      "step": 143145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.413822076882187e-06,
+      "loss": 1.1677,
+      "step": 143150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.413195311961947e-06,
+      "loss": 1.1232,
+      "step": 143155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.412568578945368e-06,
+      "loss": 1.3573,
+      "step": 143160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.41194187783603e-06,
+      "loss": 1.1291,
+      "step": 143165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4113152086375055e-06,
+      "loss": 1.2011,
+      "step": 143170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4106885713533806e-06,
+      "loss": 1.0487,
+      "step": 143175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.410061965987228e-06,
+      "loss": 1.2373,
+      "step": 143180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4094353925426315e-06,
+      "loss": 1.3713,
+      "step": 143185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4088088510231605e-06,
+      "loss": 1.4507,
+      "step": 143190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.408182341432403e-06,
+      "loss": 1.5001,
+      "step": 143195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.407555863773933e-06,
+      "loss": 1.4273,
+      "step": 143200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.406929418051321e-06,
+      "loss": 1.2989,
+      "step": 143205
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.406303004268152e-06,
+      "loss": 1.3103,
+      "step": 143210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.405676622428004e-06,
+      "loss": 1.4668,
+      "step": 143215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.405050272534448e-06,
+      "loss": 1.135,
+      "step": 143220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.404423954591068e-06,
+      "loss": 1.1267,
+      "step": 143225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.403797668601434e-06,
+      "loss": 1.5021,
+      "step": 143230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.403171414569124e-06,
+      "loss": 1.316,
+      "step": 143235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.402545192497719e-06,
+      "loss": 0.8693,
+      "step": 143240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.401919002390793e-06,
+      "loss": 1.035,
+      "step": 143245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.401292844251916e-06,
+      "loss": 1.2427,
+      "step": 143250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4006667180846675e-06,
+      "loss": 1.4908,
+      "step": 143255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.400040623892629e-06,
+      "loss": 1.3007,
+      "step": 143260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.399414561679368e-06,
+      "loss": 0.9989,
+      "step": 143265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.39878853144846e-06,
+      "loss": 1.1801,
+      "step": 143270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.398162533203483e-06,
+      "loss": 1.3589,
+      "step": 143275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.397536566948013e-06,
+      "loss": 1.0087,
+      "step": 143280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.39691063268562e-06,
+      "loss": 1.2067,
+      "step": 143285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.396284730419883e-06,
+      "loss": 0.8962,
+      "step": 143290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.395658860154375e-06,
+      "loss": 1.137,
+      "step": 143295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3950330218926614e-06,
+      "loss": 1.3293,
+      "step": 143300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.394407215638332e-06,
+      "loss": 1.116,
+      "step": 143305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.393781441394953e-06,
+      "loss": 1.1063,
+      "step": 143310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.393155699166093e-06,
+      "loss": 1.1122,
+      "step": 143315
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.392529988955333e-06,
+      "loss": 1.1687,
+      "step": 143320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.391904310766239e-06,
+      "loss": 1.1411,
+      "step": 143325
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.391278664602392e-06,
+      "loss": 1.2852,
+      "step": 143330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.390653050467356e-06,
+      "loss": 1.2009,
+      "step": 143335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.390027468364712e-06,
+      "loss": 1.3017,
+      "step": 143340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.389401918298025e-06,
+      "loss": 1.3548,
+      "step": 143345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.388776400270871e-06,
+      "loss": 1.2105,
+      "step": 143350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.388150914286825e-06,
+      "loss": 1.1942,
+      "step": 143355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.387525460349456e-06,
+      "loss": 1.131,
+      "step": 143360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.386900038462331e-06,
+      "loss": 1.2474,
+      "step": 143365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3862746486290255e-06,
+      "loss": 1.1128,
+      "step": 143370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.385649290853115e-06,
+      "loss": 1.1972,
+      "step": 143375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.385023965138165e-06,
+      "loss": 1.4473,
+      "step": 143380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.384398671487749e-06,
+      "loss": 1.192,
+      "step": 143385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.383773409905438e-06,
+      "loss": 1.1718,
+      "step": 143390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3831481803948e-06,
+      "loss": 1.1712,
+      "step": 143395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.382522982959404e-06,
+      "loss": 1.4265,
+      "step": 143400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.381897817602828e-06,
+      "loss": 1.317,
+      "step": 143405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.381272684328637e-06,
+      "loss": 1.5139,
+      "step": 143410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.380647583140397e-06,
+      "loss": 1.1752,
+      "step": 143415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.380022514041683e-06,
+      "loss": 1.5701,
+      "step": 143420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.379397477036066e-06,
+      "loss": 1.4439,
+      "step": 143425
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.378772472127109e-06,
+      "loss": 1.1067,
+      "step": 143430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.378147499318387e-06,
+      "loss": 1.378,
+      "step": 143435
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.377522558613464e-06,
+      "loss": 1.4403,
+      "step": 143440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.376897650015911e-06,
+      "loss": 1.1751,
+      "step": 143445
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.376272773529301e-06,
+      "loss": 1.082,
+      "step": 143450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.375647929157197e-06,
+      "loss": 1.4322,
+      "step": 143455
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.375023116903167e-06,
+      "loss": 1.1307,
+      "step": 143460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.374398336770778e-06,
+      "loss": 1.3713,
+      "step": 143465
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.373773588763604e-06,
+      "loss": 0.9266,
+      "step": 143470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.37314887288521e-06,
+      "loss": 1.1717,
+      "step": 143475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.372524189139159e-06,
+      "loss": 1.1,
+      "step": 143480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.371899537529021e-06,
+      "loss": 1.2483,
+      "step": 143485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.371274918058367e-06,
+      "loss": 1.48,
+      "step": 143490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.370650330730758e-06,
+      "loss": 1.0896,
+      "step": 143495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3700257755497656e-06,
+      "loss": 1.4468,
+      "step": 143500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.369401252518956e-06,
+      "loss": 1.2307,
+      "step": 143505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.368776761641889e-06,
+      "loss": 1.3965,
+      "step": 143510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.368152302922135e-06,
+      "loss": 1.4471,
+      "step": 143515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3675278763632645e-06,
+      "loss": 1.1204,
+      "step": 143520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.366903481968835e-06,
+      "loss": 1.2379,
+      "step": 143525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.36627911974242e-06,
+      "loss": 1.1641,
+      "step": 143530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.365654789687579e-06,
+      "loss": 0.8577,
+      "step": 143535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3650304918078825e-06,
+      "loss": 1.1078,
+      "step": 143540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.36440622610689e-06,
+      "loss": 1.3852,
+      "step": 143545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.363781992588171e-06,
+      "loss": 1.3707,
+      "step": 143550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3631577912552854e-06,
+      "loss": 1.1522,
+      "step": 143555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.362533622111801e-06,
+      "loss": 1.3408,
+      "step": 143560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.361909485161285e-06,
+      "loss": 1.5495,
+      "step": 143565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.361285380407298e-06,
+      "loss": 1.0651,
+      "step": 143570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.360661307853401e-06,
+      "loss": 1.2404,
+      "step": 143575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.360037267503161e-06,
+      "loss": 1.1254,
+      "step": 143580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.359413259360146e-06,
+      "loss": 1.2731,
+      "step": 143585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.35878928342791e-06,
+      "loss": 1.4272,
+      "step": 143590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.358165339710027e-06,
+      "loss": 1.5074,
+      "step": 143595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3575414282100525e-06,
+      "loss": 1.2622,
+      "step": 143600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.356917548931544e-06,
+      "loss": 1.162,
+      "step": 143605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.356293701878079e-06,
+      "loss": 1.1927,
+      "step": 143610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.355669887053213e-06,
+      "loss": 1.3404,
+      "step": 143615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.355046104460508e-06,
+      "loss": 1.2495,
+      "step": 143620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3544223541035216e-06,
+      "loss": 1.0663,
+      "step": 143625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.353798635985821e-06,
+      "loss": 1.5181,
+      "step": 143630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.35317495011097e-06,
+      "loss": 1.4896,
+      "step": 143635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3525512964825225e-06,
+      "loss": 1.0723,
+      "step": 143640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.35192767510405e-06,
+      "loss": 1.1779,
+      "step": 143645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.351304085979104e-06,
+      "loss": 1.2633,
+      "step": 143650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.350680529111249e-06,
+      "loss": 1.4585,
+      "step": 143655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.350057004504051e-06,
+      "loss": 1.1211,
+      "step": 143660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3494335121610665e-06,
+      "loss": 1.495,
+      "step": 143665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.348810052085852e-06,
+      "loss": 1.1654,
+      "step": 143670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.348186624281971e-06,
+      "loss": 1.4043,
+      "step": 143675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.347563228752989e-06,
+      "loss": 1.2643,
+      "step": 143680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.346939865502461e-06,
+      "loss": 1.0889,
+      "step": 143685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.346316534533942e-06,
+      "loss": 1.1742,
+      "step": 143690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.345693235851002e-06,
+      "loss": 1.4965,
+      "step": 143695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.345069969457189e-06,
+      "loss": 1.2134,
+      "step": 143700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.344446735356069e-06,
+      "loss": 1.3025,
+      "step": 143705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3438235335512045e-06,
+      "loss": 0.9508,
+      "step": 143710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.343200364046148e-06,
+      "loss": 1.4999,
+      "step": 143715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.342577226844458e-06,
+      "loss": 1.6428,
+      "step": 143720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.341954121949694e-06,
+      "loss": 1.0625,
+      "step": 143725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.341331049365419e-06,
+      "loss": 1.3608,
+      "step": 143730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.340708009095184e-06,
+      "loss": 1.1542,
+      "step": 143735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.340085001142553e-06,
+      "loss": 1.2337,
+      "step": 143740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.339462025511077e-06,
+      "loss": 1.4378,
+      "step": 143745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3388390822043196e-06,
+      "loss": 1.4473,
+      "step": 143750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.338216171225837e-06,
+      "loss": 1.3153,
+      "step": 143755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.337593292579186e-06,
+      "loss": 1.243,
+      "step": 143760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.33697044626792e-06,
+      "loss": 1.1669,
+      "step": 143765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.336347632295598e-06,
+      "loss": 1.2472,
+      "step": 143770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.33572485066578e-06,
+      "loss": 1.5898,
+      "step": 143775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3351021013820204e-06,
+      "loss": 1.0247,
+      "step": 143780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3344793844478716e-06,
+      "loss": 1.2451,
+      "step": 143785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.333856699866897e-06,
+      "loss": 1.3521,
+      "step": 143790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.333234047642643e-06,
+      "loss": 1.4717,
+      "step": 143795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3326114277786705e-06,
+      "loss": 1.5236,
+      "step": 143800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3319888402785395e-06,
+      "loss": 1.3963,
+      "step": 143805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.331366285145799e-06,
+      "loss": 1.1991,
+      "step": 143810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.330743762384004e-06,
+      "loss": 1.3092,
+      "step": 143815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.330121271996711e-06,
+      "loss": 1.3926,
+      "step": 143820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.329498813987479e-06,
+      "loss": 1.0941,
+      "step": 143825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.328876388359858e-06,
+      "loss": 1.4784,
+      "step": 143830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.328253995117399e-06,
+      "loss": 1.4175,
+      "step": 143835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.327631634263661e-06,
+      "loss": 1.2313,
+      "step": 143840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3270093058022025e-06,
+      "loss": 1.3885,
+      "step": 143845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.326387009736566e-06,
+      "loss": 1.4972,
+      "step": 143850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3257647460703156e-06,
+      "loss": 1.4126,
+      "step": 143855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.325142514806997e-06,
+      "loss": 1.3511,
+      "step": 143860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.324520315950167e-06,
+      "loss": 1.3097,
+      "step": 143865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.323898149503382e-06,
+      "loss": 1.4903,
+      "step": 143870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.323276015470191e-06,
+      "loss": 1.3913,
+      "step": 143875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.322653913854145e-06,
+      "loss": 1.4079,
+      "step": 143880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.322031844658797e-06,
+      "loss": 1.2401,
+      "step": 143885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.321409807887705e-06,
+      "loss": 1.1645,
+      "step": 143890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.320787803544414e-06,
+      "loss": 1.3495,
+      "step": 143895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.320165831632484e-06,
+      "loss": 1.1843,
+      "step": 143900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.31954389215546e-06,
+      "loss": 1.3972,
+      "step": 143905
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.318921985116893e-06,
+      "loss": 1.1048,
+      "step": 143910
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.318300110520337e-06,
+      "loss": 1.2969,
+      "step": 143915
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.317678268369347e-06,
+      "loss": 1.3983,
+      "step": 143920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.31705645866747e-06,
+      "loss": 1.1527,
+      "step": 143925
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.316434681418254e-06,
+      "loss": 1.343,
+      "step": 143930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.315812936625252e-06,
+      "loss": 0.9838,
+      "step": 143935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.31519122429202e-06,
+      "loss": 1.1508,
+      "step": 143940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.314569544422099e-06,
+      "loss": 1.3297,
+      "step": 143945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.313947897019049e-06,
+      "loss": 1.2135,
+      "step": 143950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.313326282086409e-06,
+      "loss": 1.0549,
+      "step": 143955
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3127046996277366e-06,
+      "loss": 1.0341,
+      "step": 143960
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.312083149646582e-06,
+      "loss": 1.0117,
+      "step": 143965
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.311461632146492e-06,
+      "loss": 1.2851,
+      "step": 143970
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.310840147131016e-06,
+      "loss": 0.9979,
+      "step": 143975
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.310218694603695e-06,
+      "loss": 1.2866,
+      "step": 143980
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3095972745680926e-06,
+      "loss": 1.5791,
+      "step": 143985
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3089758870277505e-06,
+      "loss": 1.1531,
+      "step": 143990
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3083545319862145e-06,
+      "loss": 1.2017,
+      "step": 143995
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3077332094470385e-06,
+      "loss": 1.1478,
+      "step": 144000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.307111919413762e-06,
+      "loss": 1.4995,
+      "step": 144005
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.30649066188994e-06,
+      "loss": 1.1827,
+      "step": 144010
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.305869436879122e-06,
+      "loss": 1.2298,
+      "step": 144015
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.305248244384852e-06,
+      "loss": 1.1987,
+      "step": 144020
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3046270844106744e-06,
+      "loss": 1.276,
+      "step": 144025
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.304005956960139e-06,
+      "loss": 1.4185,
+      "step": 144030
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3033848620367965e-06,
+      "loss": 0.8335,
+      "step": 144035
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.302763799644187e-06,
+      "loss": 1.5228,
+      "step": 144040
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.302142769785864e-06,
+      "loss": 1.3131,
+      "step": 144045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.301521772465367e-06,
+      "loss": 1.3348,
+      "step": 144050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.300900807686249e-06,
+      "loss": 1.4236,
+      "step": 144055
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.300279875452049e-06,
+      "loss": 1.1682,
+      "step": 144060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.299658975766321e-06,
+      "loss": 1.4554,
+      "step": 144065
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.299038108632602e-06,
+      "loss": 1.4027,
+      "step": 144070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.298417274054441e-06,
+      "loss": 1.2841,
+      "step": 144075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.297796472035388e-06,
+      "loss": 1.7688,
+      "step": 144080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.297175702578984e-06,
+      "loss": 1.0607,
+      "step": 144085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2965549656887704e-06,
+      "loss": 1.175,
+      "step": 144090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2959342613682996e-06,
+      "loss": 1.1617,
+      "step": 144095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.295313589621108e-06,
+      "loss": 1.4153,
+      "step": 144100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.294692950450745e-06,
+      "loss": 1.1851,
+      "step": 144105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.294072343860757e-06,
+      "loss": 1.1159,
+      "step": 144110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.293451769854685e-06,
+      "loss": 1.2672,
+      "step": 144115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.292831228436069e-06,
+      "loss": 1.1236,
+      "step": 144120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.292210719608456e-06,
+      "loss": 1.0009,
+      "step": 144125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.291590243375393e-06,
+      "loss": 1.4615,
+      "step": 144130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.290969799740421e-06,
+      "loss": 1.1241,
+      "step": 144135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.290349388707077e-06,
+      "loss": 1.2566,
+      "step": 144140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.28972901027891e-06,
+      "loss": 1.268,
+      "step": 144145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.289108664459465e-06,
+      "loss": 1.3028,
+      "step": 144150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.288488351252278e-06,
+      "loss": 1.4216,
+      "step": 144155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.287868070660898e-06,
+      "loss": 1.0711,
+      "step": 144160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.287247822688859e-06,
+      "loss": 1.1259,
+      "step": 144165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.286627607339708e-06,
+      "loss": 0.8753,
+      "step": 144170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.286007424616991e-06,
+      "loss": 1.2393,
+      "step": 144175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.285387274524243e-06,
+      "loss": 1.2726,
+      "step": 144180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.284767157065005e-06,
+      "loss": 1.3199,
+      "step": 144185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2841470722428205e-06,
+      "loss": 1.0789,
+      "step": 144190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.283527020061234e-06,
+      "loss": 0.95,
+      "step": 144195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.282907000523782e-06,
+      "loss": 1.0663,
+      "step": 144200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.282287013634003e-06,
+      "loss": 1.2035,
+      "step": 144205
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.281667059395444e-06,
+      "loss": 1.1712,
+      "step": 144210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.281047137811638e-06,
+      "loss": 1.0471,
+      "step": 144215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.280427248886129e-06,
+      "loss": 1.1776,
+      "step": 144220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.279807392622461e-06,
+      "loss": 1.3702,
+      "step": 144225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.279187569024168e-06,
+      "loss": 1.2046,
+      "step": 144230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.278567778094788e-06,
+      "loss": 1.1909,
+      "step": 144235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.277948019837863e-06,
+      "loss": 1.7131,
+      "step": 144240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2773282942569365e-06,
+      "loss": 0.9412,
+      "step": 144245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.276708601355539e-06,
+      "loss": 1.288,
+      "step": 144250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2760889411372185e-06,
+      "loss": 1.2136,
+      "step": 144255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.275469313605505e-06,
+      "loss": 0.9128,
+      "step": 144260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.274849718763943e-06,
+      "loss": 1.1332,
+      "step": 144265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.274230156616066e-06,
+      "loss": 1.6019,
+      "step": 144270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.273610627165418e-06,
+      "loss": 1.2927,
+      "step": 144275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2729911304155326e-06,
+      "loss": 1.2236,
+      "step": 144280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.272371666369941e-06,
+      "loss": 1.2375,
+      "step": 144285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.271752235032195e-06,
+      "loss": 1.4427,
+      "step": 144290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.271132836405824e-06,
+      "loss": 1.3094,
+      "step": 144295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.270513470494364e-06,
+      "loss": 1.1846,
+      "step": 144300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2698941373013555e-06,
+      "loss": 0.7753,
+      "step": 144305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.269274836830332e-06,
+      "loss": 1.1025,
+      "step": 144310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.26865556908483e-06,
+      "loss": 1.2499,
+      "step": 144315
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.268036334068391e-06,
+      "loss": 1.1817,
+      "step": 144320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.267417131784547e-06,
+      "loss": 1.2667,
+      "step": 144325
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.266797962236832e-06,
+      "loss": 1.281,
+      "step": 144330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.266178825428783e-06,
+      "loss": 1.4581,
+      "step": 144335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2655597213639414e-06,
+      "loss": 1.002,
+      "step": 144340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2649406500458375e-06,
+      "loss": 0.9415,
+      "step": 144345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.264321611478004e-06,
+      "loss": 1.21,
+      "step": 144350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2637026056639785e-06,
+      "loss": 1.0378,
+      "step": 144355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2630836326072996e-06,
+      "loss": 1.2504,
+      "step": 144360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.262464692311497e-06,
+      "loss": 0.9999,
+      "step": 144365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2618457847801085e-06,
+      "loss": 1.3397,
+      "step": 144370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.261226910016668e-06,
+      "loss": 1.2086,
+      "step": 144375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2606080680247016e-06,
+      "loss": 1.2478,
+      "step": 144380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.2599892588077565e-06,
+      "loss": 1.4003,
+      "step": 144385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.25937048236936e-06,
+      "loss": 1.3762,
+      "step": 144390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.258751738713043e-06,
+      "loss": 1.0979,
+      "step": 144395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.258133027842344e-06,
+      "loss": 1.3921,
+      "step": 144400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.25751434976079e-06,
+      "loss": 0.9892,
+      "step": 144405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.256895704471923e-06,
+      "loss": 1.0697,
+      "step": 144410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.256277091979265e-06,
+      "loss": 1.1621,
+      "step": 144415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.255658512286358e-06,
+      "loss": 0.957,
+      "step": 144420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.255039965396728e-06,
+      "loss": 1.0852,
+      "step": 144425
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.254421451313908e-06,
+      "loss": 1.3053,
+      "step": 144430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.253802970041437e-06,
+      "loss": 1.3591,
+      "step": 144435
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.253184521582842e-06,
+      "loss": 1.0958,
+      "step": 144440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.252566105941649e-06,
+      "loss": 1.0371,
+      "step": 144445
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.251947723121395e-06,
+      "loss": 1.2838,
+      "step": 144450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.251329373125616e-06,
+      "loss": 1.0944,
+      "step": 144455
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.250711055957833e-06,
+      "loss": 1.2555,
+      "step": 144460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.250092771621587e-06,
+      "loss": 1.0799,
+      "step": 144465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2494745201204e-06,
+      "loss": 0.9663,
+      "step": 144470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.248856301457805e-06,
+      "loss": 1.5636,
+      "step": 144475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2482381156373395e-06,
+      "loss": 0.984,
+      "step": 144480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.247619962662527e-06,
+      "loss": 1.3058,
+      "step": 144485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2470018425368985e-06,
+      "loss": 1.0345,
+      "step": 144490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2463837552639805e-06,
+      "loss": 1.1822,
+      "step": 144495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2457657008473054e-06,
+      "loss": 1.2097,
+      "step": 144500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.245147679290407e-06,
+      "loss": 1.0785,
+      "step": 144505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.244529690596807e-06,
+      "loss": 1.1766,
+      "step": 144510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.243911734770041e-06,
+      "loss": 1.1825,
+      "step": 144515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.243293811813632e-06,
+      "loss": 1.2348,
+      "step": 144520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.242675921731112e-06,
+      "loss": 1.0912,
+      "step": 144525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2420580645260125e-06,
+      "loss": 1.4334,
+      "step": 144530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.241440240201857e-06,
+      "loss": 1.5572,
+      "step": 144535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.240822448762173e-06,
+      "loss": 1.4405,
+      "step": 144540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.24020469021049e-06,
+      "loss": 1.5613,
+      "step": 144545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.239586964550338e-06,
+      "loss": 1.0009,
+      "step": 144550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.238969271785245e-06,
+      "loss": 1.3643,
+      "step": 144555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.238351611918732e-06,
+      "loss": 1.3184,
+      "step": 144560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.23773398495433e-06,
+      "loss": 1.079,
+      "step": 144565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.23711639089557e-06,
+      "loss": 1.2517,
+      "step": 144570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.23649882974597e-06,
+      "loss": 1.0934,
+      "step": 144575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.235881301509066e-06,
+      "loss": 1.3282,
+      "step": 144580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.235263806188381e-06,
+      "loss": 1.1342,
+      "step": 144585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.234646343787432e-06,
+      "loss": 1.3414,
+      "step": 144590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.23402891430976e-06,
+      "loss": 1.2678,
+      "step": 144595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.233411517758885e-06,
+      "loss": 1.3069,
+      "step": 144600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.232794154138329e-06,
+      "loss": 0.9994,
+      "step": 144605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.232176823451622e-06,
+      "loss": 1.009,
+      "step": 144610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.231559525702286e-06,
+      "loss": 1.1864,
+      "step": 144615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230942260893847e-06,
+      "loss": 1.2722,
+      "step": 144620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2303250290298325e-06,
+      "loss": 1.3578,
+      "step": 144625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.229707830113768e-06,
+      "loss": 1.1674,
+      "step": 144630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.229090664149169e-06,
+      "loss": 1.5126,
+      "step": 144635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.228473531139568e-06,
+      "loss": 1.0799,
+      "step": 144640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.227856431088491e-06,
+      "loss": 1.1762,
+      "step": 144645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.227239363999458e-06,
+      "loss": 1.329,
+      "step": 144650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.226622329875989e-06,
+      "loss": 1.2025,
+      "step": 144655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.226005328721612e-06,
+      "loss": 1.0574,
+      "step": 144660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2253883605398534e-06,
+      "loss": 1.2615,
+      "step": 144665
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2247714253342305e-06,
+      "loss": 1.4648,
+      "step": 144670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2241545231082724e-06,
+      "loss": 0.8715,
+      "step": 144675
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.223537653865498e-06,
+      "loss": 1.1998,
+      "step": 144680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2229208176094235e-06,
+      "loss": 1.4567,
+      "step": 144685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.222304014343586e-06,
+      "loss": 1.1508,
+      "step": 144690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.221687244071499e-06,
+      "loss": 1.2894,
+      "step": 144695
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2210705067966875e-06,
+      "loss": 1.1245,
+      "step": 144700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.220453802522666e-06,
+      "loss": 1.1294,
+      "step": 144705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.219837131252963e-06,
+      "loss": 1.271,
+      "step": 144710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.219220492991102e-06,
+      "loss": 0.9322,
+      "step": 144715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.218603887740597e-06,
+      "loss": 1.1653,
+      "step": 144720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.217987315504979e-06,
+      "loss": 1.3627,
+      "step": 144725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.217370776287758e-06,
+      "loss": 1.3188,
+      "step": 144730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.21675427009246e-06,
+      "loss": 1.6291,
+      "step": 144735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.21613779692261e-06,
+      "loss": 1.5745,
+      "step": 144740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.215521356781725e-06,
+      "loss": 1.2886,
+      "step": 144745
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2149049496733185e-06,
+      "loss": 1.1092,
+      "step": 144750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2142885756009176e-06,
+      "loss": 1.4341,
+      "step": 144755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.213672234568044e-06,
+      "loss": 1.1566,
+      "step": 144760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.21305592657821e-06,
+      "loss": 0.9487,
+      "step": 144765
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.212439651634943e-06,
+      "loss": 1.2506,
+      "step": 144770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.211823409741756e-06,
+      "loss": 1.2246,
+      "step": 144775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.211207200902173e-06,
+      "loss": 1.2825,
+      "step": 144780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.210591025119707e-06,
+      "loss": 1.2639,
+      "step": 144785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.209974882397884e-06,
+      "loss": 1.1385,
+      "step": 144790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2093587727402185e-06,
+      "loss": 1.3674,
+      "step": 144795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2087426961502254e-06,
+      "loss": 1.3988,
+      "step": 144800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.208126652631426e-06,
+      "loss": 1.0549,
+      "step": 144805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.207510642187341e-06,
+      "loss": 1.1372,
+      "step": 144810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.206894664821485e-06,
+      "loss": 1.1125,
+      "step": 144815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.206278720537378e-06,
+      "loss": 1.1235,
+      "step": 144820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.205662809338532e-06,
+      "loss": 1.0399,
+      "step": 144825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.205046931228468e-06,
+      "loss": 1.1638,
+      "step": 144830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.204431086210707e-06,
+      "loss": 1.1152,
+      "step": 144835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.203815274288762e-06,
+      "loss": 1.1437,
+      "step": 144840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2031994954661445e-06,
+      "loss": 1.4788,
+      "step": 144845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.202583749746376e-06,
+      "loss": 1.1244,
+      "step": 144850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.201968037132976e-06,
+      "loss": 1.157,
+      "step": 144855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.201352357629458e-06,
+      "loss": 1.059,
+      "step": 144860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.200736711239333e-06,
+      "loss": 1.2804,
+      "step": 144865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.200121097966119e-06,
+      "loss": 1.5631,
+      "step": 144870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.199505517813338e-06,
+      "loss": 1.011,
+      "step": 144875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.198889970784497e-06,
+      "loss": 1.1319,
+      "step": 144880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.198274456883117e-06,
+      "loss": 1.0761,
+      "step": 144885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.19765897611271e-06,
+      "loss": 1.2689,
+      "step": 144890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.197043528476788e-06,
+      "loss": 1.4624,
+      "step": 144895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.19642811397887e-06,
+      "loss": 1.2087,
+      "step": 144900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.195812732622471e-06,
+      "loss": 1.2497,
+      "step": 144905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1951973844111e-06,
+      "loss": 1.1897,
+      "step": 144910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.194582069348277e-06,
+      "loss": 1.2282,
+      "step": 144915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.193966787437509e-06,
+      "loss": 1.2158,
+      "step": 144920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.193351538682319e-06,
+      "loss": 1.5825,
+      "step": 144925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.19273632308621e-06,
+      "loss": 1.1632,
+      "step": 144930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1921211406527025e-06,
+      "loss": 1.1672,
+      "step": 144935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.191505991385305e-06,
+      "loss": 1.0442,
+      "step": 144940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.190890875287532e-06,
+      "loss": 1.1165,
+      "step": 144945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1902757923629e-06,
+      "loss": 1.129,
+      "step": 144950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.189660742614918e-06,
+      "loss": 1.067,
+      "step": 144955
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.189045726047095e-06,
+      "loss": 1.1701,
+      "step": 144960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.188430742662946e-06,
+      "loss": 1.0798,
+      "step": 144965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.187815792465987e-06,
+      "loss": 1.2546,
+      "step": 144970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1872008754597214e-06,
+      "loss": 1.0047,
+      "step": 144975
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.186585991647671e-06,
+      "loss": 1.2905,
+      "step": 144980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.185971141033339e-06,
+      "loss": 1.8965,
+      "step": 144985
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.185356323620236e-06,
+      "loss": 1.0672,
+      "step": 144990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1847415394118765e-06,
+      "loss": 1.1292,
+      "step": 144995
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1841267884117746e-06,
+      "loss": 1.0949,
+      "step": 145000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1835120706234355e-06,
+      "loss": 1.2493,
+      "step": 145005
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.182897386050368e-06,
+      "loss": 1.2964,
+      "step": 145010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.182282734696085e-06,
+      "loss": 1.2526,
+      "step": 145015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1816681165640995e-06,
+      "loss": 1.0965,
+      "step": 145020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.181053531657916e-06,
+      "loss": 1.1461,
+      "step": 145025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1804389799810495e-06,
+      "loss": 1.3552,
+      "step": 145030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1798244615370035e-06,
+      "loss": 1.2085,
+      "step": 145035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1792099763292895e-06,
+      "loss": 1.2768,
+      "step": 145040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.178595524361421e-06,
+      "loss": 1.1137,
+      "step": 145045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.177981105636904e-06,
+      "loss": 1.1657,
+      "step": 145050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.177366720159242e-06,
+      "loss": 1.2229,
+      "step": 145055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.176752367931946e-06,
+      "loss": 1.4591,
+      "step": 145060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.176138048958533e-06,
+      "loss": 1.2333,
+      "step": 145065
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.175523763242501e-06,
+      "loss": 1.0082,
+      "step": 145070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.174909510787359e-06,
+      "loss": 1.6802,
+      "step": 145075
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.17429529159662e-06,
+      "loss": 1.1885,
+      "step": 145080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.173681105673786e-06,
+      "loss": 1.2219,
+      "step": 145085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.173066953022366e-06,
+      "loss": 1.2912,
+      "step": 145090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.17245283364587e-06,
+      "loss": 1.4989,
+      "step": 145095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.171838747547804e-06,
+      "loss": 1.3358,
+      "step": 145100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.17122469473167e-06,
+      "loss": 1.2234,
+      "step": 145105
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.170610675200976e-06,
+      "loss": 1.1427,
+      "step": 145110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.169996688959236e-06,
+      "loss": 1.3842,
+      "step": 145115
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.169382736009946e-06,
+      "loss": 1.1662,
+      "step": 145120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.16876881635662e-06,
+      "loss": 1.1008,
+      "step": 145125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.168154930002757e-06,
+      "loss": 1.0428,
+      "step": 145130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.16754107695187e-06,
+      "loss": 1.2688,
+      "step": 145135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.166927257207457e-06,
+      "loss": 1.1699,
+      "step": 145140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.16631347077303e-06,
+      "loss": 1.0688,
+      "step": 145145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.165699717652088e-06,
+      "loss": 1.2022,
+      "step": 145150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1650859978481376e-06,
+      "loss": 1.179,
+      "step": 145155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.164472311364688e-06,
+      "loss": 1.0839,
+      "step": 145160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.16385865820524e-06,
+      "loss": 1.2894,
+      "step": 145165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1632450383732945e-06,
+      "loss": 1.1815,
+      "step": 145170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.16263145187236e-06,
+      "loss": 1.4174,
+      "step": 145175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.162017898705942e-06,
+      "loss": 1.445,
+      "step": 145180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.161404378877539e-06,
+      "loss": 1.5632,
+      "step": 145185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.160790892390661e-06,
+      "loss": 1.2876,
+      "step": 145190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.160177439248808e-06,
+      "loss": 1.3273,
+      "step": 145195
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1595640194554775e-06,
+      "loss": 1.1445,
+      "step": 145200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.15895063301418e-06,
+      "loss": 0.9588,
+      "step": 145205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.158337279928418e-06,
+      "loss": 1.1745,
+      "step": 145210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.157723960201693e-06,
+      "loss": 1.2819,
+      "step": 145215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.157110673837503e-06,
+      "loss": 0.9767,
+      "step": 145220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1564974208393515e-06,
+      "loss": 1.422,
+      "step": 145225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.155884201210748e-06,
+      "loss": 1.2437,
+      "step": 145230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1552710149551854e-06,
+      "loss": 1.6682,
+      "step": 145235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.154657862076172e-06,
+      "loss": 1.2252,
+      "step": 145240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.154044742577202e-06,
+      "loss": 1.4126,
+      "step": 145245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1534316564617824e-06,
+      "loss": 1.2026,
+      "step": 145250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1528186037334154e-06,
+      "loss": 1.2483,
+      "step": 145255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.152205584395599e-06,
+      "loss": 1.225,
+      "step": 145260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.15159259845183e-06,
+      "loss": 0.9097,
+      "step": 145265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.150979645905614e-06,
+      "loss": 1.4626,
+      "step": 145270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.150366726760453e-06,
+      "loss": 1.0611,
+      "step": 145275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.14975384101984e-06,
+      "loss": 1.1501,
+      "step": 145280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.149140988687283e-06,
+      "loss": 1.0254,
+      "step": 145285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1485281697662796e-06,
+      "loss": 1.1893,
+      "step": 145290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1479153842603225e-06,
+      "loss": 1.0326,
+      "step": 145295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.147302632172916e-06,
+      "loss": 1.0742,
+      "step": 145300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.146689913507563e-06,
+      "loss": 1.2852,
+      "step": 145305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.14607722826776e-06,
+      "loss": 1.2194,
+      "step": 145310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.145464576456999e-06,
+      "loss": 1.01,
+      "step": 145315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1448519580787846e-06,
+      "loss": 1.0763,
+      "step": 145320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.144239373136619e-06,
+      "loss": 1.088,
+      "step": 145325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1436268216339925e-06,
+      "loss": 1.0781,
+      "step": 145330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.143014303574411e-06,
+      "loss": 0.8766,
+      "step": 145335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.142401818961364e-06,
+      "loss": 1.2492,
+      "step": 145340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.141789367798352e-06,
+      "loss": 1.197,
+      "step": 145345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.141176950088877e-06,
+      "loss": 1.033,
+      "step": 145350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.140564565836435e-06,
+      "loss": 1.3404,
+      "step": 145355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.139952215044519e-06,
+      "loss": 1.3422,
+      "step": 145360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.139339897716621e-06,
+      "loss": 1.5049,
+      "step": 145365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.138727613856251e-06,
+      "loss": 1.1007,
+      "step": 145370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1381153634669e-06,
+      "loss": 1.6073,
+      "step": 145375
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.137503146552059e-06,
+      "loss": 1.1229,
+      "step": 145380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.136890963115232e-06,
+      "loss": 0.9669,
+      "step": 145385
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.136278813159908e-06,
+      "loss": 1.4042,
+      "step": 145390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.135666696689584e-06,
+      "loss": 1.0359,
+      "step": 145395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.135054613707762e-06,
+      "loss": 1.2607,
+      "step": 145400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.134442564217932e-06,
+      "loss": 1.0994,
+      "step": 145405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1338305482235875e-06,
+      "loss": 1.166,
+      "step": 145410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.133218565728224e-06,
+      "loss": 1.022,
+      "step": 145415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.132606616735343e-06,
+      "loss": 0.9769,
+      "step": 145420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.131994701248429e-06,
+      "loss": 1.4267,
+      "step": 145425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1313828192709874e-06,
+      "loss": 1.0515,
+      "step": 145430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1307709708065006e-06,
+      "loss": 1.229,
+      "step": 145435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.130159155858472e-06,
+      "loss": 1.7291,
+      "step": 145440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1295473744303894e-06,
+      "loss": 1.1095,
+      "step": 145445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.128935626525751e-06,
+      "loss": 1.0493,
+      "step": 145450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.128323912148045e-06,
+      "loss": 1.5703,
+      "step": 145455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.127712231300768e-06,
+      "loss": 1.1273,
+      "step": 145460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.127100583987416e-06,
+      "loss": 1.2203,
+      "step": 145465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.126488970211478e-06,
+      "loss": 1.4218,
+      "step": 145470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1258773899764424e-06,
+      "loss": 1.3723,
+      "step": 145475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.125265843285812e-06,
+      "loss": 1.0258,
+      "step": 145480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.124654330143069e-06,
+      "loss": 1.4214,
+      "step": 145485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.124042850551708e-06,
+      "loss": 1.2066,
+      "step": 145490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.123431404515229e-06,
+      "loss": 1.1291,
+      "step": 145495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1228199920371155e-06,
+      "loss": 1.1017,
+      "step": 145500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.122208613120858e-06,
+      "loss": 1.2284,
+      "step": 145505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.121597267769951e-06,
+      "loss": 1.0844,
+      "step": 145510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1209859559878894e-06,
+      "loss": 1.1435,
+      "step": 145515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.12037467777816e-06,
+      "loss": 1.3576,
+      "step": 145520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1197634331442494e-06,
+      "loss": 1.2119,
+      "step": 145525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.119152222089652e-06,
+      "loss": 1.0921,
+      "step": 145530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.118541044617862e-06,
+      "loss": 1.2208,
+      "step": 145535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.117929900732363e-06,
+      "loss": 1.1512,
+      "step": 145540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.117318790436651e-06,
+      "loss": 1.1444,
+      "step": 145545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.116707713734209e-06,
+      "loss": 1.3466,
+      "step": 145550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.116096670628531e-06,
+      "loss": 1.2653,
+      "step": 145555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.115485661123109e-06,
+      "loss": 1.212,
+      "step": 145560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.114874685221429e-06,
+      "loss": 1.0109,
+      "step": 145565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1142637429269765e-06,
+      "loss": 0.9504,
+      "step": 145570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.113652834243242e-06,
+      "loss": 1.3449,
+      "step": 145575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.113041959173721e-06,
+      "loss": 1.1128,
+      "step": 145580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1124311177218964e-06,
+      "loss": 1.0309,
+      "step": 145585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.111820309891253e-06,
+      "loss": 1.4336,
+      "step": 145590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.111209535685286e-06,
+      "loss": 1.1451,
+      "step": 145595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1105987951074775e-06,
+      "loss": 1.0477,
+      "step": 145600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.109988088161316e-06,
+      "loss": 1.1263,
+      "step": 145605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.109377414850295e-06,
+      "loss": 1.2454,
+      "step": 145610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.108766775177897e-06,
+      "loss": 1.3803,
+      "step": 145615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.108156169147606e-06,
+      "loss": 1.1497,
+      "step": 145620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.107545596762912e-06,
+      "loss": 1.1693,
+      "step": 145625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.106935058027306e-06,
+      "loss": 1.4914,
+      "step": 145630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.106324552944266e-06,
+      "loss": 1.2404,
+      "step": 145635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.1057140815172865e-06,
+      "loss": 1.1879,
+      "step": 145640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.105103643749846e-06,
+      "loss": 1.3718,
+      "step": 145645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.104493239645438e-06,
+      "loss": 1.2213,
+      "step": 145650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.103882869207542e-06,
+      "loss": 1.1342,
+      "step": 145655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.103272532439648e-06,
+      "loss": 1.0795,
+      "step": 145660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.10266222934524e-06,
+      "loss": 1.0772,
+      "step": 145665
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.102051959927797e-06,
+      "loss": 1.5174,
+      "step": 145670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.101441724190814e-06,
+      "loss": 1.0126,
+      "step": 145675
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.100831522137773e-06,
+      "loss": 1.2162,
+      "step": 145680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.100221353772153e-06,
+      "loss": 1.3235,
+      "step": 145685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.099611219097446e-06,
+      "loss": 1.2768,
+      "step": 145690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.099001118117129e-06,
+      "loss": 1.0602,
+      "step": 145695
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.098391050834688e-06,
+      "loss": 1.2851,
+      "step": 145700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.097781017253613e-06,
+      "loss": 1.4399,
+      "step": 145705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0971710173773825e-06,
+      "loss": 1.1727,
+      "step": 145710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.096561051209478e-06,
+      "loss": 0.9051,
+      "step": 145715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.095951118753384e-06,
+      "loss": 1.5486,
+      "step": 145720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.095341220012588e-06,
+      "loss": 1.3041,
+      "step": 145725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0947313549905694e-06,
+      "loss": 1.2939,
+      "step": 145730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0941215236908085e-06,
+      "loss": 1.1365,
+      "step": 145735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.09351172611679e-06,
+      "loss": 1.1059,
+      "step": 145740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.092901962271999e-06,
+      "loss": 1.3764,
+      "step": 145745
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.092292232159913e-06,
+      "loss": 1.1886,
+      "step": 145750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.091682535784017e-06,
+      "loss": 0.9491,
+      "step": 145755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.09107287314779e-06,
+      "loss": 1.3499,
+      "step": 145760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.090463244254713e-06,
+      "loss": 1.3133,
+      "step": 145765
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.089853649108275e-06,
+      "loss": 1.1346,
+      "step": 145770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.08924408771195e-06,
+      "loss": 1.3761,
+      "step": 145775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.088634560069218e-06,
+      "loss": 1.3381,
+      "step": 145780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.088025066183565e-06,
+      "loss": 1.3273,
+      "step": 145785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.087415606058465e-06,
+      "loss": 1.1865,
+      "step": 145790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0868061796974055e-06,
+      "loss": 1.134,
+      "step": 145795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.08619678710386e-06,
+      "loss": 1.2271,
+      "step": 145800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.085587428281315e-06,
+      "loss": 1.4587,
+      "step": 145805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.084978103233242e-06,
+      "loss": 1.2917,
+      "step": 145810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.084368811963127e-06,
+      "loss": 1.2312,
+      "step": 145815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.083759554474451e-06,
+      "loss": 1.5047,
+      "step": 145820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.083150330770691e-06,
+      "loss": 1.217,
+      "step": 145825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.08254114085532e-06,
+      "loss": 0.9217,
+      "step": 145830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.081931984731823e-06,
+      "loss": 1.0268,
+      "step": 145835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.081322862403679e-06,
+      "loss": 1.1017,
+      "step": 145840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.080713773874363e-06,
+      "loss": 1.1936,
+      "step": 145845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.080104719147358e-06,
+      "loss": 1.3276,
+      "step": 145850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.079495698226136e-06,
+      "loss": 1.1662,
+      "step": 145855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.078886711114177e-06,
+      "loss": 1.1578,
+      "step": 145860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.078277757814964e-06,
+      "loss": 1.2211,
+      "step": 145865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.07766883833197e-06,
+      "loss": 1.4317,
+      "step": 145870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.077059952668671e-06,
+      "loss": 1.3569,
+      "step": 145875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0764511008285404e-06,
+      "loss": 0.9903,
+      "step": 145880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.075842282815066e-06,
+      "loss": 1.1736,
+      "step": 145885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.075233498631719e-06,
+      "loss": 1.1295,
+      "step": 145890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.07462474828197e-06,
+      "loss": 1.2773,
+      "step": 145895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0740160317693055e-06,
+      "loss": 1.1677,
+      "step": 145900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.073407349097194e-06,
+      "loss": 1.3884,
+      "step": 145905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.072798700269113e-06,
+      "loss": 1.0403,
+      "step": 145910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0721900852885436e-06,
+      "loss": 1.4127,
+      "step": 145915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.071581504158957e-06,
+      "loss": 1.2903,
+      "step": 145920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.070972956883824e-06,
+      "loss": 1.0856,
+      "step": 145925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.070364443466625e-06,
+      "loss": 1.2995,
+      "step": 145930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.069755963910838e-06,
+      "loss": 1.7949,
+      "step": 145935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.069147518219935e-06,
+      "loss": 1.369,
+      "step": 145940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.068539106397384e-06,
+      "loss": 1.0946,
+      "step": 145945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.067930728446668e-06,
+      "loss": 1.7851,
+      "step": 145950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.067322384371259e-06,
+      "loss": 1.0529,
+      "step": 145955
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.066714074174627e-06,
+      "loss": 1.2067,
+      "step": 145960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.066105797860252e-06,
+      "loss": 1.4549,
+      "step": 145965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.065497555431606e-06,
+      "loss": 1.239,
+      "step": 145970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.064889346892153e-06,
+      "loss": 1.3342,
+      "step": 145975
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.064281172245381e-06,
+      "loss": 1.4882,
+      "step": 145980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.063673031494756e-06,
+      "loss": 1.118,
+      "step": 145985
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.063064924643747e-06,
+      "loss": 1.1756,
+      "step": 145990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.062456851695835e-06,
+      "loss": 1.2228,
+      "step": 145995
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.061848812654483e-06,
+      "loss": 1.0403,
+      "step": 146000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.061240807523169e-06,
+      "loss": 0.986,
+      "step": 146005
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.060632836305368e-06,
+      "loss": 1.2416,
+      "step": 146010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.060024899004548e-06,
+      "loss": 1.3264,
+      "step": 146015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.059416995624176e-06,
+      "loss": 1.1546,
+      "step": 146020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.058809126167729e-06,
+      "loss": 1.2363,
+      "step": 146025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.058201290638681e-06,
+      "loss": 1.1773,
+      "step": 146030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.057593489040499e-06,
+      "loss": 1.62,
+      "step": 146035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.056985721376651e-06,
+      "loss": 1.1408,
+      "step": 146040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.056377987650612e-06,
+      "loss": 1.2712,
+      "step": 146045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.055770287865854e-06,
+      "loss": 1.1009,
+      "step": 146050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.05516262202584e-06,
+      "loss": 1.2118,
+      "step": 146055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.054554990134049e-06,
+      "loss": 0.9083,
+      "step": 146060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.053947392193949e-06,
+      "loss": 1.0196,
+      "step": 146065
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.053339828208998e-06,
+      "loss": 1.1377,
+      "step": 146070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.052732298182683e-06,
+      "loss": 1.2137,
+      "step": 146075
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.052124802118466e-06,
+      "loss": 1.0982,
+      "step": 146080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.051517340019814e-06,
+      "loss": 1.0007,
+      "step": 146085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.050909911890194e-06,
+      "loss": 1.4405,
+      "step": 146090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.050302517733079e-06,
+      "loss": 1.2488,
+      "step": 146095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.049695157551941e-06,
+      "loss": 1.0735,
+      "step": 146100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.049087831350238e-06,
+      "loss": 1.2642,
+      "step": 146105
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0484805391314495e-06,
+      "loss": 1.2561,
+      "step": 146110
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.047873280899034e-06,
+      "loss": 1.5873,
+      "step": 146115
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.047266056656464e-06,
+      "loss": 1.1515,
+      "step": 146120
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.04665886640721e-06,
+      "loss": 1.3327,
+      "step": 146125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.046051710154736e-06,
+      "loss": 1.006,
+      "step": 146130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0454445879025055e-06,
+      "loss": 1.3053,
+      "step": 146135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.044837499653988e-06,
+      "loss": 1.3213,
+      "step": 146140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.044230445412655e-06,
+      "loss": 1.0091,
+      "step": 146145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0436234251819675e-06,
+      "loss": 1.1265,
+      "step": 146150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.043016438965396e-06,
+      "loss": 1.754,
+      "step": 146155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.042409486766402e-06,
+      "loss": 1.6484,
+      "step": 146160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.041802568588457e-06,
+      "loss": 1.359,
+      "step": 146165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.041195684435021e-06,
+      "loss": 1.476,
+      "step": 146170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.040588834309566e-06,
+      "loss": 1.0535,
+      "step": 146175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.039982018215554e-06,
+      "loss": 1.2136,
+      "step": 146180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.039375236156446e-06,
+      "loss": 1.1587,
+      "step": 146185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.038768488135712e-06,
+      "loss": 1.1255,
+      "step": 146190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.038161774156819e-06,
+      "loss": 1.1931,
+      "step": 146195
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.037555094223226e-06,
+      "loss": 1.1221,
+      "step": 146200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.036948448338405e-06,
+      "loss": 1.2098,
+      "step": 146205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.03634183650581e-06,
+      "loss": 0.9767,
+      "step": 146210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0357352587289125e-06,
+      "loss": 1.3346,
+      "step": 146215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.035128715011176e-06,
+      "loss": 1.1359,
+      "step": 146220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.034522205356065e-06,
+      "loss": 1.3892,
+      "step": 146225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.033915729767036e-06,
+      "loss": 1.0935,
+      "step": 146230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.033309288247557e-06,
+      "loss": 1.2574,
+      "step": 146235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.032702880801095e-06,
+      "loss": 1.5239,
+      "step": 146240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.032096507431109e-06,
+      "loss": 0.973,
+      "step": 146245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.031490168141058e-06,
+      "loss": 1.5291,
+      "step": 146250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.030883862934409e-06,
+      "loss": 1.2712,
+      "step": 146255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.030277591814626e-06,
+      "loss": 1.2624,
+      "step": 146260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.029671354785166e-06,
+      "loss": 1.3415,
+      "step": 146265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.029065151849498e-06,
+      "loss": 1.4242,
+      "step": 146270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.028458983011079e-06,
+      "loss": 1.1541,
+      "step": 146275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.027852848273364e-06,
+      "loss": 1.2198,
+      "step": 146280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.027246747639828e-06,
+      "loss": 1.0857,
+      "step": 146285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0266406811139244e-06,
+      "loss": 1.064,
+      "step": 146290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.026034648699114e-06,
+      "loss": 1.2176,
+      "step": 146295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.025428650398862e-06,
+      "loss": 1.0825,
+      "step": 146300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0248226862166205e-06,
+      "loss": 1.2009,
+      "step": 146305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.024216756155861e-06,
+      "loss": 1.2335,
+      "step": 146310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.023610860220033e-06,
+      "loss": 1.2505,
+      "step": 146315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.023004998412605e-06,
+      "loss": 1.2295,
+      "step": 146320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0223991707370304e-06,
+      "loss": 1.371,
+      "step": 146325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.021793377196771e-06,
+      "loss": 1.1046,
+      "step": 146330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.02118761779529e-06,
+      "loss": 1.5232,
+      "step": 146335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.020581892536044e-06,
+      "loss": 1.3045,
+      "step": 146340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.019976201422488e-06,
+      "loss": 1.2775,
+      "step": 146345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.019370544458085e-06,
+      "loss": 1.204,
+      "step": 146350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.018764921646294e-06,
+      "loss": 1.2436,
+      "step": 146355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.018159332990571e-06,
+      "loss": 1.3658,
+      "step": 146360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0175537784943785e-06,
+      "loss": 1.1813,
+      "step": 146365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.016948258161172e-06,
+      "loss": 1.2292,
+      "step": 146370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.016342771994403e-06,
+      "loss": 1.0407,
+      "step": 146375
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.015737319997538e-06,
+      "loss": 1.015,
+      "step": 146380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0151319021740346e-06,
+      "loss": 1.1652,
+      "step": 146385
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.014526518527347e-06,
+      "loss": 1.2217,
+      "step": 146390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.013921169060927e-06,
+      "loss": 0.9496,
+      "step": 146395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.013315853778238e-06,
+      "loss": 1.416,
+      "step": 146400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.012710572682739e-06,
+      "loss": 1.1986,
+      "step": 146405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.012105325777879e-06,
+      "loss": 1.2417,
+      "step": 146410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.011500113067123e-06,
+      "loss": 1.3019,
+      "step": 146415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.010894934553917e-06,
+      "loss": 1.5643,
+      "step": 146420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.010289790241724e-06,
+      "loss": 1.1465,
+      "step": 146425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.009684680133999e-06,
+      "loss": 0.9968,
+      "step": 146430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.009079604234198e-06,
+      "loss": 1.1665,
+      "step": 146435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.008474562545771e-06,
+      "loss": 1.3485,
+      "step": 146440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0078695550721765e-06,
+      "loss": 1.0455,
+      "step": 146445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.007264581816874e-06,
+      "loss": 1.5201,
+      "step": 146450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.006659642783314e-06,
+      "loss": 1.3016,
+      "step": 146455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.006054737974946e-06,
+      "loss": 1.6046,
+      "step": 146460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.00544986739523e-06,
+      "loss": 1.0941,
+      "step": 146465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.004845031047624e-06,
+      "loss": 1.31,
+      "step": 146470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.004240228935573e-06,
+      "loss": 1.1885,
+      "step": 146475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.003635461062538e-06,
+      "loss": 0.9708,
+      "step": 146480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0030307274319705e-06,
+      "loss": 1.4286,
+      "step": 146485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.0024260280473195e-06,
+      "loss": 1.0275,
+      "step": 146490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.001821362912042e-06,
+      "loss": 1.064,
+      "step": 146495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.001216732029594e-06,
+      "loss": 1.1332,
+      "step": 146500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.00061213540342e-06,
+      "loss": 1.2064,
+      "step": 146505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.000007573036982e-06,
+      "loss": 1.2858,
+      "step": 146510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.999403044933725e-06,
+      "loss": 1.1222,
+      "step": 146515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.998798551097107e-06,
+      "loss": 0.9791,
+      "step": 146520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.998194091530573e-06,
+      "loss": 1.2627,
+      "step": 146525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.9975896662375835e-06,
+      "loss": 1.1635,
+      "step": 146530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.996985275221581e-06,
+      "loss": 1.2687,
+      "step": 146535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.9963809184860216e-06,
+      "loss": 1.2223,
+      "step": 146540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.99577659603436e-06,
+      "loss": 1.3836,
+      "step": 146545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9951723078700425e-06,
+      "loss": 1.4675,
+      "step": 146550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994568053996518e-06,
+      "loss": 1.0794,
+      "step": 146555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.99396383441724e-06,
+      "loss": 1.2029,
+      "step": 146560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.993359649135662e-06,
+      "loss": 1.1504,
+      "step": 146565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.992755498155228e-06,
+      "loss": 1.2196,
+      "step": 146570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.992151381479395e-06,
+      "loss": 1.0237,
+      "step": 146575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.991547299111608e-06,
+      "loss": 1.0942,
+      "step": 146580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990943251055315e-06,
+      "loss": 1.185,
+      "step": 146585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990339237313967e-06,
+      "loss": 1.0327,
+      "step": 146590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989735257891019e-06,
+      "loss": 0.9476,
+      "step": 146595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989131312789913e-06,
+      "loss": 1.1564,
+      "step": 146600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9885274020141e-06,
+      "loss": 1.0109,
+      "step": 146605
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.987923525567026e-06,
+      "loss": 1.1566,
+      "step": 146610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9873196834521465e-06,
+      "loss": 0.9304,
+      "step": 146615
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.986715875672903e-06,
+      "loss": 1.0662,
+      "step": 146620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9861121022327485e-06,
+      "loss": 1.2346,
+      "step": 146625
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.985508363135126e-06,
+      "loss": 1.4067,
+      "step": 146630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984904658383485e-06,
+      "loss": 1.2095,
+      "step": 146635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984300987981278e-06,
+      "loss": 1.9489,
+      "step": 146640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983697351931947e-06,
+      "loss": 1.0846,
+      "step": 146645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983093750238937e-06,
+      "loss": 1.0958,
+      "step": 146650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.982490182905697e-06,
+      "loss": 1.1387,
+      "step": 146655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.981886649935678e-06,
+      "loss": 1.2802,
+      "step": 146660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.981283151332325e-06,
+      "loss": 1.2123,
+      "step": 146665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.980679687099077e-06,
+      "loss": 1.1533,
+      "step": 146670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.980076257239389e-06,
+      "loss": 1.1565,
+      "step": 146675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9794728617567014e-06,
+      "loss": 1.2245,
+      "step": 146680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978869500654461e-06,
+      "loss": 1.0888,
+      "step": 146685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978266173936118e-06,
+      "loss": 1.1407,
+      "step": 146690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.977662881605113e-06,
+      "loss": 1.1513,
+      "step": 146695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.97705962366489e-06,
+      "loss": 1.1451,
+      "step": 146700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.976456400118894e-06,
+      "loss": 1.0652,
+      "step": 146705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975853210970578e-06,
+      "loss": 1.3456,
+      "step": 146710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975250056223374e-06,
+      "loss": 0.7897,
+      "step": 146715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.974646935880739e-06,
+      "loss": 1.138,
+      "step": 146720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.974043849946104e-06,
+      "loss": 1.1589,
+      "step": 146725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.973440798422922e-06,
+      "loss": 1.0791,
+      "step": 146730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.972837781314637e-06,
+      "loss": 1.0941,
+      "step": 146735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.97223479862469e-06,
+      "loss": 1.2029,
+      "step": 146740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9716318503565225e-06,
+      "loss": 1.3316,
+      "step": 146745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.971028936513577e-06,
+      "loss": 1.2665,
+      "step": 146750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.970426057099304e-06,
+      "loss": 1.075,
+      "step": 146755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.969823212117141e-06,
+      "loss": 1.0559,
+      "step": 146760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.969220401570527e-06,
+      "loss": 0.9965,
+      "step": 146765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9686176254629114e-06,
+      "loss": 1.0246,
+      "step": 146770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.96801488379773e-06,
+      "loss": 1.1405,
+      "step": 146775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.967412176578428e-06,
+      "loss": 0.9879,
+      "step": 146780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.966809503808451e-06,
+      "loss": 1.0615,
+      "step": 146785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.966206865491236e-06,
+      "loss": 1.1666,
+      "step": 146790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.965604261630221e-06,
+      "loss": 1.0213,
+      "step": 146795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.965001692228854e-06,
+      "loss": 1.198,
+      "step": 146800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.964399157290574e-06,
+      "loss": 1.4378,
+      "step": 146805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.963796656818822e-06,
+      "loss": 1.1171,
+      "step": 146810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.963194190817034e-06,
+      "loss": 1.1583,
+      "step": 146815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.962591759288654e-06,
+      "loss": 1.1845,
+      "step": 146820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9619893622371265e-06,
+      "loss": 1.0586,
+      "step": 146825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.961386999665884e-06,
+      "loss": 1.3112,
+      "step": 146830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.960784671578371e-06,
+      "loss": 1.5651,
+      "step": 146835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.960182377978025e-06,
+      "loss": 1.0919,
+      "step": 146840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.959580118868284e-06,
+      "loss": 1.3511,
+      "step": 146845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.958977894252595e-06,
+      "loss": 1.0934,
+      "step": 146850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9583757041343894e-06,
+      "loss": 0.9406,
+      "step": 146855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.957773548517106e-06,
+      "loss": 1.2682,
+      "step": 146860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.957171427404184e-06,
+      "loss": 1.2591,
+      "step": 146865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.956569340799068e-06,
+      "loss": 1.4497,
+      "step": 146870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.955967288705187e-06,
+      "loss": 1.2625,
+      "step": 146875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.955365271125987e-06,
+      "loss": 1.1402,
+      "step": 146880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9547632880649035e-06,
+      "loss": 1.0918,
+      "step": 146885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9541613395253695e-06,
+      "loss": 1.2215,
+      "step": 146890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9535594255108255e-06,
+      "loss": 1.1802,
+      "step": 146895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.952957546024713e-06,
+      "loss": 1.2785,
+      "step": 146900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.952355701070465e-06,
+      "loss": 1.2654,
+      "step": 146905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.951753890651515e-06,
+      "loss": 0.846,
+      "step": 146910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.951152114771304e-06,
+      "loss": 0.9912,
+      "step": 146915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.950550373433272e-06,
+      "loss": 1.091,
+      "step": 146920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.949948666640846e-06,
+      "loss": 1.3059,
+      "step": 146925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9493469943974715e-06,
+      "loss": 1.0639,
+      "step": 146930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.948745356706576e-06,
+      "loss": 1.1537,
+      "step": 146935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.948143753571599e-06,
+      "loss": 1.0687,
+      "step": 146940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.947542184995981e-06,
+      "loss": 1.5153,
+      "step": 146945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.946940650983152e-06,
+      "loss": 1.1497,
+      "step": 146950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.946339151536547e-06,
+      "loss": 1.4376,
+      "step": 146955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.945737686659595e-06,
+      "loss": 1.5576,
+      "step": 146960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.945136256355743e-06,
+      "loss": 1.2817,
+      "step": 146965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9445348606284204e-06,
+      "loss": 1.2109,
+      "step": 146970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.943933499481057e-06,
+      "loss": 1.0505,
+      "step": 146975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.943332172917095e-06,
+      "loss": 1.2172,
+      "step": 146980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.942730880939959e-06,
+      "loss": 1.1564,
+      "step": 146985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.942129623553088e-06,
+      "loss": 1.1591,
+      "step": 146990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.941528400759919e-06,
+      "loss": 1.1486,
+      "step": 146995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.940927212563881e-06,
+      "loss": 1.2176,
+      "step": 147000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.940326058968405e-06,
+      "loss": 1.0578,
+      "step": 147005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.939724939976924e-06,
+      "loss": 1.0732,
+      "step": 147010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.939123855592879e-06,
+      "loss": 1.2083,
+      "step": 147015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9385228058196925e-06,
+      "loss": 1.1326,
+      "step": 147020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.937921790660804e-06,
+      "loss": 1.0807,
+      "step": 147025
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9373208101196395e-06,
+      "loss": 1.2314,
+      "step": 147030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.936719864199637e-06,
+      "loss": 1.3648,
+      "step": 147035
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.936118952904222e-06,
+      "loss": 1.1204,
+      "step": 147040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.935518076236833e-06,
+      "loss": 1.1242,
+      "step": 147045
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.934917234200895e-06,
+      "loss": 0.976,
+      "step": 147050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.93431642679984e-06,
+      "loss": 1.2987,
+      "step": 147055
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.933715654037105e-06,
+      "loss": 1.5795,
+      "step": 147060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.933114915916115e-06,
+      "loss": 1.3174,
+      "step": 147065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.932514212440301e-06,
+      "loss": 1.0419,
+      "step": 147070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.931913543613096e-06,
+      "loss": 1.2636,
+      "step": 147075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9313129094379254e-06,
+      "loss": 1.1163,
+      "step": 147080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.930712309918222e-06,
+      "loss": 1.1406,
+      "step": 147085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.930111745057419e-06,
+      "loss": 1.3407,
+      "step": 147090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.929511214858942e-06,
+      "loss": 1.5328,
+      "step": 147095
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.928910719326219e-06,
+      "loss": 1.0629,
+      "step": 147100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.92831025846268e-06,
+      "loss": 1.1177,
+      "step": 147105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9277098322717584e-06,
+      "loss": 1.1219,
+      "step": 147110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.927109440756881e-06,
+      "loss": 1.1857,
+      "step": 147115
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9265090839214715e-06,
+      "loss": 1.4978,
+      "step": 147120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.925908761768961e-06,
+      "loss": 1.1678,
+      "step": 147125
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.925308474302782e-06,
+      "loss": 1.144,
+      "step": 147130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9247082215263555e-06,
+      "loss": 1.5444,
+      "step": 147135
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.924108003443116e-06,
+      "loss": 1.1305,
+      "step": 147140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.923507820056483e-06,
+      "loss": 1.0625,
+      "step": 147145
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.922907671369891e-06,
+      "loss": 1.2521,
+      "step": 147150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.922307557386766e-06,
+      "loss": 1.26,
+      "step": 147155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9217074781105356e-06,
+      "loss": 1.2538,
+      "step": 147160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.921107433544621e-06,
+      "loss": 1.1059,
+      "step": 147165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.920507423692456e-06,
+      "loss": 1.0622,
+      "step": 147170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.919907448557458e-06,
+      "loss": 1.1086,
+      "step": 147175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.919307508143064e-06,
+      "loss": 1.1541,
+      "step": 147180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.918707602452691e-06,
+      "loss": 1.5491,
+      "step": 147185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.918107731489772e-06,
+      "loss": 1.2791,
+      "step": 147190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.917507895257725e-06,
+      "loss": 0.9541,
+      "step": 147195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.91690809375998e-06,
+      "loss": 1.1529,
+      "step": 147200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.916308326999965e-06,
+      "loss": 1.1002,
+      "step": 147205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.915708594981101e-06,
+      "loss": 1.3263,
+      "step": 147210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9151088977068104e-06,
+      "loss": 1.0813,
+      "step": 147215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.914509235180521e-06,
+      "loss": 1.4591,
+      "step": 147220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.913909607405659e-06,
+      "loss": 1.4723,
+      "step": 147225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.913310014385645e-06,
+      "loss": 1.1271,
+      "step": 147230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.912710456123907e-06,
+      "loss": 1.2373,
+      "step": 147235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.912110932623862e-06,
+      "loss": 1.1771,
+      "step": 147240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.911511443888943e-06,
+      "loss": 1.5021,
+      "step": 147245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.910911989922565e-06,
+      "loss": 1.4174,
+      "step": 147250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.910312570728157e-06,
+      "loss": 1.0621,
+      "step": 147255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.90971318630914e-06,
+      "loss": 1.0888,
+      "step": 147260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.909113836668929e-06,
+      "loss": 1.2224,
+      "step": 147265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9085145218109625e-06,
+      "loss": 1.0148,
+      "step": 147270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9079152417386526e-06,
+      "loss": 1.4403,
+      "step": 147275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.907315996455421e-06,
+      "loss": 0.888,
+      "step": 147280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.906716785964695e-06,
+      "loss": 1.0631,
+      "step": 147285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.906117610269889e-06,
+      "loss": 1.2378,
+      "step": 147290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.905518469374431e-06,
+      "loss": 1.2399,
+      "step": 147295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.904919363281743e-06,
+      "loss": 0.9665,
+      "step": 147300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.904320291995243e-06,
+      "loss": 1.1442,
+      "step": 147305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.90372125551835e-06,
+      "loss": 1.1709,
+      "step": 147310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.903122253854488e-06,
+      "loss": 1.1239,
+      "step": 147315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.902523287007079e-06,
+      "loss": 1.2044,
+      "step": 147320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.901924354979543e-06,
+      "loss": 1.1358,
+      "step": 147325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.901325457775295e-06,
+      "loss": 1.3584,
+      "step": 147330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.900726595397758e-06,
+      "loss": 1.3054,
+      "step": 147335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.900127767850358e-06,
+      "loss": 1.1471,
+      "step": 147340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.899528975136504e-06,
+      "loss": 0.9739,
+      "step": 147345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.898930217259625e-06,
+      "loss": 1.1789,
+      "step": 147350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.898331494223135e-06,
+      "loss": 1.271,
+      "step": 147355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.897732806030449e-06,
+      "loss": 1.0617,
+      "step": 147360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.897134152684996e-06,
+      "loss": 1.0389,
+      "step": 147365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.89653553419019e-06,
+      "loss": 1.1548,
+      "step": 147370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.895936950549445e-06,
+      "loss": 1.2058,
+      "step": 147375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8953384017661875e-06,
+      "loss": 1.353,
+      "step": 147380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.894739887843827e-06,
+      "loss": 1.2274,
+      "step": 147385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.894141408785789e-06,
+      "loss": 1.0888,
+      "step": 147390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.893542964595483e-06,
+      "loss": 1.1244,
+      "step": 147395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8929445552763364e-06,
+      "loss": 1.1141,
+      "step": 147400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.892346180831758e-06,
+      "loss": 1.4844,
+      "step": 147405
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.891747841265165e-06,
+      "loss": 1.0383,
+      "step": 147410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.891149536579984e-06,
+      "loss": 1.3181,
+      "step": 147415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.890551266779623e-06,
+      "loss": 1.3945,
+      "step": 147420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.889953031867496e-06,
+      "loss": 1.0441,
+      "step": 147425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.889354831847025e-06,
+      "loss": 1.5457,
+      "step": 147430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.888756666721628e-06,
+      "loss": 1.1421,
+      "step": 147435
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.888158536494713e-06,
+      "loss": 0.9965,
+      "step": 147440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8875604411697046e-06,
+      "loss": 1.09,
+      "step": 147445
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.886962380750009e-06,
+      "loss": 1.0901,
+      "step": 147450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.886364355239046e-06,
+      "loss": 1.012,
+      "step": 147455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8857663646402355e-06,
+      "loss": 1.1936,
+      "step": 147460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8851684089569866e-06,
+      "loss": 1.1703,
+      "step": 147465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.884570488192716e-06,
+      "loss": 1.3005,
+      "step": 147470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8839726023508325e-06,
+      "loss": 1.0897,
+      "step": 147475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.883374751434755e-06,
+      "loss": 1.3726,
+      "step": 147480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.882776935447902e-06,
+      "loss": 1.0445,
+      "step": 147485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.882179154393679e-06,
+      "loss": 1.3374,
+      "step": 147490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8815814082755065e-06,
+      "loss": 1.392,
+      "step": 147495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8809836970967915e-06,
+      "loss": 1.0697,
+      "step": 147500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.880386020860951e-06,
+      "loss": 0.983,
+      "step": 147505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.879788379571402e-06,
+      "loss": 1.1733,
+      "step": 147510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.879190773231551e-06,
+      "loss": 1.0631,
+      "step": 147515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.878593201844811e-06,
+      "loss": 1.33,
+      "step": 147520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.877995665414597e-06,
+      "loss": 1.1852,
+      "step": 147525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.877398163944323e-06,
+      "loss": 1.1504,
+      "step": 147530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.876800697437397e-06,
+      "loss": 1.2425,
+      "step": 147535
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.876203265897231e-06,
+      "loss": 1.581,
+      "step": 147540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.875605869327238e-06,
+      "loss": 1.1333,
+      "step": 147545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.875008507730832e-06,
+      "loss": 1.1015,
+      "step": 147550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.874411181111418e-06,
+      "loss": 1.2068,
+      "step": 147555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.873813889472416e-06,
+      "loss": 1.5689,
+      "step": 147560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.873216632817231e-06,
+      "loss": 1.0519,
+      "step": 147565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.872619411149265e-06,
+      "loss": 1.1241,
+      "step": 147570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.872022224471948e-06,
+      "loss": 1.0763,
+      "step": 147575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8714250727886785e-06,
+      "loss": 1.3144,
+      "step": 147580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.870827956102865e-06,
+      "loss": 1.0047,
+      "step": 147585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.870230874417924e-06,
+      "loss": 1.0555,
+      "step": 147590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.869633827737258e-06,
+      "loss": 1.2294,
+      "step": 147595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.86903681606428e-06,
+      "loss": 1.1172,
+      "step": 147600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.868439839402403e-06,
+      "loss": 1.1411,
+      "step": 147605
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.867842897755032e-06,
+      "loss": 1.6025,
+      "step": 147610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.867245991125573e-06,
+      "loss": 1.1277,
+      "step": 147615
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.866649119517438e-06,
+      "loss": 1.3212,
+      "step": 147620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.866052282934039e-06,
+      "loss": 1.3076,
+      "step": 147625
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.865455481378781e-06,
+      "loss": 1.5508,
+      "step": 147630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.864858714855066e-06,
+      "loss": 1.0004,
+      "step": 147635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.86426198336631e-06,
+      "loss": 1.2517,
+      "step": 147640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.863665286915921e-06,
+      "loss": 1.0159,
+      "step": 147645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.863068625507299e-06,
+      "loss": 1.1014,
+      "step": 147650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8624719991438605e-06,
+      "loss": 1.1345,
+      "step": 147655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.861875407829008e-06,
+      "loss": 1.2472,
+      "step": 147660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.861278851566142e-06,
+      "loss": 1.2471,
+      "step": 147665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8606823303586806e-06,
+      "loss": 0.9805,
+      "step": 147670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8600858442100265e-06,
+      "loss": 0.9167,
+      "step": 147675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.859489393123581e-06,
+      "loss": 1.028,
+      "step": 147680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.858892977102758e-06,
+      "loss": 1.1519,
+      "step": 147685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.858296596150953e-06,
+      "loss": 1.1221,
+      "step": 147690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8577002502715835e-06,
+      "loss": 1.1162,
+      "step": 147695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.857103939468046e-06,
+      "loss": 1.15,
+      "step": 147700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.856507663743751e-06,
+      "loss": 1.2128,
+      "step": 147705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.855911423102099e-06,
+      "loss": 1.3151,
+      "step": 147710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.855315217546497e-06,
+      "loss": 1.3034,
+      "step": 147715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.854719047080353e-06,
+      "loss": 1.0585,
+      "step": 147720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.854122911707069e-06,
+      "loss": 0.941,
+      "step": 147725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.853526811430045e-06,
+      "loss": 1.1329,
+      "step": 147730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.85293074625269e-06,
+      "loss": 0.9655,
+      "step": 147735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8523347161784085e-06,
+      "loss": 1.0916,
+      "step": 147740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8517387212105996e-06,
+      "loss": 1.2755,
+      "step": 147745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.851142761352673e-06,
+      "loss": 1.2352,
+      "step": 147750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.850546836608029e-06,
+      "loss": 1.2328,
+      "step": 147755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8499509469800655e-06,
+      "loss": 1.2836,
+      "step": 147760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8493550924721905e-06,
+      "loss": 1.1804,
+      "step": 147765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.848759273087809e-06,
+      "loss": 1.5862,
+      "step": 147770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.848163488830321e-06,
+      "loss": 1.3279,
+      "step": 147775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.847567739703124e-06,
+      "loss": 1.1859,
+      "step": 147780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.846972025709626e-06,
+      "loss": 1.1002,
+      "step": 147785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.846376346853229e-06,
+      "loss": 1.0943,
+      "step": 147790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.845780703137331e-06,
+      "loss": 1.3996,
+      "step": 147795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8451850945653364e-06,
+      "loss": 1.2205,
+      "step": 147800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.844589521140644e-06,
+      "loss": 1.2672,
+      "step": 147805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.843993982866655e-06,
+      "loss": 0.8922,
+      "step": 147810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.843398479746775e-06,
+      "loss": 1.3058,
+      "step": 147815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.842803011784401e-06,
+      "loss": 1.3617,
+      "step": 147820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.84220757898293e-06,
+      "loss": 0.97,
+      "step": 147825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8416121813457654e-06,
+      "loss": 1.0099,
+      "step": 147830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.841016818876312e-06,
+      "loss": 1.3489,
+      "step": 147835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8404214915779634e-06,
+      "loss": 1.1754,
+      "step": 147840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.83982619945412e-06,
+      "loss": 1.1515,
+      "step": 147845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.839230942508181e-06,
+      "loss": 1.028,
+      "step": 147850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.83863572074355e-06,
+      "loss": 1.1535,
+      "step": 147855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.83804053416362e-06,
+      "loss": 1.1526,
+      "step": 147860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8374453827717965e-06,
+      "loss": 1.0088,
+      "step": 147865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.836850266571474e-06,
+      "loss": 0.9703,
+      "step": 147870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.836255185566047e-06,
+      "loss": 1.5092,
+      "step": 147875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8356601397589185e-06,
+      "loss": 1.1285,
+      "step": 147880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.83506512915349e-06,
+      "loss": 1.4007,
+      "step": 147885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.834470153753153e-06,
+      "loss": 1.3574,
+      "step": 147890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.83387521356131e-06,
+      "loss": 1.3228,
+      "step": 147895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.833280308581352e-06,
+      "loss": 1.1316,
+      "step": 147900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.832685438816684e-06,
+      "loss": 1.0629,
+      "step": 147905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.832090604270696e-06,
+      "loss": 1.1843,
+      "step": 147910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.831495804946792e-06,
+      "loss": 1.3372,
+      "step": 147915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.830901040848361e-06,
+      "loss": 1.4458,
+      "step": 147920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.830306311978802e-06,
+      "loss": 1.4694,
+      "step": 147925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.829711618341517e-06,
+      "loss": 1.0763,
+      "step": 147930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.829116959939897e-06,
+      "loss": 1.344,
+      "step": 147935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.828522336777335e-06,
+      "loss": 1.3842,
+      "step": 147940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.82792774885723e-06,
+      "loss": 0.982,
+      "step": 147945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.82733319618298e-06,
+      "loss": 1.1212,
+      "step": 147950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826738678757974e-06,
+      "loss": 1.3568,
+      "step": 147955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8261441965856135e-06,
+      "loss": 1.305,
+      "step": 147960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8255497496692915e-06,
+      "loss": 1.1071,
+      "step": 147965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824955338012394e-06,
+      "loss": 0.9564,
+      "step": 147970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824360961618329e-06,
+      "loss": 1.1648,
+      "step": 147975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823766620490486e-06,
+      "loss": 1.2559,
+      "step": 147980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823172314632256e-06,
+      "loss": 1.3665,
+      "step": 147985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8225780440470315e-06,
+      "loss": 1.8633,
+      "step": 147990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821983808738209e-06,
+      "loss": 1.1705,
+      "step": 147995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821389608709185e-06,
+      "loss": 1.4088,
+      "step": 148000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820795443963346e-06,
+      "loss": 1.3807,
+      "step": 148005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820201314504092e-06,
+      "loss": 1.1642,
+      "step": 148010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.819607220334809e-06,
+      "loss": 1.275,
+      "step": 148015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.819013161458894e-06,
+      "loss": 1.1691,
+      "step": 148020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8184191378797394e-06,
+      "loss": 1.1426,
+      "step": 148025
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8178251496007375e-06,
+      "loss": 1.0645,
+      "step": 148030
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817231196625276e-06,
+      "loss": 1.0592,
+      "step": 148035
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.81663727895675e-06,
+      "loss": 1.0295,
+      "step": 148040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.816043396598553e-06,
+      "loss": 0.9341,
+      "step": 148045
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.815449549554076e-06,
+      "loss": 1.2447,
+      "step": 148050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814855737826704e-06,
+      "loss": 1.0661,
+      "step": 148055
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814261961419835e-06,
+      "loss": 1.2808,
+      "step": 148060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8136682203368548e-06,
+      "loss": 0.7841,
+      "step": 148065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8130745145811566e-06,
+      "loss": 1.1938,
+      "step": 148070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8124808441561335e-06,
+      "loss": 1.1151,
+      "step": 148075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.811887209065173e-06,
+      "loss": 1.101,
+      "step": 148080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8112936093116614e-06,
+      "loss": 0.987,
+      "step": 148085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8107000448989916e-06,
+      "loss": 1.0597,
+      "step": 148090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.810106515830557e-06,
+      "loss": 1.1342,
+      "step": 148095
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.80951302210974e-06,
+      "loss": 1.3138,
+      "step": 148100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8089195637399366e-06,
+      "loss": 1.1892,
+      "step": 148105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.80832614072453e-06,
+      "loss": 1.1251,
+      "step": 148110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8077327530669106e-06,
+      "loss": 0.9188,
+      "step": 148115
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.807139400770471e-06,
+      "loss": 1.28,
+      "step": 148120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.806546083838596e-06,
+      "loss": 1.2828,
+      "step": 148125
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8059528022746705e-06,
+      "loss": 1.2783,
+      "step": 148130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8053595560820867e-06,
+      "loss": 1.3485,
+      "step": 148135
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8047663452642347e-06,
+      "loss": 1.2963,
+      "step": 148140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.804173169824499e-06,
+      "loss": 1.028,
+      "step": 148145
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.803580029766264e-06,
+      "loss": 0.981,
+      "step": 148150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8029869250929184e-06,
+      "loss": 1.1931,
+      "step": 148155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8023938558078556e-06,
+      "loss": 1.3894,
+      "step": 148160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.801800821914453e-06,
+      "loss": 1.0644,
+      "step": 148165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8012078234161043e-06,
+      "loss": 1.3288,
+      "step": 148170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8006148603161942e-06,
+      "loss": 0.8706,
+      "step": 148175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8000219326181032e-06,
+      "loss": 1.375,
+      "step": 148180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.799429040325222e-06,
+      "loss": 1.2807,
+      "step": 148185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.798836183440939e-06,
+      "loss": 1.1949,
+      "step": 148190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7982433619686376e-06,
+      "loss": 1.2929,
+      "step": 148195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.797650575911699e-06,
+      "loss": 1.3932,
+      "step": 148200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.79705782527351e-06,
+      "loss": 1.471,
+      "step": 148205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.796465110057462e-06,
+      "loss": 1.2617,
+      "step": 148210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7958724302669315e-06,
+      "loss": 1.6866,
+      "step": 148215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.795279785905308e-06,
+      "loss": 1.5033,
+      "step": 148220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.794687176975972e-06,
+      "loss": 1.1601,
+      "step": 148225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.79409460348231e-06,
+      "loss": 1.3832,
+      "step": 148230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.793502065427709e-06,
+      "loss": 1.1815,
+      "step": 148235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7929095628155486e-06,
+      "loss": 1.3247,
+      "step": 148240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7923170956492096e-06,
+      "loss": 1.3804,
+      "step": 148245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.791724663932079e-06,
+      "loss": 1.1624,
+      "step": 148250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7911322676675444e-06,
+      "loss": 1.2515,
+      "step": 148255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7905399068589787e-06,
+      "loss": 1.1808,
+      "step": 148260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7899475815097754e-06,
+      "loss": 1.4548,
+      "step": 148265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.78935529162331e-06,
+      "loss": 1.2183,
+      "step": 148270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7887630372029628e-06,
+      "loss": 1.458,
+      "step": 148275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7881708182521204e-06,
+      "loss": 1.317,
+      "step": 148280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7875786347741663e-06,
+      "loss": 1.266,
+      "step": 148285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7869864867724795e-06,
+      "loss": 1.282,
+      "step": 148290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7863943742504384e-06,
+      "loss": 1.5832,
+      "step": 148295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.785802297211427e-06,
+      "loss": 1.0813,
+      "step": 148300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.785210255658831e-06,
+      "loss": 1.1821,
+      "step": 148305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.784618249596024e-06,
+      "loss": 1.0946,
+      "step": 148310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7840262790263917e-06,
+      "loss": 1.6195,
+      "step": 148315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7834343439533096e-06,
+      "loss": 1.232,
+      "step": 148320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7828424443801613e-06,
+      "loss": 1.3964,
+      "step": 148325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7822505803103305e-06,
+      "loss": 1.0836,
+      "step": 148330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7816587517471925e-06,
+      "loss": 1.3114,
+      "step": 148335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7810669586941273e-06,
+      "loss": 1.3339,
+      "step": 148340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7804752011545088e-06,
+      "loss": 1.0081,
+      "step": 148345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7798834791317284e-06,
+      "loss": 1.1607,
+      "step": 148350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7792917926291585e-06,
+      "loss": 1.0225,
+      "step": 148355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7787001416501756e-06,
+      "loss": 1.2414,
+      "step": 148360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.778108526198163e-06,
+      "loss": 1.146,
+      "step": 148365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.777516946276495e-06,
+      "loss": 1.3346,
+      "step": 148370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7769254018885527e-06,
+      "loss": 1.132,
+      "step": 148375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7763338930377156e-06,
+      "loss": 1.5454,
+      "step": 148380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7757424197273594e-06,
+      "loss": 1.1865,
+      "step": 148385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7751509819608588e-06,
+      "loss": 1.3243,
+      "step": 148390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7745595797415933e-06,
+      "loss": 1.2247,
+      "step": 148395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.773968213072946e-06,
+      "loss": 1.2014,
+      "step": 148400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.773376881958284e-06,
+      "loss": 1.421,
+      "step": 148405
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.772785586400993e-06,
+      "loss": 0.9085,
+      "step": 148410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7721943264044426e-06,
+      "loss": 0.9595,
+      "step": 148415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7716031019720145e-06,
+      "loss": 1.3851,
+      "step": 148420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.77101191310708e-06,
+      "loss": 1.4743,
+      "step": 148425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7704207598130216e-06,
+      "loss": 0.9388,
+      "step": 148430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7698296420932067e-06,
+      "loss": 1.0862,
+      "step": 148435
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.769238559951016e-06,
+      "loss": 1.1411,
+      "step": 148440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.768647513389828e-06,
+      "loss": 1.4177,
+      "step": 148445
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7680565024130145e-06,
+      "loss": 1.3132,
+      "step": 148450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7674655270239455e-06,
+      "loss": 1.1633,
+      "step": 148455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.766874587226006e-06,
+      "loss": 1.1368,
+      "step": 148460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7662836830225603e-06,
+      "loss": 1.393,
+      "step": 148465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.765692814416988e-06,
+      "loss": 1.1094,
+      "step": 148470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 1.1865,
+      "step": 148475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7645111840129643e-06,
+      "loss": 1.4755,
+      "step": 148480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7639204222212554e-06,
+      "loss": 1.1738,
+      "step": 148485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.763329696040915e-06,
+      "loss": 0.9753,
+      "step": 148490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.76273900547532e-06,
+      "loss": 1.2175,
+      "step": 148495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.762148350527839e-06,
+      "loss": 1.1158,
+      "step": 148500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7615577312018427e-06,
+      "loss": 1.1659,
+      "step": 148505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7609671475007083e-06,
+      "loss": 1.249,
+      "step": 148510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7603765994278096e-06,
+      "loss": 1.3071,
+      "step": 148515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.759786086986513e-06,
+      "loss": 1.3568,
+      "step": 148520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7591956101801973e-06,
+      "loss": 1.0464,
+      "step": 148525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.758605169012228e-06,
+      "loss": 1.3322,
+      "step": 148530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.758014763485981e-06,
+      "loss": 1.0583,
+      "step": 148535
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7574243936048303e-06,
+      "loss": 1.2683,
+      "step": 148540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7568340593721432e-06,
+      "loss": 1.2531,
+      "step": 148545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.756243760791288e-06,
+      "loss": 1.4001,
+      "step": 148550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7556534978656402e-06,
+      "loss": 1.7146,
+      "step": 148555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.755063270598572e-06,
+      "loss": 1.1104,
+      "step": 148560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7544730789934512e-06,
+      "loss": 1.2647,
+      "step": 148565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7538829230536457e-06,
+      "loss": 1.3371,
+      "step": 148570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.753292802782531e-06,
+      "loss": 1.1156,
+      "step": 148575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7527027181834707e-06,
+      "loss": 1.2541,
+      "step": 148580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7521126692598385e-06,
+      "loss": 1.2433,
+      "step": 148585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7515226560150065e-06,
+      "loss": 1.3835,
+      "step": 148590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7509326784523415e-06,
+      "loss": 1.4169,
+      "step": 148595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7503427365752077e-06,
+      "loss": 1.3044,
+      "step": 148600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.749752830386979e-06,
+      "loss": 1.0764,
+      "step": 148605
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.749162959891026e-06,
+      "loss": 1.2664,
+      "step": 148610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.748573125090712e-06,
+      "loss": 1.2632,
+      "step": 148615
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.7479833259894106e-06,
+      "loss": 1.2673,
+      "step": 148620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7473935625904832e-06,
+      "loss": 1.208,
+      "step": 148625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7468038348973056e-06,
+      "loss": 1.1417,
+      "step": 148630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.746214142913238e-06,
+      "loss": 1.4506,
+      "step": 148635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.745624486641655e-06,
+      "loss": 1.1481,
+      "step": 148640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7450348660859194e-06,
+      "loss": 1.4179,
+      "step": 148645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7444452812493927e-06,
+      "loss": 1.1284,
+      "step": 148650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.743855732135454e-06,
+      "loss": 1.0274,
+      "step": 148655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7432662187474645e-06,
+      "loss": 1.3881,
+      "step": 148660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7426767410887864e-06,
+      "loss": 1.1966,
+      "step": 148665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.742087299162794e-06,
+      "loss": 1.4557,
+      "step": 148670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7414978929728454e-06,
+      "loss": 1.148,
+      "step": 148675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7409085225223087e-06,
+      "loss": 1.3146,
+      "step": 148680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.740319187814555e-06,
+      "loss": 1.6579,
+      "step": 148685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7397298888529456e-06,
+      "loss": 1.0011,
+      "step": 148690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7391406256408425e-06,
+      "loss": 1.354,
+      "step": 148695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7385513981816125e-06,
+      "loss": 1.1634,
+      "step": 148700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.737962206478627e-06,
+      "loss": 1.2693,
+      "step": 148705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7373730505352434e-06,
+      "loss": 1.3371,
+      "step": 148710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.736783930354826e-06,
+      "loss": 1.3276,
+      "step": 148715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7361948459407404e-06,
+      "loss": 0.9313,
+      "step": 148720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7356057972963557e-06,
+      "loss": 1.4915,
+      "step": 148725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7350167844250273e-06,
+      "loss": 1.0259,
+      "step": 148730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7344278073301255e-06,
+      "loss": 1.1283,
+      "step": 148735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.733838866015008e-06,
+      "loss": 1.0978,
+      "step": 148740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.733249960483041e-06,
+      "loss": 1.2527,
+      "step": 148745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7326610907375903e-06,
+      "loss": 1.7996,
+      "step": 148750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7320722567820155e-06,
+      "loss": 1.7164,
+      "step": 148755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.731483458619676e-06,
+      "loss": 1.1296,
+      "step": 148760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7308946962539417e-06,
+      "loss": 1.032,
+      "step": 148765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.730305969688166e-06,
+      "loss": 1.3854,
+      "step": 148770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7297172789257196e-06,
+      "loss": 1.2767,
+      "step": 148775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7291286239699562e-06,
+      "loss": 1.4032,
+      "step": 148780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.728540004824245e-06,
+      "loss": 1.3267,
+      "step": 148785
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.72795142149194e-06,
+      "loss": 1.0266,
+      "step": 148790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7273628739764058e-06,
+      "loss": 1.3864,
+      "step": 148795
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.726774362281006e-06,
+      "loss": 1.4629,
+      "step": 148800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7261858864090994e-06,
+      "loss": 1.5608,
+      "step": 148805
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7255974463640423e-06,
+      "loss": 1.2104,
+      "step": 148810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.725009042149198e-06,
+      "loss": 1.4069,
+      "step": 148815
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7244206737679313e-06,
+      "loss": 1.3276,
+      "step": 148820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.723832341223593e-06,
+      "loss": 1.1172,
+      "step": 148825
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.723244044519553e-06,
+      "loss": 1.2795,
+      "step": 148830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7226557836591616e-06,
+      "loss": 1.0723,
+      "step": 148835
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.722067558645781e-06,
+      "loss": 1.2508,
+      "step": 148840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7214793694827756e-06,
+      "loss": 1.0323,
+      "step": 148845
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7208912161734987e-06,
+      "loss": 1.3016,
+      "step": 148850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.720303098721312e-06,
+      "loss": 1.2932,
+      "step": 148855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.719715017129567e-06,
+      "loss": 1.5995,
+      "step": 148860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7191269714016277e-06,
+      "loss": 1.1112,
+      "step": 148865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7185389615408554e-06,
+      "loss": 1.2793,
+      "step": 148870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7179509875506e-06,
+      "loss": 1.4392,
+      "step": 148875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7173630494342283e-06,
+      "loss": 1.0971,
+      "step": 148880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7167751471950877e-06,
+      "loss": 1.6473,
+      "step": 148885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.716187280836541e-06,
+      "loss": 1.4055,
+      "step": 148890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7155994503619473e-06,
+      "loss": 0.9259,
+      "step": 148895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7150116557746616e-06,
+      "loss": 1.3148,
+      "step": 148900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7144238970780364e-06,
+      "loss": 1.1453,
+      "step": 148905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7138361742754315e-06,
+      "loss": 1.1817,
+      "step": 148910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.713248487370207e-06,
+      "loss": 1.3374,
+      "step": 148915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7126608363657135e-06,
+      "loss": 1.2684,
+      "step": 148920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7120732212653066e-06,
+      "loss": 1.4032,
+      "step": 148925
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7114856420723423e-06,
+      "loss": 1.03,
+      "step": 148930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.710898098790181e-06,
+      "loss": 1.189,
+      "step": 148935
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7103105914221704e-06,
+      "loss": 1.4765,
+      "step": 148940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7097231199716734e-06,
+      "loss": 1.5158,
+      "step": 148945
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7091356844420403e-06,
+      "loss": 1.0339,
+      "step": 148950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7085482848366195e-06,
+      "loss": 1.3836,
+      "step": 148955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7079609211587775e-06,
+      "loss": 1.1709,
+      "step": 148960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.707373593411864e-06,
+      "loss": 1.3066,
+      "step": 148965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.706786301599228e-06,
+      "loss": 1.2229,
+      "step": 148970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.70619904572423e-06,
+      "loss": 1.105,
+      "step": 148975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7056118257902184e-06,
+      "loss": 1.5957,
+      "step": 148980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7050246418005474e-06,
+      "loss": 1.2338,
+      "step": 148985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.704437493758576e-06,
+      "loss": 1.0766,
+      "step": 148990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.703850381667652e-06,
+      "loss": 1.0626,
+      "step": 148995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.703263305531125e-06,
+      "loss": 1.0269,
+      "step": 149000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7026762653523516e-06,
+      "loss": 1.0418,
+      "step": 149005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.702089261134688e-06,
+      "loss": 1.3331,
+      "step": 149010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7015022928814812e-06,
+      "loss": 1.2378,
+      "step": 149015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.70091536059608e-06,
+      "loss": 1.2252,
+      "step": 149020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7003284642818416e-06,
+      "loss": 1.3441,
+      "step": 149025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6997416039421176e-06,
+      "loss": 1.2327,
+      "step": 149030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6991547795802553e-06,
+      "loss": 1.1552,
+      "step": 149035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.698567991199611e-06,
+      "loss": 0.9758,
+      "step": 149040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.697981238803533e-06,
+      "loss": 1.1673,
+      "step": 149045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6973945223953655e-06,
+      "loss": 1.1397,
+      "step": 149050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6968078419784716e-06,
+      "loss": 0.9755,
+      "step": 149055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6962211975561944e-06,
+      "loss": 1.2427,
+      "step": 149060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.695634589131886e-06,
+      "loss": 1.3944,
+      "step": 149065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.69504801670889e-06,
+      "loss": 1.1632,
+      "step": 149070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.694461480290561e-06,
+      "loss": 1.2357,
+      "step": 149075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.693874979880253e-06,
+      "loss": 1.0848,
+      "step": 149080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.693288515481307e-06,
+      "loss": 1.4849,
+      "step": 149085
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6927020870970775e-06,
+      "loss": 1.2467,
+      "step": 149090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.69211569473091e-06,
+      "loss": 1.3699,
+      "step": 149095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.691529338386154e-06,
+      "loss": 1.1699,
+      "step": 149100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.690943018066161e-06,
+      "loss": 1.1015,
+      "step": 149105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6903567337742774e-06,
+      "loss": 1.2881,
+      "step": 149110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6897704855138472e-06,
+      "loss": 1.2157,
+      "step": 149115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6891842732882222e-06,
+      "loss": 0.9846,
+      "step": 149120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.688598097100752e-06,
+      "loss": 1.0417,
+      "step": 149125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.688011956954778e-06,
+      "loss": 1.2741,
+      "step": 149130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.687425852853653e-06,
+      "loss": 1.1861,
+      "step": 149135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6868397848007196e-06,
+      "loss": 1.0794,
+      "step": 149140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.686253752799329e-06,
+      "loss": 1.0784,
+      "step": 149145
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6856677568528233e-06,
+      "loss": 1.0331,
+      "step": 149150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6850817969645525e-06,
+      "loss": 1.2673,
+      "step": 149155
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.684495873137862e-06,
+      "loss": 1.1127,
+      "step": 149160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.683909985376094e-06,
+      "loss": 1.1664,
+      "step": 149165
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6833241336825965e-06,
+      "loss": 1.1016,
+      "step": 149170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.682738318060719e-06,
+      "loss": 1.3752,
+      "step": 149175
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6821525385138e-06,
+      "loss": 1.3524,
+      "step": 149180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6815667950451917e-06,
+      "loss": 1.1863,
+      "step": 149185
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.680981087658232e-06,
+      "loss": 1.2771,
+      "step": 149190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6803954163562695e-06,
+      "loss": 1.0414,
+      "step": 149195
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6798097811426502e-06,
+      "loss": 1.0382,
+      "step": 149200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6792241820207177e-06,
+      "loss": 1.0311,
+      "step": 149205
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6786386189938106e-06,
+      "loss": 1.0841,
+      "step": 149210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6780530920652767e-06,
+      "loss": 1.561,
+      "step": 149215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.677467601238465e-06,
+      "loss": 1.0722,
+      "step": 149220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6768821465167114e-06,
+      "loss": 1.0574,
+      "step": 149225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.67629672790336e-06,
+      "loss": 1.0325,
+      "step": 149230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6757113454017555e-06,
+      "loss": 1.5784,
+      "step": 149235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6751259990152434e-06,
+      "loss": 1.1546,
+      "step": 149240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.67454068874716e-06,
+      "loss": 0.8982,
+      "step": 149245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.673955414600856e-06,
+      "loss": 1.3434,
+      "step": 149250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.673370176579668e-06,
+      "loss": 1.1444,
+      "step": 149255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6727849746869337e-06,
+      "loss": 1.0574,
+      "step": 149260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.672199808926006e-06,
+      "loss": 1.1626,
+      "step": 149265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6716146793002204e-06,
+      "loss": 1.3322,
+      "step": 149270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6710295858129163e-06,
+      "loss": 1.251,
+      "step": 149275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6704445284674404e-06,
+      "loss": 1.223,
+      "step": 149280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.669859507267127e-06,
+      "loss": 1.0626,
+      "step": 149285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6692745222153237e-06,
+      "loss": 1.1415,
+      "step": 149290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.668689573315365e-06,
+      "loss": 1.1193,
+      "step": 149295
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.668104660570597e-06,
+      "loss": 1.5014,
+      "step": 149300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6675197839843547e-06,
+      "loss": 1.2403,
+      "step": 149305
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6669349435599788e-06,
+      "loss": 1.1942,
+      "step": 149310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.666350139300815e-06,
+      "loss": 1.0124,
+      "step": 149315
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.665765371210198e-06,
+      "loss": 1.0323,
+      "step": 149320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6651806392914647e-06,
+      "loss": 1.1632,
+      "step": 149325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6645959435479573e-06,
+      "loss": 1.0922,
+      "step": 149330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6640112839830177e-06,
+      "loss": 1.2567,
+      "step": 149335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.663426660599978e-06,
+      "loss": 1.198,
+      "step": 149340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6628420734021843e-06,
+      "loss": 1.2816,
+      "step": 149345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6622575223929703e-06,
+      "loss": 1.3924,
+      "step": 149350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6616730075756703e-06,
+      "loss": 1.5429,
+      "step": 149355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.661088528953627e-06,
+      "loss": 1.441,
+      "step": 149360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.660504086530182e-06,
+      "loss": 1.2931,
+      "step": 149365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.659919680308668e-06,
+      "loss": 1.4004,
+      "step": 149370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.659335310292418e-06,
+      "loss": 1.261,
+      "step": 149375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6587509764847752e-06,
+      "loss": 0.9767,
+      "step": 149380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.658166678889078e-06,
+      "loss": 1.0695,
+      "step": 149385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.657582417508656e-06,
+      "loss": 1.1936,
+      "step": 149390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6569981923468533e-06,
+      "loss": 1.2016,
+      "step": 149395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6564140034069997e-06,
+      "loss": 1.2126,
+      "step": 149400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.655829850692434e-06,
+      "loss": 1.2601,
+      "step": 149405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.655245734206495e-06,
+      "loss": 1.542,
+      "step": 149410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.654661653952516e-06,
+      "loss": 1.2029,
+      "step": 149415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.654077609933827e-06,
+      "loss": 1.0747,
+      "step": 149420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6534936021537692e-06,
+      "loss": 1.0113,
+      "step": 149425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6529096306156797e-06,
+      "loss": 1.1451,
+      "step": 149430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.65232569532289e-06,
+      "loss": 1.1613,
+      "step": 149435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6517417962787316e-06,
+      "loss": 1.2236,
+      "step": 149440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6511579334865443e-06,
+      "loss": 1.3572,
+      "step": 149445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.650574106949658e-06,
+      "loss": 1.0961,
+      "step": 149450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.649990316671408e-06,
+      "loss": 1.1161,
+      "step": 149455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6494065626551313e-06,
+      "loss": 1.3147,
+      "step": 149460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6488228449041598e-06,
+      "loss": 1.2472,
+      "step": 149465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.648239163421822e-06,
+      "loss": 1.334,
+      "step": 149470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6476555182114546e-06,
+      "loss": 0.9951,
+      "step": 149475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.647071909276395e-06,
+      "loss": 1.1441,
+      "step": 149480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6464883366199677e-06,
+      "loss": 1.1458,
+      "step": 149485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6459048002455122e-06,
+      "loss": 1.4081,
+      "step": 149490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6453213001563558e-06,
+      "loss": 1.206,
+      "step": 149495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.644737836355835e-06,
+      "loss": 1.2027,
+      "step": 149500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.644154408847277e-06,
+      "loss": 1.0271,
+      "step": 149505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6435710176340177e-06,
+      "loss": 1.7328,
+      "step": 149510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.642987662719384e-06,
+      "loss": 1.0913,
+      "step": 149515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.64240434410671e-06,
+      "loss": 1.0569,
+      "step": 149520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.64182106179933e-06,
+      "loss": 1.3857,
+      "step": 149525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6412378158005724e-06,
+      "loss": 0.9941,
+      "step": 149530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6406546061137625e-06,
+      "loss": 0.9721,
+      "step": 149535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.640071432742235e-06,
+      "loss": 1.2257,
+      "step": 149540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.639488295689324e-06,
+      "loss": 0.7457,
+      "step": 149545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.638905194958353e-06,
+      "loss": 1.1956,
+      "step": 149550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6383221305526572e-06,
+      "loss": 1.3399,
+      "step": 149555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6377391024755646e-06,
+      "loss": 1.5052,
+      "step": 149560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6371561107304e-06,
+      "loss": 1.2449,
+      "step": 149565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6365731553204954e-06,
+      "loss": 0.9341,
+      "step": 149570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.635990236249185e-06,
+      "loss": 1.0884,
+      "step": 149575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.635407353519793e-06,
+      "loss": 1.7037,
+      "step": 149580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6348245071356447e-06,
+      "loss": 1.2414,
+      "step": 149585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6342416971000717e-06,
+      "loss": 1.1767,
+      "step": 149590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6336589234164067e-06,
+      "loss": 1.3099,
+      "step": 149595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6330761860879692e-06,
+      "loss": 1.1714,
+      "step": 149600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.632493485118095e-06,
+      "loss": 0.897,
+      "step": 149605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6319108205101038e-06,
+      "loss": 1.1673,
+      "step": 149610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6313281922673283e-06,
+      "loss": 1.1703,
+      "step": 149615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.630745600393096e-06,
+      "loss": 1.2367,
+      "step": 149620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6301630448907333e-06,
+      "loss": 1.401,
+      "step": 149625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.629580525763562e-06,
+      "loss": 1.2115,
+      "step": 149630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.628998043014911e-06,
+      "loss": 1.0603,
+      "step": 149635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.628415596648113e-06,
+      "loss": 1.2391,
+      "step": 149640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.627833186666487e-06,
+      "loss": 0.9782,
+      "step": 149645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6272508130733587e-06,
+      "loss": 1.3205,
+      "step": 149650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6266684758720594e-06,
+      "loss": 1.2049,
+      "step": 149655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.626086175065907e-06,
+      "loss": 1.2182,
+      "step": 149660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6255039106582313e-06,
+      "loss": 1.1224,
+      "step": 149665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6249216826523593e-06,
+      "loss": 1.1408,
+      "step": 149670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.624339491051614e-06,
+      "loss": 1.0369,
+      "step": 149675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.623757335859316e-06,
+      "loss": 1.7264,
+      "step": 149680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6231752170787928e-06,
+      "loss": 1.1672,
+      "step": 149685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6225931347133726e-06,
+      "loss": 1.1028,
+      "step": 149690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6220110887663727e-06,
+      "loss": 1.5549,
+      "step": 149695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6214290792411222e-06,
+      "loss": 1.3514,
+      "step": 149700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.62084710614094e-06,
+      "loss": 1.7089,
+      "step": 149705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6202651694691517e-06,
+      "loss": 1.5108,
+      "step": 149710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.619683269229083e-06,
+      "loss": 1.0458,
+      "step": 149715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6191014054240557e-06,
+      "loss": 0.8867,
+      "step": 149720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.618519578057388e-06,
+      "loss": 1.4479,
+      "step": 149725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.617937787132405e-06,
+      "loss": 1.4878,
+      "step": 149730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6173560326524335e-06,
+      "loss": 1.1162,
+      "step": 149735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.616774314620792e-06,
+      "loss": 1.1391,
+      "step": 149740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.616192633040799e-06,
+      "loss": 1.4426,
+      "step": 149745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.615610987915783e-06,
+      "loss": 0.9263,
+      "step": 149750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6150293792490587e-06,
+      "loss": 0.9189,
+      "step": 149755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6144478070439516e-06,
+      "loss": 1.1868,
+      "step": 149760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.613866271303784e-06,
+      "loss": 1.3835,
+      "step": 149765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.613284772031875e-06,
+      "loss": 1.2698,
+      "step": 149770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6127033092315434e-06,
+      "loss": 1.1367,
+      "step": 149775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.61212188290611e-06,
+      "loss": 1.2677,
+      "step": 149780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.611540493058899e-06,
+      "loss": 1.198,
+      "step": 149785
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6109591396932286e-06,
+      "loss": 1.2007,
+      "step": 149790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.610377822812415e-06,
+      "loss": 0.9868,
+      "step": 149795
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.60979654241978e-06,
+      "loss": 1.2554,
+      "step": 149800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.609215298518648e-06,
+      "loss": 1.0835,
+      "step": 149805
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.60863409111233e-06,
+      "loss": 1.2241,
+      "step": 149810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6080529202041516e-06,
+      "loss": 1.5294,
+      "step": 149815
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6074717857974263e-06,
+      "loss": 1.2404,
+      "step": 149820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6068906878954755e-06,
+      "loss": 1.3486,
+      "step": 149825
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6063096265016206e-06,
+      "loss": 1.1247,
+      "step": 149830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.605728601619176e-06,
+      "loss": 1.0564,
+      "step": 149835
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6051476132514585e-06,
+      "loss": 1.0915,
+      "step": 149840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.604566661401786e-06,
+      "loss": 1.1268,
+      "step": 149845
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6039857460734817e-06,
+      "loss": 1.0005,
+      "step": 149850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.603404867269855e-06,
+      "loss": 1.4485,
+      "step": 149855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6028240249942316e-06,
+      "loss": 1.2485,
+      "step": 149860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6022432192499223e-06,
+      "loss": 1.4183,
+      "step": 149865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.601662450040243e-06,
+      "loss": 1.7269,
+      "step": 149870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.601081717368512e-06,
+      "loss": 1.4646,
+      "step": 149875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.6005010212380487e-06,
+      "loss": 1.4314,
+      "step": 149880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.599920361652167e-06,
+      "loss": 1.4904,
+      "step": 149885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5993397386141793e-06,
+      "loss": 1.221,
+      "step": 149890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5987591521274046e-06,
+      "loss": 1.3724,
+      "step": 149895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5981786021951613e-06,
+      "loss": 1.3158,
+      "step": 149900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5975980888207575e-06,
+      "loss": 0.9048,
+      "step": 149905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.597017612007516e-06,
+      "loss": 1.1617,
+      "step": 149910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5964371717587444e-06,
+      "loss": 1.4257,
+      "step": 149915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.59585676807776e-06,
+      "loss": 1.1491,
+      "step": 149920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.595276400967882e-06,
+      "loss": 1.255,
+      "step": 149925
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.594696070432421e-06,
+      "loss": 1.2039,
+      "step": 149930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.594115776474689e-06,
+      "loss": 1.1137,
+      "step": 149935
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.593535519097996e-06,
+      "loss": 1.0405,
+      "step": 149940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.592955298305667e-06,
+      "loss": 1.0941,
+      "step": 149945
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5923751141010098e-06,
+      "loss": 1.3079,
+      "step": 149950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.591794966487333e-06,
+      "loss": 1.0557,
+      "step": 149955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5912148554679584e-06,
+      "loss": 1.4883,
+      "step": 149960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5906347810461894e-06,
+      "loss": 1.0366,
+      "step": 149965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.590054743225343e-06,
+      "loss": 1.2596,
+      "step": 149970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.589474742008735e-06,
+      "loss": 1.1286,
+      "step": 149975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5888947773996743e-06,
+      "loss": 1.2372,
+      "step": 149980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.58831484940147e-06,
+      "loss": 1.5509,
+      "step": 149985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5877349580174357e-06,
+      "loss": 0.969,
+      "step": 149990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5871551032508865e-06,
+      "loss": 1.2032,
+      "step": 149995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5865752851051284e-06,
+      "loss": 1.308,
+      "step": 150000
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 1.2216823101043701,
+      "eval_runtime": 6465.1035,
+      "eval_samples_per_second": 3.575,
+      "eval_steps_per_second": 1.787,
+      "step": 150000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.585995503583478e-06,
+      "loss": 1.3605,
+      "step": 150005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.585415758689239e-06,
+      "loss": 1.2273,
+      "step": 150010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5848360504257307e-06,
+      "loss": 0.9459,
+      "step": 150015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5842563787962547e-06,
+      "loss": 1.2758,
+      "step": 150020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5836767438041296e-06,
+      "loss": 1.6004,
+      "step": 150025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.583097145452661e-06,
+      "loss": 1.0701,
+      "step": 150030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.582517583745152e-06,
+      "loss": 1.0831,
+      "step": 150035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5819380586849262e-06,
+      "loss": 1.3755,
+      "step": 150040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.581358570275286e-06,
+      "loss": 1.1597,
+      "step": 150045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5807791185195373e-06,
+      "loss": 1.4752,
+      "step": 150050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5801997034209956e-06,
+      "loss": 1.2079,
+      "step": 150055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.579620324982964e-06,
+      "loss": 1.1412,
+      "step": 150060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5790409832087526e-06,
+      "loss": 1.3294,
+      "step": 150065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5784616781016747e-06,
+      "loss": 1.2602,
+      "step": 150070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5778824096650345e-06,
+      "loss": 1.1294,
+      "step": 150075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5773031779021363e-06,
+      "loss": 1.2546,
+      "step": 150080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5767239828162903e-06,
+      "loss": 1.3126,
+      "step": 150085
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.57614482441081e-06,
+      "loss": 1.1258,
+      "step": 150090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5755657026889965e-06,
+      "loss": 1.3361,
+      "step": 150095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.574986617654156e-06,
+      "loss": 1.0109,
+      "step": 150100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.574407569309598e-06,
+      "loss": 1.0637,
+      "step": 150105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5738285576586305e-06,
+      "loss": 1.3902,
+      "step": 150110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5732495827045565e-06,
+      "loss": 0.9838,
+      "step": 150115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5726706444506863e-06,
+      "loss": 1.0689,
+      "step": 150120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.572091742900321e-06,
+      "loss": 1.0763,
+      "step": 150125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5715128780567685e-06,
+      "loss": 1.118,
+      "step": 150130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5709340499233393e-06,
+      "loss": 1.0372,
+      "step": 150135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.570355258503334e-06,
+      "loss": 1.0535,
+      "step": 150140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5697765038000553e-06,
+      "loss": 1.1667,
+      "step": 150145
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5691977858168146e-06,
+      "loss": 1.2369,
+      "step": 150150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5686191045569096e-06,
+      "loss": 1.0934,
+      "step": 150155
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5680404600236528e-06,
+      "loss": 1.6634,
+      "step": 150160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5674618522203398e-06,
+      "loss": 1.0202,
+      "step": 150165
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.566883281150284e-06,
+      "loss": 1.1315,
+      "step": 150170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.56630474681678e-06,
+      "loss": 1.0682,
+      "step": 150175
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.565726249223137e-06,
+      "loss": 1.3156,
+      "step": 150180
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5651477883726616e-06,
+      "loss": 1.2365,
+      "step": 150185
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.564569364268653e-06,
+      "loss": 1.2722,
+      "step": 150190
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.563990976914411e-06,
+      "loss": 1.1927,
+      "step": 150195
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.563412626313243e-06,
+      "loss": 1.3863,
+      "step": 150200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.562834312468453e-06,
+      "loss": 1.2381,
+      "step": 150205
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.562256035383339e-06,
+      "loss": 1.2624,
+      "step": 150210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.561677795061208e-06,
+      "loss": 1.2205,
+      "step": 150215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.561099591505357e-06,
+      "loss": 1.0828,
+      "step": 150220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.56052142471909e-06,
+      "loss": 1.2357,
+      "step": 150225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5599432947057133e-06,
+      "loss": 1.1096,
+      "step": 150230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.559365201468523e-06,
+      "loss": 1.1004,
+      "step": 150235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5587871450108225e-06,
+      "loss": 1.0024,
+      "step": 150240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5582091253359054e-06,
+      "loss": 1.1183,
+      "step": 150245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5576311424470845e-06,
+      "loss": 1.3363,
+      "step": 150250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.557053196347656e-06,
+      "loss": 1.1266,
+      "step": 150255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5564752870409146e-06,
+      "loss": 1.2193,
+      "step": 150260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5558974145301696e-06,
+      "loss": 1.1491,
+      "step": 150265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.555319578818712e-06,
+      "loss": 1.06,
+      "step": 150270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.554741779909846e-06,
+      "loss": 1.13,
+      "step": 150275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5541640178068747e-06,
+      "loss": 1.0688,
+      "step": 150280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5535862925130936e-06,
+      "loss": 1.2499,
+      "step": 150285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.553008604031798e-06,
+      "loss": 1.1862,
+      "step": 150290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5524309523662915e-06,
+      "loss": 1.3107,
+      "step": 150295
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.551853337519876e-06,
+      "loss": 1.1423,
+      "step": 150300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5512757594958447e-06,
+      "loss": 1.076,
+      "step": 150305
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5506982182974947e-06,
+      "loss": 1.0433,
+      "step": 150310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.550120713928127e-06,
+      "loss": 1.0903,
+      "step": 150315
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5495432463910417e-06,
+      "loss": 1.4076,
+      "step": 150320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5489658156895314e-06,
+      "loss": 1.5061,
+      "step": 150325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.548388421826898e-06,
+      "loss": 1.6303,
+      "step": 150330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.547811064806438e-06,
+      "loss": 1.1321,
+      "step": 150335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.54723374463144e-06,
+      "loss": 1.1448,
+      "step": 150340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5466564613052146e-06,
+      "loss": 1.371,
+      "step": 150345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.546079214831052e-06,
+      "loss": 1.2415,
+      "step": 150350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5455020052122446e-06,
+      "loss": 1.2275,
+      "step": 150355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5449248324520966e-06,
+      "loss": 1.304,
+      "step": 150360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5443476965538946e-06,
+      "loss": 1.1787,
+      "step": 150365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.543770597520941e-06,
+      "loss": 1.0143,
+      "step": 150370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.543193535356533e-06,
+      "loss": 1.1757,
+      "step": 150375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.542616510063962e-06,
+      "loss": 1.6056,
+      "step": 150380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.542039521646521e-06,
+      "loss": 1.2385,
+      "step": 150385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.541462570107508e-06,
+      "loss": 1.208,
+      "step": 150390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.54088565545022e-06,
+      "loss": 1.2771,
+      "step": 150395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.540308777677949e-06,
+      "loss": 1.2224,
+      "step": 150400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5397319367939862e-06,
+      "loss": 1.0894,
+      "step": 150405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5391551328016292e-06,
+      "loss": 1.5614,
+      "step": 150410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5385783657041737e-06,
+      "loss": 1.3139,
+      "step": 150415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5380016355049087e-06,
+      "loss": 1.1283,
+      "step": 150420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.537424942207133e-06,
+      "loss": 1.0638,
+      "step": 150425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.536848285814133e-06,
+      "loss": 1.2702,
+      "step": 150430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.536271666329205e-06,
+      "loss": 1.2579,
+      "step": 150435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.535695083755647e-06,
+      "loss": 1.3462,
+      "step": 150440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5351185380967468e-06,
+      "loss": 1.3814,
+      "step": 150445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5345420293557963e-06,
+      "loss": 1.2288,
+      "step": 150450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5339655575360853e-06,
+      "loss": 1.4526,
+      "step": 150455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.533389122640908e-06,
+      "loss": 1.2118,
+      "step": 150460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.532812724673561e-06,
+      "loss": 1.56,
+      "step": 150465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.532236363637328e-06,
+      "loss": 1.0475,
+      "step": 150470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5316600395355073e-06,
+      "loss": 1.566,
+      "step": 150475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5310837523713836e-06,
+      "loss": 1.0856,
+      "step": 150480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5305075021482517e-06,
+      "loss": 1.1379,
+      "step": 150485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.529931288869404e-06,
+      "loss": 1.1726,
+      "step": 150490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5293551125381288e-06,
+      "loss": 1.035,
+      "step": 150495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5287789731577127e-06,
+      "loss": 1.0115,
+      "step": 150500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5282028707314498e-06,
+      "loss": 1.3743,
+      "step": 150505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5276268052626328e-06,
+      "loss": 1.2433,
+      "step": 150510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5270507767545435e-06,
+      "loss": 1.3185,
+      "step": 150515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5264747852104796e-06,
+      "loss": 1.0664,
+      "step": 150520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.525898830633724e-06,
+      "loss": 1.248,
+      "step": 150525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5253229130275713e-06,
+      "loss": 1.0303,
+      "step": 150530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5247470323953025e-06,
+      "loss": 1.2125,
+      "step": 150535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.524171188740215e-06,
+      "loss": 1.4211,
+      "step": 150540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5235953820655934e-06,
+      "loss": 1.2117,
+      "step": 150545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.523019612374722e-06,
+      "loss": 0.9497,
+      "step": 150550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.522443879670893e-06,
+      "loss": 1.2643,
+      "step": 150555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5218681839573954e-06,
+      "loss": 1.0848,
+      "step": 150560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.521292525237512e-06,
+      "loss": 1.4899,
+      "step": 150565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.520716903514536e-06,
+      "loss": 1.0584,
+      "step": 150570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5201413187917477e-06,
+      "loss": 1.2617,
+      "step": 150575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5195657710724384e-06,
+      "loss": 1.2681,
+      "step": 150580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.518990260359897e-06,
+      "loss": 1.199,
+      "step": 150585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.518414786657407e-06,
+      "loss": 1.32,
+      "step": 150590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5178393499682518e-06,
+      "loss": 1.1475,
+      "step": 150595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5172639502957204e-06,
+      "loss": 1.2244,
+      "step": 150600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5166885876431012e-06,
+      "loss": 1.2555,
+      "step": 150605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5161132620136784e-06,
+      "loss": 1.1183,
+      "step": 150610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.515537973410732e-06,
+      "loss": 1.1665,
+      "step": 150615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5149627218375517e-06,
+      "loss": 1.0429,
+      "step": 150620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5143875072974253e-06,
+      "loss": 1.2182,
+      "step": 150625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.513812329793631e-06,
+      "loss": 1.3887,
+      "step": 150630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5132371893294613e-06,
+      "loss": 1.1354,
+      "step": 150635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.512662085908195e-06,
+      "loss": 1.1934,
+      "step": 150640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5120870195331114e-06,
+      "loss": 1.3524,
+      "step": 150645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.511511990207508e-06,
+      "loss": 1.1367,
+      "step": 150650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5109369979346598e-06,
+      "loss": 1.5782,
+      "step": 150655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5103620427178473e-06,
+      "loss": 1.2708,
+      "step": 150660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5097871245603633e-06,
+      "loss": 0.9723,
+      "step": 150665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.509212243465481e-06,
+      "loss": 1.3923,
+      "step": 150670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.508637399436492e-06,
+      "loss": 1.3147,
+      "step": 150675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.508062592476671e-06,
+      "loss": 1.2078,
+      "step": 150680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.507487822589307e-06,
+      "loss": 1.0362,
+      "step": 150685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.506913089777678e-06,
+      "loss": 1.5293,
+      "step": 150690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.506338394045066e-06,
+      "loss": 1.5109,
+      "step": 150695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.5057637353947582e-06,
+      "loss": 0.9495,
+      "step": 150700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.505189113830033e-06,
+      "loss": 1.8354,
+      "step": 150705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.504614529354167e-06,
+      "loss": 1.3302,
+      "step": 150710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5040399819704464e-06,
+      "loss": 1.602,
+      "step": 150715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5034654716821535e-06,
+      "loss": 1.3001,
+      "step": 150720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.502890998492565e-06,
+      "loss": 1.1395,
+      "step": 150725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.502316562404966e-06,
+      "loss": 1.2308,
+      "step": 150730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5017421634226343e-06,
+      "loss": 1.2505,
+      "step": 150735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.501167801548846e-06,
+      "loss": 1.0906,
+      "step": 150740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5005934767868856e-06,
+      "loss": 1.3886,
+      "step": 150745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5000191891400357e-06,
+      "loss": 1.4095,
+      "step": 150750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4994449386115727e-06,
+      "loss": 0.9878,
+      "step": 150755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4988707252047703e-06,
+      "loss": 1.3692,
+      "step": 150760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4982965489229146e-06,
+      "loss": 1.281,
+      "step": 150765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4977224097692853e-06,
+      "loss": 1.4739,
+      "step": 150770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4971483077471546e-06,
+      "loss": 1.3348,
+      "step": 150775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4965742428598084e-06,
+      "loss": 1.27,
+      "step": 150780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4960002151105167e-06,
+      "loss": 1.3912,
+      "step": 150785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4954262245025627e-06,
+      "loss": 1.4775,
+      "step": 150790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4948522710392273e-06,
+      "loss": 1.2913,
+      "step": 150795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4942783547237848e-06,
+      "loss": 1.1408,
+      "step": 150800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.493704475559507e-06,
+      "loss": 1.3555,
+      "step": 150805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4931306335496775e-06,
+      "loss": 1.1737,
+      "step": 150810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4925568286975743e-06,
+      "loss": 1.0579,
+      "step": 150815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.491983061006472e-06,
+      "loss": 1.1989,
+      "step": 150820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4914093304796425e-06,
+      "loss": 1.1956,
+      "step": 150825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4908356371203677e-06,
+      "loss": 1.1765,
+      "step": 150830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4902619809319247e-06,
+      "loss": 1.1278,
+      "step": 150835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4896883619175847e-06,
+      "loss": 1.4374,
+      "step": 150840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.489114780080628e-06,
+      "loss": 1.2066,
+      "step": 150845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4885412354243287e-06,
+      "loss": 1.361,
+      "step": 150850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4879677279519573e-06,
+      "loss": 1.4417,
+      "step": 150855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.487394257666793e-06,
+      "loss": 1.1468,
+      "step": 150860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4868208245721135e-06,
+      "loss": 1.2308,
+      "step": 150865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4862474286711866e-06,
+      "loss": 1.1297,
+      "step": 150870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.485674069967294e-06,
+      "loss": 1.1182,
+      "step": 150875
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4851007484637036e-06,
+      "loss": 1.1536,
+      "step": 150880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4845274641636937e-06,
+      "loss": 1.1391,
+      "step": 150885
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4839542170705344e-06,
+      "loss": 1.2878,
+      "step": 150890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.483381007187504e-06,
+      "loss": 1.3368,
+      "step": 150895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4828078345178693e-06,
+      "loss": 1.1722,
+      "step": 150900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.482234699064908e-06,
+      "loss": 1.1195,
+      "step": 150905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.481661600831895e-06,
+      "loss": 1.2929,
+      "step": 150910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4810885398220993e-06,
+      "loss": 1.2474,
+      "step": 150915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4805155160387928e-06,
+      "loss": 1.1275,
+      "step": 150920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4799425294852473e-06,
+      "loss": 0.9954,
+      "step": 150925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.479369580164742e-06,
+      "loss": 0.8714,
+      "step": 150930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4787966680805387e-06,
+      "loss": 1.0681,
+      "step": 150935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4782237932359175e-06,
+      "loss": 1.2081,
+      "step": 150940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4776509556341463e-06,
+      "loss": 0.8289,
+      "step": 150945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4770781552784907e-06,
+      "loss": 0.954,
+      "step": 150950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4765053921722325e-06,
+      "loss": 1.3204,
+      "step": 150955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.475932666318638e-06,
+      "loss": 1.1681,
+      "step": 150960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4753599777209756e-06,
+      "loss": 1.4712,
+      "step": 150965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4747873263825147e-06,
+      "loss": 1.1672,
+      "step": 150970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4742147123065274e-06,
+      "loss": 1.0598,
+      "step": 150975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4736421354962878e-06,
+      "loss": 0.962,
+      "step": 150980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4730695959550586e-06,
+      "loss": 1.0285,
+      "step": 150985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.472497093686116e-06,
+      "loss": 1.254,
+      "step": 150990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.471924628692721e-06,
+      "loss": 1.5608,
+      "step": 150995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4713522009781487e-06,
+      "loss": 1.098,
+      "step": 151000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.47077981054567e-06,
+      "loss": 1.0775,
+      "step": 151005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.47020745739855e-06,
+      "loss": 1.3279,
+      "step": 151010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.469635141540055e-06,
+      "loss": 1.3041,
+      "step": 151015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4690628629734544e-06,
+      "loss": 1.3952,
+      "step": 151020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.468490621702022e-06,
+      "loss": 1.3712,
+      "step": 151025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4679184177290214e-06,
+      "loss": 1.4938,
+      "step": 151030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.467346251057716e-06,
+      "loss": 1.0746,
+      "step": 151035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4667741216913797e-06,
+      "loss": 0.9846,
+      "step": 151040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4662020296332744e-06,
+      "loss": 1.6246,
+      "step": 151045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4656299748866707e-06,
+      "loss": 1.2461,
+      "step": 151050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.465057957454836e-06,
+      "loss": 1.1749,
+      "step": 151055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4644859773410365e-06,
+      "loss": 1.3169,
+      "step": 151060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4639140345485333e-06,
+      "loss": 1.2377,
+      "step": 151065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.463342129080596e-06,
+      "loss": 1.2581,
+      "step": 151070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4627702609404944e-06,
+      "loss": 1.0713,
+      "step": 151075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.462198430131488e-06,
+      "loss": 1.2188,
+      "step": 151080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4616266366568475e-06,
+      "loss": 1.3507,
+      "step": 151085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.461054880519833e-06,
+      "loss": 1.4327,
+      "step": 151090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.460483161723712e-06,
+      "loss": 1.2463,
+      "step": 151095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4599114802717526e-06,
+      "loss": 1.1024,
+      "step": 151100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.459339836167216e-06,
+      "loss": 1.0493,
+      "step": 151105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.458768229413364e-06,
+      "loss": 1.2963,
+      "step": 151110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.458196660013463e-06,
+      "loss": 1.5021,
+      "step": 151115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.457625127970782e-06,
+      "loss": 1.2809,
+      "step": 151120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4570536332885784e-06,
+      "loss": 1.0904,
+      "step": 151125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4564821759701163e-06,
+      "loss": 1.1697,
+      "step": 151130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.455910756018662e-06,
+      "loss": 1.3042,
+      "step": 151135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4553393734374753e-06,
+      "loss": 1.0727,
+      "step": 151140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.45476802822982e-06,
+      "loss": 1.1859,
+      "step": 151145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4541967203989634e-06,
+      "loss": 1.3615,
+      "step": 151150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4536254499481646e-06,
+      "loss": 1.1326,
+      "step": 151155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4530542168806813e-06,
+      "loss": 1.2604,
+      "step": 151160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4524830211997796e-06,
+      "loss": 0.9947,
+      "step": 151165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4519118629087257e-06,
+      "loss": 1.2618,
+      "step": 151170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.451340742010776e-06,
+      "loss": 1.0148,
+      "step": 151175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.45076965850919e-06,
+      "loss": 1.3749,
+      "step": 151180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.450198612407232e-06,
+      "loss": 1.488,
+      "step": 151185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4496276037081655e-06,
+      "loss": 1.3695,
+      "step": 151190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.449056632415245e-06,
+      "loss": 1.0805,
+      "step": 151195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4484856985317394e-06,
+      "loss": 1.2803,
+      "step": 151200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4479148020608997e-06,
+      "loss": 1.6495,
+      "step": 151205
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.447343943005991e-06,
+      "loss": 1.0861,
+      "step": 151210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.446773121370275e-06,
+      "loss": 1.0078,
+      "step": 151215
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.44620233715701e-06,
+      "loss": 0.9665,
+      "step": 151220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4456315903694513e-06,
+      "loss": 1.3829,
+      "step": 151225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.445060881010861e-06,
+      "loss": 1.1179,
+      "step": 151230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4444902090845023e-06,
+      "loss": 1.5994,
+      "step": 151235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.443919574593627e-06,
+      "loss": 1.2009,
+      "step": 151240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4433489775415008e-06,
+      "loss": 1.3305,
+      "step": 151245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.442778417931378e-06,
+      "loss": 0.8719,
+      "step": 151250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.442207895766515e-06,
+      "loss": 1.4334,
+      "step": 151255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4416374110501715e-06,
+      "loss": 1.1644,
+      "step": 151260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4410669637856087e-06,
+      "loss": 1.0609,
+      "step": 151265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4404965539760814e-06,
+      "loss": 1.133,
+      "step": 151270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.439926181624843e-06,
+      "loss": 1.598,
+      "step": 151275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.439355846735155e-06,
+      "loss": 1.3495,
+      "step": 151280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4387855493102775e-06,
+      "loss": 1.0572,
+      "step": 151285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4382152893534603e-06,
+      "loss": 1.5428,
+      "step": 151290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.437645066867965e-06,
+      "loss": 0.9492,
+      "step": 151295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4370748818570444e-06,
+      "loss": 1.0689,
+      "step": 151300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.436504734323954e-06,
+      "loss": 1.2662,
+      "step": 151305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.435934624271957e-06,
+      "loss": 1.4365,
+      "step": 151310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4353645517043033e-06,
+      "loss": 1.7954,
+      "step": 151315
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.434794516624248e-06,
+      "loss": 1.1717,
+      "step": 151320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.434224519035041e-06,
+      "loss": 1.5645,
+      "step": 151325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4336545589399507e-06,
+      "loss": 1.2377,
+      "step": 151330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4330846363422233e-06,
+      "loss": 2.0925,
+      "step": 151335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4325147512451127e-06,
+      "loss": 1.0478,
+      "step": 151340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4319449036518783e-06,
+      "loss": 1.4829,
+      "step": 151345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4313750935657664e-06,
+      "loss": 1.4676,
+      "step": 151350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.430805320990037e-06,
+      "loss": 1.0967,
+      "step": 151355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4302355859279458e-06,
+      "loss": 1.1459,
+      "step": 151360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4296658883827427e-06,
+      "loss": 1.6058,
+      "step": 151365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4290962283576777e-06,
+      "loss": 1.6755,
+      "step": 151370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4285266058560074e-06,
+      "loss": 0.9958,
+      "step": 151375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4279570208809875e-06,
+      "loss": 0.9359,
+      "step": 151380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.427387473435866e-06,
+      "loss": 1.1829,
+      "step": 151385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4268179635238996e-06,
+      "loss": 1.2501,
+      "step": 151390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.426248491148334e-06,
+      "loss": 1.5099,
+      "step": 151395
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4256790563124297e-06,
+      "loss": 1.2577,
+      "step": 151400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.42510965901943e-06,
+      "loss": 1.2911,
+      "step": 151405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.424540299272594e-06,
+      "loss": 1.1113,
+      "step": 151410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.423970977075166e-06,
+      "loss": 1.8058,
+      "step": 151415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4234016924304002e-06,
+      "loss": 0.8658,
+      "step": 151420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.422832445341553e-06,
+      "loss": 1.1073,
+      "step": 151425
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.422263235811869e-06,
+      "loss": 1.1376,
+      "step": 151430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.421694063844596e-06,
+      "loss": 1.178,
+      "step": 151435
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4211249294429917e-06,
+      "loss": 1.0913,
+      "step": 151440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4205558326102993e-06,
+      "loss": 1.358,
+      "step": 151445
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4199867733497717e-06,
+      "loss": 1.1596,
+      "step": 151450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.419417751664662e-06,
+      "loss": 1.1434,
+      "step": 151455
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4188487675582173e-06,
+      "loss": 1.6389,
+      "step": 151460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.418279821033681e-06,
+      "loss": 0.9902,
+      "step": 151465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4177109120943087e-06,
+      "loss": 1.3403,
+      "step": 151470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.417142040743349e-06,
+      "loss": 1.3365,
+      "step": 151475
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4165732069840495e-06,
+      "loss": 1.8469,
+      "step": 151480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4160044108196544e-06,
+      "loss": 1.1225,
+      "step": 151485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.415435652253416e-06,
+      "loss": 1.055,
+      "step": 151490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4148669312885842e-06,
+      "loss": 1.2051,
+      "step": 151495
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4142982479284004e-06,
+      "loss": 1.541,
+      "step": 151500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4137296021761202e-06,
+      "loss": 1.1476,
+      "step": 151505
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.413160994034982e-06,
+      "loss": 0.9746,
+      "step": 151510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4125924235082374e-06,
+      "loss": 1.2874,
+      "step": 151515
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4120238905991375e-06,
+      "loss": 1.2115,
+      "step": 151520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.411455395310924e-06,
+      "loss": 1.4188,
+      "step": 151525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4108869376468403e-06,
+      "loss": 1.1653,
+      "step": 151530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.410318517610136e-06,
+      "loss": 1.322,
+      "step": 151535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.409750135204061e-06,
+      "loss": 1.3883,
+      "step": 151540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.409181790431857e-06,
+      "loss": 1.0748,
+      "step": 151545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4086134832967666e-06,
+      "loss": 1.3608,
+      "step": 151550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4080452138020413e-06,
+      "loss": 1.5333,
+      "step": 151555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4074769819509213e-06,
+      "loss": 1.3613,
+      "step": 151560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.406908787746652e-06,
+      "loss": 1.1503,
+      "step": 151565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.406340631192484e-06,
+      "loss": 1.2236,
+      "step": 151570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.405772512291656e-06,
+      "loss": 1.0343,
+      "step": 151575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.405204431047412e-06,
+      "loss": 1.4829,
+      "step": 151580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4046363874629975e-06,
+      "loss": 1.244,
+      "step": 151585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.40406838154166e-06,
+      "loss": 1.0779,
+      "step": 151590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4035004132866358e-06,
+      "loss": 1.2486,
+      "step": 151595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4029324827011747e-06,
+      "loss": 1.2855,
+      "step": 151600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.402364589788515e-06,
+      "loss": 1.4044,
+      "step": 151605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4017967345519055e-06,
+      "loss": 1.0488,
+      "step": 151610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4012289169945823e-06,
+      "loss": 1.2248,
+      "step": 151615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.400661137119794e-06,
+      "loss": 1.395,
+      "step": 151620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.40009339493078e-06,
+      "loss": 1.2834,
+      "step": 151625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.399525690430776e-06,
+      "loss": 1.441,
+      "step": 151630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3989580236230366e-06,
+      "loss": 1.3188,
+      "step": 151635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3983903945107986e-06,
+      "loss": 1.2525,
+      "step": 151640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3978228030972984e-06,
+      "loss": 1.1852,
+      "step": 151645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3972552493857835e-06,
+      "loss": 1.1732,
+      "step": 151650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.396687733379489e-06,
+      "loss": 1.376,
+      "step": 151655
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3961202550816598e-06,
+      "loss": 1.2852,
+      "step": 151660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3955528144955387e-06,
+      "loss": 1.1412,
+      "step": 151665
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3949854116243643e-06,
+      "loss": 1.6595,
+      "step": 151670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.394418046471372e-06,
+      "loss": 1.0721,
+      "step": 151675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.393850719039804e-06,
+      "loss": 1.2687,
+      "step": 151680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3932834293329065e-06,
+      "loss": 1.2339,
+      "step": 151685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.392716177353913e-06,
+      "loss": 1.1163,
+      "step": 151690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.392148963106061e-06,
+      "loss": 1.1363,
+      "step": 151695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3915817865925928e-06,
+      "loss": 1.3504,
+      "step": 151700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3910146478167504e-06,
+      "loss": 1.6929,
+      "step": 151705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3904475467817653e-06,
+      "loss": 1.1897,
+      "step": 151710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3898804834908827e-06,
+      "loss": 1.2435,
+      "step": 151715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3893134579473385e-06,
+      "loss": 1.3465,
+      "step": 151720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3887464701543625e-06,
+      "loss": 1.1658,
+      "step": 151725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3881795201152078e-06,
+      "loss": 1.0861,
+      "step": 151730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.387612607833104e-06,
+      "loss": 1.5609,
+      "step": 151735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.387045733311285e-06,
+      "loss": 1.2902,
+      "step": 151740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.386478896552995e-06,
+      "loss": 1.2582,
+      "step": 151745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3859120975614636e-06,
+      "loss": 1.0468,
+      "step": 151750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3853453363399356e-06,
+      "loss": 1.1808,
+      "step": 151755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3847786128916403e-06,
+      "loss": 1.2783,
+      "step": 151760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.38421192721982e-06,
+      "loss": 1.3569,
+      "step": 151765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.383645279327704e-06,
+      "loss": 1.2416,
+      "step": 151770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3830786692185335e-06,
+      "loss": 1.0402,
+      "step": 151775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.382512096895545e-06,
+      "loss": 1.0588,
+      "step": 151780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3819455623619724e-06,
+      "loss": 1.0912,
+      "step": 151785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3813790656210464e-06,
+      "loss": 1.4366,
+      "step": 151790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.380812606676006e-06,
+      "loss": 2.0726,
+      "step": 151795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3802461855300893e-06,
+      "loss": 1.1933,
+      "step": 151800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.379679802186523e-06,
+      "loss": 1.1532,
+      "step": 151805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3791134566485508e-06,
+      "loss": 1.1894,
+      "step": 151810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3785471489193965e-06,
+      "loss": 1.5851,
+      "step": 151815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3779808790023004e-06,
+      "loss": 1.0436,
+      "step": 151820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3774146469004985e-06,
+      "loss": 1.5495,
+      "step": 151825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3768484526172206e-06,
+      "loss": 1.2347,
+      "step": 151830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3762822961557007e-06,
+      "loss": 1.4347,
+      "step": 151835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3757161775191684e-06,
+      "loss": 1.2355,
+      "step": 151840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3751500967108586e-06,
+      "loss": 1.1186,
+      "step": 151845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.374584053734009e-06,
+      "loss": 1.209,
+      "step": 151850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.374018048591845e-06,
+      "loss": 1.2812,
+      "step": 151855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.373452081287604e-06,
+      "loss": 1.5611,
+      "step": 151860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3728861518245127e-06,
+      "loss": 1.0556,
+      "step": 151865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.372320260205807e-06,
+      "loss": 1.9508,
+      "step": 151870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.37175440643472e-06,
+      "loss": 1.2476,
+      "step": 151875
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3711885905144793e-06,
+      "loss": 1.244,
+      "step": 151880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3706228124483143e-06,
+      "loss": 1.13,
+      "step": 151885
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3700570722394586e-06,
+      "loss": 1.133,
+      "step": 151890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.369491369891147e-06,
+      "loss": 1.285,
+      "step": 151895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.368925705406606e-06,
+      "loss": 1.1149,
+      "step": 151900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.368360078789061e-06,
+      "loss": 1.5484,
+      "step": 151905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.367794490041748e-06,
+      "loss": 1.1825,
+      "step": 151910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.367228939167899e-06,
+      "loss": 0.9768,
+      "step": 151915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3666634261707364e-06,
+      "loss": 1.233,
+      "step": 151920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3660979510534975e-06,
+      "loss": 1.0732,
+      "step": 151925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.365532513819406e-06,
+      "loss": 1.5956,
+      "step": 151930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3649671144716867e-06,
+      "loss": 1.0606,
+      "step": 151935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.36440175301358e-06,
+      "loss": 1.5307,
+      "step": 151940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3638364294483085e-06,
+      "loss": 1.104,
+      "step": 151945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3632711437790968e-06,
+      "loss": 1.1768,
+      "step": 151950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3627058960091795e-06,
+      "loss": 1.2946,
+      "step": 151955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3621406861417784e-06,
+      "loss": 1.0969,
+      "step": 151960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.361575514180124e-06,
+      "loss": 1.3695,
+      "step": 151965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3610103801274476e-06,
+      "loss": 1.0532,
+      "step": 151970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3604452839869726e-06,
+      "loss": 0.9746,
+      "step": 151975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.359880225761922e-06,
+      "loss": 1.0285,
+      "step": 151980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3593152054555277e-06,
+      "loss": 1.5524,
+      "step": 151985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.358750223071019e-06,
+      "loss": 1.1675,
+      "step": 151990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3581852786116176e-06,
+      "loss": 1.2788,
+      "step": 151995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.357620372080548e-06,
+      "loss": 1.3438,
+      "step": 152000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.357055503481038e-06,
+      "loss": 1.2215,
+      "step": 152005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.356490672816317e-06,
+      "loss": 1.2327,
+      "step": 152010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3559258800896057e-06,
+      "loss": 1.2494,
+      "step": 152015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.355361125304133e-06,
+      "loss": 1.2717,
+      "step": 152020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.354796408463122e-06,
+      "loss": 1.2816,
+      "step": 152025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3542317295697925e-06,
+      "loss": 0.935,
+      "step": 152030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3536670886273793e-06,
+      "loss": 1.2214,
+      "step": 152035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3531024856391026e-06,
+      "loss": 1.1794,
+      "step": 152040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3525379206081853e-06,
+      "loss": 1.3294,
+      "step": 152045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3519733935378496e-06,
+      "loss": 1.1609,
+      "step": 152050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.35140890443132e-06,
+      "loss": 1.2987,
+      "step": 152055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.350844453291826e-06,
+      "loss": 1.013,
+      "step": 152060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.350280040122582e-06,
+      "loss": 1.5022,
+      "step": 152065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3497156649268204e-06,
+      "loss": 1.4534,
+      "step": 152070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3491513277077547e-06,
+      "loss": 1.3659,
+      "step": 152075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3485870284686116e-06,
+      "loss": 1.3393,
+      "step": 152080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3480227672126185e-06,
+      "loss": 1.348,
+      "step": 152085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.347458543942993e-06,
+      "loss": 1.1956,
+      "step": 152090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3468943586629542e-06,
+      "loss": 1.3447,
+      "step": 152095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.346330211375727e-06,
+      "loss": 1.3253,
+      "step": 152100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3457661020845354e-06,
+      "loss": 1.0955,
+      "step": 152105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.345202030792596e-06,
+      "loss": 1.7965,
+      "step": 152110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.344637997503134e-06,
+      "loss": 1.1032,
+      "step": 152115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3440740022193674e-06,
+      "loss": 1.3038,
+      "step": 152120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.34351004494452e-06,
+      "loss": 1.0743,
+      "step": 152125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3429461256818075e-06,
+      "loss": 1.446,
+      "step": 152130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3423822444344567e-06,
+      "loss": 1.295,
+      "step": 152135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3418184012056842e-06,
+      "loss": 0.9206,
+      "step": 152140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3412545959987063e-06,
+      "loss": 0.9629,
+      "step": 152145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3406908288167462e-06,
+      "loss": 1.0251,
+      "step": 152150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.340127099663026e-06,
+      "loss": 1.76,
+      "step": 152155
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.339563408540759e-06,
+      "loss": 1.0745,
+      "step": 152160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3389997554531717e-06,
+      "loss": 1.4242,
+      "step": 152165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.338436140403474e-06,
+      "loss": 1.2521,
+      "step": 152170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3378725633948884e-06,
+      "loss": 1.1798,
+      "step": 152175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3373090244306382e-06,
+      "loss": 0.9904,
+      "step": 152180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3367455235139358e-06,
+      "loss": 1.3122,
+      "step": 152185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3361820606479976e-06,
+      "loss": 1.0241,
+      "step": 152190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.335618635836044e-06,
+      "loss": 1.1132,
+      "step": 152195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.335055249081296e-06,
+      "loss": 1.122,
+      "step": 152200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.334491900386967e-06,
+      "loss": 1.0508,
+      "step": 152205
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.333928589756271e-06,
+      "loss": 1.269,
+      "step": 152210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3333653171924285e-06,
+      "loss": 1.0583,
+      "step": 152215
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3328020826986584e-06,
+      "loss": 1.2387,
+      "step": 152220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.332238886278172e-06,
+      "loss": 1.2592,
+      "step": 152225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3316757279341905e-06,
+      "loss": 1.314,
+      "step": 152230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3311126076699263e-06,
+      "loss": 1.2354,
+      "step": 152235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3305495254885945e-06,
+      "loss": 1.3546,
+      "step": 152240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3299864813934102e-06,
+      "loss": 1.1457,
+      "step": 152245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3294234753875955e-06,
+      "loss": 1.3836,
+      "step": 152250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3288605074743562e-06,
+      "loss": 1.6578,
+      "step": 152255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3282975776569147e-06,
+      "loss": 1.1662,
+      "step": 152260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3277346859384797e-06,
+      "loss": 1.4636,
+      "step": 152265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3271718323222723e-06,
+      "loss": 1.4157,
+      "step": 152270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3266090168114984e-06,
+      "loss": 1.408,
+      "step": 152275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3260462394093796e-06,
+      "loss": 1.4155,
+      "step": 152280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3254835001191234e-06,
+      "loss": 1.104,
+      "step": 152285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3249207989439458e-06,
+      "loss": 1.3249,
+      "step": 152290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3243581358870647e-06,
+      "loss": 1.3446,
+      "step": 152295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3237955109516884e-06,
+      "loss": 1.2174,
+      "step": 152300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.323232924141028e-06,
+      "loss": 1.0356,
+      "step": 152305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.322670375458299e-06,
+      "loss": 1.1146,
+      "step": 152310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3221078649067162e-06,
+      "loss": 0.8921,
+      "step": 152315
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3215453924894857e-06,
+      "loss": 1.0988,
+      "step": 152320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3209829582098264e-06,
+      "loss": 1.0442,
+      "step": 152325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3204205620709473e-06,
+      "loss": 1.3246,
+      "step": 152330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3198582040760553e-06,
+      "loss": 0.9855,
+      "step": 152335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.319295884228366e-06,
+      "loss": 1.0428,
+      "step": 152340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3187336025310945e-06,
+      "loss": 1.4446,
+      "step": 152345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3181713589874476e-06,
+      "loss": 1.05,
+      "step": 152350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3176091536006318e-06,
+      "loss": 1.1135,
+      "step": 152355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.317046986373863e-06,
+      "loss": 1.2954,
+      "step": 152360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3164848573103524e-06,
+      "loss": 0.9822,
+      "step": 152365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3159227664133064e-06,
+      "loss": 1.1919,
+      "step": 152370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.315360713685939e-06,
+      "loss": 1.5437,
+      "step": 152375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.314798699131453e-06,
+      "loss": 1.1573,
+      "step": 152380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3142367227530627e-06,
+      "loss": 0.9806,
+      "step": 152385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.313674784553981e-06,
+      "loss": 1.3818,
+      "step": 152390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3131128845374105e-06,
+      "loss": 1.2765,
+      "step": 152395
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3125510227065604e-06,
+      "loss": 1.2256,
+      "step": 152400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.31198919906464e-06,
+      "loss": 1.1573,
+      "step": 152405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.311427413614863e-06,
+      "loss": 1.0593,
+      "step": 152410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3108656663604313e-06,
+      "loss": 1.247,
+      "step": 152415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3103039573045513e-06,
+      "loss": 1.0579,
+      "step": 152420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.309742286450438e-06,
+      "loss": 1.2524,
+      "step": 152425
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3091806538012907e-06,
+      "loss": 1.2918,
+      "step": 152430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.308619059360321e-06,
+      "loss": 1.45,
+      "step": 152435
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.308057503130738e-06,
+      "loss": 0.9898,
+      "step": 152440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.307495985115746e-06,
+      "loss": 1.2289,
+      "step": 152445
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3069345053185485e-06,
+      "loss": 0.9721,
+      "step": 152450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3063730637423554e-06,
+      "loss": 1.3711,
+      "step": 152455
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3058116603903743e-06,
+      "loss": 1.3509,
+      "step": 152460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.305250295265807e-06,
+      "loss": 0.9517,
+      "step": 152465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.304688968371864e-06,
+      "loss": 1.0453,
+      "step": 152470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.304127679711745e-06,
+      "loss": 1.2903,
+      "step": 152475
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3035664292886583e-06,
+      "loss": 1.1046,
+      "step": 152480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3030052171058124e-06,
+      "loss": 0.9291,
+      "step": 152485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3024440431664082e-06,
+      "loss": 1.1815,
+      "step": 152490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.301882907473648e-06,
+      "loss": 1.0807,
+      "step": 152495
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.30132181003074e-06,
+      "loss": 1.1695,
+      "step": 152500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.300760750840889e-06,
+      "loss": 1.3212,
+      "step": 152505
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.3001997299072976e-06,
+      "loss": 1.3185,
+      "step": 152510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.299638747233167e-06,
+      "loss": 1.1105,
+      "step": 152515
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.299077802821702e-06,
+      "loss": 1.5799,
+      "step": 152520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2985168966761103e-06,
+      "loss": 1.7582,
+      "step": 152525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2979560287995883e-06,
+      "loss": 0.9523,
+      "step": 152530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2973951991953445e-06,
+      "loss": 1.3045,
+      "step": 152535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2968344078665803e-06,
+      "loss": 1.2102,
+      "step": 152540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.296273654816493e-06,
+      "loss": 1.09,
+      "step": 152545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.295712940048289e-06,
+      "loss": 1.7663,
+      "step": 152550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2951522635651734e-06,
+      "loss": 1.116,
+      "step": 152555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2945916253703446e-06,
+      "loss": 1.0684,
+      "step": 152560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.294031025467002e-06,
+      "loss": 1.2081,
+      "step": 152565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.293470463858347e-06,
+      "loss": 1.3879,
+      "step": 152570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2929099405475872e-06,
+      "loss": 1.2924,
+      "step": 152575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.292349455537915e-06,
+      "loss": 1.3437,
+      "step": 152580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.29178900883254e-06,
+      "loss": 1.2174,
+      "step": 152585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2912286004346528e-06,
+      "loss": 1.4175,
+      "step": 152590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.290668230347459e-06,
+      "loss": 1.4071,
+      "step": 152595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.290107898574162e-06,
+      "loss": 1.1166,
+      "step": 152600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2895476051179576e-06,
+      "loss": 1.4302,
+      "step": 152605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2889873499820423e-06,
+      "loss": 1.5124,
+      "step": 152610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2884271331696194e-06,
+      "loss": 0.9931,
+      "step": 152615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2878669546838894e-06,
+      "loss": 1.2105,
+      "step": 152620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2873068145280473e-06,
+      "loss": 1.0596,
+      "step": 152625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2867467127052954e-06,
+      "loss": 1.2283,
+      "step": 152630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.286186649218832e-06,
+      "loss": 1.1089,
+      "step": 152635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2856266240718504e-06,
+      "loss": 1.2115,
+      "step": 152640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.285066637267551e-06,
+      "loss": 1.3146,
+      "step": 152645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.284506688809137e-06,
+      "loss": 1.2042,
+      "step": 152650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2839467786998013e-06,
+      "loss": 1.1848,
+      "step": 152655
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.283386906942739e-06,
+      "loss": 1.1078,
+      "step": 152660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2828270735411494e-06,
+      "loss": 1.2692,
+      "step": 152665
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2822672784982332e-06,
+      "loss": 1.2703,
+      "step": 152670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.281707521817182e-06,
+      "loss": 1.1944,
+      "step": 152675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2811478035011967e-06,
+      "loss": 1.413,
+      "step": 152680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2805881235534676e-06,
+      "loss": 1.088,
+      "step": 152685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2800284819771955e-06,
+      "loss": 1.2061,
+      "step": 152690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.279468878775577e-06,
+      "loss": 1.3611,
+      "step": 152695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.278909313951807e-06,
+      "loss": 1.0169,
+      "step": 152700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.278349787509075e-06,
+      "loss": 1.0634,
+      "step": 152705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.277790299450583e-06,
+      "loss": 1.3653,
+      "step": 152710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.277230849779526e-06,
+      "loss": 1.2807,
+      "step": 152715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2766714384990962e-06,
+      "loss": 1.3462,
+      "step": 152720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2761120656124856e-06,
+      "loss": 1.336,
+      "step": 152725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.275552731122896e-06,
+      "loss": 1.2224,
+      "step": 152730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2749934350335133e-06,
+      "loss": 1.1184,
+      "step": 152735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.274434177347534e-06,
+      "loss": 1.0881,
+      "step": 152740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.273874958068157e-06,
+      "loss": 1.1083,
+      "step": 152745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2733157771985725e-06,
+      "loss": 1.2389,
+      "step": 152750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.272756634741968e-06,
+      "loss": 1.2562,
+      "step": 152755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.272197530701543e-06,
+      "loss": 1.2707,
+      "step": 152760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2716384650804912e-06,
+      "loss": 1.2288,
+      "step": 152765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2710794378819986e-06,
+      "loss": 1.5012,
+      "step": 152770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.270520449109266e-06,
+      "loss": 1.5635,
+      "step": 152775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.269961498765477e-06,
+      "loss": 1.2868,
+      "step": 152780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2694025868538315e-06,
+      "loss": 1.1527,
+      "step": 152785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.268843713377514e-06,
+      "loss": 1.0325,
+      "step": 152790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2682848783397226e-06,
+      "loss": 1.193,
+      "step": 152795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.267726081743641e-06,
+      "loss": 1.6685,
+      "step": 152800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2671673235924652e-06,
+      "loss": 1.309,
+      "step": 152805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2666086038893886e-06,
+      "loss": 1.4474,
+      "step": 152810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2660499226375987e-06,
+      "loss": 1.3387,
+      "step": 152815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2654912798402814e-06,
+      "loss": 0.9507,
+      "step": 152820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.264932675500635e-06,
+      "loss": 1.2061,
+      "step": 152825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2643741096218416e-06,
+      "loss": 1.2255,
+      "step": 152830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.263815582207095e-06,
+      "loss": 1.7915,
+      "step": 152835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2632570932595888e-06,
+      "loss": 1.5066,
+      "step": 152840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.262698642782507e-06,
+      "loss": 1.2831,
+      "step": 152845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2621402307790362e-06,
+      "loss": 1.8103,
+      "step": 152850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2615818572523696e-06,
+      "loss": 1.3783,
+      "step": 152855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2610235222056975e-06,
+      "loss": 1.0329,
+      "step": 152860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2604652256422055e-06,
+      "loss": 1.2752,
+      "step": 152865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2599069675650784e-06,
+      "loss": 0.9786,
+      "step": 152870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.259348747977509e-06,
+      "loss": 1.7154,
+      "step": 152875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2587905668826857e-06,
+      "loss": 1.0562,
+      "step": 152880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2582324242837915e-06,
+      "loss": 1.42,
+      "step": 152885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.257674320184019e-06,
+      "loss": 1.5963,
+      "step": 152890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.25711625458655e-06,
+      "loss": 1.0115,
+      "step": 152895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.256558227494573e-06,
+      "loss": 1.0436,
+      "step": 152900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2560002389112788e-06,
+      "loss": 1.2507,
+      "step": 152905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.255442288839851e-06,
+      "loss": 1.6241,
+      "step": 152910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2548843772834714e-06,
+      "loss": 1.4748,
+      "step": 152915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2543265042453307e-06,
+      "loss": 1.0019,
+      "step": 152920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2537686697286163e-06,
+      "loss": 0.979,
+      "step": 152925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.253210873736512e-06,
+      "loss": 1.2752,
+      "step": 152930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2526531162721987e-06,
+      "loss": 1.1584,
+      "step": 152935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.252095397338869e-06,
+      "loss": 1.3107,
+      "step": 152940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2515377169397e-06,
+      "loss": 1.2683,
+      "step": 152945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.250980075077881e-06,
+      "loss": 1.1496,
+      "step": 152950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.250422471756598e-06,
+      "loss": 1.1636,
+      "step": 152955
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2498649069790333e-06,
+      "loss": 0.9695,
+      "step": 152960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2493073807483676e-06,
+      "loss": 1.3029,
+      "step": 152965
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2487498930677876e-06,
+      "loss": 1.1604,
+      "step": 152970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.24819244394048e-06,
+      "loss": 1.197,
+      "step": 152975
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2476350333696214e-06,
+      "loss": 1.4843,
+      "step": 152980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2470776613584033e-06,
+      "loss": 1.2084,
+      "step": 152985
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2465203279099987e-06,
+      "loss": 1.6341,
+      "step": 152990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2459630330276005e-06,
+      "loss": 1.5448,
+      "step": 152995
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.245405776714381e-06,
+      "loss": 1.4159,
+      "step": 153000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2448485589735322e-06,
+      "loss": 1.4803,
+      "step": 153005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.244291379808231e-06,
+      "loss": 1.4177,
+      "step": 153010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.243734239221652e-06,
+      "loss": 1.1213,
+      "step": 153015
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2431771372169917e-06,
+      "loss": 1.2942,
+      "step": 153020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2426200737974243e-06,
+      "loss": 1.2206,
+      "step": 153025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2420630489661275e-06,
+      "loss": 1.3355,
+      "step": 153030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2415060627262884e-06,
+      "loss": 1.1853,
+      "step": 153035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.240949115081081e-06,
+      "loss": 1.3724,
+      "step": 153040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2403922060336914e-06,
+      "loss": 1.3332,
+      "step": 153045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.239835335587299e-06,
+      "loss": 1.2197,
+      "step": 153050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2392785037450846e-06,
+      "loss": 1.3969,
+      "step": 153055
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.238721710510222e-06,
+      "loss": 1.6457,
+      "step": 153060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.238164955885895e-06,
+      "loss": 1.095,
+      "step": 153065
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2376082398752862e-06,
+      "loss": 1.027,
+      "step": 153070
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2370515624815714e-06,
+      "loss": 1.3421,
+      "step": 153075
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.236494923707927e-06,
+      "loss": 1.1858,
+      "step": 153080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2359383235575336e-06,
+      "loss": 1.0777,
+      "step": 153085
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2353817620335748e-06,
+      "loss": 1.1033,
+      "step": 153090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.234825239139221e-06,
+      "loss": 1.3373,
+      "step": 153095
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.234268754877656e-06,
+      "loss": 0.9543,
+      "step": 153100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.233712309252053e-06,
+      "loss": 1.4634,
+      "step": 153105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.233155902265591e-06,
+      "loss": 1.4289,
+      "step": 153110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.232599533921452e-06,
+      "loss": 1.0923,
+      "step": 153115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.23204320422281e-06,
+      "loss": 1.2483,
+      "step": 153120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.231486913172839e-06,
+      "loss": 1.2248,
+      "step": 153125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2309306607747206e-06,
+      "loss": 1.6615,
+      "step": 153130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2303744470316246e-06,
+      "loss": 1.3147,
+      "step": 153135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2298182719467362e-06,
+      "loss": 1.3214,
+      "step": 153140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2292621355232234e-06,
+      "loss": 1.2375,
+      "step": 153145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2287060377642675e-06,
+      "loss": 1.3269,
+      "step": 153150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2281499786730396e-06,
+      "loss": 1.1884,
+      "step": 153155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2275939582527183e-06,
+      "loss": 1.1104,
+      "step": 153160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2270379765064807e-06,
+      "loss": 1.2734,
+      "step": 153165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.226482033437499e-06,
+      "loss": 1.5536,
+      "step": 153170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.225926129048945e-06,
+      "loss": 1.4792,
+      "step": 153175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2253702633439965e-06,
+      "loss": 1.4687,
+      "step": 153180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2248144363258306e-06,
+      "loss": 1.2427,
+      "step": 153185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2242586479976145e-06,
+      "loss": 1.1473,
+      "step": 153190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2237028983625295e-06,
+      "loss": 1.2438,
+      "step": 153195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.223147187423743e-06,
+      "loss": 1.0483,
+      "step": 153200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.22259151518443e-06,
+      "loss": 1.1289,
+      "step": 153205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2220358816477683e-06,
+      "loss": 1.2068,
+      "step": 153210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.221480286816927e-06,
+      "loss": 1.1772,
+      "step": 153215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2209247306950795e-06,
+      "loss": 1.2431,
+      "step": 153220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2203692132853914e-06,
+      "loss": 1.2697,
+      "step": 153225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2198137345910487e-06,
+      "loss": 1.1367,
+      "step": 153230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2192582946152152e-06,
+      "loss": 1.4388,
+      "step": 153235
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.218702893361061e-06,
+      "loss": 1.1895,
+      "step": 153240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2181475308317635e-06,
+      "loss": 1.5488,
+      "step": 153245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2175922070304886e-06,
+      "loss": 1.3567,
+      "step": 153250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2170369219604103e-06,
+      "loss": 0.9697,
+      "step": 153255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2164816756247018e-06,
+      "loss": 1.3715,
+      "step": 153260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2159264680265323e-06,
+      "loss": 1.3364,
+      "step": 153265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2153712991690677e-06,
+      "loss": 1.3271,
+      "step": 153270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2148161690554826e-06,
+      "loss": 1.1651,
+      "step": 153275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2142610776889495e-06,
+      "loss": 1.5075,
+      "step": 153280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2137060250726347e-06,
+      "loss": 1.1288,
+      "step": 153285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2131510112097066e-06,
+      "loss": 1.2856,
+      "step": 153290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2125960361033346e-06,
+      "loss": 1.3744,
+      "step": 153295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2120410997566952e-06,
+      "loss": 1.2847,
+      "step": 153300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2114862021729474e-06,
+      "loss": 1.3481,
+      "step": 153305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2109313433552693e-06,
+      "loss": 1.4026,
+      "step": 153310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.210376523306824e-06,
+      "loss": 1.1656,
+      "step": 153315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.209821742030773e-06,
+      "loss": 1.3722,
+      "step": 153320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2092669995303004e-06,
+      "loss": 1.0828,
+      "step": 153325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2087122958085646e-06,
+      "loss": 1.6167,
+      "step": 153330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2081576308687313e-06,
+      "loss": 1.2476,
+      "step": 153335
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2076030047139737e-06,
+      "loss": 1.5215,
+      "step": 153340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2070484173474535e-06,
+      "loss": 1.1586,
+      "step": 153345
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2064938687723402e-06,
+      "loss": 1.2115,
+      "step": 153350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2059393589918044e-06,
+      "loss": 1.0907,
+      "step": 153355
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.205384888009008e-06,
+      "loss": 1.5068,
+      "step": 153360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.204830455827116e-06,
+      "loss": 1.1958,
+      "step": 153365
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2042760624492974e-06,
+      "loss": 1.4501,
+      "step": 153370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2037217078787196e-06,
+      "loss": 1.2048,
+      "step": 153375
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.203167392118547e-06,
+      "loss": 1.0712,
+      "step": 153380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2026131151719408e-06,
+      "loss": 1.1382,
+      "step": 153385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.20205887704207e-06,
+      "loss": 1.2653,
+      "step": 153390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2015046777321025e-06,
+      "loss": 1.4585,
+      "step": 153395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.200950517245196e-06,
+      "loss": 1.5557,
+      "step": 153400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.200396395584523e-06,
+      "loss": 1.219,
+      "step": 153405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.199842312753243e-06,
+      "loss": 1.1439,
+      "step": 153410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1992882687545156e-06,
+      "loss": 1.0038,
+      "step": 153415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.198734263591515e-06,
+      "loss": 1.1012,
+      "step": 153420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1981802972674012e-06,
+      "loss": 1.478,
+      "step": 153425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.197626369785335e-06,
+      "loss": 1.3443,
+      "step": 153430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.197072481148479e-06,
+      "loss": 1.3732,
+      "step": 153435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.196518631359997e-06,
+      "loss": 1.176,
+      "step": 153440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1959648204230564e-06,
+      "loss": 0.9775,
+      "step": 153445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.195411048340813e-06,
+      "loss": 1.4405,
+      "step": 153450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1948573151164352e-06,
+      "loss": 1.2298,
+      "step": 153455
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1943036207530797e-06,
+      "loss": 1.2605,
+      "step": 153460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1937499652539116e-06,
+      "loss": 1.0493,
+      "step": 153465
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.193196348622094e-06,
+      "loss": 1.244,
+      "step": 153470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1926427708607867e-06,
+      "loss": 1.1485,
+      "step": 153475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1920892319731476e-06,
+      "loss": 1.3279,
+      "step": 153480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1915357319623395e-06,
+      "loss": 1.2553,
+      "step": 153485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1909822708315287e-06,
+      "loss": 1.2071,
+      "step": 153490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1904288485838687e-06,
+      "loss": 1.1901,
+      "step": 153495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.189875465222524e-06,
+      "loss": 1.1784,
+      "step": 153500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1893221207506506e-06,
+      "loss": 1.0482,
+      "step": 153505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.188768815171415e-06,
+      "loss": 1.1648,
+      "step": 153510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1882155484879696e-06,
+      "loss": 1.0629,
+      "step": 153515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1876623207034796e-06,
+      "loss": 1.4945,
+      "step": 153520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1871091318211024e-06,
+      "loss": 1.516,
+      "step": 153525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.186555981843993e-06,
+      "loss": 1.0638,
+      "step": 153530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.186002870775313e-06,
+      "loss": 0.8903,
+      "step": 153535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1854497986182252e-06,
+      "loss": 1.1698,
+      "step": 153540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.184896765375882e-06,
+      "loss": 1.2574,
+      "step": 153545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1843437710514456e-06,
+      "loss": 1.2616,
+      "step": 153550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.18379081564807e-06,
+      "loss": 1.297,
+      "step": 153555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.183237899168914e-06,
+      "loss": 1.575,
+      "step": 153560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.18268502161714e-06,
+      "loss": 1.1525,
+      "step": 153565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1821321829959007e-06,
+      "loss": 1.015,
+      "step": 153570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.181579383308351e-06,
+      "loss": 0.9613,
+      "step": 153575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.181026622557649e-06,
+      "loss": 1.1726,
+      "step": 153580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1804739007469565e-06,
+      "loss": 1.2938,
+      "step": 153585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.179921217879426e-06,
+      "loss": 1.3025,
+      "step": 153590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1793685739582105e-06,
+      "loss": 1.336,
+      "step": 153595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1788159689864697e-06,
+      "loss": 1.3617,
+      "step": 153600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1782634029673596e-06,
+      "loss": 1.421,
+      "step": 153605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.177710875904033e-06,
+      "loss": 1.1856,
+      "step": 153610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1771583877996505e-06,
+      "loss": 1.184,
+      "step": 153615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1766059386573623e-06,
+      "loss": 1.0756,
+      "step": 153620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.176053528480317e-06,
+      "loss": 1.3137,
+      "step": 153625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.175501157271683e-06,
+      "loss": 1.3951,
+      "step": 153630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1749488250346096e-06,
+      "loss": 1.0514,
+      "step": 153635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.174396531772245e-06,
+      "loss": 1.2651,
+      "step": 153640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.17384427748775e-06,
+      "loss": 1.1331,
+      "step": 153645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1732920621842733e-06,
+      "loss": 1.1681,
+      "step": 153650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1727398858649725e-06,
+      "loss": 1.2471,
+      "step": 153655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1721877485329975e-06,
+      "loss": 1.243,
+      "step": 153660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.171635650191506e-06,
+      "loss": 1.4153,
+      "step": 153665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.171083590843643e-06,
+      "loss": 1.1905,
+      "step": 153670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1705315704925665e-06,
+      "loss": 1.1974,
+      "step": 153675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.169979589141431e-06,
+      "loss": 1.0202,
+      "step": 153680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.169427646793386e-06,
+      "loss": 1.3799,
+      "step": 153685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.168875743451579e-06,
+      "loss": 1.2344,
+      "step": 153690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1683238791191662e-06,
+      "loss": 1.2485,
+      "step": 153695
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.167772053799302e-06,
+      "loss": 1.5513,
+      "step": 153700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1672202674951304e-06,
+      "loss": 1.5359,
+      "step": 153705
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.166668520209809e-06,
+      "loss": 1.1896,
+      "step": 153710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.166116811946486e-06,
+      "loss": 1.2942,
+      "step": 153715
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.165565142708308e-06,
+      "loss": 1.5377,
+      "step": 153720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1650135124984294e-06,
+      "loss": 1.1231,
+      "step": 153725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1644619213200033e-06,
+      "loss": 1.0391,
+      "step": 153730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.163910369176176e-06,
+      "loss": 1.176,
+      "step": 153735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1633588560700936e-06,
+      "loss": 1.2397,
+      "step": 153740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1628073820049097e-06,
+      "loss": 1.3187,
+      "step": 153745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1622559469837765e-06,
+      "loss": 0.9939,
+      "step": 153750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1617045510098356e-06,
+      "loss": 1.1925,
+      "step": 153755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1611531940862438e-06,
+      "loss": 1.3106,
+      "step": 153760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1606018762161415e-06,
+      "loss": 1.4271,
+      "step": 153765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.160050597402682e-06,
+      "loss": 1.1839,
+      "step": 153770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1594993576490154e-06,
+      "loss": 1.0944,
+      "step": 153775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1589481569582858e-06,
+      "loss": 1.0421,
+      "step": 153780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.158396995333639e-06,
+      "loss": 1.1512,
+      "step": 153785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1578458727782247e-06,
+      "loss": 1.2946,
+      "step": 153790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1572947892951942e-06,
+      "loss": 1.269,
+      "step": 153795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1567437448876914e-06,
+      "loss": 1.4273,
+      "step": 153800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1561927395588588e-06,
+      "loss": 1.1536,
+      "step": 153805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.155641773311846e-06,
+      "loss": 1.169,
+      "step": 153810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1550908461498043e-06,
+      "loss": 1.1155,
+      "step": 153815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.154539958075871e-06,
+      "loss": 0.987,
+      "step": 153820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1539891090932007e-06,
+      "loss": 1.1406,
+      "step": 153825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.153438299204934e-06,
+      "loss": 1.3816,
+      "step": 153830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1528875284142135e-06,
+      "loss": 1.1731,
+      "step": 153835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.152336796724189e-06,
+      "loss": 1.0768,
+      "step": 153840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.151786104138007e-06,
+      "loss": 1.3545,
+      "step": 153845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.151235450658806e-06,
+      "loss": 1.4114,
+      "step": 153850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1506848362897393e-06,
+      "loss": 1.0614,
+      "step": 153855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1501342610339412e-06,
+      "loss": 1.1024,
+      "step": 153860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.149583724894564e-06,
+      "loss": 1.0245,
+      "step": 153865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1490332278747447e-06,
+      "loss": 0.8816,
+      "step": 153870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1484827699776345e-06,
+      "loss": 1.4344,
+      "step": 153875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.14793235120637e-06,
+      "loss": 1.3528,
+      "step": 153880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1473819715640965e-06,
+      "loss": 1.1163,
+      "step": 153885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1468316310539604e-06,
+      "loss": 1.1213,
+      "step": 153890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.146281329679103e-06,
+      "loss": 1.0901,
+      "step": 153895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.145731067442661e-06,
+      "loss": 0.9026,
+      "step": 153900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1451808443477815e-06,
+      "loss": 1.3956,
+      "step": 153905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1446306603976097e-06,
+      "loss": 1.2712,
+      "step": 153910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.14408051559528e-06,
+      "loss": 0.9695,
+      "step": 153915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.143530409943942e-06,
+      "loss": 1.4591,
+      "step": 153920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1429803434467334e-06,
+      "loss": 1.2885,
+      "step": 153925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.142430316106788e-06,
+      "loss": 1.193,
+      "step": 153930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.141880327927261e-06,
+      "loss": 1.075,
+      "step": 153935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.141330378911286e-06,
+      "loss": 1.309,
+      "step": 153940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1407804690620035e-06,
+      "loss": 1.2257,
+      "step": 153945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.14023059838255e-06,
+      "loss": 1.0871,
+      "step": 153950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.139680766876069e-06,
+      "loss": 1.4246,
+      "step": 153955
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1391309745457044e-06,
+      "loss": 0.9791,
+      "step": 153960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.138581221394589e-06,
+      "loss": 1.1772,
+      "step": 153965
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.138031507425868e-06,
+      "loss": 1.2724,
+      "step": 153970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.137481832642675e-06,
+      "loss": 1.1665,
+      "step": 153975
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1369321970481515e-06,
+      "loss": 1.7344,
+      "step": 153980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.13638260064544e-06,
+      "loss": 1.2522,
+      "step": 153985
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.135833043437676e-06,
+      "loss": 1.358,
+      "step": 153990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1352835254279923e-06,
+      "loss": 1.1015,
+      "step": 153995
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.134734046619533e-06,
+      "loss": 1.6068,
+      "step": 154000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.134184607015438e-06,
+      "loss": 1.0841,
+      "step": 154005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1336352066188414e-06,
+      "loss": 1.1238,
+      "step": 154010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1330858454328782e-06,
+      "loss": 1.0979,
+      "step": 154015
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1325365234606907e-06,
+      "loss": 0.9733,
+      "step": 154020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.131987240705411e-06,
+      "loss": 1.0885,
+      "step": 154025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.131437997170178e-06,
+      "loss": 1.1615,
+      "step": 154030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.130888792858131e-06,
+      "loss": 1.1113,
+      "step": 154035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1303396277724028e-06,
+      "loss": 1.7899,
+      "step": 154040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.129790501916128e-06,
+      "loss": 1.1258,
+      "step": 154045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1292414152924445e-06,
+      "loss": 1.0244,
+      "step": 154050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.128692367904491e-06,
+      "loss": 0.9798,
+      "step": 154055
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1281433597553967e-06,
+      "loss": 1.431,
+      "step": 154060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.127594390848304e-06,
+      "loss": 1.3827,
+      "step": 154065
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1270454611863387e-06,
+      "loss": 1.5354,
+      "step": 154070
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1264965707726415e-06,
+      "loss": 1.2206,
+      "step": 154075
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1259477196103495e-06,
+      "loss": 1.0795,
+      "step": 154080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1253989077025928e-06,
+      "loss": 1.22,
+      "step": 154085
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1248501350525016e-06,
+      "loss": 1.1023,
+      "step": 154090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1243014016632156e-06,
+      "loss": 1.2199,
+      "step": 154095
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.123752707537868e-06,
+      "loss": 1.1276,
+      "step": 154100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1232040526795926e-06,
+      "loss": 1.0935,
+      "step": 154105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.122655437091516e-06,
+      "loss": 0.9842,
+      "step": 154110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1221068607767813e-06,
+      "loss": 1.4824,
+      "step": 154115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1215583237385107e-06,
+      "loss": 1.4378,
+      "step": 154120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1210098259798416e-06,
+      "loss": 1.8272,
+      "step": 154125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1204613675039107e-06,
+      "loss": 1.2412,
+      "step": 154130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1199129483138447e-06,
+      "loss": 1.2431,
+      "step": 154135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1193645684127727e-06,
+      "loss": 1.0769,
+      "step": 154140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1188162278038302e-06,
+      "loss": 1.3534,
+      "step": 154145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1182679264901515e-06,
+      "loss": 1.3475,
+      "step": 154150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.117719664474864e-06,
+      "loss": 1.0139,
+      "step": 154155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1171714417610964e-06,
+      "loss": 1.3364,
+      "step": 154160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.116623258351982e-06,
+      "loss": 1.1519,
+      "step": 154165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.116075114250653e-06,
+      "loss": 1.3278,
+      "step": 154170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.115527009460235e-06,
+      "loss": 0.939,
+      "step": 154175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.114978943983863e-06,
+      "loss": 1.1422,
+      "step": 154180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1144309178246624e-06,
+      "loss": 1.3858,
+      "step": 154185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1138829309857643e-06,
+      "loss": 1.6941,
+      "step": 154190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.113334983470301e-06,
+      "loss": 1.2768,
+      "step": 154195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1127870752813995e-06,
+      "loss": 0.979,
+      "step": 154200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.112239206422185e-06,
+      "loss": 1.1062,
+      "step": 154205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.111691376895789e-06,
+      "loss": 1.2783,
+      "step": 154210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.111143586705343e-06,
+      "loss": 1.1705,
+      "step": 154215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.11059583585397e-06,
+      "loss": 1.0401,
+      "step": 154220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.110048124344802e-06,
+      "loss": 1.3803,
+      "step": 154225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.109500452180967e-06,
+      "loss": 1.1452,
+      "step": 154230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.108952819365586e-06,
+      "loss": 1.4661,
+      "step": 154235
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1084052259017906e-06,
+      "loss": 1.576,
+      "step": 154240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1078576717927112e-06,
+      "loss": 1.2541,
+      "step": 154245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.107310157041472e-06,
+      "loss": 1.2433,
+      "step": 154250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1067626816511965e-06,
+      "loss": 1.2267,
+      "step": 154255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1062152456250126e-06,
+      "loss": 1.1899,
+      "step": 154260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.105667848966051e-06,
+      "loss": 1.1834,
+      "step": 154265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1051204916774303e-06,
+      "loss": 1.151,
+      "step": 154270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1045731737622842e-06,
+      "loss": 1.0421,
+      "step": 154275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.10402589522373e-06,
+      "loss": 1.2782,
+      "step": 154280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1034786560648976e-06,
+      "loss": 1.2158,
+      "step": 154285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1029314562889133e-06,
+      "loss": 1.1337,
+      "step": 154290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1023842958989015e-06,
+      "loss": 1.0833,
+      "step": 154295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1018371748979846e-06,
+      "loss": 1.1784,
+      "step": 154300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.101290093289282e-06,
+      "loss": 1.1883,
+      "step": 154305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1007430510759283e-06,
+      "loss": 1.3303,
+      "step": 154310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1001960482610437e-06,
+      "loss": 1.0418,
+      "step": 154315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.099649084847748e-06,
+      "loss": 1.4295,
+      "step": 154320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.099102160839169e-06,
+      "loss": 1.0145,
+      "step": 154325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0985552762384265e-06,
+      "loss": 1.2881,
+      "step": 154330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0980084310486447e-06,
+      "loss": 1.0786,
+      "step": 154335
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0974616252729505e-06,
+      "loss": 1.3396,
+      "step": 154340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0969148589144625e-06,
+      "loss": 1.3116,
+      "step": 154345
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0963681319763006e-06,
+      "loss": 1.4243,
+      "step": 154350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0958214444615897e-06,
+      "loss": 1.1507,
+      "step": 154355
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0952747963734554e-06,
+      "loss": 0.9816,
+      "step": 154360
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0947281877150128e-06,
+      "loss": 1.5116,
+      "step": 154365
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.094181618489388e-06,
+      "loss": 1.1953,
+      "step": 154370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.093635088699699e-06,
+      "loss": 1.2489,
+      "step": 154375
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.09308859834907e-06,
+      "loss": 1.4081,
+      "step": 154380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.092542147440618e-06,
+      "loss": 1.1046,
+      "step": 154385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0919957359774676e-06,
+      "loss": 1.3275,
+      "step": 154390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.091449363962734e-06,
+      "loss": 1.266,
+      "step": 154395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0909030313995412e-06,
+      "loss": 1.2063,
+      "step": 154400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0903567382910105e-06,
+      "loss": 0.9756,
+      "step": 154405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.08981048464026e-06,
+      "loss": 1.4404,
+      "step": 154410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.089264270450405e-06,
+      "loss": 1.045,
+      "step": 154415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.088718095724571e-06,
+      "loss": 1.049,
+      "step": 154420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.08817196046587e-06,
+      "loss": 0.9905,
+      "step": 154425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.087625864677425e-06,
+      "loss": 1.2432,
+      "step": 154430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.087079808362358e-06,
+      "loss": 1.1692,
+      "step": 154435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0865337915237826e-06,
+      "loss": 1.05,
+      "step": 154440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0859878141648146e-06,
+      "loss": 1.1815,
+      "step": 154445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0854418762885752e-06,
+      "loss": 1.2261,
+      "step": 154450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0848959778981844e-06,
+      "loss": 1.1232,
+      "step": 154455
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0843501189967585e-06,
+      "loss": 1.4877,
+      "step": 154460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0838042995874085e-06,
+      "loss": 1.3406,
+      "step": 154465
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0832585196732554e-06,
+      "loss": 1.0025,
+      "step": 154470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0827127792574205e-06,
+      "loss": 1.1985,
+      "step": 154475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0821670783430125e-06,
+      "loss": 1.1001,
+      "step": 154480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.081621416933155e-06,
+      "loss": 0.9239,
+      "step": 154485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0810757950309557e-06,
+      "loss": 1.5195,
+      "step": 154490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0805302126395364e-06,
+      "loss": 1.5435,
+      "step": 154495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.079984669762014e-06,
+      "loss": 1.1691,
+      "step": 154500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0794391664015e-06,
+      "loss": 1.0976,
+      "step": 154505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0788937025611086e-06,
+      "loss": 1.0863,
+      "step": 154510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.07834827824396e-06,
+      "loss": 1.186,
+      "step": 154515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.077802893453161e-06,
+      "loss": 1.0159,
+      "step": 154520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0772575481918355e-06,
+      "loss": 0.9933,
+      "step": 154525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0767122424630893e-06,
+      "loss": 1.2043,
+      "step": 154530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.076166976270042e-06,
+      "loss": 1.0314,
+      "step": 154535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0756217496158025e-06,
+      "loss": 1.1226,
+      "step": 154540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0750765625034873e-06,
+      "loss": 1.0408,
+      "step": 154545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.074531414936214e-06,
+      "loss": 1.127,
+      "step": 154550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0739863069170896e-06,
+      "loss": 1.0599,
+      "step": 154555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.073441238449225e-06,
+      "loss": 1.1341,
+      "step": 154560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0728962095357373e-06,
+      "loss": 1.0523,
+      "step": 154565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0723512201797413e-06,
+      "loss": 1.0848,
+      "step": 154570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.071806270384342e-06,
+      "loss": 1.3287,
+      "step": 154575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0712613601526588e-06,
+      "loss": 1.1007,
+      "step": 154580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.070716489487797e-06,
+      "loss": 1.4311,
+      "step": 154585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0701716583928744e-06,
+      "loss": 1.3327,
+      "step": 154590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.069626866870995e-06,
+      "loss": 1.2859,
+      "step": 154595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.069082114925277e-06,
+      "loss": 1.108,
+      "step": 154600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0685374025588277e-06,
+      "loss": 1.1204,
+      "step": 154605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0679927297747525e-06,
+      "loss": 1.3183,
+      "step": 154610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.067448096576172e-06,
+      "loss": 1.3112,
+      "step": 154615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.066903502966193e-06,
+      "loss": 1.0998,
+      "step": 154620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0663589489479206e-06,
+      "loss": 1.1198,
+      "step": 154625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.06581443452447e-06,
+      "loss": 1.179,
+      "step": 154630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0652699596989467e-06,
+      "loss": 1.0627,
+      "step": 154635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.064725524474461e-06,
+      "loss": 1.0994,
+      "step": 154640
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0641811288541255e-06,
+      "loss": 0.9757,
+      "step": 154645
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0636367728410476e-06,
+      "loss": 0.9643,
+      "step": 154650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0630924564383303e-06,
+      "loss": 1.25,
+      "step": 154655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0625481796490863e-06,
+      "loss": 1.0589,
+      "step": 154660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.062003942476427e-06,
+      "loss": 0.9022,
+      "step": 154665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0614597449234564e-06,
+      "loss": 1.3843,
+      "step": 154670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0609155869932793e-06,
+      "loss": 1.0945,
+      "step": 154675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0603714686890065e-06,
+      "loss": 1.4641,
+      "step": 154680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0598273900137477e-06,
+      "loss": 1.3311,
+      "step": 154685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0592833509706054e-06,
+      "loss": 0.8345,
+      "step": 154690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0587393515626905e-06,
+      "loss": 1.1957,
+      "step": 154695
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.058195391793107e-06,
+      "loss": 1.2142,
+      "step": 154700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0576514716649562e-06,
+      "loss": 1.1181,
+      "step": 154705
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0571075911813553e-06,
+      "loss": 1.3501,
+      "step": 154710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.056563750345405e-06,
+      "loss": 1.2129,
+      "step": 154715
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.056019949160206e-06,
+      "loss": 1.3843,
+      "step": 154720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0554761876288706e-06,
+      "loss": 1.4271,
+      "step": 154725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0549324657544998e-06,
+      "loss": 1.0496,
+      "step": 154730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0543887835401987e-06,
+      "loss": 0.9556,
+      "step": 154735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.053845140989078e-06,
+      "loss": 1.0897,
+      "step": 154740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0533015381042362e-06,
+      "loss": 1.4981,
+      "step": 154745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.052757974888776e-06,
+      "loss": 1.6792,
+      "step": 154750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.052214451345804e-06,
+      "loss": 0.974,
+      "step": 154755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0516709674784285e-06,
+      "loss": 1.2376,
+      "step": 154760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0511275232897487e-06,
+      "loss": 1.0249,
+      "step": 154765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0505841187828645e-06,
+      "loss": 1.0342,
+      "step": 154770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0500407539608834e-06,
+      "loss": 1.4123,
+      "step": 154775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0494974288269106e-06,
+      "loss": 0.9334,
+      "step": 154780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.048954143384042e-06,
+      "loss": 1.3136,
+      "step": 154785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0484108976353877e-06,
+      "loss": 1.5084,
+      "step": 154790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0478676915840435e-06,
+      "loss": 1.3403,
+      "step": 154795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.047324525233113e-06,
+      "loss": 1.5636,
+      "step": 154800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.046781398585702e-06,
+      "loss": 1.1268,
+      "step": 154805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.04623831164491e-06,
+      "loss": 1.3801,
+      "step": 154810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.045695264413836e-06,
+      "loss": 1.2528,
+      "step": 154815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0451522568955803e-06,
+      "loss": 1.0767,
+      "step": 154820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0446092890932454e-06,
+      "loss": 1.0895,
+      "step": 154825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.044066361009936e-06,
+      "loss": 1.0922,
+      "step": 154830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.043523472648746e-06,
+      "loss": 1.3946,
+      "step": 154835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0429806240127812e-06,
+      "loss": 1.2277,
+      "step": 154840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0424378151051358e-06,
+      "loss": 0.9998,
+      "step": 154845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.0418950459289133e-06,
+      "loss": 1.0994,
+      "step": 154850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.041352316487215e-06,
+      "loss": 1.0315,
+      "step": 154855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0408096267831376e-06,
+      "loss": 1.1675,
+      "step": 154860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0402669768197767e-06,
+      "loss": 1.1836,
+      "step": 154865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.039724366600235e-06,
+      "loss": 1.4874,
+      "step": 154870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.039181796127615e-06,
+      "loss": 1.313,
+      "step": 154875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0386392654050067e-06,
+      "loss": 1.0941,
+      "step": 154880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.038096774435515e-06,
+      "loss": 0.9549,
+      "step": 154885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0375543232222326e-06,
+      "loss": 1.2703,
+      "step": 154890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.037011911768264e-06,
+      "loss": 1.7626,
+      "step": 154895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0364695400766977e-06,
+      "loss": 1.2218,
+      "step": 154900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0359272081506398e-06,
+      "loss": 1.4332,
+      "step": 154905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0353849159931823e-06,
+      "loss": 1.0901,
+      "step": 154910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.034842663607418e-06,
+      "loss": 1.1214,
+      "step": 154915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0343004509964534e-06,
+      "loss": 1.2902,
+      "step": 154920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.03375827816338e-06,
+      "loss": 1.3402,
+      "step": 154925
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.033216145111291e-06,
+      "loss": 1.0345,
+      "step": 154930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.032674051843287e-06,
+      "loss": 0.8952,
+      "step": 154935
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0321319983624585e-06,
+      "loss": 1.1558,
+      "step": 154940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0315899846719055e-06,
+      "loss": 1.2126,
+      "step": 154945
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.031048010774723e-06,
+      "loss": 0.9357,
+      "step": 154950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0305060766740058e-06,
+      "loss": 1.4089,
+      "step": 154955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.029964182372844e-06,
+      "loss": 1.2233,
+      "step": 154960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0294223278743353e-06,
+      "loss": 0.987,
+      "step": 154965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.028880513181578e-06,
+      "loss": 1.2124,
+      "step": 154970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0283387382976614e-06,
+      "loss": 1.0875,
+      "step": 154975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0277970032256774e-06,
+      "loss": 1.0619,
+      "step": 154980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.027255307968724e-06,
+      "loss": 1.0761,
+      "step": 154985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0267136525298946e-06,
+      "loss": 0.9679,
+      "step": 154990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0261720369122783e-06,
+      "loss": 1.3923,
+      "step": 154995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0256304611189736e-06,
+      "loss": 0.9222,
+      "step": 155000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0250889251530703e-06,
+      "loss": 1.4851,
+      "step": 155005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0245474290176547e-06,
+      "loss": 1.1603,
+      "step": 155010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.024005972715831e-06,
+      "loss": 1.0821,
+      "step": 155015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0234645562506846e-06,
+      "loss": 1.4371,
+      "step": 155020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0229231796253058e-06,
+      "loss": 1.2667,
+      "step": 155025
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.022381842842791e-06,
+      "loss": 1.4975,
+      "step": 155030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.021840545906225e-06,
+      "loss": 1.0366,
+      "step": 155035
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0212992888187066e-06,
+      "loss": 1.4138,
+      "step": 155040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0207580715833194e-06,
+      "loss": 1.1118,
+      "step": 155045
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0202168942031617e-06,
+      "loss": 1.2702,
+      "step": 155050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.019675756681315e-06,
+      "loss": 1.1313,
+      "step": 155055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0191346590208748e-06,
+      "loss": 0.9476,
+      "step": 155060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0185936012249327e-06,
+      "loss": 1.1724,
+      "step": 155065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0180525832965767e-06,
+      "loss": 1.0903,
+      "step": 155070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.017511605238893e-06,
+      "loss": 1.2267,
+      "step": 155075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.016970667054974e-06,
+      "loss": 1.3396,
+      "step": 155080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0164297687479094e-06,
+      "loss": 1.2082,
+      "step": 155085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0158889103207854e-06,
+      "loss": 1.2345,
+      "step": 155090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.015348091776695e-06,
+      "loss": 1.0899,
+      "step": 155095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.014807313118724e-06,
+      "loss": 1.2926,
+      "step": 155100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0142665743499565e-06,
+      "loss": 1.0558,
+      "step": 155105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.013725875473484e-06,
+      "loss": 1.1295,
+      "step": 155110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.013185216492398e-06,
+      "loss": 1.2688,
+      "step": 155115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.012644597409782e-06,
+      "loss": 1.1795,
+      "step": 155120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.012104018228721e-06,
+      "loss": 0.9617,
+      "step": 155125
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.011563478952304e-06,
+      "loss": 1.0958,
+      "step": 155130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.011022979583621e-06,
+      "loss": 1.2594,
+      "step": 155135
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.010482520125754e-06,
+      "loss": 1.1076,
+      "step": 155140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.009942100581793e-06,
+      "loss": 1.3354,
+      "step": 155145
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0094017209548186e-06,
+      "loss": 1.1571,
+      "step": 155150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0088613812479216e-06,
+      "loss": 1.2771,
+      "step": 155155
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0083210814641895e-06,
+      "loss": 1.4954,
+      "step": 155160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.007780821606704e-06,
+      "loss": 1.1469,
+      "step": 155165
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.007240601678547e-06,
+      "loss": 1.242,
+      "step": 155170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.006700421682809e-06,
+      "loss": 1.3464,
+      "step": 155175
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0061602816225742e-06,
+      "loss": 1.2066,
+      "step": 155180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0056201815009267e-06,
+      "loss": 1.0943,
+      "step": 155185
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0050801213209467e-06,
+      "loss": 1.462,
+      "step": 155190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.004540101085722e-06,
+      "loss": 1.3547,
+      "step": 155195
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0040001207983383e-06,
+      "loss": 1.142,
+      "step": 155200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0034601804618736e-06,
+      "loss": 1.5359,
+      "step": 155205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0029202800794178e-06,
+      "loss": 1.1205,
+      "step": 155210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0023804196540505e-06,
+      "loss": 1.109,
+      "step": 155215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.001840599188851e-06,
+      "loss": 1.5179,
+      "step": 155220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.001300818686906e-06,
+      "loss": 1.2835,
+      "step": 155225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.000761078151302e-06,
+      "loss": 1.1196,
+      "step": 155230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0002213775851118e-06,
+      "loss": 1.2171,
+      "step": 155235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.999681716991427e-06,
+      "loss": 1.365,
+      "step": 155240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.999142096373321e-06,
+      "loss": 1.3623,
+      "step": 155245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9986025157338815e-06,
+      "loss": 1.2946,
+      "step": 155250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.998062975076185e-06,
+      "loss": 1.3828,
+      "step": 155255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9975234744033176e-06,
+      "loss": 1.2406,
+      "step": 155260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.996984013718355e-06,
+      "loss": 1.5454,
+      "step": 155265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.996444593024379e-06,
+      "loss": 1.2449,
+      "step": 155270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9959052123244757e-06,
+      "loss": 1.1469,
+      "step": 155275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9953658716217206e-06,
+      "loss": 0.948,
+      "step": 155280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9948265709191906e-06,
+      "loss": 1.1537,
+      "step": 155285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9942873102199675e-06,
+      "loss": 1.7801,
+      "step": 155290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9937480895271366e-06,
+      "loss": 1.5072,
+      "step": 155295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9932089088437686e-06,
+      "loss": 1.6062,
+      "step": 155300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.99266976817295e-06,
+      "loss": 1.5288,
+      "step": 155305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9921306675177553e-06,
+      "loss": 1.2924,
+      "step": 155310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.991591606881258e-06,
+      "loss": 1.1568,
+      "step": 155315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9910525862665475e-06,
+      "loss": 1.3145,
+      "step": 155320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.990513605676697e-06,
+      "loss": 1.4081,
+      "step": 155325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9899746651147845e-06,
+      "loss": 1.3091,
+      "step": 155330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.989435764583882e-06,
+      "loss": 1.3074,
+      "step": 155335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9888969040870742e-06,
+      "loss": 1.5458,
+      "step": 155340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9883580836274385e-06,
+      "loss": 1.0941,
+      "step": 155345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.987819303208045e-06,
+      "loss": 1.0388,
+      "step": 155350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.987280562831979e-06,
+      "loss": 1.1472,
+      "step": 155355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9867418625023093e-06,
+      "loss": 1.1272,
+      "step": 155360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9862032022221165e-06,
+      "loss": 1.199,
+      "step": 155365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9856645819944775e-06,
+      "loss": 1.69,
+      "step": 155370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9851260018224684e-06,
+      "loss": 1.2092,
+      "step": 155375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9845874617091575e-06,
+      "loss": 1.2415,
+      "step": 155380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9840489616576264e-06,
+      "loss": 1.2427,
+      "step": 155385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.983510501670953e-06,
+      "loss": 1.1116,
+      "step": 155390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.982972081752208e-06,
+      "loss": 1.0433,
+      "step": 155395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9824337019044624e-06,
+      "loss": 0.9652,
+      "step": 155400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.981895362130799e-06,
+      "loss": 1.3215,
+      "step": 155405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9813570624342846e-06,
+      "loss": 1.0579,
+      "step": 155410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9808188028179964e-06,
+      "loss": 1.3349,
+      "step": 155415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9802805832850114e-06,
+      "loss": 1.6747,
+      "step": 155420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9797424038383994e-06,
+      "loss": 1.4267,
+      "step": 155425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.979204264481231e-06,
+      "loss": 1.2668,
+      "step": 155430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.978666165216583e-06,
+      "loss": 1.1962,
+      "step": 155435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9781281060475296e-06,
+      "loss": 1.2751,
+      "step": 155440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9775900869771392e-06,
+      "loss": 1.2145,
+      "step": 155445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.977052108008489e-06,
+      "loss": 1.1148,
+      "step": 155450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9765141691446464e-06,
+      "loss": 1.2425,
+      "step": 155455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.975976270388684e-06,
+      "loss": 1.1457,
+      "step": 155460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9754384117436797e-06,
+      "loss": 1.291,
+      "step": 155465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9749005932126997e-06,
+      "loss": 1.0993,
+      "step": 155470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9743628147988125e-06,
+      "loss": 1.4766,
+      "step": 155475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9738250765050927e-06,
+      "loss": 1.1607,
+      "step": 155480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.973287378334613e-06,
+      "loss": 1.4006,
+      "step": 155485
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.972749720290442e-06,
+      "loss": 1.5865,
+      "step": 155490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9722121023756466e-06,
+      "loss": 1.3493,
+      "step": 155495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.971674524593301e-06,
+      "loss": 1.2491,
+      "step": 155500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.971136986946476e-06,
+      "loss": 0.9264,
+      "step": 155505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.970599489438237e-06,
+      "loss": 1.3282,
+      "step": 155510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.970062032071658e-06,
+      "loss": 1.0212,
+      "step": 155515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.969524614849807e-06,
+      "loss": 1.4533,
+      "step": 155520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.968987237775748e-06,
+      "loss": 1.492,
+      "step": 155525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9684499008525526e-06,
+      "loss": 1.0843,
+      "step": 155530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.967912604083295e-06,
+      "loss": 1.3816,
+      "step": 155535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9673753474710378e-06,
+      "loss": 1.429,
+      "step": 155540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9668381310188465e-06,
+      "loss": 1.2451,
+      "step": 155545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9663009547297927e-06,
+      "loss": 1.3621,
+      "step": 155550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9657638186069473e-06,
+      "loss": 1.1228,
+      "step": 155555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.965226722653369e-06,
+      "loss": 1.2833,
+      "step": 155560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.964689666872135e-06,
+      "loss": 1.4607,
+      "step": 155565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9641526512663023e-06,
+      "loss": 1.4244,
+      "step": 155570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.963615675838942e-06,
+      "loss": 1.2454,
+      "step": 155575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.963078740593125e-06,
+      "loss": 1.2297,
+      "step": 155580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.962541845531913e-06,
+      "loss": 1.0094,
+      "step": 155585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.962004990658368e-06,
+      "loss": 1.3405,
+      "step": 155590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9614681759755615e-06,
+      "loss": 1.3149,
+      "step": 155595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9609314014865596e-06,
+      "loss": 1.2311,
+      "step": 155600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9603946671944227e-06,
+      "loss": 1.1134,
+      "step": 155605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9598579731022214e-06,
+      "loss": 1.5147,
+      "step": 155610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9593213192130175e-06,
+      "loss": 1.6,
+      "step": 155615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9587847055298723e-06,
+      "loss": 1.3295,
+      "step": 155620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.958248132055854e-06,
+      "loss": 1.12,
+      "step": 155625
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.95771159879403e-06,
+      "loss": 1.1156,
+      "step": 155630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9571751057474596e-06,
+      "loss": 1.2342,
+      "step": 155635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9566386529192038e-06,
+      "loss": 1.2838,
+      "step": 155640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9561022403123307e-06,
+      "loss": 1.0921,
+      "step": 155645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.955565867929905e-06,
+      "loss": 1.0947,
+      "step": 155650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9550295357749837e-06,
+      "loss": 1.3173,
+      "step": 155655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.954493243850636e-06,
+      "loss": 1.4075,
+      "step": 155660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.953956992159919e-06,
+      "loss": 1.133,
+      "step": 155665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.953420780705897e-06,
+      "loss": 1.3924,
+      "step": 155670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.952884609491634e-06,
+      "loss": 1.1795,
+      "step": 155675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9523484785201917e-06,
+      "loss": 1.3803,
+      "step": 155680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9518123877946304e-06,
+      "loss": 1.283,
+      "step": 155685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.951276337318004e-06,
+      "loss": 1.0926,
+      "step": 155690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9507403270933886e-06,
+      "loss": 1.2145,
+      "step": 155695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.950204357123836e-06,
+      "loss": 1.1247,
+      "step": 155700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9496684274124053e-06,
+      "loss": 1.2684,
+      "step": 155705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.949132537962164e-06,
+      "loss": 1.2114,
+      "step": 155710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.948596688776164e-06,
+      "loss": 1.2987,
+      "step": 155715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9480608798574708e-06,
+      "loss": 1.3418,
+      "step": 155720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.947525111209144e-06,
+      "loss": 1.075,
+      "step": 155725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9469893828342433e-06,
+      "loss": 1.2358,
+      "step": 155730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9464536947358235e-06,
+      "loss": 1.121,
+      "step": 155735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9459180469169467e-06,
+      "loss": 1.0807,
+      "step": 155740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9453824393806762e-06,
+      "loss": 1.039,
+      "step": 155745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9448468721300614e-06,
+      "loss": 1.8636,
+      "step": 155750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9443113451681694e-06,
+      "loss": 1.1619,
+      "step": 155755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9437758584980515e-06,
+      "loss": 1.1355,
+      "step": 155760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.943240412122772e-06,
+      "loss": 1.5271,
+      "step": 155765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9427050060453823e-06,
+      "loss": 1.1852,
+      "step": 155770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9421696402689448e-06,
+      "loss": 1.1354,
+      "step": 155775
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.941634314796513e-06,
+      "loss": 0.9611,
+      "step": 155780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.941099029631145e-06,
+      "loss": 1.3285,
+      "step": 155785
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9405637847759006e-06,
+      "loss": 1.2316,
+      "step": 155790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9400285802338346e-06,
+      "loss": 1.081,
+      "step": 155795
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.939493416008e-06,
+      "loss": 1.2247,
+      "step": 155800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.938958292101458e-06,
+      "loss": 1.4874,
+      "step": 155805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.938423208517258e-06,
+      "loss": 1.3029,
+      "step": 155810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.937888165258461e-06,
+      "loss": 1.1513,
+      "step": 155815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.937353162328124e-06,
+      "loss": 1.0844,
+      "step": 155820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.936818199729299e-06,
+      "loss": 1.524,
+      "step": 155825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9362832774650376e-06,
+      "loss": 1.2545,
+      "step": 155830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.935748395538397e-06,
+      "loss": 1.0354,
+      "step": 155835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9352135539524375e-06,
+      "loss": 1.4419,
+      "step": 155840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.934678752710207e-06,
+      "loss": 1.435,
+      "step": 155845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9341439918147585e-06,
+      "loss": 1.1763,
+      "step": 155850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9336092712691477e-06,
+      "loss": 1.0674,
+      "step": 155855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9330745910764326e-06,
+      "loss": 1.2886,
+      "step": 155860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.932539951239658e-06,
+      "loss": 1.1691,
+      "step": 155865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9320053517618863e-06,
+      "loss": 1.4286,
+      "step": 155870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9314707926461618e-06,
+      "loss": 1.1532,
+      "step": 155875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9309362738955403e-06,
+      "loss": 1.1644,
+      "step": 155880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.930401795513078e-06,
+      "loss": 1.4172,
+      "step": 155885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9298673575018244e-06,
+      "loss": 1.0126,
+      "step": 155890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9293329598648277e-06,
+      "loss": 1.2819,
+      "step": 155895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.928798602605143e-06,
+      "loss": 1.0562,
+      "step": 155900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9282642857258237e-06,
+      "loss": 1.1504,
+      "step": 155905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9277300092299197e-06,
+      "loss": 1.0729,
+      "step": 155910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.927195773120478e-06,
+      "loss": 1.1989,
+      "step": 155915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9266615774005557e-06,
+      "loss": 1.2704,
+      "step": 155920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.926127422073197e-06,
+      "loss": 1.3323,
+      "step": 155925
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9255933071414553e-06,
+      "loss": 1.1811,
+      "step": 155930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9250592326083836e-06,
+      "loss": 1.3426,
+      "step": 155935
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9245251984770297e-06,
+      "loss": 1.1798,
+      "step": 155940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.923991204750439e-06,
+      "loss": 1.4392,
+      "step": 155945
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9234572514316637e-06,
+      "loss": 1.3324,
+      "step": 155950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9229233385237575e-06,
+      "loss": 1.1559,
+      "step": 155955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9223894660297625e-06,
+      "loss": 0.9161,
+      "step": 155960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9218556339527336e-06,
+      "loss": 1.1935,
+      "step": 155965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9213218422957125e-06,
+      "loss": 1.3588,
+      "step": 155970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9207880910617535e-06,
+      "loss": 1.2958,
+      "step": 155975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9202543802538997e-06,
+      "loss": 1.2415,
+      "step": 155980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9197207098752047e-06,
+      "loss": 1.6559,
+      "step": 155985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9191870799287125e-06,
+      "loss": 1.1167,
+      "step": 155990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9186534904174633e-06,
+      "loss": 1.3797,
+      "step": 155995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9181199413445182e-06,
+      "loss": 1.1268,
+      "step": 156000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9175864327129177e-06,
+      "loss": 1.2329,
+      "step": 156005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.917052964525704e-06,
+      "loss": 1.0721,
+      "step": 156010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.916519536785931e-06,
+      "loss": 1.0956,
+      "step": 156015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.915986149496638e-06,
+      "loss": 1.2415,
+      "step": 156020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.915452802660874e-06,
+      "loss": 1.5513,
+      "step": 156025
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9149194962816895e-06,
+      "loss": 1.5668,
+      "step": 156030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.914386230362124e-06,
+      "loss": 1.6236,
+      "step": 156035
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9138530049052215e-06,
+      "loss": 1.3347,
+      "step": 156040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9133198199140287e-06,
+      "loss": 1.1072,
+      "step": 156045
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9127866753915955e-06,
+      "loss": 1.2025,
+      "step": 156050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.912253571340963e-06,
+      "loss": 1.3463,
+      "step": 156055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.91172050776517e-06,
+      "loss": 0.9616,
+      "step": 156060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.911187484667266e-06,
+      "loss": 1.2097,
+      "step": 156065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9106545020502975e-06,
+      "loss": 1.5246,
+      "step": 156070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.910121559917302e-06,
+      "loss": 1.1828,
+      "step": 156075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9095886582713273e-06,
+      "loss": 1.1833,
+      "step": 156080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.909055797115413e-06,
+      "loss": 1.1275,
+      "step": 156085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9085229764526045e-06,
+      "loss": 1.1497,
+      "step": 156090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.907990196285946e-06,
+      "loss": 1.151,
+      "step": 156095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.907457456618479e-06,
+      "loss": 1.1823,
+      "step": 156100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.90692475745324e-06,
+      "loss": 1.0272,
+      "step": 156105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9063920987932804e-06,
+      "loss": 1.2674,
+      "step": 156110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.905859480641633e-06,
+      "loss": 1.2245,
+      "step": 156115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9053269030013474e-06,
+      "loss": 1.181,
+      "step": 156120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9047943658754576e-06,
+      "loss": 1.0679,
+      "step": 156125
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.904261869267011e-06,
+      "loss": 1.2929,
+      "step": 156130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.903729413179043e-06,
+      "loss": 1.1214,
+      "step": 156135
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.903196997614597e-06,
+      "loss": 1.2978,
+      "step": 156140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.902664622576715e-06,
+      "loss": 1.604,
+      "step": 156145
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.902132288068437e-06,
+      "loss": 0.8644,
+      "step": 156150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9015999940927963e-06,
+      "loss": 1.4196,
+      "step": 156155
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9010677406528387e-06,
+      "loss": 1.1777,
+      "step": 156160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9005355277516056e-06,
+      "loss": 1.0587,
+      "step": 156165
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9000033553921303e-06,
+      "loss": 1.0426,
+      "step": 156170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.899471223577457e-06,
+      "loss": 1.4349,
+      "step": 156175
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.898939132310619e-06,
+      "loss": 1.539,
+      "step": 156180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8984070815946584e-06,
+      "loss": 1.1355,
+      "step": 156185
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8978750714326154e-06,
+      "loss": 1.4417,
+      "step": 156190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8973431018275255e-06,
+      "loss": 1.5915,
+      "step": 156195
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.896811172782428e-06,
+      "loss": 1.1575,
+      "step": 156200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8962792843003553e-06,
+      "loss": 1.2082,
+      "step": 156205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8957474363843486e-06,
+      "loss": 1.145,
+      "step": 156210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.895215629037448e-06,
+      "loss": 1.2004,
+      "step": 156215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8946838622626837e-06,
+      "loss": 1.0025,
+      "step": 156220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8941521360631008e-06,
+      "loss": 1.2804,
+      "step": 156225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8936204504417264e-06,
+      "loss": 1.0843,
+      "step": 156230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8930888054016027e-06,
+      "loss": 1.4622,
+      "step": 156235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.892557200945767e-06,
+      "loss": 1.1006,
+      "step": 156240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8920256370772525e-06,
+      "loss": 1.3792,
+      "step": 156245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8914941137990916e-06,
+      "loss": 1.0703,
+      "step": 156250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8909626311143224e-06,
+      "loss": 1.2326,
+      "step": 156255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.890431189025984e-06,
+      "loss": 1.2074,
+      "step": 156260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.889899787537106e-06,
+      "loss": 1.2842,
+      "step": 156265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8893684266507227e-06,
+      "loss": 1.059,
+      "step": 156270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8888371063698695e-06,
+      "loss": 1.1406,
+      "step": 156275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.888305826697584e-06,
+      "loss": 1.2442,
+      "step": 156280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8877745876368946e-06,
+      "loss": 1.1486,
+      "step": 156285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8872433891908414e-06,
+      "loss": 1.1087,
+      "step": 156290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.886712231362453e-06,
+      "loss": 1.1677,
+      "step": 156295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.886181114154758e-06,
+      "loss": 1.3455,
+      "step": 156300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8856500375708008e-06,
+      "loss": 0.9305,
+      "step": 156305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.885119001613609e-06,
+      "loss": 1.4301,
+      "step": 156310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.884588006286211e-06,
+      "loss": 1.5604,
+      "step": 156315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8840570515916455e-06,
+      "loss": 1.3638,
+      "step": 156320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.883526137532938e-06,
+      "loss": 1.1035,
+      "step": 156325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.882995264113124e-06,
+      "loss": 1.4514,
+      "step": 156330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.882464431335239e-06,
+      "loss": 1.2201,
+      "step": 156335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.88193363920231e-06,
+      "loss": 1.3977,
+      "step": 156340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.881402887717365e-06,
+      "loss": 1.806,
+      "step": 156345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.880872176883438e-06,
+      "loss": 1.1702,
+      "step": 156350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8803415067035635e-06,
+      "loss": 1.0369,
+      "step": 156355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8798108771807677e-06,
+      "loss": 0.9816,
+      "step": 156360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8792802883180793e-06,
+      "loss": 1.5799,
+      "step": 156365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8787497401185292e-06,
+      "loss": 0.9761,
+      "step": 156370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.878219232585152e-06,
+      "loss": 1.2316,
+      "step": 156375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8776887657209705e-06,
+      "loss": 1.2136,
+      "step": 156380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8771583395290194e-06,
+      "loss": 1.2045,
+      "step": 156385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8766279540123253e-06,
+      "loss": 1.4916,
+      "step": 156390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.876097609173911e-06,
+      "loss": 1.1338,
+      "step": 156395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8755673050168178e-06,
+      "loss": 1.3669,
+      "step": 156400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8750370415440666e-06,
+      "loss": 1.2767,
+      "step": 156405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8745068187586854e-06,
+      "loss": 1.225,
+      "step": 156410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8739766366637e-06,
+      "loss": 1.1027,
+      "step": 156415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.873446495262141e-06,
+      "loss": 1.5709,
+      "step": 156420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8729163945570394e-06,
+      "loss": 1.1133,
+      "step": 156425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8723863345514146e-06,
+      "loss": 1.2914,
+      "step": 156430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8718563152483005e-06,
+      "loss": 1.3757,
+      "step": 156435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.871326336650717e-06,
+      "loss": 1.118,
+      "step": 156440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8707963987616948e-06,
+      "loss": 1.1259,
+      "step": 156445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8702665015842636e-06,
+      "loss": 1.5226,
+      "step": 156450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8697366451214447e-06,
+      "loss": 1.2032,
+      "step": 156455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.869206829376261e-06,
+      "loss": 1.1687,
+      "step": 156460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.868677054351743e-06,
+      "loss": 1.0766,
+      "step": 156465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.868147320050916e-06,
+      "loss": 1.3377,
+      "step": 156470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8676176264768017e-06,
+      "loss": 1.0914,
+      "step": 156475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.86708797363243e-06,
+      "loss": 1.203,
+      "step": 156480
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8665583615208192e-06,
+      "loss": 1.1186,
+      "step": 156485
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.866028790144999e-06,
+      "loss": 1.3045,
+      "step": 156490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.86549925950799e-06,
+      "loss": 1.0644,
+      "step": 156495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.86496976961282e-06,
+      "loss": 1.4954,
+      "step": 156500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8644403204625094e-06,
+      "loss": 1.2601,
+      "step": 156505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8639109120600805e-06,
+      "loss": 1.0346,
+      "step": 156510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.863381544408558e-06,
+      "loss": 1.2934,
+      "step": 156515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8628522175109675e-06,
+      "loss": 1.1469,
+      "step": 156520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.862322931370327e-06,
+      "loss": 1.173,
+      "step": 156525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8617936859896654e-06,
+      "loss": 1.5858,
+      "step": 156530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.861264481371998e-06,
+      "loss": 1.1261,
+      "step": 156535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8607353175203502e-06,
+      "loss": 0.977,
+      "step": 156540
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8602061944377458e-06,
+      "loss": 1.0,
+      "step": 156545
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.859677112127204e-06,
+      "loss": 1.4697,
+      "step": 156550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.859148070591744e-06,
+      "loss": 0.9889,
+      "step": 156555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8586190698343885e-06,
+      "loss": 1.0256,
+      "step": 156560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8580901098581636e-06,
+      "loss": 1.1049,
+      "step": 156565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.857561190666084e-06,
+      "loss": 1.0377,
+      "step": 156570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8570323122611697e-06,
+      "loss": 1.458,
+      "step": 156575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8565034746464413e-06,
+      "loss": 1.4926,
+      "step": 156580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8559746778249243e-06,
+      "loss": 1.167,
+      "step": 156585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.855445921799631e-06,
+      "loss": 1.3246,
+      "step": 156590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8549172065735873e-06,
+      "loss": 1.0559,
+      "step": 156595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.85438853214981e-06,
+      "loss": 1.2799,
+      "step": 156600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8538598985313102e-06,
+      "loss": 1.0691,
+      "step": 156605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.853331305721121e-06,
+      "loss": 1.0499,
+      "step": 156610
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8528027537222537e-06,
+      "loss": 1.5055,
+      "step": 156615
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8522742425377238e-06,
+      "loss": 1.3418,
+      "step": 156620
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8517457721705545e-06,
+      "loss": 1.4949,
+      "step": 156625
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.85121734262376e-06,
+      "loss": 1.2738,
+      "step": 156630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8506889539003614e-06,
+      "loss": 1.4289,
+      "step": 156635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8501606060033715e-06,
+      "loss": 1.259,
+      "step": 156640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.849632298935813e-06,
+      "loss": 1.3103,
+      "step": 156645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8491040327006968e-06,
+      "loss": 1.2525,
+      "step": 156650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.848575807301043e-06,
+      "loss": 1.1221,
+      "step": 156655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.84804762273987e-06,
+      "loss": 1.3327,
+      "step": 156660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.847519479020193e-06,
+      "loss": 1.1743,
+      "step": 156665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.846991376145023e-06,
+      "loss": 1.2828,
+      "step": 156670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.846463314117379e-06,
+      "loss": 1.1737,
+      "step": 156675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8459352929402805e-06,
+      "loss": 1.5487,
+      "step": 156680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.845407312616737e-06,
+      "loss": 1.1993,
+      "step": 156685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8448793731497682e-06,
+      "loss": 0.8738,
+      "step": 156690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.844351474542387e-06,
+      "loss": 1.1289,
+      "step": 156695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8438236167976042e-06,
+      "loss": 1.5425,
+      "step": 156700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8432957999184375e-06,
+      "loss": 1.2025,
+      "step": 156705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8427680239079036e-06,
+      "loss": 1.1521,
+      "step": 156710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.842240288769015e-06,
+      "loss": 1.0168,
+      "step": 156715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.841712594504781e-06,
+      "loss": 0.9196,
+      "step": 156720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.841184941118217e-06,
+      "loss": 1.1261,
+      "step": 156725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.840657328612341e-06,
+      "loss": 1.2864,
+      "step": 156730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.84012975699016e-06,
+      "loss": 1.1989,
+      "step": 156735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8396022262546906e-06,
+      "loss": 1.5645,
+      "step": 156740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8390747364089423e-06,
+      "loss": 1.2454,
+      "step": 156745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.838547287455927e-06,
+      "loss": 1.2245,
+      "step": 156750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8380198793986624e-06,
+      "loss": 1.0258,
+      "step": 156755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8374925122401566e-06,
+      "loss": 1.0384,
+      "step": 156760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.836965185983418e-06,
+      "loss": 1.4523,
+      "step": 156765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.83643790063146e-06,
+      "loss": 1.404,
+      "step": 156770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8359106561872984e-06,
+      "loss": 1.2042,
+      "step": 156775
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8353834526539405e-06,
+      "loss": 1.3996,
+      "step": 156780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8348562900343936e-06,
+      "loss": 1.3521,
+      "step": 156785
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.834329168331673e-06,
+      "loss": 1.3902,
+      "step": 156790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.833802087548785e-06,
+      "loss": 1.1623,
+      "step": 156795
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8332750476887405e-06,
+      "loss": 1.0358,
+      "step": 156800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8327480487545533e-06,
+      "loss": 1.2608,
+      "step": 156805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8322210907492297e-06,
+      "loss": 1.0685,
+      "step": 156810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8316941736757763e-06,
+      "loss": 1.249,
+      "step": 156815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.831167297537204e-06,
+      "loss": 1.0774,
+      "step": 156820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.830640462336526e-06,
+      "loss": 1.2059,
+      "step": 156825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8301136680767427e-06,
+      "loss": 1.0102,
+      "step": 156830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8295869147608712e-06,
+      "loss": 1.2002,
+      "step": 156835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.829060202391911e-06,
+      "loss": 1.4309,
+      "step": 156840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.828533530972878e-06,
+      "loss": 1.0575,
+      "step": 156845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.828006900506772e-06,
+      "loss": 1.0256,
+      "step": 156850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8274803109966075e-06,
+      "loss": 1.2849,
+      "step": 156855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8269537624453847e-06,
+      "loss": 1.1978,
+      "step": 156860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8264272548561133e-06,
+      "loss": 1.1277,
+      "step": 156865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8259007882318048e-06,
+      "loss": 1.3717,
+      "step": 156870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8253743625754615e-06,
+      "loss": 1.0959,
+      "step": 156875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8248479778900863e-06,
+      "loss": 1.5432,
+      "step": 156880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8243216341786872e-06,
+      "loss": 1.2868,
+      "step": 156885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8237953314442744e-06,
+      "loss": 1.1431,
+      "step": 156890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.823269069689848e-06,
+      "loss": 1.328,
+      "step": 156895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.822742848918417e-06,
+      "loss": 1.2204,
+      "step": 156900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8222166691329855e-06,
+      "loss": 1.3615,
+      "step": 156905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8216905303365548e-06,
+      "loss": 1.432,
+      "step": 156910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8211644325321306e-06,
+      "loss": 1.1369,
+      "step": 156915
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.820638375722723e-06,
+      "loss": 1.2354,
+      "step": 156920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.820112359911331e-06,
+      "loss": 1.1516,
+      "step": 156925
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8195863851009563e-06,
+      "loss": 1.0924,
+      "step": 156930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.819060451294606e-06,
+      "loss": 1.062,
+      "step": 156935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.818534558495285e-06,
+      "loss": 1.0586,
+      "step": 156940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8180087067059914e-06,
+      "loss": 1.3439,
+      "step": 156945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.817482895929734e-06,
+      "loss": 1.7076,
+      "step": 156950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8169571261695095e-06,
+      "loss": 1.0407,
+      "step": 156955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.816431397428323e-06,
+      "loss": 1.3993,
+      "step": 156960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8159057097091803e-06,
+      "loss": 1.0147,
+      "step": 156965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8153800630150797e-06,
+      "loss": 1.3442,
+      "step": 156970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.81485445734902e-06,
+      "loss": 1.6316,
+      "step": 156975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8143288927140057e-06,
+      "loss": 1.3902,
+      "step": 156980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.813803369113042e-06,
+      "loss": 1.3162,
+      "step": 156985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8132778865491262e-06,
+      "loss": 1.073,
+      "step": 156990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8127524450252553e-06,
+      "loss": 1.3024,
+      "step": 156995
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8122270445444366e-06,
+      "loss": 1.1313,
+      "step": 157000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8117016851096655e-06,
+      "loss": 1.1467,
+      "step": 157005
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8111763667239433e-06,
+      "loss": 1.4943,
+      "step": 157010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8106510893902738e-06,
+      "loss": 1.0788,
+      "step": 157015
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8101258531116526e-06,
+      "loss": 1.2027,
+      "step": 157020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8096006578910773e-06,
+      "loss": 1.3274,
+      "step": 157025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.80907550373155e-06,
+      "loss": 1.3217,
+      "step": 157030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.808550390636072e-06,
+      "loss": 1.2525,
+      "step": 157035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8080253186076366e-06,
+      "loss": 1.5938,
+      "step": 157040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8075002876492475e-06,
+      "loss": 1.1659,
+      "step": 157045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8069752977638976e-06,
+      "loss": 1.087,
+      "step": 157050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.806450348954587e-06,
+      "loss": 0.8899,
+      "step": 157055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8059254412243163e-06,
+      "loss": 1.2219,
+      "step": 157060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.805400574576083e-06,
+      "loss": 1.5385,
+      "step": 157065
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8048757490128762e-06,
+      "loss": 1.1455,
+      "step": 157070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.804350964537701e-06,
+      "loss": 1.6173,
+      "step": 157075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.803826221153554e-06,
+      "loss": 1.2743,
+      "step": 157080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8033015188634295e-06,
+      "loss": 1.3162,
+      "step": 157085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.802776857670322e-06,
+      "loss": 1.2323,
+      "step": 157090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.802252237577232e-06,
+      "loss": 1.2292,
+      "step": 157095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.80172765858715e-06,
+      "loss": 0.9815,
+      "step": 157100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8012031207030753e-06,
+      "loss": 1.1906,
+      "step": 157105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.800678623928005e-06,
+      "loss": 1.1108,
+      "step": 157110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8001541682649325e-06,
+      "loss": 1.3129,
+      "step": 157115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7996297537168483e-06,
+      "loss": 1.1574,
+      "step": 157120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7991053802867516e-06,
+      "loss": 1.285,
+      "step": 157125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7985810479776397e-06,
+      "loss": 1.2923,
+      "step": 157130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7980567567925e-06,
+      "loss": 1.1,
+      "step": 157135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7975325067343317e-06,
+      "loss": 1.5084,
+      "step": 157140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.797008297806125e-06,
+      "loss": 1.0249,
+      "step": 157145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.796484130010877e-06,
+      "loss": 1.3849,
+      "step": 157150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7959600033515766e-06,
+      "loss": 1.349,
+      "step": 157155
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7954359178312228e-06,
+      "loss": 1.3543,
+      "step": 157160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7949118734528013e-06,
+      "loss": 1.3284,
+      "step": 157165
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7943878702193084e-06,
+      "loss": 1.5138,
+      "step": 157170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7938639081337393e-06,
+      "loss": 1.179,
+      "step": 157175
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.793339987199083e-06,
+      "loss": 1.1193,
+      "step": 157180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.792816107418328e-06,
+      "loss": 1.3377,
+      "step": 157185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.792292268794471e-06,
+      "loss": 1.217,
+      "step": 157190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.791768471330504e-06,
+      "loss": 1.4377,
+      "step": 157195
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.791244715029412e-06,
+      "loss": 1.7512,
+      "step": 157200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7907209998941943e-06,
+      "loss": 1.1826,
+      "step": 157205
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.790197325927836e-06,
+      "loss": 1.3524,
+      "step": 157210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7896736931333267e-06,
+      "loss": 1.1319,
+      "step": 157215
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7891501015136578e-06,
+      "loss": 1.082,
+      "step": 157220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7886265510718247e-06,
+      "loss": 1.5501,
+      "step": 157225
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7881030418108112e-06,
+      "loss": 1.201,
+      "step": 157230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7875795737336065e-06,
+      "loss": 1.0138,
+      "step": 157235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7870561468432013e-06,
+      "loss": 1.2464,
+      "step": 157240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7865327611425875e-06,
+      "loss": 1.3958,
+      "step": 157245
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7860094166347486e-06,
+      "loss": 1.1035,
+      "step": 157250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7854861133226787e-06,
+      "loss": 1.3107,
+      "step": 157255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7849628512093608e-06,
+      "loss": 1.2262,
+      "step": 157260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.784439630297785e-06,
+      "loss": 1.525,
+      "step": 157265
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7839164505909434e-06,
+      "loss": 1.3614,
+      "step": 157270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.783393312091821e-06,
+      "loss": 1.5794,
+      "step": 157275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7828702148034004e-06,
+      "loss": 1.1274,
+      "step": 157280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.782347158728672e-06,
+      "loss": 1.2357,
+      "step": 157285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7818241438706273e-06,
+      "loss": 1.168,
+      "step": 157290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7813011702322477e-06,
+      "loss": 1.3851,
+      "step": 157295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7807782378165193e-06,
+      "loss": 1.4156,
+      "step": 157300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7802553466264324e-06,
+      "loss": 0.9753,
+      "step": 157305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7797324966649687e-06,
+      "loss": 1.3188,
+      "step": 157310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7792096879351147e-06,
+      "loss": 1.0646,
+      "step": 157315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.77868692043986e-06,
+      "loss": 1.4438,
+      "step": 157320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.778164194182188e-06,
+      "loss": 1.5783,
+      "step": 157325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7776415091650787e-06,
+      "loss": 1.4334,
+      "step": 157330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7771188653915217e-06,
+      "loss": 1.0996,
+      "step": 157335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.776596262864504e-06,
+      "loss": 1.1648,
+      "step": 157340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7760737015870033e-06,
+      "loss": 1.1972,
+      "step": 157345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7755511815620106e-06,
+      "loss": 1.245,
+      "step": 157350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7750287027925025e-06,
+      "loss": 1.2815,
+      "step": 157355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7745062652814704e-06,
+      "loss": 0.9719,
+      "step": 157360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.77398386903189e-06,
+      "loss": 1.325,
+      "step": 157365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7734615140467513e-06,
+      "loss": 1.0764,
+      "step": 157370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7729392003290347e-06,
+      "loss": 1.3613,
+      "step": 157375
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.772416927881716e-06,
+      "loss": 1.3184,
+      "step": 157380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.77189469670779e-06,
+      "loss": 1.2233,
+      "step": 157385
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7713725068102326e-06,
+      "loss": 1.6128,
+      "step": 157390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.770850358192024e-06,
+      "loss": 1.2091,
+      "step": 157395
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.77032825085615e-06,
+      "loss": 1.0548,
+      "step": 157400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7698061848055878e-06,
+      "loss": 1.2656,
+      "step": 157405
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.769284160043322e-06,
+      "loss": 1.0274,
+      "step": 157410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.768762176572334e-06,
+      "loss": 1.1136,
+      "step": 157415
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.768240234395604e-06,
+      "loss": 1.1539,
+      "step": 157420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7677183335161096e-06,
+      "loss": 1.2566,
+      "step": 157425
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.767196473936833e-06,
+      "loss": 1.1987,
+      "step": 157430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.766674655660757e-06,
+      "loss": 1.451,
+      "step": 157435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.76615287869086e-06,
+      "loss": 1.2463,
+      "step": 157440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7656311430301174e-06,
+      "loss": 1.3785,
+      "step": 157445
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7651094486815112e-06,
+      "loss": 1.4998,
+      "step": 157450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.764587795648026e-06,
+      "loss": 1.7303,
+      "step": 157455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.764066183932632e-06,
+      "loss": 1.1903,
+      "step": 157460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7635446135383147e-06,
+      "loss": 1.2008,
+      "step": 157465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.763023084468047e-06,
+      "loss": 1.0597,
+      "step": 157470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7625015967248113e-06,
+      "loss": 1.4918,
+      "step": 157475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7619801503115863e-06,
+      "loss": 1.0755,
+      "step": 157480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7614587452313466e-06,
+      "loss": 1.5057,
+      "step": 157485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.760937381487069e-06,
+      "loss": 1.1764,
+      "step": 157490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.760416059081735e-06,
+      "loss": 1.4355,
+      "step": 157495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.759894778018316e-06,
+      "loss": 1.2585,
+      "step": 157500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7593735382997956e-06,
+      "loss": 1.2568,
+      "step": 157505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7588523399291434e-06,
+      "loss": 1.4041,
+      "step": 157510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7583311829093417e-06,
+      "loss": 1.1284,
+      "step": 157515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7578100672433605e-06,
+      "loss": 1.2845,
+      "step": 157520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7572889929341797e-06,
+      "loss": 1.0345,
+      "step": 157525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7567679599847774e-06,
+      "loss": 1.2515,
+      "step": 157530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.756246968398125e-06,
+      "loss": 1.2447,
+      "step": 157535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.755726018177196e-06,
+      "loss": 0.9747,
+      "step": 157540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7552051093249677e-06,
+      "loss": 1.5836,
+      "step": 157545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7546842418444186e-06,
+      "loss": 1.1402,
+      "step": 157550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7541634157385155e-06,
+      "loss": 1.0262,
+      "step": 157555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7536426310102393e-06,
+      "loss": 1.1142,
+      "step": 157560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7531218876625587e-06,
+      "loss": 1.2608,
+      "step": 157565
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7526011856984503e-06,
+      "loss": 0.9616,
+      "step": 157570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7520805251208892e-06,
+      "loss": 1.1728,
+      "step": 157575
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.751559905932847e-06,
+      "loss": 1.4607,
+      "step": 157580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.751039328137297e-06,
+      "loss": 0.9411,
+      "step": 157585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7505187917372044e-06,
+      "loss": 1.2047,
+      "step": 157590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.749998296735554e-06,
+      "loss": 1.2981,
+      "step": 157595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7494778431353142e-06,
+      "loss": 1.0375,
+      "step": 157600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.748957430939453e-06,
+      "loss": 1.4096,
+      "step": 157605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7484370601509468e-06,
+      "loss": 1.2498,
+      "step": 157610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7479167307727616e-06,
+      "loss": 1.0264,
+      "step": 157615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7473964428078735e-06,
+      "loss": 1.5853,
+      "step": 157620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.746876196259255e-06,
+      "loss": 1.1091,
+      "step": 157625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7463559911298753e-06,
+      "loss": 1.4846,
+      "step": 157630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7458358274227004e-06,
+      "loss": 1.1002,
+      "step": 157635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.745315705140705e-06,
+      "loss": 1.1891,
+      "step": 157640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7447956242868623e-06,
+      "loss": 1.2037,
+      "step": 157645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.744275584864138e-06,
+      "loss": 1.265,
+      "step": 157650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7437555868755004e-06,
+      "loss": 1.607,
+      "step": 157655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7432356303239204e-06,
+      "loss": 1.2513,
+      "step": 157660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.742715715212372e-06,
+      "loss": 1.0963,
+      "step": 157665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7421958415438166e-06,
+      "loss": 1.0704,
+      "step": 157670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7416760093212292e-06,
+      "loss": 1.393,
+      "step": 157675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7411562185475758e-06,
+      "loss": 0.9384,
+      "step": 157680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.740636469225819e-06,
+      "loss": 1.3202,
+      "step": 157685
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7401167613589386e-06,
+      "loss": 1.1971,
+      "step": 157690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.739597094949895e-06,
+      "loss": 1.5294,
+      "step": 157695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7390774700016555e-06,
+      "loss": 1.2878,
+      "step": 157700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7385578865171914e-06,
+      "loss": 1.389,
+      "step": 157705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.738038344499464e-06,
+      "loss": 1.2993,
+      "step": 157710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7375188439514445e-06,
+      "loss": 1.1121,
+      "step": 157715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7369993848761003e-06,
+      "loss": 1.3057,
+      "step": 157720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7364799672763964e-06,
+      "loss": 1.4185,
+      "step": 157725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7359605911552955e-06,
+      "loss": 1.0212,
+      "step": 157730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7354412565157675e-06,
+      "loss": 1.1702,
+      "step": 157735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7349219633607793e-06,
+      "loss": 1.1686,
+      "step": 157740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7344027116932946e-06,
+      "loss": 1.196,
+      "step": 157745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.733883501516276e-06,
+      "loss": 1.308,
+      "step": 157750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7333643328326886e-06,
+      "loss": 1.2149,
+      "step": 157755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7328452056455046e-06,
+      "loss": 1.2359,
+      "step": 157760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.732326119957679e-06,
+      "loss": 1.4042,
+      "step": 157765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7318070757721825e-06,
+      "loss": 1.1522,
+      "step": 157770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7312880730919745e-06,
+      "loss": 1.1287,
+      "step": 157775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.730769111920021e-06,
+      "loss": 0.9728,
+      "step": 157780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7302501922592896e-06,
+      "loss": 1.138,
+      "step": 157785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.729731314112738e-06,
+      "loss": 1.1021,
+      "step": 157790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7292124774833317e-06,
+      "loss": 1.407,
+      "step": 157795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7286936823740307e-06,
+      "loss": 1.3131,
+      "step": 157800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.728174928787798e-06,
+      "loss": 1.167,
+      "step": 157805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.727656216727601e-06,
+      "loss": 1.3348,
+      "step": 157810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7271375461963966e-06,
+      "loss": 1.2926,
+      "step": 157815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.72661891719715e-06,
+      "loss": 0.9939,
+      "step": 157820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7261003297328193e-06,
+      "loss": 1.3258,
+      "step": 157825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.725581783806368e-06,
+      "loss": 1.2738,
+      "step": 157830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.725063279420761e-06,
+      "loss": 1.1383,
+      "step": 157835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7245448165789545e-06,
+      "loss": 1.2292,
+      "step": 157840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7240263952839075e-06,
+      "loss": 1.128,
+      "step": 157845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7235080155385827e-06,
+      "loss": 1.2183,
+      "step": 157850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.722989677345944e-06,
+      "loss": 1.0303,
+      "step": 157855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7224713807089455e-06,
+      "loss": 1.3736,
+      "step": 157860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.721953125630552e-06,
+      "loss": 1.2819,
+      "step": 157865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7214349121137183e-06,
+      "loss": 1.188,
+      "step": 157870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.720916740161409e-06,
+      "loss": 1.1951,
+      "step": 157875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7203986097765778e-06,
+      "loss": 1.1751,
+      "step": 157880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.719880520962188e-06,
+      "loss": 1.2404,
+      "step": 157885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7193624737211955e-06,
+      "loss": 1.2912,
+      "step": 157890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.718844468056554e-06,
+      "loss": 1.1115,
+      "step": 157895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7183265039712325e-06,
+      "loss": 1.367,
+      "step": 157900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.717808581468184e-06,
+      "loss": 1.0528,
+      "step": 157905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7172907005503613e-06,
+      "loss": 0.9841,
+      "step": 157910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.716772861220729e-06,
+      "loss": 1.6634,
+      "step": 157915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.716255063482238e-06,
+      "loss": 1.3837,
+      "step": 157920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.715737307337848e-06,
+      "loss": 1.4576,
+      "step": 157925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.715219592790519e-06,
+      "loss": 1.4791,
+      "step": 157930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7147019198432047e-06,
+      "loss": 1.2515,
+      "step": 157935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7141842884988566e-06,
+      "loss": 1.0383,
+      "step": 157940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7136666987604366e-06,
+      "loss": 1.1226,
+      "step": 157945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7131491506309003e-06,
+      "loss": 1.0038,
+      "step": 157950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7126316441132027e-06,
+      "loss": 1.4154,
+      "step": 157955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7121141792102934e-06,
+      "loss": 1.1471,
+      "step": 157960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7115967559251334e-06,
+      "loss": 1.3183,
+      "step": 157965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7110793742606787e-06,
+      "loss": 1.1199,
+      "step": 157970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.710562034219878e-06,
+      "loss": 1.1665,
+      "step": 157975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7100447358056914e-06,
+      "loss": 1.122,
+      "step": 157980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7095274790210703e-06,
+      "loss": 1.0536,
+      "step": 157985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.709010263868963e-06,
+      "loss": 1.4032,
+      "step": 157990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7084930903523345e-06,
+      "loss": 1.2065,
+      "step": 157995
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7079759584741318e-06,
+      "loss": 1.1323,
+      "step": 158000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.707458868237306e-06,
+      "loss": 1.021,
+      "step": 158005
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.706941819644815e-06,
+      "loss": 1.2077,
+      "step": 158010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.706424812699605e-06,
+      "loss": 1.0987,
+      "step": 158015
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.705907847404636e-06,
+      "loss": 1.3054,
+      "step": 158020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.705390923762854e-06,
+      "loss": 1.1665,
+      "step": 158025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7048740417772145e-06,
+      "loss": 1.4141,
+      "step": 158030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.704357201450667e-06,
+      "loss": 1.3537,
+      "step": 158035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.703840402786162e-06,
+      "loss": 0.9926,
+      "step": 158040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.703323645786656e-06,
+      "loss": 1.6871,
+      "step": 158045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7028069304550974e-06,
+      "loss": 1.3122,
+      "step": 158050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7022902567944332e-06,
+      "loss": 1.5466,
+      "step": 158055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7017736248076156e-06,
+      "loss": 1.4428,
+      "step": 158060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7012570344975997e-06,
+      "loss": 1.2931,
+      "step": 158065
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.700740485867328e-06,
+      "loss": 1.2796,
+      "step": 158070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7002239789197583e-06,
+      "loss": 1.0388,
+      "step": 158075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6997075136578344e-06,
+      "loss": 1.4394,
+      "step": 158080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6991910900845055e-06,
+      "loss": 1.3966,
+      "step": 158085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.698674708202721e-06,
+      "loss": 1.0356,
+      "step": 158090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6981583680154342e-06,
+      "loss": 1.3989,
+      "step": 158095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.697642069525591e-06,
+      "loss": 1.0873,
+      "step": 158100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6971258127361364e-06,
+      "loss": 1.0021,
+      "step": 158105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6966095976500204e-06,
+      "loss": 1.1395,
+      "step": 158110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6960934242701952e-06,
+      "loss": 1.1307,
+      "step": 158115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6955772925996014e-06,
+      "loss": 1.1452,
+      "step": 158120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6950612026411937e-06,
+      "loss": 1.1391,
+      "step": 158125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6945451543979127e-06,
+      "loss": 1.8551,
+      "step": 158130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6940291478727077e-06,
+      "loss": 1.159,
+      "step": 158135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.693513183068529e-06,
+      "loss": 1.2383,
+      "step": 158140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.692997259988319e-06,
+      "loss": 1.2739,
+      "step": 158145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6924813786350223e-06,
+      "loss": 0.9628,
+      "step": 158150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6919655390115873e-06,
+      "loss": 1.1855,
+      "step": 158155
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6914497411209628e-06,
+      "loss": 1.037,
+      "step": 158160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.690933984966091e-06,
+      "loss": 1.0099,
+      "step": 158165
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.690418270549915e-06,
+      "loss": 1.1205,
+      "step": 158170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6899025978753813e-06,
+      "loss": 1.4766,
+      "step": 158175
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6893869669454387e-06,
+      "loss": 1.1119,
+      "step": 158180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.688871377763026e-06,
+      "loss": 1.6163,
+      "step": 158185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.688355830331093e-06,
+      "loss": 1.1558,
+      "step": 158190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.687840324652581e-06,
+      "loss": 1.1956,
+      "step": 158195
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.687324860730429e-06,
+      "loss": 1.3504,
+      "step": 158200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.686809438567587e-06,
+      "loss": 1.0699,
+      "step": 158205
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6862940581669996e-06,
+      "loss": 1.5806,
+      "step": 158210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.685778719531602e-06,
+      "loss": 1.3713,
+      "step": 158215
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6852634226643458e-06,
+      "loss": 1.3781,
+      "step": 158220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6847481675681675e-06,
+      "loss": 1.2548,
+      "step": 158225
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6842329542460143e-06,
+      "loss": 1.0607,
+      "step": 158230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.683717782700822e-06,
+      "loss": 1.4224,
+      "step": 158235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6832026529355406e-06,
+      "loss": 1.0925,
+      "step": 158240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6826875649531026e-06,
+      "loss": 1.2859,
+      "step": 158245
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.682172518756455e-06,
+      "loss": 0.9824,
+      "step": 158250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6816575143485425e-06,
+      "loss": 1.1506,
+      "step": 158255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6811425517322997e-06,
+      "loss": 1.2383,
+      "step": 158260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6806276309106683e-06,
+      "loss": 1.3593,
+      "step": 158265
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6801127518865887e-06,
+      "loss": 1.2672,
+      "step": 158270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6795979146630057e-06,
+      "loss": 1.1727,
+      "step": 158275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6790831192428526e-06,
+      "loss": 1.4933,
+      "step": 158280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6785683656290753e-06,
+      "loss": 0.9559,
+      "step": 158285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6780536538246104e-06,
+      "loss": 1.1184,
+      "step": 158290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.677538983832392e-06,
+      "loss": 1.2144,
+      "step": 158295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.67702435565537e-06,
+      "loss": 1.5657,
+      "step": 158300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.676509769296478e-06,
+      "loss": 1.2591,
+      "step": 158305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.675995224758654e-06,
+      "loss": 1.1175,
+      "step": 158310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6754807220448332e-06,
+      "loss": 1.1481,
+      "step": 158315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6749662611579574e-06,
+      "loss": 1.3619,
+      "step": 158320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6744518421009667e-06,
+      "loss": 1.2002,
+      "step": 158325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6739374648767925e-06,
+      "loss": 1.4979,
+      "step": 158330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6734231294883793e-06,
+      "loss": 1.3184,
+      "step": 158335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.672908835938658e-06,
+      "loss": 1.1271,
+      "step": 158340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6723945842305677e-06,
+      "loss": 1.1375,
+      "step": 158345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6718803743670485e-06,
+      "loss": 1.2003,
+      "step": 158350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6713662063510337e-06,
+      "loss": 1.0008,
+      "step": 158355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.670852080185458e-06,
+      "loss": 1.1553,
+      "step": 158360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6703379958732566e-06,
+      "loss": 1.1851,
+      "step": 158365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.669823953417372e-06,
+      "loss": 1.1539,
+      "step": 158370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6693099528207344e-06,
+      "loss": 1.1583,
+      "step": 158375
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6687959940862773e-06,
+      "loss": 1.0266,
+      "step": 158380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6682820772169417e-06,
+      "loss": 1.4641,
+      "step": 158385
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6677682022156548e-06,
+      "loss": 1.2059,
+      "step": 158390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6672543690853558e-06,
+      "loss": 1.1155,
+      "step": 158395
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6667405778289813e-06,
+      "loss": 1.1094,
+      "step": 158400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.666226828449462e-06,
+      "loss": 1.1406,
+      "step": 158405
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.665713120949729e-06,
+      "loss": 0.9485,
+      "step": 158410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.665199455332719e-06,
+      "loss": 1.3279,
+      "step": 158415
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.664685831601368e-06,
+      "loss": 1.2141,
+      "step": 158420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.664172249758603e-06,
+      "loss": 1.5752,
+      "step": 158425
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.663658709807363e-06,
+      "loss": 1.342,
+      "step": 158430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.663145211750575e-06,
+      "loss": 1.3359,
+      "step": 158435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.662631755591174e-06,
+      "loss": 1.6977,
+      "step": 158440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.662118341332095e-06,
+      "loss": 1.4848,
+      "step": 158445
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6616049689762668e-06,
+      "loss": 1.2863,
+      "step": 158450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6610916385266184e-06,
+      "loss": 1.1734,
+      "step": 158455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6605783499860837e-06,
+      "loss": 1.3146,
+      "step": 158460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6600651033575973e-06,
+      "loss": 1.5428,
+      "step": 158465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.659551898644087e-06,
+      "loss": 1.2794,
+      "step": 158470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6590387358484817e-06,
+      "loss": 1.127,
+      "step": 158475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6585256149737148e-06,
+      "loss": 1.2233,
+      "step": 158480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6580125360227127e-06,
+      "loss": 1.1411,
+      "step": 158485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6574994989984084e-06,
+      "loss": 1.1263,
+      "step": 158490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6569865039037333e-06,
+      "loss": 1.1462,
+      "step": 158495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.656473550741616e-06,
+      "loss": 1.1902,
+      "step": 158500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6559606395149805e-06,
+      "loss": 1.0486,
+      "step": 158505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.65544777022676e-06,
+      "loss": 1.1818,
+      "step": 158510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6549349428798852e-06,
+      "loss": 1.2101,
+      "step": 158515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6544221574772833e-06,
+      "loss": 1.1323,
+      "step": 158520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.653909414021879e-06,
+      "loss": 1.4158,
+      "step": 158525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6533967125166027e-06,
+      "loss": 1.246,
+      "step": 158530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.652884052964385e-06,
+      "loss": 1.3744,
+      "step": 158535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6523714353681494e-06,
+      "loss": 1.3407,
+      "step": 158540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.651858859730827e-06,
+      "loss": 1.4676,
+      "step": 158545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6513463260553397e-06,
+      "loss": 1.302,
+      "step": 158550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6508338343446173e-06,
+      "loss": 1.0506,
+      "step": 158555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6503213846015894e-06,
+      "loss": 1.1215,
+      "step": 158560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.649808976829179e-06,
+      "loss": 0.9926,
+      "step": 158565
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.649296611030311e-06,
+      "loss": 1.034,
+      "step": 158570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6487842872079117e-06,
+      "loss": 1.3606,
+      "step": 158575
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6482720053649112e-06,
+      "loss": 1.2235,
+      "step": 158580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6477597655042297e-06,
+      "loss": 1.1844,
+      "step": 158585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6472475676287967e-06,
+      "loss": 1.2205,
+      "step": 158590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6467354117415344e-06,
+      "loss": 1.1915,
+      "step": 158595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.646223297845365e-06,
+      "loss": 1.0779,
+      "step": 158600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.645711225943216e-06,
+      "loss": 1.3993,
+      "step": 158605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6451991960380154e-06,
+      "loss": 1.0488,
+      "step": 158610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6446872081326814e-06,
+      "loss": 0.9912,
+      "step": 158615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6441752622301374e-06,
+      "loss": 1.3529,
+      "step": 158620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6436633583333092e-06,
+      "loss": 1.2617,
+      "step": 158625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6431514964451224e-06,
+      "loss": 1.3709,
+      "step": 158630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.642639676568495e-06,
+      "loss": 1.4196,
+      "step": 158635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6421278987063546e-06,
+      "loss": 1.1988,
+      "step": 158640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6416161628616187e-06,
+      "loss": 0.9805,
+      "step": 158645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6411044690372123e-06,
+      "loss": 0.9283,
+      "step": 158650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6405928172360595e-06,
+      "loss": 1.5072,
+      "step": 158655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6400812074610803e-06,
+      "loss": 1.1646,
+      "step": 158660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.639569639715197e-06,
+      "loss": 1.3029,
+      "step": 158665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.639058114001323e-06,
+      "loss": 1.1691,
+      "step": 158670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6385466303223928e-06,
+      "loss": 0.9973,
+      "step": 158675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6380351886813204e-06,
+      "loss": 1.4126,
+      "step": 158680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6375237890810233e-06,
+      "loss": 1.1632,
+      "step": 158685
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6370124315244292e-06,
+      "loss": 0.9623,
+      "step": 158690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6365011160144503e-06,
+      "loss": 1.3028,
+      "step": 158695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.635989842554011e-06,
+      "loss": 1.61,
+      "step": 158700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.635478611146034e-06,
+      "loss": 1.516,
+      "step": 158705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6349674217934353e-06,
+      "loss": 0.933,
+      "step": 158710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6344562744991298e-06,
+      "loss": 1.3322,
+      "step": 158715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6339451692660413e-06,
+      "loss": 1.4346,
+      "step": 158720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.63343410609709e-06,
+      "loss": 1.0951,
+      "step": 158725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6329230849951893e-06,
+      "loss": 1.2387,
+      "step": 158730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6324121059632634e-06,
+      "loss": 1.4128,
+      "step": 158735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.631901169004224e-06,
+      "loss": 1.2296,
+      "step": 158740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6313902741209953e-06,
+      "loss": 1.2946,
+      "step": 158745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.630879421316488e-06,
+      "loss": 1.0771,
+      "step": 158750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6303686105936254e-06,
+      "loss": 1.1413,
+      "step": 158755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6298578419553188e-06,
+      "loss": 1.4402,
+      "step": 158760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.629347115404489e-06,
+      "loss": 0.9573,
+      "step": 158765
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6288364309440537e-06,
+      "loss": 1.1257,
+      "step": 158770
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6283257885769263e-06,
+      "loss": 1.2299,
+      "step": 158775
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.627815188306022e-06,
+      "loss": 1.4471,
+      "step": 158780
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6273046301342597e-06,
+      "loss": 1.1267,
+      "step": 158785
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6267941140645516e-06,
+      "loss": 1.4637,
+      "step": 158790
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6262836400998136e-06,
+      "loss": 1.3376,
+      "step": 158795
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6257732082429665e-06,
+      "loss": 1.4921,
+      "step": 158800
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6252628184969196e-06,
+      "loss": 1.3203,
+      "step": 158805
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6247524708645856e-06,
+      "loss": 1.4016,
+      "step": 158810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6242421653488816e-06,
+      "loss": 1.2201,
+      "step": 158815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6237319019527253e-06,
+      "loss": 1.2134,
+      "step": 158820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6232216806790265e-06,
+      "loss": 0.9401,
+      "step": 158825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.622711501530698e-06,
+      "loss": 1.1885,
+      "step": 158830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6222013645106537e-06,
+      "loss": 1.7081,
+      "step": 158835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.621691269621811e-06,
+      "loss": 1.5149,
+      "step": 158840
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.621181216867076e-06,
+      "loss": 1.1241,
+      "step": 158845
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6206712062493687e-06,
+      "loss": 1.3689,
+      "step": 158850
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6201612377715947e-06,
+      "loss": 1.0189,
+      "step": 158855
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6196513114366694e-06,
+      "loss": 1.1982,
+      "step": 158860
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.619141427247507e-06,
+      "loss": 1.0058,
+      "step": 158865
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6186315852070177e-06,
+      "loss": 1.0243,
+      "step": 158870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6181217853181084e-06,
+      "loss": 1.2544,
+      "step": 158875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.617612027583695e-06,
+      "loss": 1.124,
+      "step": 158880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6171023120066897e-06,
+      "loss": 1.3802,
+      "step": 158885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.616592638590001e-06,
+      "loss": 1.0075,
+      "step": 158890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.616083007336536e-06,
+      "loss": 1.1291,
+      "step": 158895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.615573418249213e-06,
+      "loss": 1.5528,
+      "step": 158900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.615063871330934e-06,
+      "loss": 1.4435,
+      "step": 158905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6145543665846117e-06,
+      "loss": 1.6625,
+      "step": 158910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.61404490401316e-06,
+      "loss": 1.5014,
+      "step": 158915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6135354836194836e-06,
+      "loss": 1.3183,
+      "step": 158920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6130261054064898e-06,
+      "loss": 1.2787,
+      "step": 158925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6125167693770903e-06,
+      "loss": 1.0341,
+      "step": 158930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.612007475534196e-06,
+      "loss": 1.3992,
+      "step": 158935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.611498223880711e-06,
+      "loss": 1.2514,
+      "step": 158940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.610989014419547e-06,
+      "loss": 1.1313,
+      "step": 158945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6104798471536066e-06,
+      "loss": 1.1284,
+      "step": 158950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.609970722085804e-06,
+      "loss": 1.3069,
+      "step": 158955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6094616392190406e-06,
+      "loss": 1.1096,
+      "step": 158960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6089525985562293e-06,
+      "loss": 1.2442,
+      "step": 158965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6084436001002735e-06,
+      "loss": 1.2498,
+      "step": 158970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6079346438540744e-06,
+      "loss": 1.4133,
+      "step": 158975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6074257298205507e-06,
+      "loss": 1.2055,
+      "step": 158980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6069168580026016e-06,
+      "loss": 1.1314,
+      "step": 158985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6064080284031314e-06,
+      "loss": 1.1222,
+      "step": 158990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6058992410250506e-06,
+      "loss": 1.3016,
+      "step": 158995
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6053904958712585e-06,
+      "loss": 1.1265,
+      "step": 159000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6048817929446648e-06,
+      "loss": 1.1687,
+      "step": 159005
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.604373132248176e-06,
+      "loss": 1.3552,
+      "step": 159010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.6038645137846952e-06,
+      "loss": 1.2569,
+      "step": 159015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6033559375571216e-06,
+      "loss": 1.0447,
+      "step": 159020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.602847403568365e-06,
+      "loss": 1.1312,
+      "step": 159025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6023389118213317e-06,
+      "loss": 1.695,
+      "step": 159030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6018304623189206e-06,
+      "loss": 1.0915,
+      "step": 159035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.601322055064035e-06,
+      "loss": 1.2581,
+      "step": 159040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.600813690059579e-06,
+      "loss": 1.204,
+      "step": 159045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6003053673084598e-06,
+      "loss": 1.2971,
+      "step": 159050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5997970868135738e-06,
+      "loss": 1.1796,
+      "step": 159055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5992888485778302e-06,
+      "loss": 1.4359,
+      "step": 159060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.598780652604127e-06,
+      "loss": 1.7089,
+      "step": 159065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.598272498895361e-06,
+      "loss": 1.3039,
+      "step": 159070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5977643874544468e-06,
+      "loss": 1.2704,
+      "step": 159075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5972563182842793e-06,
+      "loss": 1.1285,
+      "step": 159080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.596748291387756e-06,
+      "loss": 1.3293,
+      "step": 159085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5962403067677855e-06,
+      "loss": 1.5723,
+      "step": 159090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5957323644272613e-06,
+      "loss": 1.0272,
+      "step": 159095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5952244643690916e-06,
+      "loss": 1.7019,
+      "step": 159100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.594716606596169e-06,
+      "loss": 1.3171,
+      "step": 159105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5942087911114012e-06,
+      "loss": 1.1505,
+      "step": 159110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.59370101791768e-06,
+      "loss": 1.5349,
+      "step": 159115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5931932870179113e-06,
+      "loss": 1.4637,
+      "step": 159120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5926855984149947e-06,
+      "loss": 1.4008,
+      "step": 159125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5921779521118283e-06,
+      "loss": 1.1684,
+      "step": 159130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5916703481113058e-06,
+      "loss": 1.2083,
+      "step": 159135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5911627864163313e-06,
+      "loss": 1.14,
+      "step": 159140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5906552670298046e-06,
+      "loss": 1.4313,
+      "step": 159145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.590147789954619e-06,
+      "loss": 1.3072,
+      "step": 159150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5896403551936767e-06,
+      "loss": 1.173,
+      "step": 159155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5891329627498718e-06,
+      "loss": 1.1095,
+      "step": 159160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5886256126261033e-06,
+      "loss": 1.485,
+      "step": 159165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5881183048252724e-06,
+      "loss": 1.063,
+      "step": 159170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5876110393502718e-06,
+      "loss": 1.3591,
+      "step": 159175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5871038162039986e-06,
+      "loss": 1.2013,
+      "step": 159180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.586596635389347e-06,
+      "loss": 1.119,
+      "step": 159185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5860894969092155e-06,
+      "loss": 1.2229,
+      "step": 159190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5855824007665042e-06,
+      "loss": 1.3087,
+      "step": 159195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.585075346964102e-06,
+      "loss": 1.1103,
+      "step": 159200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.584568335504911e-06,
+      "loss": 1.3325,
+      "step": 159205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5840613663918203e-06,
+      "loss": 1.2907,
+      "step": 159210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.583554439627727e-06,
+      "loss": 1.3039,
+      "step": 159215
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.583047555215531e-06,
+      "loss": 1.1326,
+      "step": 159220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5825407131581216e-06,
+      "loss": 1.0093,
+      "step": 159225
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5820339134583915e-06,
+      "loss": 1.2457,
+      "step": 159230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5815271561192377e-06,
+      "loss": 1.2402,
+      "step": 159235
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5810204411435567e-06,
+      "loss": 1.3915,
+      "step": 159240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5805137685342394e-06,
+      "loss": 1.3007,
+      "step": 159245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5800071382941772e-06,
+      "loss": 1.4568,
+      "step": 159250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.579500550426264e-06,
+      "loss": 1.1639,
+      "step": 159255
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.578994004933396e-06,
+      "loss": 1.194,
+      "step": 159260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.578487501818462e-06,
+      "loss": 1.4744,
+      "step": 159265
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.577981041084359e-06,
+      "loss": 1.2873,
+      "step": 159270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.577474622733975e-06,
+      "loss": 1.3351,
+      "step": 159275
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5769682467701984e-06,
+      "loss": 1.3035,
+      "step": 159280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.576461913195931e-06,
+      "loss": 1.5227,
+      "step": 159285
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.575955622014059e-06,
+      "loss": 1.219,
+      "step": 159290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.57544937322747e-06,
+      "loss": 1.3389,
+      "step": 159295
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5749431668390613e-06,
+      "loss": 1.3379,
+      "step": 159300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5744370028517184e-06,
+      "loss": 1.7101,
+      "step": 159305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.573930881268334e-06,
+      "loss": 1.0965,
+      "step": 159310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5734248020918008e-06,
+      "loss": 0.9888,
+      "step": 159315
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5729187653250076e-06,
+      "loss": 1.1645,
+      "step": 159320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.572412770970839e-06,
+      "loss": 1.2293,
+      "step": 159325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5719068190321883e-06,
+      "loss": 1.2399,
+      "step": 159330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5714009095119474e-06,
+      "loss": 1.1951,
+      "step": 159335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.570895042413004e-06,
+      "loss": 1.0502,
+      "step": 159340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.570389217738242e-06,
+      "loss": 1.332,
+      "step": 159345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5698834354905544e-06,
+      "loss": 1.5779,
+      "step": 159350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5693776956728313e-06,
+      "loss": 1.3246,
+      "step": 159355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.568871998287954e-06,
+      "loss": 1.1208,
+      "step": 159360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5683663433388185e-06,
+      "loss": 1.0982,
+      "step": 159365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5678607308283077e-06,
+      "loss": 1.2372,
+      "step": 159370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5673551607593038e-06,
+      "loss": 1.0529,
+      "step": 159375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.566849633134706e-06,
+      "loss": 1.2986,
+      "step": 159380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5663441479573946e-06,
+      "loss": 1.2708,
+      "step": 159385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5658387052302568e-06,
+      "loss": 1.5847,
+      "step": 159390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5653333049561748e-06,
+      "loss": 1.4317,
+      "step": 159395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.564827947138038e-06,
+      "loss": 0.9837,
+      "step": 159400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.564322631778736e-06,
+      "loss": 1.358,
+      "step": 159405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.563817358881149e-06,
+      "loss": 0.9063,
+      "step": 159410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5633121284481664e-06,
+      "loss": 1.034,
+      "step": 159415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.562806940482668e-06,
+      "loss": 0.9493,
+      "step": 159420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.562301794987543e-06,
+      "loss": 1.3953,
+      "step": 159425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.561796691965678e-06,
+      "loss": 1.0537,
+      "step": 159430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5612916314199545e-06,
+      "loss": 1.1106,
+      "step": 159435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.560786613353253e-06,
+      "loss": 1.3044,
+      "step": 159440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5602816377684614e-06,
+      "loss": 1.2279,
+      "step": 159445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.559776704668467e-06,
+      "loss": 1.1742,
+      "step": 159450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.559271814056146e-06,
+      "loss": 1.0461,
+      "step": 159455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5587669659343873e-06,
+      "loss": 1.5527,
+      "step": 159460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5582621603060687e-06,
+      "loss": 1.2555,
+      "step": 159465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55775739717408e-06,
+      "loss": 1.7819,
+      "step": 159470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5572526765412955e-06,
+      "loss": 1.1389,
+      "step": 159475
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5567479984106037e-06,
+      "loss": 1.3043,
+      "step": 159480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.556243362784885e-06,
+      "loss": 1.0911,
+      "step": 159485
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5557387696670167e-06,
+      "loss": 1.445,
+      "step": 159490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5552342190598846e-06,
+      "loss": 1.163,
+      "step": 159495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5547297109663717e-06,
+      "loss": 1.0132,
+      "step": 159500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5542252453893535e-06,
+      "loss": 1.223,
+      "step": 159505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.553720822331717e-06,
+      "loss": 1.1221,
+      "step": 159510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.553216441796337e-06,
+      "loss": 1.3296,
+      "step": 159515
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.552712103786096e-06,
+      "loss": 1.1384,
+      "step": 159520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.552207808303878e-06,
+      "loss": 1.402,
+      "step": 159525
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5517035553525583e-06,
+      "loss": 1.2505,
+      "step": 159530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.551199344935016e-06,
+      "loss": 1.0792,
+      "step": 159535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55069517705413e-06,
+      "loss": 1.3519,
+      "step": 159540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.550191051712785e-06,
+      "loss": 1.264,
+      "step": 159545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.549686968913856e-06,
+      "loss": 1.3023,
+      "step": 159550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.549182928660219e-06,
+      "loss": 1.2802,
+      "step": 159555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5486789309547542e-06,
+      "loss": 1.0828,
+      "step": 159560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5481749758003436e-06,
+      "loss": 1.141,
+      "step": 159565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.547671063199858e-06,
+      "loss": 1.1721,
+      "step": 159570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.547167193156184e-06,
+      "loss": 1.237,
+      "step": 159575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.546663365672192e-06,
+      "loss": 1.287,
+      "step": 159580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5461595807507556e-06,
+      "loss": 1.0702,
+      "step": 159585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.545655838394763e-06,
+      "loss": 1.2255,
+      "step": 159590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5451521386070855e-06,
+      "loss": 1.3459,
+      "step": 159595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5446484813905946e-06,
+      "loss": 1.3247,
+      "step": 159600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.544144866748175e-06,
+      "loss": 1.287,
+      "step": 159605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5436412946826948e-06,
+      "loss": 1.5011,
+      "step": 159610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.543137765197036e-06,
+      "loss": 1.176,
+      "step": 159615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5426342782940683e-06,
+      "loss": 1.1408,
+      "step": 159620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5421308339766736e-06,
+      "loss": 1.1687,
+      "step": 159625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.54162743224772e-06,
+      "loss": 1.2122,
+      "step": 159630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.541124073110084e-06,
+      "loss": 1.4552,
+      "step": 159635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5406207565666453e-06,
+      "loss": 1.3223,
+      "step": 159640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5401174826202735e-06,
+      "loss": 1.1251,
+      "step": 159645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.539614251273841e-06,
+      "loss": 1.3388,
+      "step": 159650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.539111062530223e-06,
+      "loss": 1.0043,
+      "step": 159655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.538607916392297e-06,
+      "loss": 1.4079,
+      "step": 159660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5381048128629305e-06,
+      "loss": 1.1085,
+      "step": 159665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5376017519450014e-06,
+      "loss": 1.0316,
+      "step": 159670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.53709873364138e-06,
+      "loss": 1.6785,
+      "step": 159675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.536595757954933e-06,
+      "loss": 1.4521,
+      "step": 159680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.536092824888543e-06,
+      "loss": 1.1016,
+      "step": 159685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5355899344450786e-06,
+      "loss": 1.2642,
+      "step": 159690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.53508708662741e-06,
+      "loss": 1.1129,
+      "step": 159695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5345842814384068e-06,
+      "loss": 1.5584,
+      "step": 159700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.534081518880943e-06,
+      "loss": 1.0933,
+      "step": 159705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.533578798957892e-06,
+      "loss": 1.2925,
+      "step": 159710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5330761216721177e-06,
+      "loss": 1.177,
+      "step": 159715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5325734870264996e-06,
+      "loss": 1.4856,
+      "step": 159720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5320708950238993e-06,
+      "loss": 1.0246,
+      "step": 159725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5315683456671904e-06,
+      "loss": 1.0183,
+      "step": 159730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5310658389592468e-06,
+      "loss": 1.148,
+      "step": 159735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5305633749029345e-06,
+      "loss": 1.2737,
+      "step": 159740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.53006095350112e-06,
+      "loss": 1.19,
+      "step": 159745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.529558574756674e-06,
+      "loss": 1.0569,
+      "step": 159750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5290562386724704e-06,
+      "loss": 1.4027,
+      "step": 159755
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.528553945251374e-06,
+      "loss": 1.8109,
+      "step": 159760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.52805169449625e-06,
+      "loss": 1.2274,
+      "step": 159765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5275494864099725e-06,
+      "loss": 1.2633,
+      "step": 159770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5270473209954037e-06,
+      "loss": 1.4197,
+      "step": 159775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5265451982554136e-06,
+      "loss": 1.0883,
+      "step": 159780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5260431181928736e-06,
+      "loss": 1.0121,
+      "step": 159785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.525541080810647e-06,
+      "loss": 1.9418,
+      "step": 159790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.525039086111598e-06,
+      "loss": 1.0185,
+      "step": 159795
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5245371340985967e-06,
+      "loss": 1.2656,
+      "step": 159800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5240352247745114e-06,
+      "loss": 0.662,
+      "step": 159805
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5235333581422026e-06,
+      "loss": 1.3403,
+      "step": 159810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.523031534204544e-06,
+      "loss": 1.1761,
+      "step": 159815
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.522529752964393e-06,
+      "loss": 1.2349,
+      "step": 159820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.522028014424619e-06,
+      "loss": 1.5901,
+      "step": 159825
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.52152631858809e-06,
+      "loss": 1.3978,
+      "step": 159830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5210246654576677e-06,
+      "loss": 1.2916,
+      "step": 159835
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5205230550362146e-06,
+      "loss": 1.2294,
+      "step": 159840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.520021487326597e-06,
+      "loss": 1.2754,
+      "step": 159845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5195199623316837e-06,
+      "loss": 1.4596,
+      "step": 159850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.519018480054334e-06,
+      "loss": 1.3706,
+      "step": 159855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.51851704049741e-06,
+      "loss": 1.1728,
+      "step": 159860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.518015643663778e-06,
+      "loss": 1.1268,
+      "step": 159865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.517514289556303e-06,
+      "loss": 1.0035,
+      "step": 159870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.517012978177843e-06,
+      "loss": 1.3669,
+      "step": 159875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5165117095312675e-06,
+      "loss": 1.0334,
+      "step": 159880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5160104836194345e-06,
+      "loss": 1.2776,
+      "step": 159885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5155093004452036e-06,
+      "loss": 0.9259,
+      "step": 159890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5150081600114405e-06,
+      "loss": 1.2151,
+      "step": 159895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.51450706232101e-06,
+      "loss": 1.2366,
+      "step": 159900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5140060073767703e-06,
+      "loss": 1.5009,
+      "step": 159905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5135049951815795e-06,
+      "loss": 1.1811,
+      "step": 159910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5130040257383015e-06,
+      "loss": 1.0444,
+      "step": 159915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5125030990498e-06,
+      "loss": 1.4259,
+      "step": 159920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5120022151189306e-06,
+      "loss": 1.393,
+      "step": 159925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5115013739485594e-06,
+      "loss": 1.5743,
+      "step": 159930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.511000575541539e-06,
+      "loss": 1.185,
+      "step": 159935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5104998199007334e-06,
+      "loss": 1.2288,
+      "step": 159940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5099991070290066e-06,
+      "loss": 1.4358,
+      "step": 159945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5094984369292118e-06,
+      "loss": 1.1251,
+      "step": 159950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.508997809604207e-06,
+      "loss": 1.2805,
+      "step": 159955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.508497225056854e-06,
+      "loss": 0.9494,
+      "step": 159960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.507996683290014e-06,
+      "loss": 1.2194,
+      "step": 159965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5074961843065402e-06,
+      "loss": 1.0939,
+      "step": 159970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.506995728109295e-06,
+      "loss": 1.0145,
+      "step": 159975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.506495314701135e-06,
+      "loss": 1.6767,
+      "step": 159980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.505994944084914e-06,
+      "loss": 1.3416,
+      "step": 159985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.505494616263493e-06,
+      "loss": 1.2672,
+      "step": 159990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5049943312397307e-06,
+      "loss": 1.0854,
+      "step": 159995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.504494089016483e-06,
+      "loss": 0.9096,
+      "step": 160000
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 1.2214573621749878,
+      "eval_runtime": 6463.4208,
+      "eval_samples_per_second": 3.576,
+      "eval_steps_per_second": 1.788,
+      "step": 160000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.503993889596602e-06,
+      "loss": 1.2029,
+      "step": 160005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5034937329829465e-06,
+      "loss": 1.2474,
+      "step": 160010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5029936191783786e-06,
+      "loss": 1.0499,
+      "step": 160015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.502493548185746e-06,
+      "loss": 1.1192,
+      "step": 160020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5019935200079106e-06,
+      "loss": 1.4667,
+      "step": 160025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.501493534647722e-06,
+      "loss": 1.5632,
+      "step": 160030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5009935921080385e-06,
+      "loss": 1.1061,
+      "step": 160035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5004936923917176e-06,
+      "loss": 1.2785,
+      "step": 160040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4999938355016116e-06,
+      "loss": 1.6448,
+      "step": 160045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4994940214405717e-06,
+      "loss": 1.1799,
+      "step": 160050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4989942502114552e-06,
+      "loss": 1.0388,
+      "step": 160055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4984945218171185e-06,
+      "loss": 1.1423,
+      "step": 160060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4979948362604135e-06,
+      "loss": 1.5012,
+      "step": 160065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4974951935441895e-06,
+      "loss": 1.3244,
+      "step": 160070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.496995593671306e-06,
+      "loss": 1.2051,
+      "step": 160075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.49649603664461e-06,
+      "loss": 1.3028,
+      "step": 160080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.495996522466958e-06,
+      "loss": 1.2854,
+      "step": 160085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4954970511412047e-06,
+      "loss": 0.9058,
+      "step": 160090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4949976226702e-06,
+      "loss": 1.1128,
+      "step": 160095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4944982370567917e-06,
+      "loss": 1.2472,
+      "step": 160100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4939988943038364e-06,
+      "loss": 1.3859,
+      "step": 160105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.493499594414186e-06,
+      "loss": 1.0808,
+      "step": 160110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4930003373906884e-06,
+      "loss": 1.1414,
+      "step": 160115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4925011232361996e-06,
+      "loss": 1.2574,
+      "step": 160120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.492001951953563e-06,
+      "loss": 1.1173,
+      "step": 160125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4915028235456374e-06,
+      "loss": 1.0308,
+      "step": 160130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4910037380152662e-06,
+      "loss": 1.1826,
+      "step": 160135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4905046953653057e-06,
+      "loss": 1.2591,
+      "step": 160140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4900056955985985e-06,
+      "loss": 1.0878,
+      "step": 160145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4895067387179995e-06,
+      "loss": 1.2549,
+      "step": 160150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.489007824726358e-06,
+      "loss": 0.9017,
+      "step": 160155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4885089536265216e-06,
+      "loss": 1.1657,
+      "step": 160160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4880101254213374e-06,
+      "loss": 1.0507,
+      "step": 160165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4875113401136586e-06,
+      "loss": 1.0014,
+      "step": 160170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4870125977063263e-06,
+      "loss": 0.9103,
+      "step": 160175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.486513898202194e-06,
+      "loss": 1.0877,
+      "step": 160180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4860152416041116e-06,
+      "loss": 1.1874,
+      "step": 160185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4855166279149235e-06,
+      "loss": 1.2235,
+      "step": 160190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.485018057137474e-06,
+      "loss": 0.9709,
+      "step": 160195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.484519529274613e-06,
+      "loss": 1.2239,
+      "step": 160200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4840210443291913e-06,
+      "loss": 0.9265,
+      "step": 160205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4835226023040514e-06,
+      "loss": 0.9974,
+      "step": 160210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.483024203202037e-06,
+      "loss": 1.063,
+      "step": 160215
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.482525847025998e-06,
+      "loss": 1.1517,
+      "step": 160220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4820275337787824e-06,
+      "loss": 1.2844,
+      "step": 160225
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4815292634632303e-06,
+      "loss": 1.3092,
+      "step": 160230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.481031036082193e-06,
+      "loss": 1.4906,
+      "step": 160235
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4805328516385096e-06,
+      "loss": 1.2517,
+      "step": 160240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4800347101350275e-06,
+      "loss": 1.2663,
+      "step": 160245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4795366115745955e-06,
+      "loss": 1.1035,
+      "step": 160250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.479038555960054e-06,
+      "loss": 1.6011,
+      "step": 160255
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4785405432942443e-06,
+      "loss": 0.997,
+      "step": 160260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.478042573580014e-06,
+      "loss": 1.0511,
+      "step": 160265
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.477544646820209e-06,
+      "loss": 1.0361,
+      "step": 160270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4770467630176696e-06,
+      "loss": 1.1952,
+      "step": 160275
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.476548922175237e-06,
+      "loss": 1.3057,
+      "step": 160280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.47605112429576e-06,
+      "loss": 1.1216,
+      "step": 160285
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4755533693820743e-06,
+      "loss": 1.0732,
+      "step": 160290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4750556574370256e-06,
+      "loss": 1.3691,
+      "step": 160295
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4745579884634596e-06,
+      "loss": 1.2626,
+      "step": 160300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4740603624642157e-06,
+      "loss": 1.6251,
+      "step": 160305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4735627794421314e-06,
+      "loss": 1.3899,
+      "step": 160310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.473065239400051e-06,
+      "loss": 1.1449,
+      "step": 160315
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.472567742340821e-06,
+      "loss": 1.3418,
+      "step": 160320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4720702882672743e-06,
+      "loss": 1.0572,
+      "step": 160325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4715728771822577e-06,
+      "loss": 1.2265,
+      "step": 160330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4710755090886064e-06,
+      "loss": 1.2176,
+      "step": 160335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.470578183989166e-06,
+      "loss": 1.1618,
+      "step": 160340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4700809018867723e-06,
+      "loss": 1.1162,
+      "step": 160345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4695836627842685e-06,
+      "loss": 1.1438,
+      "step": 160350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4690864666844916e-06,
+      "loss": 1.6987,
+      "step": 160355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.468589313590276e-06,
+      "loss": 1.2376,
+      "step": 160360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.468092203504473e-06,
+      "loss": 1.0565,
+      "step": 160365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4675951364299146e-06,
+      "loss": 1.0749,
+      "step": 160370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4670981123694347e-06,
+      "loss": 1.3135,
+      "step": 160375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.466601131325881e-06,
+      "loss": 1.2911,
+      "step": 160380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4661041933020823e-06,
+      "loss": 1.3015,
+      "step": 160385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4656072983008815e-06,
+      "loss": 0.8815,
+      "step": 160390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4651104463251187e-06,
+      "loss": 0.9982,
+      "step": 160395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4646136373776275e-06,
+      "loss": 1.298,
+      "step": 160400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4641168714612418e-06,
+      "loss": 1.175,
+      "step": 160405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4636201485788037e-06,
+      "loss": 1.2834,
+      "step": 160410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4631234687331496e-06,
+      "loss": 1.1689,
+      "step": 160415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.462626831927114e-06,
+      "loss": 1.5505,
+      "step": 160420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.46213023816353e-06,
+      "loss": 1.2052,
+      "step": 160425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4616336874452384e-06,
+      "loss": 1.3851,
+      "step": 160430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.461137179775074e-06,
+      "loss": 1.1727,
+      "step": 160435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.46064071515587e-06,
+      "loss": 1.3179,
+      "step": 160440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.460144293590464e-06,
+      "loss": 1.3053,
+      "step": 160445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4596479150816866e-06,
+      "loss": 1.0627,
+      "step": 160450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.459151579632376e-06,
+      "loss": 1.1896,
+      "step": 160455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4586552872453685e-06,
+      "loss": 1.2502,
+      "step": 160460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.458159037923494e-06,
+      "loss": 1.3768,
+      "step": 160465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.457662831669586e-06,
+      "loss": 1.1645,
+      "step": 160470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.457166668486484e-06,
+      "loss": 1.0539,
+      "step": 160475
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4566705483770127e-06,
+      "loss": 1.0392,
+      "step": 160480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.456174471344013e-06,
+      "loss": 0.9895,
+      "step": 160485
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.455678437390312e-06,
+      "loss": 1.1874,
+      "step": 160490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.455182446518747e-06,
+      "loss": 1.19,
+      "step": 160495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.454686498732146e-06,
+      "loss": 1.2435,
+      "step": 160500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.454190594033343e-06,
+      "loss": 1.0408,
+      "step": 160505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4536947324251735e-06,
+      "loss": 1.1245,
+      "step": 160510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4531989139104662e-06,
+      "loss": 1.1665,
+      "step": 160515
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4527031384920486e-06,
+      "loss": 1.4361,
+      "step": 160520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4522074061727553e-06,
+      "loss": 1.0027,
+      "step": 160525
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4517117169554205e-06,
+      "loss": 1.1984,
+      "step": 160530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.451216070842869e-06,
+      "loss": 1.2878,
+      "step": 160535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.450720467837937e-06,
+      "loss": 0.9925,
+      "step": 160540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4502249079434483e-06,
+      "loss": 1.1945,
+      "step": 160545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4497293911622356e-06,
+      "loss": 1.1431,
+      "step": 160550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.449233917497131e-06,
+      "loss": 1.0447,
+      "step": 160555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4487384869509643e-06,
+      "loss": 1.0983,
+      "step": 160560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4482430995265604e-06,
+      "loss": 1.3568,
+      "step": 160565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.447747755226746e-06,
+      "loss": 0.892,
+      "step": 160570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4472524540543587e-06,
+      "loss": 1.2031,
+      "step": 160575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4467571960122216e-06,
+      "loss": 1.127,
+      "step": 160580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4462619811031617e-06,
+      "loss": 1.2533,
+      "step": 160585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.44576680933001e-06,
+      "loss": 1.033,
+      "step": 160590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4452716806955922e-06,
+      "loss": 1.3633,
+      "step": 160595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4447765952027346e-06,
+      "loss": 1.147,
+      "step": 160600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4442815528542696e-06,
+      "loss": 1.26,
+      "step": 160605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4437865536530214e-06,
+      "loss": 1.309,
+      "step": 160610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4432915976018125e-06,
+      "loss": 0.9044,
+      "step": 160615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4427966847034735e-06,
+      "loss": 1.266,
+      "step": 160620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4423018149608326e-06,
+      "loss": 1.0558,
+      "step": 160625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4418069883767137e-06,
+      "loss": 1.4554,
+      "step": 160630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4413122049539383e-06,
+      "loss": 1.1811,
+      "step": 160635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4408174646953364e-06,
+      "loss": 1.1999,
+      "step": 160640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4403227676037357e-06,
+      "loss": 1.4014,
+      "step": 160645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4398281136819544e-06,
+      "loss": 0.8966,
+      "step": 160650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4393335029328245e-06,
+      "loss": 1.5463,
+      "step": 160655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.438838935359166e-06,
+      "loss": 1.1916,
+      "step": 160660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.438344410963799e-06,
+      "loss": 1.3642,
+      "step": 160665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.437849929749557e-06,
+      "loss": 1.1114,
+      "step": 160670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.43735549171926e-06,
+      "loss": 1.2609,
+      "step": 160675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4368610968757277e-06,
+      "loss": 1.2317,
+      "step": 160680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4363667452217897e-06,
+      "loss": 1.2827,
+      "step": 160685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4358724367602627e-06,
+      "loss": 0.9965,
+      "step": 160690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4353781714939715e-06,
+      "loss": 1.1727,
+      "step": 160695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4348839494257436e-06,
+      "loss": 1.0661,
+      "step": 160700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4343897705583975e-06,
+      "loss": 1.299,
+      "step": 160705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4338956348947507e-06,
+      "loss": 1.0925,
+      "step": 160710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.43340154243763e-06,
+      "loss": 1.3757,
+      "step": 160715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.43290749318986e-06,
+      "loss": 1.3234,
+      "step": 160720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.432413487154257e-06,
+      "loss": 1.3539,
+      "step": 160725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4319195243336413e-06,
+      "loss": 1.4096,
+      "step": 160730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4314256047308347e-06,
+      "loss": 1.2394,
+      "step": 160735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4309317283486623e-06,
+      "loss": 1.4366,
+      "step": 160740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4304378951899365e-06,
+      "loss": 1.1885,
+      "step": 160745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4299441052574858e-06,
+      "loss": 1.1243,
+      "step": 160750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.429450358554125e-06,
+      "loss": 1.0699,
+      "step": 160755
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4289566550826695e-06,
+      "loss": 1.2674,
+      "step": 160760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.428462994845949e-06,
+      "loss": 1.7309,
+      "step": 160765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4279693778467772e-06,
+      "loss": 0.8463,
+      "step": 160770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4274758040879732e-06,
+      "loss": 1.4103,
+      "step": 160775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4269822735723526e-06,
+      "loss": 1.0951,
+      "step": 160780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4264887863027352e-06,
+      "loss": 1.1977,
+      "step": 160785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.425995342281945e-06,
+      "loss": 1.1886,
+      "step": 160790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.42550194151279e-06,
+      "loss": 1.4707,
+      "step": 160795
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4250085839980974e-06,
+      "loss": 1.3464,
+      "step": 160800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.424515269740676e-06,
+      "loss": 1.1688,
+      "step": 160805
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4240219987433476e-06,
+      "loss": 1.5835,
+      "step": 160810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4235287710089316e-06,
+      "loss": 1.096,
+      "step": 160815
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.423035586540241e-06,
+      "loss": 1.1054,
+      "step": 160820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.422542445340089e-06,
+      "loss": 1.3132,
+      "step": 160825
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.422049347411296e-06,
+      "loss": 0.9217,
+      "step": 160830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.421556292756679e-06,
+      "loss": 0.9372,
+      "step": 160835
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.421063281379049e-06,
+      "loss": 0.9624,
+      "step": 160840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4205703132812276e-06,
+      "loss": 1.3098,
+      "step": 160845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.420077388466023e-06,
+      "loss": 0.9842,
+      "step": 160850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.419584506936257e-06,
+      "loss": 1.5145,
+      "step": 160855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4190916686947365e-06,
+      "loss": 1.2888,
+      "step": 160860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4185988737442833e-06,
+      "loss": 1.1669,
+      "step": 160865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4181061220877087e-06,
+      "loss": 1.1199,
+      "step": 160870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.417613413727823e-06,
+      "loss": 1.3178,
+      "step": 160875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.417120748667442e-06,
+      "loss": 1.3186,
+      "step": 160880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.416628126909384e-06,
+      "loss": 1.3108,
+      "step": 160885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4161355484564553e-06,
+      "loss": 1.3393,
+      "step": 160890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4156430133114738e-06,
+      "loss": 0.9431,
+      "step": 160895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4151505214772473e-06,
+      "loss": 1.1873,
+      "step": 160900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4146580729565906e-06,
+      "loss": 1.0883,
+      "step": 160905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4141656677523196e-06,
+      "loss": 1.0476,
+      "step": 160910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.413673305867242e-06,
+      "loss": 1.3656,
+      "step": 160915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.413180987304168e-06,
+      "loss": 1.164,
+      "step": 160920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4126887120659105e-06,
+      "loss": 1.3898,
+      "step": 160925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4121964801552843e-06,
+      "loss": 1.321,
+      "step": 160930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.411704291575098e-06,
+      "loss": 1.0029,
+      "step": 160935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.411212146328158e-06,
+      "loss": 1.4815,
+      "step": 160940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4107200444172784e-06,
+      "loss": 1.4112,
+      "step": 160945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4102279858452715e-06,
+      "loss": 1.2172,
+      "step": 160950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4097359706149436e-06,
+      "loss": 1.5833,
+      "step": 160955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.409243998729107e-06,
+      "loss": 1.2423,
+      "step": 160960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.408752070190571e-06,
+      "loss": 1.2787,
+      "step": 160965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.408260185002138e-06,
+      "loss": 1.1874,
+      "step": 160970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4077683431666276e-06,
+      "loss": 1.2028,
+      "step": 160975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.407276544686844e-06,
+      "loss": 1.2075,
+      "step": 160980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4067847895655916e-06,
+      "loss": 0.9651,
+      "step": 160985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.406293077805686e-06,
+      "loss": 1.2781,
+      "step": 160990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.405801409409927e-06,
+      "loss": 1.0755,
+      "step": 160995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.40530978438113e-06,
+      "loss": 1.6391,
+      "step": 161000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.404818202722097e-06,
+      "loss": 1.4563,
+      "step": 161005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.404326664435639e-06,
+      "loss": 1.284,
+      "step": 161010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4038351695245588e-06,
+      "loss": 1.2185,
+      "step": 161015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4033437179916643e-06,
+      "loss": 1.1799,
+      "step": 161020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.402852309839767e-06,
+      "loss": 1.2501,
+      "step": 161025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.402360945071669e-06,
+      "loss": 1.2713,
+      "step": 161030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.401869623690173e-06,
+      "loss": 1.16,
+      "step": 161035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4013783456980887e-06,
+      "loss": 1.3079,
+      "step": 161040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4008871110982233e-06,
+      "loss": 0.9108,
+      "step": 161045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4003959198933767e-06,
+      "loss": 1.0002,
+      "step": 161050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.39990477208636e-06,
+      "loss": 1.2855,
+      "step": 161055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.399413667679975e-06,
+      "loss": 1.0207,
+      "step": 161060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3989226066770233e-06,
+      "loss": 1.1484,
+      "step": 161065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3984315890803113e-06,
+      "loss": 1.203,
+      "step": 161070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3979406148926466e-06,
+      "loss": 1.1573,
+      "step": 161075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3974496841168295e-06,
+      "loss": 0.9951,
+      "step": 161080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.396958796755662e-06,
+      "loss": 1.0125,
+      "step": 161085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.396467952811947e-06,
+      "loss": 1.0527,
+      "step": 161090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.3959771522884945e-06,
+      "loss": 1.1922,
+      "step": 161095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.395486395188098e-06,
+      "loss": 1.0239,
+      "step": 161100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3949956815135678e-06,
+      "loss": 1.32,
+      "step": 161105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3945050112676994e-06,
+      "loss": 1.2063,
+      "step": 161110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.394014384453298e-06,
+      "loss": 1.0549,
+      "step": 161115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3935238010731687e-06,
+      "loss": 1.2011,
+      "step": 161120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.39303326113011e-06,
+      "loss": 1.1619,
+      "step": 161125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3925427646269197e-06,
+      "loss": 1.2519,
+      "step": 161130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3920523115664006e-06,
+      "loss": 1.2822,
+      "step": 161135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3915619019513593e-06,
+      "loss": 1.3058,
+      "step": 161140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3910715357845906e-06,
+      "loss": 1.0772,
+      "step": 161145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3905812130688934e-06,
+      "loss": 1.1858,
+      "step": 161150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3900909338070698e-06,
+      "loss": 1.2618,
+      "step": 161155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3896006980019238e-06,
+      "loss": 1.1384,
+      "step": 161160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.389110505656248e-06,
+      "loss": 1.2018,
+      "step": 161165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3886203567728473e-06,
+      "loss": 1.4424,
+      "step": 161170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.388130251354518e-06,
+      "loss": 1.4171,
+      "step": 161175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.387640189404057e-06,
+      "loss": 1.3408,
+      "step": 161180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.387150170924264e-06,
+      "loss": 1.2965,
+      "step": 161185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3866601959179415e-06,
+      "loss": 1.2664,
+      "step": 161190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3861702643878804e-06,
+      "loss": 1.1441,
+      "step": 161195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3856803763368864e-06,
+      "loss": 1.0076,
+      "step": 161200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3851905317677482e-06,
+      "loss": 1.1425,
+      "step": 161205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3847007306832715e-06,
+      "loss": 1.0828,
+      "step": 161210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3842109730862473e-06,
+      "loss": 1.0489,
+      "step": 161215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3837212589794767e-06,
+      "loss": 1.1724,
+      "step": 161220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.383231588365752e-06,
+      "loss": 1.4715,
+      "step": 161225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3827419612478708e-06,
+      "loss": 1.2726,
+      "step": 161230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.382252377628633e-06,
+      "loss": 0.9121,
+      "step": 161235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3817628375108316e-06,
+      "loss": 1.1475,
+      "step": 161240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.381273340897259e-06,
+      "loss": 1.0052,
+      "step": 161245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.380783887790713e-06,
+      "loss": 1.1741,
+      "step": 161250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.380294478193993e-06,
+      "loss": 1.0369,
+      "step": 161255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.379805112109885e-06,
+      "loss": 1.3014,
+      "step": 161260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3793157895411933e-06,
+      "loss": 1.1413,
+      "step": 161265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.378826510490706e-06,
+      "loss": 1.3265,
+      "step": 161270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.378337274961213e-06,
+      "loss": 1.3791,
+      "step": 161275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3778480829555193e-06,
+      "loss": 1.418,
+      "step": 161280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3773589344764126e-06,
+      "loss": 1.3042,
+      "step": 161285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3768698295266857e-06,
+      "loss": 1.2007,
+      "step": 161290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.37638076810913e-06,
+      "loss": 1.2243,
+      "step": 161295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.37589175022654e-06,
+      "loss": 1.356,
+      "step": 161300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3754027758817123e-06,
+      "loss": 0.9888,
+      "step": 161305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3749138450774324e-06,
+      "loss": 1.2725,
+      "step": 161310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.374424957816498e-06,
+      "loss": 1.4708,
+      "step": 161315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.373936114101696e-06,
+      "loss": 1.205,
+      "step": 161320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3734473139358206e-06,
+      "loss": 1.7783,
+      "step": 161325
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.372958557321666e-06,
+      "loss": 1.073,
+      "step": 161330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3724698442620208e-06,
+      "loss": 1.3424,
+      "step": 161335
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3719811747596713e-06,
+      "loss": 1.1755,
+      "step": 161340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3714925488174124e-06,
+      "loss": 1.0491,
+      "step": 161345
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3710039664380368e-06,
+      "loss": 1.1762,
+      "step": 161350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3705154276243325e-06,
+      "loss": 1.2677,
+      "step": 161355
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3700269323790846e-06,
+      "loss": 1.2942,
+      "step": 161360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3695384807050903e-06,
+      "loss": 1.27,
+      "step": 161365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3690500726051322e-06,
+      "loss": 1.2415,
+      "step": 161370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3685617080820035e-06,
+      "loss": 1.3716,
+      "step": 161375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.368073387138494e-06,
+      "loss": 1.1739,
+      "step": 161380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3675851097773904e-06,
+      "loss": 1.3012,
+      "step": 161385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.367096876001478e-06,
+      "loss": 1.033,
+      "step": 161390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3666086858135484e-06,
+      "loss": 1.0019,
+      "step": 161395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3661205392163913e-06,
+      "loss": 1.384,
+      "step": 161400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3656324362127893e-06,
+      "loss": 1.2016,
+      "step": 161405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.365144376805534e-06,
+      "loss": 1.2076,
+      "step": 161410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364656360997408e-06,
+      "loss": 1.1762,
+      "step": 161415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3641683887912027e-06,
+      "loss": 0.9581,
+      "step": 161420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3636804601897034e-06,
+      "loss": 1.0969,
+      "step": 161425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3631925751956976e-06,
+      "loss": 1.3473,
+      "step": 161430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3627047338119657e-06,
+      "loss": 1.0199,
+      "step": 161435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3622169360412984e-06,
+      "loss": 1.272,
+      "step": 161440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3617291818864827e-06,
+      "loss": 1.0622,
+      "step": 161445
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3612414713503016e-06,
+      "loss": 1.8339,
+      "step": 161450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.360753804435537e-06,
+      "loss": 1.0698,
+      "step": 161455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3602661811449812e-06,
+      "loss": 1.3732,
+      "step": 161460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3597786014814108e-06,
+      "loss": 1.2439,
+      "step": 161465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.359291065447613e-06,
+      "loss": 1.3565,
+      "step": 161470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3588035730463775e-06,
+      "loss": 1.3696,
+      "step": 161475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.358316124280482e-06,
+      "loss": 1.1563,
+      "step": 161480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3578287191527092e-06,
+      "loss": 1.0872,
+      "step": 161485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3573413576658457e-06,
+      "loss": 1.2543,
+      "step": 161490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.356854039822675e-06,
+      "loss": 1.4179,
+      "step": 161495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3563667656259794e-06,
+      "loss": 0.9333,
+      "step": 161500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3558795350785366e-06,
+      "loss": 1.1772,
+      "step": 161505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.355392348183134e-06,
+      "loss": 1.1885,
+      "step": 161510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.354905204942557e-06,
+      "loss": 1.0385,
+      "step": 161515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3544181053595784e-06,
+      "loss": 1.2476,
+      "step": 161520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3539310494369884e-06,
+      "loss": 0.9101,
+      "step": 161525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.353444037177561e-06,
+      "loss": 1.3107,
+      "step": 161530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.352957068584082e-06,
+      "loss": 1.1413,
+      "step": 161535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3524701436593336e-06,
+      "loss": 1.025,
+      "step": 161540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3519832624060936e-06,
+      "loss": 1.0743,
+      "step": 161545
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3514964248271387e-06,
+      "loss": 1.3891,
+      "step": 161550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.351009630925255e-06,
+      "loss": 1.3088,
+      "step": 161555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.350522880703222e-06,
+      "loss": 1.0859,
+      "step": 161560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3500361741638154e-06,
+      "loss": 1.0554,
+      "step": 161565
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3495495113098187e-06,
+      "loss": 1.512,
+      "step": 161570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3490628921440106e-06,
+      "loss": 1.2474,
+      "step": 161575
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3485763166691644e-06,
+      "loss": 1.205,
+      "step": 161580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3480897848880634e-06,
+      "loss": 1.1032,
+      "step": 161585
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3476032968034877e-06,
+      "loss": 1.1282,
+      "step": 161590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3471168524182133e-06,
+      "loss": 1.1288,
+      "step": 161595
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3466304517350147e-06,
+      "loss": 0.7797,
+      "step": 161600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3461440947566727e-06,
+      "loss": 1.0038,
+      "step": 161605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3456577814859673e-06,
+      "loss": 1.2743,
+      "step": 161610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3451715119256703e-06,
+      "loss": 1.2041,
+      "step": 161615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.344685286078564e-06,
+      "loss": 1.0693,
+      "step": 161620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.344199103947419e-06,
+      "loss": 1.0967,
+      "step": 161625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3437129655350157e-06,
+      "loss": 1.2117,
+      "step": 161630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3432268708441307e-06,
+      "loss": 1.2769,
+      "step": 161635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3427408198775392e-06,
+      "loss": 1.1561,
+      "step": 161640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.342254812638016e-06,
+      "loss": 1.1562,
+      "step": 161645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3417688491283307e-06,
+      "loss": 1.1072,
+      "step": 161650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3412829293512706e-06,
+      "loss": 1.2524,
+      "step": 161655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.340797053309605e-06,
+      "loss": 1.1786,
+      "step": 161660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.340311221006103e-06,
+      "loss": 1.098,
+      "step": 161665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3398254324435475e-06,
+      "loss": 1.2808,
+      "step": 161670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3393396876247053e-06,
+      "loss": 1.2992,
+      "step": 161675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.338853986552355e-06,
+      "loss": 1.454,
+      "step": 161680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.338368329229271e-06,
+      "loss": 1.3369,
+      "step": 161685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3378827156582252e-06,
+      "loss": 1.0085,
+      "step": 161690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3373971458419863e-06,
+      "loss": 1.246,
+      "step": 161695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3369116197833318e-06,
+      "loss": 1.3926,
+      "step": 161700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.336426137485036e-06,
+      "loss": 1.1966,
+      "step": 161705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3359406989498665e-06,
+      "loss": 1.1157,
+      "step": 161710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3354553041806005e-06,
+      "loss": 1.2553,
+      "step": 161715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.334969953180004e-06,
+      "loss": 1.1822,
+      "step": 161720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3344846459508554e-06,
+      "loss": 1.2559,
+      "step": 161725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3339993824959184e-06,
+      "loss": 1.139,
+      "step": 161730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3335141628179714e-06,
+      "loss": 1.1052,
+      "step": 161735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3330289869197785e-06,
+      "loss": 1.3346,
+      "step": 161740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.332543854804115e-06,
+      "loss": 1.28,
+      "step": 161745
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3320587664737525e-06,
+      "loss": 1.1573,
+      "step": 161750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.331573721931458e-06,
+      "loss": 1.1768,
+      "step": 161755
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3310887211799992e-06,
+      "loss": 1.0831,
+      "step": 161760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3306037642221525e-06,
+      "loss": 1.1418,
+      "step": 161765
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.33011885106068e-06,
+      "loss": 1.2651,
+      "step": 161770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3296339816983537e-06,
+      "loss": 1.2764,
+      "step": 161775
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3291491561379455e-06,
+      "loss": 1.0574,
+      "step": 161780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3286643743822223e-06,
+      "loss": 1.1354,
+      "step": 161785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3281796364339484e-06,
+      "loss": 1.1562,
+      "step": 161790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.327694942295895e-06,
+      "loss": 1.5427,
+      "step": 161795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.327210291970833e-06,
+      "loss": 1.4699,
+      "step": 161800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3267256854615262e-06,
+      "loss": 1.4222,
+      "step": 161805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3262411227707404e-06,
+      "loss": 1.0922,
+      "step": 161810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3257566039012457e-06,
+      "loss": 1.2411,
+      "step": 161815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3252721288558113e-06,
+      "loss": 1.1742,
+      "step": 161820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3247876976371974e-06,
+      "loss": 1.5557,
+      "step": 161825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.324303310248177e-06,
+      "loss": 1.1077,
+      "step": 161830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.323818966691511e-06,
+      "loss": 1.6146,
+      "step": 161835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3233346669699663e-06,
+      "loss": 1.4362,
+      "step": 161840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3228504110863127e-06,
+      "loss": 1.2298,
+      "step": 161845
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.322366199043312e-06,
+      "loss": 0.958,
+      "step": 161850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3218820308437282e-06,
+      "loss": 1.077,
+      "step": 161855
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3213979064903268e-06,
+      "loss": 1.237,
+      "step": 161860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3209138259858766e-06,
+      "loss": 1.2725,
+      "step": 161865
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3204297893331374e-06,
+      "loss": 1.1613,
+      "step": 161870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3199457965348726e-06,
+      "loss": 1.2097,
+      "step": 161875
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.319461847593851e-06,
+      "loss": 1.3788,
+      "step": 161880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.318977942512831e-06,
+      "loss": 1.0275,
+      "step": 161885
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.318494081294577e-06,
+      "loss": 1.3734,
+      "step": 161890
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3180102639418567e-06,
+      "loss": 1.0803,
+      "step": 161895
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.31752649045743e-06,
+      "loss": 1.0217,
+      "step": 161900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.317042760844055e-06,
+      "loss": 1.3699,
+      "step": 161905
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3165590751044996e-06,
+      "loss": 1.02,
+      "step": 161910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.316075433241526e-06,
+      "loss": 1.0956,
+      "step": 161915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.315591835257892e-06,
+      "loss": 1.3244,
+      "step": 161920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.315108281156365e-06,
+      "loss": 1.1567,
+      "step": 161925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3146247709397006e-06,
+      "loss": 1.2504,
+      "step": 161930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3141413046106607e-06,
+      "loss": 1.6985,
+      "step": 161935
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3136578821720124e-06,
+      "loss": 1.0945,
+      "step": 161940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3131745036265105e-06,
+      "loss": 1.5429,
+      "step": 161945
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.312691168976917e-06,
+      "loss": 1.5276,
+      "step": 161950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.312207878225986e-06,
+      "loss": 1.2358,
+      "step": 161955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3117246313764887e-06,
+      "loss": 1.5523,
+      "step": 161960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.311241428431179e-06,
+      "loss": 0.9128,
+      "step": 161965
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3107582693928133e-06,
+      "loss": 1.4854,
+      "step": 161970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3102751542641557e-06,
+      "loss": 1.0591,
+      "step": 161975
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3097920830479604e-06,
+      "loss": 1.5572,
+      "step": 161980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.309309055746989e-06,
+      "loss": 1.7045,
+      "step": 161985
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.308826072364002e-06,
+      "loss": 1.0909,
+      "step": 161990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.308343132901755e-06,
+      "loss": 1.4184,
+      "step": 161995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.307860237363002e-06,
+      "loss": 1.0109,
+      "step": 162000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3073773857505044e-06,
+      "loss": 1.2685,
+      "step": 162005
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3068945780670227e-06,
+      "loss": 1.0943,
+      "step": 162010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3064118143153113e-06,
+      "loss": 1.0913,
+      "step": 162015
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.305929094498123e-06,
+      "loss": 1.2147,
+      "step": 162020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.305446418618218e-06,
+      "loss": 1.1401,
+      "step": 162025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.304963786678355e-06,
+      "loss": 1.2141,
+      "step": 162030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.304481198681284e-06,
+      "loss": 1.2852,
+      "step": 162035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.303998654629769e-06,
+      "loss": 1.3603,
+      "step": 162040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3035161545265593e-06,
+      "loss": 1.5661,
+      "step": 162045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.303033698374407e-06,
+      "loss": 1.4101,
+      "step": 162050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.302551286176078e-06,
+      "loss": 1.5633,
+      "step": 162055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3020689179343216e-06,
+      "loss": 1.0169,
+      "step": 162060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.301586593651889e-06,
+      "loss": 1.0658,
+      "step": 162065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.301104313331539e-06,
+      "loss": 1.0392,
+      "step": 162070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3006220769760234e-06,
+      "loss": 1.3315,
+      "step": 162075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.300139884588095e-06,
+      "loss": 1.1998,
+      "step": 162080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2996577361705143e-06,
+      "loss": 1.3034,
+      "step": 162085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2991756317260273e-06,
+      "loss": 1.121,
+      "step": 162090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2986935712573877e-06,
+      "loss": 1.3797,
+      "step": 162095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.29821155476735e-06,
+      "loss": 1.1095,
+      "step": 162100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.29772958225867e-06,
+      "loss": 1.0783,
+      "step": 162105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.297247653734096e-06,
+      "loss": 1.7012,
+      "step": 162110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2967657691963774e-06,
+      "loss": 1.1543,
+      "step": 162115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.296283928648271e-06,
+      "loss": 1.1634,
+      "step": 162120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2958021320925284e-06,
+      "loss": 1.3471,
+      "step": 162125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2953203795318967e-06,
+      "loss": 1.5208,
+      "step": 162130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2948386709691327e-06,
+      "loss": 1.18,
+      "step": 162135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.294357006406982e-06,
+      "loss": 1.2948,
+      "step": 162140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2938753858481965e-06,
+      "loss": 1.4369,
+      "step": 162145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.29339380929553e-06,
+      "loss": 1.1135,
+      "step": 162150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2929122767517296e-06,
+      "loss": 1.2332,
+      "step": 162155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2924307882195464e-06,
+      "loss": 1.3422,
+      "step": 162160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.291949343701726e-06,
+      "loss": 1.1875,
+      "step": 162165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2914679432010203e-06,
+      "loss": 1.1617,
+      "step": 162170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.290986586720181e-06,
+      "loss": 0.9998,
+      "step": 162175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2905052742619528e-06,
+      "loss": 1.0092,
+      "step": 162180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.290024005829088e-06,
+      "loss": 1.1581,
+      "step": 162185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.28954278142433e-06,
+      "loss": 1.1365,
+      "step": 162190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.289061601050431e-06,
+      "loss": 1.5359,
+      "step": 162195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2885804647101396e-06,
+      "loss": 1.297,
+      "step": 162200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2880993724062006e-06,
+      "loss": 1.6189,
+      "step": 162205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.287618324141361e-06,
+      "loss": 1.1465,
+      "step": 162210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2871373199183676e-06,
+      "loss": 1.0538,
+      "step": 162215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2866563597399716e-06,
+      "loss": 1.0766,
+      "step": 162220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2861754436089136e-06,
+      "loss": 1.5292,
+      "step": 162225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2856945715279454e-06,
+      "loss": 1.0877,
+      "step": 162230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.285213743499808e-06,
+      "loss": 1.2454,
+      "step": 162235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2847329595272515e-06,
+      "loss": 1.3745,
+      "step": 162240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.284252219613018e-06,
+      "loss": 1.1236,
+      "step": 162245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2837715237598557e-06,
+      "loss": 1.2697,
+      "step": 162250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2832908719705094e-06,
+      "loss": 1.0006,
+      "step": 162255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2828102642477156e-06,
+      "loss": 1.2901,
+      "step": 162260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2823297005942325e-06,
+      "loss": 1.0675,
+      "step": 162265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2818491810127986e-06,
+      "loss": 1.1015,
+      "step": 162270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2813687055061528e-06,
+      "loss": 1.2637,
+      "step": 162275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2808882740770466e-06,
+      "loss": 1.1104,
+      "step": 162280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2804078867282178e-06,
+      "loss": 1.1257,
+      "step": 162285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.27992754346241e-06,
+      "loss": 1.2232,
+      "step": 162290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2794472442823733e-06,
+      "loss": 1.2118,
+      "step": 162295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.278966989190844e-06,
+      "loss": 1.5216,
+      "step": 162300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2784867781905638e-06,
+      "loss": 1.0225,
+      "step": 162305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.278006611284277e-06,
+      "loss": 1.2813,
+      "step": 162310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2775264884747282e-06,
+      "loss": 1.1242,
+      "step": 162315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2770464097646574e-06,
+      "loss": 1.2094,
+      "step": 162320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2765663751568024e-06,
+      "loss": 1.3548,
+      "step": 162325
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.276086384653906e-06,
+      "loss": 1.2585,
+      "step": 162330
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.275606438258715e-06,
+      "loss": 1.0861,
+      "step": 162335
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.275126535973963e-06,
+      "loss": 1.2888,
+      "step": 162340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2746466778023956e-06,
+      "loss": 1.3872,
+      "step": 162345
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.274166863746752e-06,
+      "loss": 1.2369,
+      "step": 162350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.273687093809763e-06,
+      "loss": 1.3365,
+      "step": 162355
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2732073679941836e-06,
+      "loss": 1.3782,
+      "step": 162360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.272727686302746e-06,
+      "loss": 1.579,
+      "step": 162365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.272248048738186e-06,
+      "loss": 1.1006,
+      "step": 162370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2717684553032493e-06,
+      "loss": 1.4487,
+      "step": 162375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2712889060006683e-06,
+      "loss": 1.3212,
+      "step": 162380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2708094008331883e-06,
+      "loss": 1.4426,
+      "step": 162385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2703299398035395e-06,
+      "loss": 1.2215,
+      "step": 162390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2698505229144673e-06,
+      "loss": 1.3123,
+      "step": 162395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2693711501687044e-06,
+      "loss": 1.2395,
+      "step": 162400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.268891821568989e-06,
+      "loss": 1.0575,
+      "step": 162405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.268412537118063e-06,
+      "loss": 1.1111,
+      "step": 162410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2679332968186607e-06,
+      "loss": 1.3936,
+      "step": 162415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.267454100673514e-06,
+      "loss": 1.0321,
+      "step": 162420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.266974948685363e-06,
+      "loss": 1.1354,
+      "step": 162425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2664958408569483e-06,
+      "loss": 0.949,
+      "step": 162430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.266016777190998e-06,
+      "loss": 1.3881,
+      "step": 162435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.265537757690255e-06,
+      "loss": 1.3765,
+      "step": 162440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.265058782357451e-06,
+      "loss": 1.1357,
+      "step": 162445
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2645798511953188e-06,
+      "loss": 0.9858,
+      "step": 162450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.264100964206597e-06,
+      "loss": 1.3539,
+      "step": 162455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2636221213940214e-06,
+      "loss": 1.214,
+      "step": 162460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.263143322760325e-06,
+      "loss": 1.2974,
+      "step": 162465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2626645683082384e-06,
+      "loss": 1.0584,
+      "step": 162470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2621858580404986e-06,
+      "loss": 1.0791,
+      "step": 162475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2617071919598423e-06,
+      "loss": 1.0678,
+      "step": 162480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2612285700689973e-06,
+      "loss": 1.5823,
+      "step": 162485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.260749992370702e-06,
+      "loss": 1.3202,
+      "step": 162490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2602714588676842e-06,
+      "loss": 1.2787,
+      "step": 162495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2597929695626795e-06,
+      "loss": 1.201,
+      "step": 162500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2593145244584226e-06,
+      "loss": 1.106,
+      "step": 162505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.258836123557643e-06,
+      "loss": 1.5223,
+      "step": 162510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2583577668630696e-06,
+      "loss": 1.3653,
+      "step": 162515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2578794543774373e-06,
+      "loss": 1.0994,
+      "step": 162520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.257401186103481e-06,
+      "loss": 1.2324,
+      "step": 162525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.256922962043928e-06,
+      "loss": 0.8718,
+      "step": 162530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2564447822015055e-06,
+      "loss": 1.3357,
+      "step": 162535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.25596664657895e-06,
+      "loss": 1.2643,
+      "step": 162540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.255488555178992e-06,
+      "loss": 1.2334,
+      "step": 162545
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.255010508004357e-06,
+      "loss": 1.2217,
+      "step": 162550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2545325050577806e-06,
+      "loss": 0.99,
+      "step": 162555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2540545463419894e-06,
+      "loss": 1.3022,
+      "step": 162560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2535766318597096e-06,
+      "loss": 1.1373,
+      "step": 162565
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2530987616136747e-06,
+      "loss": 1.2111,
+      "step": 162570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2526209356066154e-06,
+      "loss": 1.0259,
+      "step": 162575
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2521431538412542e-06,
+      "loss": 1.0558,
+      "step": 162580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2516654163203257e-06,
+      "loss": 1.2008,
+      "step": 162585
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.251187723046552e-06,
+      "loss": 1.2062,
+      "step": 162590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2507100740226673e-06,
+      "loss": 1.099,
+      "step": 162595
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.250232469251392e-06,
+      "loss": 1.109,
+      "step": 162600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.249754908735462e-06,
+      "loss": 1.227,
+      "step": 162605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.249277392477597e-06,
+      "loss": 1.2847,
+      "step": 162610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2487999204805257e-06,
+      "loss": 1.0141,
+      "step": 162615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2483224927469793e-06,
+      "loss": 1.212,
+      "step": 162620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2478451092796817e-06,
+      "loss": 1.428,
+      "step": 162625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2473677700813533e-06,
+      "loss": 1.1775,
+      "step": 162630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.246890475154726e-06,
+      "loss": 1.18,
+      "step": 162635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.246413224502527e-06,
+      "loss": 1.2195,
+      "step": 162640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2459360181274746e-06,
+      "loss": 1.2027,
+      "step": 162645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.245458856032302e-06,
+      "loss": 1.4649,
+      "step": 162650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2449817382197302e-06,
+      "loss": 1.2427,
+      "step": 162655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2445046646924785e-06,
+      "loss": 1.056,
+      "step": 162660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.244027635453281e-06,
+      "loss": 1.185,
+      "step": 162665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.243550650504858e-06,
+      "loss": 1.1535,
+      "step": 162670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2430737098499323e-06,
+      "loss": 1.4747,
+      "step": 162675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.242596813491226e-06,
+      "loss": 1.28,
+      "step": 162680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2421199614314625e-06,
+      "loss": 1.3997,
+      "step": 162685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.241643153673371e-06,
+      "loss": 1.26,
+      "step": 162690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.241166390219666e-06,
+      "loss": 1.425,
+      "step": 162695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2406896710730772e-06,
+      "loss": 1.099,
+      "step": 162700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.240212996236321e-06,
+      "loss": 1.0985,
+      "step": 162705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2397363657121218e-06,
+      "loss": 1.3168,
+      "step": 162710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2392597795032047e-06,
+      "loss": 1.3569,
+      "step": 162715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2387832376122885e-06,
+      "loss": 1.0092,
+      "step": 162720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.238306740042092e-06,
+      "loss": 1.0294,
+      "step": 162725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2378302867953385e-06,
+      "loss": 1.382,
+      "step": 162730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2373538778747515e-06,
+      "loss": 1.262,
+      "step": 162735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2368775132830502e-06,
+      "loss": 1.2972,
+      "step": 162740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.236401193022949e-06,
+      "loss": 1.4074,
+      "step": 162745
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2359249170971764e-06,
+      "loss": 1.4779,
+      "step": 162750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2354486855084456e-06,
+      "loss": 1.4584,
+      "step": 162755
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.234972498259479e-06,
+      "loss": 0.9154,
+      "step": 162760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2344963553529984e-06,
+      "loss": 1.2213,
+      "step": 162765
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.234020256791721e-06,
+      "loss": 1.1476,
+      "step": 162770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2335442025783616e-06,
+      "loss": 1.1618,
+      "step": 162775
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.233068192715643e-06,
+      "loss": 1.1145,
+      "step": 162780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2325922272062837e-06,
+      "loss": 1.0858,
+      "step": 162785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2321163060529983e-06,
+      "loss": 1.1022,
+      "step": 162790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2316404292585093e-06,
+      "loss": 1.206,
+      "step": 162795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.231164596825529e-06,
+      "loss": 1.0954,
+      "step": 162800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.230688808756778e-06,
+      "loss": 1.3595,
+      "step": 162805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2302130650549747e-06,
+      "loss": 1.2338,
+      "step": 162810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.229737365722835e-06,
+      "loss": 1.2635,
+      "step": 162815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2292617107630697e-06,
+      "loss": 1.2522,
+      "step": 162820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2287861001784005e-06,
+      "loss": 0.9999,
+      "step": 162825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.228310533971545e-06,
+      "loss": 1.0889,
+      "step": 162830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.227835012145216e-06,
+      "loss": 1.0662,
+      "step": 162835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.227359534702128e-06,
+      "loss": 1.4403,
+      "step": 162840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.226884101644996e-06,
+      "loss": 1.12,
+      "step": 162845
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.22640871297654e-06,
+      "loss": 1.4772,
+      "step": 162850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.225933368699469e-06,
+      "loss": 1.1562,
+      "step": 162855
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2254580688165017e-06,
+      "loss": 1.066,
+      "step": 162860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.224982813330351e-06,
+      "loss": 1.3895,
+      "step": 162865
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.224507602243726e-06,
+      "loss": 1.246,
+      "step": 162870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.224032435559347e-06,
+      "loss": 1.2374,
+      "step": 162875
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.223557313279926e-06,
+      "loss": 1.1604,
+      "step": 162880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.223082235408176e-06,
+      "loss": 1.3617,
+      "step": 162885
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.222607201946807e-06,
+      "loss": 1.3217,
+      "step": 162890
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2221322128985332e-06,
+      "loss": 1.2162,
+      "step": 162895
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2216572682660707e-06,
+      "loss": 1.1411,
+      "step": 162900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.221182368052126e-06,
+      "loss": 1.1297,
+      "step": 162905
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.220707512259417e-06,
+      "loss": 1.435,
+      "step": 162910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2202327008906488e-06,
+      "loss": 1.2877,
+      "step": 162915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2197579339485377e-06,
+      "loss": 1.261,
+      "step": 162920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2192832114357945e-06,
+      "loss": 1.2222,
+      "step": 162925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.21880853335513e-06,
+      "loss": 1.1895,
+      "step": 162930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2183338997092506e-06,
+      "loss": 1.2133,
+      "step": 162935
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2178593105008705e-06,
+      "loss": 1.3919,
+      "step": 162940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.217384765732702e-06,
+      "loss": 1.447,
+      "step": 162945
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2169102654074493e-06,
+      "loss": 1.1323,
+      "step": 162950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2164358095278284e-06,
+      "loss": 1.064,
+      "step": 162955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2159613980965444e-06,
+      "loss": 1.2071,
+      "step": 162960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2154870311163055e-06,
+      "loss": 1.2708,
+      "step": 162965
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2150127085898232e-06,
+      "loss": 1.1906,
+      "step": 162970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.214538430519807e-06,
+      "loss": 1.6264,
+      "step": 162975
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.214064196908965e-06,
+      "loss": 1.1543,
+      "step": 162980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2135900077600005e-06,
+      "loss": 1.498,
+      "step": 162985
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2131158630756257e-06,
+      "loss": 1.1187,
+      "step": 162990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2126417628585495e-06,
+      "loss": 1.4534,
+      "step": 162995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2121677071114742e-06,
+      "loss": 1.19,
+      "step": 163000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.211693695837114e-06,
+      "loss": 1.2081,
+      "step": 163005
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2112197290381686e-06,
+      "loss": 1.4694,
+      "step": 163010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.210745806717347e-06,
+      "loss": 1.2207,
+      "step": 163015
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.21027192887736e-06,
+      "loss": 1.5334,
+      "step": 163020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2097980955209096e-06,
+      "loss": 1.1046,
+      "step": 163025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2093243066507018e-06,
+      "loss": 0.9713,
+      "step": 163030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.208850562269438e-06,
+      "loss": 1.2378,
+      "step": 163035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.208376862379832e-06,
+      "loss": 1.5336,
+      "step": 163040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2079032069845852e-06,
+      "loss": 1.1533,
+      "step": 163045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2074295960864e-06,
+      "loss": 1.1877,
+      "step": 163050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2069560296879854e-06,
+      "loss": 1.1559,
+      "step": 163055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2064825077920403e-06,
+      "loss": 1.2088,
+      "step": 163060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2060090304012726e-06,
+      "loss": 1.4165,
+      "step": 163065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2055355975183877e-06,
+      "loss": 1.3648,
+      "step": 163070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2050622091460872e-06,
+      "loss": 1.0777,
+      "step": 163075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2045888652870707e-06,
+      "loss": 1.0146,
+      "step": 163080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.204115565944045e-06,
+      "loss": 1.0811,
+      "step": 163085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.203642311119715e-06,
+      "loss": 1.154,
+      "step": 163090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.203169100816778e-06,
+      "loss": 1.1143,
+      "step": 163095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.202695935037942e-06,
+      "loss": 1.3115,
+      "step": 163100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2022228137859025e-06,
+      "loss": 1.1433,
+      "step": 163105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2017497370633677e-06,
+      "loss": 1.0685,
+      "step": 163110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2012767048730345e-06,
+      "loss": 1.5067,
+      "step": 163115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.200803717217608e-06,
+      "loss": 1.3746,
+      "step": 163120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.200330774099784e-06,
+      "loss": 1.0632,
+      "step": 163125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1998578755222666e-06,
+      "loss": 1.4279,
+      "step": 163130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1993850214877587e-06,
+      "loss": 1.0944,
+      "step": 163135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1989122119989582e-06,
+      "loss": 1.4311,
+      "step": 163140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1984394470585614e-06,
+      "loss": 1.1633,
+      "step": 163145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.197966726669275e-06,
+      "loss": 1.2257,
+      "step": 163150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1974940508337916e-06,
+      "loss": 1.0713,
+      "step": 163155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1970214195548135e-06,
+      "loss": 1.1982,
+      "step": 163160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.196548832835044e-06,
+      "loss": 1.0147,
+      "step": 163165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.196076290677176e-06,
+      "loss": 1.252,
+      "step": 163170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1956037930839082e-06,
+      "loss": 1.0153,
+      "step": 163175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1951313400579397e-06,
+      "loss": 1.2198,
+      "step": 163180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.194658931601973e-06,
+      "loss": 1.3653,
+      "step": 163185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1941865677187e-06,
+      "loss": 0.9002,
+      "step": 163190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.193714248410819e-06,
+      "loss": 1.2838,
+      "step": 163195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1932419736810286e-06,
+      "loss": 1.1989,
+      "step": 163200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.192769743532026e-06,
+      "loss": 1.0001,
+      "step": 163205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.192297557966506e-06,
+      "loss": 1.0717,
+      "step": 163210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.191825416987169e-06,
+      "loss": 1.0339,
+      "step": 163215
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1913533205967063e-06,
+      "loss": 1.2018,
+      "step": 163220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1908812687978156e-06,
+      "loss": 0.9397,
+      "step": 163225
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.190409261593195e-06,
+      "loss": 1.0935,
+      "step": 163230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1899372989855385e-06,
+      "loss": 1.2427,
+      "step": 163235
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.189465380977538e-06,
+      "loss": 1.3288,
+      "step": 163240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1889935075718916e-06,
+      "loss": 1.2203,
+      "step": 163245
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.188521678771295e-06,
+      "loss": 1.541,
+      "step": 163250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.188049894578442e-06,
+      "loss": 1.0315,
+      "step": 163255
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.187578154996023e-06,
+      "loss": 1.0645,
+      "step": 163260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1871064600267365e-06,
+      "loss": 1.0569,
+      "step": 163265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1866348096732725e-06,
+      "loss": 0.9901,
+      "step": 163270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1861632039383253e-06,
+      "loss": 1.497,
+      "step": 163275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.185691642824592e-06,
+      "loss": 1.1322,
+      "step": 163280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.185220126334763e-06,
+      "loss": 0.975,
+      "step": 163285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.184748654471527e-06,
+      "loss": 1.2681,
+      "step": 163290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.18427722723758e-06,
+      "loss": 1.095,
+      "step": 163295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.183805844635616e-06,
+      "loss": 1.4241,
+      "step": 163300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.183334506668322e-06,
+      "loss": 1.2792,
+      "step": 163305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1828632133383953e-06,
+      "loss": 1.3449,
+      "step": 163310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1823919646485203e-06,
+      "loss": 1.2596,
+      "step": 163315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.181920760601396e-06,
+      "loss": 1.3584,
+      "step": 163320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1814496011997056e-06,
+      "loss": 1.5701,
+      "step": 163325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.180978486446146e-06,
+      "loss": 1.1715,
+      "step": 163330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.180507416343405e-06,
+      "loss": 1.117,
+      "step": 163335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.180036390894167e-06,
+      "loss": 1.1159,
+      "step": 163340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1795654101011323e-06,
+      "loss": 1.0196,
+      "step": 163345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.179094473966985e-06,
+      "loss": 1.1976,
+      "step": 163350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1786235824944123e-06,
+      "loss": 1.2393,
+      "step": 163355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.178152735686109e-06,
+      "loss": 1.1654,
+      "step": 163360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1776819335447573e-06,
+      "loss": 1.817,
+      "step": 163365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1772111760730485e-06,
+      "loss": 1.26,
+      "step": 163370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1767404632736757e-06,
+      "loss": 1.0968,
+      "step": 163375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1762697951493205e-06,
+      "loss": 1.1288,
+      "step": 163380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1757991717026715e-06,
+      "loss": 1.0571,
+      "step": 163385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1753285929364175e-06,
+      "loss": 1.1028,
+      "step": 163390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1748580588532474e-06,
+      "loss": 1.3729,
+      "step": 163395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.174387569455848e-06,
+      "loss": 1.266,
+      "step": 163400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1739171247469017e-06,
+      "loss": 1.0017,
+      "step": 163405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.173446724729097e-06,
+      "loss": 1.0417,
+      "step": 163410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1729763694051243e-06,
+      "loss": 1.0572,
+      "step": 163415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1725060587776636e-06,
+      "loss": 1.0222,
+      "step": 163420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.172035792849405e-06,
+      "loss": 1.1129,
+      "step": 163425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.171565571623031e-06,
+      "loss": 1.3074,
+      "step": 163430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1710953951012282e-06,
+      "loss": 0.9385,
+      "step": 163435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1706252632866832e-06,
+      "loss": 1.1327,
+      "step": 163440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1701551761820795e-06,
+      "loss": 1.4031,
+      "step": 163445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1696851337900983e-06,
+      "loss": 1.1745,
+      "step": 163450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1692151361134293e-06,
+      "loss": 1.3648,
+      "step": 163455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.16874518315475e-06,
+      "loss": 1.2574,
+      "step": 163460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1682752749167526e-06,
+      "loss": 1.2705,
+      "step": 163465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1678054114021107e-06,
+      "loss": 1.2975,
+      "step": 163470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1673355926135165e-06,
+      "loss": 1.1241,
+      "step": 163475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.166865818553646e-06,
+      "loss": 1.0619,
+      "step": 163480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1663960892251844e-06,
+      "loss": 1.1438,
+      "step": 163485
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.165926404630817e-06,
+      "loss": 1.2003,
+      "step": 163490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.165456764773224e-06,
+      "loss": 1.0597,
+      "step": 163495
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.164987169655084e-06,
+      "loss": 1.3903,
+      "step": 163500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1645176192790807e-06,
+      "loss": 1.1436,
+      "step": 163505
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1640481136478996e-06,
+      "loss": 1.3108,
+      "step": 163510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1635786527642156e-06,
+      "loss": 1.1512,
+      "step": 163515
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.163109236630715e-06,
+      "loss": 1.41,
+      "step": 163520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1626398652500724e-06,
+      "loss": 1.2018,
+      "step": 163525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1621705386249713e-06,
+      "loss": 1.2103,
+      "step": 163530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1617012567580952e-06,
+      "loss": 1.1281,
+      "step": 163535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1612320196521217e-06,
+      "loss": 1.0874,
+      "step": 163540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1607628273097283e-06,
+      "loss": 1.2913,
+      "step": 163545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1602936797335906e-06,
+      "loss": 1.0157,
+      "step": 163550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.159824576926398e-06,
+      "loss": 1.3799,
+      "step": 163555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.159355518890823e-06,
+      "loss": 0.7783,
+      "step": 163560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.158886505629544e-06,
+      "loss": 1.8364,
+      "step": 163565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.158417537145242e-06,
+      "loss": 1.1117,
+      "step": 163570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1579486134405916e-06,
+      "loss": 0.9805,
+      "step": 163575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.157479734518272e-06,
+      "loss": 1.1782,
+      "step": 163580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1570109003809637e-06,
+      "loss": 1.5261,
+      "step": 163585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.156542111031341e-06,
+      "loss": 1.2204,
+      "step": 163590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.15607336647208e-06,
+      "loss": 1.0644,
+      "step": 163595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1556046667058583e-06,
+      "loss": 1.4958,
+      "step": 163600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.155136011735356e-06,
+      "loss": 1.4209,
+      "step": 163605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.154667401563246e-06,
+      "loss": 1.0463,
+      "step": 163610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.154198836192203e-06,
+      "loss": 1.0872,
+      "step": 163615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1537303156249034e-06,
+      "loss": 1.2699,
+      "step": 163620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.153261839864027e-06,
+      "loss": 1.1295,
+      "step": 163625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1527934089122437e-06,
+      "loss": 1.0173,
+      "step": 163630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1523250227722324e-06,
+      "loss": 1.2579,
+      "step": 163635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1518566814466667e-06,
+      "loss": 1.3214,
+      "step": 163640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1513883849382157e-06,
+      "loss": 1.3509,
+      "step": 163645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1509201332495623e-06,
+      "loss": 1.4306,
+      "step": 163650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.150451926383378e-06,
+      "loss": 1.1541,
+      "step": 163655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1499837643423315e-06,
+      "loss": 1.1421,
+      "step": 163660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1495156471291035e-06,
+      "loss": 1.3772,
+      "step": 163665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.14904757474636e-06,
+      "loss": 1.1868,
+      "step": 163670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1485795471967773e-06,
+      "loss": 1.1737,
+      "step": 163675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.148111564483032e-06,
+      "loss": 1.2343,
+      "step": 163680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1476436266077915e-06,
+      "loss": 1.0865,
+      "step": 163685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1471757335737275e-06,
+      "loss": 1.1341,
+      "step": 163690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1467078853835135e-06,
+      "loss": 0.973,
+      "step": 163695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1462400820398243e-06,
+      "loss": 1.1313,
+      "step": 163700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1457723235453286e-06,
+      "loss": 1.3898,
+      "step": 163705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.145304609902694e-06,
+      "loss": 1.3253,
+      "step": 163710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1448369411145954e-06,
+      "loss": 1.1582,
+      "step": 163715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1443693171837042e-06,
+      "loss": 1.1977,
+      "step": 163720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.143901738112688e-06,
+      "loss": 1.492,
+      "step": 163725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.143434203904221e-06,
+      "loss": 1.0952,
+      "step": 163730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1429667145609702e-06,
+      "loss": 1.4773,
+      "step": 163735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1424992700855994e-06,
+      "loss": 1.8335,
+      "step": 163740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1420318704807896e-06,
+      "loss": 1.1875,
+      "step": 163745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1415645157492037e-06,
+      "loss": 1.3082,
+      "step": 163750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.141097205893512e-06,
+      "loss": 1.537,
+      "step": 163755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.14062994091638e-06,
+      "loss": 1.3246,
+      "step": 163760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.140162720820477e-06,
+      "loss": 1.1524,
+      "step": 163765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1396955456084757e-06,
+      "loss": 1.4129,
+      "step": 163770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1392284152830365e-06,
+      "loss": 1.0945,
+      "step": 163775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1387613298468356e-06,
+      "loss": 1.1146,
+      "step": 163780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1382942893025316e-06,
+      "loss": 1.1054,
+      "step": 163785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1378272936527968e-06,
+      "loss": 1.115,
+      "step": 163790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1373603429002986e-06,
+      "loss": 1.1096,
+      "step": 163795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1368934370477024e-06,
+      "loss": 1.3226,
+      "step": 163800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.13642657609767e-06,
+      "loss": 1.153,
+      "step": 163805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1359597600528723e-06,
+      "loss": 1.3993,
+      "step": 163810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.135492988915977e-06,
+      "loss": 1.2817,
+      "step": 163815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1350262626896433e-06,
+      "loss": 1.2173,
+      "step": 163820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.134559581376543e-06,
+      "loss": 1.3877,
+      "step": 163825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1340929449793347e-06,
+      "loss": 1.4968,
+      "step": 163830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1336263535006896e-06,
+      "loss": 1.0924,
+      "step": 163835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.133159806943266e-06,
+      "loss": 1.3679,
+      "step": 163840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1326933053097342e-06,
+      "loss": 1.2206,
+      "step": 163845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1322268486027542e-06,
+      "loss": 1.1384,
+      "step": 163850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.131760436824989e-06,
+      "loss": 1.1667,
+      "step": 163855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.131294069979103e-06,
+      "loss": 1.3788,
+      "step": 163860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.130827748067763e-06,
+      "loss": 1.0697,
+      "step": 163865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.130361471093626e-06,
+      "loss": 1.415,
+      "step": 163870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.129895239059361e-06,
+      "loss": 1.2754,
+      "step": 163875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1294290519676233e-06,
+      "loss": 1.205,
+      "step": 163880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1289629098210797e-06,
+      "loss": 1.2475,
+      "step": 163885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1284968126223938e-06,
+      "loss": 1.1183,
+      "step": 163890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1280307603742255e-06,
+      "loss": 1.0655,
+      "step": 163895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.127564753079232e-06,
+      "loss": 1.1377,
+      "step": 163900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1270987907400785e-06,
+      "loss": 1.1864,
+      "step": 163905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1266328733594277e-06,
+      "loss": 1.4333,
+      "step": 163910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1261670009399383e-06,
+      "loss": 1.2467,
+      "step": 163915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.125701173484267e-06,
+      "loss": 1.6096,
+      "step": 163920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.125235390995077e-06,
+      "loss": 1.1235,
+      "step": 163925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1247696534750316e-06,
+      "loss": 1.1271,
+      "step": 163930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1243039609267846e-06,
+      "loss": 1.2051,
+      "step": 163935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1238383133530017e-06,
+      "loss": 1.268,
+      "step": 163940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1233727107563362e-06,
+      "loss": 1.3894,
+      "step": 163945
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1229071531394454e-06,
+      "loss": 1.1018,
+      "step": 163950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.122441640504995e-06,
+      "loss": 1.5824,
+      "step": 163955
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.121976172855641e-06,
+      "loss": 0.9513,
+      "step": 163960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.121510750194037e-06,
+      "loss": 1.172,
+      "step": 163965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1210453725228475e-06,
+      "loss": 1.3752,
+      "step": 163970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.120580039844723e-06,
+      "loss": 1.058,
+      "step": 163975
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.120114752162328e-06,
+      "loss": 1.6303,
+      "step": 163980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.119649509478313e-06,
+      "loss": 1.466,
+      "step": 163985
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.11918431179534e-06,
+      "loss": 1.1184,
+      "step": 163990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1187191591160616e-06,
+      "loss": 1.1796,
+      "step": 163995
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.118254051443135e-06,
+      "loss": 1.4495,
+      "step": 164000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.11778898877922e-06,
+      "loss": 1.4247,
+      "step": 164005
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.117323971126969e-06,
+      "loss": 1.3583,
+      "step": 164010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1168589984890355e-06,
+      "loss": 1.3506,
+      "step": 164015
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1163940708680763e-06,
+      "loss": 1.2828,
+      "step": 164020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.115929188266751e-06,
+      "loss": 1.2146,
+      "step": 164025
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.115464350687707e-06,
+      "loss": 1.5418,
+      "step": 164030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.114999558133606e-06,
+      "loss": 1.2614,
+      "step": 164035
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1145348106070974e-06,
+      "loss": 1.073,
+      "step": 164040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.114070108110834e-06,
+      "loss": 0.8939,
+      "step": 164045
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.113605450647471e-06,
+      "loss": 1.3098,
+      "step": 164050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.113140838219666e-06,
+      "loss": 1.3136,
+      "step": 164055
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1126762708300684e-06,
+      "loss": 1.2588,
+      "step": 164060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.11221174848133e-06,
+      "loss": 1.1804,
+      "step": 164065
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1117472711761033e-06,
+      "loss": 1.9094,
+      "step": 164070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.111282838917046e-06,
+      "loss": 1.391,
+      "step": 164075
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1108184517068044e-06,
+      "loss": 1.3611,
+      "step": 164080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.110354109548034e-06,
+      "loss": 1.0918,
+      "step": 164085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.109889812443383e-06,
+      "loss": 1.5172,
+      "step": 164090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.109425560395506e-06,
+      "loss": 1.0932,
+      "step": 164095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.108961353407054e-06,
+      "loss": 1.2019,
+      "step": 164100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1084971914806784e-06,
+      "loss": 1.4083,
+      "step": 164105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1080330746190247e-06,
+      "loss": 1.2885,
+      "step": 164110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1075690028247466e-06,
+      "loss": 1.2774,
+      "step": 164115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.107104976100498e-06,
+      "loss": 1.3316,
+      "step": 164120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.106640994448924e-06,
+      "loss": 1.4008,
+      "step": 164125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.106177057872674e-06,
+      "loss": 1.497,
+      "step": 164130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1057131663744014e-06,
+      "loss": 1.1263,
+      "step": 164135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.105249319956749e-06,
+      "loss": 1.2248,
+      "step": 164140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1047855186223697e-06,
+      "loss": 1.2695,
+      "step": 164145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.104321762373914e-06,
+      "loss": 1.1178,
+      "step": 164150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1038580512140273e-06,
+      "loss": 1.1174,
+      "step": 164155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1033943851453563e-06,
+      "loss": 1.1048,
+      "step": 164160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1029307641705497e-06,
+      "loss": 0.9398,
+      "step": 164165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.10246718829226e-06,
+      "loss": 0.9583,
+      "step": 164170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1020036575131266e-06,
+      "loss": 1.1316,
+      "step": 164175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1015401718358032e-06,
+      "loss": 1.061,
+      "step": 164180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.101076731262931e-06,
+      "loss": 1.2621,
+      "step": 164185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.100613335797158e-06,
+      "loss": 1.6549,
+      "step": 164190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1001499854411356e-06,
+      "loss": 1.2222,
+      "step": 164195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0996866801975056e-06,
+      "loss": 1.1876,
+      "step": 164200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.099223420068911e-06,
+      "loss": 1.156,
+      "step": 164205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0987602050580002e-06,
+      "loss": 1.5521,
+      "step": 164210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0982970351674205e-06,
+      "loss": 1.4642,
+      "step": 164215
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.097833910399815e-06,
+      "loss": 1.1458,
+      "step": 164220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0973708307578254e-06,
+      "loss": 1.3963,
+      "step": 164225
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0969077962440985e-06,
+      "loss": 1.2844,
+      "step": 164230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.096444806861282e-06,
+      "loss": 1.108,
+      "step": 164235
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0959818626120143e-06,
+      "loss": 1.2923,
+      "step": 164240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0955189634989426e-06,
+      "loss": 0.9723,
+      "step": 164245
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0950561095247102e-06,
+      "loss": 1.215,
+      "step": 164250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.094593300691956e-06,
+      "loss": 1.4839,
+      "step": 164255
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.094130537003326e-06,
+      "loss": 1.28,
+      "step": 164260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0936678184614656e-06,
+      "loss": 1.3072,
+      "step": 164265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.093205145069015e-06,
+      "loss": 1.2651,
+      "step": 164270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.092742516828612e-06,
+      "loss": 1.2168,
+      "step": 164275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0922799337429023e-06,
+      "loss": 1.1539,
+      "step": 164280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0918173958145303e-06,
+      "loss": 1.0804,
+      "step": 164285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0913549030461324e-06,
+      "loss": 0.9029,
+      "step": 164290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0908924554403542e-06,
+      "loss": 1.1451,
+      "step": 164295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.09043005299983e-06,
+      "loss": 1.1677,
+      "step": 164300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0899676957272054e-06,
+      "loss": 1.0903,
+      "step": 164305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0895053836251224e-06,
+      "loss": 1.3278,
+      "step": 164310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.089043116696218e-06,
+      "loss": 1.1853,
+      "step": 164315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.08858089494313e-06,
+      "loss": 1.0218,
+      "step": 164320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0881187183685003e-06,
+      "loss": 1.3679,
+      "step": 164325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.08765658697497e-06,
+      "loss": 1.145,
+      "step": 164330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.087194500765175e-06,
+      "loss": 1.2922,
+      "step": 164335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.086732459741757e-06,
+      "loss": 1.0599,
+      "step": 164340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0862704639073526e-06,
+      "loss": 1.3829,
+      "step": 164345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0858085132645977e-06,
+      "loss": 0.8875,
+      "step": 164350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.085346607816133e-06,
+      "loss": 1.2517,
+      "step": 164355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0848847475645995e-06,
+      "loss": 1.0438,
+      "step": 164360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0844229325126308e-06,
+      "loss": 1.4477,
+      "step": 164365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.083961162662862e-06,
+      "loss": 1.0789,
+      "step": 164370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.083499438017932e-06,
+      "loss": 1.269,
+      "step": 164375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0830377585804807e-06,
+      "loss": 1.3029,
+      "step": 164380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0825761243531407e-06,
+      "loss": 1.113,
+      "step": 164385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.082114535338551e-06,
+      "loss": 1.2055,
+      "step": 164390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0816529915393437e-06,
+      "loss": 1.2364,
+      "step": 164395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0811914929581558e-06,
+      "loss": 1.3641,
+      "step": 164400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.080730039597627e-06,
+      "loss": 1.1699,
+      "step": 164405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.080268631460389e-06,
+      "loss": 1.1925,
+      "step": 164410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0798072685490744e-06,
+      "loss": 1.3083,
+      "step": 164415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0793459508663192e-06,
+      "loss": 1.1666,
+      "step": 164420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.078884678414762e-06,
+      "loss": 1.1711,
+      "step": 164425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0784234511970346e-06,
+      "loss": 1.159,
+      "step": 164430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0779622692157665e-06,
+      "loss": 1.6466,
+      "step": 164435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.077501132473597e-06,
+      "loss": 1.3208,
+      "step": 164440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.077040040973155e-06,
+      "loss": 1.222,
+      "step": 164445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0765789947170756e-06,
+      "loss": 1.4542,
+      "step": 164450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0761179937079945e-06,
+      "loss": 1.0872,
+      "step": 164455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.075657037948542e-06,
+      "loss": 1.7347,
+      "step": 164460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0751961274413458e-06,
+      "loss": 1.3885,
+      "step": 164465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.074735262189044e-06,
+      "loss": 1.2205,
+      "step": 164470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0742744421942675e-06,
+      "loss": 1.1342,
+      "step": 164475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0738136674596455e-06,
+      "loss": 1.0156,
+      "step": 164480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0733529379878126e-06,
+      "loss": 1.3324,
+      "step": 164485
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0728922537813945e-06,
+      "loss": 1.1259,
+      "step": 164490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.072431614843029e-06,
+      "loss": 1.2142,
+      "step": 164495
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0719710211753397e-06,
+      "loss": 1.2343,
+      "step": 164500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.071510472780962e-06,
+      "loss": 1.013,
+      "step": 164505
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.071049969662522e-06,
+      "loss": 1.2179,
+      "step": 164510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0705895118226514e-06,
+      "loss": 1.2497,
+      "step": 164515
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0701290992639813e-06,
+      "loss": 1.4949,
+      "step": 164520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0696687319891397e-06,
+      "loss": 0.9505,
+      "step": 164525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0692084100007524e-06,
+      "loss": 1.4903,
+      "step": 164530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0687481333014516e-06,
+      "loss": 1.4826,
+      "step": 164535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.068287901893866e-06,
+      "loss": 1.2029,
+      "step": 164540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0678277157806202e-06,
+      "loss": 1.2141,
+      "step": 164545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0673675749643472e-06,
+      "loss": 1.4483,
+      "step": 164550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.066907479447672e-06,
+      "loss": 1.2401,
+      "step": 164555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0664474292332193e-06,
+      "loss": 0.9974,
+      "step": 164560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.065987424323619e-06,
+      "loss": 1.1233,
+      "step": 164565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0655274647215006e-06,
+      "loss": 1.1542,
+      "step": 164570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0650675504294882e-06,
+      "loss": 1.1181,
+      "step": 164575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0646076814502047e-06,
+      "loss": 1.359,
+      "step": 164580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0641478577862796e-06,
+      "loss": 1.1208,
+      "step": 164585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.06368807944034e-06,
+      "loss": 1.3199,
+      "step": 164590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0632283464150094e-06,
+      "loss": 1.2434,
+      "step": 164595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0627686587129146e-06,
+      "loss": 1.3481,
+      "step": 164600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.062309016336679e-06,
+      "loss": 1.0054,
+      "step": 164605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0618494192889273e-06,
+      "loss": 0.9756,
+      "step": 164610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0613898675722877e-06,
+      "loss": 1.3415,
+      "step": 164615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0609303611893817e-06,
+      "loss": 1.3163,
+      "step": 164620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0604709001428304e-06,
+      "loss": 1.2309,
+      "step": 164625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0600114844352605e-06,
+      "loss": 1.3827,
+      "step": 164630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0595521140692986e-06,
+      "loss": 1.3091,
+      "step": 164635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0590927890475655e-06,
+      "loss": 1.1962,
+      "step": 164640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0586335093726796e-06,
+      "loss": 1.4029,
+      "step": 164645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0581742750472712e-06,
+      "loss": 1.6602,
+      "step": 164650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.057715086073956e-06,
+      "loss": 1.3939,
+      "step": 164655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0572559424553597e-06,
+      "loss": 1.1088,
+      "step": 164660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.056796844194107e-06,
+      "loss": 1.661,
+      "step": 164665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.056337791292816e-06,
+      "loss": 1.2077,
+      "step": 164670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0558787837541074e-06,
+      "loss": 1.4589,
+      "step": 164675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.055419821580602e-06,
+      "loss": 1.205,
+      "step": 164680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0549609047749266e-06,
+      "loss": 1.465,
+      "step": 164685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0545020333396937e-06,
+      "loss": 1.0613,
+      "step": 164690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0540432072775306e-06,
+      "loss": 1.0649,
+      "step": 164695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.053584426591052e-06,
+      "loss": 1.6946,
+      "step": 164700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0531256912828836e-06,
+      "loss": 1.4464,
+      "step": 164705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0526670013556394e-06,
+      "loss": 1.0198,
+      "step": 164710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0522083568119423e-06,
+      "loss": 1.1883,
+      "step": 164715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.051749757654411e-06,
+      "loss": 1.2584,
+      "step": 164720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.051291203885658e-06,
+      "loss": 1.0692,
+      "step": 164725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0508326955083134e-06,
+      "loss": 1.3262,
+      "step": 164730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0503742325249885e-06,
+      "loss": 0.9086,
+      "step": 164735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.049915814938301e-06,
+      "loss": 1.339,
+      "step": 164740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.049457442750872e-06,
+      "loss": 1.33,
+      "step": 164745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0489991159653157e-06,
+      "loss": 1.1042,
+      "step": 164750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0485408345842495e-06,
+      "loss": 0.8928,
+      "step": 164755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.048082598610295e-06,
+      "loss": 1.4669,
+      "step": 164760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.047624408046066e-06,
+      "loss": 0.9493,
+      "step": 164765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0471662628941757e-06,
+      "loss": 0.9725,
+      "step": 164770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.046708163157244e-06,
+      "loss": 1.4604,
+      "step": 164775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0462501088378893e-06,
+      "loss": 1.087,
+      "step": 164780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0457920999387236e-06,
+      "loss": 1.022,
+      "step": 164785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0453341364623613e-06,
+      "loss": 1.0898,
+      "step": 164790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.044876218411419e-06,
+      "loss": 0.9853,
+      "step": 164795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0444183457885146e-06,
+      "loss": 1.4456,
+      "step": 164800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0439605185962586e-06,
+      "loss": 1.125,
+      "step": 164805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0435027368372694e-06,
+      "loss": 1.0916,
+      "step": 164810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0430450005141566e-06,
+      "loss": 1.4709,
+      "step": 164815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.042587309629536e-06,
+      "loss": 1.0043,
+      "step": 164820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0421296641860245e-06,
+      "loss": 0.998,
+      "step": 164825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.041672064186233e-06,
+      "loss": 1.1959,
+      "step": 164830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0412145096327718e-06,
+      "loss": 1.362,
+      "step": 164835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0407570005282583e-06,
+      "loss": 1.1674,
+      "step": 164840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.040299536875302e-06,
+      "loss": 1.1245,
+      "step": 164845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0398421186765185e-06,
+      "loss": 1.4389,
+      "step": 164850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.039384745934515e-06,
+      "loss": 1.4417,
+      "step": 164855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.038927418651908e-06,
+      "loss": 1.4526,
+      "step": 164860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0384701368313066e-06,
+      "loss": 1.2615,
+      "step": 164865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0380129004753214e-06,
+      "loss": 1.5541,
+      "step": 164870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.037555709586568e-06,
+      "loss": 1.1605,
+      "step": 164875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0370985641676543e-06,
+      "loss": 0.9776,
+      "step": 164880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0366414642211884e-06,
+      "loss": 1.2246,
+      "step": 164885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.036184409749782e-06,
+      "loss": 1.1676,
+      "step": 164890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.035727400756049e-06,
+      "loss": 1.2953,
+      "step": 164895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0352704372425937e-06,
+      "loss": 1.1591,
+      "step": 164900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.03481351921203e-06,
+      "loss": 1.0324,
+      "step": 164905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0343566466669627e-06,
+      "loss": 1.1397,
+      "step": 164910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0338998196100048e-06,
+      "loss": 1.1944,
+      "step": 164915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.033443038043764e-06,
+      "loss": 1.113,
+      "step": 164920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0329863019708497e-06,
+      "loss": 1.1454,
+      "step": 164925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0325296113938686e-06,
+      "loss": 1.1784,
+      "step": 164930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0320729663154228e-06,
+      "loss": 1.2176,
+      "step": 164935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0316163667381316e-06,
+      "loss": 1.1662,
+      "step": 164940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.031159812664596e-06,
+      "loss": 1.4004,
+      "step": 164945
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0307033040974224e-06,
+      "loss": 1.1351,
+      "step": 164950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.030246841039222e-06,
+      "loss": 1.2804,
+      "step": 164955
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0297904234925947e-06,
+      "loss": 1.3081,
+      "step": 164960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.029334051460152e-06,
+      "loss": 1.1997,
+      "step": 164965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0288777249445013e-06,
+      "loss": 1.2865,
+      "step": 164970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0284214439482465e-06,
+      "loss": 0.8936,
+      "step": 164975
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0279652084739885e-06,
+      "loss": 1.186,
+      "step": 164980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0275090185243383e-06,
+      "loss": 1.1586,
+      "step": 164985
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0270528741019023e-06,
+      "loss": 1.181,
+      "step": 164990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.026596775209282e-06,
+      "loss": 1.0711,
+      "step": 164995
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0261407218490804e-06,
+      "loss": 1.1322,
+      "step": 165000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0256847140239043e-06,
+      "loss": 1.0542,
+      "step": 165005
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.02522875173636e-06,
+      "loss": 1.1613,
+      "step": 165010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.024772834989046e-06,
+      "loss": 1.1725,
+      "step": 165015
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0243169637845715e-06,
+      "loss": 0.9398,
+      "step": 165020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.023861138125536e-06,
+      "loss": 1.0488,
+      "step": 165025
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0234053580145386e-06,
+      "loss": 1.1363,
+      "step": 165030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0229496234541925e-06,
+      "loss": 1.0947,
+      "step": 165035
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.022493934447094e-06,
+      "loss": 1.0918,
+      "step": 165040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0220382909958436e-06,
+      "loss": 1.1176,
+      "step": 165045
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.021582693103048e-06,
+      "loss": 1.133,
+      "step": 165050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.021127140771304e-06,
+      "loss": 1.3217,
+      "step": 165055
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.020671634003215e-06,
+      "loss": 1.4067,
+      "step": 165060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.020216172801386e-06,
+      "loss": 1.2846,
+      "step": 165065
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0197607571684143e-06,
+      "loss": 1.3475,
+      "step": 165070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.019305387106898e-06,
+      "loss": 1.3074,
+      "step": 165075
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.018850062619441e-06,
+      "loss": 1.2031,
+      "step": 165080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.018394783708645e-06,
+      "loss": 1.3408,
+      "step": 165085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.017939550377107e-06,
+      "loss": 1.2905,
+      "step": 165090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0174843626274252e-06,
+      "loss": 1.1975,
+      "step": 165095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.017029220462202e-06,
+      "loss": 1.3016,
+      "step": 165100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0165741238840374e-06,
+      "loss": 1.6264,
+      "step": 165105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.016119072895526e-06,
+      "loss": 1.3871,
+      "step": 165110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.015664067499271e-06,
+      "loss": 1.1633,
+      "step": 165115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0152091076978665e-06,
+      "loss": 1.083,
+      "step": 165120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.014754193493913e-06,
+      "loss": 1.0332,
+      "step": 165125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.01429932489001e-06,
+      "loss": 1.0818,
+      "step": 165130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.013844501888753e-06,
+      "loss": 1.0118,
+      "step": 165135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.013389724492739e-06,
+      "loss": 1.1582,
+      "step": 165140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0129349927045626e-06,
+      "loss": 1.2699,
+      "step": 165145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0124803065268237e-06,
+      "loss": 1.5655,
+      "step": 165150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.012025665962121e-06,
+      "loss": 1.1463,
+      "step": 165155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0115710710130453e-06,
+      "loss": 1.1865,
+      "step": 165160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.011116521682198e-06,
+      "loss": 1.1722,
+      "step": 165165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0106620179721705e-06,
+      "loss": 1.037,
+      "step": 165170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.010207559885559e-06,
+      "loss": 1.155,
+      "step": 165175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.009753147424962e-06,
+      "loss": 0.9828,
+      "step": 165180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0092987805929732e-06,
+      "loss": 1.0845,
+      "step": 165185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.008844459392182e-06,
+      "loss": 1.0205,
+      "step": 165190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0083901838251886e-06,
+      "loss": 1.0003,
+      "step": 165195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0079359538945864e-06,
+      "loss": 0.85,
+      "step": 165200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.007481769602967e-06,
+      "loss": 1.2834,
+      "step": 165205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.007027630952928e-06,
+      "loss": 1.0399,
+      "step": 165210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0065735379470587e-06,
+      "loss": 1.3814,
+      "step": 165215
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0061194905879543e-06,
+      "loss": 1.1248,
+      "step": 165220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0056654888782057e-06,
+      "loss": 0.8697,
+      "step": 165225
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0052115328204092e-06,
+      "loss": 1.0024,
+      "step": 165230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.004757622417155e-06,
+      "loss": 0.9339,
+      "step": 165235
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.00430375767103e-06,
+      "loss": 1.0106,
+      "step": 165240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0038499385846364e-06,
+      "loss": 1.1271,
+      "step": 165245
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.0033961651605603e-06,
+      "loss": 1.4966,
+      "step": 165250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0029424374013907e-06,
+      "loss": 1.4683,
+      "step": 165255
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.002488755309723e-06,
+      "loss": 1.2949,
+      "step": 165260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.002035118888144e-06,
+      "loss": 1.0835,
+      "step": 165265
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.001581528139245e-06,
+      "loss": 1.0405,
+      "step": 165270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0011279830656207e-06,
+      "loss": 1.3558,
+      "step": 165275
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0006744836698587e-06,
+      "loss": 1.2478,
+      "step": 165280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.000221029954544e-06,
+      "loss": 1.1209,
+      "step": 165285
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9997676219222707e-06,
+      "loss": 0.9621,
+      "step": 165290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.99931425957563e-06,
+      "loss": 1.2245,
+      "step": 165295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9988609429172067e-06,
+      "loss": 1.1736,
+      "step": 165300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9984076719495893e-06,
+      "loss": 1.2173,
+      "step": 165305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9979544466753676e-06,
+      "loss": 1.456,
+      "step": 165310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9975012670971326e-06,
+      "loss": 1.3131,
+      "step": 165315
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9970481332174674e-06,
+      "loss": 1.154,
+      "step": 165320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9965950450389637e-06,
+      "loss": 1.2562,
+      "step": 165325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9961420025642077e-06,
+      "loss": 1.2503,
+      "step": 165330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.995689005795779e-06,
+      "loss": 1.0163,
+      "step": 165335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.995236054736278e-06,
+      "loss": 1.0088,
+      "step": 165340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9947831493882843e-06,
+      "loss": 1.4474,
+      "step": 165345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9943302897543813e-06,
+      "loss": 1.3113,
+      "step": 165350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9938774758371606e-06,
+      "loss": 1.2696,
+      "step": 165355
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9934247076392035e-06,
+      "loss": 1.0886,
+      "step": 165360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9929719851631e-06,
+      "loss": 1.1743,
+      "step": 165365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9925193084114303e-06,
+      "loss": 1.4365,
+      "step": 165370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.992066677386786e-06,
+      "loss": 1.2406,
+      "step": 165375
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.991614092091744e-06,
+      "loss": 1.3698,
+      "step": 165380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.991161552528893e-06,
+      "loss": 1.2952,
+      "step": 165385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9907090587008205e-06,
+      "loss": 1.2917,
+      "step": 165390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9902566106101074e-06,
+      "loss": 1.2164,
+      "step": 165395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.989804208259334e-06,
+      "loss": 0.8856,
+      "step": 165400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9893518516510857e-06,
+      "loss": 1.1846,
+      "step": 165405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.988899540787951e-06,
+      "loss": 1.277,
+      "step": 165410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9884472756725047e-06,
+      "loss": 1.1994,
+      "step": 165415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.987995056307338e-06,
+      "loss": 1.3159,
+      "step": 165420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9875428826950273e-06,
+      "loss": 0.9988,
+      "step": 165425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9870907548381545e-06,
+      "loss": 1.4684,
+      "step": 165430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.986638672739303e-06,
+      "loss": 1.0984,
+      "step": 165435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9861866364010573e-06,
+      "loss": 1.2414,
+      "step": 165440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9857346458259953e-06,
+      "loss": 1.1077,
+      "step": 165445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9852827010166977e-06,
+      "loss": 1.2742,
+      "step": 165450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.984830801975746e-06,
+      "loss": 1.2412,
+      "step": 165455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.984378948705724e-06,
+      "loss": 1.0314,
+      "step": 165460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9839271412092064e-06,
+      "loss": 1.1662,
+      "step": 165465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9834753794887794e-06,
+      "loss": 1.2664,
+      "step": 165470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9830236635470167e-06,
+      "loss": 1.1522,
+      "step": 165475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.982571993386502e-06,
+      "loss": 1.4537,
+      "step": 165480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9821203690098147e-06,
+      "loss": 1.3453,
+      "step": 165485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.981668790419533e-06,
+      "loss": 1.2559,
+      "step": 165490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.981217257618232e-06,
+      "loss": 1.4845,
+      "step": 165495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9807657706084948e-06,
+      "loss": 1.2246,
+      "step": 165500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9803143293928993e-06,
+      "loss": 0.9066,
+      "step": 165505
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.979862933974024e-06,
+      "loss": 1.1505,
+      "step": 165510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9794115843544402e-06,
+      "loss": 1.3092,
+      "step": 165515
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9789602805367313e-06,
+      "loss": 1.6324,
+      "step": 165520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978509022523477e-06,
+      "loss": 1.2875,
+      "step": 165525
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9780578103172467e-06,
+      "loss": 0.9738,
+      "step": 165530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.977606643920623e-06,
+      "loss": 0.9983,
+      "step": 165535
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9771555233361815e-06,
+      "loss": 1.3008,
+      "step": 165540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9767044485664945e-06,
+      "loss": 1.1127,
+      "step": 165545
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9762534196141393e-06,
+      "loss": 1.4905,
+      "step": 165550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9758024364816964e-06,
+      "loss": 1.5657,
+      "step": 165555
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.975351499171734e-06,
+      "loss": 0.9801,
+      "step": 165560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9749006076868337e-06,
+      "loss": 1.1047,
+      "step": 165565
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9744497620295644e-06,
+      "loss": 1.163,
+      "step": 165570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9739989622025056e-06,
+      "loss": 1.2519,
+      "step": 165575
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973548208208228e-06,
+      "loss": 1.6124,
+      "step": 165580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9730975000493092e-06,
+      "loss": 0.9903,
+      "step": 165585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.972646837728319e-06,
+      "loss": 1.2694,
+      "step": 165590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9721962212478318e-06,
+      "loss": 1.4611,
+      "step": 165595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9717456506104248e-06,
+      "loss": 1.4624,
+      "step": 165600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9712951258186684e-06,
+      "loss": 1.3562,
+      "step": 165605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.970844646875133e-06,
+      "loss": 1.3555,
+      "step": 165610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9703942137823928e-06,
+      "loss": 1.2544,
+      "step": 165615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9699438265430226e-06,
+      "loss": 1.1202,
+      "step": 165620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9694934851595902e-06,
+      "loss": 1.2975,
+      "step": 165625
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9690431896346716e-06,
+      "loss": 1.3319,
+      "step": 165630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9685929399708357e-06,
+      "loss": 1.1468,
+      "step": 165635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9681427361706486e-06,
+      "loss": 1.4944,
+      "step": 165640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9676925782366917e-06,
+      "loss": 0.7529,
+      "step": 165645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9672424661715306e-06,
+      "loss": 1.3837,
+      "step": 165650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.966792399977735e-06,
+      "loss": 1.2728,
+      "step": 165655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.966342379657873e-06,
+      "loss": 1.3414,
+      "step": 165660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9658924052145166e-06,
+      "loss": 1.0594,
+      "step": 165665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.965442476650239e-06,
+      "loss": 1.2439,
+      "step": 165670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.964992593967603e-06,
+      "loss": 1.3373,
+      "step": 165675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9645427571691842e-06,
+      "loss": 1.1436,
+      "step": 165680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.964092966257546e-06,
+      "loss": 1.3682,
+      "step": 165685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9636432212352576e-06,
+      "loss": 1.2734,
+      "step": 165690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9631935221048924e-06,
+      "loss": 1.1228,
+      "step": 165695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962743868869015e-06,
+      "loss": 1.2581,
+      "step": 165700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962294261530189e-06,
+      "loss": 1.4436,
+      "step": 165705
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961844700090987e-06,
+      "loss": 1.2858,
+      "step": 165710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961395184553978e-06,
+      "loss": 1.3748,
+      "step": 165715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.960945714921725e-06,
+      "loss": 1.5046,
+      "step": 165720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.960496291196794e-06,
+      "loss": 0.9921,
+      "step": 165725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9600469133817557e-06,
+      "loss": 1.1187,
+      "step": 165730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.959597581479171e-06,
+      "loss": 1.2429,
+      "step": 165735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9591482954916086e-06,
+      "loss": 1.6871,
+      "step": 165740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9586990554216365e-06,
+      "loss": 1.1021,
+      "step": 165745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9582498612718193e-06,
+      "loss": 1.2117,
+      "step": 165750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.957800713044716e-06,
+      "loss": 1.4414,
+      "step": 165755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.957351610742898e-06,
+      "loss": 1.3562,
+      "step": 165760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9569025543689303e-06,
+      "loss": 1.4725,
+      "step": 165765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9564535439253716e-06,
+      "loss": 1.2325,
+      "step": 165770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.956004579414792e-06,
+      "loss": 1.1941,
+      "step": 165775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9555556608397506e-06,
+      "loss": 0.9659,
+      "step": 165780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.955106788202813e-06,
+      "loss": 1.2018,
+      "step": 165785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.954657961506545e-06,
+      "loss": 1.0803,
+      "step": 165790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.954209180753508e-06,
+      "loss": 1.3487,
+      "step": 165795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.953760445946261e-06,
+      "loss": 1.564,
+      "step": 165800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9533117570873695e-06,
+      "loss": 1.7204,
+      "step": 165805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9528631141793984e-06,
+      "loss": 1.3289,
+      "step": 165810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9524145172249065e-06,
+      "loss": 1.2111,
+      "step": 165815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9519659662264544e-06,
+      "loss": 1.1707,
+      "step": 165820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951517461186605e-06,
+      "loss": 1.1778,
+      "step": 165825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9510690021079224e-06,
+      "loss": 0.9788,
+      "step": 165830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.950620588992962e-06,
+      "loss": 1.0662,
+      "step": 165835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.950172221844291e-06,
+      "loss": 0.9885,
+      "step": 165840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.949723900664465e-06,
+      "loss": 1.1939,
+      "step": 165845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.949275625456043e-06,
+      "loss": 1.0131,
+      "step": 165850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9488273962215874e-06,
+      "loss": 1.2471,
+      "step": 165855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9483792129636616e-06,
+      "loss": 1.5018,
+      "step": 165860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.947931075684819e-06,
+      "loss": 1.1053,
+      "step": 165865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9474829843876197e-06,
+      "loss": 1.4479,
+      "step": 165870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9470349390746236e-06,
+      "loss": 1.5503,
+      "step": 165875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9465869397483903e-06,
+      "loss": 1.1229,
+      "step": 165880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9461389864114764e-06,
+      "loss": 1.2019,
+      "step": 165885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.945691079066442e-06,
+      "loss": 1.4213,
+      "step": 165890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.945243217715842e-06,
+      "loss": 1.2743,
+      "step": 165895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.944795402362234e-06,
+      "loss": 1.2354,
+      "step": 165900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.944347633008181e-06,
+      "loss": 1.2384,
+      "step": 165905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.943899909656235e-06,
+      "loss": 1.1792,
+      "step": 165910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.943452232308951e-06,
+      "loss": 1.2815,
+      "step": 165915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9430046009688876e-06,
+      "loss": 1.0601,
+      "step": 165920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9425570156386043e-06,
+      "loss": 1.0809,
+      "step": 165925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9421094763206516e-06,
+      "loss": 1.2127,
+      "step": 165930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.941661983017591e-06,
+      "loss": 1.1146,
+      "step": 165935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9412145357319733e-06,
+      "loss": 1.6463,
+      "step": 165940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9407671344663536e-06,
+      "loss": 1.3961,
+      "step": 165945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9403197792232874e-06,
+      "loss": 0.9557,
+      "step": 165950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.939872470005334e-06,
+      "loss": 1.509,
+      "step": 165955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9394252068150432e-06,
+      "loss": 1.2633,
+      "step": 165960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9389779896549666e-06,
+      "loss": 1.1345,
+      "step": 165965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.938530818527662e-06,
+      "loss": 1.1692,
+      "step": 165970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.938083693435685e-06,
+      "loss": 1.402,
+      "step": 165975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.937636614381584e-06,
+      "loss": 1.5772,
+      "step": 165980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.937189581367915e-06,
+      "loss": 1.3666,
+      "step": 165985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9367425943972294e-06,
+      "loss": 1.306,
+      "step": 165990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.93629565347208e-06,
+      "loss": 1.1253,
+      "step": 165995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9358487585950227e-06,
+      "loss": 1.3646,
+      "step": 166000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.935401909768605e-06,
+      "loss": 1.3517,
+      "step": 166005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.934955106995381e-06,
+      "loss": 1.1814,
+      "step": 166010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.934508350277895e-06,
+      "loss": 1.3144,
+      "step": 166015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9340616396187107e-06,
+      "loss": 1.3351,
+      "step": 166020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9336149750203714e-06,
+      "loss": 1.33,
+      "step": 166025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9331683564854278e-06,
+      "loss": 0.9169,
+      "step": 166030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9327217840164337e-06,
+      "loss": 0.9987,
+      "step": 166035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9322752576159344e-06,
+      "loss": 1.3568,
+      "step": 166040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9318287772864818e-06,
+      "loss": 1.3365,
+      "step": 166045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.93138234303063e-06,
+      "loss": 1.1036,
+      "step": 166050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9309359548509253e-06,
+      "loss": 1.2535,
+      "step": 166055
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.930489612749913e-06,
+      "loss": 1.265,
+      "step": 166060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9300433167301447e-06,
+      "loss": 1.1736,
+      "step": 166065
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9295970667941733e-06,
+      "loss": 1.3094,
+      "step": 166070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.92915086294454e-06,
+      "loss": 1.1523,
+      "step": 166075
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9287047051837992e-06,
+      "loss": 1.2484,
+      "step": 166080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9282585935144927e-06,
+      "loss": 1.5601,
+      "step": 166085
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9278125279391734e-06,
+      "loss": 1.0815,
+      "step": 166090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9273665084603844e-06,
+      "loss": 1.4724,
+      "step": 166095
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9269205350806763e-06,
+      "loss": 0.9179,
+      "step": 166100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.926474607802592e-06,
+      "loss": 1.2596,
+      "step": 166105
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9260287266286794e-06,
+      "loss": 1.6302,
+      "step": 166110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9255828915614894e-06,
+      "loss": 1.1105,
+      "step": 166115
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9251371026035628e-06,
+      "loss": 1.0742,
+      "step": 166120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9246913597574436e-06,
+      "loss": 1.1428,
+      "step": 166125
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.924245663025683e-06,
+      "loss": 1.4617,
+      "step": 166130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9238000124108213e-06,
+      "loss": 1.0354,
+      "step": 166135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.923354407915404e-06,
+      "loss": 1.2089,
+      "step": 166140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9229088495419813e-06,
+      "loss": 1.6004,
+      "step": 166145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.922463337293092e-06,
+      "loss": 1.1955,
+      "step": 166150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.922017871171279e-06,
+      "loss": 1.2712,
+      "step": 166155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.921572451179089e-06,
+      "loss": 1.172,
+      "step": 166160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9211270773190684e-06,
+      "loss": 1.3964,
+      "step": 166165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.920681749593757e-06,
+      "loss": 1.1914,
+      "step": 166170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9202364680056963e-06,
+      "loss": 1.1284,
+      "step": 166175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9197912325574296e-06,
+      "loss": 1.0946,
+      "step": 166180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9193460432515055e-06,
+      "loss": 1.3646,
+      "step": 166185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.918900900090458e-06,
+      "loss": 1.2299,
+      "step": 166190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.918455803076835e-06,
+      "loss": 1.7115,
+      "step": 166195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.918010752213174e-06,
+      "loss": 1.3331,
+      "step": 166200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9175657475020182e-06,
+      "loss": 1.2194,
+      "step": 166205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.917120788945912e-06,
+      "loss": 1.2063,
+      "step": 166210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9166758765473937e-06,
+      "loss": 0.9974,
+      "step": 166215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.916231010309001e-06,
+      "loss": 1.4059,
+      "step": 166220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9157861902332776e-06,
+      "loss": 0.9543,
+      "step": 166225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9153414163227644e-06,
+      "loss": 1.6557,
+      "step": 166230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9148966885800015e-06,
+      "loss": 1.4432,
+      "step": 166235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9144520070075235e-06,
+      "loss": 1.1794,
+      "step": 166240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9140073716078765e-06,
+      "loss": 1.2587,
+      "step": 166245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9135627823835933e-06,
+      "loss": 1.149,
+      "step": 166250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.913118239337217e-06,
+      "loss": 1.2111,
+      "step": 166255
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912673742471287e-06,
+      "loss": 1.159,
+      "step": 166260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912229291788339e-06,
+      "loss": 1.2847,
+      "step": 166265
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9117848872909105e-06,
+      "loss": 1.4703,
+      "step": 166270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.91134052898154e-06,
+      "loss": 1.2364,
+      "step": 166275
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9108962168627675e-06,
+      "loss": 1.2212,
+      "step": 166280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9104519509371265e-06,
+      "loss": 1.2343,
+      "step": 166285
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9100077312071585e-06,
+      "loss": 1.2114,
+      "step": 166290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.909563557675396e-06,
+      "loss": 1.2266,
+      "step": 166295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9091194303443774e-06,
+      "loss": 1.2425,
+      "step": 166300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.908675349216638e-06,
+      "loss": 1.1106,
+      "step": 166305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.908231314294715e-06,
+      "loss": 1.0897,
+      "step": 166310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.907787325581144e-06,
+      "loss": 1.2058,
+      "step": 166315
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9073433830784548e-06,
+      "loss": 0.9567,
+      "step": 166320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9068994867891932e-06,
+      "loss": 1.0268,
+      "step": 166325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9064556367158882e-06,
+      "loss": 1.0561,
+      "step": 166330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.906011832861072e-06,
+      "loss": 1.3223,
+      "step": 166335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9055680752272843e-06,
+      "loss": 1.1191,
+      "step": 166340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9051243638170536e-06,
+      "loss": 1.1237,
+      "step": 166345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.904680698632917e-06,
+      "loss": 1.2021,
+      "step": 166350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9042370796774102e-06,
+      "loss": 1.2005,
+      "step": 166355
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9037935069530633e-06,
+      "loss": 1.3625,
+      "step": 166360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9033499804624079e-06,
+      "loss": 1.4961,
+      "step": 166365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9029065002079793e-06,
+      "loss": 1.5533,
+      "step": 166370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9024630661923115e-06,
+      "loss": 1.4382,
+      "step": 166375
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9020196784179356e-06,
+      "loss": 1.0741,
+      "step": 166380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9015763368873797e-06,
+      "loss": 1.0499,
+      "step": 166385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9011330416031791e-06,
+      "loss": 1.5575,
+      "step": 166390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9006897925678669e-06,
+      "loss": 1.338,
+      "step": 166395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9002465897839694e-06,
+      "loss": 0.9628,
+      "step": 166400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8998034332540238e-06,
+      "loss": 1.1191,
+      "step": 166405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.899360322980557e-06,
+      "loss": 1.3791,
+      "step": 166410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8989172589660943e-06,
+      "loss": 0.9666,
+      "step": 166415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8984742412131774e-06,
+      "loss": 1.3492,
+      "step": 166420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8980312697243287e-06,
+      "loss": 1.2024,
+      "step": 166425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.897588344502077e-06,
+      "loss": 1.288,
+      "step": 166430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8971454655489562e-06,
+      "loss": 1.1016,
+      "step": 166435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.896702632867491e-06,
+      "loss": 1.0258,
+      "step": 166440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8962598464602144e-06,
+      "loss": 1.0295,
+      "step": 166445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8958171063296493e-06,
+      "loss": 0.9907,
+      "step": 166450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8953744124783314e-06,
+      "loss": 1.2668,
+      "step": 166455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8949317649087807e-06,
+      "loss": 1.2186,
+      "step": 166460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8944891636235295e-06,
+      "loss": 1.1483,
+      "step": 166465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8940466086251074e-06,
+      "loss": 1.1399,
+      "step": 166470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.89360409991604e-06,
+      "loss": 1.3233,
+      "step": 166475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8931616374988493e-06,
+      "loss": 1.5779,
+      "step": 166480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.892719221376066e-06,
+      "loss": 1.8135,
+      "step": 166485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.89227685155022e-06,
+      "loss": 1.3514,
+      "step": 166490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8918345280238304e-06,
+      "loss": 1.2612,
+      "step": 166495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8913922507994287e-06,
+      "loss": 1.1753,
+      "step": 166500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8909500198795372e-06,
+      "loss": 1.1013,
+      "step": 166505
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8905078352666816e-06,
+      "loss": 1.4119,
+      "step": 166510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8900656969633913e-06,
+      "loss": 1.0625,
+      "step": 166515
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8896236049721873e-06,
+      "loss": 1.4235,
+      "step": 166520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8891815592955954e-06,
+      "loss": 1.1231,
+      "step": 166525
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8887395599361358e-06,
+      "loss": 1.4895,
+      "step": 166530
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8882976068963366e-06,
+      "loss": 1.4257,
+      "step": 166535
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8878557001787235e-06,
+      "loss": 1.068,
+      "step": 166540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8874138397858143e-06,
+      "loss": 1.6871,
+      "step": 166545
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8869720257201385e-06,
+      "loss": 1.0909,
+      "step": 166550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8865302579842138e-06,
+      "loss": 1.2483,
+      "step": 166555
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8860885365805648e-06,
+      "loss": 1.1687,
+      "step": 166560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8856468615117164e-06,
+      "loss": 1.1917,
+      "step": 166565
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8852052327801894e-06,
+      "loss": 1.4341,
+      "step": 166570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8847636503885025e-06,
+      "loss": 1.2522,
+      "step": 166575
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8843221143391798e-06,
+      "loss": 1.1646,
+      "step": 166580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8838806246347452e-06,
+      "loss": 0.9497,
+      "step": 166585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8834391812777154e-06,
+      "loss": 1.3126,
+      "step": 166590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8829977842706148e-06,
+      "loss": 1.0635,
+      "step": 166595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.882556433615962e-06,
+      "loss": 1.0541,
+      "step": 166600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8821151293162788e-06,
+      "loss": 1.1423,
+      "step": 166605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8816738713740823e-06,
+      "loss": 1.294,
+      "step": 166610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8812326597918973e-06,
+      "loss": 1.166,
+      "step": 166615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8807914945722405e-06,
+      "loss": 1.0842,
+      "step": 166620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8803503757176268e-06,
+      "loss": 1.3544,
+      "step": 166625
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8799093032305837e-06,
+      "loss": 1.3019,
+      "step": 166630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8794682771136263e-06,
+      "loss": 1.0822,
+      "step": 166635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.879027297369269e-06,
+      "loss": 1.1564,
+      "step": 166640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8785863640000379e-06,
+      "loss": 1.0917,
+      "step": 166645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.878145477008444e-06,
+      "loss": 1.3843,
+      "step": 166650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8777046363970064e-06,
+      "loss": 1.2256,
+      "step": 166655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8772638421682476e-06,
+      "loss": 1.564,
+      "step": 166660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8768230943246812e-06,
+      "loss": 1.3428,
+      "step": 166665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8763823928688208e-06,
+      "loss": 1.2552,
+      "step": 166670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8759417378031873e-06,
+      "loss": 1.0352,
+      "step": 166675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8755011291302972e-06,
+      "loss": 1.2347,
+      "step": 166680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8750605668526667e-06,
+      "loss": 1.3488,
+      "step": 166685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8746200509728075e-06,
+      "loss": 1.4396,
+      "step": 166690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8741795814932374e-06,
+      "loss": 1.3353,
+      "step": 166695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8737391584164754e-06,
+      "loss": 1.1545,
+      "step": 166700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.873298781745031e-06,
+      "loss": 1.2511,
+      "step": 166705
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8728584514814241e-06,
+      "loss": 1.353,
+      "step": 166710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.872418167628166e-06,
+      "loss": 1.2491,
+      "step": 166715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8719779301877671e-06,
+      "loss": 1.2885,
+      "step": 166720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.871537739162751e-06,
+      "loss": 1.1565,
+      "step": 166725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8710975945556264e-06,
+      "loss": 1.023,
+      "step": 166730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.870657496368904e-06,
+      "loss": 1.1979,
+      "step": 166735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8702174446051024e-06,
+      "loss": 1.163,
+      "step": 166740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8697774392667289e-06,
+      "loss": 1.0525,
+      "step": 166745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8693374803563026e-06,
+      "loss": 1.1461,
+      "step": 166750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8688975678763288e-06,
+      "loss": 1.283,
+      "step": 166755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.868457701829326e-06,
+      "loss": 1.221,
+      "step": 166760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8680178822178008e-06,
+      "loss": 1.5526,
+      "step": 166765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8675781090442668e-06,
+      "loss": 1.1697,
+      "step": 166770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8671383823112389e-06,
+      "loss": 1.4182,
+      "step": 166775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8666987020212258e-06,
+      "loss": 1.0779,
+      "step": 166780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8662590681767334e-06,
+      "loss": 1.1578,
+      "step": 166785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8658194807802776e-06,
+      "loss": 1.3368,
+      "step": 166790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8653799398343697e-06,
+      "loss": 1.1427,
+      "step": 166795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8649404453415154e-06,
+      "loss": 1.297,
+      "step": 166800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8645009973042295e-06,
+      "loss": 1.272,
+      "step": 166805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8640615957250153e-06,
+      "loss": 0.9203,
+      "step": 166810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8636222406063886e-06,
+      "loss": 1.4838,
+      "step": 166815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8631829319508521e-06,
+      "loss": 1.1141,
+      "step": 166820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8627436697609192e-06,
+      "loss": 1.1732,
+      "step": 166825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8623044540390978e-06,
+      "loss": 1.3455,
+      "step": 166830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8618652847878916e-06,
+      "loss": 1.0291,
+      "step": 166835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8614261620098117e-06,
+      "loss": 1.1582,
+      "step": 166840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8609870857073685e-06,
+      "loss": 1.3766,
+      "step": 166845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8605480558830646e-06,
+      "loss": 2.1117,
+      "step": 166850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.860109072539411e-06,
+      "loss": 1.1372,
+      "step": 166855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8596701356789104e-06,
+      "loss": 1.3194,
+      "step": 166860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8592312453040718e-06,
+      "loss": 1.1191,
+      "step": 166865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.858792401417404e-06,
+      "loss": 1.5265,
+      "step": 166870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8583536040214101e-06,
+      "loss": 1.5274,
+      "step": 166875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.857914853118593e-06,
+      "loss": 1.3185,
+      "step": 166880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8574761487114622e-06,
+      "loss": 1.0626,
+      "step": 166885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8570374908025235e-06,
+      "loss": 1.3493,
+      "step": 166890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8565988793942813e-06,
+      "loss": 1.9308,
+      "step": 166895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.856160314489237e-06,
+      "loss": 1.3596,
+      "step": 166900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8557217960898977e-06,
+      "loss": 0.9984,
+      "step": 166905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.855283324198769e-06,
+      "loss": 0.9845,
+      "step": 166910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.854844898818351e-06,
+      "loss": 1.2131,
+      "step": 166915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.854406519951153e-06,
+      "loss": 2.0091,
+      "step": 166920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.853968187599674e-06,
+      "loss": 1.4484,
+      "step": 166925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8535299017664122e-06,
+      "loss": 1.086,
+      "step": 166930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8530916624538819e-06,
+      "loss": 1.3008,
+      "step": 166935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8526534696645803e-06,
+      "loss": 1.2281,
+      "step": 166940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8522153234010077e-06,
+      "loss": 1.1446,
+      "step": 166945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8517772236656695e-06,
+      "loss": 0.9082,
+      "step": 166950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8513391704610629e-06,
+      "loss": 1.3402,
+      "step": 166955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8509011637896956e-06,
+      "loss": 1.5972,
+      "step": 166960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8504632036540637e-06,
+      "loss": 1.063,
+      "step": 166965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8500252900566717e-06,
+      "loss": 1.2201,
+      "step": 166970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8495874230000166e-06,
+      "loss": 1.2612,
+      "step": 166975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8491496024866007e-06,
+      "loss": 1.4123,
+      "step": 166980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8487118285189276e-06,
+      "loss": 1.2765,
+      "step": 166985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8482741010994943e-06,
+      "loss": 1.1978,
+      "step": 166990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8478364202307975e-06,
+      "loss": 1.0975,
+      "step": 166995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8473987859153386e-06,
+      "loss": 1.5374,
+      "step": 167000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8469611981556213e-06,
+      "loss": 1.1985,
+      "step": 167005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8465236569541367e-06,
+      "loss": 1.1689,
+      "step": 167010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8460861623133918e-06,
+      "loss": 1.1971,
+      "step": 167015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8456487142358793e-06,
+      "loss": 1.0417,
+      "step": 167020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8452113127240934e-06,
+      "loss": 1.1176,
+      "step": 167025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8447739577805423e-06,
+      "loss": 1.2586,
+      "step": 167030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8443366494077186e-06,
+      "loss": 1.1764,
+      "step": 167035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.843899387608119e-06,
+      "loss": 0.9548,
+      "step": 167040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8434621723842372e-06,
+      "loss": 1.206,
+      "step": 167045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8430250037385743e-06,
+      "loss": 1.2594,
+      "step": 167050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8425878816736287e-06,
+      "loss": 1.5264,
+      "step": 167055
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8421508061918902e-06,
+      "loss": 1.1678,
+      "step": 167060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8417137772958616e-06,
+      "loss": 1.1352,
+      "step": 167065
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8412767949880327e-06,
+      "loss": 1.1975,
+      "step": 167070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.840839859270901e-06,
+      "loss": 1.4664,
+      "step": 167075
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.840402970146965e-06,
+      "loss": 1.4842,
+      "step": 167080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8399661276187176e-06,
+      "loss": 1.1464,
+      "step": 167085
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8395293316886487e-06,
+      "loss": 1.2738,
+      "step": 167090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8390925823592564e-06,
+      "loss": 1.4272,
+      "step": 167095
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.838655879633038e-06,
+      "loss": 1.4579,
+      "step": 167100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8382192235124841e-06,
+      "loss": 1.2786,
+      "step": 167105
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8377826140000854e-06,
+      "loss": 1.3672,
+      "step": 167110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.837346051098341e-06,
+      "loss": 1.2106,
+      "step": 167115
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8369095348097377e-06,
+      "loss": 1.6847,
+      "step": 167120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8364730651367725e-06,
+      "loss": 1.3003,
+      "step": 167125
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8360366420819386e-06,
+      "loss": 0.9883,
+      "step": 167130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8356002656477268e-06,
+      "loss": 1.1732,
+      "step": 167135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8351639358366258e-06,
+      "loss": 1.2665,
+      "step": 167140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8347276526511305e-06,
+      "loss": 1.0596,
+      "step": 167145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.834291416093734e-06,
+      "loss": 1.1748,
+      "step": 167150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.833855226166924e-06,
+      "loss": 1.2525,
+      "step": 167155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8334190828731947e-06,
+      "loss": 1.227,
+      "step": 167160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8329829862150328e-06,
+      "loss": 1.2591,
+      "step": 167165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.83254693619493e-06,
+      "loss": 1.5074,
+      "step": 167170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8321109328153809e-06,
+      "loss": 1.148,
+      "step": 167175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8316749760788698e-06,
+      "loss": 1.4005,
+      "step": 167180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8312390659878866e-06,
+      "loss": 1.4733,
+      "step": 167185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8308032025449218e-06,
+      "loss": 1.0707,
+      "step": 167190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8303673857524662e-06,
+      "loss": 1.3708,
+      "step": 167195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8299316156130075e-06,
+      "loss": 1.3596,
+      "step": 167200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8294958921290306e-06,
+      "loss": 1.2059,
+      "step": 167205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.829060215303028e-06,
+      "loss": 1.3015,
+      "step": 167210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8286245851374873e-06,
+      "loss": 1.0479,
+      "step": 167215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8281890016348935e-06,
+      "loss": 1.2603,
+      "step": 167220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.827753464797738e-06,
+      "loss": 1.0768,
+      "step": 167225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8273179746285053e-06,
+      "loss": 1.1885,
+      "step": 167230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8268825311296813e-06,
+      "loss": 1.2813,
+      "step": 167235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8264471343037527e-06,
+      "loss": 1.0838,
+      "step": 167240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.826011784153211e-06,
+      "loss": 1.1611,
+      "step": 167245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8255764806805376e-06,
+      "loss": 1.2286,
+      "step": 167250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.825141223888216e-06,
+      "loss": 1.0368,
+      "step": 167255
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8247060137787354e-06,
+      "loss": 1.4607,
+      "step": 167260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8242708503545836e-06,
+      "loss": 1.3513,
+      "step": 167265
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.82383573361824e-06,
+      "loss": 1.1944,
+      "step": 167270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8234006635721936e-06,
+      "loss": 1.0165,
+      "step": 167275
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8229656402189254e-06,
+      "loss": 1.2143,
+      "step": 167280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8225306635609208e-06,
+      "loss": 0.9839,
+      "step": 167285
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8220957336006672e-06,
+      "loss": 1.2465,
+      "step": 167290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8216608503406463e-06,
+      "loss": 1.3866,
+      "step": 167295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8212260137833372e-06,
+      "loss": 1.1726,
+      "step": 167300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8207912239312264e-06,
+      "loss": 1.856,
+      "step": 167305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8203564807868e-06,
+      "loss": 1.0681,
+      "step": 167310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8199217843525353e-06,
+      "loss": 1.0623,
+      "step": 167315
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8194871346309184e-06,
+      "loss": 1.4019,
+      "step": 167320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.81905253162443e-06,
+      "loss": 1.4513,
+      "step": 167325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8186179753355503e-06,
+      "loss": 1.1848,
+      "step": 167330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8181834657667618e-06,
+      "loss": 1.4361,
+      "step": 167335
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.817749002920548e-06,
+      "loss": 1.1649,
+      "step": 167340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8173145867993892e-06,
+      "loss": 1.2492,
+      "step": 167345
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8168802174057631e-06,
+      "loss": 1.2008,
+      "step": 167350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8164458947421516e-06,
+      "loss": 1.272,
+      "step": 167355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8160116188110377e-06,
+      "loss": 1.2423,
+      "step": 167360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8155773896148977e-06,
+      "loss": 1.2518,
+      "step": 167365
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8151432071562148e-06,
+      "loss": 1.3909,
+      "step": 167370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8147090714374639e-06,
+      "loss": 1.1575,
+      "step": 167375
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.814274982461126e-06,
+      "loss": 1.5125,
+      "step": 167380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8138409402296842e-06,
+      "loss": 1.1899,
+      "step": 167385
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8134069447456127e-06,
+      "loss": 1.4704,
+      "step": 167390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8129729960113885e-06,
+      "loss": 1.2662,
+      "step": 167395
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.812539094029493e-06,
+      "loss": 1.3024,
+      "step": 167400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.812105238802404e-06,
+      "loss": 1.0723,
+      "step": 167405
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8116714303325988e-06,
+      "loss": 1.1545,
+      "step": 167410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8112376686225508e-06,
+      "loss": 0.9623,
+      "step": 167415
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8108039536747424e-06,
+      "loss": 1.2531,
+      "step": 167420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.810370285491646e-06,
+      "loss": 1.4966,
+      "step": 167425
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8099366640757398e-06,
+      "loss": 1.2162,
+      "step": 167430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.809503089429504e-06,
+      "loss": 1.0082,
+      "step": 167435
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8090695615554099e-06,
+      "loss": 1.2575,
+      "step": 167440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.808636080455931e-06,
+      "loss": 1.2644,
+      "step": 167445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8082026461335468e-06,
+      "loss": 1.2379,
+      "step": 167450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8077692585907336e-06,
+      "loss": 1.1085,
+      "step": 167455
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.807335917829961e-06,
+      "loss": 1.0771,
+      "step": 167460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8069026238537102e-06,
+      "loss": 1.0043,
+      "step": 167465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8064693766644491e-06,
+      "loss": 1.311,
+      "step": 167470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8060361762646572e-06,
+      "loss": 1.0557,
+      "step": 167475
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8056030226568034e-06,
+      "loss": 1.4422,
+      "step": 167480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8051699158433667e-06,
+      "loss": 0.8847,
+      "step": 167485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.804736855826814e-06,
+      "loss": 1.494,
+      "step": 167490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8043038426096227e-06,
+      "loss": 1.1366,
+      "step": 167495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.803870876194267e-06,
+      "loss": 1.3509,
+      "step": 167500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.803437956583216e-06,
+      "loss": 1.216,
+      "step": 167505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8030050837789404e-06,
+      "loss": 1.3845,
+      "step": 167510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8025722577839156e-06,
+      "loss": 1.2786,
+      "step": 167515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.802139478600614e-06,
+      "loss": 1.31,
+      "step": 167520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.801706746231503e-06,
+      "loss": 1.0311,
+      "step": 167525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.801274060679058e-06,
+      "loss": 1.2034,
+      "step": 167530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8008414219457483e-06,
+      "loss": 1.1587,
+      "step": 167535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8004088300340406e-06,
+      "loss": 1.3303,
+      "step": 167540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7999762849464097e-06,
+      "loss": 1.0853,
+      "step": 167545
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.799543786685327e-06,
+      "loss": 1.3462,
+      "step": 167550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7991113352532597e-06,
+      "loss": 1.0067,
+      "step": 167555
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.798678930652674e-06,
+      "loss": 1.1818,
+      "step": 167560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7982465728860443e-06,
+      "loss": 1.2219,
+      "step": 167565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7978142619558404e-06,
+      "loss": 1.2354,
+      "step": 167570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7973819978645246e-06,
+      "loss": 0.9056,
+      "step": 167575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.796949780614573e-06,
+      "loss": 1.2762,
+      "step": 167580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.796517610208448e-06,
+      "loss": 0.8156,
+      "step": 167585
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7960854866486188e-06,
+      "loss": 1.4879,
+      "step": 167590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7956534099375578e-06,
+      "loss": 1.122,
+      "step": 167595
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7952213800777274e-06,
+      "loss": 1.2524,
+      "step": 167600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7947893970715935e-06,
+      "loss": 1.3328,
+      "step": 167605
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7943574609216252e-06,
+      "loss": 0.9993,
+      "step": 167610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7939255716302929e-06,
+      "loss": 0.9734,
+      "step": 167615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7934937292000576e-06,
+      "loss": 1.214,
+      "step": 167620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7930619336333855e-06,
+      "loss": 1.2338,
+      "step": 167625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7926301849327466e-06,
+      "loss": 0.9549,
+      "step": 167630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7921984831006e-06,
+      "loss": 1.2783,
+      "step": 167635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7917668281394162e-06,
+      "loss": 0.9742,
+      "step": 167640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7913352200516599e-06,
+      "loss": 1.147,
+      "step": 167645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7909036588397955e-06,
+      "loss": 0.979,
+      "step": 167650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7904721445062835e-06,
+      "loss": 1.4015,
+      "step": 167655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7900406770535906e-06,
+      "loss": 1.0565,
+      "step": 167660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7896092564841838e-06,
+      "loss": 1.2145,
+      "step": 167665
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7891778828005223e-06,
+      "loss": 0.8699,
+      "step": 167670
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7887465560050731e-06,
+      "loss": 1.2641,
+      "step": 167675
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7883152761002953e-06,
+      "loss": 1.0577,
+      "step": 167680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.787884043088657e-06,
+      "loss": 1.5912,
+      "step": 167685
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7874528569726146e-06,
+      "loss": 1.0809,
+      "step": 167690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7870217177546355e-06,
+      "loss": 1.3754,
+      "step": 167695
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7865906254371812e-06,
+      "loss": 1.2265,
+      "step": 167700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.786159580022706e-06,
+      "loss": 1.1626,
+      "step": 167705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7857285815136826e-06,
+      "loss": 1.0823,
+      "step": 167710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7852976299125669e-06,
+      "loss": 1.055,
+      "step": 167715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7848667252218178e-06,
+      "loss": 1.1248,
+      "step": 167720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7844358674439012e-06,
+      "loss": 1.3187,
+      "step": 167725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7840050565812716e-06,
+      "loss": 1.5468,
+      "step": 167730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7835742926363918e-06,
+      "loss": 1.0004,
+      "step": 167735
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7831435756117244e-06,
+      "loss": 1.0587,
+      "step": 167740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.782712905509727e-06,
+      "loss": 1.3317,
+      "step": 167745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7822822823328568e-06,
+      "loss": 1.3488,
+      "step": 167750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7818517060835738e-06,
+      "loss": 1.22,
+      "step": 167755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7814211767643396e-06,
+      "loss": 1.0723,
+      "step": 167760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7809906943776123e-06,
+      "loss": 0.9318,
+      "step": 167765
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7805602589258453e-06,
+      "loss": 1.4696,
+      "step": 167770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7801298704114999e-06,
+      "loss": 1.2523,
+      "step": 167775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7796995288370378e-06,
+      "loss": 1.0869,
+      "step": 167780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.779269234204909e-06,
+      "loss": 1.3221,
+      "step": 167785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.778838986517577e-06,
+      "loss": 1.3126,
+      "step": 167790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7784087857774945e-06,
+      "loss": 1.462,
+      "step": 167795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7779786319871195e-06,
+      "loss": 1.3471,
+      "step": 167800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.777548525148911e-06,
+      "loss": 1.2461,
+      "step": 167805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7771184652653228e-06,
+      "loss": 1.442,
+      "step": 167810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7766884523388084e-06,
+      "loss": 1.237,
+      "step": 167815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.776258486371828e-06,
+      "loss": 1.1226,
+      "step": 167820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.775828567366833e-06,
+      "loss": 0.9974,
+      "step": 167825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7753986953262826e-06,
+      "loss": 1.2016,
+      "step": 167830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7749688702526268e-06,
+      "loss": 1.6762,
+      "step": 167835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7745390921483253e-06,
+      "loss": 0.9657,
+      "step": 167840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7741093610158267e-06,
+      "loss": 1.2052,
+      "step": 167845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.773679676857588e-06,
+      "loss": 1.1229,
+      "step": 167850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7732500396760654e-06,
+      "loss": 1.222,
+      "step": 167855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7728204494737101e-06,
+      "loss": 1.0485,
+      "step": 167860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7723909062529721e-06,
+      "loss": 1.6987,
+      "step": 167865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7719614100163074e-06,
+      "loss": 1.3584,
+      "step": 167870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7715319607661718e-06,
+      "loss": 1.1722,
+      "step": 167875
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7711025585050112e-06,
+      "loss": 1.328,
+      "step": 167880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7706732032352837e-06,
+      "loss": 1.2524,
+      "step": 167885
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7702438949594346e-06,
+      "loss": 1.053,
+      "step": 167890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7698146336799206e-06,
+      "loss": 0.9949,
+      "step": 167895
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7693854193991944e-06,
+      "loss": 1.4011,
+      "step": 167900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7689562521197034e-06,
+      "loss": 1.391,
+      "step": 167905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7685271318439001e-06,
+      "loss": 1.2087,
+      "step": 167910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.768098058574228e-06,
+      "loss": 1.3958,
+      "step": 167915
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7676690323131495e-06,
+      "loss": 1.1926,
+      "step": 167920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7672400530631083e-06,
+      "loss": 0.9793,
+      "step": 167925
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7668111208265526e-06,
+      "loss": 1.2651,
+      "step": 167930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7663822356059358e-06,
+      "loss": 1.0283,
+      "step": 167935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7659533974037014e-06,
+      "loss": 1.1436,
+      "step": 167940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7655246062223031e-06,
+      "loss": 1.3905,
+      "step": 167945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7650958620641901e-06,
+      "loss": 1.082,
+      "step": 167950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7646671649318092e-06,
+      "loss": 1.1534,
+      "step": 167955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7642385148276053e-06,
+      "loss": 1.1086,
+      "step": 167960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7638099117540285e-06,
+      "loss": 1.2766,
+      "step": 167965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7633813557135303e-06,
+      "loss": 1.4445,
+      "step": 167970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7629528467085555e-06,
+      "loss": 1.1126,
+      "step": 167975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7625243847415462e-06,
+      "loss": 1.1859,
+      "step": 167980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7620959698149542e-06,
+      "loss": 1.48,
+      "step": 167985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7616676019312273e-06,
+      "loss": 1.2906,
+      "step": 167990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.761239281092807e-06,
+      "loss": 1.2631,
+      "step": 167995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7608110073021434e-06,
+      "loss": 1.2221,
+      "step": 168000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7603827805616812e-06,
+      "loss": 1.1665,
+      "step": 168005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7599546008738588e-06,
+      "loss": 1.1687,
+      "step": 168010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.759526468241134e-06,
+      "loss": 1.1325,
+      "step": 168015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7590983826659446e-06,
+      "loss": 1.3034,
+      "step": 168020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7586703441507335e-06,
+      "loss": 1.562,
+      "step": 168025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7582423526979497e-06,
+      "loss": 1.1546,
+      "step": 168030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7578144083100324e-06,
+      "loss": 1.2551,
+      "step": 168035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7573865109894284e-06,
+      "loss": 1.231,
+      "step": 168040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7569586607385825e-06,
+      "loss": 1.1535,
+      "step": 168045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.756530857559936e-06,
+      "loss": 1.603,
+      "step": 168050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7561031014559304e-06,
+      "loss": 1.222,
+      "step": 168055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7556753924290104e-06,
+      "loss": 1.091,
+      "step": 168060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7552477304816196e-06,
+      "loss": 1.0482,
+      "step": 168065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7548201156161992e-06,
+      "loss": 1.0744,
+      "step": 168070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7543925478351875e-06,
+      "loss": 0.9133,
+      "step": 168075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7539650271410303e-06,
+      "loss": 1.0977,
+      "step": 168080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.75353755353617e-06,
+      "loss": 1.0493,
+      "step": 168085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7531101270230433e-06,
+      "loss": 1.205,
+      "step": 168090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7526827476040954e-06,
+      "loss": 1.3009,
+      "step": 168095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.752255415281765e-06,
+      "loss": 1.2432,
+      "step": 168100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.751828130058487e-06,
+      "loss": 1.1043,
+      "step": 168105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.751400891936712e-06,
+      "loss": 1.4518,
+      "step": 168110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7509737009188755e-06,
+      "loss": 1.399,
+      "step": 168115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7505465570074143e-06,
+      "loss": 1.1574,
+      "step": 168120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7501194602047677e-06,
+      "loss": 1.0526,
+      "step": 168125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.749692410513376e-06,
+      "loss": 1.2948,
+      "step": 168130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7492654079356797e-06,
+      "loss": 1.1677,
+      "step": 168135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7488384524741142e-06,
+      "loss": 1.4286,
+      "step": 168140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7484115441311211e-06,
+      "loss": 1.3919,
+      "step": 168145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.747984682909134e-06,
+      "loss": 1.1899,
+      "step": 168150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7475578688105922e-06,
+      "loss": 0.9731,
+      "step": 168155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7471311018379367e-06,
+      "loss": 1.6014,
+      "step": 168160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7467043819936015e-06,
+      "loss": 1.008,
+      "step": 168165
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7462777092800198e-06,
+      "loss": 0.9555,
+      "step": 168170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7458510836996323e-06,
+      "loss": 1.007,
+      "step": 168175
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7454245052548768e-06,
+      "loss": 1.3166,
+      "step": 168180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.744997973948186e-06,
+      "loss": 1.2644,
+      "step": 168185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7445714897819976e-06,
+      "loss": 1.15,
+      "step": 168190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7441450527587445e-06,
+      "loss": 1.1596,
+      "step": 168195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7437186628808656e-06,
+      "loss": 1.3119,
+      "step": 168200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7432923201507924e-06,
+      "loss": 1.5066,
+      "step": 168205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7428660245709617e-06,
+      "loss": 1.2947,
+      "step": 168210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7424397761438083e-06,
+      "loss": 1.2202,
+      "step": 168215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7420135748717626e-06,
+      "loss": 1.3657,
+      "step": 168220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7415874207572602e-06,
+      "loss": 1.3633,
+      "step": 168225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7411613138027383e-06,
+      "loss": 1.6214,
+      "step": 168230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.740735254010626e-06,
+      "loss": 1.1308,
+      "step": 168235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.740309241383359e-06,
+      "loss": 1.129,
+      "step": 168240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7398832759233663e-06,
+      "loss": 1.1271,
+      "step": 168245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.739457357633083e-06,
+      "loss": 1.1809,
+      "step": 168250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7390314865149427e-06,
+      "loss": 1.1635,
+      "step": 168255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7386056625713766e-06,
+      "loss": 1.3946,
+      "step": 168260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7381798858048138e-06,
+      "loss": 1.3938,
+      "step": 168265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7377541562176869e-06,
+      "loss": 1.7566,
+      "step": 168270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7373284738124297e-06,
+      "loss": 1.0739,
+      "step": 168275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.736902838591471e-06,
+      "loss": 1.2789,
+      "step": 168280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7364772505572391e-06,
+      "loss": 1.0665,
+      "step": 168285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7360517097121676e-06,
+      "loss": 0.985,
+      "step": 168290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7356262160586866e-06,
+      "loss": 1.1863,
+      "step": 168295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7352007695992234e-06,
+      "loss": 1.2765,
+      "step": 168300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7347753703362114e-06,
+      "loss": 1.501,
+      "step": 168305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7343500182720774e-06,
+      "loss": 0.9583,
+      "step": 168310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7339247134092452e-06,
+      "loss": 1.1816,
+      "step": 168315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.733499455750155e-06,
+      "loss": 1.1263,
+      "step": 168320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7330742452972281e-06,
+      "loss": 1.2331,
+      "step": 168325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7326490820528907e-06,
+      "loss": 1.2471,
+      "step": 168330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7322239660195773e-06,
+      "loss": 1.1064,
+      "step": 168335
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.731798897199708e-06,
+      "loss": 1.0452,
+      "step": 168340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7313738755957177e-06,
+      "loss": 1.0894,
+      "step": 168345
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7309489012100268e-06,
+      "loss": 1.1902,
+      "step": 168350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7305239740450686e-06,
+      "loss": 1.2177,
+      "step": 168355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7300990941032625e-06,
+      "loss": 1.0087,
+      "step": 168360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7296742613870388e-06,
+      "loss": 1.0918,
+      "step": 168365
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7292494758988266e-06,
+      "loss": 1.2915,
+      "step": 168370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7288247376410472e-06,
+      "loss": 0.9912,
+      "step": 168375
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7284000466161254e-06,
+      "loss": 1.4074,
+      "step": 168380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7279754028264883e-06,
+      "loss": 1.2997,
+      "step": 168385
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7275508062745639e-06,
+      "loss": 1.1961,
+      "step": 168390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.72712625696277e-06,
+      "loss": 1.1624,
+      "step": 168395
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7267017548935384e-06,
+      "loss": 1.2918,
+      "step": 168400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.726277300069289e-06,
+      "loss": 0.9976,
+      "step": 168405
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7258528924924434e-06,
+      "loss": 1.1826,
+      "step": 168410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7254285321654285e-06,
+      "loss": 1.1692,
+      "step": 168415
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7250042190906691e-06,
+      "loss": 1.488,
+      "step": 168420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7245799532705876e-06,
+      "loss": 1.5553,
+      "step": 168425
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.724155734707602e-06,
+      "loss": 0.957,
+      "step": 168430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7237315634041384e-06,
+      "loss": 1.1292,
+      "step": 168435
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7233074393626214e-06,
+      "loss": 1.0676,
+      "step": 168440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.722883362585468e-06,
+      "loss": 1.204,
+      "step": 168445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.722459333075106e-06,
+      "loss": 1.1779,
+      "step": 168450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7220353508339494e-06,
+      "loss": 1.188,
+      "step": 168455
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7216114158644248e-06,
+      "loss": 1.0809,
+      "step": 168460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7211875281689528e-06,
+      "loss": 1.2872,
+      "step": 168465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7207636877499534e-06,
+      "loss": 1.3613,
+      "step": 168470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7203398946098438e-06,
+      "loss": 1.3334,
+      "step": 168475
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7199161487510463e-06,
+      "loss": 1.2695,
+      "step": 168480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7194924501759835e-06,
+      "loss": 1.1635,
+      "step": 168485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7190687988870725e-06,
+      "loss": 1.3184,
+      "step": 168490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71864519488673e-06,
+      "loss": 1.5757,
+      "step": 168495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7182216381773787e-06,
+      "loss": 1.0483,
+      "step": 168500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7177981287614376e-06,
+      "loss": 1.1155,
+      "step": 168505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7173746666413215e-06,
+      "loss": 1.2517,
+      "step": 168510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716951251819453e-06,
+      "loss": 1.1281,
+      "step": 168515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7165278842982491e-06,
+      "loss": 1.3129,
+      "step": 168520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7161045640801232e-06,
+      "loss": 1.2026,
+      "step": 168525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7156812911674958e-06,
+      "loss": 1.1079,
+      "step": 168530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715258065562786e-06,
+      "loss": 1.1036,
+      "step": 168535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7148348872684062e-06,
+      "loss": 1.1512,
+      "step": 168540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714411756286778e-06,
+      "loss": 1.2294,
+      "step": 168545
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7139886726203125e-06,
+      "loss": 1.5888,
+      "step": 168550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7135656362714316e-06,
+      "loss": 1.3529,
+      "step": 168555
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7131426472425438e-06,
+      "loss": 1.0544,
+      "step": 168560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7127197055360723e-06,
+      "loss": 1.5502,
+      "step": 168565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7122968111544258e-06,
+      "loss": 1.1023,
+      "step": 168570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7118739641000226e-06,
+      "loss": 1.1816,
+      "step": 168575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7114511643752785e-06,
+      "loss": 1.1609,
+      "step": 168580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711028411982607e-06,
+      "loss": 1.3003,
+      "step": 168585
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7106057069244186e-06,
+      "loss": 1.3249,
+      "step": 168590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7101830492031301e-06,
+      "loss": 1.0308,
+      "step": 168595
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7097604388211576e-06,
+      "loss": 0.9525,
+      "step": 168600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7093378757809086e-06,
+      "loss": 0.9355,
+      "step": 168605
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.708915360084803e-06,
+      "loss": 1.2671,
+      "step": 168610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7084928917352496e-06,
+      "loss": 1.1397,
+      "step": 168615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7080704707346573e-06,
+      "loss": 1.1504,
+      "step": 168620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7076480970854458e-06,
+      "loss": 1.0973,
+      "step": 168625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7072257707900253e-06,
+      "loss": 1.2112,
+      "step": 168630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7068034918508048e-06,
+      "loss": 0.9438,
+      "step": 168635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7063812602701958e-06,
+      "loss": 1.1853,
+      "step": 168640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7059590760506094e-06,
+      "loss": 1.1841,
+      "step": 168645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7055369391944598e-06,
+      "loss": 1.6015,
+      "step": 168650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7051148497041537e-06,
+      "loss": 1.2652,
+      "step": 168655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7046928075821056e-06,
+      "loss": 1.2417,
+      "step": 168660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7042708128307205e-06,
+      "loss": 1.2088,
+      "step": 168665
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7038488654524122e-06,
+      "loss": 1.3904,
+      "step": 168670
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7034269654495894e-06,
+      "loss": 1.482,
+      "step": 168675
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7030051128246628e-06,
+      "loss": 1.1259,
+      "step": 168680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7025833075800358e-06,
+      "loss": 1.1527,
+      "step": 168685
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.702161549718122e-06,
+      "loss": 1.304,
+      "step": 168690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7017398392413298e-06,
+      "loss": 1.2614,
+      "step": 168695
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.701318176152068e-06,
+      "loss": 1.0553,
+      "step": 168700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7008965604527394e-06,
+      "loss": 1.3581,
+      "step": 168705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7004749921457574e-06,
+      "loss": 1.3248,
+      "step": 168710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7000534712335248e-06,
+      "loss": 1.4401,
+      "step": 168715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6996319977184516e-06,
+      "loss": 1.2234,
+      "step": 168720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.699210571602946e-06,
+      "loss": 1.2085,
+      "step": 168725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6987891928894129e-06,
+      "loss": 1.274,
+      "step": 168730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6983678615802556e-06,
+      "loss": 1.1615,
+      "step": 168735
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6979465776778836e-06,
+      "loss": 1.2518,
+      "step": 168740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6975253411847038e-06,
+      "loss": 1.2147,
+      "step": 168745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6971041521031172e-06,
+      "loss": 1.1235,
+      "step": 168750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6966830104355335e-06,
+      "loss": 1.3552,
+      "step": 168755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6962619161843551e-06,
+      "loss": 1.1563,
+      "step": 168760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6958408693519857e-06,
+      "loss": 1.2268,
+      "step": 168765
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.695419869940834e-06,
+      "loss": 1.1728,
+      "step": 168770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6949989179533022e-06,
+      "loss": 1.1783,
+      "step": 168775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.694578013391791e-06,
+      "loss": 1.0595,
+      "step": 168780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6941571562587055e-06,
+      "loss": 1.3147,
+      "step": 168785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6937363465564537e-06,
+      "loss": 1.2989,
+      "step": 168790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6933155842874338e-06,
+      "loss": 1.3207,
+      "step": 168795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.692894869454048e-06,
+      "loss": 1.2522,
+      "step": 168800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6924742020587027e-06,
+      "loss": 1.2109,
+      "step": 168805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6920535821037952e-06,
+      "loss": 1.2112,
+      "step": 168810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.69163300959173e-06,
+      "loss": 1.2212,
+      "step": 168815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6912124845249122e-06,
+      "loss": 1.593,
+      "step": 168820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6907920069057393e-06,
+      "loss": 1.1664,
+      "step": 168825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6903715767366114e-06,
+      "loss": 1.2614,
+      "step": 168830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.689951194019931e-06,
+      "loss": 1.1413,
+      "step": 168835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6895308587581004e-06,
+      "loss": 1.3687,
+      "step": 168840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6891105709535195e-06,
+      "loss": 1.6711,
+      "step": 168845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6886903306085844e-06,
+      "loss": 1.6668,
+      "step": 168850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6882701377256982e-06,
+      "loss": 1.2222,
+      "step": 168855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6878499923072623e-06,
+      "loss": 1.3803,
+      "step": 168860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.68742989435567e-06,
+      "loss": 1.6342,
+      "step": 168865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6870098438733273e-06,
+      "loss": 1.331,
+      "step": 168870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.686589840862628e-06,
+      "loss": 1.3303,
+      "step": 168875
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.686169885325971e-06,
+      "loss": 1.0986,
+      "step": 168880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6857499772657582e-06,
+      "loss": 1.3138,
+      "step": 168885
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.685330116684385e-06,
+      "loss": 1.1547,
+      "step": 168890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6849103035842462e-06,
+      "loss": 1.1117,
+      "step": 168895
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6844905379677423e-06,
+      "loss": 1.7624,
+      "step": 168900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6840708198372713e-06,
+      "loss": 1.1458,
+      "step": 168905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.683651149195228e-06,
+      "loss": 0.975,
+      "step": 168910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6832315260440102e-06,
+      "loss": 0.9074,
+      "step": 168915
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.682811950386015e-06,
+      "loss": 0.9708,
+      "step": 168920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6823924222236332e-06,
+      "loss": 1.0179,
+      "step": 168925
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6819729415592655e-06,
+      "loss": 1.3068,
+      "step": 168930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6815535083953082e-06,
+      "loss": 1.081,
+      "step": 168935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6811341227341548e-06,
+      "loss": 1.0862,
+      "step": 168940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6807147845781969e-06,
+      "loss": 1.5101,
+      "step": 168945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6802954939298322e-06,
+      "loss": 1.2068,
+      "step": 168950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6798762507914579e-06,
+      "loss": 1.1487,
+      "step": 168955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.679457055165462e-06,
+      "loss": 1.6021,
+      "step": 168960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6790379070542452e-06,
+      "loss": 1.0718,
+      "step": 168965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6786188064601938e-06,
+      "loss": 1.2939,
+      "step": 168970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6781997533857053e-06,
+      "loss": 1.2649,
+      "step": 168975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6777807478331742e-06,
+      "loss": 1.2243,
+      "step": 168980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.677361789804991e-06,
+      "loss": 1.3171,
+      "step": 168985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.676942879303546e-06,
+      "loss": 1.2737,
+      "step": 168990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6765240163312346e-06,
+      "loss": 1.0865,
+      "step": 168995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6761052008904488e-06,
+      "loss": 1.1063,
+      "step": 169000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6756864329835808e-06,
+      "loss": 1.0149,
+      "step": 169005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6752677126130169e-06,
+      "loss": 1.3293,
+      "step": 169010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6748490397811545e-06,
+      "loss": 1.3897,
+      "step": 169015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.67443041449038e-06,
+      "loss": 1.7407,
+      "step": 169020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6740118367430858e-06,
+      "loss": 1.1012,
+      "step": 169025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6735933065416643e-06,
+      "loss": 1.1765,
+      "step": 169030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6731748238885026e-06,
+      "loss": 1.5208,
+      "step": 169035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.67275638878599e-06,
+      "loss": 1.0933,
+      "step": 169040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6723380012365165e-06,
+      "loss": 1.5369,
+      "step": 169045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.671919661242475e-06,
+      "loss": 1.5488,
+      "step": 169050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6715013688062486e-06,
+      "loss": 1.5501,
+      "step": 169055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6710831239302315e-06,
+      "loss": 1.1121,
+      "step": 169060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6706649266168073e-06,
+      "loss": 1.3715,
+      "step": 169065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6702467768683684e-06,
+      "loss": 1.2101,
+      "step": 169070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6698286746872983e-06,
+      "loss": 1.1637,
+      "step": 169075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6694106200759897e-06,
+      "loss": 1.4781,
+      "step": 169080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.668992613036824e-06,
+      "loss": 1.4456,
+      "step": 169085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6685746535721914e-06,
+      "loss": 1.1422,
+      "step": 169090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6681567416844812e-06,
+      "loss": 1.3089,
+      "step": 169095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6677388773760772e-06,
+      "loss": 1.2892,
+      "step": 169100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.667321060649364e-06,
+      "loss": 1.1233,
+      "step": 169105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6669032915067307e-06,
+      "loss": 1.1779,
+      "step": 169110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.66648556995056e-06,
+      "loss": 1.3563,
+      "step": 169115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6660678959832387e-06,
+      "loss": 0.8973,
+      "step": 169120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6656502696071542e-06,
+      "loss": 0.9184,
+      "step": 169125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6652326908246897e-06,
+      "loss": 1.3299,
+      "step": 169130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6648151596382268e-06,
+      "loss": 1.1958,
+      "step": 169135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6643976760501523e-06,
+      "loss": 1.1558,
+      "step": 169140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6639802400628524e-06,
+      "loss": 1.2292,
+      "step": 169145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6635628516787094e-06,
+      "loss": 1.8316,
+      "step": 169150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.663145510900105e-06,
+      "loss": 1.0236,
+      "step": 169155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6627282177294224e-06,
+      "loss": 1.1766,
+      "step": 169160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6623109721690478e-06,
+      "loss": 1.02,
+      "step": 169165
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6618937742213604e-06,
+      "loss": 1.209,
+      "step": 169170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6614766238887469e-06,
+      "loss": 1.1708,
+      "step": 169175
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6610595211735836e-06,
+      "loss": 1.2768,
+      "step": 169180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.660642466078256e-06,
+      "loss": 1.5145,
+      "step": 169185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.660225458605148e-06,
+      "loss": 1.134,
+      "step": 169190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6598084987566377e-06,
+      "loss": 1.4081,
+      "step": 169195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.659391586535104e-06,
+      "loss": 1.2971,
+      "step": 169200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6589747219429309e-06,
+      "loss": 1.4465,
+      "step": 169205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6585579049825007e-06,
+      "loss": 1.3767,
+      "step": 169210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6581411356561916e-06,
+      "loss": 1.306,
+      "step": 169215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6577244139663807e-06,
+      "loss": 0.8264,
+      "step": 169220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6573077399154525e-06,
+      "loss": 1.1943,
+      "step": 169225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6568911135057818e-06,
+      "loss": 1.0447,
+      "step": 169230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6564745347397504e-06,
+      "loss": 1.1795,
+      "step": 169235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6560580036197381e-06,
+      "loss": 1.2903,
+      "step": 169240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6556415201481246e-06,
+      "loss": 1.4709,
+      "step": 169245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.655225084327282e-06,
+      "loss": 1.2516,
+      "step": 169250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6548086961595933e-06,
+      "loss": 1.5421,
+      "step": 169255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6543923556474374e-06,
+      "loss": 1.1217,
+      "step": 169260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6539760627931879e-06,
+      "loss": 1.1757,
+      "step": 169265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6535598175992262e-06,
+      "loss": 1.4632,
+      "step": 169270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6531436200679241e-06,
+      "loss": 1.1814,
+      "step": 169275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6527274702016626e-06,
+      "loss": 0.9379,
+      "step": 169280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.652311368002818e-06,
+      "loss": 1.0907,
+      "step": 169285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.651895313473766e-06,
+      "loss": 1.3885,
+      "step": 169290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6514793066168821e-06,
+      "loss": 1.4928,
+      "step": 169295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.651063347434536e-06,
+      "loss": 1.1632,
+      "step": 169300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6506474359291136e-06,
+      "loss": 1.2009,
+      "step": 169305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6502315721029849e-06,
+      "loss": 1.1824,
+      "step": 169310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6498157559585238e-06,
+      "loss": 1.3268,
+      "step": 169315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6493999874981071e-06,
+      "loss": 1.1891,
+      "step": 169320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.648984266724105e-06,
+      "loss": 1.21,
+      "step": 169325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.648568593638895e-06,
+      "loss": 1.1296,
+      "step": 169330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6481529682448527e-06,
+      "loss": 1.1444,
+      "step": 169335
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6477373905443493e-06,
+      "loss": 1.1478,
+      "step": 169340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6473218605397546e-06,
+      "loss": 1.5026,
+      "step": 169345
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.646906378233445e-06,
+      "loss": 1.3178,
+      "step": 169350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6464909436277953e-06,
+      "loss": 1.1652,
+      "step": 169355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6460755567251752e-06,
+      "loss": 1.1715,
+      "step": 169360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6456602175279546e-06,
+      "loss": 1.4379,
+      "step": 169365
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6452449260385062e-06,
+      "loss": 1.2008,
+      "step": 169370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.644829682259207e-06,
+      "loss": 1.4362,
+      "step": 169375
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6444144861924206e-06,
+      "loss": 1.1144,
+      "step": 169380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.643999337840525e-06,
+      "loss": 1.3883,
+      "step": 169385
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6435842372058863e-06,
+      "loss": 1.104,
+      "step": 169390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6431691842908726e-06,
+      "loss": 1.3067,
+      "step": 169395
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6427541790978608e-06,
+      "loss": 1.19,
+      "step": 169400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6423392216292189e-06,
+      "loss": 1.4422,
+      "step": 169405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6419243118873118e-06,
+      "loss": 1.6175,
+      "step": 169410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6415094498745155e-06,
+      "loss": 1.3414,
+      "step": 169415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6410946355931934e-06,
+      "loss": 1.2851,
+      "step": 169420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.640679869045717e-06,
+      "loss": 1.3634,
+      "step": 169425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6402651502344568e-06,
+      "loss": 1.3441,
+      "step": 169430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6398504791617786e-06,
+      "loss": 1.2405,
+      "step": 169435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6394358558300483e-06,
+      "loss": 1.3667,
+      "step": 169440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.639021280241636e-06,
+      "loss": 1.2387,
+      "step": 169445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6386067523989125e-06,
+      "loss": 1.7218,
+      "step": 169450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6381922723042409e-06,
+      "loss": 1.2268,
+      "step": 169455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6377778399599865e-06,
+      "loss": 1.0999,
+      "step": 169460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.637363455368518e-06,
+      "loss": 1.3343,
+      "step": 169465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.636949118532205e-06,
+      "loss": 1.9811,
+      "step": 169470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6365348294534078e-06,
+      "loss": 1.0574,
+      "step": 169475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6361205881344966e-06,
+      "loss": 1.1984,
+      "step": 169480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.635706394577834e-06,
+      "loss": 1.1058,
+      "step": 169485
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.635292248785787e-06,
+      "loss": 1.083,
+      "step": 169490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6348781507607226e-06,
+      "loss": 1.5247,
+      "step": 169495
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.634464100505002e-06,
+      "loss": 1.3411,
+      "step": 169500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6340500980209916e-06,
+      "loss": 1.4463,
+      "step": 169505
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6336361433110526e-06,
+      "loss": 1.1749,
+      "step": 169510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6332222363775508e-06,
+      "loss": 1.1254,
+      "step": 169515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.632808377222852e-06,
+      "loss": 1.5432,
+      "step": 169520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6323945658493156e-06,
+      "loss": 1.142,
+      "step": 169525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6319808022593098e-06,
+      "loss": 1.0217,
+      "step": 169530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6315670864551913e-06,
+      "loss": 1.2455,
+      "step": 169535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6311534184393263e-06,
+      "loss": 1.4199,
+      "step": 169540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6307397982140783e-06,
+      "loss": 1.5135,
+      "step": 169545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6303262257818076e-06,
+      "loss": 1.266,
+      "step": 169550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6299127011448734e-06,
+      "loss": 1.2798,
+      "step": 169555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6294992243056395e-06,
+      "loss": 1.578,
+      "step": 169560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6290857952664707e-06,
+      "loss": 1.0472,
+      "step": 169565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6286724140297205e-06,
+      "loss": 0.9353,
+      "step": 169570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.628259080597756e-06,
+      "loss": 0.997,
+      "step": 169575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6278457949729332e-06,
+      "loss": 1.1868,
+      "step": 169580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6274325571576156e-06,
+      "loss": 1.1457,
+      "step": 169585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.627019367154159e-06,
+      "loss": 1.5058,
+      "step": 169590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6266062249649272e-06,
+      "loss": 1.4503,
+      "step": 169595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6261931305922785e-06,
+      "loss": 1.5953,
+      "step": 169600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6257800840385651e-06,
+      "loss": 1.0173,
+      "step": 169605
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6253670853061576e-06,
+      "loss": 1.2672,
+      "step": 169610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6249541343974073e-06,
+      "loss": 0.9669,
+      "step": 169615
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6245412313146713e-06,
+      "loss": 1.198,
+      "step": 169620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6241283760603122e-06,
+      "loss": 1.1346,
+      "step": 169625
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6237155686366824e-06,
+      "loss": 1.4793,
+      "step": 169630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6233028090461422e-06,
+      "loss": 1.3283,
+      "step": 169635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.62289009729105e-06,
+      "loss": 1.1451,
+      "step": 169640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6224774333737615e-06,
+      "loss": 1.5393,
+      "step": 169645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6220648172966302e-06,
+      "loss": 1.1636,
+      "step": 169650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6216522490620157e-06,
+      "loss": 1.2341,
+      "step": 169655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6212397286722747e-06,
+      "loss": 1.2393,
+      "step": 169660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6208272561297623e-06,
+      "loss": 1.1274,
+      "step": 169665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6204148314368296e-06,
+      "loss": 1.1157,
+      "step": 169670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.620002454595836e-06,
+      "loss": 1.3998,
+      "step": 169675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6195901256091374e-06,
+      "loss": 1.1586,
+      "step": 169680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.619177844479084e-06,
+      "loss": 0.9804,
+      "step": 169685
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6187656112080353e-06,
+      "loss": 1.4123,
+      "step": 169690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6183534257983436e-06,
+      "loss": 1.3298,
+      "step": 169695
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.617941288252356e-06,
+      "loss": 0.917,
+      "step": 169700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6175291985724374e-06,
+      "loss": 1.2166,
+      "step": 169705
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6171171567609356e-06,
+      "loss": 1.0357,
+      "step": 169710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6167051628202013e-06,
+      "loss": 1.1266,
+      "step": 169715
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6162932167525914e-06,
+      "loss": 0.9721,
+      "step": 169720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6158813185604549e-06,
+      "loss": 1.0143,
+      "step": 169725
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.615469468246147e-06,
+      "loss": 1.3394,
+      "step": 169730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6150576658120154e-06,
+      "loss": 1.056,
+      "step": 169735
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6146459112604173e-06,
+      "loss": 1.1605,
+      "step": 169740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6142342045937e-06,
+      "loss": 1.4673,
+      "step": 169745
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6138225458142143e-06,
+      "loss": 0.9575,
+      "step": 169750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6134109349243143e-06,
+      "loss": 1.1566,
+      "step": 169755
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6129993719263493e-06,
+      "loss": 1.039,
+      "step": 169760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6125878568226672e-06,
+      "loss": 1.5476,
+      "step": 169765
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6121763896156185e-06,
+      "loss": 1.2113,
+      "step": 169770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6117649703075566e-06,
+      "loss": 1.3758,
+      "step": 169775
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6113535989008267e-06,
+      "loss": 1.0385,
+      "step": 169780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6109422753977811e-06,
+      "loss": 1.4905,
+      "step": 169785
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6105309998007646e-06,
+      "loss": 1.0871,
+      "step": 169790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.610119772112132e-06,
+      "loss": 1.3224,
+      "step": 169795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6097085923342249e-06,
+      "loss": 1.1928,
+      "step": 169800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6092974604693967e-06,
+      "loss": 1.2981,
+      "step": 169805
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6088863765199936e-06,
+      "loss": 1.1329,
+      "step": 169810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.608475340488359e-06,
+      "loss": 1.0503,
+      "step": 169815
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6080643523768447e-06,
+      "loss": 1.6476,
+      "step": 169820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6076534121877985e-06,
+      "loss": 1.303,
+      "step": 169825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.607242519923563e-06,
+      "loss": 1.3037,
+      "step": 169830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6068316755864888e-06,
+      "loss": 1.1771,
+      "step": 169835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.606420879178917e-06,
+      "loss": 1.2079,
+      "step": 169840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6060101307031973e-06,
+      "loss": 1.039,
+      "step": 169845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6055994301616774e-06,
+      "loss": 1.1931,
+      "step": 169850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6051887775566988e-06,
+      "loss": 1.1884,
+      "step": 169855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6047781728906053e-06,
+      "loss": 1.109,
+      "step": 169860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.604367616165744e-06,
+      "loss": 0.9493,
+      "step": 169865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6039571073844616e-06,
+      "loss": 1.3154,
+      "step": 169870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6035466465491001e-06,
+      "loss": 1.2844,
+      "step": 169875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6031362336620005e-06,
+      "loss": 1.2709,
+      "step": 169880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.60272586872551e-06,
+      "loss": 1.5311,
+      "step": 169885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.602315551741973e-06,
+      "loss": 1.1245,
+      "step": 169890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6019052827137283e-06,
+      "loss": 1.5795,
+      "step": 169895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6014950616431246e-06,
+      "loss": 1.1947,
+      "step": 169900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6010848885325004e-06,
+      "loss": 1.0698,
+      "step": 169905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6006747633841968e-06,
+      "loss": 1.0153,
+      "step": 169910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6002646862005577e-06,
+      "loss": 1.1899,
+      "step": 169915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5998546569839278e-06,
+      "loss": 1.125,
+      "step": 169920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5994446757366432e-06,
+      "loss": 1.4881,
+      "step": 169925
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.599034742461051e-06,
+      "loss": 1.1683,
+      "step": 169930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5986248571594854e-06,
+      "loss": 1.5294,
+      "step": 169935
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5982150198342937e-06,
+      "loss": 1.3195,
+      "step": 169940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.59780523048781e-06,
+      "loss": 1.1078,
+      "step": 169945
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5973954891223808e-06,
+      "loss": 1.3246,
+      "step": 169950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.59698579574034e-06,
+      "loss": 1.3197,
+      "step": 169955
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5965761503440292e-06,
+      "loss": 1.2589,
+      "step": 169960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5961665529357917e-06,
+      "loss": 1.555,
+      "step": 169965
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5957570035179626e-06,
+      "loss": 1.0977,
+      "step": 169970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5953475020928788e-06,
+      "loss": 1.2787,
+      "step": 169975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5949380486628818e-06,
+      "loss": 1.0374,
+      "step": 169980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.594528643230312e-06,
+      "loss": 1.0446,
+      "step": 169985
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5941192857975007e-06,
+      "loss": 1.0364,
+      "step": 169990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5937099763667929e-06,
+      "loss": 1.0778,
+      "step": 169995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5933007149405222e-06,
+      "loss": 1.2817,
+      "step": 170000
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 1.2211291790008545,
+      "eval_runtime": 6463.0234,
+      "eval_samples_per_second": 3.576,
+      "eval_steps_per_second": 1.788,
+      "step": 170000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5928915015210212e-06,
+      "loss": 1.0794,
+      "step": 170005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5924823361106368e-06,
+      "loss": 1.3021,
+      "step": 170010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5920732187116993e-06,
+      "loss": 1.4418,
+      "step": 170015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.591664149326546e-06,
+      "loss": 1.285,
+      "step": 170020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5912551279575117e-06,
+      "loss": 1.1177,
+      "step": 170025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.590846154606932e-06,
+      "loss": 1.1346,
+      "step": 170030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5904372292771463e-06,
+      "loss": 1.2488,
+      "step": 170035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.590028351970485e-06,
+      "loss": 1.3129,
+      "step": 170040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5896195226892873e-06,
+      "loss": 1.3351,
+      "step": 170045
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5892107414358836e-06,
+      "loss": 1.2241,
+      "step": 170050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.58880200821261e-06,
+      "loss": 1.121,
+      "step": 170055
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.588393323021804e-06,
+      "loss": 1.2051,
+      "step": 170060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5879846858657955e-06,
+      "loss": 1.5462,
+      "step": 170065
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5875760967469168e-06,
+      "loss": 1.3776,
+      "step": 170070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5871675556675026e-06,
+      "loss": 1.2298,
+      "step": 170075
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.58675906262989e-06,
+      "loss": 1.2475,
+      "step": 170080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5863506176364075e-06,
+      "loss": 1.386,
+      "step": 170085
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.585942220689386e-06,
+      "loss": 1.4094,
+      "step": 170090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5855338717911627e-06,
+      "loss": 1.5878,
+      "step": 170095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5851255709440638e-06,
+      "loss": 1.2019,
+      "step": 170100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.584717318150424e-06,
+      "loss": 0.9845,
+      "step": 170105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5843091134125776e-06,
+      "loss": 1.1602,
+      "step": 170110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5839009567328523e-06,
+      "loss": 0.9605,
+      "step": 170115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5834928481135757e-06,
+      "loss": 1.5585,
+      "step": 170120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.583084787557083e-06,
+      "loss": 1.2465,
+      "step": 170125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5826767750657046e-06,
+      "loss": 1.0725,
+      "step": 170130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5822688106417672e-06,
+      "loss": 1.7898,
+      "step": 170135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5818608942876046e-06,
+      "loss": 1.0366,
+      "step": 170140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5814530260055426e-06,
+      "loss": 1.2333,
+      "step": 170145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.581045205797911e-06,
+      "loss": 1.2472,
+      "step": 170150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5806374336670415e-06,
+      "loss": 1.2336,
+      "step": 170155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5802297096152608e-06,
+      "loss": 0.996,
+      "step": 170160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5798220336448943e-06,
+      "loss": 1.5146,
+      "step": 170165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.579414405758274e-06,
+      "loss": 1.0798,
+      "step": 170170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5790068259577274e-06,
+      "loss": 1.0355,
+      "step": 170175
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.578599294245582e-06,
+      "loss": 1.3203,
+      "step": 170180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5781918106241622e-06,
+      "loss": 1.1223,
+      "step": 170185
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5777843750957955e-06,
+      "loss": 1.11,
+      "step": 170190
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.577376987662813e-06,
+      "loss": 0.8627,
+      "step": 170195
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5769696483275355e-06,
+      "loss": 1.5207,
+      "step": 170200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.576562357092295e-06,
+      "loss": 1.3016,
+      "step": 170205
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5761551139594122e-06,
+      "loss": 0.9055,
+      "step": 170210
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.575747918931213e-06,
+      "loss": 1.2993,
+      "step": 170215
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5753407720100245e-06,
+      "loss": 1.1612,
+      "step": 170220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5749336731981735e-06,
+      "loss": 1.5063,
+      "step": 170225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5745266224979827e-06,
+      "loss": 1.6379,
+      "step": 170230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5741196199117747e-06,
+      "loss": 1.2455,
+      "step": 170235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5737126654418744e-06,
+      "loss": 1.244,
+      "step": 170240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.573305759090611e-06,
+      "loss": 1.291,
+      "step": 170245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5728989008603e-06,
+      "loss": 1.3037,
+      "step": 170250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5724920907532716e-06,
+      "loss": 1.353,
+      "step": 170255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.572085328771844e-06,
+      "loss": 1.1969,
+      "step": 170260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5716786149183415e-06,
+      "loss": 1.0098,
+      "step": 170265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5712719491950913e-06,
+      "loss": 1.4101,
+      "step": 170270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5708653316044099e-06,
+      "loss": 1.159,
+      "step": 170275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5704587621486189e-06,
+      "loss": 1.6658,
+      "step": 170280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5700522408300434e-06,
+      "loss": 1.2663,
+      "step": 170285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5696457676510046e-06,
+      "loss": 1.239,
+      "step": 170290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5692393426138218e-06,
+      "loss": 1.0574,
+      "step": 170295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5688329657208178e-06,
+      "loss": 0.9374,
+      "step": 170300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.568426636974314e-06,
+      "loss": 1.0737,
+      "step": 170305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.568020356376625e-06,
+      "loss": 1.2015,
+      "step": 170310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5676141239300758e-06,
+      "loss": 1.0892,
+      "step": 170315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5672079396369877e-06,
+      "loss": 0.9862,
+      "step": 170320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.566801803499678e-06,
+      "loss": 0.9396,
+      "step": 170325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5663957155204634e-06,
+      "loss": 1.0603,
+      "step": 170330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5659896757016646e-06,
+      "loss": 1.3099,
+      "step": 170335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.565583684045604e-06,
+      "loss": 1.0492,
+      "step": 170340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5651777405545942e-06,
+      "loss": 1.1632,
+      "step": 170345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.564771845230959e-06,
+      "loss": 1.1408,
+      "step": 170350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5643659980770109e-06,
+      "loss": 1.3016,
+      "step": 170355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5639601990950703e-06,
+      "loss": 1.3075,
+      "step": 170360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5635544482874566e-06,
+      "loss": 1.3997,
+      "step": 170365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5631487456564842e-06,
+      "loss": 1.2968,
+      "step": 170370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5627430912044706e-06,
+      "loss": 1.5813,
+      "step": 170375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5623374849337269e-06,
+      "loss": 1.1794,
+      "step": 170380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5619319268465794e-06,
+      "loss": 1.2201,
+      "step": 170385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5615264169453393e-06,
+      "loss": 1.2368,
+      "step": 170390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5611209552323192e-06,
+      "loss": 1.2541,
+      "step": 170395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5607155417098395e-06,
+      "loss": 1.2396,
+      "step": 170400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5603101763802108e-06,
+      "loss": 1.1943,
+      "step": 170405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.559904859245751e-06,
+      "loss": 1.1469,
+      "step": 170410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5594995903087762e-06,
+      "loss": 1.0341,
+      "step": 170415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5590943695715976e-06,
+      "loss": 1.1956,
+      "step": 170420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5586891970365282e-06,
+      "loss": 1.2455,
+      "step": 170425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5582840727058846e-06,
+      "loss": 1.4155,
+      "step": 170430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5578789965819807e-06,
+      "loss": 1.2367,
+      "step": 170435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5574739686671258e-06,
+      "loss": 1.3438,
+      "step": 170440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.557068988963638e-06,
+      "loss": 0.9441,
+      "step": 170445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5566640574738256e-06,
+      "loss": 1.3612,
+      "step": 170450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5562591742000043e-06,
+      "loss": 1.1782,
+      "step": 170455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5558543391444825e-06,
+      "loss": 0.9394,
+      "step": 170460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5554495523095759e-06,
+      "loss": 1.3772,
+      "step": 170465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5550448136975926e-06,
+      "loss": 0.9795,
+      "step": 170470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5546401233108455e-06,
+      "loss": 1.0901,
+      "step": 170475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.554235481151648e-06,
+      "loss": 1.7519,
+      "step": 170480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5538308872223084e-06,
+      "loss": 1.0097,
+      "step": 170485
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.553426341525135e-06,
+      "loss": 1.1091,
+      "step": 170490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5530218440624422e-06,
+      "loss": 1.0691,
+      "step": 170495
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5526173948365365e-06,
+      "loss": 1.3579,
+      "step": 170500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5522129938497277e-06,
+      "loss": 1.2914,
+      "step": 170505
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.55180864110433e-06,
+      "loss": 1.2877,
+      "step": 170510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5514043366026476e-06,
+      "loss": 0.974,
+      "step": 170515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5510000803469882e-06,
+      "loss": 1.7119,
+      "step": 170520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.550595872339663e-06,
+      "loss": 1.4118,
+      "step": 170525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5501917125829825e-06,
+      "loss": 1.0721,
+      "step": 170530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5497876010792522e-06,
+      "loss": 1.1597,
+      "step": 170535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5493835378307765e-06,
+      "loss": 1.1543,
+      "step": 170540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5489795228398674e-06,
+      "loss": 1.4647,
+      "step": 170545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5485755561088312e-06,
+      "loss": 1.2466,
+      "step": 170550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5481716376399736e-06,
+      "loss": 1.3969,
+      "step": 170555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5477677674356029e-06,
+      "loss": 1.2826,
+      "step": 170560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5473639454980227e-06,
+      "loss": 1.2538,
+      "step": 170565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5469601718295402e-06,
+      "loss": 1.187,
+      "step": 170570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5465564464324634e-06,
+      "loss": 1.377,
+      "step": 170575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5461527693090971e-06,
+      "loss": 1.2729,
+      "step": 170580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.545749140461742e-06,
+      "loss": 1.2206,
+      "step": 170585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.545345559892707e-06,
+      "loss": 1.3439,
+      "step": 170590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5449420276042993e-06,
+      "loss": 1.4311,
+      "step": 170595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5445385435988192e-06,
+      "loss": 1.2727,
+      "step": 170600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5441351078785694e-06,
+      "loss": 1.2627,
+      "step": 170605
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.543731720445859e-06,
+      "loss": 1.123,
+      "step": 170610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5433283813029865e-06,
+      "loss": 1.0773,
+      "step": 170615
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5429250904522563e-06,
+      "loss": 1.2212,
+      "step": 170620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5425218478959758e-06,
+      "loss": 1.2457,
+      "step": 170625
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.542118653636444e-06,
+      "loss": 1.3677,
+      "step": 170630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5417155076759615e-06,
+      "loss": 1.0534,
+      "step": 170635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5413124100168332e-06,
+      "loss": 1.1693,
+      "step": 170640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5409093606613623e-06,
+      "loss": 1.233,
+      "step": 170645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5405063596118475e-06,
+      "loss": 1.5022,
+      "step": 170650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5401034068705923e-06,
+      "loss": 1.3024,
+      "step": 170655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5397005024398947e-06,
+      "loss": 1.1615,
+      "step": 170660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.539297646322061e-06,
+      "loss": 1.345,
+      "step": 170665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5388948385193846e-06,
+      "loss": 1.3655,
+      "step": 170670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5384920790341728e-06,
+      "loss": 1.1692,
+      "step": 170675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5380893678687225e-06,
+      "loss": 1.1047,
+      "step": 170680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5376867050253274e-06,
+      "loss": 1.0702,
+      "step": 170685
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.537284090506298e-06,
+      "loss": 1.3534,
+      "step": 170690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.536881524313928e-06,
+      "loss": 1.0834,
+      "step": 170695
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5364790064505143e-06,
+      "loss": 1.2771,
+      "step": 170700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5360765369183605e-06,
+      "loss": 1.1735,
+      "step": 170705
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5356741157197586e-06,
+      "loss": 1.0649,
+      "step": 170710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5352717428570107e-06,
+      "loss": 1.0172,
+      "step": 170715
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5348694183324153e-06,
+      "loss": 1.0887,
+      "step": 170720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5344671421482692e-06,
+      "loss": 1.1457,
+      "step": 170725
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5340649143068665e-06,
+      "loss": 1.455,
+      "step": 170730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5336627348105071e-06,
+      "loss": 1.13,
+      "step": 170735
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5332606036614895e-06,
+      "loss": 1.5731,
+      "step": 170740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5328585208621072e-06,
+      "loss": 1.1365,
+      "step": 170745
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.532456486414654e-06,
+      "loss": 1.2662,
+      "step": 170750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5320545003214293e-06,
+      "loss": 1.2523,
+      "step": 170755
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.53165256258473e-06,
+      "loss": 1.0813,
+      "step": 170760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5312506732068467e-06,
+      "loss": 1.3665,
+      "step": 170765
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5308488321900794e-06,
+      "loss": 1.0316,
+      "step": 170770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5304470395367189e-06,
+      "loss": 1.1768,
+      "step": 170775
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5300452952490597e-06,
+      "loss": 1.3758,
+      "step": 170780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5296435993294012e-06,
+      "loss": 1.413,
+      "step": 170785
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5292419517800328e-06,
+      "loss": 1.216,
+      "step": 170790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5288403526032468e-06,
+      "loss": 1.2739,
+      "step": 170795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5284388018013407e-06,
+      "loss": 1.1567,
+      "step": 170800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5280372993766035e-06,
+      "loss": 1.228,
+      "step": 170805
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5276358453313323e-06,
+      "loss": 0.9437,
+      "step": 170810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.527234439667814e-06,
+      "loss": 1.1324,
+      "step": 170815
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5268330823883471e-06,
+      "loss": 1.3144,
+      "step": 170820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5264317734952183e-06,
+      "loss": 1.1202,
+      "step": 170825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5260305129907228e-06,
+      "loss": 1.3574,
+      "step": 170830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5256293008771516e-06,
+      "loss": 1.2374,
+      "step": 170835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5252281371567957e-06,
+      "loss": 1.6767,
+      "step": 170840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5248270218319427e-06,
+      "loss": 1.3121,
+      "step": 170845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5244259549048856e-06,
+      "loss": 1.3859,
+      "step": 170850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5240249363779181e-06,
+      "loss": 0.9311,
+      "step": 170855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5236239662533236e-06,
+      "loss": 1.6039,
+      "step": 170860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5232230445333985e-06,
+      "loss": 1.1124,
+      "step": 170865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.522822171220426e-06,
+      "loss": 1.0846,
+      "step": 170870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5224213463166992e-06,
+      "loss": 1.1038,
+      "step": 170875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5220205698245084e-06,
+      "loss": 1.1644,
+      "step": 170880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.521619841746139e-06,
+      "loss": 1.227,
+      "step": 170885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5212191620838812e-06,
+      "loss": 1.119,
+      "step": 170890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.520818530840018e-06,
+      "loss": 1.1482,
+      "step": 170895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5204179480168458e-06,
+      "loss": 1.3993,
+      "step": 170900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5200174136166478e-06,
+      "loss": 1.0621,
+      "step": 170905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5196169276417093e-06,
+      "loss": 1.1713,
+      "step": 170910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5192164900943219e-06,
+      "loss": 0.9366,
+      "step": 170915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5188161009767666e-06,
+      "loss": 1.211,
+      "step": 170920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5184157602913329e-06,
+      "loss": 1.3297,
+      "step": 170925
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5180154680403102e-06,
+      "loss": 1.2161,
+      "step": 170930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5176152242259812e-06,
+      "loss": 1.107,
+      "step": 170935
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5172150288506294e-06,
+      "loss": 1.0603,
+      "step": 170940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5168148819165419e-06,
+      "loss": 1.3529,
+      "step": 170945
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5164147834260056e-06,
+      "loss": 1.311,
+      "step": 170950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5160147333813047e-06,
+      "loss": 1.7339,
+      "step": 170955
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5156147317847202e-06,
+      "loss": 1.0488,
+      "step": 170960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5152147786385395e-06,
+      "loss": 1.4267,
+      "step": 170965
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5148148739450475e-06,
+      "loss": 1.2483,
+      "step": 170970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5144150177065242e-06,
+      "loss": 1.9699,
+      "step": 170975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5140152099252582e-06,
+      "loss": 1.3277,
+      "step": 170980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5136154506035283e-06,
+      "loss": 1.1012,
+      "step": 170985
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5132157397436153e-06,
+      "loss": 1.1506,
+      "step": 170990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5128160773478084e-06,
+      "loss": 1.0347,
+      "step": 170995
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.512416463418388e-06,
+      "loss": 1.2682,
+      "step": 171000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.512016897957631e-06,
+      "loss": 1.5041,
+      "step": 171005
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5116173809678258e-06,
+      "loss": 1.3413,
+      "step": 171010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5112179124512483e-06,
+      "loss": 1.2757,
+      "step": 171015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.510818492410182e-06,
+      "loss": 1.1513,
+      "step": 171020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.510419120846911e-06,
+      "loss": 1.2199,
+      "step": 171025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5100197977637131e-06,
+      "loss": 1.3295,
+      "step": 171030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5096205231628657e-06,
+      "loss": 1.4766,
+      "step": 171035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5092212970466513e-06,
+      "loss": 1.3015,
+      "step": 171040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5088221194173526e-06,
+      "loss": 1.0065,
+      "step": 171045
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5084229902772474e-06,
+      "loss": 2.0508,
+      "step": 171050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.508023909628611e-06,
+      "loss": 1.1185,
+      "step": 171055
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5076248774737257e-06,
+      "loss": 1.0404,
+      "step": 171060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5072258938148721e-06,
+      "loss": 1.2728,
+      "step": 171065
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.506826958654325e-06,
+      "loss": 1.2881,
+      "step": 171070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5064280719943658e-06,
+      "loss": 0.9578,
+      "step": 171075
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5060292338372706e-06,
+      "loss": 1.2467,
+      "step": 171080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5056304441853121e-06,
+      "loss": 1.1121,
+      "step": 171085
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5052317030407771e-06,
+      "loss": 1.4928,
+      "step": 171090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5048330104059383e-06,
+      "loss": 1.1766,
+      "step": 171095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5044343662830717e-06,
+      "loss": 1.0979,
+      "step": 171100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5040357706744524e-06,
+      "loss": 1.0667,
+      "step": 171105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5036372235823592e-06,
+      "loss": 1.2354,
+      "step": 171110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5032387250090686e-06,
+      "loss": 1.3398,
+      "step": 171115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5028402749568527e-06,
+      "loss": 1.0294,
+      "step": 171120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.502441873427991e-06,
+      "loss": 1.0289,
+      "step": 171125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.502043520424754e-06,
+      "loss": 1.2647,
+      "step": 171130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.50164521594942e-06,
+      "loss": 1.3016,
+      "step": 171135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5012469600042657e-06,
+      "loss": 1.4858,
+      "step": 171140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5008487525915605e-06,
+      "loss": 1.0904,
+      "step": 171145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5004505937135794e-06,
+      "loss": 1.3211,
+      "step": 171150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.500052483372596e-06,
+      "loss": 1.1575,
+      "step": 171155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4996544215708875e-06,
+      "loss": 1.3126,
+      "step": 171160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.499256408310722e-06,
+      "loss": 1.1305,
+      "step": 171165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4988584435943764e-06,
+      "loss": 1.2337,
+      "step": 171170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4984605274241193e-06,
+      "loss": 1.3102,
+      "step": 171175
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4980626598022275e-06,
+      "loss": 1.1277,
+      "step": 171180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.497664840730968e-06,
+      "loss": 1.0601,
+      "step": 171185
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.497267070212618e-06,
+      "loss": 1.3504,
+      "step": 171190
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4968693482494457e-06,
+      "loss": 1.1813,
+      "step": 171195
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4964716748437214e-06,
+      "loss": 1.2762,
+      "step": 171200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4960740499977167e-06,
+      "loss": 1.4815,
+      "step": 171205
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4956764737137053e-06,
+      "loss": 1.3167,
+      "step": 171210
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4952789459939532e-06,
+      "loss": 1.0357,
+      "step": 171215
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4948814668407352e-06,
+      "loss": 1.1371,
+      "step": 171220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4944840362563161e-06,
+      "loss": 1.1586,
+      "step": 171225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4940866542429677e-06,
+      "loss": 1.2134,
+      "step": 171230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4936893208029625e-06,
+      "loss": 1.2375,
+      "step": 171235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.493292035938566e-06,
+      "loss": 1.237,
+      "step": 171240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.492894799652045e-06,
+      "loss": 1.2327,
+      "step": 171245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4924976119456691e-06,
+      "loss": 1.4896,
+      "step": 171250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4921004728217114e-06,
+      "loss": 1.273,
+      "step": 171255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4917033822824355e-06,
+      "loss": 1.2051,
+      "step": 171260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4913063403301075e-06,
+      "loss": 1.4025,
+      "step": 171265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4909093469669965e-06,
+      "loss": 1.3095,
+      "step": 171270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.490512402195372e-06,
+      "loss": 1.0298,
+      "step": 171275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4901155060174955e-06,
+      "loss": 1.117,
+      "step": 171280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4897186584356392e-06,
+      "loss": 1.1982,
+      "step": 171285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4893218594520675e-06,
+      "loss": 1.2577,
+      "step": 171290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4889251090690404e-06,
+      "loss": 1.2742,
+      "step": 171295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4885284072888328e-06,
+      "loss": 1.2309,
+      "step": 171300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4881317541137063e-06,
+      "loss": 1.3416,
+      "step": 171305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4877351495459236e-06,
+      "loss": 1.4783,
+      "step": 171310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4873385935877539e-06,
+      "loss": 1.3541,
+      "step": 171315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4869420862414564e-06,
+      "loss": 1.3219,
+      "step": 171320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4865456275093005e-06,
+      "loss": 1.1593,
+      "step": 171325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4861492173935465e-06,
+      "loss": 1.3491,
+      "step": 171330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4857528558964618e-06,
+      "loss": 1.3507,
+      "step": 171335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4853565430203055e-06,
+      "loss": 1.1704,
+      "step": 171340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4849602787673423e-06,
+      "loss": 1.4763,
+      "step": 171345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4845640631398383e-06,
+      "loss": 1.1573,
+      "step": 171350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.484167896140054e-06,
+      "loss": 1.2278,
+      "step": 171355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4837717777702487e-06,
+      "loss": 1.6312,
+      "step": 171360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4833757080326872e-06,
+      "loss": 1.6163,
+      "step": 171365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.482979686929633e-06,
+      "loss": 1.2252,
+      "step": 171370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4825837144633438e-06,
+      "loss": 1.301,
+      "step": 171375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4821877906360848e-06,
+      "loss": 1.3711,
+      "step": 171380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4817919154501147e-06,
+      "loss": 1.3898,
+      "step": 171385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4813960889076895e-06,
+      "loss": 1.5672,
+      "step": 171390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4810003110110792e-06,
+      "loss": 1.0707,
+      "step": 171395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4806045817625391e-06,
+      "loss": 1.3006,
+      "step": 171400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4802089011643295e-06,
+      "loss": 1.2813,
+      "step": 171405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4798132692187073e-06,
+      "loss": 1.265,
+      "step": 171410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.479417685927933e-06,
+      "loss": 1.1957,
+      "step": 171415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4790221512942693e-06,
+      "loss": 1.1637,
+      "step": 171420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4786266653199698e-06,
+      "loss": 1.2942,
+      "step": 171425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4782312280072975e-06,
+      "loss": 1.2809,
+      "step": 171430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4778358393585058e-06,
+      "loss": 1.3243,
+      "step": 171435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.477440499375855e-06,
+      "loss": 1.2085,
+      "step": 171440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.477045208061606e-06,
+      "loss": 1.3232,
+      "step": 171445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4766499654180123e-06,
+      "loss": 0.8244,
+      "step": 171450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4762547714473296e-06,
+      "loss": 1.3666,
+      "step": 171455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4758596261518165e-06,
+      "loss": 1.1829,
+      "step": 171460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4754645295337323e-06,
+      "loss": 1.2634,
+      "step": 171465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4750694815953303e-06,
+      "loss": 1.3644,
+      "step": 171470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4746744823388649e-06,
+      "loss": 1.2442,
+      "step": 171475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4742795317665925e-06,
+      "loss": 1.1424,
+      "step": 171480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.473884629880773e-06,
+      "loss": 1.5183,
+      "step": 171485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4734897766836552e-06,
+      "loss": 1.389,
+      "step": 171490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.473094972177499e-06,
+      "loss": 1.4595,
+      "step": 171495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4727002163645565e-06,
+      "loss": 1.0905,
+      "step": 171500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4723055092470806e-06,
+      "loss": 1.0501,
+      "step": 171505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4719108508273261e-06,
+      "loss": 1.4768,
+      "step": 171510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.47151624110755e-06,
+      "loss": 1.1825,
+      "step": 171515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.471121680090002e-06,
+      "loss": 1.1976,
+      "step": 171520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4707271677769375e-06,
+      "loss": 1.3253,
+      "step": 171525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4703327041706062e-06,
+      "loss": 1.1643,
+      "step": 171530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4699382892732628e-06,
+      "loss": 1.1038,
+      "step": 171535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4695439230871623e-06,
+      "loss": 1.1093,
+      "step": 171540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.469149605614555e-06,
+      "loss": 1.088,
+      "step": 171545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.468755336857688e-06,
+      "loss": 1.1031,
+      "step": 171550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4683611168188172e-06,
+      "loss": 1.0698,
+      "step": 171555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4679669455001954e-06,
+      "loss": 1.2205,
+      "step": 171560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.467572822904072e-06,
+      "loss": 0.8593,
+      "step": 171565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4671787490326938e-06,
+      "loss": 1.0288,
+      "step": 171570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4667847238883148e-06,
+      "loss": 0.9962,
+      "step": 171575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4663907474731865e-06,
+      "loss": 1.1497,
+      "step": 171580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4659968197895546e-06,
+      "loss": 1.2156,
+      "step": 171585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4656029408396731e-06,
+      "loss": 1.228,
+      "step": 171590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4652091106257894e-06,
+      "loss": 1.2671,
+      "step": 171595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4648153291501499e-06,
+      "loss": 1.3031,
+      "step": 171600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4644215964150054e-06,
+      "loss": 1.4463,
+      "step": 171605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4640279124226065e-06,
+      "loss": 1.3246,
+      "step": 171610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4636342771751999e-06,
+      "loss": 1.359,
+      "step": 171615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4632406906750307e-06,
+      "loss": 1.0854,
+      "step": 171620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4628471529243481e-06,
+      "loss": 0.8564,
+      "step": 171625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4624536639254016e-06,
+      "loss": 1.3072,
+      "step": 171630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.462060223680436e-06,
+      "loss": 1.4452,
+      "step": 171635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4616668321916993e-06,
+      "loss": 1.2802,
+      "step": 171640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4612734894614366e-06,
+      "loss": 1.2584,
+      "step": 171645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.460880195491894e-06,
+      "loss": 1.3086,
+      "step": 171650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4604869502853202e-06,
+      "loss": 1.3593,
+      "step": 171655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4600937538439596e-06,
+      "loss": 1.1229,
+      "step": 171660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4597006061700546e-06,
+      "loss": 1.1664,
+      "step": 171665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4593075072658525e-06,
+      "loss": 0.9472,
+      "step": 171670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4589144571336001e-06,
+      "loss": 1.5814,
+      "step": 171675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4585214557755378e-06,
+      "loss": 1.2726,
+      "step": 171680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.458128503193914e-06,
+      "loss": 1.2927,
+      "step": 171685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4577355993909715e-06,
+      "loss": 1.1545,
+      "step": 171690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4573427443689502e-06,
+      "loss": 1.2171,
+      "step": 171695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4569499381300967e-06,
+      "loss": 0.957,
+      "step": 171700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4565571806766566e-06,
+      "loss": 1.2731,
+      "step": 171705
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4561644720108704e-06,
+      "loss": 1.1704,
+      "step": 171710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4557718121349774e-06,
+      "loss": 1.1344,
+      "step": 171715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4553792010512225e-06,
+      "loss": 1.1058,
+      "step": 171720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4549866387618516e-06,
+      "loss": 1.4716,
+      "step": 171725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4545941252690998e-06,
+      "loss": 0.9994,
+      "step": 171730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.454201660575214e-06,
+      "loss": 1.6584,
+      "step": 171735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4538092446824314e-06,
+      "loss": 1.4578,
+      "step": 171740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4534168775929936e-06,
+      "loss": 1.1382,
+      "step": 171745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4530245593091452e-06,
+      "loss": 1.2446,
+      "step": 171750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4526322898331236e-06,
+      "loss": 1.3514,
+      "step": 171755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4522400691671668e-06,
+      "loss": 1.2091,
+      "step": 171760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4518478973135163e-06,
+      "loss": 1.1798,
+      "step": 171765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.451455774274414e-06,
+      "loss": 1.0357,
+      "step": 171770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4510637000520978e-06,
+      "loss": 1.3613,
+      "step": 171775
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4506716746488036e-06,
+      "loss": 1.4786,
+      "step": 171780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4502796980667743e-06,
+      "loss": 1.6678,
+      "step": 171785
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4498877703082437e-06,
+      "loss": 1.0796,
+      "step": 171790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4494958913754532e-06,
+      "loss": 0.9816,
+      "step": 171795
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4491040612706419e-06,
+      "loss": 1.2252,
+      "step": 171800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4487122799960463e-06,
+      "loss": 1.3838,
+      "step": 171805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4483205475538997e-06,
+      "loss": 1.4135,
+      "step": 171810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4479288639464428e-06,
+      "loss": 1.3158,
+      "step": 171815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4475372291759139e-06,
+      "loss": 1.0129,
+      "step": 171820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4471456432445451e-06,
+      "loss": 1.326,
+      "step": 171825
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4467541061545776e-06,
+      "loss": 1.1397,
+      "step": 171830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4463626179082424e-06,
+      "loss": 1.2159,
+      "step": 171835
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.445971178507779e-06,
+      "loss": 1.0817,
+      "step": 171840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4455797879554189e-06,
+      "loss": 1.2153,
+      "step": 171845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4451884462534015e-06,
+      "loss": 1.2181,
+      "step": 171850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4447971534039562e-06,
+      "loss": 0.9947,
+      "step": 171855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4444059094093221e-06,
+      "loss": 1.2569,
+      "step": 171860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4440147142717331e-06,
+      "loss": 0.9326,
+      "step": 171865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.443623567993422e-06,
+      "loss": 1.1477,
+      "step": 171870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.443232470576621e-06,
+      "loss": 1.1464,
+      "step": 171875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4428414220235643e-06,
+      "loss": 1.1439,
+      "step": 171880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4424504223364878e-06,
+      "loss": 1.3479,
+      "step": 171885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4420594715176195e-06,
+      "loss": 1.3811,
+      "step": 171890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4416685695691967e-06,
+      "loss": 1.3412,
+      "step": 171895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.44127771649345e-06,
+      "loss": 1.1903,
+      "step": 171900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4408869122926084e-06,
+      "loss": 1.1459,
+      "step": 171905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4404961569689058e-06,
+      "loss": 1.3675,
+      "step": 171910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.440105450524577e-06,
+      "loss": 1.2484,
+      "step": 171915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4397147929618494e-06,
+      "loss": 1.3148,
+      "step": 171920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.439324184282952e-06,
+      "loss": 1.3003,
+      "step": 171925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4389336244901186e-06,
+      "loss": 1.3861,
+      "step": 171930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.438543113585581e-06,
+      "loss": 1.2783,
+      "step": 171935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4381526515715638e-06,
+      "loss": 1.2152,
+      "step": 171940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.437762238450303e-06,
+      "loss": 1.7079,
+      "step": 171945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4373718742240229e-06,
+      "loss": 1.3079,
+      "step": 171950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4369815588949542e-06,
+      "loss": 1.0626,
+      "step": 171955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.436591292465329e-06,
+      "loss": 1.3758,
+      "step": 171960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4362010749373745e-06,
+      "loss": 1.4467,
+      "step": 171965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4358109063133141e-06,
+      "loss": 0.9574,
+      "step": 171970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4354207865953806e-06,
+      "loss": 1.1586,
+      "step": 171975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4350307157858024e-06,
+      "loss": 1.1776,
+      "step": 171980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4346406938868063e-06,
+      "loss": 1.0842,
+      "step": 171985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4342507209006151e-06,
+      "loss": 1.3243,
+      "step": 171990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4338607968294628e-06,
+      "loss": 1.3725,
+      "step": 171995
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4334709216755704e-06,
+      "loss": 1.1817,
+      "step": 172000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4330810954411668e-06,
+      "loss": 1.2125,
+      "step": 172005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4326913181284796e-06,
+      "loss": 1.5643,
+      "step": 172010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.432301589739733e-06,
+      "loss": 1.4507,
+      "step": 172015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4319119102771494e-06,
+      "loss": 1.242,
+      "step": 172020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4315222797429584e-06,
+      "loss": 1.1514,
+      "step": 172025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4311326981393858e-06,
+      "loss": 1.2072,
+      "step": 172030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.430743165468652e-06,
+      "loss": 1.1653,
+      "step": 172035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4303536817329856e-06,
+      "loss": 1.0223,
+      "step": 172040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4299642469346064e-06,
+      "loss": 1.4147,
+      "step": 172045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4295748610757442e-06,
+      "loss": 1.3392,
+      "step": 172050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.429185524158616e-06,
+      "loss": 1.389,
+      "step": 172055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4287962361854501e-06,
+      "loss": 1.1535,
+      "step": 172060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.428406997158469e-06,
+      "loss": 1.5069,
+      "step": 172065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.428017807079889e-06,
+      "loss": 1.5228,
+      "step": 172070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4276286659519422e-06,
+      "loss": 1.323,
+      "step": 172075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4272395737768463e-06,
+      "loss": 1.2595,
+      "step": 172080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4268505305568214e-06,
+      "loss": 1.1632,
+      "step": 172085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4264615362940937e-06,
+      "loss": 1.0438,
+      "step": 172090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.426072590990879e-06,
+      "loss": 1.2872,
+      "step": 172095
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4256836946494012e-06,
+      "loss": 1.7062,
+      "step": 172100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4252948472718842e-06,
+      "loss": 1.3383,
+      "step": 172105
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4249060488605448e-06,
+      "loss": 1.4616,
+      "step": 172110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4245172994176015e-06,
+      "loss": 1.3786,
+      "step": 172115
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.424128598945278e-06,
+      "loss": 1.474,
+      "step": 172120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4237399474457936e-06,
+      "loss": 1.3028,
+      "step": 172125
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4233513449213676e-06,
+      "loss": 1.349,
+      "step": 172130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4229627913742162e-06,
+      "loss": 1.1254,
+      "step": 172135
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4225742868065607e-06,
+      "loss": 1.2592,
+      "step": 172140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4221858312206217e-06,
+      "loss": 1.2033,
+      "step": 172145
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4217974246186117e-06,
+      "loss": 1.6904,
+      "step": 172150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4214090670027558e-06,
+      "loss": 1.0535,
+      "step": 172155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4210207583752656e-06,
+      "loss": 1.245,
+      "step": 172160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4206324987383603e-06,
+      "loss": 1.2206,
+      "step": 172165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4202442880942613e-06,
+      "loss": 1.2056,
+      "step": 172170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4198561264451815e-06,
+      "loss": 1.0877,
+      "step": 172175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.419468013793336e-06,
+      "loss": 1.1322,
+      "step": 172180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4190799501409446e-06,
+      "loss": 1.4334,
+      "step": 172185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4186919354902207e-06,
+      "loss": 1.1281,
+      "step": 172190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4183039698433832e-06,
+      "loss": 1.1061,
+      "step": 172195
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4179160532026426e-06,
+      "loss": 1.3851,
+      "step": 172200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4175281855702205e-06,
+      "loss": 1.2655,
+      "step": 172205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4171403669483263e-06,
+      "loss": 1.074,
+      "step": 172210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4167525973391772e-06,
+      "loss": 1.0809,
+      "step": 172215
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.416364876744989e-06,
+      "loss": 1.0905,
+      "step": 172220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4159772051679743e-06,
+      "loss": 1.1295,
+      "step": 172225
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.415589582610344e-06,
+      "loss": 1.1349,
+      "step": 172230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4152020090743146e-06,
+      "loss": 1.0322,
+      "step": 172235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4148144845621015e-06,
+      "loss": 1.2159,
+      "step": 172240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4144270090759128e-06,
+      "loss": 1.1081,
+      "step": 172245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4140395826179664e-06,
+      "loss": 1.7471,
+      "step": 172250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4136522051904688e-06,
+      "loss": 1.0172,
+      "step": 172255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4132648767956358e-06,
+      "loss": 1.1101,
+      "step": 172260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4128775974356811e-06,
+      "loss": 1.2847,
+      "step": 172265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4124903671128143e-06,
+      "loss": 1.2824,
+      "step": 172270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4121031858292455e-06,
+      "loss": 1.1373,
+      "step": 172275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4117160535871822e-06,
+      "loss": 0.9058,
+      "step": 172280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4113289703888444e-06,
+      "loss": 1.1093,
+      "step": 172285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4109419362364374e-06,
+      "loss": 1.3532,
+      "step": 172290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4105549511321692e-06,
+      "loss": 1.1564,
+      "step": 172295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4101680150782548e-06,
+      "loss": 1.2504,
+      "step": 172300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.409781128076898e-06,
+      "loss": 1.2861,
+      "step": 172305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4093942901303114e-06,
+      "loss": 1.1195,
+      "step": 172310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4090075012407068e-06,
+      "loss": 1.1466,
+      "step": 172315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4086207614102898e-06,
+      "loss": 1.1513,
+      "step": 172320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4082340706412667e-06,
+      "loss": 1.1528,
+      "step": 172325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.407847428935848e-06,
+      "loss": 1.4377,
+      "step": 172330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.407460836296245e-06,
+      "loss": 1.2536,
+      "step": 172335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4070742927246616e-06,
+      "loss": 1.2497,
+      "step": 172340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4066877982233029e-06,
+      "loss": 1.0678,
+      "step": 172345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4063013527943793e-06,
+      "loss": 1.2794,
+      "step": 172350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4059149564400986e-06,
+      "loss": 1.2309,
+      "step": 172355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.405528609162665e-06,
+      "loss": 1.2316,
+      "step": 172360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.405142310964287e-06,
+      "loss": 1.2699,
+      "step": 172365
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.404756061847169e-06,
+      "loss": 1.2785,
+      "step": 172370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4043698618135126e-06,
+      "loss": 1.287,
+      "step": 172375
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.403983710865532e-06,
+      "loss": 1.1511,
+      "step": 172380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4035976090054281e-06,
+      "loss": 1.5068,
+      "step": 172385
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4032115562354031e-06,
+      "loss": 1.134,
+      "step": 172390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.402825552557665e-06,
+      "loss": 1.0829,
+      "step": 172395
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4024395979744166e-06,
+      "loss": 1.5049,
+      "step": 172400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4020536924878614e-06,
+      "loss": 1.1516,
+      "step": 172405
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4016678361002055e-06,
+      "loss": 1.3308,
+      "step": 172410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4012820288136519e-06,
+      "loss": 1.1199,
+      "step": 172415
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4008962706304007e-06,
+      "loss": 1.7016,
+      "step": 172420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.400510561552656e-06,
+      "loss": 1.0684,
+      "step": 172425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4001249015826246e-06,
+      "loss": 1.2809,
+      "step": 172430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3997392907225048e-06,
+      "loss": 1.0556,
+      "step": 172435
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3993537289744963e-06,
+      "loss": 1.0852,
+      "step": 172440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3989682163408047e-06,
+      "loss": 0.9387,
+      "step": 172445
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.398582752823633e-06,
+      "loss": 1.1446,
+      "step": 172450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3981973384251778e-06,
+      "loss": 1.0576,
+      "step": 172455
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3978119731476447e-06,
+      "loss": 1.0115,
+      "step": 172460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3974266569932292e-06,
+      "loss": 1.6537,
+      "step": 172465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3970413899641344e-06,
+      "loss": 1.1449,
+      "step": 172470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3966561720625637e-06,
+      "loss": 1.1201,
+      "step": 172475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3962710032907123e-06,
+      "loss": 1.1398,
+      "step": 172480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3958858836507815e-06,
+      "loss": 1.3531,
+      "step": 172485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3955008131449677e-06,
+      "loss": 1.3099,
+      "step": 172490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.395115791775472e-06,
+      "loss": 1.1748,
+      "step": 172495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3947308195444963e-06,
+      "loss": 1.2457,
+      "step": 172500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3943458964542333e-06,
+      "loss": 1.1223,
+      "step": 172505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3939610225068857e-06,
+      "loss": 1.2114,
+      "step": 172510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.393576197704648e-06,
+      "loss": 1.1276,
+      "step": 172515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.393191422049719e-06,
+      "loss": 1.1726,
+      "step": 172520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3928066955442975e-06,
+      "loss": 1.1357,
+      "step": 172525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3924220181905801e-06,
+      "loss": 1.3583,
+      "step": 172530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3920373899907602e-06,
+      "loss": 1.0403,
+      "step": 172535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3916528109470363e-06,
+      "loss": 1.2565,
+      "step": 172540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3912682810616074e-06,
+      "loss": 1.1873,
+      "step": 172545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3908838003366642e-06,
+      "loss": 1.0746,
+      "step": 172550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3904993687744072e-06,
+      "loss": 1.3524,
+      "step": 172555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3901149863770281e-06,
+      "loss": 1.1289,
+      "step": 172560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3897306531467247e-06,
+      "loss": 1.1556,
+      "step": 172565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.389346369085688e-06,
+      "loss": 1.0115,
+      "step": 172570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3889621341961167e-06,
+      "loss": 1.2288,
+      "step": 172575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3885779484802041e-06,
+      "loss": 1.1489,
+      "step": 172580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3881938119401384e-06,
+      "loss": 1.1449,
+      "step": 172585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.387809724578122e-06,
+      "loss": 1.0505,
+      "step": 172590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3874256863963441e-06,
+      "loss": 1.1332,
+      "step": 172595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3870416973969958e-06,
+      "loss": 1.3278,
+      "step": 172600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3866577575822737e-06,
+      "loss": 1.3557,
+      "step": 172605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3862738669543673e-06,
+      "loss": 1.358,
+      "step": 172610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3858900255154694e-06,
+      "loss": 0.9583,
+      "step": 172615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3855062332677748e-06,
+      "loss": 1.1936,
+      "step": 172620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.385122490213473e-06,
+      "loss": 1.2232,
+      "step": 172625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3847387963547533e-06,
+      "loss": 1.1176,
+      "step": 172630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3843551516938082e-06,
+      "loss": 1.3183,
+      "step": 172635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3839715562328315e-06,
+      "loss": 1.0658,
+      "step": 172640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3835880099740119e-06,
+      "loss": 1.4056,
+      "step": 172645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.383204512919536e-06,
+      "loss": 1.1174,
+      "step": 172650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3828210650715978e-06,
+      "loss": 1.3882,
+      "step": 172655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.382437666432388e-06,
+      "loss": 1.423,
+      "step": 172660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3820543170040923e-06,
+      "loss": 1.0274,
+      "step": 172665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3816710167889025e-06,
+      "loss": 1.2088,
+      "step": 172670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3812877657890078e-06,
+      "loss": 1.2657,
+      "step": 172675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.380904564006591e-06,
+      "loss": 1.2722,
+      "step": 172680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3805214114438493e-06,
+      "loss": 1.1567,
+      "step": 172685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3801383081029661e-06,
+      "loss": 1.0694,
+      "step": 172690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3797552539861269e-06,
+      "loss": 1.3163,
+      "step": 172695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.379372249095524e-06,
+      "loss": 1.0975,
+      "step": 172700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3789892934333404e-06,
+      "loss": 1.1663,
+      "step": 172705
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3786063870017673e-06,
+      "loss": 1.5879,
+      "step": 172710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3782235298029857e-06,
+      "loss": 1.3195,
+      "step": 172715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3778407218391877e-06,
+      "loss": 1.2364,
+      "step": 172720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.377457963112554e-06,
+      "loss": 1.2083,
+      "step": 172725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.377075253625273e-06,
+      "loss": 1.2279,
+      "step": 172730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3766925933795328e-06,
+      "loss": 1.6328,
+      "step": 172735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.376309982377516e-06,
+      "loss": 1.2923,
+      "step": 172740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3759274206214045e-06,
+      "loss": 1.0527,
+      "step": 172745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3755449081133864e-06,
+      "loss": 1.4581,
+      "step": 172750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3751624448556478e-06,
+      "loss": 1.0421,
+      "step": 172755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.374780030850369e-06,
+      "loss": 1.1688,
+      "step": 172760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.374397666099736e-06,
+      "loss": 1.8063,
+      "step": 172765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3740153506059327e-06,
+      "loss": 1.1047,
+      "step": 172770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3736330843711387e-06,
+      "loss": 1.4049,
+      "step": 172775
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3732508673975397e-06,
+      "loss": 1.1008,
+      "step": 172780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.372868699687321e-06,
+      "loss": 1.1466,
+      "step": 172785
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3724865812426613e-06,
+      "loss": 1.3735,
+      "step": 172790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3721045120657406e-06,
+      "loss": 1.378,
+      "step": 172795
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3717224921587446e-06,
+      "loss": 1.1405,
+      "step": 172800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3713405215238562e-06,
+      "loss": 1.1925,
+      "step": 172805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3709586001632525e-06,
+      "loss": 1.1186,
+      "step": 172810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3705767280791183e-06,
+      "loss": 1.5465,
+      "step": 172815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3701949052736296e-06,
+      "loss": 1.1716,
+      "step": 172820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3698131317489693e-06,
+      "loss": 1.32,
+      "step": 172825
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3694314075073211e-06,
+      "loss": 1.4218,
+      "step": 172830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3690497325508612e-06,
+      "loss": 1.0258,
+      "step": 172835
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3686681068817664e-06,
+      "loss": 1.2288,
+      "step": 172840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3682865305022197e-06,
+      "loss": 1.2419,
+      "step": 172845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3679050034144014e-06,
+      "loss": 1.2064,
+      "step": 172850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3675235256204888e-06,
+      "loss": 1.2887,
+      "step": 172855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.367142097122658e-06,
+      "loss": 1.5176,
+      "step": 172860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.366760717923088e-06,
+      "loss": 1.0076,
+      "step": 172865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3663793880239606e-06,
+      "loss": 1.1589,
+      "step": 172870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3659981074274486e-06,
+      "loss": 0.9219,
+      "step": 172875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3656168761357325e-06,
+      "loss": 1.2064,
+      "step": 172880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3652356941509893e-06,
+      "loss": 1.2304,
+      "step": 172885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3648545614753916e-06,
+      "loss": 1.2236,
+      "step": 172890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.364473478111118e-06,
+      "loss": 1.1516,
+      "step": 172895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3640924440603487e-06,
+      "loss": 1.2155,
+      "step": 172900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3637114593252542e-06,
+      "loss": 1.3758,
+      "step": 172905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.363330523908014e-06,
+      "loss": 1.0703,
+      "step": 172910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3629496378108e-06,
+      "loss": 1.1447,
+      "step": 172915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3625688010357906e-06,
+      "loss": 1.5671,
+      "step": 172920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3621880135851573e-06,
+      "loss": 1.3139,
+      "step": 172925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3618072754610788e-06,
+      "loss": 1.3655,
+      "step": 172930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.361426586665724e-06,
+      "loss": 1.2534,
+      "step": 172935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3610459472012706e-06,
+      "loss": 0.9825,
+      "step": 172940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3606653570698925e-06,
+      "loss": 1.136,
+      "step": 172945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3602848162737624e-06,
+      "loss": 1.2086,
+      "step": 172950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3599043248150502e-06,
+      "loss": 0.916,
+      "step": 172955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3595238826959323e-06,
+      "loss": 1.307,
+      "step": 172960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3591434899185818e-06,
+      "loss": 1.4387,
+      "step": 172965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3587631464851669e-06,
+      "loss": 1.314,
+      "step": 172970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3583828523978648e-06,
+      "loss": 1.1138,
+      "step": 172975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3580026076588449e-06,
+      "loss": 1.2064,
+      "step": 172980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.357622412270273e-06,
+      "loss": 1.223,
+      "step": 172985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.357242266234331e-06,
+      "loss": 1.1167,
+      "step": 172990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3568621695531836e-06,
+      "loss": 1.3821,
+      "step": 172995
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3564821222290015e-06,
+      "loss": 1.5543,
+      "step": 173000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3561021242639538e-06,
+      "loss": 1.0812,
+      "step": 173005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3557221756602123e-06,
+      "loss": 1.3241,
+      "step": 173010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3553422764199486e-06,
+      "loss": 1.2065,
+      "step": 173015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3549624265453277e-06,
+      "loss": 1.2192,
+      "step": 173020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.354582626038523e-06,
+      "loss": 1.4673,
+      "step": 173025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.354202874901699e-06,
+      "loss": 1.4677,
+      "step": 173030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3538231731370276e-06,
+      "loss": 1.035,
+      "step": 173035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.353443520746678e-06,
+      "loss": 1.0464,
+      "step": 173040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3530639177328164e-06,
+      "loss": 1.1022,
+      "step": 173045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3526843640976085e-06,
+      "loss": 0.8979,
+      "step": 173050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3523048598432242e-06,
+      "loss": 1.3892,
+      "step": 173055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3519254049718322e-06,
+      "loss": 1.4135,
+      "step": 173060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3515459994855972e-06,
+      "loss": 1.1822,
+      "step": 173065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3511666433866844e-06,
+      "loss": 1.1183,
+      "step": 173070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3507873366772639e-06,
+      "loss": 1.3449,
+      "step": 173075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3504080793594975e-06,
+      "loss": 1.2018,
+      "step": 173080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3500288714355537e-06,
+      "loss": 1.4196,
+      "step": 173085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.349649712907599e-06,
+      "loss": 1.3871,
+      "step": 173090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.349270603777797e-06,
+      "loss": 1.0766,
+      "step": 173095
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3488915440483109e-06,
+      "loss": 1.498,
+      "step": 173100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3485125337213067e-06,
+      "loss": 0.8356,
+      "step": 173105
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.348133572798952e-06,
+      "loss": 1.0602,
+      "step": 173110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3477546612834058e-06,
+      "loss": 1.0787,
+      "step": 173115
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3473757991768355e-06,
+      "loss": 1.3141,
+      "step": 173120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3469969864814015e-06,
+      "loss": 1.3718,
+      "step": 173125
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3466182231992686e-06,
+      "loss": 1.1815,
+      "step": 173130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3462395093326019e-06,
+      "loss": 1.3121,
+      "step": 173135
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.345860844883563e-06,
+      "loss": 1.4098,
+      "step": 173140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3454822298543101e-06,
+      "loss": 1.1538,
+      "step": 173145
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3451036642470084e-06,
+      "loss": 1.37,
+      "step": 173150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3447251480638224e-06,
+      "loss": 1.6774,
+      "step": 173155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3443466813069116e-06,
+      "loss": 1.0585,
+      "step": 173160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3439682639784335e-06,
+      "loss": 1.2817,
+      "step": 173165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3435898960805538e-06,
+      "loss": 1.3855,
+      "step": 173170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.343211577615432e-06,
+      "loss": 1.3059,
+      "step": 173175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3428333085852273e-06,
+      "loss": 1.1219,
+      "step": 173180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3424550889921028e-06,
+      "loss": 1.8099,
+      "step": 173185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3420769188382155e-06,
+      "loss": 1.843,
+      "step": 173190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3416987981257235e-06,
+      "loss": 1.041,
+      "step": 173195
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3413207268567884e-06,
+      "loss": 1.31,
+      "step": 173200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.340942705033572e-06,
+      "loss": 1.3152,
+      "step": 173205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3405647326582294e-06,
+      "loss": 1.1719,
+      "step": 173210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3401868097329173e-06,
+      "loss": 1.0594,
+      "step": 173215
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3398089362597965e-06,
+      "loss": 1.4934,
+      "step": 173220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3394311122410275e-06,
+      "loss": 1.5711,
+      "step": 173225
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3390533376787619e-06,
+      "loss": 0.9846,
+      "step": 173230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.338675612575162e-06,
+      "loss": 1.4411,
+      "step": 173235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3382979369323811e-06,
+      "loss": 1.6242,
+      "step": 173240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3379203107525773e-06,
+      "loss": 1.2694,
+      "step": 173245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.33754273403791e-06,
+      "loss": 1.3439,
+      "step": 173250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3371652067905328e-06,
+      "loss": 1.1796,
+      "step": 173255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3367877290125997e-06,
+      "loss": 1.3661,
+      "step": 173260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3364103007062678e-06,
+      "loss": 1.4235,
+      "step": 173265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3360329218736945e-06,
+      "loss": 1.209,
+      "step": 173270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3356555925170323e-06,
+      "loss": 1.2154,
+      "step": 173275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3352783126384384e-06,
+      "loss": 1.2285,
+      "step": 173280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3349010822400655e-06,
+      "loss": 1.0823,
+      "step": 173285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3345239013240651e-06,
+      "loss": 1.218,
+      "step": 173290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3341467698925946e-06,
+      "loss": 1.4463,
+      "step": 173295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.33376968794781e-06,
+      "loss": 1.3294,
+      "step": 173300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3333926554918608e-06,
+      "loss": 1.5567,
+      "step": 173305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3330156725268984e-06,
+      "loss": 1.2417,
+      "step": 173310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3326387390550787e-06,
+      "loss": 1.5611,
+      "step": 173315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3322618550785548e-06,
+      "loss": 1.2491,
+      "step": 173320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3318850205994772e-06,
+      "loss": 1.4718,
+      "step": 173325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3315082356199993e-06,
+      "loss": 1.1873,
+      "step": 173330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3311315001422697e-06,
+      "loss": 1.4054,
+      "step": 173335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3307548141684412e-06,
+      "loss": 1.1202,
+      "step": 173340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3303781777006685e-06,
+      "loss": 1.2269,
+      "step": 173345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.330001590741099e-06,
+      "loss": 1.3814,
+      "step": 173350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3296250532918842e-06,
+      "loss": 1.3102,
+      "step": 173355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.329248565355169e-06,
+      "loss": 1.345,
+      "step": 173360
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3288721269331118e-06,
+      "loss": 1.3394,
+      "step": 173365
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3284957380278596e-06,
+      "loss": 1.4591,
+      "step": 173370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3281193986415575e-06,
+      "loss": 1.2218,
+      "step": 173375
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3277431087763604e-06,
+      "loss": 1.1679,
+      "step": 173380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3273668684344132e-06,
+      "loss": 1.7652,
+      "step": 173385
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3269906776178643e-06,
+      "loss": 1.193,
+      "step": 173390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3266145363288675e-06,
+      "loss": 1.0538,
+      "step": 173395
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3262384445695653e-06,
+      "loss": 1.1737,
+      "step": 173400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.325862402342104e-06,
+      "loss": 1.4222,
+      "step": 173405
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3254864096486354e-06,
+      "loss": 1.537,
+      "step": 173410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.325110466491306e-06,
+      "loss": 1.329,
+      "step": 173415
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3247345728722604e-06,
+      "loss": 1.188,
+      "step": 173420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3243587287936476e-06,
+      "loss": 1.1752,
+      "step": 173425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3239829342576116e-06,
+      "loss": 1.5826,
+      "step": 173430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3236071892663004e-06,
+      "loss": 1.296,
+      "step": 173435
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3232314938218582e-06,
+      "loss": 1.2506,
+      "step": 173440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.322855847926432e-06,
+      "loss": 1.3889,
+      "step": 173445
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3224802515821633e-06,
+      "loss": 0.99,
+      "step": 173450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3221047047912005e-06,
+      "loss": 1.1646,
+      "step": 173455
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3217292075556898e-06,
+      "loss": 1.2345,
+      "step": 173460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3213537598777714e-06,
+      "loss": 1.2059,
+      "step": 173465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3209783617595894e-06,
+      "loss": 1.2697,
+      "step": 173470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.320603013203292e-06,
+      "loss": 1.2521,
+      "step": 173475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3202277142110165e-06,
+      "loss": 1.4402,
+      "step": 173480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3198524647849099e-06,
+      "loss": 1.1863,
+      "step": 173485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3194772649271158e-06,
+      "loss": 1.2336,
+      "step": 173490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3191021146397765e-06,
+      "loss": 1.1944,
+      "step": 173495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3187270139250297e-06,
+      "loss": 1.4691,
+      "step": 173500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3183519627850216e-06,
+      "loss": 1.2608,
+      "step": 173505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3179769612218952e-06,
+      "loss": 0.9859,
+      "step": 173510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3176020092377895e-06,
+      "loss": 1.1269,
+      "step": 173515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3172271068348451e-06,
+      "loss": 1.407,
+      "step": 173520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3168522540152028e-06,
+      "loss": 1.2019,
+      "step": 173525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3164774507810073e-06,
+      "loss": 1.5394,
+      "step": 173530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3161026971343938e-06,
+      "loss": 1.1598,
+      "step": 173535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3157279930775069e-06,
+      "loss": 1.4563,
+      "step": 173540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3153533386124818e-06,
+      "loss": 1.1663,
+      "step": 173545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.31497873374146e-06,
+      "loss": 1.5,
+      "step": 173550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.314604178466582e-06,
+      "loss": 1.2951,
+      "step": 173555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3142296727899873e-06,
+      "loss": 1.2309,
+      "step": 173560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3138552167138097e-06,
+      "loss": 1.0137,
+      "step": 173565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3134808102401909e-06,
+      "loss": 1.9115,
+      "step": 173570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3131064533712713e-06,
+      "loss": 1.36,
+      "step": 173575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3127321461091858e-06,
+      "loss": 1.0341,
+      "step": 173580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3123578884560695e-06,
+      "loss": 1.5424,
+      "step": 173585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.311983680414065e-06,
+      "loss": 1.256,
+      "step": 173590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3116095219853043e-06,
+      "loss": 1.4615,
+      "step": 173595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3112354131719262e-06,
+      "loss": 1.7938,
+      "step": 173600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3108613539760694e-06,
+      "loss": 1.2519,
+      "step": 173605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3104873443998688e-06,
+      "loss": 1.1562,
+      "step": 173610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3101133844454561e-06,
+      "loss": 1.2803,
+      "step": 173615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3097394741149694e-06,
+      "loss": 1.2745,
+      "step": 173620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.309365613410547e-06,
+      "loss": 1.3486,
+      "step": 173625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3089918023343184e-06,
+      "loss": 1.11,
+      "step": 173630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3086180408884241e-06,
+      "loss": 1.1085,
+      "step": 173635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3082443290749935e-06,
+      "loss": 1.1304,
+      "step": 173640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3078706668961616e-06,
+      "loss": 1.6032,
+      "step": 173645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3074970543540666e-06,
+      "loss": 1.451,
+      "step": 173650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3071234914508379e-06,
+      "loss": 1.5791,
+      "step": 173655
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3067499781886094e-06,
+      "loss": 1.3661,
+      "step": 173660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3063765145695107e-06,
+      "loss": 1.2511,
+      "step": 173665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.306003100595682e-06,
+      "loss": 1.4027,
+      "step": 173670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3056297362692516e-06,
+      "loss": 1.1551,
+      "step": 173675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3052564215923492e-06,
+      "loss": 1.1378,
+      "step": 173680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3048831565671115e-06,
+      "loss": 1.2366,
+      "step": 173685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3045099411956652e-06,
+      "loss": 1.0356,
+      "step": 173690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3041367754801449e-06,
+      "loss": 1.4484,
+      "step": 173695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.303763659422682e-06,
+      "loss": 1.3782,
+      "step": 173700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3033905930254065e-06,
+      "loss": 1.2129,
+      "step": 173705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3030175762904462e-06,
+      "loss": 0.9618,
+      "step": 173710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3026446092199319e-06,
+      "loss": 1.3397,
+      "step": 173715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3022716918159984e-06,
+      "loss": 1.154,
+      "step": 173720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3018988240807707e-06,
+      "loss": 0.8446,
+      "step": 173725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.301526006016377e-06,
+      "loss": 1.5856,
+      "step": 173730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3011532376249492e-06,
+      "loss": 1.6883,
+      "step": 173735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3007805189086164e-06,
+      "loss": 1.1056,
+      "step": 173740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.300407849869504e-06,
+      "loss": 1.1777,
+      "step": 173745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3000352305097442e-06,
+      "loss": 1.0791,
+      "step": 173750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.29966266083146e-06,
+      "loss": 1.1669,
+      "step": 173755
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.299290140836781e-06,
+      "loss": 1.1358,
+      "step": 173760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2989176705278372e-06,
+      "loss": 1.1954,
+      "step": 173765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2985452499067541e-06,
+      "loss": 1.626,
+      "step": 173770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2981728789756553e-06,
+      "loss": 1.3495,
+      "step": 173775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2978005577366714e-06,
+      "loss": 1.0007,
+      "step": 173780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2974282861919251e-06,
+      "loss": 1.051,
+      "step": 173785
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2970560643435448e-06,
+      "loss": 1.4333,
+      "step": 173790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2966838921936563e-06,
+      "loss": 1.0124,
+      "step": 173795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2963117697443838e-06,
+      "loss": 1.2951,
+      "step": 173800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2959396969978511e-06,
+      "loss": 1.4194,
+      "step": 173805
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.295567673956184e-06,
+      "loss": 0.922,
+      "step": 173810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2951957006215087e-06,
+      "loss": 1.2627,
+      "step": 173815
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2948237769959483e-06,
+      "loss": 1.2475,
+      "step": 173820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2944519030816239e-06,
+      "loss": 1.1071,
+      "step": 173825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2940800788806618e-06,
+      "loss": 1.1399,
+      "step": 173830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2937083043951858e-06,
+      "loss": 1.1907,
+      "step": 173835
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2933365796273179e-06,
+      "loss": 1.0442,
+      "step": 173840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2929649045791815e-06,
+      "loss": 1.4507,
+      "step": 173845
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2925932792528971e-06,
+      "loss": 1.3708,
+      "step": 173850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.292221703650589e-06,
+      "loss": 0.894,
+      "step": 173855
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2918501777743796e-06,
+      "loss": 1.1889,
+      "step": 173860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2914787016263908e-06,
+      "loss": 1.6086,
+      "step": 173865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2911072752087405e-06,
+      "loss": 1.1047,
+      "step": 173870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2907358985235508e-06,
+      "loss": 1.2022,
+      "step": 173875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.290364571572944e-06,
+      "loss": 1.3111,
+      "step": 173880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.289993294359041e-06,
+      "loss": 1.2123,
+      "step": 173885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2896220668839599e-06,
+      "loss": 1.2179,
+      "step": 173890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2892508891498223e-06,
+      "loss": 0.8866,
+      "step": 173895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2888797611587467e-06,
+      "loss": 1.0857,
+      "step": 173900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2885086829128525e-06,
+      "loss": 1.0146,
+      "step": 173905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.288137654414261e-06,
+      "loss": 1.2867,
+      "step": 173910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2877666756650887e-06,
+      "loss": 1.2078,
+      "step": 173915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2873957466674537e-06,
+      "loss": 1.0759,
+      "step": 173920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2870248674234741e-06,
+      "loss": 1.5005,
+      "step": 173925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.286654037935271e-06,
+      "loss": 1.1271,
+      "step": 173930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2862832582049577e-06,
+      "loss": 1.1788,
+      "step": 173935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2859125282346563e-06,
+      "loss": 1.3341,
+      "step": 173940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.285541848026479e-06,
+      "loss": 1.3024,
+      "step": 173945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.285171217582547e-06,
+      "loss": 1.264,
+      "step": 173950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2848006369049714e-06,
+      "loss": 0.9486,
+      "step": 173955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2844301059958752e-06,
+      "loss": 1.3007,
+      "step": 173960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.28405962485737e-06,
+      "loss": 1.0282,
+      "step": 173965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2836891934915685e-06,
+      "loss": 1.1755,
+      "step": 173970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2833188119005947e-06,
+      "loss": 1.4798,
+      "step": 173975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2829484800865578e-06,
+      "loss": 0.9221,
+      "step": 173980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2825781980515717e-06,
+      "loss": 1.0833,
+      "step": 173985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2822079657977538e-06,
+      "loss": 1.1741,
+      "step": 173990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2818377833272166e-06,
+      "loss": 1.6494,
+      "step": 173995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2814676506420742e-06,
+      "loss": 0.8888,
+      "step": 174000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2810975677444425e-06,
+      "loss": 1.0692,
+      "step": 174005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2807275346364345e-06,
+      "loss": 1.381,
+      "step": 174010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2803575513201583e-06,
+      "loss": 1.3112,
+      "step": 174015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.279987617797731e-06,
+      "loss": 1.0657,
+      "step": 174020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2796177340712668e-06,
+      "loss": 1.303,
+      "step": 174025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.279247900142876e-06,
+      "loss": 1.2625,
+      "step": 174030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2788781160146668e-06,
+      "loss": 0.9949,
+      "step": 174035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2785083816887557e-06,
+      "loss": 1.1548,
+      "step": 174040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2781386971672538e-06,
+      "loss": 1.2636,
+      "step": 174045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2777690624522687e-06,
+      "loss": 0.9135,
+      "step": 174050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2773994775459165e-06,
+      "loss": 1.0427,
+      "step": 174055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.277029942450304e-06,
+      "loss": 1.1932,
+      "step": 174060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.276660457167539e-06,
+      "loss": 1.2653,
+      "step": 174065
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.276291021699737e-06,
+      "loss": 1.5658,
+      "step": 174070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2759216360490068e-06,
+      "loss": 1.2745,
+      "step": 174075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2755523002174542e-06,
+      "loss": 1.4143,
+      "step": 174080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.275183014207192e-06,
+      "loss": 1.2428,
+      "step": 174085
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2748137780203263e-06,
+      "loss": 1.4594,
+      "step": 174090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2744445916589676e-06,
+      "loss": 1.4431,
+      "step": 174095
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2740754551252222e-06,
+      "loss": 1.3652,
+      "step": 174100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2737063684212003e-06,
+      "loss": 1.6282,
+      "step": 174105
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.273337331549007e-06,
+      "loss": 1.0464,
+      "step": 174110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.272968344510751e-06,
+      "loss": 1.3769,
+      "step": 174115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2725994073085423e-06,
+      "loss": 1.0901,
+      "step": 174120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.272230519944485e-06,
+      "loss": 1.0044,
+      "step": 174125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2718616824206831e-06,
+      "loss": 1.3377,
+      "step": 174130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2714928947392458e-06,
+      "loss": 1.1603,
+      "step": 174135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2711241569022802e-06,
+      "loss": 1.1776,
+      "step": 174140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2707554689118884e-06,
+      "loss": 1.4166,
+      "step": 174145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2703868307701806e-06,
+      "loss": 1.3603,
+      "step": 174150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2700182424792574e-06,
+      "loss": 1.1842,
+      "step": 174155
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.269649704041227e-06,
+      "loss": 1.7203,
+      "step": 174160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2692812154581902e-06,
+      "loss": 1.3242,
+      "step": 174165
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2689127767322562e-06,
+      "loss": 1.3819,
+      "step": 174170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2685443878655258e-06,
+      "loss": 1.0475,
+      "step": 174175
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2681760488601013e-06,
+      "loss": 0.8974,
+      "step": 174180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2678077597180883e-06,
+      "loss": 1.0831,
+      "step": 174185
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2674395204415912e-06,
+      "loss": 1.3515,
+      "step": 174190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2670713310327088e-06,
+      "loss": 0.9922,
+      "step": 174195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.266703191493549e-06,
+      "loss": 1.2027,
+      "step": 174200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2663351018262082e-06,
+      "loss": 1.2262,
+      "step": 174205
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2659670620327924e-06,
+      "loss": 1.1036,
+      "step": 174210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2655990721154042e-06,
+      "loss": 1.5492,
+      "step": 174215
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2652311320761424e-06,
+      "loss": 1.0735,
+      "step": 174220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.264863241917108e-06,
+      "loss": 1.2007,
+      "step": 174225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.264495401640401e-06,
+      "loss": 1.376,
+      "step": 174230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.264127611248127e-06,
+      "loss": 1.049,
+      "step": 174235
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2637598707423826e-06,
+      "loss": 1.134,
+      "step": 174240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.263392180125267e-06,
+      "loss": 1.2387,
+      "step": 174245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2630245393988795e-06,
+      "loss": 1.0896,
+      "step": 174250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2626569485653239e-06,
+      "loss": 1.1695,
+      "step": 174255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2622894076266945e-06,
+      "loss": 1.107,
+      "step": 174260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2619219165850938e-06,
+      "loss": 1.3579,
+      "step": 174265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.261554475442619e-06,
+      "loss": 1.1654,
+      "step": 174270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2611870842013641e-06,
+      "loss": 1.1536,
+      "step": 174275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.260819742863436e-06,
+      "loss": 1.0305,
+      "step": 174280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2604524514309268e-06,
+      "loss": 1.4683,
+      "step": 174285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2600852099059325e-06,
+      "loss": 1.2117,
+      "step": 174290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2597180182905544e-06,
+      "loss": 1.1986,
+      "step": 174295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2593508765868857e-06,
+      "loss": 1.092,
+      "step": 174300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2589837847970266e-06,
+      "loss": 0.9948,
+      "step": 174305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2586167429230701e-06,
+      "loss": 1.2567,
+      "step": 174310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2582497509671144e-06,
+      "loss": 1.18,
+      "step": 174315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2578828089312534e-06,
+      "loss": 1.3055,
+      "step": 174320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2575159168175822e-06,
+      "loss": 1.0193,
+      "step": 174325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2571490746282e-06,
+      "loss": 1.078,
+      "step": 174330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2567822823651998e-06,
+      "loss": 1.6245,
+      "step": 174335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2564155400306721e-06,
+      "loss": 1.5546,
+      "step": 174340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2560488476267152e-06,
+      "loss": 1.3001,
+      "step": 174345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.255682205155424e-06,
+      "loss": 1.3087,
+      "step": 174350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2553156126188892e-06,
+      "loss": 1.3939,
+      "step": 174355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2549490700192069e-06,
+      "loss": 1.2735,
+      "step": 174360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2545825773584686e-06,
+      "loss": 1.1834,
+      "step": 174365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2542161346387637e-06,
+      "loss": 1.4577,
+      "step": 174370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.253849741862193e-06,
+      "loss": 1.4002,
+      "step": 174375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2534833990308449e-06,
+      "loss": 1.4171,
+      "step": 174380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2531171061468106e-06,
+      "loss": 1.2017,
+      "step": 174385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.25275086321218e-06,
+      "loss": 1.4909,
+      "step": 174390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2523846702290465e-06,
+      "loss": 1.1129,
+      "step": 174395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2520185271995034e-06,
+      "loss": 1.0733,
+      "step": 174400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2516524341256375e-06,
+      "loss": 1.1693,
+      "step": 174405
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.251286391009544e-06,
+      "loss": 1.2069,
+      "step": 174410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.250920397853308e-06,
+      "loss": 1.1288,
+      "step": 174415
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.250554454659023e-06,
+      "loss": 1.0066,
+      "step": 174420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2501885614287802e-06,
+      "loss": 1.3688,
+      "step": 174425
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.249822718164666e-06,
+      "loss": 1.1538,
+      "step": 174430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2494569248687694e-06,
+      "loss": 1.3146,
+      "step": 174435
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2490911815431795e-06,
+      "loss": 1.4528,
+      "step": 174440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2487254881899869e-06,
+      "loss": 1.2652,
+      "step": 174445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.24835984481128e-06,
+      "loss": 1.18,
+      "step": 174450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2479942514091425e-06,
+      "loss": 1.237,
+      "step": 174455
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2476287079856685e-06,
+      "loss": 1.3721,
+      "step": 174460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2472632145429387e-06,
+      "loss": 1.1861,
+      "step": 174465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2468977710830443e-06,
+      "loss": 0.9389,
+      "step": 174470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.246532377608073e-06,
+      "loss": 1.5985,
+      "step": 174475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2461670341201104e-06,
+      "loss": 1.1994,
+      "step": 174480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2458017406212398e-06,
+      "loss": 1.439,
+      "step": 174485
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2454364971135502e-06,
+      "loss": 1.2506,
+      "step": 174490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.245071303599129e-06,
+      "loss": 1.1216,
+      "step": 174495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.244706160080057e-06,
+      "loss": 1.0574,
+      "step": 174500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2443410665584242e-06,
+      "loss": 1.1622,
+      "step": 174505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2439760230363106e-06,
+      "loss": 1.1497,
+      "step": 174510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2436110295158044e-06,
+      "loss": 1.0486,
+      "step": 174515
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2432460859989902e-06,
+      "loss": 1.4786,
+      "step": 174520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2428811924879502e-06,
+      "loss": 1.4679,
+      "step": 174525
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2425163489847668e-06,
+      "loss": 1.399,
+      "step": 174530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2421515554915265e-06,
+      "loss": 1.0702,
+      "step": 174535
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2417868120103115e-06,
+      "loss": 1.3033,
+      "step": 174540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2414221185432063e-06,
+      "loss": 1.1239,
+      "step": 174545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2410574750922888e-06,
+      "loss": 1.0745,
+      "step": 174550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2406928816596432e-06,
+      "loss": 1.3673,
+      "step": 174555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2403283382473551e-06,
+      "loss": 1.2569,
+      "step": 174560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2399638448575024e-06,
+      "loss": 1.1973,
+      "step": 174565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2395994014921687e-06,
+      "loss": 1.0729,
+      "step": 174570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2392350081534354e-06,
+      "loss": 1.064,
+      "step": 174575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.238870664843379e-06,
+      "loss": 1.1452,
+      "step": 174580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2385063715640833e-06,
+      "loss": 1.2424,
+      "step": 174585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.238142128317631e-06,
+      "loss": 1.3917,
+      "step": 174590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2377779351061003e-06,
+      "loss": 1.168,
+      "step": 174595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2374137919315676e-06,
+      "loss": 1.3951,
+      "step": 174600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2370496987961155e-06,
+      "loss": 1.2861,
+      "step": 174605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2366856557018247e-06,
+      "loss": 0.8129,
+      "step": 174610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.23632166265077e-06,
+      "loss": 1.1175,
+      "step": 174615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2359577196450346e-06,
+      "loss": 1.1712,
+      "step": 174620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2355938266866919e-06,
+      "loss": 1.0293,
+      "step": 174625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2352299837778226e-06,
+      "loss": 1.3282,
+      "step": 174630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2348661909205063e-06,
+      "loss": 1.5357,
+      "step": 174635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2345024481168177e-06,
+      "loss": 1.094,
+      "step": 174640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2341387553688333e-06,
+      "loss": 1.085,
+      "step": 174645
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2337751126786312e-06,
+      "loss": 1.1627,
+      "step": 174650
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.233411520048291e-06,
+      "loss": 1.0539,
+      "step": 174655
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.233047977479883e-06,
+      "loss": 1.4346,
+      "step": 174660
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2326844849754892e-06,
+      "loss": 1.3242,
+      "step": 174665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2323210425371823e-06,
+      "loss": 1.1228,
+      "step": 174670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.231957650167035e-06,
+      "loss": 1.1414,
+      "step": 174675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2315943078671267e-06,
+      "loss": 1.4315,
+      "step": 174680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2312310156395325e-06,
+      "loss": 1.2475,
+      "step": 174685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.230867773486325e-06,
+      "loss": 1.4036,
+      "step": 174690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2305045814095774e-06,
+      "loss": 1.5265,
+      "step": 174695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2301414394113653e-06,
+      "loss": 1.1718,
+      "step": 174700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2297783474937653e-06,
+      "loss": 1.1076,
+      "step": 174705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2294153056588453e-06,
+      "loss": 1.0529,
+      "step": 174710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.229052313908683e-06,
+      "loss": 1.4473,
+      "step": 174715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2286893722453475e-06,
+      "loss": 1.1089,
+      "step": 174720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.228326480670914e-06,
+      "loss": 1.2017,
+      "step": 174725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.227963639187455e-06,
+      "loss": 1.3758,
+      "step": 174730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2276008477970424e-06,
+      "loss": 1.5732,
+      "step": 174735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2272381065017457e-06,
+      "loss": 1.2049,
+      "step": 174740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2268754153036377e-06,
+      "loss": 1.1688,
+      "step": 174745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2265127742047912e-06,
+      "loss": 1.2151,
+      "step": 174750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2261501832072753e-06,
+      "loss": 0.9969,
+      "step": 174755
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2257876423131598e-06,
+      "loss": 1.2,
+      "step": 174760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2254251515245176e-06,
+      "loss": 0.9893,
+      "step": 174765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2250627108434155e-06,
+      "loss": 0.9984,
+      "step": 174770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2247003202719243e-06,
+      "loss": 1.2197,
+      "step": 174775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2243379798121169e-06,
+      "loss": 1.369,
+      "step": 174780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2239756894660592e-06,
+      "loss": 1.3416,
+      "step": 174785
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2236134492358186e-06,
+      "loss": 1.1104,
+      "step": 174790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2232512591234658e-06,
+      "loss": 1.0579,
+      "step": 174795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2228891191310699e-06,
+      "loss": 1.2402,
+      "step": 174800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2225270292606973e-06,
+      "loss": 1.0151,
+      "step": 174805
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2221649895144172e-06,
+      "loss": 1.1252,
+      "step": 174810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.221802999894295e-06,
+      "loss": 1.5815,
+      "step": 174815
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2214410604024007e-06,
+      "loss": 1.4742,
+      "step": 174820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2210791710407977e-06,
+      "loss": 1.3295,
+      "step": 174825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2207173318115561e-06,
+      "loss": 1.4517,
+      "step": 174830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.220355542716739e-06,
+      "loss": 1.1255,
+      "step": 174835
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2199938037584136e-06,
+      "loss": 1.5715,
+      "step": 174840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.219632114938648e-06,
+      "loss": 1.3806,
+      "step": 174845
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2192704762595052e-06,
+      "loss": 1.5025,
+      "step": 174850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2189088877230492e-06,
+      "loss": 1.3947,
+      "step": 174855
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.218547349331346e-06,
+      "loss": 1.0929,
+      "step": 174860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2181858610864627e-06,
+      "loss": 1.2447,
+      "step": 174865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.217824422990459e-06,
+      "loss": 1.1657,
+      "step": 174870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2174630350454042e-06,
+      "loss": 0.9357,
+      "step": 174875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2171016972533589e-06,
+      "loss": 2.0119,
+      "step": 174880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.216740409616385e-06,
+      "loss": 1.1462,
+      "step": 174885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.216379172136547e-06,
+      "loss": 1.0727,
+      "step": 174890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2160179848159103e-06,
+      "loss": 1.0899,
+      "step": 174895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2156568476565356e-06,
+      "loss": 1.0475,
+      "step": 174900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.215295760660482e-06,
+      "loss": 1.4348,
+      "step": 174905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.214934723829816e-06,
+      "loss": 1.3817,
+      "step": 174910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2145737371665988e-06,
+      "loss": 1.1428,
+      "step": 174915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2142128006728893e-06,
+      "loss": 1.2067,
+      "step": 174920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.213851914350752e-06,
+      "loss": 1.3405,
+      "step": 174925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2134910782022435e-06,
+      "loss": 1.1846,
+      "step": 174930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2131302922294274e-06,
+      "loss": 1.207,
+      "step": 174935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2127695564343656e-06,
+      "loss": 1.459,
+      "step": 174940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.212408870819115e-06,
+      "loss": 1.0562,
+      "step": 174945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2120482353857354e-06,
+      "loss": 1.5391,
+      "step": 174950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2116876501362862e-06,
+      "loss": 1.2647,
+      "step": 174955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.211327115072829e-06,
+      "loss": 1.1902,
+      "step": 174960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.210966630197421e-06,
+      "loss": 1.0658,
+      "step": 174965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2106061955121195e-06,
+      "loss": 1.3902,
+      "step": 174970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.210245811018985e-06,
+      "loss": 1.1144,
+      "step": 174975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.209885476720074e-06,
+      "loss": 1.3574,
+      "step": 174980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2095251926174433e-06,
+      "loss": 1.1597,
+      "step": 174985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2091649587131549e-06,
+      "loss": 1.2118,
+      "step": 174990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2088047750092613e-06,
+      "loss": 1.4004,
+      "step": 174995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.20844464150782e-06,
+      "loss": 0.9968,
+      "step": 175000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.208084558210888e-06,
+      "loss": 1.3807,
+      "step": 175005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.207724525120524e-06,
+      "loss": 1.0432,
+      "step": 175010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2073645422387792e-06,
+      "loss": 1.1522,
+      "step": 175015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2070046095677147e-06,
+      "loss": 1.5352,
+      "step": 175020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2066447271093818e-06,
+      "loss": 1.2688,
+      "step": 175025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.206284894865838e-06,
+      "loss": 1.3335,
+      "step": 175030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2059251128391359e-06,
+      "loss": 1.0109,
+      "step": 175035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2055653810313328e-06,
+      "loss": 1.067,
+      "step": 175040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2052056994444815e-06,
+      "loss": 1.1468,
+      "step": 175045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2048460680806317e-06,
+      "loss": 1.6022,
+      "step": 175050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.204486486941846e-06,
+      "loss": 1.8824,
+      "step": 175055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2041269560301728e-06,
+      "loss": 1.0451,
+      "step": 175060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2037674753476635e-06,
+      "loss": 1.1452,
+      "step": 175065
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.203408044896376e-06,
+      "loss": 1.3775,
+      "step": 175070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2030486646783568e-06,
+      "loss": 1.1485,
+      "step": 175075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2026893346956603e-06,
+      "loss": 1.0981,
+      "step": 175080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2023300549503425e-06,
+      "loss": 1.1773,
+      "step": 175085
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2019708254444518e-06,
+      "loss": 1.4548,
+      "step": 175090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2016116461800387e-06,
+      "loss": 1.1442,
+      "step": 175095
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2012525171591537e-06,
+      "loss": 1.0896,
+      "step": 175100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2008934383838522e-06,
+      "loss": 1.13,
+      "step": 175105
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.200534409856181e-06,
+      "loss": 1.3688,
+      "step": 175110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2001754315781899e-06,
+      "loss": 1.3227,
+      "step": 175115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1998165035519294e-06,
+      "loss": 1.1859,
+      "step": 175120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1994576257794533e-06,
+      "loss": 1.3131,
+      "step": 175125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1990987982628044e-06,
+      "loss": 1.1838,
+      "step": 175130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1987400210040368e-06,
+      "loss": 1.1947,
+      "step": 175135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1983812940051952e-06,
+      "loss": 1.2675,
+      "step": 175140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1980226172683318e-06,
+      "loss": 1.1461,
+      "step": 175145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1976639907954946e-06,
+      "loss": 0.9793,
+      "step": 175150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1973054145887297e-06,
+      "loss": 1.165,
+      "step": 175155
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1969468886500846e-06,
+      "loss": 1.2198,
+      "step": 175160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1965884129816085e-06,
+      "loss": 1.1559,
+      "step": 175165
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1962299875853467e-06,
+      "loss": 1.3669,
+      "step": 175170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1958716124633473e-06,
+      "loss": 1.4067,
+      "step": 175175
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1955132876176556e-06,
+      "loss": 1.1592,
+      "step": 175180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1951550130503199e-06,
+      "loss": 1.0704,
+      "step": 175185
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1947967887633827e-06,
+      "loss": 1.4249,
+      "step": 175190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1944386147588916e-06,
+      "loss": 1.4328,
+      "step": 175195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1940804910388947e-06,
+      "loss": 1.6127,
+      "step": 175200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.193722417605434e-06,
+      "loss": 1.0696,
+      "step": 175205
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1933643944605533e-06,
+      "loss": 1.1553,
+      "step": 175210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1930064216062987e-06,
+      "loss": 1.1832,
+      "step": 175215
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1926484990447152e-06,
+      "loss": 1.1631,
+      "step": 175220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1922906267778444e-06,
+      "loss": 1.5034,
+      "step": 175225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.191932804807734e-06,
+      "loss": 1.1539,
+      "step": 175230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1915750331364218e-06,
+      "loss": 1.2426,
+      "step": 175235
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1912173117659542e-06,
+      "loss": 1.066,
+      "step": 175240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1908596406983753e-06,
+      "loss": 1.2996,
+      "step": 175245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1905020199357265e-06,
+      "loss": 0.9204,
+      "step": 175250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1901444494800485e-06,
+      "loss": 1.0711,
+      "step": 175255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1897869293333809e-06,
+      "loss": 1.1774,
+      "step": 175260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1894294594977719e-06,
+      "loss": 0.9923,
+      "step": 175265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1890720399752598e-06,
+      "loss": 1.2118,
+      "step": 175270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1887146707678832e-06,
+      "loss": 1.4412,
+      "step": 175275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.188357351877688e-06,
+      "loss": 1.0864,
+      "step": 175280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1880000833067084e-06,
+      "loss": 1.2287,
+      "step": 175285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1876428650569882e-06,
+      "loss": 1.5063,
+      "step": 175290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.187285697130569e-06,
+      "loss": 1.1418,
+      "step": 175295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1869285795294882e-06,
+      "loss": 1.1857,
+      "step": 175300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1865715122557843e-06,
+      "loss": 1.3461,
+      "step": 175305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1862144953114963e-06,
+      "loss": 1.089,
+      "step": 175310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1858575286986663e-06,
+      "loss": 1.7011,
+      "step": 175315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1855006124193313e-06,
+      "loss": 0.9857,
+      "step": 175320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1851437464755255e-06,
+      "loss": 1.2464,
+      "step": 175325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1847869308692905e-06,
+      "loss": 1.1781,
+      "step": 175330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1844301656026657e-06,
+      "loss": 1.0365,
+      "step": 175335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.184073450677684e-06,
+      "loss": 1.2447,
+      "step": 175340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.183716786096386e-06,
+      "loss": 1.076,
+      "step": 175345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1833601718608078e-06,
+      "loss": 1.0541,
+      "step": 175350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1830036079729802e-06,
+      "loss": 1.2216,
+      "step": 175355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.182647094434949e-06,
+      "loss": 1.2502,
+      "step": 175360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1822906312487447e-06,
+      "loss": 0.9466,
+      "step": 175365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.181934218416403e-06,
+      "loss": 1.1875,
+      "step": 175370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1815778559399604e-06,
+      "loss": 1.5244,
+      "step": 175375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.181221543821449e-06,
+      "loss": 1.1424,
+      "step": 175380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1808652820629062e-06,
+      "loss": 1.405,
+      "step": 175385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.180509070666368e-06,
+      "loss": 0.7842,
+      "step": 175390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.180152909633867e-06,
+      "loss": 1.2151,
+      "step": 175395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1797967989674342e-06,
+      "loss": 1.183,
+      "step": 175400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1794407386691065e-06,
+      "loss": 1.2138,
+      "step": 175405
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1790847287409168e-06,
+      "loss": 1.3406,
+      "step": 175410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1787287691848991e-06,
+      "loss": 1.2319,
+      "step": 175415
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.178372860003082e-06,
+      "loss": 1.3275,
+      "step": 175420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1780170011975e-06,
+      "loss": 1.4136,
+      "step": 175425
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1776611927701886e-06,
+      "loss": 1.2625,
+      "step": 175430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1773054347231749e-06,
+      "loss": 1.3662,
+      "step": 175435
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.176949727058495e-06,
+      "loss": 1.2285,
+      "step": 175440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1765940697781752e-06,
+      "loss": 1.1549,
+      "step": 175445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1762384628842482e-06,
+      "loss": 1.324,
+      "step": 175450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1758829063787481e-06,
+      "loss": 1.2404,
+      "step": 175455
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1755274002637028e-06,
+      "loss": 1.1653,
+      "step": 175460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1751719445411425e-06,
+      "loss": 1.11,
+      "step": 175465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.174816539213094e-06,
+      "loss": 1.2606,
+      "step": 175470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1744611842815901e-06,
+      "loss": 1.0143,
+      "step": 175475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1741058797486615e-06,
+      "loss": 1.3692,
+      "step": 175480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1737506256163344e-06,
+      "loss": 1.1436,
+      "step": 175485
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1733954218866394e-06,
+      "loss": 1.4116,
+      "step": 175490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1730402685616017e-06,
+      "loss": 1.3126,
+      "step": 175495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1726851656432526e-06,
+      "loss": 1.2477,
+      "step": 175500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1723301131336206e-06,
+      "loss": 1.1573,
+      "step": 175505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1719751110347322e-06,
+      "loss": 1.376,
+      "step": 175510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.171620159348611e-06,
+      "loss": 1.3129,
+      "step": 175515
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1712652580772876e-06,
+      "loss": 1.2545,
+      "step": 175520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1709104072227894e-06,
+      "loss": 0.9136,
+      "step": 175525
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1705556067871405e-06,
+      "loss": 1.1864,
+      "step": 175530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.17020085677237e-06,
+      "loss": 0.9857,
+      "step": 175535
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1698461571804998e-06,
+      "loss": 1.236,
+      "step": 175540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1694915080135583e-06,
+      "loss": 1.1421,
+      "step": 175545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1691369092735693e-06,
+      "loss": 1.1797,
+      "step": 175550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1687823609625592e-06,
+      "loss": 1.227,
+      "step": 175555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.168427863082552e-06,
+      "loss": 1.1692,
+      "step": 175560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.16807341563557e-06,
+      "loss": 1.2628,
+      "step": 175565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1677190186236386e-06,
+      "loss": 1.2055,
+      "step": 175570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.167364672048784e-06,
+      "loss": 1.3216,
+      "step": 175575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1670103759130258e-06,
+      "loss": 1.0695,
+      "step": 175580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.166656130218392e-06,
+      "loss": 1.6252,
+      "step": 175585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1663019349669002e-06,
+      "loss": 0.9225,
+      "step": 175590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1659477901605753e-06,
+      "loss": 1.2357,
+      "step": 175595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1655936958014425e-06,
+      "loss": 1.1752,
+      "step": 175600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.165239651891521e-06,
+      "loss": 1.1637,
+      "step": 175605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1648856584328306e-06,
+      "loss": 1.1728,
+      "step": 175610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.164531715427396e-06,
+      "loss": 1.3381,
+      "step": 175615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1641778228772382e-06,
+      "loss": 1.2058,
+      "step": 175620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1638239807843787e-06,
+      "loss": 1.2374,
+      "step": 175625
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1634701891508338e-06,
+      "loss": 1.3772,
+      "step": 175630
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1631164479786271e-06,
+      "loss": 1.0672,
+      "step": 175635
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1627627572697809e-06,
+      "loss": 0.9903,
+      "step": 175640
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.1624091170263097e-06,
+      "loss": 1.7828,
+      "step": 175645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1620555272502387e-06,
+      "loss": 1.2303,
+      "step": 175650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1617019879435832e-06,
+      "loss": 1.0659,
+      "step": 175655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1613484991083601e-06,
+      "loss": 1.0194,
+      "step": 175660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1609950607465935e-06,
+      "loss": 1.2058,
+      "step": 175665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.160641672860301e-06,
+      "loss": 1.2046,
+      "step": 175670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.160288335451496e-06,
+      "loss": 1.2517,
+      "step": 175675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1599350485222004e-06,
+      "loss": 1.1625,
+      "step": 175680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1595818120744285e-06,
+      "loss": 1.2598,
+      "step": 175685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1592286261102025e-06,
+      "loss": 1.3929,
+      "step": 175690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1588754906315326e-06,
+      "loss": 1.3515,
+      "step": 175695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.158522405640442e-06,
+      "loss": 1.2464,
+      "step": 175700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1581693711389409e-06,
+      "loss": 1.1232,
+      "step": 175705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1578163871290493e-06,
+      "loss": 1.2455,
+      "step": 175710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1574634536127826e-06,
+      "loss": 1.4735,
+      "step": 175715
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.157110570592157e-06,
+      "loss": 1.226,
+      "step": 175720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.156757738069183e-06,
+      "loss": 1.2865,
+      "step": 175725
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.156404956045879e-06,
+      "loss": 1.0691,
+      "step": 175730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1560522245242623e-06,
+      "loss": 1.0722,
+      "step": 175735
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1556995435063412e-06,
+      "loss": 1.1849,
+      "step": 175740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.155346912994134e-06,
+      "loss": 1.43,
+      "step": 175745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.154994332989654e-06,
+      "loss": 1.1512,
+      "step": 175750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1546418034949114e-06,
+      "loss": 1.2397,
+      "step": 175755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1542893245119225e-06,
+      "loss": 1.2943,
+      "step": 175760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1539368960427e-06,
+      "loss": 1.342,
+      "step": 175765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1535845180892557e-06,
+      "loss": 1.1402,
+      "step": 175770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1532321906536004e-06,
+      "loss": 1.1817,
+      "step": 175775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1528799137377478e-06,
+      "loss": 1.1882,
+      "step": 175780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1525276873437109e-06,
+      "loss": 1.0453,
+      "step": 175785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1521755114734967e-06,
+      "loss": 1.1901,
+      "step": 175790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1518233861291228e-06,
+      "loss": 1.2332,
+      "step": 175795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.151471311312593e-06,
+      "loss": 1.4451,
+      "step": 175800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1511192870259224e-06,
+      "loss": 1.2381,
+      "step": 175805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1507673132711217e-06,
+      "loss": 1.3475,
+      "step": 175810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1504153900501991e-06,
+      "loss": 1.246,
+      "step": 175815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.150063517365162e-06,
+      "loss": 1.1624,
+      "step": 175820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.149711695218022e-06,
+      "loss": 1.249,
+      "step": 175825
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1493599236107912e-06,
+      "loss": 1.2875,
+      "step": 175830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1490082025454753e-06,
+      "loss": 1.2403,
+      "step": 175835
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1486565320240805e-06,
+      "loss": 1.3192,
+      "step": 175840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.148304912048618e-06,
+      "loss": 2.0144,
+      "step": 175845
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1479533426210975e-06,
+      "loss": 1.0632,
+      "step": 175850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1476018237435227e-06,
+      "loss": 1.165,
+      "step": 175855
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.147250355417905e-06,
+      "loss": 1.4217,
+      "step": 175860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.146898937646248e-06,
+      "loss": 1.0833,
+      "step": 175865
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1465475704305584e-06,
+      "loss": 1.2816,
+      "step": 175870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1461962537728443e-06,
+      "loss": 1.1979,
+      "step": 175875
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1458449876751144e-06,
+      "loss": 1.4003,
+      "step": 175880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1454937721393678e-06,
+      "loss": 1.2284,
+      "step": 175885
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1451426071676176e-06,
+      "loss": 1.0786,
+      "step": 175890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.144791492761863e-06,
+      "loss": 1.0078,
+      "step": 175895
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1444404289241152e-06,
+      "loss": 1.0377,
+      "step": 175900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.144089415656372e-06,
+      "loss": 1.0316,
+      "step": 175905
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1437384529606443e-06,
+      "loss": 1.269,
+      "step": 175910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1433875408389305e-06,
+      "loss": 1.1709,
+      "step": 175915
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.143036679293238e-06,
+      "loss": 1.4819,
+      "step": 175920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1426858683255716e-06,
+      "loss": 1.0063,
+      "step": 175925
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.142335107937933e-06,
+      "loss": 0.9645,
+      "step": 175930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.141984398132323e-06,
+      "loss": 1.4332,
+      "step": 175935
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1416337389107468e-06,
+      "loss": 0.9867,
+      "step": 175940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1412831302752082e-06,
+      "loss": 1.1663,
+      "step": 175945
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1409325722277053e-06,
+      "loss": 1.2939,
+      "step": 175950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1405820647702438e-06,
+      "loss": 1.0908,
+      "step": 175955
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1402316079048248e-06,
+      "loss": 1.1278,
+      "step": 175960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1398812016334438e-06,
+      "loss": 1.4515,
+      "step": 175965
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1395308459581112e-06,
+      "loss": 1.3015,
+      "step": 175970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1391805408808221e-06,
+      "loss": 1.4155,
+      "step": 175975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1388302864035794e-06,
+      "loss": 1.0839,
+      "step": 175980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1384800825283782e-06,
+      "loss": 0.9978,
+      "step": 175985
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1381299292572234e-06,
+      "loss": 1.4361,
+      "step": 175990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1377798265921135e-06,
+      "loss": 1.1355,
+      "step": 175995
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1374297745350471e-06,
+      "loss": 1.2733,
+      "step": 176000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1370797730880246e-06,
+      "loss": 1.7147,
+      "step": 176005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.136729822253041e-06,
+      "loss": 1.4022,
+      "step": 176010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1363799220320982e-06,
+      "loss": 1.1436,
+      "step": 176015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1360300724271943e-06,
+      "loss": 1.2771,
+      "step": 176020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1356802734403272e-06,
+      "loss": 1.001,
+      "step": 176025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1353305250734914e-06,
+      "loss": 1.1029,
+      "step": 176030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1349808273286856e-06,
+      "loss": 1.2806,
+      "step": 176035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.134631180207909e-06,
+      "loss": 1.1759,
+      "step": 176040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.134281583713156e-06,
+      "loss": 1.3268,
+      "step": 176045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1339320378464247e-06,
+      "loss": 1.6571,
+      "step": 176050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1335825426097102e-06,
+      "loss": 0.9863,
+      "step": 176055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1332330980050055e-06,
+      "loss": 0.8708,
+      "step": 176060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1328837040343099e-06,
+      "loss": 1.2143,
+      "step": 176065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1325343606996197e-06,
+      "loss": 1.2044,
+      "step": 176070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1321850680029267e-06,
+      "loss": 1.339,
+      "step": 176075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1318358259462247e-06,
+      "loss": 1.3225,
+      "step": 176080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.13148663453151e-06,
+      "loss": 1.3316,
+      "step": 176085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.131137493760779e-06,
+      "loss": 1.3313,
+      "step": 176090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1307884036360205e-06,
+      "loss": 1.0442,
+      "step": 176095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1304393641592338e-06,
+      "loss": 1.122,
+      "step": 176100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1300903753324055e-06,
+      "loss": 1.1455,
+      "step": 176105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.129741437157532e-06,
+      "loss": 1.0848,
+      "step": 176110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1293925496366087e-06,
+      "loss": 1.2455,
+      "step": 176115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1290437127716236e-06,
+      "loss": 1.357,
+      "step": 176120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1286949265645696e-06,
+      "loss": 1.3388,
+      "step": 176125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1283461910174377e-06,
+      "loss": 1.0853,
+      "step": 176130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1279975061322234e-06,
+      "loss": 1.2385,
+      "step": 176135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1276488719109157e-06,
+      "loss": 1.2362,
+      "step": 176140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1273002883555018e-06,
+      "loss": 1.5329,
+      "step": 176145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1269517554679787e-06,
+      "loss": 1.386,
+      "step": 176150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1266032732503307e-06,
+      "loss": 1.1689,
+      "step": 176155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1262548417045505e-06,
+      "loss": 1.2275,
+      "step": 176160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1259064608326298e-06,
+      "loss": 1.359,
+      "step": 176165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1255581306365571e-06,
+      "loss": 1.2256,
+      "step": 176170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1252098511183175e-06,
+      "loss": 1.473,
+      "step": 176175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1248616222799037e-06,
+      "loss": 1.1037,
+      "step": 176180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1245134441233052e-06,
+      "loss": 1.3467,
+      "step": 176185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1241653166505062e-06,
+      "loss": 1.197,
+      "step": 176190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1238172398634994e-06,
+      "loss": 1.2682,
+      "step": 176195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1234692137642689e-06,
+      "loss": 1.1681,
+      "step": 176200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1231212383548052e-06,
+      "loss": 1.6486,
+      "step": 176205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1227733136370911e-06,
+      "loss": 1.4935,
+      "step": 176210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1224254396131186e-06,
+      "loss": 1.2229,
+      "step": 176215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.122077616284869e-06,
+      "loss": 1.1493,
+      "step": 176220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1217298436543322e-06,
+      "loss": 1.4043,
+      "step": 176225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1213821217234943e-06,
+      "loss": 1.214,
+      "step": 176230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1210344504943405e-06,
+      "loss": 1.1566,
+      "step": 176235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1206868299688533e-06,
+      "loss": 1.1618,
+      "step": 176240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1203392601490204e-06,
+      "loss": 1.0216,
+      "step": 176245
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1199917410368278e-06,
+      "loss": 1.0926,
+      "step": 176250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1196442726342571e-06,
+      "loss": 0.9905,
+      "step": 176255
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1192968549432947e-06,
+      "loss": 1.1405,
+      "step": 176260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1189494879659247e-06,
+      "loss": 1.3536,
+      "step": 176265
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1186021717041284e-06,
+      "loss": 1.2857,
+      "step": 176270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1182549061598892e-06,
+      "loss": 1.3491,
+      "step": 176275
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1179076913351938e-06,
+      "loss": 1.3084,
+      "step": 176280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1175605272320233e-06,
+      "loss": 1.6899,
+      "step": 176285
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1172134138523582e-06,
+      "loss": 1.1521,
+      "step": 176290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1168663511981814e-06,
+      "loss": 1.1292,
+      "step": 176295
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1165193392714768e-06,
+      "loss": 1.0715,
+      "step": 176300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.116172378074224e-06,
+      "loss": 1.1129,
+      "step": 176305
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1158254676084067e-06,
+      "loss": 1.0653,
+      "step": 176310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1154786078760016e-06,
+      "loss": 1.43,
+      "step": 176315
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1151317988789933e-06,
+      "loss": 1.4614,
+      "step": 176320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1147850406193628e-06,
+      "loss": 1.3771,
+      "step": 176325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1144383330990881e-06,
+      "loss": 1.2691,
+      "step": 176330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.114091676320148e-06,
+      "loss": 1.1464,
+      "step": 176335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1137450702845243e-06,
+      "loss": 1.1932,
+      "step": 176340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1133985149941972e-06,
+      "loss": 1.8268,
+      "step": 176345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1130520104511443e-06,
+      "loss": 1.0884,
+      "step": 176350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.112705556657343e-06,
+      "loss": 1.3342,
+      "step": 176355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1123591536147739e-06,
+      "loss": 1.1329,
+      "step": 176360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1120128013254128e-06,
+      "loss": 1.2924,
+      "step": 176365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1116664997912397e-06,
+      "loss": 1.3751,
+      "step": 176370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1113202490142316e-06,
+      "loss": 1.1107,
+      "step": 176375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.110974048996367e-06,
+      "loss": 1.2692,
+      "step": 176380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1106278997396202e-06,
+      "loss": 1.1508,
+      "step": 176385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1102818012459681e-06,
+      "loss": 1.288,
+      "step": 176390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1099357535173905e-06,
+      "loss": 1.2753,
+      "step": 176395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.10958975655586e-06,
+      "loss": 1.2649,
+      "step": 176400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1092438103633551e-06,
+      "loss": 1.0676,
+      "step": 176405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1088979149418476e-06,
+      "loss": 1.1458,
+      "step": 176410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1085520702933172e-06,
+      "loss": 1.2773,
+      "step": 176415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1082062764197355e-06,
+      "loss": 1.1993,
+      "step": 176420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1078605333230796e-06,
+      "loss": 1.0536,
+      "step": 176425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1075148410053216e-06,
+      "loss": 1.0439,
+      "step": 176430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1071691994684365e-06,
+      "loss": 1.0679,
+      "step": 176435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1068236087144e-06,
+      "loss": 1.5811,
+      "step": 176440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1064780687451838e-06,
+      "loss": 1.3144,
+      "step": 176445
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1061325795627587e-06,
+      "loss": 1.318,
+      "step": 176450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1057871411691035e-06,
+      "loss": 1.1413,
+      "step": 176455
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1054417535661843e-06,
+      "loss": 1.4719,
+      "step": 176460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1050964167559775e-06,
+      "loss": 1.1066,
+      "step": 176465
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1047511307404556e-06,
+      "loss": 1.1251,
+      "step": 176470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1044058955215898e-06,
+      "loss": 1.0716,
+      "step": 176475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.104060711101349e-06,
+      "loss": 1.362,
+      "step": 176480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1037155774817055e-06,
+      "loss": 1.1677,
+      "step": 176485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1033704946646329e-06,
+      "loss": 1.1947,
+      "step": 176490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1030254626520999e-06,
+      "loss": 1.0821,
+      "step": 176495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1026804814460756e-06,
+      "loss": 1.208,
+      "step": 176500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.10233555104853e-06,
+      "loss": 1.2003,
+      "step": 176505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.101990671461437e-06,
+      "loss": 1.099,
+      "step": 176510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1016458426867604e-06,
+      "loss": 1.2688,
+      "step": 176515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1013010647264743e-06,
+      "loss": 1.2076,
+      "step": 176520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1009563375825438e-06,
+      "loss": 1.5936,
+      "step": 176525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1006116612569385e-06,
+      "loss": 1.2028,
+      "step": 176530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1002670357516287e-06,
+      "loss": 1.1288,
+      "step": 176535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.099922461068582e-06,
+      "loss": 1.2684,
+      "step": 176540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0995779372097615e-06,
+      "loss": 1.4481,
+      "step": 176545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0992334641771396e-06,
+      "loss": 1.3031,
+      "step": 176550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0988890419726828e-06,
+      "loss": 1.524,
+      "step": 176555
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0985446705983583e-06,
+      "loss": 1.4166,
+      "step": 176560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0982003500561277e-06,
+      "loss": 1.3363,
+      "step": 176565
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0978560803479643e-06,
+      "loss": 1.1917,
+      "step": 176570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0975118614758284e-06,
+      "loss": 1.2777,
+      "step": 176575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0971676934416875e-06,
+      "loss": 1.0058,
+      "step": 176580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0968235762475098e-06,
+      "loss": 1.2828,
+      "step": 176585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0964795098952585e-06,
+      "loss": 1.2428,
+      "step": 176590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0961354943868952e-06,
+      "loss": 1.0267,
+      "step": 176595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0957915297243881e-06,
+      "loss": 1.1801,
+      "step": 176600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0954476159097017e-06,
+      "loss": 1.4994,
+      "step": 176605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0951037529447972e-06,
+      "loss": 1.4178,
+      "step": 176610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0947599408316411e-06,
+      "loss": 1.1206,
+      "step": 176615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0944161795721942e-06,
+      "loss": 1.4194,
+      "step": 176620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0940724691684202e-06,
+      "loss": 1.0373,
+      "step": 176625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0937288096222853e-06,
+      "loss": 1.2201,
+      "step": 176630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0933852009357494e-06,
+      "loss": 0.9791,
+      "step": 176635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0930416431107726e-06,
+      "loss": 1.0087,
+      "step": 176640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0926981361493162e-06,
+      "loss": 1.0448,
+      "step": 176645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.092354680053348e-06,
+      "loss": 1.1218,
+      "step": 176650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0920112748248258e-06,
+      "loss": 1.2023,
+      "step": 176655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0916679204657078e-06,
+      "loss": 1.276,
+      "step": 176660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.091324616977959e-06,
+      "loss": 1.2034,
+      "step": 176665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.090981364363537e-06,
+      "loss": 1.2741,
+      "step": 176670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0906381626244034e-06,
+      "loss": 0.9221,
+      "step": 176675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.090295011762519e-06,
+      "loss": 0.8808,
+      "step": 176680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.089951911779843e-06,
+      "loss": 1.0688,
+      "step": 176685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0896088626783307e-06,
+      "loss": 1.5925,
+      "step": 176690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0892658644599452e-06,
+      "loss": 1.1363,
+      "step": 176695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0889229171266457e-06,
+      "loss": 1.145,
+      "step": 176700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0885800206803898e-06,
+      "loss": 1.1688,
+      "step": 176705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0882371751231324e-06,
+      "loss": 1.3839,
+      "step": 176710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0878943804568342e-06,
+      "loss": 1.3207,
+      "step": 176715
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0875516366834549e-06,
+      "loss": 1.449,
+      "step": 176720
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0872089438049471e-06,
+      "loss": 0.9956,
+      "step": 176725
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0868663018232717e-06,
+      "loss": 1.3248,
+      "step": 176730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0865237107403836e-06,
+      "loss": 1.1915,
+      "step": 176735
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.086181170558236e-06,
+      "loss": 1.1884,
+      "step": 176740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0858386812787912e-06,
+      "loss": 1.0112,
+      "step": 176745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0854962429040028e-06,
+      "loss": 1.2692,
+      "step": 176750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0851538554358232e-06,
+      "loss": 1.4934,
+      "step": 176755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.084811518876212e-06,
+      "loss": 1.9673,
+      "step": 176760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.08446923322712e-06,
+      "loss": 1.4856,
+      "step": 176765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0841269984905045e-06,
+      "loss": 1.342,
+      "step": 176770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0837848146683205e-06,
+      "loss": 1.028,
+      "step": 176775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.083442681762521e-06,
+      "loss": 1.7935,
+      "step": 176780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0831005997750576e-06,
+      "loss": 1.201,
+      "step": 176785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0827585687078867e-06,
+      "loss": 1.1527,
+      "step": 176790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0824165885629612e-06,
+      "loss": 1.5775,
+      "step": 176795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0820746593422338e-06,
+      "loss": 1.0941,
+      "step": 176800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0817327810476553e-06,
+      "loss": 1.1066,
+      "step": 176805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0813909536811784e-06,
+      "loss": 1.1245,
+      "step": 176810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0810491772447585e-06,
+      "loss": 1.1135,
+      "step": 176815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0807074517403427e-06,
+      "loss": 1.0232,
+      "step": 176820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0803657771698861e-06,
+      "loss": 1.2809,
+      "step": 176825
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0800241535353372e-06,
+      "loss": 1.2204,
+      "step": 176830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0796825808386468e-06,
+      "loss": 1.3122,
+      "step": 176835
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.07934105908177e-06,
+      "loss": 1.5934,
+      "step": 176840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0789995882666538e-06,
+      "loss": 1.1351,
+      "step": 176845
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0786581683952468e-06,
+      "loss": 1.285,
+      "step": 176850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0783167994694987e-06,
+      "loss": 1.0621,
+      "step": 176855
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0779754814913613e-06,
+      "loss": 1.6363,
+      "step": 176860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0776342144627827e-06,
+      "loss": 1.0806,
+      "step": 176865
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0772929983857118e-06,
+      "loss": 1.2664,
+      "step": 176870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0769518332620977e-06,
+      "loss": 0.9336,
+      "step": 176875
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0766107190938858e-06,
+      "loss": 1.0238,
+      "step": 176880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0762696558830267e-06,
+      "loss": 1.255,
+      "step": 176885
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0759286436314688e-06,
+      "loss": 1.4014,
+      "step": 176890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.075587682341158e-06,
+      "loss": 1.2419,
+      "step": 176895
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0752467720140404e-06,
+      "loss": 1.2404,
+      "step": 176900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0749059126520632e-06,
+      "loss": 1.3042,
+      "step": 176905
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0745651042571769e-06,
+      "loss": 1.2629,
+      "step": 176910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0742243468313218e-06,
+      "loss": 1.1111,
+      "step": 176915
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0738836403764474e-06,
+      "loss": 1.1468,
+      "step": 176920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0735429848944978e-06,
+      "loss": 1.3189,
+      "step": 176925
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0732023803874203e-06,
+      "loss": 1.2955,
+      "step": 176930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0728618268571566e-06,
+      "loss": 1.2476,
+      "step": 176935
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0725213243056554e-06,
+      "loss": 1.1456,
+      "step": 176940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0721808727348582e-06,
+      "loss": 1.364,
+      "step": 176945
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0718404721467069e-06,
+      "loss": 1.0844,
+      "step": 176950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.071500122543152e-06,
+      "loss": 1.1332,
+      "step": 176955
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0711598239261345e-06,
+      "loss": 1.1219,
+      "step": 176960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.070819576297595e-06,
+      "loss": 1.1024,
+      "step": 176965
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0704793796594793e-06,
+      "loss": 1.7383,
+      "step": 176970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0701392340137274e-06,
+      "loss": 1.2758,
+      "step": 176975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.069799139362283e-06,
+      "loss": 1.3045,
+      "step": 176980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0694590957070915e-06,
+      "loss": 1.2635,
+      "step": 176985
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0691191030500914e-06,
+      "loss": 1.1567,
+      "step": 176990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.068779161393223e-06,
+      "loss": 1.4139,
+      "step": 176995
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0684392707384285e-06,
+      "loss": 1.1944,
+      "step": 177000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0680994310876524e-06,
+      "loss": 1.2079,
+      "step": 177005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0677596424428327e-06,
+      "loss": 1.286,
+      "step": 177010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0674199048059075e-06,
+      "loss": 1.3217,
+      "step": 177015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0670802181788188e-06,
+      "loss": 1.1364,
+      "step": 177020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.066740582563508e-06,
+      "loss": 1.1036,
+      "step": 177025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0664009979619128e-06,
+      "loss": 1.2954,
+      "step": 177030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0660614643759736e-06,
+      "loss": 0.8919,
+      "step": 177035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0657219818076292e-06,
+      "loss": 1.408,
+      "step": 177040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0653825502588133e-06,
+      "loss": 1.3699,
+      "step": 177045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0650431697314722e-06,
+      "loss": 1.1831,
+      "step": 177050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0647038402275411e-06,
+      "loss": 1.0493,
+      "step": 177055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.064364561748955e-06,
+      "loss": 1.105,
+      "step": 177060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0640253342976547e-06,
+      "loss": 1.2435,
+      "step": 177065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.063686157875574e-06,
+      "loss": 1.1749,
+      "step": 177070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.063347032484654e-06,
+      "loss": 1.6983,
+      "step": 177075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0630079581268272e-06,
+      "loss": 1.3322,
+      "step": 177080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.062668934804033e-06,
+      "loss": 1.1977,
+      "step": 177085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0623299625182048e-06,
+      "loss": 1.3312,
+      "step": 177090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0619910412712796e-06,
+      "loss": 1.3101,
+      "step": 177095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.061652171065195e-06,
+      "loss": 1.3059,
+      "step": 177100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0613133519018836e-06,
+      "loss": 1.6839,
+      "step": 177105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0609745837832786e-06,
+      "loss": 1.328,
+      "step": 177110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.060635866711317e-06,
+      "loss": 1.5017,
+      "step": 177115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0602972006879352e-06,
+      "loss": 1.2603,
+      "step": 177120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0599585857150619e-06,
+      "loss": 1.4738,
+      "step": 177125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0596200217946351e-06,
+      "loss": 1.1484,
+      "step": 177130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0592815089285857e-06,
+      "loss": 1.2491,
+      "step": 177135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0589430471188488e-06,
+      "loss": 1.1023,
+      "step": 177140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0586046363673542e-06,
+      "loss": 1.8016,
+      "step": 177145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0582662766760377e-06,
+      "loss": 1.3073,
+      "step": 177150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0579279680468313e-06,
+      "loss": 1.1251,
+      "step": 177155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0575897104816636e-06,
+      "loss": 1.1331,
+      "step": 177160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0572515039824671e-06,
+      "loss": 1.3485,
+      "step": 177165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0569133485511774e-06,
+      "loss": 1.1421,
+      "step": 177170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0565752441897203e-06,
+      "loss": 0.9878,
+      "step": 177175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0562371909000313e-06,
+      "loss": 1.2369,
+      "step": 177180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0558991886840354e-06,
+      "loss": 1.2589,
+      "step": 177185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0555612375436664e-06,
+      "loss": 1.2524,
+      "step": 177190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0552233374808552e-06,
+      "loss": 1.2548,
+      "step": 177195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0548854884975301e-06,
+      "loss": 1.238,
+      "step": 177200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0545476905956175e-06,
+      "loss": 1.1442,
+      "step": 177205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0542099437770491e-06,
+      "loss": 1.5232,
+      "step": 177210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0538722480437557e-06,
+      "loss": 1.1938,
+      "step": 177215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0535346033976636e-06,
+      "loss": 1.1829,
+      "step": 177220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0531970098406986e-06,
+      "loss": 1.1896,
+      "step": 177225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0528594673747904e-06,
+      "loss": 1.4333,
+      "step": 177230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.052521976001869e-06,
+      "loss": 1.143,
+      "step": 177235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.052184535723858e-06,
+      "loss": 1.2092,
+      "step": 177240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0518471465426872e-06,
+      "loss": 1.1223,
+      "step": 177245
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0515098084602826e-06,
+      "loss": 1.1543,
+      "step": 177250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0511725214785672e-06,
+      "loss": 1.158,
+      "step": 177255
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0508352855994708e-06,
+      "loss": 1.1061,
+      "step": 177260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0504981008249193e-06,
+      "loss": 1.052,
+      "step": 177265
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0501609671568359e-06,
+      "loss": 1.0798,
+      "step": 177270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0498238845971486e-06,
+      "loss": 0.8294,
+      "step": 177275
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0494868531477786e-06,
+      "loss": 1.2243,
+      "step": 177280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0491498728106553e-06,
+      "loss": 1.3155,
+      "step": 177285
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0488129435876982e-06,
+      "loss": 1.1292,
+      "step": 177290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0484760654808357e-06,
+      "loss": 1.1652,
+      "step": 177295
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0481392384919874e-06,
+      "loss": 1.0927,
+      "step": 177300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0478024626230787e-06,
+      "loss": 1.0066,
+      "step": 177305
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0474657378760356e-06,
+      "loss": 1.308,
+      "step": 177310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0471290642527776e-06,
+      "loss": 1.0619,
+      "step": 177315
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0467924417552267e-06,
+      "loss": 1.2813,
+      "step": 177320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0464558703853056e-06,
+      "loss": 1.3197,
+      "step": 177325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0461193501449395e-06,
+      "loss": 0.8348,
+      "step": 177330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.045782881036046e-06,
+      "loss": 1.0706,
+      "step": 177335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0454464630605498e-06,
+      "loss": 1.3162,
+      "step": 177340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0451100962203708e-06,
+      "loss": 1.2737,
+      "step": 177345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0447737805174252e-06,
+      "loss": 1.2031,
+      "step": 177350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0444375159536424e-06,
+      "loss": 1.227,
+      "step": 177355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0441013025309378e-06,
+      "loss": 1.0203,
+      "step": 177360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0437651402512317e-06,
+      "loss": 1.405,
+      "step": 177365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0434290291164418e-06,
+      "loss": 1.3987,
+      "step": 177370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0430929691284896e-06,
+      "loss": 1.2076,
+      "step": 177375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0427569602892962e-06,
+      "loss": 1.4041,
+      "step": 177380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0424210026007765e-06,
+      "loss": 1.0294,
+      "step": 177385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0420850960648521e-06,
+      "loss": 1.1258,
+      "step": 177390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0417492406834372e-06,
+      "loss": 1.0402,
+      "step": 177395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0414134364584538e-06,
+      "loss": 1.554,
+      "step": 177400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.041077683391819e-06,
+      "loss": 0.8852,
+      "step": 177405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0407419814854502e-06,
+      "loss": 1.2336,
+      "step": 177410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0404063307412605e-06,
+      "loss": 1.4452,
+      "step": 177415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0400707311611703e-06,
+      "loss": 1.3313,
+      "step": 177420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.039735182747097e-06,
+      "loss": 1.6367,
+      "step": 177425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0393996855009547e-06,
+      "loss": 0.9542,
+      "step": 177430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0390642394246586e-06,
+      "loss": 1.1812,
+      "step": 177435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.038728844520127e-06,
+      "loss": 1.3049,
+      "step": 177440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.038393500789271e-06,
+      "loss": 1.2454,
+      "step": 177445
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0380582082340084e-06,
+      "loss": 0.9766,
+      "step": 177450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.037722966856256e-06,
+      "loss": 1.1925,
+      "step": 177455
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0373877766579254e-06,
+      "loss": 1.1254,
+      "step": 177460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0370526376409295e-06,
+      "loss": 1.3777,
+      "step": 177465
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0367175498071836e-06,
+      "loss": 1.1357,
+      "step": 177470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0363825131586037e-06,
+      "loss": 1.1566,
+      "step": 177475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0360475276970982e-06,
+      "loss": 1.0904,
+      "step": 177480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.035712593424585e-06,
+      "loss": 1.1321,
+      "step": 177485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.035377710342973e-06,
+      "loss": 1.052,
+      "step": 177490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0350428784541743e-06,
+      "loss": 1.4669,
+      "step": 177495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0347080977601054e-06,
+      "loss": 1.261,
+      "step": 177500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0343733682626755e-06,
+      "loss": 1.1867,
+      "step": 177505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0340386899637943e-06,
+      "loss": 1.4531,
+      "step": 177510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0337040628653738e-06,
+      "loss": 1.142,
+      "step": 177515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0333694869693278e-06,
+      "loss": 1.1705,
+      "step": 177520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0330349622775647e-06,
+      "loss": 1.1933,
+      "step": 177525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0327004887919923e-06,
+      "loss": 1.7828,
+      "step": 177530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.032366066514524e-06,
+      "loss": 1.2326,
+      "step": 177535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.032031695447071e-06,
+      "loss": 0.9821,
+      "step": 177540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0316973755915372e-06,
+      "loss": 1.2248,
+      "step": 177545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0313631069498386e-06,
+      "loss": 1.2585,
+      "step": 177550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0310288895238795e-06,
+      "loss": 1.3132,
+      "step": 177555
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0306947233155674e-06,
+      "loss": 1.146,
+      "step": 177560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0303606083268126e-06,
+      "loss": 1.1939,
+      "step": 177565
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0300265445595248e-06,
+      "loss": 1.0665,
+      "step": 177570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0296925320156103e-06,
+      "loss": 1.225,
+      "step": 177575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0293585706969744e-06,
+      "loss": 1.1504,
+      "step": 177580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0290246606055266e-06,
+      "loss": 1.3068,
+      "step": 177585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.028690801743174e-06,
+      "loss": 1.382,
+      "step": 177590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0283569941118198e-06,
+      "loss": 1.416,
+      "step": 177595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0280232377133748e-06,
+      "loss": 1.2597,
+      "step": 177600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0276895325497416e-06,
+      "loss": 1.2514,
+      "step": 177605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0273558786228266e-06,
+      "loss": 1.2874,
+      "step": 177610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0270222759345373e-06,
+      "loss": 0.9907,
+      "step": 177615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0266887244867763e-06,
+      "loss": 1.3422,
+      "step": 177620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.026355224281448e-06,
+      "loss": 1.4602,
+      "step": 177625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.026021775320457e-06,
+      "loss": 1.1786,
+      "step": 177630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0256883776057114e-06,
+      "loss": 1.0594,
+      "step": 177635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0253550311391093e-06,
+      "loss": 1.3381,
+      "step": 177640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.025021735922559e-06,
+      "loss": 1.4134,
+      "step": 177645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0246884919579625e-06,
+      "loss": 1.1457,
+      "step": 177650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0243552992472195e-06,
+      "loss": 1.1311,
+      "step": 177655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0240221577922349e-06,
+      "loss": 1.1541,
+      "step": 177660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.023689067594914e-06,
+      "loss": 1.1894,
+      "step": 177665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0233560286571565e-06,
+      "loss": 0.977,
+      "step": 177670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0230230409808628e-06,
+      "loss": 1.3113,
+      "step": 177675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.022690104567935e-06,
+      "loss": 1.1413,
+      "step": 177680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0223572194202769e-06,
+      "loss": 1.4145,
+      "step": 177685
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0220243855397861e-06,
+      "loss": 1.125,
+      "step": 177690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0216916029283675e-06,
+      "loss": 1.4313,
+      "step": 177695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0213588715879163e-06,
+      "loss": 1.249,
+      "step": 177700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0210261915203345e-06,
+      "loss": 1.2457,
+      "step": 177705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.020693562727525e-06,
+      "loss": 1.0815,
+      "step": 177710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.020360985211385e-06,
+      "loss": 1.2126,
+      "step": 177715
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0200284589738118e-06,
+      "loss": 1.152,
+      "step": 177720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0196959840167054e-06,
+      "loss": 1.0912,
+      "step": 177725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0193635603419672e-06,
+      "loss": 1.4726,
+      "step": 177730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0190311879514936e-06,
+      "loss": 1.1036,
+      "step": 177735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0186988668471797e-06,
+      "loss": 1.4268,
+      "step": 177740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0183665970309287e-06,
+      "loss": 1.2713,
+      "step": 177745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.018034378504632e-06,
+      "loss": 1.218,
+      "step": 177750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0177022112701917e-06,
+      "loss": 1.2416,
+      "step": 177755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0173700953295041e-06,
+      "loss": 1.1122,
+      "step": 177760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.017038030684464e-06,
+      "loss": 1.2193,
+      "step": 177765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.016706017336967e-06,
+      "loss": 1.0582,
+      "step": 177770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0163740552889102e-06,
+      "loss": 1.1642,
+      "step": 177775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.016042144542191e-06,
+      "loss": 1.4916,
+      "step": 177780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0157102850987012e-06,
+      "loss": 1.1184,
+      "step": 177785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0153784769603402e-06,
+      "loss": 1.234,
+      "step": 177790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0150467201289981e-06,
+      "loss": 1.5381,
+      "step": 177795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0147150146065743e-06,
+      "loss": 1.3268,
+      "step": 177800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0143833603949583e-06,
+      "loss": 1.6387,
+      "step": 177805
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0140517574960485e-06,
+      "loss": 1.1119,
+      "step": 177810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0137202059117347e-06,
+      "loss": 1.2907,
+      "step": 177815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0133887056439108e-06,
+      "loss": 1.2983,
+      "step": 177820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.013057256694473e-06,
+      "loss": 1.2752,
+      "step": 177825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0127258590653121e-06,
+      "loss": 1.3688,
+      "step": 177830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.012394512758319e-06,
+      "loss": 1.1643,
+      "step": 177835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0120632177753887e-06,
+      "loss": 1.2589,
+      "step": 177840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0117319741184094e-06,
+      "loss": 1.1749,
+      "step": 177845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0114007817892746e-06,
+      "loss": 1.1656,
+      "step": 177850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0110696407898791e-06,
+      "loss": 1.2318,
+      "step": 177855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.01073855112211e-06,
+      "loss": 1.3176,
+      "step": 177860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0104075127878565e-06,
+      "loss": 1.1598,
+      "step": 177865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0100765257890111e-06,
+      "loss": 1.1224,
+      "step": 177870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0097455901274666e-06,
+      "loss": 1.0886,
+      "step": 177875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0094147058051096e-06,
+      "loss": 1.3513,
+      "step": 177880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0090838728238284e-06,
+      "loss": 1.0868,
+      "step": 177885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0087530911855137e-06,
+      "loss": 1.2596,
+      "step": 177890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0084223608920574e-06,
+      "loss": 1.1197,
+      "step": 177895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0080916819453435e-06,
+      "loss": 1.0863,
+      "step": 177900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0077610543472637e-06,
+      "loss": 1.7898,
+      "step": 177905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0074304780997035e-06,
+      "loss": 1.7497,
+      "step": 177910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0070999532045523e-06,
+      "loss": 1.3622,
+      "step": 177915
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0067694796636984e-06,
+      "loss": 1.2076,
+      "step": 177920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0064390574790284e-06,
+      "loss": 1.1406,
+      "step": 177925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.006108686652426e-06,
+      "loss": 1.2578,
+      "step": 177930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0057783671857801e-06,
+      "loss": 1.2826,
+      "step": 177935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.00544809908098e-06,
+      "loss": 1.0805,
+      "step": 177940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0051178823399088e-06,
+      "loss": 1.056,
+      "step": 177945
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0047877169644504e-06,
+      "loss": 1.1218,
+      "step": 177950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0044576029564946e-06,
+      "loss": 1.1227,
+      "step": 177955
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0041275403179217e-06,
+      "loss": 1.0739,
+      "step": 177960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0037975290506196e-06,
+      "loss": 1.2774,
+      "step": 177965
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0034675691564732e-06,
+      "loss": 1.0766,
+      "step": 177970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0031376606373667e-06,
+      "loss": 1.0415,
+      "step": 177975
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0028078034951805e-06,
+      "loss": 1.2734,
+      "step": 177980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0024779977318022e-06,
+      "loss": 1.05,
+      "step": 177985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0021482433491147e-06,
+      "loss": 1.1484,
+      "step": 177990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0018185403489977e-06,
+      "loss": 1.1579,
+      "step": 177995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0014888887333395e-06,
+      "loss": 1.1764,
+      "step": 178000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0011592885040178e-06,
+      "loss": 1.0689,
+      "step": 178005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0008297396629185e-06,
+      "loss": 1.7246,
+      "step": 178010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0005002422119192e-06,
+      "loss": 1.2219,
+      "step": 178015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.000170796152906e-06,
+      "loss": 1.2053,
+      "step": 178020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.998414014877579e-07,
+      "loss": 1.2728,
+      "step": 178025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.995120582183526e-07,
+      "loss": 1.2737,
+      "step": 178030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.99182766346578e-07,
+      "loss": 1.2216,
+      "step": 178035
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.988535258743116e-07,
+      "loss": 1.2358,
+      "step": 178040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.985243368034292e-07,
+      "loss": 1.1439,
+      "step": 178045
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.981951991358184e-07,
+      "loss": 1.255,
+      "step": 178050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.978661128733514e-07,
+      "loss": 1.5188,
+      "step": 178055
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.975370780179105e-07,
+      "loss": 1.2269,
+      "step": 178060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.97208094571378e-07,
+      "loss": 1.3346,
+      "step": 178065
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.968791625356288e-07,
+      "loss": 1.5301,
+      "step": 178070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.965502819125394e-07,
+      "loss": 1.1378,
+      "step": 178075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.962214527039915e-07,
+      "loss": 1.1922,
+      "step": 178080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.958926749118635e-07,
+      "loss": 1.4038,
+      "step": 178085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.955639485380308e-07,
+      "loss": 1.192,
+      "step": 178090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.952352735843695e-07,
+      "loss": 1.026,
+      "step": 178095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.94906650052757e-07,
+      "loss": 0.7517,
+      "step": 178100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.945780779450742e-07,
+      "loss": 1.4719,
+      "step": 178105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.942495572631916e-07,
+      "loss": 1.079,
+      "step": 178110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.9392108800899e-07,
+      "loss": 1.1855,
+      "step": 178115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.935926701843402e-07,
+      "loss": 1.088,
+      "step": 178120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.932643037911217e-07,
+      "loss": 1.1937,
+      "step": 178125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.929359888312095e-07,
+      "loss": 1.2572,
+      "step": 178130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.92607725306477e-07,
+      "loss": 1.1182,
+      "step": 178135
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.922795132187978e-07,
+      "loss": 1.143,
+      "step": 178140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.919513525700486e-07,
+      "loss": 1.1899,
+      "step": 178145
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.91623243362101e-07,
+      "loss": 1.1941,
+      "step": 178150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.912951855968312e-07,
+      "loss": 1.5734,
+      "step": 178155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.909671792761088e-07,
+      "loss": 1.2389,
+      "step": 178160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.906392244018126e-07,
+      "loss": 1.1725,
+      "step": 178165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 1.3179,
+      "step": 178170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.899834689999743e-07,
+      "loss": 1.1095,
+      "step": 178175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.896556684761815e-07,
+      "loss": 1.1916,
+      "step": 178180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.893279194063022e-07,
+      "loss": 1.2149,
+      "step": 178185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.890002217922035e-07,
+      "loss": 1.1009,
+      "step": 178190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.886725756357619e-07,
+      "loss": 1.1661,
+      "step": 178195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.883449809388467e-07,
+      "loss": 1.0994,
+      "step": 178200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.88017437703328e-07,
+      "loss": 1.0452,
+      "step": 178205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.876899459310774e-07,
+      "loss": 1.1734,
+      "step": 178210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.873625056239644e-07,
+      "loss": 1.4253,
+      "step": 178215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.870351167838576e-07,
+      "loss": 1.1303,
+      "step": 178220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.867077794126302e-07,
+      "loss": 1.1677,
+      "step": 178225
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.86380493512149e-07,
+      "loss": 1.5796,
+      "step": 178230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.86053259084283e-07,
+      "loss": 1.2095,
+      "step": 178235
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.857260761308984e-07,
+      "loss": 1.221,
+      "step": 178240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.853989446538693e-07,
+      "loss": 1.0221,
+      "step": 178245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.850718646550605e-07,
+      "loss": 1.455,
+      "step": 178250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.847448361363387e-07,
+      "loss": 1.5503,
+      "step": 178255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.844178590995746e-07,
+      "loss": 1.1402,
+      "step": 178260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.840909335466309e-07,
+      "loss": 1.262,
+      "step": 178265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.837640594793774e-07,
+      "loss": 1.5509,
+      "step": 178270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.834372368996836e-07,
+      "loss": 0.9562,
+      "step": 178275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.831104658094115e-07,
+      "loss": 1.3098,
+      "step": 178280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.827837462104272e-07,
+      "loss": 1.3183,
+      "step": 178285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.824570781045972e-07,
+      "loss": 1.2871,
+      "step": 178290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.821304614937888e-07,
+      "loss": 1.2274,
+      "step": 178295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.81803896379867e-07,
+      "loss": 1.3127,
+      "step": 178300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.814773827646928e-07,
+      "loss": 1.0951,
+      "step": 178305
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.811509206501334e-07,
+      "loss": 1.4446,
+      "step": 178310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.808245100380553e-07,
+      "loss": 1.7574,
+      "step": 178315
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.80498150930318e-07,
+      "loss": 1.4054,
+      "step": 178320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.801718433287887e-07,
+      "loss": 1.1679,
+      "step": 178325
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.798455872353307e-07,
+      "loss": 1.4556,
+      "step": 178330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.795193826518013e-07,
+      "loss": 1.6527,
+      "step": 178335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.79193229580072e-07,
+      "loss": 1.37,
+      "step": 178340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.78867128022002e-07,
+      "loss": 1.1572,
+      "step": 178345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.7854107797945e-07,
+      "loss": 1.2239,
+      "step": 178350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.782150794542822e-07,
+      "loss": 1.3729,
+      "step": 178355
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.778891324483576e-07,
+      "loss": 1.0204,
+      "step": 178360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.77563236963538e-07,
+      "loss": 1.0151,
+      "step": 178365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.772373930016877e-07,
+      "loss": 1.2348,
+      "step": 178370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.76911600564664e-07,
+      "loss": 1.1925,
+      "step": 178375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.765858596543265e-07,
+      "loss": 1.4312,
+      "step": 178380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.76260170272537e-07,
+      "loss": 1.3137,
+      "step": 178385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.759345324211566e-07,
+      "loss": 1.5257,
+      "step": 178390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.756089461020447e-07,
+      "loss": 1.1227,
+      "step": 178395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.752834113170572e-07,
+      "loss": 1.3824,
+      "step": 178400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.749579280680555e-07,
+      "loss": 1.1338,
+      "step": 178405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.746324963568998e-07,
+      "loss": 1.1073,
+      "step": 178410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.743071161854457e-07,
+      "loss": 1.3475,
+      "step": 178415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.739817875555535e-07,
+      "loss": 1.1278,
+      "step": 178420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.736565104690797e-07,
+      "loss": 1.4067,
+      "step": 178425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.733312849278797e-07,
+      "loss": 1.3789,
+      "step": 178430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.730061109338163e-07,
+      "loss": 1.4601,
+      "step": 178435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.726809884887423e-07,
+      "loss": 1.3221,
+      "step": 178440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.72355917594514e-07,
+      "loss": 1.1676,
+      "step": 178445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.720308982529901e-07,
+      "loss": 1.16,
+      "step": 178450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.717059304660249e-07,
+      "loss": 1.0344,
+      "step": 178455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.713810142354763e-07,
+      "loss": 1.3224,
+      "step": 178460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.710561495631954e-07,
+      "loss": 1.0293,
+      "step": 178465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.70731336451043e-07,
+      "loss": 1.0039,
+      "step": 178470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.704065749008684e-07,
+      "loss": 1.2354,
+      "step": 178475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.700818649145293e-07,
+      "loss": 1.1463,
+      "step": 178480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.69757206493881e-07,
+      "loss": 1.1287,
+      "step": 178485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.694325996407772e-07,
+      "loss": 1.1384,
+      "step": 178490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.691080443570667e-07,
+      "loss": 1.2201,
+      "step": 178495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.687835406446077e-07,
+      "loss": 1.254,
+      "step": 178500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.684590885052536e-07,
+      "loss": 1.3134,
+      "step": 178505
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.681346879408548e-07,
+      "loss": 1.0491,
+      "step": 178510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.678103389532656e-07,
+      "loss": 1.0286,
+      "step": 178515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.674860415443355e-07,
+      "loss": 1.0143,
+      "step": 178520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.671617957159208e-07,
+      "loss": 1.6659,
+      "step": 178525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.668376014698677e-07,
+      "loss": 1.1671,
+      "step": 178530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.665134588080327e-07,
+      "loss": 1.0102,
+      "step": 178535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.661893677322653e-07,
+      "loss": 1.3665,
+      "step": 178540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.65865328244413e-07,
+      "loss": 1.1641,
+      "step": 178545
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.655413403463288e-07,
+      "loss": 1.453,
+      "step": 178550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.652174040398642e-07,
+      "loss": 1.0636,
+      "step": 178555
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.64893519326866e-07,
+      "loss": 0.8411,
+      "step": 178560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.645696862091868e-07,
+      "loss": 1.3659,
+      "step": 178565
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.642459046886732e-07,
+      "loss": 1.2632,
+      "step": 178570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.639221747671757e-07,
+      "loss": 1.3093,
+      "step": 178575
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.63598496446544e-07,
+      "loss": 1.5949,
+      "step": 178580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.632748697286243e-07,
+      "loss": 1.1804,
+      "step": 178585
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.629512946152653e-07,
+      "loss": 1.399,
+      "step": 178590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.626277711083132e-07,
+      "loss": 1.1269,
+      "step": 178595
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.62304299209621e-07,
+      "loss": 1.1851,
+      "step": 178600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.619808789210305e-07,
+      "loss": 1.4597,
+      "step": 178605
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.616575102443892e-07,
+      "loss": 1.5743,
+      "step": 178610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.613341931815457e-07,
+      "loss": 0.9961,
+      "step": 178615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.61010927734346e-07,
+      "loss": 0.9695,
+      "step": 178620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.606877139046345e-07,
+      "loss": 1.1849,
+      "step": 178625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.603645516942595e-07,
+      "loss": 1.5872,
+      "step": 178630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.600414411050651e-07,
+      "loss": 1.2152,
+      "step": 178635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.597183821388923e-07,
+      "loss": 1.044,
+      "step": 178640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.593953747975938e-07,
+      "loss": 1.0206,
+      "step": 178645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.590724190830093e-07,
+      "loss": 1.2024,
+      "step": 178650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.587495149969828e-07,
+      "loss": 1.0684,
+      "step": 178655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.584266625413618e-07,
+      "loss": 1.0346,
+      "step": 178660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.58103861717985e-07,
+      "loss": 1.2877,
+      "step": 178665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.577811125287007e-07,
+      "loss": 0.9982,
+      "step": 178670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.574584149753475e-07,
+      "loss": 1.0377,
+      "step": 178675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.571357690597715e-07,
+      "loss": 1.4572,
+      "step": 178680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.568131747838138e-07,
+      "loss": 1.7911,
+      "step": 178685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.56490632149315e-07,
+      "loss": 1.0065,
+      "step": 178690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.561681411581214e-07,
+      "loss": 1.1229,
+      "step": 178695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.558457018120726e-07,
+      "loss": 1.3205,
+      "step": 178700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.555233141130061e-07,
+      "loss": 1.0325,
+      "step": 178705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.55200978062767e-07,
+      "loss": 1.0976,
+      "step": 178710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.54878693663197e-07,
+      "loss": 1.2595,
+      "step": 178715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.545564609161317e-07,
+      "loss": 1.3032,
+      "step": 178720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.542342798234173e-07,
+      "loss": 1.2918,
+      "step": 178725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.539121503868898e-07,
+      "loss": 1.163,
+      "step": 178730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.535900726083858e-07,
+      "loss": 1.0433,
+      "step": 178735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.532680464897515e-07,
+      "loss": 1.251,
+      "step": 178740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.529460720328232e-07,
+      "loss": 1.4886,
+      "step": 178745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.526241492394384e-07,
+      "loss": 1.3312,
+      "step": 178750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.523022781114333e-07,
+      "loss": 0.9983,
+      "step": 178755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.519804586506487e-07,
+      "loss": 1.5207,
+      "step": 178760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.516586908589254e-07,
+      "loss": 1.342,
+      "step": 178765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.51336974738094e-07,
+      "loss": 1.154,
+      "step": 178770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.510153102899988e-07,
+      "loss": 1.0132,
+      "step": 178775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.506936975164705e-07,
+      "loss": 1.2057,
+      "step": 178780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.503721364193485e-07,
+      "loss": 1.0918,
+      "step": 178785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.500506270004706e-07,
+      "loss": 1.4364,
+      "step": 178790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.497291692616717e-07,
+      "loss": 1.0802,
+      "step": 178795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.494077632047838e-07,
+      "loss": 1.2497,
+      "step": 178800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.490864088316465e-07,
+      "loss": 1.4619,
+      "step": 178805
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.487651061440962e-07,
+      "loss": 1.2021,
+      "step": 178810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.484438551439645e-07,
+      "loss": 1.1125,
+      "step": 178815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.481226558330847e-07,
+      "loss": 0.9745,
+      "step": 178820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.47801508213294e-07,
+      "loss": 1.3417,
+      "step": 178825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.474804122864268e-07,
+      "loss": 1.3151,
+      "step": 178830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.471593680543134e-07,
+      "loss": 1.1303,
+      "step": 178835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.468383755187915e-07,
+      "loss": 1.4058,
+      "step": 178840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.465174346816907e-07,
+      "loss": 1.4204,
+      "step": 178845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.461965455448441e-07,
+      "loss": 1.771,
+      "step": 178850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.458757081100844e-07,
+      "loss": 1.2435,
+      "step": 178855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.455549223792459e-07,
+      "loss": 1.3694,
+      "step": 178860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.452341883541572e-07,
+      "loss": 1.275,
+      "step": 178865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.449135060366533e-07,
+      "loss": 1.2217,
+      "step": 178870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.445928754285616e-07,
+      "loss": 1.2592,
+      "step": 178875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.442722965317141e-07,
+      "loss": 1.3768,
+      "step": 178880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.43951769347945e-07,
+      "loss": 1.1303,
+      "step": 178885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.436312938790826e-07,
+      "loss": 1.1921,
+      "step": 178890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.433108701269544e-07,
+      "loss": 1.2969,
+      "step": 178895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.429904980933924e-07,
+      "loss": 1.1954,
+      "step": 178900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.426701777802272e-07,
+      "loss": 1.2083,
+      "step": 178905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.423499091892874e-07,
+      "loss": 1.5014,
+      "step": 178910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.420296923223992e-07,
+      "loss": 1.1307,
+      "step": 178915
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.417095271813937e-07,
+      "loss": 1.4059,
+      "step": 178920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.413894137681001e-07,
+      "loss": 1.5508,
+      "step": 178925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.410693520843439e-07,
+      "loss": 1.1293,
+      "step": 178930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.407493421319547e-07,
+      "loss": 1.1641,
+      "step": 178935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.404293839127609e-07,
+      "loss": 1.3978,
+      "step": 178940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.401094774285846e-07,
+      "loss": 1.2807,
+      "step": 178945
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.397896226812575e-07,
+      "loss": 1.1947,
+      "step": 178950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.39469819672606e-07,
+      "loss": 1.092,
+      "step": 178955
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.391500684044552e-07,
+      "loss": 1.236,
+      "step": 178960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.388303688786282e-07,
+      "loss": 1.262,
+      "step": 178965
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.385107210969546e-07,
+      "loss": 1.2574,
+      "step": 178970
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.381911250612596e-07,
+      "loss": 1.2443,
+      "step": 178975
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.378715807733652e-07,
+      "loss": 1.2048,
+      "step": 178980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.375520882351007e-07,
+      "loss": 1.1701,
+      "step": 178985
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.372326474482851e-07,
+      "loss": 1.2294,
+      "step": 178990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.369132584147467e-07,
+      "loss": 1.2044,
+      "step": 178995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.365939211363096e-07,
+      "loss": 1.231,
+      "step": 179000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.362746356147956e-07,
+      "loss": 1.3766,
+      "step": 179005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.359554018520268e-07,
+      "loss": 1.596,
+      "step": 179010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.356362198498281e-07,
+      "loss": 1.2489,
+      "step": 179015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.353170896100238e-07,
+      "loss": 1.1965,
+      "step": 179020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.349980111344315e-07,
+      "loss": 1.1336,
+      "step": 179025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.346789844248794e-07,
+      "loss": 1.6642,
+      "step": 179030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.343600094831851e-07,
+      "loss": 1.4123,
+      "step": 179035
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.340410863111693e-07,
+      "loss": 1.3592,
+      "step": 179040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.337222149106562e-07,
+      "loss": 1.0648,
+      "step": 179045
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.334033952834664e-07,
+      "loss": 1.1095,
+      "step": 179050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.330846274314198e-07,
+      "loss": 1.1812,
+      "step": 179055
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.327659113563348e-07,
+      "loss": 1.189,
+      "step": 179060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.324472470600332e-07,
+      "loss": 1.3213,
+      "step": 179065
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.32128634544337e-07,
+      "loss": 1.1189,
+      "step": 179070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.318100738110614e-07,
+      "loss": 1.2344,
+      "step": 179075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.314915648620293e-07,
+      "loss": 1.2614,
+      "step": 179080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.311731076990571e-07,
+      "loss": 1.4771,
+      "step": 179085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.308547023239634e-07,
+      "loss": 0.9784,
+      "step": 179090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.305363487385687e-07,
+      "loss": 1.0366,
+      "step": 179095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.302180469446897e-07,
+      "loss": 1.086,
+      "step": 179100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.298997969441414e-07,
+      "loss": 1.2015,
+      "step": 179105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.295815987387447e-07,
+      "loss": 1.376,
+      "step": 179110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.292634523303168e-07,
+      "loss": 1.1073,
+      "step": 179115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.289453577206742e-07,
+      "loss": 1.5205,
+      "step": 179120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.286273149116299e-07,
+      "loss": 1.5263,
+      "step": 179125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.283093239050045e-07,
+      "loss": 1.1148,
+      "step": 179130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.2799138470261e-07,
+      "loss": 1.4467,
+      "step": 179135
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.276734973062651e-07,
+      "loss": 1.0481,
+      "step": 179140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.273556617177848e-07,
+      "loss": 0.9603,
+      "step": 179145
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.270378779389844e-07,
+      "loss": 1.3588,
+      "step": 179150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.267201459716746e-07,
+      "loss": 1.0407,
+      "step": 179155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.26402465817674e-07,
+      "loss": 1.1228,
+      "step": 179160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.260848374787968e-07,
+      "loss": 1.3726,
+      "step": 179165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.257672609568535e-07,
+      "loss": 1.6548,
+      "step": 179170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.254497362536629e-07,
+      "loss": 1.0768,
+      "step": 179175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.251322633710325e-07,
+      "loss": 1.2961,
+      "step": 179180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.248148423107794e-07,
+      "loss": 1.2051,
+      "step": 179185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.244974730747125e-07,
+      "loss": 1.4075,
+      "step": 179190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.241801556646491e-07,
+      "loss": 1.3189,
+      "step": 179195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.238628900823954e-07,
+      "loss": 1.248,
+      "step": 179200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.235456763297668e-07,
+      "loss": 1.2401,
+      "step": 179205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.232285144085762e-07,
+      "loss": 1.1309,
+      "step": 179210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.229114043206333e-07,
+      "loss": 1.5857,
+      "step": 179215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.225943460677456e-07,
+      "loss": 1.1604,
+      "step": 179220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.222773396517271e-07,
+      "loss": 1.3383,
+      "step": 179225
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.219603850743885e-07,
+      "loss": 1.0999,
+      "step": 179230
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.216434823375375e-07,
+      "loss": 1.1818,
+      "step": 179235
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.213266314429869e-07,
+      "loss": 1.4928,
+      "step": 179240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.210098323925442e-07,
+      "loss": 1.1437,
+      "step": 179245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.206930851880169e-07,
+      "loss": 1.3584,
+      "step": 179250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.203763898312157e-07,
+      "loss": 1.4646,
+      "step": 179255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.200597463239491e-07,
+      "loss": 1.1702,
+      "step": 179260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.197431546680268e-07,
+      "loss": 0.917,
+      "step": 179265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.19426614865252e-07,
+      "loss": 1.1311,
+      "step": 179270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.191101269174352e-07,
+      "loss": 1.1007,
+      "step": 179275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.187936908263862e-07,
+      "loss": 1.2085,
+      "step": 179280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.184773065939079e-07,
+      "loss": 1.1309,
+      "step": 179285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.1816097422181e-07,
+      "loss": 1.5168,
+      "step": 179290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.178446937118957e-07,
+      "loss": 1.2281,
+      "step": 179295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.175284650659733e-07,
+      "loss": 1.0854,
+      "step": 179300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.172122882858503e-07,
+      "loss": 1.1995,
+      "step": 179305
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.168961633733298e-07,
+      "loss": 1.3829,
+      "step": 179310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.165800903302158e-07,
+      "loss": 1.3942,
+      "step": 179315
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.162640691583158e-07,
+      "loss": 1.6848,
+      "step": 179320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.15948099859435e-07,
+      "loss": 1.4896,
+      "step": 179325
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.156321824353764e-07,
+      "loss": 0.852,
+      "step": 179330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.153163168879419e-07,
+      "loss": 1.3727,
+      "step": 179335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.150005032189402e-07,
+      "loss": 1.3504,
+      "step": 179340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.146847414301696e-07,
+      "loss": 1.1359,
+      "step": 179345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.143690315234355e-07,
+      "loss": 1.0817,
+      "step": 179350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.14053373500543e-07,
+      "loss": 1.4151,
+      "step": 179355
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.13737767363293e-07,
+      "loss": 1.0002,
+      "step": 179360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.134222131134851e-07,
+      "loss": 1.4244,
+      "step": 179365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.131067107529246e-07,
+      "loss": 1.2368,
+      "step": 179370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.127912602834132e-07,
+      "loss": 1.7969,
+      "step": 179375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.124758617067487e-07,
+      "loss": 1.5764,
+      "step": 179380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.121605150247381e-07,
+      "loss": 1.2857,
+      "step": 179385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.118452202391759e-07,
+      "loss": 1.4327,
+      "step": 179390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.115299773518682e-07,
+      "loss": 1.1315,
+      "step": 179395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.112147863646103e-07,
+      "loss": 1.0687,
+      "step": 179400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.108996472792064e-07,
+      "loss": 1.225,
+      "step": 179405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.105845600974517e-07,
+      "loss": 1.0831,
+      "step": 179410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.102695248211491e-07,
+      "loss": 1.0503,
+      "step": 179415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.099545414520972e-07,
+      "loss": 1.1809,
+      "step": 179420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.096396099920946e-07,
+      "loss": 1.3043,
+      "step": 179425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.093247304429364e-07,
+      "loss": 1.0733,
+      "step": 179430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.090099028064259e-07,
+      "loss": 1.1786,
+      "step": 179435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.086951270843569e-07,
+      "loss": 1.3172,
+      "step": 179440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.083804032785282e-07,
+      "loss": 0.9558,
+      "step": 179445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.080657313907393e-07,
+      "loss": 1.6579,
+      "step": 179450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.077511114227843e-07,
+      "loss": 1.3722,
+      "step": 179455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.074365433764599e-07,
+      "loss": 1.0061,
+      "step": 179460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.07122027253563e-07,
+      "loss": 1.3001,
+      "step": 179465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.068075630558926e-07,
+      "loss": 1.2803,
+      "step": 179470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.064931507852404e-07,
+      "loss": 1.7236,
+      "step": 179475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.061787904434027e-07,
+      "loss": 1.2246,
+      "step": 179480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.058644820321738e-07,
+      "loss": 1.1885,
+      "step": 179485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.055502255533532e-07,
+      "loss": 1.3687,
+      "step": 179490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.052360210087297e-07,
+      "loss": 1.2418,
+      "step": 179495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.049218684001026e-07,
+      "loss": 1.0679,
+      "step": 179500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.046077677292608e-07,
+      "loss": 1.8104,
+      "step": 179505
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.042937189980017e-07,
+      "loss": 1.5234,
+      "step": 179510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.039797222081182e-07,
+      "loss": 1.2324,
+      "step": 179515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.036657773614033e-07,
+      "loss": 0.9781,
+      "step": 179520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.033518844596478e-07,
+      "loss": 1.4364,
+      "step": 179525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.030380435046471e-07,
+      "loss": 1.3366,
+      "step": 179530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.027242544981906e-07,
+      "loss": 1.063,
+      "step": 179535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.024105174420739e-07,
+      "loss": 1.478,
+      "step": 179540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.02096832338083e-07,
+      "loss": 1.2794,
+      "step": 179545
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.017831991880155e-07,
+      "loss": 0.864,
+      "step": 179550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.014696179936567e-07,
+      "loss": 1.4206,
+      "step": 179555
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.011560887568005e-07,
+      "loss": 1.2451,
+      "step": 179560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.00842611479239e-07,
+      "loss": 1.1773,
+      "step": 179565
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.005291861627596e-07,
+      "loss": 1.1037,
+      "step": 179570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.002158128091521e-07,
+      "loss": 0.9643,
+      "step": 179575
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.999024914202059e-07,
+      "loss": 0.9409,
+      "step": 179580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.995892219977121e-07,
+      "loss": 1.4429,
+      "step": 179585
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.992760045434578e-07,
+      "loss": 1.0644,
+      "step": 179590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.989628390592331e-07,
+      "loss": 1.2583,
+      "step": 179595
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.986497255468241e-07,
+      "loss": 0.9767,
+      "step": 179600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.983366640080204e-07,
+      "loss": 1.3496,
+      "step": 179605
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.980236544446119e-07,
+      "loss": 1.0274,
+      "step": 179610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.977106968583838e-07,
+      "loss": 1.6626,
+      "step": 179615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.973977912511234e-07,
+      "loss": 1.3079,
+      "step": 179620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.970849376246138e-07,
+      "loss": 1.088,
+      "step": 179625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.96772135980648e-07,
+      "loss": 1.0885,
+      "step": 179630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.964593863210103e-07,
+      "loss": 1.3064,
+      "step": 179635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.961466886474845e-07,
+      "loss": 1.3369,
+      "step": 179640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.958340429618584e-07,
+      "loss": 1.1741,
+      "step": 179645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.955214492659159e-07,
+      "loss": 1.143,
+      "step": 179650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.952089075614412e-07,
+      "loss": 1.2036,
+      "step": 179655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.94896417850224e-07,
+      "loss": 1.2836,
+      "step": 179660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.94583980134045e-07,
+      "loss": 1.1971,
+      "step": 179665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.942715944146863e-07,
+      "loss": 1.6701,
+      "step": 179670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.939592606939351e-07,
+      "loss": 1.2715,
+      "step": 179675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.936469789735758e-07,
+      "loss": 1.1692,
+      "step": 179680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.933347492553901e-07,
+      "loss": 1.3277,
+      "step": 179685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.9302257154116e-07,
+      "loss": 1.6446,
+      "step": 179690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.927104458326686e-07,
+      "loss": 1.158,
+      "step": 179695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.923983721317009e-07,
+      "loss": 0.9684,
+      "step": 179700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.920863504400346e-07,
+      "loss": 1.1176,
+      "step": 179705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.917743807594559e-07,
+      "loss": 1.6064,
+      "step": 179710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.914624630917445e-07,
+      "loss": 1.1701,
+      "step": 179715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.91150597438678e-07,
+      "loss": 1.0145,
+      "step": 179720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.908387838020449e-07,
+      "loss": 1.0669,
+      "step": 179725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.905270221836216e-07,
+      "loss": 1.4038,
+      "step": 179730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.902153125851865e-07,
+      "loss": 1.3987,
+      "step": 179735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.899036550085227e-07,
+      "loss": 1.1862,
+      "step": 179740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.895920494554078e-07,
+      "loss": 1.4765,
+      "step": 179745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.892804959276225e-07,
+      "loss": 0.9753,
+      "step": 179750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.889689944269475e-07,
+      "loss": 1.1763,
+      "step": 179755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.886575449551604e-07,
+      "loss": 1.3728,
+      "step": 179760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.883461475140365e-07,
+      "loss": 1.646,
+      "step": 179765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.880348021053575e-07,
+      "loss": 1.6805,
+      "step": 179770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.877235087309022e-07,
+      "loss": 1.2326,
+      "step": 179775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.874122673924468e-07,
+      "loss": 1.2678,
+      "step": 179780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.871010780917666e-07,
+      "loss": 0.9949,
+      "step": 179785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.867899408306402e-07,
+      "loss": 1.3569,
+      "step": 179790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.864788556108472e-07,
+      "loss": 1.1521,
+      "step": 179795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.861678224341608e-07,
+      "loss": 1.1011,
+      "step": 179800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.858568413023583e-07,
+      "loss": 1.4349,
+      "step": 179805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.855459122172139e-07,
+      "loss": 1.0628,
+      "step": 179810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.85235035180505e-07,
+      "loss": 1.4172,
+      "step": 179815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.849242101940081e-07,
+      "loss": 1.3891,
+      "step": 179820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.846134372594961e-07,
+      "loss": 1.1278,
+      "step": 179825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.843027163787443e-07,
+      "loss": 1.2447,
+      "step": 179830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.839920475535257e-07,
+      "loss": 1.4837,
+      "step": 179835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.836814307856146e-07,
+      "loss": 1.189,
+      "step": 179840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.833708660767882e-07,
+      "loss": 0.9787,
+      "step": 179845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.830603534288163e-07,
+      "loss": 1.3675,
+      "step": 179850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.827498928434741e-07,
+      "loss": 1.0937,
+      "step": 179855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.824394843225326e-07,
+      "loss": 0.9757,
+      "step": 179860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.821291278677646e-07,
+      "loss": 1.2394,
+      "step": 179865
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.818188234809455e-07,
+      "loss": 0.9439,
+      "step": 179870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.81508571163846e-07,
+      "loss": 1.06,
+      "step": 179875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.811983709182348e-07,
+      "loss": 1.1978,
+      "step": 179880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.808882227458848e-07,
+      "loss": 1.2729,
+      "step": 179885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.805781266485702e-07,
+      "loss": 0.9993,
+      "step": 179890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.802680826280574e-07,
+      "loss": 1.2602,
+      "step": 179895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.799580906861205e-07,
+      "loss": 1.2854,
+      "step": 179900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.796481508245269e-07,
+      "loss": 1.3818,
+      "step": 179905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.793382630450497e-07,
+      "loss": 1.2156,
+      "step": 179910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.790284273494543e-07,
+      "loss": 1.3605,
+      "step": 179915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.787186437395135e-07,
+      "loss": 1.3961,
+      "step": 179920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.784089122169958e-07,
+      "loss": 1.1885,
+      "step": 179925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.780992327836658e-07,
+      "loss": 1.2793,
+      "step": 179930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.777896054412994e-07,
+      "loss": 1.3744,
+      "step": 179935
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.774800301916597e-07,
+      "loss": 1.1807,
+      "step": 179940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.771705070365144e-07,
+      "loss": 1.0112,
+      "step": 179945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.768610359776341e-07,
+      "loss": 1.1543,
+      "step": 179950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.765516170167832e-07,
+      "loss": 1.0349,
+      "step": 179955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.762422501557278e-07,
+      "loss": 1.2128,
+      "step": 179960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.759329353962398e-07,
+      "loss": 1.0558,
+      "step": 179965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.756236727400813e-07,
+      "loss": 1.0271,
+      "step": 179970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.753144621890175e-07,
+      "loss": 1.128,
+      "step": 179975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.750053037448158e-07,
+      "loss": 1.4226,
+      "step": 179980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.746961974092438e-07,
+      "loss": 1.2692,
+      "step": 179985
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.743871431840645e-07,
+      "loss": 1.1493,
+      "step": 179990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.740781410710408e-07,
+      "loss": 1.2534,
+      "step": 179995
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.737691910719393e-07,
+      "loss": 1.2904,
+      "step": 180000
+    },
+    {
+      "epoch": 0.87,
+      "eval_loss": 1.22108793258667,
+      "eval_runtime": 6463.9358,
+      "eval_samples_per_second": 3.575,
+      "eval_steps_per_second": 1.788,
+      "step": 180000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.734602931885261e-07,
+      "loss": 1.0966,
+      "step": 180005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.731514474225622e-07,
+      "loss": 1.5609,
+      "step": 180010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.728426537758128e-07,
+      "loss": 1.3764,
+      "step": 180015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.72533912250042e-07,
+      "loss": 1.2545,
+      "step": 180020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.722252228470085e-07,
+      "loss": 1.4091,
+      "step": 180025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.719165855684808e-07,
+      "loss": 1.0139,
+      "step": 180030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.716080004162198e-07,
+      "loss": 1.062,
+      "step": 180035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.71299467391985e-07,
+      "loss": 1.2692,
+      "step": 180040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.709909864975419e-07,
+      "loss": 1.5154,
+      "step": 180045
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.70682557734649e-07,
+      "loss": 1.2114,
+      "step": 180050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.703741811050693e-07,
+      "loss": 1.1412,
+      "step": 180055
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.700658566105624e-07,
+      "loss": 1.1809,
+      "step": 180060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.697575842528927e-07,
+      "loss": 1.2216,
+      "step": 180065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.694493640338153e-07,
+      "loss": 1.6952,
+      "step": 180070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.691411959550921e-07,
+      "loss": 1.3047,
+      "step": 180075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.688330800184874e-07,
+      "loss": 1.1708,
+      "step": 180080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.685250162257553e-07,
+      "loss": 1.1904,
+      "step": 180085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.682170045786553e-07,
+      "loss": 1.2882,
+      "step": 180090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.679090450789484e-07,
+      "loss": 1.6883,
+      "step": 180095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.676011377283944e-07,
+      "loss": 1.3467,
+      "step": 180100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.672932825287472e-07,
+      "loss": 0.9985,
+      "step": 180105
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.669854794817701e-07,
+      "loss": 1.205,
+      "step": 180110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.66677728589218e-07,
+      "loss": 1.4081,
+      "step": 180115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.663700298528455e-07,
+      "loss": 1.4894,
+      "step": 180120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.660623832744142e-07,
+      "loss": 1.3037,
+      "step": 180125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.657547888556806e-07,
+      "loss": 1.0778,
+      "step": 180130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.654472465983998e-07,
+      "loss": 1.4101,
+      "step": 180135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.651397565043263e-07,
+      "loss": 1.7207,
+      "step": 180140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.648323185752184e-07,
+      "loss": 1.3266,
+      "step": 180145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.645249328128325e-07,
+      "loss": 1.279,
+      "step": 180150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.642175992189217e-07,
+      "loss": 0.9831,
+      "step": 180155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.639103177952435e-07,
+      "loss": 1.2488,
+      "step": 180160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.636030885435498e-07,
+      "loss": 1.0696,
+      "step": 180165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.63295911465597e-07,
+      "loss": 1.2437,
+      "step": 180170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.629887865631392e-07,
+      "loss": 1.0082,
+      "step": 180175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.626817138379307e-07,
+      "loss": 1.1541,
+      "step": 180180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.623746932917221e-07,
+      "loss": 1.345,
+      "step": 180185
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.620677249262687e-07,
+      "loss": 1.1804,
+      "step": 180190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.617608087433249e-07,
+      "loss": 1.2705,
+      "step": 180195
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.614539447446424e-07,
+      "loss": 1.0496,
+      "step": 180200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.61147132931971e-07,
+      "loss": 1.2234,
+      "step": 180205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.608403733070647e-07,
+      "loss": 1.3202,
+      "step": 180210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.605336658716779e-07,
+      "loss": 1.3219,
+      "step": 180215
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.602270106275579e-07,
+      "loss": 1.1247,
+      "step": 180220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.599204075764578e-07,
+      "loss": 1.0491,
+      "step": 180225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.596138567201296e-07,
+      "loss": 1.0611,
+      "step": 180230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.593073580603195e-07,
+      "loss": 1.1444,
+      "step": 180235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.590009115987819e-07,
+      "loss": 1.3574,
+      "step": 180240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.586945173372674e-07,
+      "loss": 1.2239,
+      "step": 180245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.583881752775213e-07,
+      "loss": 1.1102,
+      "step": 180250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.580818854212969e-07,
+      "loss": 1.4118,
+      "step": 180255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.577756477703413e-07,
+      "loss": 1.6041,
+      "step": 180260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.574694623264058e-07,
+      "loss": 1.4227,
+      "step": 180265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.57163329091234e-07,
+      "loss": 1.2915,
+      "step": 180270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.568572480665794e-07,
+      "loss": 1.5694,
+      "step": 180275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.565512192541859e-07,
+      "loss": 0.952,
+      "step": 180280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.562452426558032e-07,
+      "loss": 0.8932,
+      "step": 180285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.55939318273179e-07,
+      "loss": 1.4282,
+      "step": 180290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.556334461080607e-07,
+      "loss": 1.1563,
+      "step": 180295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.553276261621912e-07,
+      "loss": 1.3598,
+      "step": 180300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.550218584373194e-07,
+      "loss": 1.0767,
+      "step": 180305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.547161429351936e-07,
+      "loss": 1.2396,
+      "step": 180310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.544104796575548e-07,
+      "loss": 1.1931,
+      "step": 180315
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.541048686061537e-07,
+      "loss": 1.1689,
+      "step": 180320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.537993097827324e-07,
+      "loss": 1.258,
+      "step": 180325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.534938031890327e-07,
+      "loss": 1.0272,
+      "step": 180330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.531883488268067e-07,
+      "loss": 1.1249,
+      "step": 180335
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.52882946697795e-07,
+      "loss": 1.1236,
+      "step": 180340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.52577596803742e-07,
+      "loss": 1.1955,
+      "step": 180345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.522722991463883e-07,
+      "loss": 1.2208,
+      "step": 180350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.519670537274805e-07,
+      "loss": 1.1468,
+      "step": 180355
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.516618605487625e-07,
+      "loss": 1.1353,
+      "step": 180360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.513567196119743e-07,
+      "loss": 1.2202,
+      "step": 180365
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.51051630918861e-07,
+      "loss": 1.255,
+      "step": 180370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.507465944711624e-07,
+      "loss": 1.4138,
+      "step": 180375
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.504416102706214e-07,
+      "loss": 0.9054,
+      "step": 180380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.501366783189813e-07,
+      "loss": 1.2902,
+      "step": 180385
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.498317986179827e-07,
+      "loss": 1.4388,
+      "step": 180390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.495269711693643e-07,
+      "loss": 1.0419,
+      "step": 180395
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.492221959748681e-07,
+      "loss": 1.0237,
+      "step": 180400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.48917473036237e-07,
+      "loss": 0.899,
+      "step": 180405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.486128023552087e-07,
+      "loss": 1.2486,
+      "step": 180410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.483081839335216e-07,
+      "loss": 1.3074,
+      "step": 180415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.480036177729189e-07,
+      "loss": 1.3671,
+      "step": 180420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.476991038751369e-07,
+      "loss": 1.4164,
+      "step": 180425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.473946422419144e-07,
+      "loss": 1.441,
+      "step": 180430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.470902328749942e-07,
+      "loss": 1.5856,
+      "step": 180435
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.467858757761104e-07,
+      "loss": 1.0901,
+      "step": 180440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.46481570947002e-07,
+      "loss": 1.2203,
+      "step": 180445
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.461773183894062e-07,
+      "loss": 1.1664,
+      "step": 180450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.458731181050628e-07,
+      "loss": 1.087,
+      "step": 180455
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.455689700957059e-07,
+      "loss": 0.8575,
+      "step": 180460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.452648743630765e-07,
+      "loss": 1.0464,
+      "step": 180465
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.449608309089053e-07,
+      "loss": 1.6076,
+      "step": 180470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.446568397349331e-07,
+      "loss": 1.0549,
+      "step": 180475
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.443529008428963e-07,
+      "loss": 1.478,
+      "step": 180480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.440490142345281e-07,
+      "loss": 1.3131,
+      "step": 180485
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.437451799115637e-07,
+      "loss": 1.3482,
+      "step": 180490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.434413978757383e-07,
+      "loss": 1.1562,
+      "step": 180495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.431376681287895e-07,
+      "loss": 1.3703,
+      "step": 180500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.428339906724492e-07,
+      "loss": 0.9179,
+      "step": 180505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.425303655084505e-07,
+      "loss": 1.1361,
+      "step": 180510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.422267926385285e-07,
+      "loss": 1.3354,
+      "step": 180515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.419232720644199e-07,
+      "loss": 1.1994,
+      "step": 180520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.41619803787852e-07,
+      "loss": 1.3867,
+      "step": 180525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.413163878105635e-07,
+      "loss": 1.0818,
+      "step": 180530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.410130241342839e-07,
+      "loss": 1.186,
+      "step": 180535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.407097127607445e-07,
+      "loss": 1.3141,
+      "step": 180540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.40406453691679e-07,
+      "loss": 1.2778,
+      "step": 180545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.401032469288217e-07,
+      "loss": 1.2516,
+      "step": 180550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.398000924739002e-07,
+      "loss": 1.1364,
+      "step": 180555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.394969903286465e-07,
+      "loss": 1.1567,
+      "step": 180560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.391939404947913e-07,
+      "loss": 1.2315,
+      "step": 180565
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.388909429740677e-07,
+      "loss": 1.3322,
+      "step": 180570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.385879977682021e-07,
+      "loss": 1.448,
+      "step": 180575
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.382851048789287e-07,
+      "loss": 1.3822,
+      "step": 180580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.379822643079727e-07,
+      "loss": 1.0515,
+      "step": 180585
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.376794760570661e-07,
+      "loss": 1.3517,
+      "step": 180590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.373767401279398e-07,
+      "loss": 1.3661,
+      "step": 180595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.370740565223201e-07,
+      "loss": 1.316,
+      "step": 180600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.367714252419345e-07,
+      "loss": 1.146,
+      "step": 180605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.364688462885117e-07,
+      "loss": 1.3615,
+      "step": 180610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.361663196637826e-07,
+      "loss": 1.2101,
+      "step": 180615
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.358638453694712e-07,
+      "loss": 1.2006,
+      "step": 180620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.355614234073073e-07,
+      "loss": 1.3395,
+      "step": 180625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.352590537790172e-07,
+      "loss": 1.2393,
+      "step": 180630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.349567364863254e-07,
+      "loss": 1.2782,
+      "step": 180635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.346544715309601e-07,
+      "loss": 1.0494,
+      "step": 180640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.343522589146491e-07,
+      "loss": 1.0993,
+      "step": 180645
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.340500986391164e-07,
+      "loss": 1.0312,
+      "step": 180650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.337479907060863e-07,
+      "loss": 1.0655,
+      "step": 180655
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.334459351172841e-07,
+      "loss": 1.3611,
+      "step": 180660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.331439318744383e-07,
+      "loss": 1.0269,
+      "step": 180665
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.328419809792698e-07,
+      "loss": 1.3611,
+      "step": 180670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.32540082433505e-07,
+      "loss": 1.0618,
+      "step": 180675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.322382362388659e-07,
+      "loss": 1.1953,
+      "step": 180680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.319364423970777e-07,
+      "loss": 1.2322,
+      "step": 180685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.316347009098647e-07,
+      "loss": 1.1036,
+      "step": 180690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.313330117789486e-07,
+      "loss": 1.1366,
+      "step": 180695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.310313750060527e-07,
+      "loss": 1.256,
+      "step": 180700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.307297905928958e-07,
+      "loss": 1.6646,
+      "step": 180705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.304282585412071e-07,
+      "loss": 1.3629,
+      "step": 180710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.301267788527045e-07,
+      "loss": 1.3739,
+      "step": 180715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.298253515291077e-07,
+      "loss": 1.3,
+      "step": 180720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.29523976572143e-07,
+      "loss": 1.0234,
+      "step": 180725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.292226539835269e-07,
+      "loss": 1.0853,
+      "step": 180730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.289213837649823e-07,
+      "loss": 1.6027,
+      "step": 180735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.286201659182303e-07,
+      "loss": 1.4111,
+      "step": 180740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.283190004449903e-07,
+      "loss": 1.4661,
+      "step": 180745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.280178873469802e-07,
+      "loss": 1.0544,
+      "step": 180750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.277168266259216e-07,
+      "loss": 1.1214,
+      "step": 180755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.274158182835335e-07,
+      "loss": 0.9792,
+      "step": 180760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.271148623215341e-07,
+      "loss": 1.1666,
+      "step": 180765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.268139587416446e-07,
+      "loss": 1.3253,
+      "step": 180770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.265131075455779e-07,
+      "loss": 1.3117,
+      "step": 180775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.262123087350581e-07,
+      "loss": 1.6249,
+      "step": 180780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.259115623117986e-07,
+      "loss": 1.6547,
+      "step": 180785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.256108682775199e-07,
+      "loss": 1.3554,
+      "step": 180790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.253102266339352e-07,
+      "loss": 1.1362,
+      "step": 180795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.250096373827643e-07,
+      "loss": 1.4134,
+      "step": 180800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.247091005257257e-07,
+      "loss": 1.0932,
+      "step": 180805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.244086160645314e-07,
+      "loss": 1.1052,
+      "step": 180810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.241081840008991e-07,
+      "loss": 1.1232,
+      "step": 180815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.238078043365449e-07,
+      "loss": 1.0547,
+      "step": 180820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.235074770731821e-07,
+      "loss": 1.1919,
+      "step": 180825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.23207202212527e-07,
+      "loss": 1.2739,
+      "step": 180830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.22906979756296e-07,
+      "loss": 1.1912,
+      "step": 180835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.226068097062023e-07,
+      "loss": 1.2792,
+      "step": 180840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.223066920639589e-07,
+      "loss": 0.9971,
+      "step": 180845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.220066268312799e-07,
+      "loss": 1.5594,
+      "step": 180850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.217066140098806e-07,
+      "loss": 1.1308,
+      "step": 180855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.214066536014742e-07,
+      "loss": 1.1836,
+      "step": 180860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.211067456077704e-07,
+      "loss": 1.0219,
+      "step": 180865
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.208068900304844e-07,
+      "loss": 1.3606,
+      "step": 180870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.205070868713283e-07,
+      "loss": 1.1853,
+      "step": 180875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.202073361320139e-07,
+      "loss": 1.4213,
+      "step": 180880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.199076378142535e-07,
+      "loss": 1.4589,
+      "step": 180885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.196079919197574e-07,
+      "loss": 1.3133,
+      "step": 180890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.19308398450237e-07,
+      "loss": 1.2649,
+      "step": 180895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.190088574074051e-07,
+      "loss": 1.3541,
+      "step": 180900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.187093687929703e-07,
+      "loss": 1.1024,
+      "step": 180905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.184099326086415e-07,
+      "loss": 1.2512,
+      "step": 180910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.181105488561303e-07,
+      "loss": 0.9586,
+      "step": 180915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.178112175371489e-07,
+      "loss": 0.9186,
+      "step": 180920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.175119386534037e-07,
+      "loss": 1.4826,
+      "step": 180925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.172127122066021e-07,
+      "loss": 1.7717,
+      "step": 180930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.169135381984572e-07,
+      "loss": 1.0127,
+      "step": 180935
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.166144166306722e-07,
+      "loss": 1.3194,
+      "step": 180940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.16315347504959e-07,
+      "loss": 1.4578,
+      "step": 180945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.160163308230262e-07,
+      "loss": 1.3531,
+      "step": 180950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.157173665865802e-07,
+      "loss": 1.085,
+      "step": 180955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.154184547973265e-07,
+      "loss": 1.0986,
+      "step": 180960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.151195954569723e-07,
+      "loss": 1.4579,
+      "step": 180965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.148207885672277e-07,
+      "loss": 1.193,
+      "step": 180970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.145220341297943e-07,
+      "loss": 1.0803,
+      "step": 180975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.142233321463833e-07,
+      "loss": 1.1323,
+      "step": 180980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.139246826186942e-07,
+      "loss": 0.9934,
+      "step": 180985
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.13626085548438e-07,
+      "loss": 1.0179,
+      "step": 180990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.133275409373176e-07,
+      "loss": 1.1809,
+      "step": 180995
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.130290487870396e-07,
+      "loss": 1.1224,
+      "step": 181000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.127306090993059e-07,
+      "loss": 1.1123,
+      "step": 181005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.124322218758184e-07,
+      "loss": 1.5107,
+      "step": 181010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.121338871182882e-07,
+      "loss": 0.9723,
+      "step": 181015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.11835604828416e-07,
+      "loss": 1.168,
+      "step": 181020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.115373750079014e-07,
+      "loss": 0.9773,
+      "step": 181025
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.112391976584533e-07,
+      "loss": 1.4921,
+      "step": 181030
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.10941072781769e-07,
+      "loss": 0.9599,
+      "step": 181035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.10643000379554e-07,
+      "loss": 1.1912,
+      "step": 181040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.103449804535113e-07,
+      "loss": 1.4488,
+      "step": 181045
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.100470130053418e-07,
+      "loss": 1.1286,
+      "step": 181050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.097490980367439e-07,
+      "loss": 1.3782,
+      "step": 181055
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.094512355494233e-07,
+      "loss": 1.4006,
+      "step": 181060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.091534255450795e-07,
+      "loss": 1.5045,
+      "step": 181065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.088556680254133e-07,
+      "loss": 0.9966,
+      "step": 181070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.085579629921236e-07,
+      "loss": 1.2011,
+      "step": 181075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.082603104469122e-07,
+      "loss": 1.2275,
+      "step": 181080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.079627103914789e-07,
+      "loss": 1.3373,
+      "step": 181085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.076651628275212e-07,
+      "loss": 1.5177,
+      "step": 181090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.07367667756741e-07,
+      "loss": 1.404,
+      "step": 181095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.070702251808338e-07,
+      "loss": 1.2256,
+      "step": 181100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.067728351015003e-07,
+      "loss": 1.1615,
+      "step": 181105
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.064754975204403e-07,
+      "loss": 1.2464,
+      "step": 181110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.061782124393492e-07,
+      "loss": 1.1809,
+      "step": 181115
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.058809798599243e-07,
+      "loss": 0.9551,
+      "step": 181120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.055837997838656e-07,
+      "loss": 1.1199,
+      "step": 181125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.052866722128672e-07,
+      "loss": 1.29,
+      "step": 181130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.049895971486277e-07,
+      "loss": 1.5811,
+      "step": 181135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.046925745928447e-07,
+      "loss": 0.9742,
+      "step": 181140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.043956045472123e-07,
+      "loss": 1.1931,
+      "step": 181145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.040986870134249e-07,
+      "loss": 1.3425,
+      "step": 181150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.038018219931809e-07,
+      "loss": 1.2582,
+      "step": 181155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.035050094881758e-07,
+      "loss": 1.4814,
+      "step": 181160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.032082495001037e-07,
+      "loss": 1.1424,
+      "step": 181165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.029115420306577e-07,
+      "loss": 1.2139,
+      "step": 181170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.026148870815331e-07,
+      "loss": 1.3193,
+      "step": 181175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.023182846544264e-07,
+      "loss": 1.37,
+      "step": 181180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.020217347510273e-07,
+      "loss": 1.6075,
+      "step": 181185
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.017252373730322e-07,
+      "loss": 1.05,
+      "step": 181190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.01428792522132e-07,
+      "loss": 1.4517,
+      "step": 181195
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.01132400200022e-07,
+      "loss": 1.4181,
+      "step": 181200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.008360604083943e-07,
+      "loss": 1.312,
+      "step": 181205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.005397731489406e-07,
+      "loss": 1.1521,
+      "step": 181210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.002435384233531e-07,
+      "loss": 1.2597,
+      "step": 181215
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.999473562333204e-07,
+      "loss": 1.4443,
+      "step": 181220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.996512265805378e-07,
+      "loss": 1.4697,
+      "step": 181225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.993551494666963e-07,
+      "loss": 1.0699,
+      "step": 181230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.990591248934831e-07,
+      "loss": 1.4912,
+      "step": 181235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.987631528625938e-07,
+      "loss": 1.4277,
+      "step": 181240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.984672333757137e-07,
+      "loss": 0.9257,
+      "step": 181245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.981713664345347e-07,
+      "loss": 1.0595,
+      "step": 181250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.978755520407477e-07,
+      "loss": 1.1366,
+      "step": 181255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.975797901960414e-07,
+      "loss": 1.0474,
+      "step": 181260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.972840809021021e-07,
+      "loss": 1.4283,
+      "step": 181265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.969884241606207e-07,
+      "loss": 1.1134,
+      "step": 181270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.96692819973287e-07,
+      "loss": 1.2298,
+      "step": 181275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.963972683417854e-07,
+      "loss": 1.5322,
+      "step": 181280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.961017692678064e-07,
+      "loss": 1.3517,
+      "step": 181285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.958063227530355e-07,
+      "loss": 1.1323,
+      "step": 181290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.955109287991636e-07,
+      "loss": 1.0729,
+      "step": 181295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.952155874078727e-07,
+      "loss": 1.5825,
+      "step": 181300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.949202985808524e-07,
+      "loss": 1.2402,
+      "step": 181305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.946250623197893e-07,
+      "loss": 1.2765,
+      "step": 181310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.943298786263643e-07,
+      "loss": 1.459,
+      "step": 181315
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.940347475022703e-07,
+      "loss": 1.3264,
+      "step": 181320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.937396689491905e-07,
+      "loss": 1.4102,
+      "step": 181325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.934446429688058e-07,
+      "loss": 1.0434,
+      "step": 181330
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.93149669562806e-07,
+      "loss": 1.0329,
+      "step": 181335
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.928547487328719e-07,
+      "loss": 1.0255,
+      "step": 181340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.925598804806889e-07,
+      "loss": 1.3596,
+      "step": 181345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.922650648079433e-07,
+      "loss": 1.1609,
+      "step": 181350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.919703017163172e-07,
+      "loss": 1.2803,
+      "step": 181355
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.916755912074903e-07,
+      "loss": 1.2248,
+      "step": 181360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.913809332831491e-07,
+      "loss": 1.4316,
+      "step": 181365
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.910863279449777e-07,
+      "loss": 1.3846,
+      "step": 181370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.907917751946559e-07,
+      "loss": 1.3055,
+      "step": 181375
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.904972750338646e-07,
+      "loss": 1.3386,
+      "step": 181380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.90202827464287e-07,
+      "loss": 1.0208,
+      "step": 181385
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.899084324876072e-07,
+      "loss": 1.352,
+      "step": 181390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.896140901055016e-07,
+      "loss": 1.3963,
+      "step": 181395
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.893198003196545e-07,
+      "loss": 1.0452,
+      "step": 181400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.890255631317467e-07,
+      "loss": 1.121,
+      "step": 181405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.887313785434525e-07,
+      "loss": 1.44,
+      "step": 181410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.884372465564605e-07,
+      "loss": 1.2571,
+      "step": 181415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.88143167172446e-07,
+      "loss": 1.2353,
+      "step": 181420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.878491403930877e-07,
+      "loss": 1.2987,
+      "step": 181425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.875551662200664e-07,
+      "loss": 1.0921,
+      "step": 181430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.872612446550587e-07,
+      "loss": 1.1145,
+      "step": 181435
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.869673756997453e-07,
+      "loss": 1.2501,
+      "step": 181440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.866735593558028e-07,
+      "loss": 1.3585,
+      "step": 181445
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.863797956249108e-07,
+      "loss": 1.6905,
+      "step": 181450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.860860845087448e-07,
+      "loss": 1.5948,
+      "step": 181455
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.857924260089822e-07,
+      "loss": 1.6743,
+      "step": 181460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.854988201273028e-07,
+      "loss": 1.0857,
+      "step": 181465
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.852052668653809e-07,
+      "loss": 1.3103,
+      "step": 181470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.849117662248917e-07,
+      "loss": 1.7399,
+      "step": 181475
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.846183182075128e-07,
+      "loss": 0.9975,
+      "step": 181480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.843249228149208e-07,
+      "loss": 1.0375,
+      "step": 181485
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.840315800487886e-07,
+      "loss": 1.215,
+      "step": 181490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.837382899107948e-07,
+      "loss": 1.1161,
+      "step": 181495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.834450524026116e-07,
+      "loss": 1.1768,
+      "step": 181500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.831518675259153e-07,
+      "loss": 1.0362,
+      "step": 181505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.828587352823768e-07,
+      "loss": 1.6479,
+      "step": 181510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.825656556736749e-07,
+      "loss": 1.3609,
+      "step": 181515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.822726287014803e-07,
+      "loss": 1.3145,
+      "step": 181520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.819796543674651e-07,
+      "loss": 1.3147,
+      "step": 181525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.816867326733035e-07,
+      "loss": 1.1145,
+      "step": 181530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.813938636206719e-07,
+      "loss": 1.3148,
+      "step": 181535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.811010472112368e-07,
+      "loss": 1.2563,
+      "step": 181540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.808082834466745e-07,
+      "loss": 1.1202,
+      "step": 181545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.805155723286551e-07,
+      "loss": 1.4651,
+      "step": 181550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.802229138588491e-07,
+      "loss": 1.4628,
+      "step": 181555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.79930308038932e-07,
+      "loss": 1.2391,
+      "step": 181560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.796377548705713e-07,
+      "loss": 1.1447,
+      "step": 181565
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.793452543554358e-07,
+      "loss": 1.129,
+      "step": 181570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.790528064951997e-07,
+      "loss": 1.0007,
+      "step": 181575
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.787604112915326e-07,
+      "loss": 1.2567,
+      "step": 181580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.784680687461033e-07,
+      "loss": 1.5836,
+      "step": 181585
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.781757788605793e-07,
+      "loss": 1.1863,
+      "step": 181590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.778835416366315e-07,
+      "loss": 0.9921,
+      "step": 181595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.775913570759297e-07,
+      "loss": 1.1849,
+      "step": 181600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.772992251801415e-07,
+      "loss": 1.3316,
+      "step": 181605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.770071459509354e-07,
+      "loss": 1.282,
+      "step": 181610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.767151193899802e-07,
+      "loss": 1.1276,
+      "step": 181615
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.76423145498939e-07,
+      "loss": 1.6296,
+      "step": 181620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.761312242794861e-07,
+      "loss": 1.1933,
+      "step": 181625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.758393557332843e-07,
+      "loss": 1.2313,
+      "step": 181630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.755475398619994e-07,
+      "loss": 1.0945,
+      "step": 181635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.752557766673008e-07,
+      "loss": 1.1801,
+      "step": 181640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.749640661508528e-07,
+      "loss": 1.1683,
+      "step": 181645
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.74672408314322e-07,
+      "loss": 1.1831,
+      "step": 181650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.743808031593725e-07,
+      "loss": 1.0345,
+      "step": 181655
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.740892506876729e-07,
+      "loss": 1.2573,
+      "step": 181660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.737977509008832e-07,
+      "loss": 1.4041,
+      "step": 181665
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.735063038006718e-07,
+      "loss": 0.8953,
+      "step": 181670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.73214909388702e-07,
+      "loss": 1.3804,
+      "step": 181675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.729235676666392e-07,
+      "loss": 1.221,
+      "step": 181680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.726322786361428e-07,
+      "loss": 1.2113,
+      "step": 181685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.723410422988797e-07,
+      "loss": 1.1231,
+      "step": 181690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.72049858656514e-07,
+      "loss": 1.1415,
+      "step": 181695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.717587277107053e-07,
+      "loss": 0.9512,
+      "step": 181700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.71467649463119e-07,
+      "loss": 1.4087,
+      "step": 181705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.711766239154162e-07,
+      "loss": 1.2589,
+      "step": 181710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.708856510692553e-07,
+      "loss": 1.3321,
+      "step": 181715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.705947309263051e-07,
+      "loss": 1.2843,
+      "step": 181720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.703038634882221e-07,
+      "loss": 1.2504,
+      "step": 181725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.700130487566693e-07,
+      "loss": 1.1762,
+      "step": 181730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.697222867333043e-07,
+      "loss": 1.1573,
+      "step": 181735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.694315774197891e-07,
+      "loss": 1.2817,
+      "step": 181740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.691409208177858e-07,
+      "loss": 1.131,
+      "step": 181745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.688503169289519e-07,
+      "loss": 1.3352,
+      "step": 181750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.685597657549482e-07,
+      "loss": 1.2625,
+      "step": 181755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.682692672974324e-07,
+      "loss": 1.4151,
+      "step": 181760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.679788215580653e-07,
+      "loss": 1.1478,
+      "step": 181765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.676884285385044e-07,
+      "loss": 1.0906,
+      "step": 181770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.673980882404098e-07,
+      "loss": 1.266,
+      "step": 181775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.671078006654353e-07,
+      "loss": 1.2543,
+      "step": 181780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.66817565815241e-07,
+      "loss": 1.2539,
+      "step": 181785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.665273836914866e-07,
+      "loss": 1.1186,
+      "step": 181790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.662372542958274e-07,
+      "loss": 1.2378,
+      "step": 181795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.659471776299176e-07,
+      "loss": 1.5175,
+      "step": 181800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.656571536954182e-07,
+      "loss": 1.2282,
+      "step": 181805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.653671824939812e-07,
+      "loss": 1.2134,
+      "step": 181810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.650772640272652e-07,
+      "loss": 1.2292,
+      "step": 181815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.647873982969256e-07,
+      "loss": 1.2622,
+      "step": 181820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.644975853046177e-07,
+      "loss": 0.9542,
+      "step": 181825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.642078250519936e-07,
+      "loss": 1.1325,
+      "step": 181830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.639181175407107e-07,
+      "loss": 1.3127,
+      "step": 181835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.636284627724255e-07,
+      "loss": 1.0352,
+      "step": 181840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.633388607487869e-07,
+      "loss": 1.0955,
+      "step": 181845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.630493114714532e-07,
+      "loss": 1.346,
+      "step": 181850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.627598149420745e-07,
+      "loss": 1.3493,
+      "step": 181855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.62470371162306e-07,
+      "loss": 1.6101,
+      "step": 181860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.621809801338009e-07,
+      "loss": 1.382,
+      "step": 181865
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.618916418582112e-07,
+      "loss": 1.2172,
+      "step": 181870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.616023563371877e-07,
+      "loss": 1.0016,
+      "step": 181875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.613131235723837e-07,
+      "loss": 1.1111,
+      "step": 181880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.61023943565452e-07,
+      "loss": 1.2536,
+      "step": 181885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.607348163180428e-07,
+      "loss": 1.2567,
+      "step": 181890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.604457418318056e-07,
+      "loss": 1.1308,
+      "step": 181895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.601567201083925e-07,
+      "loss": 1.6409,
+      "step": 181900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.598677511494568e-07,
+      "loss": 1.3395,
+      "step": 181905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.595788349566435e-07,
+      "loss": 1.1931,
+      "step": 181910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.592899715316071e-07,
+      "loss": 1.3098,
+      "step": 181915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.59001160875995e-07,
+      "loss": 1.1417,
+      "step": 181920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.58712402991455e-07,
+      "loss": 1.1124,
+      "step": 181925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.584236978796378e-07,
+      "loss": 1.1531,
+      "step": 181930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.581350455421932e-07,
+      "loss": 1.1518,
+      "step": 181935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.578464459807677e-07,
+      "loss": 1.6912,
+      "step": 181940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.57557899197009e-07,
+      "loss": 1.1199,
+      "step": 181945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.572694051925655e-07,
+      "loss": 1.3248,
+      "step": 181950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.569809639690862e-07,
+      "loss": 1.1348,
+      "step": 181955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.566925755282162e-07,
+      "loss": 1.4425,
+      "step": 181960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.564042398716043e-07,
+      "loss": 1.1497,
+      "step": 181965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.561159570008947e-07,
+      "loss": 1.1902,
+      "step": 181970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.558277269177339e-07,
+      "loss": 1.1921,
+      "step": 181975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.555395496237716e-07,
+      "loss": 1.4289,
+      "step": 181980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5525142512065e-07,
+      "loss": 1.0232,
+      "step": 181985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.549633534100131e-07,
+      "loss": 1.4671,
+      "step": 181990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.546753344935088e-07,
+      "loss": 1.2133,
+      "step": 181995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.54387368372782e-07,
+      "loss": 1.5992,
+      "step": 182000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.54099455049474e-07,
+      "loss": 1.6535,
+      "step": 182005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.538115945252334e-07,
+      "loss": 0.9227,
+      "step": 182010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.535237868017009e-07,
+      "loss": 1.0753,
+      "step": 182015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.532360318805199e-07,
+      "loss": 1.1502,
+      "step": 182020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.529483297633344e-07,
+      "loss": 0.9492,
+      "step": 182025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.526606804517888e-07,
+      "loss": 1.0897,
+      "step": 182030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.523730839475252e-07,
+      "loss": 1.2463,
+      "step": 182035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.520855402521831e-07,
+      "loss": 1.172,
+      "step": 182040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.51798049367406e-07,
+      "loss": 1.2628,
+      "step": 182045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.515106112948389e-07,
+      "loss": 1.4514,
+      "step": 182050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.512232260361185e-07,
+      "loss": 0.8585,
+      "step": 182055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.50935893592889e-07,
+      "loss": 1.0784,
+      "step": 182060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.50648613966789e-07,
+      "loss": 1.754,
+      "step": 182065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.503613871594606e-07,
+      "loss": 1.0879,
+      "step": 182070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.500742131725447e-07,
+      "loss": 1.0763,
+      "step": 182075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.497870920076799e-07,
+      "loss": 1.1421,
+      "step": 182080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.495000236665051e-07,
+      "loss": 1.3604,
+      "step": 182085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.492130081506599e-07,
+      "loss": 1.5028,
+      "step": 182090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.489260454617863e-07,
+      "loss": 1.2134,
+      "step": 182095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.486391356015199e-07,
+      "loss": 1.1007,
+      "step": 182100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.483522785714992e-07,
+      "loss": 1.2685,
+      "step": 182105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.48065474373364e-07,
+      "loss": 1.3253,
+      "step": 182110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.477787230087496e-07,
+      "loss": 1.1214,
+      "step": 182115
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.474920244792949e-07,
+      "loss": 1.1599,
+      "step": 182120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.472053787866395e-07,
+      "loss": 1.1025,
+      "step": 182125
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.469187859324178e-07,
+      "loss": 1.5419,
+      "step": 182130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.46632245918264e-07,
+      "loss": 0.9955,
+      "step": 182135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.463457587458178e-07,
+      "loss": 1.183,
+      "step": 182140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.460593244167158e-07,
+      "loss": 1.3574,
+      "step": 182145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.45772942932591e-07,
+      "loss": 1.3117,
+      "step": 182150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.454866142950823e-07,
+      "loss": 1.1383,
+      "step": 182155
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.452003385058204e-07,
+      "loss": 1.3706,
+      "step": 182160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.44914115566443e-07,
+      "loss": 1.2271,
+      "step": 182165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.446279454785832e-07,
+      "loss": 1.492,
+      "step": 182170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.443418282438774e-07,
+      "loss": 1.2563,
+      "step": 182175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.440557638639567e-07,
+      "loss": 1.1529,
+      "step": 182180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.437697523404552e-07,
+      "loss": 1.219,
+      "step": 182185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.434837936750083e-07,
+      "loss": 1.0662,
+      "step": 182190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.431978878692481e-07,
+      "loss": 1.1466,
+      "step": 182195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.429120349248053e-07,
+      "loss": 1.1444,
+      "step": 182200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.426262348433133e-07,
+      "loss": 1.3796,
+      "step": 182205
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.423404876264073e-07,
+      "loss": 1.3254,
+      "step": 182210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.420547932757139e-07,
+      "loss": 1.2781,
+      "step": 182215
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.417691517928682e-07,
+      "loss": 1.229,
+      "step": 182220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.414835631795004e-07,
+      "loss": 1.6317,
+      "step": 182225
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.411980274372399e-07,
+      "loss": 1.1592,
+      "step": 182230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.409125445677178e-07,
+      "loss": 1.1688,
+      "step": 182235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.406271145725685e-07,
+      "loss": 1.0549,
+      "step": 182240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.403417374534172e-07,
+      "loss": 0.9,
+      "step": 182245
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.400564132118926e-07,
+      "loss": 1.2987,
+      "step": 182250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.397711418496279e-07,
+      "loss": 1.2888,
+      "step": 182255
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.394859233682516e-07,
+      "loss": 1.3451,
+      "step": 182260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.392007577693894e-07,
+      "loss": 1.2876,
+      "step": 182265
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.389156450546742e-07,
+      "loss": 1.1959,
+      "step": 182270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.386305852257292e-07,
+      "loss": 1.1761,
+      "step": 182275
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.383455782841853e-07,
+      "loss": 0.936,
+      "step": 182280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.380606242316724e-07,
+      "loss": 1.0804,
+      "step": 182285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.377757230698135e-07,
+      "loss": 1.35,
+      "step": 182290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.374908748002352e-07,
+      "loss": 1.1222,
+      "step": 182295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.372060794245661e-07,
+      "loss": 1.1175,
+      "step": 182300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.369213369444339e-07,
+      "loss": 1.3102,
+      "step": 182305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.366366473614639e-07,
+      "loss": 1.1393,
+      "step": 182310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.363520106772782e-07,
+      "loss": 1.2793,
+      "step": 182315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.360674268935076e-07,
+      "loss": 1.2581,
+      "step": 182320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.357828960117719e-07,
+      "loss": 1.2126,
+      "step": 182325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.354984180336989e-07,
+      "loss": 1.1056,
+      "step": 182330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.352139929609147e-07,
+      "loss": 1.0663,
+      "step": 182335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.349296207950418e-07,
+      "loss": 1.2313,
+      "step": 182340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.346453015377031e-07,
+      "loss": 1.1342,
+      "step": 182345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.343610351905218e-07,
+      "loss": 1.0083,
+      "step": 182350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.340768217551253e-07,
+      "loss": 1.1138,
+      "step": 182355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.337926612331314e-07,
+      "loss": 1.0701,
+      "step": 182360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.335085536261666e-07,
+      "loss": 1.3915,
+      "step": 182365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.332244989358517e-07,
+      "loss": 1.1207,
+      "step": 182370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.329404971638099e-07,
+      "loss": 1.0196,
+      "step": 182375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.326565483116599e-07,
+      "loss": 1.2836,
+      "step": 182380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.323726523810282e-07,
+      "loss": 1.6281,
+      "step": 182385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.320888093735323e-07,
+      "loss": 1.1793,
+      "step": 182390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.318050192907911e-07,
+      "loss": 0.8747,
+      "step": 182395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.31521282134432e-07,
+      "loss": 1.1494,
+      "step": 182400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.312375979060704e-07,
+      "loss": 1.1639,
+      "step": 182405
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.309539666073273e-07,
+      "loss": 1.2176,
+      "step": 182410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.306703882398226e-07,
+      "loss": 1.2969,
+      "step": 182415
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.303868628051747e-07,
+      "loss": 1.4803,
+      "step": 182420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.301033903050025e-07,
+      "loss": 1.5177,
+      "step": 182425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.298199707409281e-07,
+      "loss": 1.1594,
+      "step": 182430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.29536604114568e-07,
+      "loss": 1.1082,
+      "step": 182435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.292532904275384e-07,
+      "loss": 1.1742,
+      "step": 182440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.289700296814584e-07,
+      "loss": 1.2709,
+      "step": 182445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.286868218779475e-07,
+      "loss": 1.2727,
+      "step": 182450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.284036670186212e-07,
+      "loss": 1.1252,
+      "step": 182455
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.28120565105096e-07,
+      "loss": 1.0496,
+      "step": 182460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.278375161389883e-07,
+      "loss": 1.1762,
+      "step": 182465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.275545201219181e-07,
+      "loss": 1.4739,
+      "step": 182470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.272715770554961e-07,
+      "loss": 1.181,
+      "step": 182475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.269886869413434e-07,
+      "loss": 1.207,
+      "step": 182480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.267058497810708e-07,
+      "loss": 1.3656,
+      "step": 182485
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.264230655762949e-07,
+      "loss": 1.0769,
+      "step": 182490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.261403343286333e-07,
+      "loss": 1.0555,
+      "step": 182495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.258576560396979e-07,
+      "loss": 1.1993,
+      "step": 182500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.25575030711102e-07,
+      "loss": 1.0819,
+      "step": 182505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.252924583444631e-07,
+      "loss": 1.6069,
+      "step": 182510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.250099389413901e-07,
+      "loss": 1.0587,
+      "step": 182515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.247274725035014e-07,
+      "loss": 1.1175,
+      "step": 182520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.244450590324059e-07,
+      "loss": 1.1333,
+      "step": 182525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.24162698529719e-07,
+      "loss": 1.3093,
+      "step": 182530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.238803909970504e-07,
+      "loss": 0.983,
+      "step": 182535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.235981364360145e-07,
+      "loss": 1.1567,
+      "step": 182540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.233159348482244e-07,
+      "loss": 1.2414,
+      "step": 182545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.230337862352898e-07,
+      "loss": 1.2075,
+      "step": 182550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.227516905988197e-07,
+      "loss": 1.2147,
+      "step": 182555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.22469647940427e-07,
+      "loss": 1.012,
+      "step": 182560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.221876582617249e-07,
+      "loss": 1.0312,
+      "step": 182565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2190572156432e-07,
+      "loss": 1.5153,
+      "step": 182570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.216238378498241e-07,
+      "loss": 0.9746,
+      "step": 182575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.213420071198463e-07,
+      "loss": 1.1346,
+      "step": 182580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.21060229375995e-07,
+      "loss": 1.3447,
+      "step": 182585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.207785046198823e-07,
+      "loss": 0.9427,
+      "step": 182590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.204968328531159e-07,
+      "loss": 1.1678,
+      "step": 182595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.202152140773033e-07,
+      "loss": 1.1123,
+      "step": 182600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.199336482940489e-07,
+      "loss": 1.4762,
+      "step": 182605
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.19652135504969e-07,
+      "loss": 1.2408,
+      "step": 182610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.19370675711667e-07,
+      "loss": 1.304,
+      "step": 182615
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.19089268915747e-07,
+      "loss": 1.3389,
+      "step": 182620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.188079151188221e-07,
+      "loss": 1.1672,
+      "step": 182625
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.185266143224934e-07,
+      "loss": 1.2916,
+      "step": 182630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.182453665283707e-07,
+      "loss": 1.237,
+      "step": 182635
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.179641717380592e-07,
+      "loss": 1.1726,
+      "step": 182640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.176830299531656e-07,
+      "loss": 1.1628,
+      "step": 182645
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.17401941175293e-07,
+      "loss": 1.0764,
+      "step": 182650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.171209054060479e-07,
+      "loss": 1.4234,
+      "step": 182655
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.168399226470369e-07,
+      "loss": 1.0731,
+      "step": 182660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.165589928998629e-07,
+      "loss": 0.9672,
+      "step": 182665
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.162781161661281e-07,
+      "loss": 0.9428,
+      "step": 182670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.159972924474389e-07,
+      "loss": 1.1025,
+      "step": 182675
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.157165217454009e-07,
+      "loss": 1.3689,
+      "step": 182680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.154358040616127e-07,
+      "loss": 1.0904,
+      "step": 182685
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.151551393976819e-07,
+      "loss": 1.333,
+      "step": 182690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.148745277552083e-07,
+      "loss": 1.6588,
+      "step": 182695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.145939691357928e-07,
+      "loss": 1.2208,
+      "step": 182700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.14313463541042e-07,
+      "loss": 0.9006,
+      "step": 182705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.140330109725568e-07,
+      "loss": 1.3317,
+      "step": 182710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.137526114319349e-07,
+      "loss": 1.1983,
+      "step": 182715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.134722649207826e-07,
+      "loss": 1.3775,
+      "step": 182720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.131919714406965e-07,
+      "loss": 1.2581,
+      "step": 182725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.129117309932798e-07,
+      "loss": 0.9551,
+      "step": 182730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.126315435801334e-07,
+      "loss": 1.1663,
+      "step": 182735
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.123514092028561e-07,
+      "loss": 1.1525,
+      "step": 182740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.120713278630464e-07,
+      "loss": 1.0224,
+      "step": 182745
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.117912995623044e-07,
+      "loss": 1.1194,
+      "step": 182750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.11511324302232e-07,
+      "loss": 1.2215,
+      "step": 182755
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.112314020844247e-07,
+      "loss": 1.0169,
+      "step": 182760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.10951532910481e-07,
+      "loss": 1.3511,
+      "step": 182765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.10671716782001e-07,
+      "loss": 1.0713,
+      "step": 182770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.103919537005822e-07,
+      "loss": 1.2669,
+      "step": 182775
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.101122436678209e-07,
+      "loss": 1.2765,
+      "step": 182780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.098325866853162e-07,
+      "loss": 1.7422,
+      "step": 182785
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.095529827546632e-07,
+      "loss": 1.3178,
+      "step": 182790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.092734318774596e-07,
+      "loss": 1.6189,
+      "step": 182795
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.089939340553031e-07,
+      "loss": 1.4687,
+      "step": 182800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.087144892897879e-07,
+      "loss": 1.3389,
+      "step": 182805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.084350975825117e-07,
+      "loss": 1.3903,
+      "step": 182810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.081557589350652e-07,
+      "loss": 1.1143,
+      "step": 182815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.078764733490484e-07,
+      "loss": 1.2843,
+      "step": 182820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.075972408260556e-07,
+      "loss": 1.5332,
+      "step": 182825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.073180613676789e-07,
+      "loss": 1.12,
+      "step": 182830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.070389349755158e-07,
+      "loss": 1.2502,
+      "step": 182835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.067598616511572e-07,
+      "loss": 1.2766,
+      "step": 182840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.064808413961988e-07,
+      "loss": 1.2752,
+      "step": 182845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.062018742122335e-07,
+      "loss": 1.2676,
+      "step": 182850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.059229601008555e-07,
+      "loss": 1.169,
+      "step": 182855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.056440990636537e-07,
+      "loss": 1.1711,
+      "step": 182860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.053652911022235e-07,
+      "loss": 1.3496,
+      "step": 182865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.05086536218158e-07,
+      "loss": 1.1495,
+      "step": 182870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.048078344130471e-07,
+      "loss": 1.2322,
+      "step": 182875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.045291856884839e-07,
+      "loss": 1.3158,
+      "step": 182880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.042505900460561e-07,
+      "loss": 1.5798,
+      "step": 182885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.039720474873602e-07,
+      "loss": 1.2402,
+      "step": 182890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.036935580139814e-07,
+      "loss": 1.1177,
+      "step": 182895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.034151216275142e-07,
+      "loss": 1.1861,
+      "step": 182900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.031367383295462e-07,
+      "loss": 1.2014,
+      "step": 182905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.028584081216672e-07,
+      "loss": 1.0574,
+      "step": 182910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.025801310054669e-07,
+      "loss": 1.3002,
+      "step": 182915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.023019069825354e-07,
+      "loss": 1.1798,
+      "step": 182920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.020237360544601e-07,
+      "loss": 1.3199,
+      "step": 182925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.017456182228321e-07,
+      "loss": 1.1145,
+      "step": 182930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.014675534892357e-07,
+      "loss": 1.2431,
+      "step": 182935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.011895418552605e-07,
+      "loss": 1.3246,
+      "step": 182940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.009115833224955e-07,
+      "loss": 1.0387,
+      "step": 182945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.006336778925282e-07,
+      "loss": 1.2584,
+      "step": 182950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.003558255669419e-07,
+      "loss": 1.5929,
+      "step": 182955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.000780263473262e-07,
+      "loss": 1.1501,
+      "step": 182960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.998002802352677e-07,
+      "loss": 1.1182,
+      "step": 182965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.99522587232353e-07,
+      "loss": 1.1118,
+      "step": 182970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.992449473401641e-07,
+      "loss": 1.4327,
+      "step": 182975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.989673605602887e-07,
+      "loss": 1.0115,
+      "step": 182980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.986898268943143e-07,
+      "loss": 1.2692,
+      "step": 182985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.98412346343822e-07,
+      "loss": 1.2837,
+      "step": 182990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.981349189104003e-07,
+      "loss": 1.1878,
+      "step": 182995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.978575445956304e-07,
+      "loss": 1.53,
+      "step": 183000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.975802234010942e-07,
+      "loss": 1.1833,
+      "step": 183005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.973029553283817e-07,
+      "loss": 1.3347,
+      "step": 183010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.970257403790726e-07,
+      "loss": 1.2416,
+      "step": 183015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.96748578554749e-07,
+      "loss": 1.1314,
+      "step": 183020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.964714698569952e-07,
+      "loss": 1.0493,
+      "step": 183025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.961944142873934e-07,
+      "loss": 1.2217,
+      "step": 183030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.959174118475254e-07,
+      "loss": 1.3346,
+      "step": 183035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.956404625389723e-07,
+      "loss": 1.0315,
+      "step": 183040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.953635663633185e-07,
+      "loss": 1.1478,
+      "step": 183045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.950867233221415e-07,
+      "loss": 1.3022,
+      "step": 183050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.948099334170233e-07,
+      "loss": 1.2133,
+      "step": 183055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.945331966495472e-07,
+      "loss": 1.2892,
+      "step": 183060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.94256513021292e-07,
+      "loss": 1.1552,
+      "step": 183065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.939798825338351e-07,
+      "loss": 1.104,
+      "step": 183070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.937033051887587e-07,
+      "loss": 0.9807,
+      "step": 183075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.934267809876427e-07,
+      "loss": 1.0476,
+      "step": 183080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.931503099320646e-07,
+      "loss": 1.1337,
+      "step": 183085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.928738920236056e-07,
+      "loss": 1.029,
+      "step": 183090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.925975272638418e-07,
+      "loss": 1.1952,
+      "step": 183095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.923212156543491e-07,
+      "loss": 0.8565,
+      "step": 183100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.920449571967125e-07,
+      "loss": 1.0717,
+      "step": 183105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.917687518925042e-07,
+      "loss": 1.2186,
+      "step": 183110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.914925997433031e-07,
+      "loss": 1.0234,
+      "step": 183115
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.912165007506855e-07,
+      "loss": 1.3026,
+      "step": 183120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.909404549162269e-07,
+      "loss": 1.3441,
+      "step": 183125
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.90664462241507e-07,
+      "loss": 1.4741,
+      "step": 183130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.903885227280993e-07,
+      "loss": 1.1324,
+      "step": 183135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.901126363775812e-07,
+      "loss": 1.1158,
+      "step": 183140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.898368031915259e-07,
+      "loss": 1.4359,
+      "step": 183145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.8956102317151e-07,
+      "loss": 1.0378,
+      "step": 183150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.892852963191099e-07,
+      "loss": 1.31,
+      "step": 183155
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.890096226358989e-07,
+      "loss": 1.1466,
+      "step": 183160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.88734002123449e-07,
+      "loss": 1.1903,
+      "step": 183165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.884584347833356e-07,
+      "loss": 1.253,
+      "step": 183170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.881829206171353e-07,
+      "loss": 1.5152,
+      "step": 183175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.879074596264179e-07,
+      "loss": 1.2052,
+      "step": 183180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.876320518127566e-07,
+      "loss": 1.1634,
+      "step": 183185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.873566971777246e-07,
+      "loss": 1.0123,
+      "step": 183190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.870813957228962e-07,
+      "loss": 1.2534,
+      "step": 183195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.8680614744984e-07,
+      "loss": 1.2639,
+      "step": 183200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.865309523601326e-07,
+      "loss": 1.1636,
+      "step": 183205
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.862558104553418e-07,
+      "loss": 1.3065,
+      "step": 183210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.859807217370384e-07,
+      "loss": 1.2877,
+      "step": 183215
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.857056862067946e-07,
+      "loss": 1.5184,
+      "step": 183220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.854307038661833e-07,
+      "loss": 1.487,
+      "step": 183225
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.851557747167703e-07,
+      "loss": 1.4852,
+      "step": 183230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.848808987601296e-07,
+      "loss": 1.0075,
+      "step": 183235
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.846060759978268e-07,
+      "loss": 1.163,
+      "step": 183240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.843313064314338e-07,
+      "loss": 1.1928,
+      "step": 183245
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.840565900625217e-07,
+      "loss": 1.0181,
+      "step": 183250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.837819268926571e-07,
+      "loss": 1.5069,
+      "step": 183255
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.835073169234064e-07,
+      "loss": 1.2933,
+      "step": 183260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.832327601563405e-07,
+      "loss": 0.9419,
+      "step": 183265
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.829582565930281e-07,
+      "loss": 1.1581,
+      "step": 183270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.82683806235035e-07,
+      "loss": 1.0742,
+      "step": 183275
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.824094090839262e-07,
+      "loss": 1.1621,
+      "step": 183280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.821350651412729e-07,
+      "loss": 1.1943,
+      "step": 183285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.818607744086402e-07,
+      "loss": 1.3443,
+      "step": 183290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.815865368875929e-07,
+      "loss": 1.3638,
+      "step": 183295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.813123525797005e-07,
+      "loss": 1.4886,
+      "step": 183300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.810382214865263e-07,
+      "loss": 1.4289,
+      "step": 183305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.807641436096324e-07,
+      "loss": 1.6648,
+      "step": 183310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.80490118950592e-07,
+      "loss": 1.0019,
+      "step": 183315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.802161475109648e-07,
+      "loss": 1.5193,
+      "step": 183320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.799422292923164e-07,
+      "loss": 1.2205,
+      "step": 183325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.796683642962087e-07,
+      "loss": 1.077,
+      "step": 183330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.793945525242073e-07,
+      "loss": 1.2111,
+      "step": 183335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.791207939778788e-07,
+      "loss": 1.2829,
+      "step": 183340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.788470886587817e-07,
+      "loss": 1.1784,
+      "step": 183345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.785734365684826e-07,
+      "loss": 1.0669,
+      "step": 183350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.782998377085414e-07,
+      "loss": 1.4089,
+      "step": 183355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.780262920805214e-07,
+      "loss": 1.4566,
+      "step": 183360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.777527996859868e-07,
+      "loss": 1.5516,
+      "step": 183365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.774793605264973e-07,
+      "loss": 1.1086,
+      "step": 183370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.772059746036141e-07,
+      "loss": 1.3829,
+      "step": 183375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.769326419188981e-07,
+      "loss": 1.156,
+      "step": 183380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.766593624739138e-07,
+      "loss": 1.0283,
+      "step": 183385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.763861362702173e-07,
+      "loss": 1.1522,
+      "step": 183390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.761129633093721e-07,
+      "loss": 1.3604,
+      "step": 183395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.758398435929358e-07,
+      "loss": 1.1039,
+      "step": 183400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.755667771224683e-07,
+      "loss": 1.1069,
+      "step": 183405
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.752937638995305e-07,
+      "loss": 1.1784,
+      "step": 183410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.750208039256812e-07,
+      "loss": 1.3109,
+      "step": 183415
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.747478972024791e-07,
+      "loss": 0.9854,
+      "step": 183420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.744750437314795e-07,
+      "loss": 1.2407,
+      "step": 183425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.742022435142437e-07,
+      "loss": 1.1944,
+      "step": 183430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.739294965523313e-07,
+      "loss": 1.2605,
+      "step": 183435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.736568028472956e-07,
+      "loss": 1.565,
+      "step": 183440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.733841624006965e-07,
+      "loss": 1.0008,
+      "step": 183445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.731115752140893e-07,
+      "loss": 1.0716,
+      "step": 183450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.728390412890318e-07,
+      "loss": 1.2771,
+      "step": 183455
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.725665606270804e-07,
+      "loss": 1.4043,
+      "step": 183460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.722941332297917e-07,
+      "loss": 1.1788,
+      "step": 183465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.720217590987188e-07,
+      "loss": 1.2638,
+      "step": 183470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.717494382354183e-07,
+      "loss": 1.3841,
+      "step": 183475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.71477170641448e-07,
+      "loss": 1.5965,
+      "step": 183480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.712049563183598e-07,
+      "loss": 1.1853,
+      "step": 183485
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.70932795267707e-07,
+      "loss": 1.22,
+      "step": 183490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.706606874910482e-07,
+      "loss": 1.305,
+      "step": 183495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.703886329899323e-07,
+      "loss": 1.2683,
+      "step": 183500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.701166317659158e-07,
+      "loss": 1.3238,
+      "step": 183505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.698446838205519e-07,
+      "loss": 1.2001,
+      "step": 183510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.695727891553938e-07,
+      "loss": 1.3018,
+      "step": 183515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.693009477719925e-07,
+      "loss": 1.3688,
+      "step": 183520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.690291596719012e-07,
+      "loss": 1.2776,
+      "step": 183525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.68757424856673e-07,
+      "loss": 1.2564,
+      "step": 183530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.684857433278569e-07,
+      "loss": 1.2526,
+      "step": 183535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.68214115087008e-07,
+      "loss": 1.1731,
+      "step": 183540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.679425401356743e-07,
+      "loss": 1.2356,
+      "step": 183545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.676710184754098e-07,
+      "loss": 1.0826,
+      "step": 183550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.673995501077612e-07,
+      "loss": 1.1544,
+      "step": 183555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.671281350342828e-07,
+      "loss": 0.9677,
+      "step": 183560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.668567732565201e-07,
+      "loss": 1.7574,
+      "step": 183565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.665854647760262e-07,
+      "loss": 1.5863,
+      "step": 183570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.663142095943497e-07,
+      "loss": 1.5519,
+      "step": 183575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.660430077130397e-07,
+      "loss": 1.0769,
+      "step": 183580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.657718591336437e-07,
+      "loss": 1.0782,
+      "step": 183585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.655007638577093e-07,
+      "loss": 1.0864,
+      "step": 183590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.652297218867887e-07,
+      "loss": 1.2557,
+      "step": 183595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.64958733222425e-07,
+      "loss": 1.1252,
+      "step": 183600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.646877978661692e-07,
+      "loss": 1.2569,
+      "step": 183605
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.644169158195679e-07,
+      "loss": 1.0994,
+      "step": 183610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.641460870841653e-07,
+      "loss": 1.2562,
+      "step": 183615
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.638753116615094e-07,
+      "loss": 1.1507,
+      "step": 183620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.636045895531496e-07,
+      "loss": 0.9276,
+      "step": 183625
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.633339207606282e-07,
+      "loss": 1.0365,
+      "step": 183630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.630633052854907e-07,
+      "loss": 1.126,
+      "step": 183635
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.627927431292824e-07,
+      "loss": 1.0116,
+      "step": 183640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.625222342935523e-07,
+      "loss": 0.86,
+      "step": 183645
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.622517787798411e-07,
+      "loss": 1.0834,
+      "step": 183650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.619813765896954e-07,
+      "loss": 1.3735,
+      "step": 183655
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.617110277246575e-07,
+      "loss": 1.4823,
+      "step": 183660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.614407321862726e-07,
+      "loss": 1.2085,
+      "step": 183665
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.611704899760851e-07,
+      "loss": 1.1328,
+      "step": 183670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.609003010956361e-07,
+      "loss": 1.1735,
+      "step": 183675
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.606301655464686e-07,
+      "loss": 1.4949,
+      "step": 183680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.60360083330126e-07,
+      "loss": 1.3544,
+      "step": 183685
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.600900544481526e-07,
+      "loss": 1.5355,
+      "step": 183690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.598200789020881e-07,
+      "loss": 1.0557,
+      "step": 183695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.595501566934736e-07,
+      "loss": 1.1448,
+      "step": 183700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.592802878238525e-07,
+      "loss": 1.3816,
+      "step": 183705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.590104722947633e-07,
+      "loss": 1.3967,
+      "step": 183710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.587407101077481e-07,
+      "loss": 1.1883,
+      "step": 183715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.584710012643503e-07,
+      "loss": 1.17,
+      "step": 183720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.582013457661063e-07,
+      "loss": 0.9805,
+      "step": 183725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.57931743614556e-07,
+      "loss": 1.3486,
+      "step": 183730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.576621948112405e-07,
+      "loss": 1.0928,
+      "step": 183735
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.573926993576995e-07,
+      "loss": 1.1142,
+      "step": 183740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.571232572554698e-07,
+      "loss": 1.2726,
+      "step": 183745
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.568538685060933e-07,
+      "loss": 1.4842,
+      "step": 183750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.565845331111043e-07,
+      "loss": 1.4312,
+      "step": 183755
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.563152510720449e-07,
+      "loss": 0.9602,
+      "step": 183760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.560460223904497e-07,
+      "loss": 1.1598,
+      "step": 183765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.557768470678583e-07,
+      "loss": 1.3804,
+      "step": 183770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.555077251058051e-07,
+      "loss": 1.2988,
+      "step": 183775
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.5523865650583e-07,
+      "loss": 1.3513,
+      "step": 183780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.549696412694684e-07,
+      "loss": 1.0511,
+      "step": 183785
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.54700679398258e-07,
+      "loss": 1.0936,
+      "step": 183790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.544317708937297e-07,
+      "loss": 1.1906,
+      "step": 183795
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.541629157574258e-07,
+      "loss": 1.1471,
+      "step": 183800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.538941139908761e-07,
+      "loss": 0.8605,
+      "step": 183805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.53625365595617e-07,
+      "loss": 1.1022,
+      "step": 183810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.533566705731864e-07,
+      "loss": 1.5183,
+      "step": 183815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.530880289251163e-07,
+      "loss": 1.4246,
+      "step": 183820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.528194406529398e-07,
+      "loss": 1.3921,
+      "step": 183825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.525509057581902e-07,
+      "loss": 1.1342,
+      "step": 183830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.522824242424053e-07,
+      "loss": 1.1425,
+      "step": 183835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.520139961071148e-07,
+      "loss": 1.0492,
+      "step": 183840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.51745621353852e-07,
+      "loss": 1.0523,
+      "step": 183845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.514772999841478e-07,
+      "loss": 1.287,
+      "step": 183850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.51209031999539e-07,
+      "loss": 1.1991,
+      "step": 183855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.50940817401553e-07,
+      "loss": 1.2965,
+      "step": 183860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.506726561917254e-07,
+      "loss": 1.4665,
+      "step": 183865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.504045483715838e-07,
+      "loss": 1.2892,
+      "step": 183870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.501364939426602e-07,
+      "loss": 1.1698,
+      "step": 183875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.498684929064881e-07,
+      "loss": 1.1587,
+      "step": 183880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.496005452645959e-07,
+      "loss": 1.8799,
+      "step": 183885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.493326510185116e-07,
+      "loss": 1.281,
+      "step": 183890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.490648101697683e-07,
+      "loss": 1.4357,
+      "step": 183895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.487970227198947e-07,
+      "loss": 0.8898,
+      "step": 183900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.485292886704208e-07,
+      "loss": 1.3685,
+      "step": 183905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.482616080228721e-07,
+      "loss": 1.3172,
+      "step": 183910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.479939807787805e-07,
+      "loss": 1.2496,
+      "step": 183915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.477264069396727e-07,
+      "loss": 0.9828,
+      "step": 183920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.474588865070774e-07,
+      "loss": 1.0806,
+      "step": 183925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.471914194825235e-07,
+      "loss": 1.3766,
+      "step": 183930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.469240058675364e-07,
+      "loss": 1.2076,
+      "step": 183935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.466566456636425e-07,
+      "loss": 1.155,
+      "step": 183940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.463893388723696e-07,
+      "loss": 1.185,
+      "step": 183945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.461220854952466e-07,
+      "loss": 1.1847,
+      "step": 183950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.458548855337965e-07,
+      "loss": 1.302,
+      "step": 183955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.455877389895459e-07,
+      "loss": 1.2325,
+      "step": 183960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.453206458640204e-07,
+      "loss": 1.2885,
+      "step": 183965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.450536061587453e-07,
+      "loss": 1.2716,
+      "step": 183970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.447866198752473e-07,
+      "loss": 1.0194,
+      "step": 183975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.445196870150494e-07,
+      "loss": 1.1592,
+      "step": 183980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.44252807579675e-07,
+      "loss": 1.0202,
+      "step": 183985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.439859815706473e-07,
+      "loss": 1.0633,
+      "step": 183990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.437192089894939e-07,
+      "loss": 1.1932,
+      "step": 183995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.43452489837737e-07,
+      "loss": 1.0637,
+      "step": 184000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.431858241168964e-07,
+      "loss": 1.1741,
+      "step": 184005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.429192118284999e-07,
+      "loss": 1.1663,
+      "step": 184010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.426526529740651e-07,
+      "loss": 1.3496,
+      "step": 184015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.423861475551163e-07,
+      "loss": 1.1962,
+      "step": 184020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.421196955731779e-07,
+      "loss": 1.1714,
+      "step": 184025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.418532970297686e-07,
+      "loss": 1.3824,
+      "step": 184030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.415869519264084e-07,
+      "loss": 1.4924,
+      "step": 184035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.413206602646215e-07,
+      "loss": 1.1041,
+      "step": 184040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.410544220459281e-07,
+      "loss": 1.2015,
+      "step": 184045
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.407882372718477e-07,
+      "loss": 1.2418,
+      "step": 184050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.405221059438982e-07,
+      "loss": 1.042,
+      "step": 184055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.402560280636017e-07,
+      "loss": 1.0829,
+      "step": 184060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.399900036324791e-07,
+      "loss": 1.1446,
+      "step": 184065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.39724032652047e-07,
+      "loss": 1.2405,
+      "step": 184070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.394581151238255e-07,
+      "loss": 1.005,
+      "step": 184075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.391922510493343e-07,
+      "loss": 1.3998,
+      "step": 184080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.389264404300854e-07,
+      "loss": 1.1844,
+      "step": 184085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.386606832676057e-07,
+      "loss": 1.1004,
+      "step": 184090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.383949795634081e-07,
+      "loss": 1.2203,
+      "step": 184095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.381293293190082e-07,
+      "loss": 1.2712,
+      "step": 184100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.378637325359272e-07,
+      "loss": 1.236,
+      "step": 184105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.375981892156779e-07,
+      "loss": 1.2282,
+      "step": 184110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.373326993597784e-07,
+      "loss": 0.9663,
+      "step": 184115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.370672629697472e-07,
+      "loss": 1.4714,
+      "step": 184120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.368018800470976e-07,
+      "loss": 1.1535,
+      "step": 184125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.36536550593343e-07,
+      "loss": 1.1552,
+      "step": 184130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.362712746100019e-07,
+      "loss": 1.1701,
+      "step": 184135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.360060520985889e-07,
+      "loss": 1.1514,
+      "step": 184140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.357408830606182e-07,
+      "loss": 1.134,
+      "step": 184145
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.35475767497602e-07,
+      "loss": 1.0134,
+      "step": 184150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.352107054110557e-07,
+      "loss": 1.0098,
+      "step": 184155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.349456968024947e-07,
+      "loss": 1.1669,
+      "step": 184160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.346807416734302e-07,
+      "loss": 1.5033,
+      "step": 184165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.344158400253763e-07,
+      "loss": 1.3015,
+      "step": 184170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.341509918598443e-07,
+      "loss": 1.38,
+      "step": 184175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.338861971783472e-07,
+      "loss": 1.464,
+      "step": 184180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.336214559823983e-07,
+      "loss": 1.1775,
+      "step": 184185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.333567682735098e-07,
+      "loss": 1.1002,
+      "step": 184190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.330921340531915e-07,
+      "loss": 1.1316,
+      "step": 184195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.328275533229544e-07,
+      "loss": 1.2272,
+      "step": 184200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.325630260843085e-07,
+      "loss": 1.3226,
+      "step": 184205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.322985523387692e-07,
+      "loss": 1.1228,
+      "step": 184210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.320341320878409e-07,
+      "loss": 1.2334,
+      "step": 184215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.31769765333039e-07,
+      "loss": 1.4159,
+      "step": 184220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.315054520758679e-07,
+      "loss": 0.8114,
+      "step": 184225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.312411923178396e-07,
+      "loss": 1.3121,
+      "step": 184230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.309769860604653e-07,
+      "loss": 1.1358,
+      "step": 184235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.307128333052514e-07,
+      "loss": 0.9642,
+      "step": 184240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.304487340537046e-07,
+      "loss": 1.0179,
+      "step": 184245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.301846883073348e-07,
+      "loss": 1.1083,
+      "step": 184250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.299206960676518e-07,
+      "loss": 0.9877,
+      "step": 184255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.296567573361601e-07,
+      "loss": 0.9119,
+      "step": 184260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.293928721143694e-07,
+      "loss": 1.4475,
+      "step": 184265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.291290404037842e-07,
+      "loss": 1.5005,
+      "step": 184270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.288652622059133e-07,
+      "loss": 1.0443,
+      "step": 184275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.286015375222609e-07,
+      "loss": 1.243,
+      "step": 184280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.283378663543349e-07,
+      "loss": 1.3119,
+      "step": 184285
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.280742487036417e-07,
+      "loss": 1.1255,
+      "step": 184290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.278106845716813e-07,
+      "loss": 1.3736,
+      "step": 184295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.275471739599659e-07,
+      "loss": 1.2563,
+      "step": 184300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.272837168699974e-07,
+      "loss": 1.3546,
+      "step": 184305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.270203133032782e-07,
+      "loss": 1.223,
+      "step": 184310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.267569632613157e-07,
+      "loss": 1.2528,
+      "step": 184315
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.264936667456112e-07,
+      "loss": 1.5339,
+      "step": 184320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.262304237576688e-07,
+      "loss": 1.3723,
+      "step": 184325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.259672342989942e-07,
+      "loss": 1.1217,
+      "step": 184330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.257040983710894e-07,
+      "loss": 1.2138,
+      "step": 184335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.254410159754532e-07,
+      "loss": 1.1956,
+      "step": 184340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.251779871135921e-07,
+      "loss": 1.3835,
+      "step": 184345
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.249150117870073e-07,
+      "loss": 1.1779,
+      "step": 184350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.246520899972019e-07,
+      "loss": 0.9384,
+      "step": 184355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.243892217456726e-07,
+      "loss": 1.2317,
+      "step": 184360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.241264070339237e-07,
+      "loss": 1.1468,
+      "step": 184365
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.238636458634584e-07,
+      "loss": 1.2446,
+      "step": 184370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.236009382357722e-07,
+      "loss": 1.7465,
+      "step": 184375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.233382841523694e-07,
+      "loss": 1.3267,
+      "step": 184380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.230756836147489e-07,
+      "loss": 1.1371,
+      "step": 184385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.228131366244061e-07,
+      "loss": 1.2029,
+      "step": 184390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.225506431828465e-07,
+      "loss": 1.4546,
+      "step": 184395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.222882032915667e-07,
+      "loss": 1.4778,
+      "step": 184400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.220258169520643e-07,
+      "loss": 1.2326,
+      "step": 184405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.217634841658393e-07,
+      "loss": 1.189,
+      "step": 184410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.215012049343872e-07,
+      "loss": 1.0228,
+      "step": 184415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.212389792592088e-07,
+      "loss": 1.4196,
+      "step": 184420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.209768071417999e-07,
+      "loss": 1.2658,
+      "step": 184425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.207146885836579e-07,
+      "loss": 1.5649,
+      "step": 184430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.204526235862796e-07,
+      "loss": 1.2798,
+      "step": 184435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.201906121511603e-07,
+      "loss": 1.1228,
+      "step": 184440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.199286542798e-07,
+      "loss": 1.1817,
+      "step": 184445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.19666749973692e-07,
+      "loss": 1.2875,
+      "step": 184450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.194048992343304e-07,
+      "loss": 1.1597,
+      "step": 184455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.191431020632122e-07,
+      "loss": 0.8609,
+      "step": 184460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.188813584618347e-07,
+      "loss": 1.2594,
+      "step": 184465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.186196684316881e-07,
+      "loss": 1.0888,
+      "step": 184470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.183580319742699e-07,
+      "loss": 0.7944,
+      "step": 184475
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.180964490910723e-07,
+      "loss": 1.1708,
+      "step": 184480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.17834919783592e-07,
+      "loss": 1.1818,
+      "step": 184485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.175734440533188e-07,
+      "loss": 1.2129,
+      "step": 184490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.173120219017492e-07,
+      "loss": 1.0642,
+      "step": 184495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.170506533303744e-07,
+      "loss": 1.479,
+      "step": 184500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.167893383406854e-07,
+      "loss": 1.1468,
+      "step": 184505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.165280769341753e-07,
+      "loss": 1.1446,
+      "step": 184510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.162668691123397e-07,
+      "loss": 1.291,
+      "step": 184515
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.16005714876664e-07,
+      "loss": 1.0152,
+      "step": 184520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.157446142286461e-07,
+      "loss": 1.188,
+      "step": 184525
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.154835671697712e-07,
+      "loss": 1.5516,
+      "step": 184530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.152225737015339e-07,
+      "loss": 1.4554,
+      "step": 184535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.149616338254239e-07,
+      "loss": 1.3046,
+      "step": 184540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.147007475429312e-07,
+      "loss": 0.9822,
+      "step": 184545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.144399148555424e-07,
+      "loss": 1.5725,
+      "step": 184550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.141791357647509e-07,
+      "loss": 1.6172,
+      "step": 184555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.139184102720464e-07,
+      "loss": 1.2021,
+      "step": 184560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.136577383789155e-07,
+      "loss": 0.9622,
+      "step": 184565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.13397120086846e-07,
+      "loss": 1.4206,
+      "step": 184570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.131365553973278e-07,
+      "loss": 1.1361,
+      "step": 184575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.128760443118498e-07,
+      "loss": 1.0281,
+      "step": 184580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.126155868318973e-07,
+      "loss": 0.9658,
+      "step": 184585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.123551829589602e-07,
+      "loss": 1.3805,
+      "step": 184590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.120948326945242e-07,
+      "loss": 1.0042,
+      "step": 184595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.118345360400723e-07,
+      "loss": 1.2179,
+      "step": 184600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.11574292997098e-07,
+      "loss": 1.0953,
+      "step": 184605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.113141035670844e-07,
+      "loss": 1.2286,
+      "step": 184610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.11053967751516e-07,
+      "loss": 1.2218,
+      "step": 184615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.107938855518803e-07,
+      "loss": 1.2286,
+      "step": 184620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.105338569696595e-07,
+      "loss": 1.0363,
+      "step": 184625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.102738820063425e-07,
+      "loss": 1.0667,
+      "step": 184630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.100139606634103e-07,
+      "loss": 1.1365,
+      "step": 184635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.097540929423507e-07,
+      "loss": 1.3609,
+      "step": 184640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.094942788446434e-07,
+      "loss": 1.3633,
+      "step": 184645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.092345183717751e-07,
+      "loss": 1.1394,
+      "step": 184650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.089748115252303e-07,
+      "loss": 0.9042,
+      "step": 184655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.087151583064899e-07,
+      "loss": 1.0261,
+      "step": 184660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.08455558717036e-07,
+      "loss": 1.1941,
+      "step": 184665
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.08196012758352e-07,
+      "loss": 1.4509,
+      "step": 184670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.07936520431921e-07,
+      "loss": 1.0996,
+      "step": 184675
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.07677081739223e-07,
+      "loss": 0.8744,
+      "step": 184680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.074176966817424e-07,
+      "loss": 1.1021,
+      "step": 184685
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.071583652609581e-07,
+      "loss": 1.202,
+      "step": 184690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.068990874783487e-07,
+      "loss": 1.4441,
+      "step": 184695
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.066398633353999e-07,
+      "loss": 1.2011,
+      "step": 184700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.063806928335902e-07,
+      "loss": 1.2897,
+      "step": 184705
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.061215759743999e-07,
+      "loss": 1.2713,
+      "step": 184710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.058625127593054e-07,
+      "loss": 1.0467,
+      "step": 184715
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.056035031897889e-07,
+      "loss": 1.0608,
+      "step": 184720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.053445472673314e-07,
+      "loss": 1.1916,
+      "step": 184725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.050856449934073e-07,
+      "loss": 1.1971,
+      "step": 184730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.048267963694987e-07,
+      "loss": 0.9508,
+      "step": 184735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.045680013970812e-07,
+      "loss": 1.3068,
+      "step": 184740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.043092600776334e-07,
+      "loss": 1.9032,
+      "step": 184745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.040505724126345e-07,
+      "loss": 1.5835,
+      "step": 184750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.037919384035606e-07,
+      "loss": 1.066,
+      "step": 184755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.035333580518877e-07,
+      "loss": 1.0982,
+      "step": 184760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.03274831359092e-07,
+      "loss": 1.7413,
+      "step": 184765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.030163583266535e-07,
+      "loss": 1.3985,
+      "step": 184770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.027579389560456e-07,
+      "loss": 1.2975,
+      "step": 184775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.024995732487426e-07,
+      "loss": 1.091,
+      "step": 184780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.022412612062234e-07,
+      "loss": 1.4913,
+      "step": 184785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.01983002829959e-07,
+      "loss": 1.1614,
+      "step": 184790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.017247981214269e-07,
+      "loss": 1.1945,
+      "step": 184795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.014666470821018e-07,
+      "loss": 1.3536,
+      "step": 184800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.012085497134579e-07,
+      "loss": 1.3889,
+      "step": 184805
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.009505060169673e-07,
+      "loss": 1.2518,
+      "step": 184810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.006925159941035e-07,
+      "loss": 1.1043,
+      "step": 184815
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.004345796463429e-07,
+      "loss": 1.1109,
+      "step": 184820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.001766969751554e-07,
+      "loss": 1.1826,
+      "step": 184825
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.999188679820156e-07,
+      "loss": 1.2136,
+      "step": 184830
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.996610926683932e-07,
+      "loss": 1.1346,
+      "step": 184835
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.994033710357628e-07,
+      "loss": 1.2201,
+      "step": 184840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.991457030855963e-07,
+      "loss": 1.3636,
+      "step": 184845
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.988880888193638e-07,
+      "loss": 1.3084,
+      "step": 184850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.986305282385363e-07,
+      "loss": 1.2168,
+      "step": 184855
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.983730213445837e-07,
+      "loss": 1.2561,
+      "step": 184860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.981155681389794e-07,
+      "loss": 1.1316,
+      "step": 184865
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.978581686231933e-07,
+      "loss": 1.1633,
+      "step": 184870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.976008227986918e-07,
+      "loss": 1.3026,
+      "step": 184875
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.973435306669461e-07,
+      "loss": 1.1404,
+      "step": 184880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.970862922294285e-07,
+      "loss": 1.2382,
+      "step": 184885
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.968291074876032e-07,
+      "loss": 1.1991,
+      "step": 184890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.965719764429434e-07,
+      "loss": 1.029,
+      "step": 184895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.963148990969147e-07,
+      "loss": 1.0458,
+      "step": 184900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.960578754509849e-07,
+      "loss": 1.2079,
+      "step": 184905
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.958009055066228e-07,
+      "loss": 1.0035,
+      "step": 184910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.955439892652959e-07,
+      "loss": 1.245,
+      "step": 184915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.952871267284721e-07,
+      "loss": 1.0856,
+      "step": 184920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.950303178976147e-07,
+      "loss": 1.4189,
+      "step": 184925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.947735627741935e-07,
+      "loss": 1.4288,
+      "step": 184930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.945168613596752e-07,
+      "loss": 1.6505,
+      "step": 184935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.942602136555231e-07,
+      "loss": 1.2431,
+      "step": 184940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.940036196632048e-07,
+      "loss": 1.6223,
+      "step": 184945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.937470793841838e-07,
+      "loss": 1.1186,
+      "step": 184950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.934905928199264e-07,
+      "loss": 1.3339,
+      "step": 184955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.932341599718982e-07,
+      "loss": 1.4375,
+      "step": 184960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.929777808415627e-07,
+      "loss": 1.0808,
+      "step": 184965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.927214554303817e-07,
+      "loss": 1.1137,
+      "step": 184970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.92465183739821e-07,
+      "loss": 1.1818,
+      "step": 184975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.922089657713459e-07,
+      "loss": 1.3987,
+      "step": 184980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.919528015264153e-07,
+      "loss": 1.3444,
+      "step": 184985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.91696691006497e-07,
+      "loss": 1.0831,
+      "step": 184990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.914406342130496e-07,
+      "loss": 1.288,
+      "step": 184995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.911846311475356e-07,
+      "loss": 1.2052,
+      "step": 185000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.90928681811418e-07,
+      "loss": 1.3979,
+      "step": 185005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.906727862061601e-07,
+      "loss": 1.2655,
+      "step": 185010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.904169443332208e-07,
+      "loss": 1.3879,
+      "step": 185015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.901611561940612e-07,
+      "loss": 1.3023,
+      "step": 185020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.899054217901412e-07,
+      "loss": 0.9495,
+      "step": 185025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.896497411229252e-07,
+      "loss": 1.5575,
+      "step": 185030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.893941141938686e-07,
+      "loss": 1.2739,
+      "step": 185035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.891385410044348e-07,
+      "loss": 1.4746,
+      "step": 185040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.888830215560793e-07,
+      "loss": 1.314,
+      "step": 185045
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.886275558502641e-07,
+      "loss": 1.2568,
+      "step": 185050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.883721438884493e-07,
+      "loss": 1.3532,
+      "step": 185055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.881167856720916e-07,
+      "loss": 1.154,
+      "step": 185060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.878614812026473e-07,
+      "loss": 1.228,
+      "step": 185065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.876062304815755e-07,
+      "loss": 1.3009,
+      "step": 185070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.873510335103372e-07,
+      "loss": 1.1218,
+      "step": 185075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.870958902903867e-07,
+      "loss": 1.278,
+      "step": 185080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.868408008231807e-07,
+      "loss": 1.2138,
+      "step": 185085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.865857651101781e-07,
+      "loss": 1.1952,
+      "step": 185090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.863307831528309e-07,
+      "loss": 1.3351,
+      "step": 185095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.860758549525992e-07,
+      "loss": 1.3211,
+      "step": 185100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.858209805109394e-07,
+      "loss": 1.0176,
+      "step": 185105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.85566159829305e-07,
+      "loss": 1.1277,
+      "step": 185110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.853113929091492e-07,
+      "loss": 1.1897,
+      "step": 185115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.850566797519297e-07,
+      "loss": 1.199,
+      "step": 185120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.848020203591021e-07,
+      "loss": 1.1283,
+      "step": 185125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.845474147321162e-07,
+      "loss": 1.0718,
+      "step": 185130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.842928628724309e-07,
+      "loss": 1.1419,
+      "step": 185135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.840383647814962e-07,
+      "loss": 1.2052,
+      "step": 185140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.837839204607687e-07,
+      "loss": 1.3895,
+      "step": 185145
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.835295299116983e-07,
+      "loss": 1.2855,
+      "step": 185150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.832751931357395e-07,
+      "loss": 1.0958,
+      "step": 185155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.830209101343431e-07,
+      "loss": 1.1946,
+      "step": 185160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.827666809089627e-07,
+      "loss": 1.149,
+      "step": 185165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.825125054610503e-07,
+      "loss": 1.3632,
+      "step": 185170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.82258383792057e-07,
+      "loss": 1.2661,
+      "step": 185175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.820043159034327e-07,
+      "loss": 1.1705,
+      "step": 185180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.817503017966297e-07,
+      "loss": 1.0026,
+      "step": 185185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.814963414730979e-07,
+      "loss": 1.0495,
+      "step": 185190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.812424349342871e-07,
+      "loss": 1.3664,
+      "step": 185195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.809885821816486e-07,
+      "loss": 1.1849,
+      "step": 185200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.807347832166321e-07,
+      "loss": 0.9552,
+      "step": 185205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.804810380406845e-07,
+      "loss": 1.148,
+      "step": 185210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.802273466552566e-07,
+      "loss": 1.1083,
+      "step": 185215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.799737090617986e-07,
+      "loss": 1.2315,
+      "step": 185220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.797201252617579e-07,
+      "loss": 0.9646,
+      "step": 185225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.794665952565793e-07,
+      "loss": 0.9924,
+      "step": 185230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.792131190477146e-07,
+      "loss": 1.1824,
+      "step": 185235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.789596966366107e-07,
+      "loss": 1.4435,
+      "step": 185240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.787063280247118e-07,
+      "loss": 1.4777,
+      "step": 185245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.784530132134691e-07,
+      "loss": 1.1545,
+      "step": 185250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.781997522043259e-07,
+      "loss": 1.1778,
+      "step": 185255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.779465449987298e-07,
+      "loss": 1.5998,
+      "step": 185260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.776933915981275e-07,
+      "loss": 1.0887,
+      "step": 185265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.774402920039646e-07,
+      "loss": 1.1577,
+      "step": 185270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.771872462176831e-07,
+      "loss": 1.2468,
+      "step": 185275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.769342542407319e-07,
+      "loss": 1.0182,
+      "step": 185280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.766813160745544e-07,
+      "loss": 1.0578,
+      "step": 185285
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.764284317205959e-07,
+      "loss": 1.4058,
+      "step": 185290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.761756011802966e-07,
+      "loss": 1.4225,
+      "step": 185295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.759228244551063e-07,
+      "loss": 1.221,
+      "step": 185300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.756701015464628e-07,
+      "loss": 1.0464,
+      "step": 185305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.754174324558126e-07,
+      "loss": 1.2799,
+      "step": 185310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.751648171845992e-07,
+      "loss": 1.3007,
+      "step": 185315
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.749122557342646e-07,
+      "loss": 1.0695,
+      "step": 185320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.746597481062477e-07,
+      "loss": 1.1697,
+      "step": 185325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.744072943019941e-07,
+      "loss": 1.1859,
+      "step": 185330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.741548943229447e-07,
+      "loss": 1.1961,
+      "step": 185335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.739025481705396e-07,
+      "loss": 1.4782,
+      "step": 185340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.736502558462232e-07,
+      "loss": 1.3577,
+      "step": 185345
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.733980173514309e-07,
+      "loss": 1.356,
+      "step": 185350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.731458326876083e-07,
+      "loss": 1.297,
+      "step": 185355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.72893701856192e-07,
+      "loss": 1.1917,
+      "step": 185360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.726416248586242e-07,
+      "loss": 1.2317,
+      "step": 185365
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.723896016963426e-07,
+      "loss": 1.5217,
+      "step": 185370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.72137632370785e-07,
+      "loss": 0.9189,
+      "step": 185375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.718857168833946e-07,
+      "loss": 1.3897,
+      "step": 185380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.716338552356082e-07,
+      "loss": 1.1854,
+      "step": 185385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.713820474288622e-07,
+      "loss": 1.0427,
+      "step": 185390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.711302934645979e-07,
+      "loss": 1.4529,
+      "step": 185395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.708785933442484e-07,
+      "loss": 1.2862,
+      "step": 185400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.706269470692538e-07,
+      "loss": 1.4648,
+      "step": 185405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.703753546410529e-07,
+      "loss": 1.232,
+      "step": 185410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7012381606108e-07,
+      "loss": 1.1271,
+      "step": 185415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.698723313307708e-07,
+      "loss": 1.3752,
+      "step": 185420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.696209004515619e-07,
+      "loss": 1.2604,
+      "step": 185425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.693695234248919e-07,
+      "loss": 1.151,
+      "step": 185430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.691182002521945e-07,
+      "loss": 1.3638,
+      "step": 185435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.688669309349026e-07,
+      "loss": 1.4609,
+      "step": 185440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.686157154744531e-07,
+      "loss": 1.1709,
+      "step": 185445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.683645538722827e-07,
+      "loss": 1.1056,
+      "step": 185450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.68113446129821e-07,
+      "loss": 1.0051,
+      "step": 185455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.678623922485061e-07,
+      "loss": 1.3961,
+      "step": 185460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.676113922297688e-07,
+      "loss": 1.3684,
+      "step": 185465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.673604460750448e-07,
+      "loss": 1.1893,
+      "step": 185470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.671095537857663e-07,
+      "loss": 1.0165,
+      "step": 185475
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.668587153633654e-07,
+      "loss": 1.5676,
+      "step": 185480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.666079308092742e-07,
+      "loss": 1.1471,
+      "step": 185485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.663572001249262e-07,
+      "loss": 1.4509,
+      "step": 185490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.661065233117524e-07,
+      "loss": 1.4361,
+      "step": 185495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.658559003711827e-07,
+      "loss": 1.2054,
+      "step": 185500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.656053313046528e-07,
+      "loss": 0.937,
+      "step": 185505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.653548161135902e-07,
+      "loss": 1.0021,
+      "step": 185510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.65104354799425e-07,
+      "loss": 1.2262,
+      "step": 185515
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.648539473635884e-07,
+      "loss": 1.6032,
+      "step": 185520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.646035938075112e-07,
+      "loss": 1.1757,
+      "step": 185525
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.643532941326224e-07,
+      "loss": 1.1951,
+      "step": 185530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.641030483403509e-07,
+      "loss": 1.2565,
+      "step": 185535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.638528564321255e-07,
+      "loss": 1.4688,
+      "step": 185540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.636027184093762e-07,
+      "loss": 1.0423,
+      "step": 185545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.633526342735308e-07,
+      "loss": 1.345,
+      "step": 185550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.631026040260179e-07,
+      "loss": 1.2014,
+      "step": 185555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.628526276682633e-07,
+      "loss": 1.3973,
+      "step": 185560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.626027052016958e-07,
+      "loss": 1.2028,
+      "step": 185565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.623528366277453e-07,
+      "loss": 1.3191,
+      "step": 185570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.621030219478352e-07,
+      "loss": 1.4374,
+      "step": 185575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.618532611633931e-07,
+      "loss": 1.5536,
+      "step": 185580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.616035542758424e-07,
+      "loss": 1.9063,
+      "step": 185585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.613539012866154e-07,
+      "loss": 1.1645,
+      "step": 185590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.611043021971341e-07,
+      "loss": 1.0392,
+      "step": 185595
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.608547570088218e-07,
+      "loss": 1.1316,
+      "step": 185600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.606052657231076e-07,
+      "loss": 1.5455,
+      "step": 185605
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.603558283414123e-07,
+      "loss": 0.9722,
+      "step": 185610
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.601064448651638e-07,
+      "loss": 1.0248,
+      "step": 185615
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.598571152957843e-07,
+      "loss": 1.291,
+      "step": 185620
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.596078396346993e-07,
+      "loss": 1.1863,
+      "step": 185625
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.593586178833299e-07,
+      "loss": 1.0776,
+      "step": 185630
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.591094500430993e-07,
+      "loss": 1.3582,
+      "step": 185635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.588603361154344e-07,
+      "loss": 0.9675,
+      "step": 185640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.586112761017526e-07,
+      "loss": 1.102,
+      "step": 185645
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.583622700034797e-07,
+      "loss": 1.4789,
+      "step": 185650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.581133178220354e-07,
+      "loss": 1.1707,
+      "step": 185655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.578644195588434e-07,
+      "loss": 1.137,
+      "step": 185660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.576155752153234e-07,
+      "loss": 1.2025,
+      "step": 185665
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.573667847928976e-07,
+      "loss": 1.6557,
+      "step": 185670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.571180482929872e-07,
+      "loss": 1.5469,
+      "step": 185675
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.568693657170076e-07,
+      "loss": 1.0139,
+      "step": 185680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.566207370663868e-07,
+      "loss": 1.1044,
+      "step": 185685
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.5637216234254e-07,
+      "loss": 1.1763,
+      "step": 185690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.561236415468862e-07,
+      "loss": 1.1023,
+      "step": 185695
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.558751746808477e-07,
+      "loss": 1.0847,
+      "step": 185700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.556267617458399e-07,
+      "loss": 1.1349,
+      "step": 185705
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.55378402743283e-07,
+      "loss": 1.1164,
+      "step": 185710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.551300976745966e-07,
+      "loss": 1.468,
+      "step": 185715
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.548818465411976e-07,
+      "loss": 1.236,
+      "step": 185720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.546336493445015e-07,
+      "loss": 1.234,
+      "step": 185725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.543855060859271e-07,
+      "loss": 1.1308,
+      "step": 185730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.541374167668934e-07,
+      "loss": 1.4224,
+      "step": 185735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.538893813888158e-07,
+      "loss": 1.1978,
+      "step": 185740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.536413999531087e-07,
+      "loss": 1.1115,
+      "step": 185745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.533934724611901e-07,
+      "loss": 1.268,
+      "step": 185750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.531455989144774e-07,
+      "loss": 1.1166,
+      "step": 185755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.528977793143831e-07,
+      "loss": 1.1867,
+      "step": 185760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.526500136623248e-07,
+      "loss": 1.195,
+      "step": 185765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.524023019597157e-07,
+      "loss": 1.5085,
+      "step": 185770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.521546442079684e-07,
+      "loss": 1.2791,
+      "step": 185775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.519070404085025e-07,
+      "loss": 1.5069,
+      "step": 185780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.516594905627293e-07,
+      "loss": 1.1338,
+      "step": 185785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.514119946720598e-07,
+      "loss": 1.4419,
+      "step": 185790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.511645527379117e-07,
+      "loss": 1.2269,
+      "step": 185795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.509171647616951e-07,
+      "loss": 0.9245,
+      "step": 185800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.506698307448255e-07,
+      "loss": 1.0699,
+      "step": 185805
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.504225506887106e-07,
+      "loss": 1.4244,
+      "step": 185810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.501753245947683e-07,
+      "loss": 1.2864,
+      "step": 185815
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.499281524644052e-07,
+      "loss": 1.5512,
+      "step": 185820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.496810342990355e-07,
+      "loss": 1.1042,
+      "step": 185825
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.494339701000706e-07,
+      "loss": 1.0877,
+      "step": 185830
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.491869598689214e-07,
+      "loss": 1.2796,
+      "step": 185835
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.489400036069969e-07,
+      "loss": 1.0156,
+      "step": 185840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.486931013157082e-07,
+      "loss": 1.177,
+      "step": 185845
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.484462529964663e-07,
+      "loss": 1.5888,
+      "step": 185850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.481994586506778e-07,
+      "loss": 1.7476,
+      "step": 185855
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.479527182797561e-07,
+      "loss": 1.2431,
+      "step": 185860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.477060318851069e-07,
+      "loss": 1.8354,
+      "step": 185865
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.47459399468141e-07,
+      "loss": 1.0896,
+      "step": 185870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.472128210302652e-07,
+      "loss": 1.3866,
+      "step": 185875
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.469662965728895e-07,
+      "loss": 1.1119,
+      "step": 185880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.467198260974205e-07,
+      "loss": 1.2621,
+      "step": 185885
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.464734096052648e-07,
+      "loss": 1.2271,
+      "step": 185890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.462270470978303e-07,
+      "loss": 1.1802,
+      "step": 185895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.459807385765259e-07,
+      "loss": 1.0665,
+      "step": 185900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.457344840427536e-07,
+      "loss": 1.0885,
+      "step": 185905
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.454882834979258e-07,
+      "loss": 1.1037,
+      "step": 185910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.452421369434424e-07,
+      "loss": 1.0681,
+      "step": 185915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.449960443807112e-07,
+      "loss": 1.3549,
+      "step": 185920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.447500058111399e-07,
+      "loss": 1.1036,
+      "step": 185925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.445040212361318e-07,
+      "loss": 0.9732,
+      "step": 185930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.442580906570905e-07,
+      "loss": 1.26,
+      "step": 185935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.440122140754201e-07,
+      "loss": 1.0819,
+      "step": 185940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.437663914925284e-07,
+      "loss": 1.4446,
+      "step": 185945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.435206229098156e-07,
+      "loss": 1.2474,
+      "step": 185950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.432749083286859e-07,
+      "loss": 1.3311,
+      "step": 185955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.430292477505416e-07,
+      "loss": 1.3266,
+      "step": 185960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.427836411767895e-07,
+      "loss": 1.035,
+      "step": 185965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.425380886088272e-07,
+      "loss": 1.4369,
+      "step": 185970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.422925900480602e-07,
+      "loss": 1.1477,
+      "step": 185975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.420471454958898e-07,
+      "loss": 1.0264,
+      "step": 185980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.418017549537147e-07,
+      "loss": 1.0607,
+      "step": 185985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.415564184229416e-07,
+      "loss": 1.4913,
+      "step": 185990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.413111359049672e-07,
+      "loss": 1.0846,
+      "step": 185995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.410659074011937e-07,
+      "loss": 0.9932,
+      "step": 186000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.408207329130222e-07,
+      "loss": 1.1086,
+      "step": 186005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.405756124418504e-07,
+      "loss": 1.2246,
+      "step": 186010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.403305459890817e-07,
+      "loss": 1.242,
+      "step": 186015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.400855335561117e-07,
+      "loss": 1.2186,
+      "step": 186020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.398405751443425e-07,
+      "loss": 1.4654,
+      "step": 186025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.395956707551697e-07,
+      "loss": 1.2289,
+      "step": 186030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.393508203899955e-07,
+      "loss": 1.6196,
+      "step": 186035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.391060240502167e-07,
+      "loss": 1.4618,
+      "step": 186040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.388612817372318e-07,
+      "loss": 1.2894,
+      "step": 186045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.386165934524356e-07,
+      "loss": 1.2735,
+      "step": 186050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.383719591972281e-07,
+      "loss": 1.2697,
+      "step": 186055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.381273789730068e-07,
+      "loss": 1.2687,
+      "step": 186060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.378828527811664e-07,
+      "loss": 1.0604,
+      "step": 186065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.376383806231045e-07,
+      "loss": 1.5025,
+      "step": 186070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.373939625002178e-07,
+      "loss": 1.195,
+      "step": 186075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.371495984138975e-07,
+      "loss": 1.257,
+      "step": 186080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.369052883655457e-07,
+      "loss": 1.1569,
+      "step": 186085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.366610323565547e-07,
+      "loss": 1.1851,
+      "step": 186090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.364168303883188e-07,
+      "loss": 0.9494,
+      "step": 186095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.361726824622315e-07,
+      "loss": 1.5081,
+      "step": 186100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.359285885796884e-07,
+      "loss": 1.1973,
+      "step": 186105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.356845487420848e-07,
+      "loss": 1.0602,
+      "step": 186110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.354405629508109e-07,
+      "loss": 1.1831,
+      "step": 186115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.351966312072642e-07,
+      "loss": 1.2631,
+      "step": 186120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.349527535128329e-07,
+      "loss": 1.0626,
+      "step": 186125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.347089298689134e-07,
+      "loss": 1.295,
+      "step": 186130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.344651602768968e-07,
+      "loss": 1.3693,
+      "step": 186135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.342214447381766e-07,
+      "loss": 1.1338,
+      "step": 186140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.339777832541415e-07,
+      "loss": 1.2338,
+      "step": 186145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.337341758261838e-07,
+      "loss": 1.4382,
+      "step": 186150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.334906224556969e-07,
+      "loss": 1.8467,
+      "step": 186155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.332471231440705e-07,
+      "loss": 1.0376,
+      "step": 186160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.330036778926939e-07,
+      "loss": 0.9971,
+      "step": 186165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.327602867029569e-07,
+      "loss": 1.4371,
+      "step": 186170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.325169495762528e-07,
+      "loss": 1.3236,
+      "step": 186175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.322736665139683e-07,
+      "loss": 1.1921,
+      "step": 186180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.320304375174934e-07,
+      "loss": 1.1398,
+      "step": 186185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.317872625882181e-07,
+      "loss": 1.463,
+      "step": 186190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.31544141727528e-07,
+      "loss": 0.9989,
+      "step": 186195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.31301074936813e-07,
+      "loss": 1.0965,
+      "step": 186200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.310580622174644e-07,
+      "loss": 0.944,
+      "step": 186205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.308151035708653e-07,
+      "loss": 1.0202,
+      "step": 186210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.305721989984059e-07,
+      "loss": 1.188,
+      "step": 186215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.303293485014715e-07,
+      "loss": 1.3942,
+      "step": 186220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.300865520814491e-07,
+      "loss": 1.2004,
+      "step": 186225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.298438097397274e-07,
+      "loss": 1.0604,
+      "step": 186230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.29601121477692e-07,
+      "loss": 1.4159,
+      "step": 186235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.293584872967262e-07,
+      "loss": 1.1051,
+      "step": 186240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.291159071982166e-07,
+      "loss": 1.1452,
+      "step": 186245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.288733811835511e-07,
+      "loss": 1.8108,
+      "step": 186250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.28630909254112e-07,
+      "loss": 1.3053,
+      "step": 186255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.283884914112836e-07,
+      "loss": 1.2139,
+      "step": 186260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.281461276564515e-07,
+      "loss": 1.0994,
+      "step": 186265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.279038179910012e-07,
+      "loss": 1.6106,
+      "step": 186270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.276615624163118e-07,
+      "loss": 0.9985,
+      "step": 186275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.274193609337719e-07,
+      "loss": 1.4963,
+      "step": 186280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.271772135447628e-07,
+      "loss": 1.4226,
+      "step": 186285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.269351202506634e-07,
+      "loss": 1.248,
+      "step": 186290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.266930810528626e-07,
+      "loss": 1.2579,
+      "step": 186295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.264510959527392e-07,
+      "loss": 1.4172,
+      "step": 186300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.262091649516754e-07,
+      "loss": 1.0589,
+      "step": 186305
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.259672880510524e-07,
+      "loss": 1.1576,
+      "step": 186310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.257254652522503e-07,
+      "loss": 1.1864,
+      "step": 186315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.254836965566545e-07,
+      "loss": 1.135,
+      "step": 186320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.252419819656406e-07,
+      "loss": 0.9784,
+      "step": 186325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.250003214805922e-07,
+      "loss": 1.4411,
+      "step": 186330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.247587151028865e-07,
+      "loss": 1.3346,
+      "step": 186335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.245171628339052e-07,
+      "loss": 1.2325,
+      "step": 186340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.24275664675028e-07,
+      "loss": 1.5023,
+      "step": 186345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.240342206276339e-07,
+      "loss": 1.1953,
+      "step": 186350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.237928306931006e-07,
+      "loss": 0.9072,
+      "step": 186355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.23551494872806e-07,
+      "loss": 1.1402,
+      "step": 186360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2331021316813e-07,
+      "loss": 1.1557,
+      "step": 186365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.230689855804494e-07,
+      "loss": 1.2828,
+      "step": 186370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.22827812111143e-07,
+      "loss": 1.1659,
+      "step": 186375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.225866927615874e-07,
+      "loss": 1.1319,
+      "step": 186380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.223456275331584e-07,
+      "loss": 1.1729,
+      "step": 186385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.221046164272326e-07,
+      "loss": 1.3401,
+      "step": 186390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.218636594451887e-07,
+      "loss": 1.295,
+      "step": 186395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.216227565884014e-07,
+      "loss": 1.2033,
+      "step": 186400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.213819078582449e-07,
+      "loss": 1.124,
+      "step": 186405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.21141113256095e-07,
+      "loss": 1.1065,
+      "step": 186410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.209003727833295e-07,
+      "loss": 1.3821,
+      "step": 186415
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.206596864413193e-07,
+      "loss": 1.1617,
+      "step": 186420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.204190542314424e-07,
+      "loss": 1.468,
+      "step": 186425
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.20178476155071e-07,
+      "loss": 1.2777,
+      "step": 186430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.199379522135783e-07,
+      "loss": 1.4455,
+      "step": 186435
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1969748240834e-07,
+      "loss": 0.9912,
+      "step": 186440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.194570667407284e-07,
+      "loss": 1.2426,
+      "step": 186445
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.192167052121144e-07,
+      "loss": 1.5221,
+      "step": 186450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.189763978238727e-07,
+      "loss": 1.158,
+      "step": 186455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.187361445773764e-07,
+      "loss": 1.4157,
+      "step": 186460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.184959454739957e-07,
+      "loss": 1.1569,
+      "step": 186465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.182558005151017e-07,
+      "loss": 1.1873,
+      "step": 186470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.180157097020688e-07,
+      "loss": 1.216,
+      "step": 186475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.177756730362649e-07,
+      "loss": 1.4194,
+      "step": 186480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.175356905190609e-07,
+      "loss": 1.0348,
+      "step": 186485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.172957621518304e-07,
+      "loss": 1.5093,
+      "step": 186490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.170558879359411e-07,
+      "loss": 1.6329,
+      "step": 186495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.168160678727618e-07,
+      "loss": 1.2177,
+      "step": 186500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.165763019636627e-07,
+      "loss": 1.1009,
+      "step": 186505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.163365902100159e-07,
+      "loss": 1.3475,
+      "step": 186510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.160969326131859e-07,
+      "loss": 1.2622,
+      "step": 186515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.15857329174545e-07,
+      "loss": 1.3167,
+      "step": 186520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.156177798954587e-07,
+      "loss": 1.2411,
+      "step": 186525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.15378284777297e-07,
+      "loss": 1.0154,
+      "step": 186530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.151388438214256e-07,
+      "loss": 0.885,
+      "step": 186535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.148994570292142e-07,
+      "loss": 1.2926,
+      "step": 186540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.146601244020266e-07,
+      "loss": 1.3565,
+      "step": 186545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.144208459412326e-07,
+      "loss": 1.4749,
+      "step": 186550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.141816216481976e-07,
+      "loss": 1.1185,
+      "step": 186555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.139424515242885e-07,
+      "loss": 1.4687,
+      "step": 186560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.137033355708676e-07,
+      "loss": 1.1242,
+      "step": 186565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.134642737893037e-07,
+      "loss": 1.5105,
+      "step": 186570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.132252661809622e-07,
+      "loss": 1.1637,
+      "step": 186575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.129863127472057e-07,
+      "loss": 1.3393,
+      "step": 186580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.127474134894006e-07,
+      "loss": 1.2548,
+      "step": 186585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.125085684089115e-07,
+      "loss": 1.2493,
+      "step": 186590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.122697775070995e-07,
+      "loss": 1.5736,
+      "step": 186595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.120310407853302e-07,
+      "loss": 1.2646,
+      "step": 186600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.11792358244968e-07,
+      "loss": 1.2384,
+      "step": 186605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.115537298873752e-07,
+      "loss": 1.1607,
+      "step": 186610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.113151557139118e-07,
+      "loss": 1.6136,
+      "step": 186615
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.110766357259434e-07,
+      "loss": 1.1444,
+      "step": 186620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.108381699248322e-07,
+      "loss": 1.3363,
+      "step": 186625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.10599758311937e-07,
+      "loss": 1.1061,
+      "step": 186630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.103614008886227e-07,
+      "loss": 1.056,
+      "step": 186635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.101230976562465e-07,
+      "loss": 1.3311,
+      "step": 186640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.098848486161723e-07,
+      "loss": 1.2304,
+      "step": 186645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.096466537697619e-07,
+      "loss": 1.1549,
+      "step": 186650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.094085131183723e-07,
+      "loss": 1.2334,
+      "step": 186655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.091704266633646e-07,
+      "loss": 1.3319,
+      "step": 186660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.089323944060976e-07,
+      "loss": 1.3669,
+      "step": 186665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.086944163479324e-07,
+      "loss": 1.2649,
+      "step": 186670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.08456492490228e-07,
+      "loss": 1.2844,
+      "step": 186675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.0821862283434e-07,
+      "loss": 1.2709,
+      "step": 186680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.079808073816306e-07,
+      "loss": 1.1411,
+      "step": 186685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.077430461334543e-07,
+      "loss": 1.2706,
+      "step": 186690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.07505339091171e-07,
+      "loss": 1.5054,
+      "step": 186695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.072676862561387e-07,
+      "loss": 1.17,
+      "step": 186700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.070300876297141e-07,
+      "loss": 1.0531,
+      "step": 186705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.067925432132525e-07,
+      "loss": 1.3332,
+      "step": 186710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.065550530081109e-07,
+      "loss": 1.0468,
+      "step": 186715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.06317617015647e-07,
+      "loss": 1.07,
+      "step": 186720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.060802352372152e-07,
+      "loss": 1.1956,
+      "step": 186725
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.058429076741723e-07,
+      "loss": 1.1309,
+      "step": 186730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.056056343278715e-07,
+      "loss": 0.9678,
+      "step": 186735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.053684151996707e-07,
+      "loss": 1.1483,
+      "step": 186740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.051312502909211e-07,
+      "loss": 1.1588,
+      "step": 186745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.048941396029805e-07,
+      "loss": 1.1082,
+      "step": 186750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.046570831371999e-07,
+      "loss": 1.36,
+      "step": 186755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.044200808949351e-07,
+      "loss": 1.1108,
+      "step": 186760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.041831328775404e-07,
+      "loss": 0.8528,
+      "step": 186765
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.039462390863659e-07,
+      "loss": 1.1072,
+      "step": 186770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.03709399522766e-07,
+      "loss": 1.343,
+      "step": 186775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.03472614188093e-07,
+      "loss": 1.1175,
+      "step": 186780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.032358830836992e-07,
+      "loss": 1.1395,
+      "step": 186785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.029992062109356e-07,
+      "loss": 1.1066,
+      "step": 186790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.027625835711558e-07,
+      "loss": 1.0571,
+      "step": 186795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.025260151657097e-07,
+      "loss": 1.0484,
+      "step": 186800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.022895009959472e-07,
+      "loss": 1.0251,
+      "step": 186805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.020530410632196e-07,
+      "loss": 1.3643,
+      "step": 186810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.018166353688803e-07,
+      "loss": 1.1062,
+      "step": 186815
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.015802839142747e-07,
+      "loss": 1.3732,
+      "step": 186820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.013439867007541e-07,
+      "loss": 1.2289,
+      "step": 186825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.011077437296686e-07,
+      "loss": 1.3292,
+      "step": 186830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.00871555002368e-07,
+      "loss": 1.1154,
+      "step": 186835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.00635420520198e-07,
+      "loss": 1.2949,
+      "step": 186840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.003993402845109e-07,
+      "loss": 1.1462,
+      "step": 186845
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.001633142966511e-07,
+      "loss": 1.0794,
+      "step": 186850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.999273425579687e-07,
+      "loss": 1.2266,
+      "step": 186855
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.996914250698126e-07,
+      "loss": 1.5585,
+      "step": 186860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.994555618335273e-07,
+      "loss": 1.5153,
+      "step": 186865
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.992197528504594e-07,
+      "loss": 0.8664,
+      "step": 186870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.989839981219591e-07,
+      "loss": 1.4962,
+      "step": 186875
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.987482976493674e-07,
+      "loss": 1.1306,
+      "step": 186880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.985126514340355e-07,
+      "loss": 1.0591,
+      "step": 186885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.982770594773057e-07,
+      "loss": 1.4005,
+      "step": 186890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.980415217805245e-07,
+      "loss": 1.0459,
+      "step": 186895
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.978060383450368e-07,
+      "loss": 1.1568,
+      "step": 186900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.975706091721866e-07,
+      "loss": 1.0761,
+      "step": 186905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.973352342633208e-07,
+      "loss": 1.1204,
+      "step": 186910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.970999136197818e-07,
+      "loss": 1.364,
+      "step": 186915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.968646472429128e-07,
+      "loss": 1.0457,
+      "step": 186920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.966294351340572e-07,
+      "loss": 1.2014,
+      "step": 186925
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.963942772945607e-07,
+      "loss": 1.1339,
+      "step": 186930
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.96159173725762e-07,
+      "loss": 1.0451,
+      "step": 186935
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.95924124429008e-07,
+      "loss": 1.3381,
+      "step": 186940
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.956891294056387e-07,
+      "loss": 1.3658,
+      "step": 186945
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.954541886569953e-07,
+      "loss": 1.4497,
+      "step": 186950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.95219302184422e-07,
+      "loss": 1.1353,
+      "step": 186955
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.949844699892592e-07,
+      "loss": 0.9553,
+      "step": 186960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.947496920728467e-07,
+      "loss": 1.2349,
+      "step": 186965
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.945149684365236e-07,
+      "loss": 1.229,
+      "step": 186970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.942802990816353e-07,
+      "loss": 1.3769,
+      "step": 186975
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.940456840095187e-07,
+      "loss": 1.0675,
+      "step": 186980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.938111232215126e-07,
+      "loss": 1.3221,
+      "step": 186985
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.935766167189592e-07,
+      "loss": 1.1084,
+      "step": 186990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.933421645031956e-07,
+      "loss": 1.0996,
+      "step": 186995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.931077665755612e-07,
+      "loss": 1.1259,
+      "step": 187000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.928734229373966e-07,
+      "loss": 1.2216,
+      "step": 187005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.926391335900382e-07,
+      "loss": 1.2116,
+      "step": 187010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.924048985348218e-07,
+      "loss": 1.1222,
+      "step": 187015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.921707177730872e-07,
+      "loss": 1.3649,
+      "step": 187020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.919365913061736e-07,
+      "loss": 1.5411,
+      "step": 187025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.917025191354152e-07,
+      "loss": 1.1976,
+      "step": 187030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.914685012621489e-07,
+      "loss": 1.0661,
+      "step": 187035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.912345376877114e-07,
+      "loss": 1.4054,
+      "step": 187040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.910006284134394e-07,
+      "loss": 1.2404,
+      "step": 187045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.907667734406685e-07,
+      "loss": 1.1236,
+      "step": 187050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.905329727707342e-07,
+      "loss": 1.6043,
+      "step": 187055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.90299226404971e-07,
+      "loss": 1.5536,
+      "step": 187060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.900655343447124e-07,
+      "loss": 1.208,
+      "step": 187065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.898318965912963e-07,
+      "loss": 1.1401,
+      "step": 187070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.895983131460558e-07,
+      "loss": 1.3863,
+      "step": 187075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.893647840103221e-07,
+      "loss": 1.1956,
+      "step": 187080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.891313091854333e-07,
+      "loss": 1.2476,
+      "step": 187085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.888978886727181e-07,
+      "loss": 1.2669,
+      "step": 187090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.886645224735109e-07,
+      "loss": 1.2453,
+      "step": 187095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.884312105891475e-07,
+      "loss": 1.3287,
+      "step": 187100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.881979530209569e-07,
+      "loss": 1.2202,
+      "step": 187105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.879647497702711e-07,
+      "loss": 1.3348,
+      "step": 187110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.877316008384225e-07,
+      "loss": 1.1538,
+      "step": 187115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.874985062267435e-07,
+      "loss": 1.2944,
+      "step": 187120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.87265465936565e-07,
+      "loss": 1.109,
+      "step": 187125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.870324799692149e-07,
+      "loss": 1.258,
+      "step": 187130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.867995483260268e-07,
+      "loss": 1.038,
+      "step": 187135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.865666710083305e-07,
+      "loss": 1.4052,
+      "step": 187140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.863338480174551e-07,
+      "loss": 1.2275,
+      "step": 187145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.861010793547316e-07,
+      "loss": 1.3803,
+      "step": 187150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.858683650214857e-07,
+      "loss": 1.0943,
+      "step": 187155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.856357050190496e-07,
+      "loss": 1.298,
+      "step": 187160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.854030993487513e-07,
+      "loss": 1.5649,
+      "step": 187165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.851705480119196e-07,
+      "loss": 1.5227,
+      "step": 187170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.849380510098811e-07,
+      "loss": 1.2851,
+      "step": 187175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.847056083439627e-07,
+      "loss": 1.0413,
+      "step": 187180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.844732200154933e-07,
+      "loss": 1.0142,
+      "step": 187185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.842408860258019e-07,
+      "loss": 1.2813,
+      "step": 187190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.840086063762106e-07,
+      "loss": 1.1317,
+      "step": 187195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.837763810680496e-07,
+      "loss": 0.9233,
+      "step": 187200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.835442101026433e-07,
+      "loss": 1.0259,
+      "step": 187205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.833120934813174e-07,
+      "loss": 1.1184,
+      "step": 187210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.830800312053996e-07,
+      "loss": 1.1443,
+      "step": 187215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.828480232762134e-07,
+      "loss": 1.0963,
+      "step": 187220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.826160696950833e-07,
+      "loss": 1.3713,
+      "step": 187225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.823841704633336e-07,
+      "loss": 1.2558,
+      "step": 187230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.821523255822902e-07,
+      "loss": 1.2361,
+      "step": 187235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.819205350532751e-07,
+      "loss": 1.4583,
+      "step": 187240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.816887988776153e-07,
+      "loss": 1.2112,
+      "step": 187245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.814571170566296e-07,
+      "loss": 1.3241,
+      "step": 187250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.812254895916457e-07,
+      "loss": 1.0258,
+      "step": 187255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.809939164839816e-07,
+      "loss": 1.0507,
+      "step": 187260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.80762397734964e-07,
+      "loss": 1.6215,
+      "step": 187265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.80530933345913e-07,
+      "loss": 1.3483,
+      "step": 187270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.802995233181473e-07,
+      "loss": 1.1058,
+      "step": 187275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.800681676529951e-07,
+      "loss": 1.2749,
+      "step": 187280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.798368663517727e-07,
+      "loss": 1.2007,
+      "step": 187285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.796056194158017e-07,
+      "loss": 1.0855,
+      "step": 187290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.793744268464051e-07,
+      "loss": 1.3365,
+      "step": 187295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.791432886448988e-07,
+      "loss": 1.1712,
+      "step": 187300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.78912204812605e-07,
+      "loss": 1.3376,
+      "step": 187305
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.786811753508458e-07,
+      "loss": 1.2515,
+      "step": 187310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.78450200260937e-07,
+      "loss": 1.0456,
+      "step": 187315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.782192795441986e-07,
+      "loss": 1.2687,
+      "step": 187320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.779884132019486e-07,
+      "loss": 1.2959,
+      "step": 187325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.777576012355078e-07,
+      "loss": 1.5632,
+      "step": 187330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.775268436461933e-07,
+      "loss": 0.9534,
+      "step": 187335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.772961404353194e-07,
+      "loss": 0.9054,
+      "step": 187340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.770654916042073e-07,
+      "loss": 1.2427,
+      "step": 187345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7683489715417476e-07,
+      "loss": 1.3047,
+      "step": 187350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.766043570865353e-07,
+      "loss": 1.3006,
+      "step": 187355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7637387140260893e-07,
+      "loss": 1.5288,
+      "step": 187360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7614344010370903e-07,
+      "loss": 1.2951,
+      "step": 187365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7591306319115016e-07,
+      "loss": 1.2036,
+      "step": 187370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7568274066625343e-07,
+      "loss": 1.0255,
+      "step": 187375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.754524725303311e-07,
+      "loss": 1.1536,
+      "step": 187380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7522225878469553e-07,
+      "loss": 1.4055,
+      "step": 187385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.749920994306667e-07,
+      "loss": 1.3067,
+      "step": 187390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7476199446955365e-07,
+      "loss": 1.1612,
+      "step": 187395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.745319439026741e-07,
+      "loss": 1.3989,
+      "step": 187400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.743019477313393e-07,
+      "loss": 1.3125,
+      "step": 187405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.740720059568649e-07,
+      "loss": 0.9561,
+      "step": 187410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7384211858056196e-07,
+      "loss": 1.4516,
+      "step": 187415
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7361228560374396e-07,
+      "loss": 1.6182,
+      "step": 187420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7338250702772424e-07,
+      "loss": 1.4111,
+      "step": 187425
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7315278285381515e-07,
+      "loss": 1.3703,
+      "step": 187430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.729231130833267e-07,
+      "loss": 1.0804,
+      "step": 187435
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.726934977175712e-07,
+      "loss": 1.2233,
+      "step": 187440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7246393675786094e-07,
+      "loss": 1.4318,
+      "step": 187445
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.722344302055049e-07,
+      "loss": 1.5859,
+      "step": 187450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7200497806181636e-07,
+      "loss": 1.0928,
+      "step": 187455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.717755803281043e-07,
+      "loss": 1.3909,
+      "step": 187460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.715462370056767e-07,
+      "loss": 1.4619,
+      "step": 187465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.713169480958446e-07,
+      "loss": 1.3566,
+      "step": 187470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.710877135999192e-07,
+      "loss": 1.1568,
+      "step": 187475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7085853351920843e-07,
+      "loss": 1.0007,
+      "step": 187480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7062940785502e-07,
+      "loss": 1.5723,
+      "step": 187485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.704003366086618e-07,
+      "loss": 1.219,
+      "step": 187490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.70171319781445e-07,
+      "loss": 1.3151,
+      "step": 187495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.699423573746742e-07,
+      "loss": 1.0333,
+      "step": 187500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6971344938966047e-07,
+      "loss": 1.0857,
+      "step": 187505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6948459582770613e-07,
+      "loss": 1.3291,
+      "step": 187510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.692557966901223e-07,
+      "loss": 1.3185,
+      "step": 187515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6902705197821364e-07,
+      "loss": 0.9416,
+      "step": 187520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6879836169328784e-07,
+      "loss": 1.0753,
+      "step": 187525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.685697258366484e-07,
+      "loss": 1.1607,
+      "step": 187530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.68341144409602e-07,
+      "loss": 1.1852,
+      "step": 187535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6811261741345535e-07,
+      "loss": 1.0491,
+      "step": 187540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6788414484951304e-07,
+      "loss": 1.3455,
+      "step": 187545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6765572671907735e-07,
+      "loss": 1.2779,
+      "step": 187550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6742736302345494e-07,
+      "loss": 1.4133,
+      "step": 187555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6719905376395035e-07,
+      "loss": 0.9396,
+      "step": 187560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.669707989418659e-07,
+      "loss": 1.4541,
+      "step": 187565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.667425985585061e-07,
+      "loss": 0.9536,
+      "step": 187570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6651445261517326e-07,
+      "loss": 0.9811,
+      "step": 187575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.662863611131696e-07,
+      "loss": 1.134,
+      "step": 187580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.660583240537986e-07,
+      "loss": 1.2471,
+      "step": 187585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.658303414383636e-07,
+      "loss": 1.0671,
+      "step": 187590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.656024132681636e-07,
+      "loss": 1.1309,
+      "step": 187595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6537453954450416e-07,
+      "loss": 1.4454,
+      "step": 187600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6514672026868215e-07,
+      "loss": 1.1216,
+      "step": 187605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6491895544200305e-07,
+      "loss": 1.3119,
+      "step": 187610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6469124506576256e-07,
+      "loss": 1.3495,
+      "step": 187615
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.644635891412663e-07,
+      "loss": 1.4913,
+      "step": 187620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.642359876698088e-07,
+      "loss": 1.2767,
+      "step": 187625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6400844065269346e-07,
+      "loss": 1.5143,
+      "step": 187630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.637809480912203e-07,
+      "loss": 1.1397,
+      "step": 187635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6355350998668725e-07,
+      "loss": 1.1951,
+      "step": 187640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6332612634039097e-07,
+      "loss": 1.2576,
+      "step": 187645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.630987971536327e-07,
+      "loss": 1.1734,
+      "step": 187650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6287152242771027e-07,
+      "loss": 1.2341,
+      "step": 187655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.626443021639193e-07,
+      "loss": 1.157,
+      "step": 187660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.62417136363561e-07,
+      "loss": 1.1289,
+      "step": 187665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.62190025027931e-07,
+      "loss": 1.1109,
+      "step": 187670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6196296815832264e-07,
+      "loss": 1.0342,
+      "step": 187675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.617359657560394e-07,
+      "loss": 1.3136,
+      "step": 187680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6150901782237247e-07,
+      "loss": 1.6629,
+      "step": 187685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6128212435862077e-07,
+      "loss": 1.2604,
+      "step": 187690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.610552853660766e-07,
+      "loss": 0.961,
+      "step": 187695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.608285008460367e-07,
+      "loss": 1.339,
+      "step": 187700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6060177079979895e-07,
+      "loss": 0.967,
+      "step": 187705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.603750952286534e-07,
+      "loss": 0.9568,
+      "step": 187710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6014847413390016e-07,
+      "loss": 1.0001,
+      "step": 187715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.599219075168271e-07,
+      "loss": 1.2478,
+      "step": 187720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.59695395378732e-07,
+      "loss": 0.9996,
+      "step": 187725
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.594689377209083e-07,
+      "loss": 1.2908,
+      "step": 187730
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.592425345446494e-07,
+      "loss": 1.2122,
+      "step": 187735
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.590161858512443e-07,
+      "loss": 1.0353,
+      "step": 187740
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5878989164198975e-07,
+      "loss": 1.1978,
+      "step": 187745
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.585636519181769e-07,
+      "loss": 1.2085,
+      "step": 187750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.58337466681098e-07,
+      "loss": 1.5265,
+      "step": 187755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.581113359320433e-07,
+      "loss": 0.935,
+      "step": 187760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.57885259672306e-07,
+      "loss": 1.5027,
+      "step": 187765
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.576592379031741e-07,
+      "loss": 1.3226,
+      "step": 187770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.57433270625941e-07,
+      "loss": 1.2974,
+      "step": 187775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5720735784189786e-07,
+      "loss": 1.1639,
+      "step": 187780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.569814995523325e-07,
+      "loss": 1.188,
+      "step": 187785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5675569575853507e-07,
+      "loss": 1.6723,
+      "step": 187790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5652994646179563e-07,
+      "loss": 1.2384,
+      "step": 187795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.563042516634042e-07,
+      "loss": 1.1901,
+      "step": 187800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.560786113646465e-07,
+      "loss": 1.074,
+      "step": 187805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5585302556681477e-07,
+      "loss": 1.0804,
+      "step": 187810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5562749427119467e-07,
+      "loss": 1.2562,
+      "step": 187815
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.554020174790752e-07,
+      "loss": 1.1361,
+      "step": 187820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.551765951917453e-07,
+      "loss": 1.3261,
+      "step": 187825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5495122741048946e-07,
+      "loss": 1.1648,
+      "step": 187830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.547259141365956e-07,
+      "loss": 1.1438,
+      "step": 187835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5450065537135046e-07,
+      "loss": 1.2911,
+      "step": 187840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5427545111604296e-07,
+      "loss": 1.222,
+      "step": 187845
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.540503013719555e-07,
+      "loss": 0.8942,
+      "step": 187850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5382520614037476e-07,
+      "loss": 1.3228,
+      "step": 187855
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.536001654225852e-07,
+      "loss": 1.669,
+      "step": 187860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5337517921987597e-07,
+      "loss": 1.3423,
+      "step": 187865
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.53150247533527e-07,
+      "loss": 1.1089,
+      "step": 187870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.529253703648262e-07,
+      "loss": 1.8098,
+      "step": 187875
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.52700547715057e-07,
+      "loss": 1.2631,
+      "step": 187880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.524757795855017e-07,
+      "loss": 1.1937,
+      "step": 187885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.522510659774437e-07,
+      "loss": 1.3604,
+      "step": 187890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.520264068921698e-07,
+      "loss": 0.9927,
+      "step": 187895
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5180180233095894e-07,
+      "loss": 1.0173,
+      "step": 187900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5157725229509565e-07,
+      "loss": 1.4645,
+      "step": 187905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.513527567858611e-07,
+      "loss": 1.0613,
+      "step": 187910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5112831580453986e-07,
+      "loss": 1.0909,
+      "step": 187915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.509039293524109e-07,
+      "loss": 1.2021,
+      "step": 187920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.506795974307565e-07,
+      "loss": 0.906,
+      "step": 187925
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.504553200408568e-07,
+      "loss": 1.4166,
+      "step": 187930
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5023109718399407e-07,
+      "loss": 1.1201,
+      "step": 187935
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.5000692886144947e-07,
+      "loss": 1.216,
+      "step": 187940
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4978281507450093e-07,
+      "loss": 1.2434,
+      "step": 187945
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4955875582442854e-07,
+      "loss": 1.2944,
+      "step": 187950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4933475111251233e-07,
+      "loss": 1.138,
+      "step": 187955
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4911080094003247e-07,
+      "loss": 1.2144,
+      "step": 187960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.488869053082656e-07,
+      "loss": 1.3355,
+      "step": 187965
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4866306421849304e-07,
+      "loss": 1.2368,
+      "step": 187970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4843927767199256e-07,
+      "loss": 1.6502,
+      "step": 187975
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4821554567003875e-07,
+      "loss": 1.1791,
+      "step": 187980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4799186821391174e-07,
+      "loss": 1.1845,
+      "step": 187985
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.477682453048904e-07,
+      "loss": 1.1994,
+      "step": 187990
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.475446769442493e-07,
+      "loss": 1.0534,
+      "step": 187995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.473211631332652e-07,
+      "loss": 1.5755,
+      "step": 188000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4709770387321495e-07,
+      "loss": 1.3404,
+      "step": 188005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4687429916537626e-07,
+      "loss": 1.2176,
+      "step": 188010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.466509490110227e-07,
+      "loss": 1.149,
+      "step": 188015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.464276534114309e-07,
+      "loss": 1.102,
+      "step": 188020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.462044123678744e-07,
+      "loss": 1.1037,
+      "step": 188025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.459812258816298e-07,
+      "loss": 1.1258,
+      "step": 188030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4575809395397187e-07,
+      "loss": 1.1412,
+      "step": 188035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4553501658617494e-07,
+      "loss": 1.2467,
+      "step": 188040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4531199377950916e-07,
+      "loss": 1.0808,
+      "step": 188045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4508902553525244e-07,
+      "loss": 1.3208,
+      "step": 188050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.448661118546771e-07,
+      "loss": 1.066,
+      "step": 188055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.446432527390554e-07,
+      "loss": 1.0873,
+      "step": 188060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4442044818965855e-07,
+      "loss": 1.0902,
+      "step": 188065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.441976982077623e-07,
+      "loss": 0.9747,
+      "step": 188070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4397500279463434e-07,
+      "loss": 1.6415,
+      "step": 188075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.437523619515505e-07,
+      "loss": 1.2199,
+      "step": 188080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.435297756797807e-07,
+      "loss": 1.2832,
+      "step": 188085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.433072439805963e-07,
+      "loss": 1.0929,
+      "step": 188090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4308476685526626e-07,
+      "loss": 1.2969,
+      "step": 188095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.428623443050628e-07,
+      "loss": 1.2066,
+      "step": 188100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4263997633125613e-07,
+      "loss": 1.2373,
+      "step": 188105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.424176629351151e-07,
+      "loss": 1.2482,
+      "step": 188110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.4219540411791104e-07,
+      "loss": 1.3892,
+      "step": 188115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.419731998809096e-07,
+      "loss": 1.3425,
+      "step": 188120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4175105022538413e-07,
+      "loss": 1.0955,
+      "step": 188125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.415289551526003e-07,
+      "loss": 1.3774,
+      "step": 188130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4130691466382713e-07,
+      "loss": 1.4686,
+      "step": 188135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4108492876033247e-07,
+      "loss": 0.9781,
+      "step": 188140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.408629974433842e-07,
+      "loss": 1.1571,
+      "step": 188145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4064112071425026e-07,
+      "loss": 1.1095,
+      "step": 188150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.404192985741973e-07,
+      "loss": 1.2349,
+      "step": 188155
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.401975310244899e-07,
+      "loss": 1.1715,
+      "step": 188160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.399758180663993e-07,
+      "loss": 1.1597,
+      "step": 188165
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3975415970118563e-07,
+      "loss": 1.0038,
+      "step": 188170
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.395325559301189e-07,
+      "loss": 1.1823,
+      "step": 188175
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.393110067544637e-07,
+      "loss": 1.2912,
+      "step": 188180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3908951217548567e-07,
+      "loss": 1.1882,
+      "step": 188185
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3886807219444717e-07,
+      "loss": 0.8138,
+      "step": 188190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.386466868126149e-07,
+      "loss": 1.2393,
+      "step": 188195
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.384253560312535e-07,
+      "loss": 1.245,
+      "step": 188200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3820407985162627e-07,
+      "loss": 1.1147,
+      "step": 188205
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3798285827499454e-07,
+      "loss": 1.0745,
+      "step": 188210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.377616913026239e-07,
+      "loss": 1.15,
+      "step": 188215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3754057893577894e-07,
+      "loss": 0.9997,
+      "step": 188220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3731952117571975e-07,
+      "loss": 1.2406,
+      "step": 188225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.370985180237097e-07,
+      "loss": 1.1232,
+      "step": 188230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.36877569481009e-07,
+      "loss": 1.3862,
+      "step": 188235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.36656675548881e-07,
+      "loss": 1.2921,
+      "step": 188240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.364358362285892e-07,
+      "loss": 1.0059,
+      "step": 188245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.362150515213925e-07,
+      "loss": 1.4273,
+      "step": 188250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3599432142855005e-07,
+      "loss": 1.0055,
+      "step": 188255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3577364595132396e-07,
+      "loss": 1.3422,
+      "step": 188260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3555302509097565e-07,
+      "loss": 0.9588,
+      "step": 188265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3533245884876396e-07,
+      "loss": 1.2902,
+      "step": 188270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3511194722594684e-07,
+      "loss": 1.0094,
+      "step": 188275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3489149022378664e-07,
+      "loss": 1.3784,
+      "step": 188280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.34671087843539e-07,
+      "loss": 1.4697,
+      "step": 188285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3445074008646503e-07,
+      "loss": 1.1446,
+      "step": 188290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3423044695382275e-07,
+      "loss": 1.2097,
+      "step": 188295
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3401020844687004e-07,
+      "loss": 1.8392,
+      "step": 188300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3379002456686246e-07,
+      "loss": 1.1554,
+      "step": 188305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.335698953150591e-07,
+      "loss": 1.1908,
+      "step": 188310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.333498206927178e-07,
+      "loss": 1.3599,
+      "step": 188315
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.331298007010931e-07,
+      "loss": 1.445,
+      "step": 188320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3290983534144514e-07,
+      "loss": 1.1,
+      "step": 188325
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.326899246150262e-07,
+      "loss": 1.0457,
+      "step": 188330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3247006852309313e-07,
+      "loss": 1.1305,
+      "step": 188335
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.322502670669038e-07,
+      "loss": 1.3164,
+      "step": 188340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.320305202477104e-07,
+      "loss": 1.2184,
+      "step": 188345
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.318108280667699e-07,
+      "loss": 1.0268,
+      "step": 188350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3159119052533226e-07,
+      "loss": 1.2143,
+      "step": 188355
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.313716076246588e-07,
+      "loss": 1.0527,
+      "step": 188360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3115207936599957e-07,
+      "loss": 1.0717,
+      "step": 188365
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.309326057506069e-07,
+      "loss": 1.1043,
+      "step": 188370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.307131867797376e-07,
+      "loss": 1.4038,
+      "step": 188375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.304938224546418e-07,
+      "loss": 1.1244,
+      "step": 188380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3027451277657284e-07,
+      "loss": 1.2129,
+      "step": 188385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3005525774678535e-07,
+      "loss": 1.1773,
+      "step": 188390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2983605736652945e-07,
+      "loss": 1.0932,
+      "step": 188395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.296169116370552e-07,
+      "loss": 1.2625,
+      "step": 188400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.293978205596161e-07,
+      "loss": 1.1827,
+      "step": 188405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.291787841354644e-07,
+      "loss": 1.0685,
+      "step": 188410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.289598023658492e-07,
+      "loss": 1.2598,
+      "step": 188415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.287408752520206e-07,
+      "loss": 1.5416,
+      "step": 188420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.285220027952286e-07,
+      "loss": 1.2936,
+      "step": 188425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2830318499672675e-07,
+      "loss": 1.2857,
+      "step": 188430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2808442185775956e-07,
+      "loss": 1.0875,
+      "step": 188435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.278657133795794e-07,
+      "loss": 1.1224,
+      "step": 188440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.276470595634341e-07,
+      "loss": 0.9456,
+      "step": 188445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2742846041057273e-07,
+      "loss": 1.2845,
+      "step": 188450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2720991592224427e-07,
+      "loss": 1.3465,
+      "step": 188455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.269914260996955e-07,
+      "loss": 1.2668,
+      "step": 188460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2677299094417423e-07,
+      "loss": 1.1274,
+      "step": 188465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.265546104569285e-07,
+      "loss": 1.4063,
+      "step": 188470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2633628463920384e-07,
+      "loss": 1.0614,
+      "step": 188475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2611801349224933e-07,
+      "loss": 1.4516,
+      "step": 188480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.258997970173107e-07,
+      "loss": 1.0536,
+      "step": 188485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2568163521563345e-07,
+      "loss": 1.2439,
+      "step": 188490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2546352808846337e-07,
+      "loss": 1.0723,
+      "step": 188495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.25245475637045e-07,
+      "loss": 1.1838,
+      "step": 188500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2502747786262624e-07,
+      "loss": 1.0379,
+      "step": 188505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.248095347664516e-07,
+      "loss": 1.4743,
+      "step": 188510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2459164634976236e-07,
+      "loss": 1.0593,
+      "step": 188515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.243738126138053e-07,
+      "loss": 1.3697,
+      "step": 188520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2415603355982495e-07,
+      "loss": 1.3394,
+      "step": 188525
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2393830918906255e-07,
+      "loss": 1.1429,
+      "step": 188530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2372063950276485e-07,
+      "loss": 1.027,
+      "step": 188535
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2350302450217207e-07,
+      "loss": 1.1143,
+      "step": 188540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2328546418852644e-07,
+      "loss": 1.4944,
+      "step": 188545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.230679585630737e-07,
+      "loss": 1.2113,
+      "step": 188550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2285050762705393e-07,
+      "loss": 1.0868,
+      "step": 188555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.226331113817084e-07,
+      "loss": 1.1196,
+      "step": 188560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2241576982827715e-07,
+      "loss": 1.0795,
+      "step": 188565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2219848296800594e-07,
+      "loss": 1.3819,
+      "step": 188570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2198125080213147e-07,
+      "loss": 1.3068,
+      "step": 188575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.217640733318951e-07,
+      "loss": 1.01,
+      "step": 188580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.21546950558539e-07,
+      "loss": 1.2685,
+      "step": 188585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.213298824833001e-07,
+      "loss": 1.1391,
+      "step": 188590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.211128691074196e-07,
+      "loss": 1.2412,
+      "step": 188595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2089591043213864e-07,
+      "loss": 0.8764,
+      "step": 188600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.20679006458693e-07,
+      "loss": 1.3009,
+      "step": 188605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.204621571883216e-07,
+      "loss": 1.2652,
+      "step": 188610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.202453626222636e-07,
+      "loss": 1.1463,
+      "step": 188615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.200286227617589e-07,
+      "loss": 1.3054,
+      "step": 188620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1981193760804116e-07,
+      "loss": 1.3096,
+      "step": 188625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1959530716235265e-07,
+      "loss": 1.0554,
+      "step": 188630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.193787314259257e-07,
+      "loss": 1.1977,
+      "step": 188635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.191622104000004e-07,
+      "loss": 1.0813,
+      "step": 188640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1894574408581026e-07,
+      "loss": 1.0581,
+      "step": 188645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.187293324845942e-07,
+      "loss": 1.0669,
+      "step": 188650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.185129755975881e-07,
+      "loss": 1.1856,
+      "step": 188655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.18296673426023e-07,
+      "loss": 1.2317,
+      "step": 188660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1808042597114016e-07,
+      "loss": 1.5452,
+      "step": 188665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.17864233234172e-07,
+      "loss": 1.02,
+      "step": 188670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1764809521635086e-07,
+      "loss": 1.207,
+      "step": 188675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.174320119189146e-07,
+      "loss": 1.6828,
+      "step": 188680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.172159833430944e-07,
+      "loss": 1.1079,
+      "step": 188685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.17000009490125e-07,
+      "loss": 1.3699,
+      "step": 188690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.167840903612419e-07,
+      "loss": 1.1884,
+      "step": 188695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.165682259576753e-07,
+      "loss": 1.2036,
+      "step": 188700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.163524162806565e-07,
+      "loss": 1.4509,
+      "step": 188705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1613666133142105e-07,
+      "loss": 1.3066,
+      "step": 188710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1592096111120137e-07,
+      "loss": 1.4177,
+      "step": 188715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.157053156212265e-07,
+      "loss": 0.9521,
+      "step": 188720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1548972486272987e-07,
+      "loss": 1.2009,
+      "step": 188725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.152741888369405e-07,
+      "loss": 1.0034,
+      "step": 188730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1505870754509183e-07,
+      "loss": 1.3339,
+      "step": 188735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1484328098841287e-07,
+      "loss": 1.3853,
+      "step": 188740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.146279091681349e-07,
+      "loss": 1.3505,
+      "step": 188745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.14412592085488e-07,
+      "loss": 1.4571,
+      "step": 188750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.141973297416968e-07,
+      "loss": 1.0834,
+      "step": 188755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1398212213799805e-07,
+      "loss": 1.0907,
+      "step": 188760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.137669692756174e-07,
+      "loss": 1.5299,
+      "step": 188765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1355187115578175e-07,
+      "loss": 1.0945,
+      "step": 188770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.133368277797223e-07,
+      "loss": 1.2608,
+      "step": 188775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1312183914866353e-07,
+      "loss": 1.0788,
+      "step": 188780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.129069052638379e-07,
+      "loss": 1.1602,
+      "step": 188785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1269202612646886e-07,
+      "loss": 0.9851,
+      "step": 188790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.124772017377865e-07,
+      "loss": 1.5427,
+      "step": 188795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.122624320990143e-07,
+      "loss": 1.4719,
+      "step": 188800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1204771721138016e-07,
+      "loss": 1.126,
+      "step": 188805
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1183305707611197e-07,
+      "loss": 1.1995,
+      "step": 188810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1161845169443326e-07,
+      "loss": 1.3649,
+      "step": 188815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.114039010675697e-07,
+      "loss": 0.8768,
+      "step": 188820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.11189405196748e-07,
+      "loss": 1.2647,
+      "step": 188825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.109749640831917e-07,
+      "loss": 1.2311,
+      "step": 188830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.107605777281254e-07,
+      "loss": 1.4291,
+      "step": 188835
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.105462461327758e-07,
+      "loss": 1.2415,
+      "step": 188840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1033196929836316e-07,
+      "loss": 1.5338,
+      "step": 188845
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1011774722611307e-07,
+      "loss": 1.4242,
+      "step": 188850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0990357991724795e-07,
+      "loss": 1.2068,
+      "step": 188855
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0968946737299343e-07,
+      "loss": 1.0385,
+      "step": 188860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0947540959456965e-07,
+      "loss": 1.2452,
+      "step": 188865
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.092614065831979e-07,
+      "loss": 1.2744,
+      "step": 188870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0904745834010275e-07,
+      "loss": 1.106,
+      "step": 188875
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.088335648665054e-07,
+      "loss": 1.8589,
+      "step": 188880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.08619726163626e-07,
+      "loss": 1.2167,
+      "step": 188885
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0840594223268915e-07,
+      "loss": 1.1165,
+      "step": 188890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.081922130749105e-07,
+      "loss": 1.1513,
+      "step": 188895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.079785386915136e-07,
+      "loss": 1.1656,
+      "step": 188900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0776491908372073e-07,
+      "loss": 1.5461,
+      "step": 188905
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.075513542527476e-07,
+      "loss": 1.2253,
+      "step": 188910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.073378441998155e-07,
+      "loss": 1.1033,
+      "step": 188915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.071243889261434e-07,
+      "loss": 1.2874,
+      "step": 188920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.069109884329525e-07,
+      "loss": 1.2187,
+      "step": 188925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.066976427214586e-07,
+      "loss": 1.1963,
+      "step": 188930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0648435179287957e-07,
+      "loss": 1.095,
+      "step": 188935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0627111564843556e-07,
+      "loss": 1.2194,
+      "step": 188940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.060579342893445e-07,
+      "loss": 1.4409,
+      "step": 188945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.058448077168209e-07,
+      "loss": 1.2699,
+      "step": 188950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.05631735932086e-07,
+      "loss": 1.078,
+      "step": 188955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.054187189363534e-07,
+      "loss": 1.1155,
+      "step": 188960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0520575673083874e-07,
+      "loss": 0.8619,
+      "step": 188965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.04992849316761e-07,
+      "loss": 1.1334,
+      "step": 188970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0477999669533585e-07,
+      "loss": 1.1534,
+      "step": 188975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0456719886777575e-07,
+      "loss": 1.1927,
+      "step": 188980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0435445583529964e-07,
+      "loss": 1.1349,
+      "step": 188985
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.041417675991199e-07,
+      "loss": 1.2636,
+      "step": 188990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.039291341604523e-07,
+      "loss": 1.2189,
+      "step": 188995
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.037165555205091e-07,
+      "loss": 1.2074,
+      "step": 189000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.035040316805072e-07,
+      "loss": 1.5162,
+      "step": 189005
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.032915626416578e-07,
+      "loss": 1.16,
+      "step": 189010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0307914840517547e-07,
+      "loss": 1.295,
+      "step": 189015
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.028667889722737e-07,
+      "loss": 1.0284,
+      "step": 189020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.026544843441649e-07,
+      "loss": 1.1191,
+      "step": 189025
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.024422345220591e-07,
+      "loss": 1.6252,
+      "step": 189030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.022300395071699e-07,
+      "loss": 0.996,
+      "step": 189035
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.020178993007118e-07,
+      "loss": 1.0956,
+      "step": 189040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.018058139038916e-07,
+      "loss": 0.8901,
+      "step": 189045
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0159378331792397e-07,
+      "loss": 1.0668,
+      "step": 189050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.01381807544019e-07,
+      "loss": 1.4875,
+      "step": 189055
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.011698865833835e-07,
+      "loss": 1.413,
+      "step": 189060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.009580204372332e-07,
+      "loss": 0.8934,
+      "step": 189065
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0074620910677485e-07,
+      "loss": 1.0351,
+      "step": 189070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.005344525932198e-07,
+      "loss": 1.3744,
+      "step": 189075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.003227508977747e-07,
+      "loss": 1.3546,
+      "step": 189080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0011110402164877e-07,
+      "loss": 1.0658,
+      "step": 189085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9989951196605426e-07,
+      "loss": 1.4795,
+      "step": 189090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9968797473219467e-07,
+      "loss": 1.2225,
+      "step": 189095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9947649232128236e-07,
+      "loss": 1.45,
+      "step": 189100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.99265064734522e-07,
+      "loss": 1.0726,
+      "step": 189105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.990536919731214e-07,
+      "loss": 1.2777,
+      "step": 189110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.988423740382896e-07,
+      "loss": 1.2339,
+      "step": 189115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.986311109312313e-07,
+      "loss": 1.3415,
+      "step": 189120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.984199026531521e-07,
+      "loss": 1.0624,
+      "step": 189125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9820874920526e-07,
+      "loss": 1.1491,
+      "step": 189130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9799765058876174e-07,
+      "loss": 1.2172,
+      "step": 189135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.977866068048608e-07,
+      "loss": 1.0914,
+      "step": 189140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9757561785476183e-07,
+      "loss": 1.2546,
+      "step": 189145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9736468373967274e-07,
+      "loss": 0.9926,
+      "step": 189150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9715380446079367e-07,
+      "loss": 1.4011,
+      "step": 189155
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9694298001933254e-07,
+      "loss": 1.1827,
+      "step": 189160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9673221041649166e-07,
+      "loss": 1.1931,
+      "step": 189165
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9652149565347573e-07,
+      "loss": 1.0214,
+      "step": 189170
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.963108357314871e-07,
+      "loss": 1.3725,
+      "step": 189175
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.961002306517281e-07,
+      "loss": 1.1785,
+      "step": 189180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.958896804154033e-07,
+      "loss": 1.2447,
+      "step": 189185
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9567918502371297e-07,
+      "loss": 1.4722,
+      "step": 189190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9546874447786156e-07,
+      "loss": 1.1772,
+      "step": 189195
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9525835877904816e-07,
+      "loss": 1.4337,
+      "step": 189200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.950480279284752e-07,
+      "loss": 1.1311,
+      "step": 189205
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9483775192734496e-07,
+      "loss": 1.1354,
+      "step": 189210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.946275307768588e-07,
+      "loss": 1.0905,
+      "step": 189215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9441736447821344e-07,
+      "loss": 1.1209,
+      "step": 189220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.942072530326113e-07,
+      "loss": 1.6515,
+      "step": 189225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9399719644125366e-07,
+      "loss": 1.0802,
+      "step": 189230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9378719470533844e-07,
+      "loss": 1.1668,
+      "step": 189235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9357724782606356e-07,
+      "loss": 1.1352,
+      "step": 189240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9336735580463023e-07,
+      "loss": 1.0168,
+      "step": 189245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.931575186422365e-07,
+      "loss": 1.0249,
+      "step": 189250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.929477363400802e-07,
+      "loss": 1.133,
+      "step": 189255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.927380088993604e-07,
+      "loss": 1.2129,
+      "step": 189260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9252833632127285e-07,
+      "loss": 1.1845,
+      "step": 189265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9231871860701654e-07,
+      "loss": 0.948,
+      "step": 189270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9210915575778605e-07,
+      "loss": 1.3941,
+      "step": 189275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9189964777478273e-07,
+      "loss": 1.3058,
+      "step": 189280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.916901946591989e-07,
+      "loss": 1.1673,
+      "step": 189285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9148079641223133e-07,
+      "loss": 1.1816,
+      "step": 189290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9127145303507586e-07,
+      "loss": 1.3269,
+      "step": 189295
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9106216452893033e-07,
+      "loss": 1.2449,
+      "step": 189300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.90852930894986e-07,
+      "loss": 1.3885,
+      "step": 189305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9064375213444193e-07,
+      "loss": 0.9586,
+      "step": 189310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9043462824848833e-07,
+      "loss": 1.1393,
+      "step": 189315
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9022555923832194e-07,
+      "loss": 1.1712,
+      "step": 189320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9001654510513744e-07,
+      "loss": 1.0362,
+      "step": 189325
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.898075858501271e-07,
+      "loss": 1.0849,
+      "step": 189330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.895986814744834e-07,
+      "loss": 1.1061,
+      "step": 189335
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8938983197940095e-07,
+      "loss": 1.1101,
+      "step": 189340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.891810373660709e-07,
+      "loss": 1.5391,
+      "step": 189345
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.889722976356869e-07,
+      "loss": 1.2661,
+      "step": 189350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.887636127894412e-07,
+      "loss": 1.3315,
+      "step": 189355
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.88554982828524e-07,
+      "loss": 1.318,
+      "step": 189360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8834640775412656e-07,
+      "loss": 1.2074,
+      "step": 189365
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.881378875674402e-07,
+      "loss": 1.235,
+      "step": 189370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8792942226965837e-07,
+      "loss": 1.0923,
+      "step": 189375
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.87721011861969e-07,
+      "loss": 1.3648,
+      "step": 189380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8751265634556224e-07,
+      "loss": 1.1773,
+      "step": 189385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.873043557216272e-07,
+      "loss": 1.24,
+      "step": 189390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.870961099913562e-07,
+      "loss": 1.2407,
+      "step": 189395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.868879191559349e-07,
+      "loss": 0.9431,
+      "step": 189400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8667978321655586e-07,
+      "loss": 1.389,
+      "step": 189405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8647170217440464e-07,
+      "loss": 1.1743,
+      "step": 189410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.862636760306704e-07,
+      "loss": 1.0844,
+      "step": 189415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.860557047865432e-07,
+      "loss": 0.9096,
+      "step": 189420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.858477884432077e-07,
+      "loss": 1.0597,
+      "step": 189425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8563992700185183e-07,
+      "loss": 1.1989,
+      "step": 189430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.854321204636635e-07,
+      "loss": 1.0296,
+      "step": 189435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.852243688298296e-07,
+      "loss": 1.1537,
+      "step": 189440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8501667210153584e-07,
+      "loss": 1.2691,
+      "step": 189445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.848090302799667e-07,
+      "loss": 1.2038,
+      "step": 189450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8460144336631145e-07,
+      "loss": 1.215,
+      "step": 189455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8439391136175233e-07,
+      "loss": 1.9917,
+      "step": 189460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8418643426747504e-07,
+      "loss": 1.0016,
+      "step": 189465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.839790120846676e-07,
+      "loss": 1.6601,
+      "step": 189470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8377164481451126e-07,
+      "loss": 1.3604,
+      "step": 189475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8356433245818947e-07,
+      "loss": 1.1686,
+      "step": 189480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8335707501688803e-07,
+      "loss": 0.9196,
+      "step": 189485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8314987249179146e-07,
+      "loss": 1.4438,
+      "step": 189490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8294272488408005e-07,
+      "loss": 1.1974,
+      "step": 189495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.827356321949394e-07,
+      "loss": 1.319,
+      "step": 189500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8252859442554857e-07,
+      "loss": 0.9664,
+      "step": 189505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.823216115770956e-07,
+      "loss": 1.6555,
+      "step": 189510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.821146836507561e-07,
+      "loss": 1.1291,
+      "step": 189515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.81907810647717e-07,
+      "loss": 1.1802,
+      "step": 189520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8170099256915615e-07,
+      "loss": 1.4693,
+      "step": 189525
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.81494229416256e-07,
+      "loss": 1.1087,
+      "step": 189530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8128752119019786e-07,
+      "loss": 1.1053,
+      "step": 189535
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.810808678921618e-07,
+      "loss": 1.3171,
+      "step": 189540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8087426952332583e-07,
+      "loss": 1.363,
+      "step": 189545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8066772608487126e-07,
+      "loss": 1.0267,
+      "step": 189550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.804612375779793e-07,
+      "loss": 1.4801,
+      "step": 189555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.802548040038267e-07,
+      "loss": 1.1252,
+      "step": 189560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.800484253635939e-07,
+      "loss": 1.1985,
+      "step": 189565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.798421016584597e-07,
+      "loss": 0.9984,
+      "step": 189570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7963583288959884e-07,
+      "loss": 1.0906,
+      "step": 189575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7942961905819254e-07,
+      "loss": 1.439,
+      "step": 189580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.792234601654188e-07,
+      "loss": 1.22,
+      "step": 189585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.790173562124533e-07,
+      "loss": 1.1297,
+      "step": 189590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7881130720047175e-07,
+      "loss": 1.4354,
+      "step": 189595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.786053131306533e-07,
+      "loss": 1.1855,
+      "step": 189600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.783993740041736e-07,
+      "loss": 1.0629,
+      "step": 189605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.781934898222084e-07,
+      "loss": 0.9802,
+      "step": 189610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.779876605859334e-07,
+      "loss": 1.1934,
+      "step": 189615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7778188629652437e-07,
+      "loss": 1.1544,
+      "step": 189620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7757616695515473e-07,
+      "loss": 0.91,
+      "step": 189625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7737050256300254e-07,
+      "loss": 1.1601,
+      "step": 189630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7716489312124015e-07,
+      "loss": 1.1878,
+      "step": 189635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7695933863104107e-07,
+      "loss": 1.2757,
+      "step": 189640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7675383909357987e-07,
+      "loss": 0.951,
+      "step": 189645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.765483945100312e-07,
+      "loss": 1.5917,
+      "step": 189650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.763430048815664e-07,
+      "loss": 1.3694,
+      "step": 189655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7613767020935885e-07,
+      "loss": 1.1874,
+      "step": 189660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.759323904945822e-07,
+      "loss": 1.3389,
+      "step": 189665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.757271657384076e-07,
+      "loss": 1.2318,
+      "step": 189670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7552199594200646e-07,
+      "loss": 1.1296,
+      "step": 189675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7531688110655216e-07,
+      "loss": 1.3493,
+      "step": 189680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.751118212332161e-07,
+      "loss": 1.1847,
+      "step": 189685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7490681632316616e-07,
+      "loss": 1.3772,
+      "step": 189690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.747018663775759e-07,
+      "loss": 1.1262,
+      "step": 189695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.744969713976154e-07,
+      "loss": 1.209,
+      "step": 189700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7429213138445385e-07,
+      "loss": 1.1382,
+      "step": 189705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7408734633926247e-07,
+      "loss": 1.1548,
+      "step": 189710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.738826162632081e-07,
+      "loss": 1.0162,
+      "step": 189715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7367794115746315e-07,
+      "loss": 1.3748,
+      "step": 189720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7347332102319335e-07,
+      "loss": 1.0759,
+      "step": 189725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7326875586157e-07,
+      "loss": 1.0379,
+      "step": 189730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.730642456737588e-07,
+      "loss": 1.2235,
+      "step": 189735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.728597904609277e-07,
+      "loss": 0.9972,
+      "step": 189740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.726553902242469e-07,
+      "loss": 1.0993,
+      "step": 189745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7245104496488107e-07,
+      "loss": 1.4296,
+      "step": 189750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7224675468399805e-07,
+      "loss": 1.1287,
+      "step": 189755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.720425193827648e-07,
+      "loss": 1.3074,
+      "step": 189760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.718383390623459e-07,
+      "loss": 1.1603,
+      "step": 189765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.716342137239093e-07,
+      "loss": 1.1943,
+      "step": 189770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.714301433686207e-07,
+      "loss": 1.5533,
+      "step": 189775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.712261279976448e-07,
+      "loss": 1.1172,
+      "step": 189780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7102216761214507e-07,
+      "loss": 1.1808,
+      "step": 189785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.708182622132883e-07,
+      "loss": 1.0567,
+      "step": 189790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.706144118022392e-07,
+      "loss": 1.2506,
+      "step": 189795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.704106163801613e-07,
+      "loss": 0.9803,
+      "step": 189800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.702068759482158e-07,
+      "loss": 0.9305,
+      "step": 189805
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.700031905075696e-07,
+      "loss": 1.4475,
+      "step": 189810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.697995600593851e-07,
+      "loss": 1.1185,
+      "step": 189815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.695959846048247e-07,
+      "loss": 1.0643,
+      "step": 189820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.693924641450508e-07,
+      "loss": 1.5846,
+      "step": 189825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6918899868122583e-07,
+      "loss": 1.1872,
+      "step": 189830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6898558821451105e-07,
+      "loss": 1.1083,
+      "step": 189835
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6878223274606995e-07,
+      "loss": 1.5095,
+      "step": 189840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6857893227706164e-07,
+      "loss": 1.3742,
+      "step": 189845
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.683756868086474e-07,
+      "loss": 1.3149,
+      "step": 189850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.681724963419897e-07,
+      "loss": 1.0952,
+      "step": 189855
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.679693608782464e-07,
+      "loss": 1.2994,
+      "step": 189860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6776628041858e-07,
+      "loss": 1.5355,
+      "step": 189865
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6756325496414726e-07,
+      "loss": 1.1412,
+      "step": 189870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.673602845161106e-07,
+      "loss": 1.2329,
+      "step": 189875
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.671573690756258e-07,
+      "loss": 0.8764,
+      "step": 189880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.669545086438542e-07,
+      "loss": 1.2338,
+      "step": 189885
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.667517032219547e-07,
+      "loss": 1.3413,
+      "step": 189890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6654895281108436e-07,
+      "loss": 1.072,
+      "step": 189895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.663462574123988e-07,
+      "loss": 0.8859,
+      "step": 189900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6614361702705827e-07,
+      "loss": 1.2538,
+      "step": 189905
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6594103165622066e-07,
+      "loss": 1.1907,
+      "step": 189910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.657385013010395e-07,
+      "loss": 1.2692,
+      "step": 189915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6553602596267503e-07,
+      "loss": 1.0364,
+      "step": 189920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.653336056422807e-07,
+      "loss": 1.1231,
+      "step": 189925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6513124034101344e-07,
+      "loss": 1.2287,
+      "step": 189930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6492893006003003e-07,
+      "loss": 0.9915,
+      "step": 189935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6472667480048406e-07,
+      "loss": 1.2788,
+      "step": 189940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6452447456353236e-07,
+      "loss": 1.2613,
+      "step": 189945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.643223293503251e-07,
+      "loss": 0.956,
+      "step": 189950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6412023916202245e-07,
+      "loss": 1.0316,
+      "step": 189955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6391820399977574e-07,
+      "loss": 1.1997,
+      "step": 189960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6371622386473846e-07,
+      "loss": 1.414,
+      "step": 189965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.635142987580642e-07,
+      "loss": 1.8463,
+      "step": 189970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6331242868090645e-07,
+      "loss": 1.1371,
+      "step": 189975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.631106136344176e-07,
+      "loss": 1.1508,
+      "step": 189980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6290885361975117e-07,
+      "loss": 1.1583,
+      "step": 189985
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6270714863805846e-07,
+      "loss": 1.1399,
+      "step": 189990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.625054986904908e-07,
+      "loss": 1.0703,
+      "step": 189995
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.623039037781995e-07,
+      "loss": 0.9066,
+      "step": 190000
+    },
+    {
+      "epoch": 0.91,
+      "eval_loss": 1.22099769115448,
+      "eval_runtime": 6466.515,
+      "eval_samples_per_second": 3.574,
+      "eval_steps_per_second": 1.787,
+      "step": 190000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.62102363902338e-07,
+      "loss": 1.5463,
+      "step": 190005
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.619008790640566e-07,
+      "loss": 1.4156,
+      "step": 190010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6169944926450317e-07,
+      "loss": 1.1033,
+      "step": 190015
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.614980745048302e-07,
+      "loss": 1.0627,
+      "step": 190020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.612967547861879e-07,
+      "loss": 0.9883,
+      "step": 190025
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.610954901097241e-07,
+      "loss": 1.2954,
+      "step": 190030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.6089428047659136e-07,
+      "loss": 1.2469,
+      "step": 190035
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.606931258879365e-07,
+      "loss": 1.0682,
+      "step": 190040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.604920263449052e-07,
+      "loss": 1.485,
+      "step": 190045
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.602909818486522e-07,
+      "loss": 1.039,
+      "step": 190050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.600899924003221e-07,
+      "loss": 1.5338,
+      "step": 190055
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5988905800106165e-07,
+      "loss": 1.0923,
+      "step": 190060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.596881786520212e-07,
+      "loss": 1.2695,
+      "step": 190065
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5948735435434536e-07,
+      "loss": 1.4023,
+      "step": 190070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.59286585109182e-07,
+      "loss": 1.2562,
+      "step": 190075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.590858709176781e-07,
+      "loss": 1.2149,
+      "step": 190080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5888521178097936e-07,
+      "loss": 1.2641,
+      "step": 190085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.586846077002304e-07,
+      "loss": 1.2588,
+      "step": 190090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5848405867657807e-07,
+      "loss": 1.385,
+      "step": 190095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.582835647111682e-07,
+      "loss": 1.1799,
+      "step": 190100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5808312580514536e-07,
+      "loss": 1.0143,
+      "step": 190105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.578827419596531e-07,
+      "loss": 1.0732,
+      "step": 190110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.576824131758361e-07,
+      "loss": 1.1919,
+      "step": 190115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.574821394548389e-07,
+      "loss": 1.2986,
+      "step": 190120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.572819207978051e-07,
+      "loss": 1.2533,
+      "step": 190125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.570817572058782e-07,
+      "loss": 1.1132,
+      "step": 190130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.568816486802007e-07,
+      "loss": 1.4506,
+      "step": 190135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.566815952219149e-07,
+      "loss": 1.3331,
+      "step": 190140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5648159683216444e-07,
+      "loss": 1.0505,
+      "step": 190145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5628165351209163e-07,
+      "loss": 1.2669,
+      "step": 190150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5608176526283675e-07,
+      "loss": 1.3503,
+      "step": 190155
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5588193208554223e-07,
+      "loss": 1.1448,
+      "step": 190160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5568215398134817e-07,
+      "loss": 1.1694,
+      "step": 190165
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.554824309513971e-07,
+      "loss": 1.1709,
+      "step": 190170
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.552827629968281e-07,
+      "loss": 1.102,
+      "step": 190175
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5508315011878236e-07,
+      "loss": 1.1948,
+      "step": 190180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.5488359231839907e-07,
+      "loss": 1.3525,
+      "step": 190185
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.546840895968173e-07,
+      "loss": 1.1524,
+      "step": 190190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.544846419551795e-07,
+      "loss": 1.3594,
+      "step": 190195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.542852493946214e-07,
+      "loss": 1.3536,
+      "step": 190200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5408591191628097e-07,
+      "loss": 1.2109,
+      "step": 190205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.538866295212984e-07,
+      "loss": 1.3122,
+      "step": 190210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.536874022108128e-07,
+      "loss": 1.4662,
+      "step": 190215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.534882299859588e-07,
+      "loss": 0.9885,
+      "step": 190220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.532891128478777e-07,
+      "loss": 1.1544,
+      "step": 190225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.53090050797702e-07,
+      "loss": 0.8526,
+      "step": 190230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.528910438365718e-07,
+      "loss": 1.1904,
+      "step": 190235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5269209196562183e-07,
+      "loss": 1.2051,
+      "step": 190240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5249319518599e-07,
+      "loss": 1.1055,
+      "step": 190245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5229435349881103e-07,
+      "loss": 1.1998,
+      "step": 190250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.520955669052173e-07,
+      "loss": 1.3152,
+      "step": 190255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.518968354063501e-07,
+      "loss": 1.4871,
+      "step": 190260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.516981590033408e-07,
+      "loss": 1.0996,
+      "step": 190265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5149953769732295e-07,
+      "loss": 1.1383,
+      "step": 190270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5130097148943443e-07,
+      "loss": 1.4357,
+      "step": 190275
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.511024603808044e-07,
+      "loss": 1.1636,
+      "step": 190280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.509040043725698e-07,
+      "loss": 1.175,
+      "step": 190285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.50705603465864e-07,
+      "loss": 1.0482,
+      "step": 190290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.505072576618196e-07,
+      "loss": 1.2387,
+      "step": 190295
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.503089669615678e-07,
+      "loss": 1.2178,
+      "step": 190300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5011073136624107e-07,
+      "loss": 1.6275,
+      "step": 190305
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4991255087697405e-07,
+      "loss": 1.2969,
+      "step": 190310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4971442549489587e-07,
+      "loss": 1.1772,
+      "step": 190315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.495163552211378e-07,
+      "loss": 1.0179,
+      "step": 190320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.493183400568323e-07,
+      "loss": 1.7174,
+      "step": 190325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.491203800031118e-07,
+      "loss": 1.2709,
+      "step": 190330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.48922475061102e-07,
+      "loss": 1.1657,
+      "step": 190335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4872462523193764e-07,
+      "loss": 1.1666,
+      "step": 190340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4852683051674664e-07,
+      "loss": 1.3383,
+      "step": 190345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.483290909166559e-07,
+      "loss": 1.517,
+      "step": 190350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4813140643279895e-07,
+      "loss": 1.285,
+      "step": 190355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4793377706630384e-07,
+      "loss": 1.1069,
+      "step": 190360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4773620281829625e-07,
+      "loss": 1.1745,
+      "step": 190365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4753868368990864e-07,
+      "loss": 1.346,
+      "step": 190370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473412196822645e-07,
+      "loss": 0.9863,
+      "step": 190375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471438107964964e-07,
+      "loss": 1.0263,
+      "step": 190380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4694645703372666e-07,
+      "loss": 0.959,
+      "step": 190385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467491583950855e-07,
+      "loss": 1.2414,
+      "step": 190390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465519148816976e-07,
+      "loss": 1.2875,
+      "step": 190395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.463547264946909e-07,
+      "loss": 1.1058,
+      "step": 190400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.461575932351924e-07,
+      "loss": 1.147,
+      "step": 190405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4596051510432547e-07,
+      "loss": 1.1755,
+      "step": 190410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.45763492103216e-07,
+      "loss": 1.4227,
+      "step": 190415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4556652423298973e-07,
+      "loss": 1.4871,
+      "step": 190420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.453696114947713e-07,
+      "loss": 1.2703,
+      "step": 190425
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.451727538896854e-07,
+      "loss": 1.275,
+      "step": 190430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4497595141885555e-07,
+      "loss": 1.3102,
+      "step": 190435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4477920408340746e-07,
+      "loss": 1.3984,
+      "step": 190440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4458251188446035e-07,
+      "loss": 1.0744,
+      "step": 190445
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.443858748231421e-07,
+      "loss": 1.0852,
+      "step": 190450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.441892929005741e-07,
+      "loss": 1.534,
+      "step": 190455
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.439927661178788e-07,
+      "loss": 1.2306,
+      "step": 190460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4379629447617746e-07,
+      "loss": 1.1347,
+      "step": 190465
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.435998779765926e-07,
+      "loss": 1.2177,
+      "step": 190470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.434035166202465e-07,
+      "loss": 1.288,
+      "step": 190475
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4320721040825956e-07,
+      "loss": 1.194,
+      "step": 190480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.430109593417552e-07,
+      "loss": 1.2177,
+      "step": 190485
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4281476342185036e-07,
+      "loss": 1.3665,
+      "step": 190490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.426186226496675e-07,
+      "loss": 1.2242,
+      "step": 190495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4242253702632786e-07,
+      "loss": 0.9319,
+      "step": 190500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4222650655294956e-07,
+      "loss": 1.3749,
+      "step": 190505
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4203053123065154e-07,
+      "loss": 1.7022,
+      "step": 190510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4183461106055416e-07,
+      "loss": 1.3896,
+      "step": 190515
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4163874604377645e-07,
+      "loss": 1.0297,
+      "step": 190520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4144293618143644e-07,
+      "loss": 1.0725,
+      "step": 190525
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4124718147465096e-07,
+      "loss": 1.3732,
+      "step": 190530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.410514819245392e-07,
+      "loss": 1.2112,
+      "step": 190535
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4085583753221905e-07,
+      "loss": 1.6195,
+      "step": 190540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.406602482988064e-07,
+      "loss": 1.4411,
+      "step": 190545
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4046471422542026e-07,
+      "loss": 1.2636,
+      "step": 190550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4026923531317645e-07,
+      "loss": 0.9318,
+      "step": 190555
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.400738115631896e-07,
+      "loss": 1.2449,
+      "step": 190560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.398784429765767e-07,
+      "loss": 1.1488,
+      "step": 190565
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.396831295544545e-07,
+      "loss": 1.1996,
+      "step": 190570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3948787129793657e-07,
+      "loss": 1.2031,
+      "step": 190575
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.392926682081399e-07,
+      "loss": 1.27,
+      "step": 190580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.390975202861768e-07,
+      "loss": 1.1449,
+      "step": 190585
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3890242753316315e-07,
+      "loss": 1.2237,
+      "step": 190590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.387073899502136e-07,
+      "loss": 1.3218,
+      "step": 190595
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.385124075384416e-07,
+      "loss": 1.428,
+      "step": 190600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3831748029895863e-07,
+      "loss": 1.1713,
+      "step": 190605
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.381226082328803e-07,
+      "loss": 1.0839,
+      "step": 190610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.379277913413193e-07,
+      "loss": 1.0962,
+      "step": 190615
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.377330296253878e-07,
+      "loss": 1.2671,
+      "step": 190620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.375383230861951e-07,
+      "loss": 1.068,
+      "step": 190625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.373436717248568e-07,
+      "loss": 1.3892,
+      "step": 190630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.371490755424833e-07,
+      "loss": 1.8024,
+      "step": 190635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.369545345401859e-07,
+      "loss": 1.2111,
+      "step": 190640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.367600487190759e-07,
+      "loss": 0.964,
+      "step": 190645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3656561808026345e-07,
+      "loss": 1.7248,
+      "step": 190650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.363712426248567e-07,
+      "loss": 1.1133,
+      "step": 190655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3617692235397013e-07,
+      "loss": 1.1414,
+      "step": 190660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.359826572687108e-07,
+      "loss": 1.2949,
+      "step": 190665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.357884473701878e-07,
+      "loss": 1.5446,
+      "step": 190670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3559429265951013e-07,
+      "loss": 1.5207,
+      "step": 190675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.35400193137787e-07,
+      "loss": 1.2218,
+      "step": 190680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3520614880612755e-07,
+      "loss": 0.9591,
+      "step": 190685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.350121596656386e-07,
+      "loss": 1.2367,
+      "step": 190690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.348182257174304e-07,
+      "loss": 1.0387,
+      "step": 190695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3462434696260647e-07,
+      "loss": 1.3765,
+      "step": 190700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.34430523402276e-07,
+      "loss": 1.4864,
+      "step": 190705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3423675503754695e-07,
+      "loss": 1.1086,
+      "step": 190710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3404304186952507e-07,
+      "loss": 1.1275,
+      "step": 190715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.338493838993151e-07,
+      "loss": 0.9872,
+      "step": 190720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.33655781128025e-07,
+      "loss": 1.0585,
+      "step": 190725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3346223355675944e-07,
+      "loss": 1.3754,
+      "step": 190730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.332687411866231e-07,
+      "loss": 1.0806,
+      "step": 190735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.330753040187218e-07,
+      "loss": 0.9717,
+      "step": 190740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3288192205416123e-07,
+      "loss": 1.1166,
+      "step": 190745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3268859529404286e-07,
+      "loss": 1.357,
+      "step": 190750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.324953237394712e-07,
+      "loss": 1.0266,
+      "step": 190755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.323021073915533e-07,
+      "loss": 1.2688,
+      "step": 190760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3210894625138934e-07,
+      "loss": 1.5159,
+      "step": 190765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3191584032008175e-07,
+      "loss": 1.138,
+      "step": 190770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3172278959873626e-07,
+      "loss": 1.3023,
+      "step": 190775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.315297940884532e-07,
+      "loss": 1.1651,
+      "step": 190780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3133685379033496e-07,
+      "loss": 1.0242,
+      "step": 190785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3114396870548515e-07,
+      "loss": 1.4785,
+      "step": 190790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.309511388350017e-07,
+      "loss": 1.3953,
+      "step": 190795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.307583641799883e-07,
+      "loss": 1.4301,
+      "step": 190800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3056564474154617e-07,
+      "loss": 2.018,
+      "step": 190805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.303729805207745e-07,
+      "loss": 1.0651,
+      "step": 190810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3018037151877346e-07,
+      "loss": 1.4768,
+      "step": 190815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2998781773664335e-07,
+      "loss": 1.1319,
+      "step": 190820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2979531917548433e-07,
+      "loss": 0.9118,
+      "step": 190825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2960287583639563e-07,
+      "loss": 1.2173,
+      "step": 190830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.294104877204751e-07,
+      "loss": 0.9599,
+      "step": 190835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.29218154828822e-07,
+      "loss": 1.2302,
+      "step": 190840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.290258771625354e-07,
+      "loss": 1.2729,
+      "step": 190845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.288336547227111e-07,
+      "loss": 1.173,
+      "step": 190850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.286414875104493e-07,
+      "loss": 1.1947,
+      "step": 190855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.28449375526847e-07,
+      "loss": 1.3975,
+      "step": 190860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2825731877299985e-07,
+      "loss": 1.3301,
+      "step": 190865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.280653172500048e-07,
+      "loss": 1.2655,
+      "step": 190870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2787337095895986e-07,
+      "loss": 1.1944,
+      "step": 190875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.276814799009598e-07,
+      "loss": 1.2431,
+      "step": 190880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.274896440771014e-07,
+      "loss": 1.2761,
+      "step": 190885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2729786348847824e-07,
+      "loss": 1.1872,
+      "step": 190890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.271061381361873e-07,
+      "loss": 1.0196,
+      "step": 190895
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.269144680213232e-07,
+      "loss": 1.1476,
+      "step": 190900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.267228531449806e-07,
+      "loss": 1.1538,
+      "step": 190905
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.265312935082521e-07,
+      "loss": 1.3237,
+      "step": 190910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.263397891122322e-07,
+      "loss": 1.0559,
+      "step": 190915
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2614833995801676e-07,
+      "loss": 1.1612,
+      "step": 190920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.259569460466983e-07,
+      "loss": 0.9797,
+      "step": 190925
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2576560737936693e-07,
+      "loss": 1.2309,
+      "step": 190930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.255743239571174e-07,
+      "loss": 1.384,
+      "step": 190935
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.253830957810422e-07,
+      "loss": 0.9924,
+      "step": 190940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2519192285223267e-07,
+      "loss": 1.2458,
+      "step": 190945
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.250008051717823e-07,
+      "loss": 1.5279,
+      "step": 190950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2480974274078037e-07,
+      "loss": 1.0788,
+      "step": 190955
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.246187355603181e-07,
+      "loss": 1.4123,
+      "step": 190960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2442778363148684e-07,
+      "loss": 1.1652,
+      "step": 190965
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.24236886955378e-07,
+      "loss": 1.2926,
+      "step": 190970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.240460455330796e-07,
+      "loss": 1.1948,
+      "step": 190975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2385525936568297e-07,
+      "loss": 1.2794,
+      "step": 190980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.236645284542761e-07,
+      "loss": 1.2514,
+      "step": 190985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.234738527999504e-07,
+      "loss": 1.1379,
+      "step": 190990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2328323240379267e-07,
+      "loss": 1.3995,
+      "step": 190995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2309266726689437e-07,
+      "loss": 1.3819,
+      "step": 191000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2290215739033904e-07,
+      "loss": 1.3705,
+      "step": 191005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2271170277521804e-07,
+      "loss": 1.2219,
+      "step": 191010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2252130342261933e-07,
+      "loss": 1.6664,
+      "step": 191015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.223309593336288e-07,
+      "loss": 1.2287,
+      "step": 191020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.221406705093322e-07,
+      "loss": 1.167,
+      "step": 191025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.219504369508175e-07,
+      "loss": 1.1266,
+      "step": 191030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.217602586591728e-07,
+      "loss": 0.9792,
+      "step": 191035
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2157013563548276e-07,
+      "loss": 1.3665,
+      "step": 191040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.21380067880831e-07,
+      "loss": 1.2754,
+      "step": 191045
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2119005539630657e-07,
+      "loss": 1.2971,
+      "step": 191050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2100009818299084e-07,
+      "loss": 1.2542,
+      "step": 191055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2081019624197073e-07,
+      "loss": 1.1064,
+      "step": 191060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.206203495743321e-07,
+      "loss": 1.382,
+      "step": 191065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.204305581811562e-07,
+      "loss": 1.1515,
+      "step": 191070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.202408220635278e-07,
+      "loss": 1.5832,
+      "step": 191075
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2005114122253045e-07,
+      "loss": 0.9472,
+      "step": 191080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.198615156592488e-07,
+      "loss": 1.4391,
+      "step": 191085
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1967194537476207e-07,
+      "loss": 1.2617,
+      "step": 191090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1948243037015713e-07,
+      "loss": 1.1226,
+      "step": 191095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.192929706465131e-07,
+      "loss": 1.5228,
+      "step": 191100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.191035662049147e-07,
+      "loss": 1.0788,
+      "step": 191105
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.189142170464399e-07,
+      "loss": 1.0981,
+      "step": 191110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.187249231721723e-07,
+      "loss": 1.3503,
+      "step": 191115
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.185356845831922e-07,
+      "loss": 1.038,
+      "step": 191120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.183465012805809e-07,
+      "loss": 1.1937,
+      "step": 191125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.181573732654197e-07,
+      "loss": 1.1197,
+      "step": 191130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.179683005387868e-07,
+      "loss": 1.1796,
+      "step": 191135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.177792831017612e-07,
+      "loss": 1.1931,
+      "step": 191140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.175903209554243e-07,
+      "loss": 1.2739,
+      "step": 191145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.17401414100853e-07,
+      "loss": 1.177,
+      "step": 191150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.172125625391287e-07,
+      "loss": 1.5097,
+      "step": 191155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1702376627132823e-07,
+      "loss": 1.6913,
+      "step": 191160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.168350252985319e-07,
+      "loss": 1.2289,
+      "step": 191165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1664633962181337e-07,
+      "loss": 1.2294,
+      "step": 191170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.164577092422527e-07,
+      "loss": 1.2639,
+      "step": 191175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.16269134160927e-07,
+      "loss": 1.2979,
+      "step": 191180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.160806143789141e-07,
+      "loss": 1.513,
+      "step": 191185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.158921498972878e-07,
+      "loss": 1.4786,
+      "step": 191190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1570374071712707e-07,
+      "loss": 1.0097,
+      "step": 191195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.155153868395067e-07,
+      "loss": 1.0677,
+      "step": 191200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.153270882655013e-07,
+      "loss": 1.1234,
+      "step": 191205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1513884499618787e-07,
+      "loss": 1.4188,
+      "step": 191210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1495065703263994e-07,
+      "loss": 1.3324,
+      "step": 191215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1476252437593333e-07,
+      "loss": 1.1953,
+      "step": 191220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.145744470271428e-07,
+      "loss": 1.4868,
+      "step": 191225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.143864249873407e-07,
+      "loss": 1.4159,
+      "step": 191230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1419845825760184e-07,
+      "loss": 1.6564,
+      "step": 191235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.140105468389987e-07,
+      "loss": 1.289,
+      "step": 191240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1382269073260584e-07,
+      "loss": 1.1598,
+      "step": 191245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.136348899394948e-07,
+      "loss": 1.0435,
+      "step": 191250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1344714446073786e-07,
+      "loss": 1.2955,
+      "step": 191255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.132594542974088e-07,
+      "loss": 1.1468,
+      "step": 191260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1307181945057665e-07,
+      "loss": 1.2044,
+      "step": 191265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.12884239921315e-07,
+      "loss": 1.3009,
+      "step": 191270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1269671571069636e-07,
+      "loss": 1.1617,
+      "step": 191275
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1250924681978877e-07,
+      "loss": 1.2886,
+      "step": 191280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.123218332496625e-07,
+      "loss": 1.1206,
+      "step": 191285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1213447500138994e-07,
+      "loss": 1.0956,
+      "step": 191290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.119471720760414e-07,
+      "loss": 1.4305,
+      "step": 191295
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1175992447468385e-07,
+      "loss": 1.1664,
+      "step": 191300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1157273219838855e-07,
+      "loss": 1.1952,
+      "step": 191305
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.113855952482225e-07,
+      "loss": 1.0682,
+      "step": 191310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.111985136252571e-07,
+      "loss": 1.4068,
+      "step": 191315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.110114873305592e-07,
+      "loss": 1.4269,
+      "step": 191320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1082451636519793e-07,
+      "loss": 1.341,
+      "step": 191325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.106376007302392e-07,
+      "loss": 1.1046,
+      "step": 191330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.104507404267498e-07,
+      "loss": 1.3075,
+      "step": 191335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1026393545580014e-07,
+      "loss": 1.6341,
+      "step": 191340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.100771858184548e-07,
+      "loss": 1.4805,
+      "step": 191345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0989049151577966e-07,
+      "loss": 1.2516,
+      "step": 191350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.097038525488427e-07,
+      "loss": 1.0659,
+      "step": 191355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.095172689187076e-07,
+      "loss": 1.182,
+      "step": 191360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0933074062644007e-07,
+      "loss": 1.0782,
+      "step": 191365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.091442676731082e-07,
+      "loss": 1.0824,
+      "step": 191370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.089578500597745e-07,
+      "loss": 1.2519,
+      "step": 191375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0877148778750254e-07,
+      "loss": 1.3504,
+      "step": 191380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.085851808573592e-07,
+      "loss": 1.3427,
+      "step": 191385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.08398929270407e-07,
+      "loss": 1.1381,
+      "step": 191390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.082127330277096e-07,
+      "loss": 1.3186,
+      "step": 191395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0802659213032826e-07,
+      "loss": 1.2636,
+      "step": 191400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0784050657932884e-07,
+      "loss": 1.1511,
+      "step": 191405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0765447637577496e-07,
+      "loss": 1.3199,
+      "step": 191410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.074685015207246e-07,
+      "loss": 1.2859,
+      "step": 191415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0728258201524477e-07,
+      "loss": 1.2899,
+      "step": 191420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.070967178603934e-07,
+      "loss": 1.474,
+      "step": 191425
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0691090905723084e-07,
+      "loss": 1.5061,
+      "step": 191430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0672515560682403e-07,
+      "loss": 1.1011,
+      "step": 191435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0653945751022875e-07,
+      "loss": 1.0947,
+      "step": 191440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0635381476850633e-07,
+      "loss": 1.3901,
+      "step": 191445
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0616822738271825e-07,
+      "loss": 1.3303,
+      "step": 191450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.059826953539224e-07,
+      "loss": 1.1651,
+      "step": 191455
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0579721868317926e-07,
+      "loss": 1.3605,
+      "step": 191460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0561179737155e-07,
+      "loss": 1.2691,
+      "step": 191465
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.054264314200905e-07,
+      "loss": 1.2507,
+      "step": 191470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.052411208298589e-07,
+      "loss": 1.3353,
+      "step": 191475
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0505586560191537e-07,
+      "loss": 1.1659,
+      "step": 191480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.04870665737319e-07,
+      "loss": 1.0827,
+      "step": 191485
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.046855212371247e-07,
+      "loss": 1.3316,
+      "step": 191490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0450043210239033e-07,
+      "loss": 1.3026,
+      "step": 191495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.043153983341729e-07,
+      "loss": 1.0345,
+      "step": 191500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.041304199335304e-07,
+      "loss": 1.1399,
+      "step": 191505
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0394549690151764e-07,
+      "loss": 1.1007,
+      "step": 191510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.037606292391915e-07,
+      "loss": 1.4899,
+      "step": 191515
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.035758169476066e-07,
+      "loss": 1.5319,
+      "step": 191520
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0339106002781894e-07,
+      "loss": 1.4114,
+      "step": 191525
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.032063584808853e-07,
+      "loss": 1.3168,
+      "step": 191530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0302171230785824e-07,
+      "loss": 1.1854,
+      "step": 191535
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.028371215097925e-07,
+      "loss": 1.2144,
+      "step": 191540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0265258608774164e-07,
+      "loss": 1.3256,
+      "step": 191545
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0246810604276033e-07,
+      "loss": 1.3162,
+      "step": 191550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0228368137590336e-07,
+      "loss": 0.9888,
+      "step": 191555
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0209931208822095e-07,
+      "loss": 1.229,
+      "step": 191560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0191499818076785e-07,
+      "loss": 1.4189,
+      "step": 191565
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0173073965459656e-07,
+      "loss": 1.2934,
+      "step": 191570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0154653651075836e-07,
+      "loss": 1.2698,
+      "step": 191575
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.01362388750307e-07,
+      "loss": 1.4595,
+      "step": 191580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0117829637429265e-07,
+      "loss": 1.2963,
+      "step": 191585
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.009942593837656e-07,
+      "loss": 1.4009,
+      "step": 191590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.008102777797783e-07,
+      "loss": 1.5624,
+      "step": 191595
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0062635156338227e-07,
+      "loss": 1.5747,
+      "step": 191600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.004424807356254e-07,
+      "loss": 1.4251,
+      "step": 191605
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.002586652975592e-07,
+      "loss": 1.201,
+      "step": 191610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0007490525023277e-07,
+      "loss": 0.952,
+      "step": 191615
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9989120059469636e-07,
+      "loss": 1.3814,
+      "step": 191620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9970755133199804e-07,
+      "loss": 1.3642,
+      "step": 191625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.99523957463187e-07,
+      "loss": 1.449,
+      "step": 191630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.993404189893123e-07,
+      "loss": 1.2231,
+      "step": 191635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.991569359114188e-07,
+      "loss": 1.2258,
+      "step": 191640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9897350823056004e-07,
+      "loss": 1.1475,
+      "step": 191645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9879013594777963e-07,
+      "loss": 1.1408,
+      "step": 191650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9860681906412336e-07,
+      "loss": 1.7137,
+      "step": 191655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9842355758064154e-07,
+      "loss": 1.2102,
+      "step": 191660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9824035149837894e-07,
+      "loss": 1.0988,
+      "step": 191665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.980572008183813e-07,
+      "loss": 1.1628,
+      "step": 191670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.978741055416967e-07,
+      "loss": 1.2306,
+      "step": 191675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.976910656693699e-07,
+      "loss": 1.0641,
+      "step": 191680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9750808120244336e-07,
+      "loss": 1.1356,
+      "step": 191685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.97325152141964e-07,
+      "loss": 1.0852,
+      "step": 191690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.971422784889788e-07,
+      "loss": 1.2684,
+      "step": 191695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9695946024452913e-07,
+      "loss": 1.1521,
+      "step": 191700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9677669740965867e-07,
+      "loss": 1.3739,
+      "step": 191705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.965939899854131e-07,
+      "loss": 1.6334,
+      "step": 191710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9641133797283505e-07,
+      "loss": 1.035,
+      "step": 191715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9622874137296585e-07,
+      "loss": 1.129,
+      "step": 191720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.960462001868514e-07,
+      "loss": 1.3751,
+      "step": 191725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9586371441553295e-07,
+      "loss": 1.0263,
+      "step": 191730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9568128406004983e-07,
+      "loss": 1.1264,
+      "step": 191735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.954989091214477e-07,
+      "loss": 1.5011,
+      "step": 191740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9531658960076594e-07,
+      "loss": 1.0484,
+      "step": 191745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9513432549904576e-07,
+      "loss": 1.0575,
+      "step": 191750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.949521168173297e-07,
+      "loss": 1.4718,
+      "step": 191755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9476996355665476e-07,
+      "loss": 1.3175,
+      "step": 191760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9458786571806565e-07,
+      "loss": 1.2454,
+      "step": 191765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9440582330259817e-07,
+      "loss": 1.1274,
+      "step": 191770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.942238363112948e-07,
+      "loss": 1.207,
+      "step": 191775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9404190474519256e-07,
+      "loss": 1.1272,
+      "step": 191780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.938600286053317e-07,
+      "loss": 1.2016,
+      "step": 191785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9367820789275025e-07,
+      "loss": 1.5488,
+      "step": 191790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9349644260848855e-07,
+      "loss": 1.3562,
+      "step": 191795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9331473275358127e-07,
+      "loss": 0.9208,
+      "step": 191800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9313307832906645e-07,
+      "loss": 1.3804,
+      "step": 191805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.929514793359856e-07,
+      "loss": 1.26,
+      "step": 191810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9276993577536996e-07,
+      "loss": 0.9327,
+      "step": 191815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.92588447648261e-07,
+      "loss": 1.0363,
+      "step": 191820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.924070149556923e-07,
+      "loss": 1.4069,
+      "step": 191825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9222563769870093e-07,
+      "loss": 1.0179,
+      "step": 191830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.920443158783226e-07,
+      "loss": 0.8626,
+      "step": 191835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9186304949559317e-07,
+      "loss": 1.2244,
+      "step": 191840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9168183855154743e-07,
+      "loss": 1.3338,
+      "step": 191845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.915006830472189e-07,
+      "loss": 1.2502,
+      "step": 191850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.913195829836424e-07,
+      "loss": 1.1625,
+      "step": 191855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.911385383618548e-07,
+      "loss": 1.1329,
+      "step": 191860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9095754918288755e-07,
+      "loss": 1.2986,
+      "step": 191865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9077661544777536e-07,
+      "loss": 1.1303,
+      "step": 191870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.905957371575496e-07,
+      "loss": 1.1424,
+      "step": 191875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9041491431324396e-07,
+      "loss": 1.3419,
+      "step": 191880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9023414691589316e-07,
+      "loss": 1.1562,
+      "step": 191885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.900534349665274e-07,
+      "loss": 1.1699,
+      "step": 191890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8987277846617814e-07,
+      "loss": 1.0051,
+      "step": 191895
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8969217741587783e-07,
+      "loss": 1.0432,
+      "step": 191900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.895116318166591e-07,
+      "loss": 1.68,
+      "step": 191905
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8933114166955214e-07,
+      "loss": 1.1011,
+      "step": 191910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.891507069755861e-07,
+      "loss": 1.1998,
+      "step": 191915
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8897032773579136e-07,
+      "loss": 1.1287,
+      "step": 191920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.887900039512015e-07,
+      "loss": 1.4236,
+      "step": 191925
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.886097356228423e-07,
+      "loss": 1.0412,
+      "step": 191930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.884295227517464e-07,
+      "loss": 1.352,
+      "step": 191935
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8824936533894063e-07,
+      "loss": 1.0928,
+      "step": 191940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.880692633854532e-07,
+      "loss": 1.0076,
+      "step": 191945
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.878892168923153e-07,
+      "loss": 1.051,
+      "step": 191950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8770922586055405e-07,
+      "loss": 1.3098,
+      "step": 191955
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8752929029119526e-07,
+      "loss": 1.3741,
+      "step": 191960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.873494101852692e-07,
+      "loss": 1.3376,
+      "step": 191965
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.871695855438017e-07,
+      "loss": 1.1806,
+      "step": 191970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8698981636781974e-07,
+      "loss": 1.1432,
+      "step": 191975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.868101026583503e-07,
+      "loss": 1.1161,
+      "step": 191980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.866304444164203e-07,
+      "loss": 1.0642,
+      "step": 191985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.864508416430534e-07,
+      "loss": 1.3175,
+      "step": 191990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8627129433927647e-07,
+      "loss": 1.294,
+      "step": 191995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8609180250611544e-07,
+      "loss": 1.0697,
+      "step": 192000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8591236614459615e-07,
+      "loss": 1.1325,
+      "step": 192005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.857329852557411e-07,
+      "loss": 1.1006,
+      "step": 192010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8555365984057395e-07,
+      "loss": 1.4688,
+      "step": 192015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8537438990012265e-07,
+      "loss": 1.2008,
+      "step": 192020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8519517543540766e-07,
+      "loss": 1.1249,
+      "step": 192025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8501601644745357e-07,
+      "loss": 1.3335,
+      "step": 192030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.848369129372841e-07,
+      "loss": 1.4539,
+      "step": 192035
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.846578649059184e-07,
+      "loss": 0.9572,
+      "step": 192040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8447887235438456e-07,
+      "loss": 1.1166,
+      "step": 192045
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.842999352837017e-07,
+      "loss": 1.1814,
+      "step": 192050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.841210536948913e-07,
+      "loss": 1.3063,
+      "step": 192055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8394222758897474e-07,
+      "loss": 1.1659,
+      "step": 192060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.837634569669734e-07,
+      "loss": 1.6459,
+      "step": 192065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.835847418299098e-07,
+      "loss": 1.4276,
+      "step": 192070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8340608217880203e-07,
+      "loss": 1.3349,
+      "step": 192075
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8322747801467377e-07,
+      "loss": 1.221,
+      "step": 192080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8304892933854077e-07,
+      "loss": 1.0157,
+      "step": 192085
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.828704361514245e-07,
+      "loss": 1.3908,
+      "step": 192090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8269199845434635e-07,
+      "loss": 1.0117,
+      "step": 192095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.825136162483233e-07,
+      "loss": 1.2639,
+      "step": 192100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8233528953437226e-07,
+      "loss": 1.0034,
+      "step": 192105
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.821570183135136e-07,
+      "loss": 1.0955,
+      "step": 192110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8197880258676644e-07,
+      "loss": 0.9687,
+      "step": 192115
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8180064235514783e-07,
+      "loss": 1.269,
+      "step": 192120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.816225376196724e-07,
+      "loss": 1.1081,
+      "step": 192125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8144448838136053e-07,
+      "loss": 1.1732,
+      "step": 192130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.812664946412269e-07,
+      "loss": 1.32,
+      "step": 192135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8108855640028854e-07,
+      "loss": 1.0971,
+      "step": 192140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.809106736595635e-07,
+      "loss": 1.1913,
+      "step": 192145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.807328464200654e-07,
+      "loss": 1.2665,
+      "step": 192150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.805550746828101e-07,
+      "loss": 1.0988,
+      "step": 192155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.803773584488123e-07,
+      "loss": 1.2305,
+      "step": 192160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.801996977190879e-07,
+      "loss": 1.306,
+      "step": 192165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8002209249465174e-07,
+      "loss": 1.4512,
+      "step": 192170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.798445427765173e-07,
+      "loss": 1.9291,
+      "step": 192175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7966704856569715e-07,
+      "loss": 1.2114,
+      "step": 192180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7948960986320607e-07,
+      "loss": 1.2068,
+      "step": 192185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7931222667005987e-07,
+      "loss": 1.4347,
+      "step": 192190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7913489898726774e-07,
+      "loss": 1.1447,
+      "step": 192195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7895762681584336e-07,
+      "loss": 1.0084,
+      "step": 192200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7878041015679926e-07,
+      "loss": 1.4827,
+      "step": 192205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7860324901114897e-07,
+      "loss": 1.0365,
+      "step": 192210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7842614337990293e-07,
+      "loss": 1.6012,
+      "step": 192215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7824909326407135e-07,
+      "loss": 1.1268,
+      "step": 192220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.780720986646657e-07,
+      "loss": 1.1759,
+      "step": 192225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.778951595826984e-07,
+      "loss": 1.2041,
+      "step": 192230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7771827601917875e-07,
+      "loss": 1.3059,
+      "step": 192235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.775414479751171e-07,
+      "loss": 1.0877,
+      "step": 192240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7736467545152246e-07,
+      "loss": 1.2986,
+      "step": 192245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7718795844940525e-07,
+      "loss": 1.1559,
+      "step": 192250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.770112969697725e-07,
+      "loss": 1.2845,
+      "step": 192255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.768346910136377e-07,
+      "loss": 1.28,
+      "step": 192260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.7665814058200456e-07,
+      "loss": 0.9988,
+      "step": 192265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.764816456758834e-07,
+      "loss": 1.2889,
+      "step": 192270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.763052062962823e-07,
+      "loss": 1.4527,
+      "step": 192275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7612882244420826e-07,
+      "loss": 1.3497,
+      "step": 192280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.759524941206682e-07,
+      "loss": 1.4138,
+      "step": 192285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7577622132667017e-07,
+      "loss": 1.4611,
+      "step": 192290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7560000406321896e-07,
+      "loss": 1.2165,
+      "step": 192295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.754238423313227e-07,
+      "loss": 1.1056,
+      "step": 192300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.752477361319872e-07,
+      "loss": 1.1029,
+      "step": 192305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7507168546621833e-07,
+      "loss": 1.6385,
+      "step": 192310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.748956903350186e-07,
+      "loss": 1.2584,
+      "step": 192315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.747197507393951e-07,
+      "loss": 1.1047,
+      "step": 192320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.745438666803546e-07,
+      "loss": 1.3003,
+      "step": 192325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.743680381588976e-07,
+      "loss": 1.1428,
+      "step": 192330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7419226517603203e-07,
+      "loss": 1.4243,
+      "step": 192335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7401654773275945e-07,
+      "loss": 1.3843,
+      "step": 192340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.738408858300823e-07,
+      "loss": 1.0376,
+      "step": 192345
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.736652794690042e-07,
+      "loss": 1.0988,
+      "step": 192350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7348972865053005e-07,
+      "loss": 1.1327,
+      "step": 192355
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.733142333756622e-07,
+      "loss": 1.1704,
+      "step": 192360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7313879364539886e-07,
+      "loss": 1.0689,
+      "step": 192365
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7296340946074586e-07,
+      "loss": 1.1318,
+      "step": 192370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.727880808227035e-07,
+      "loss": 1.135,
+      "step": 192375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.726128077322732e-07,
+      "loss": 1.0161,
+      "step": 192380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.724375901904552e-07,
+      "loss": 1.2201,
+      "step": 192385
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7226242819825113e-07,
+      "loss": 1.475,
+      "step": 192390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7208732175665995e-07,
+      "loss": 1.1657,
+      "step": 192395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7191227086668324e-07,
+      "loss": 0.966,
+      "step": 192400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.717372755293202e-07,
+      "loss": 1.41,
+      "step": 192405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.715623357455677e-07,
+      "loss": 1.1789,
+      "step": 192410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.713874515164272e-07,
+      "loss": 1.2453,
+      "step": 192415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7121262284289793e-07,
+      "loss": 1.3299,
+      "step": 192420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.71037849725978e-07,
+      "loss": 1.0882,
+      "step": 192425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7086313216666216e-07,
+      "loss": 1.4332,
+      "step": 192430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.706884701659529e-07,
+      "loss": 1.265,
+      "step": 192435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.705138637248439e-07,
+      "loss": 1.0873,
+      "step": 192440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7033931284433437e-07,
+      "loss": 1.102,
+      "step": 192445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7016481752542124e-07,
+      "loss": 1.1367,
+      "step": 192450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.699903777690993e-07,
+      "loss": 1.1051,
+      "step": 192455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6981599357636557e-07,
+      "loss": 1.1231,
+      "step": 192460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.696416649482159e-07,
+      "loss": 0.9799,
+      "step": 192465
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6946739188564606e-07,
+      "loss": 1.2632,
+      "step": 192470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6929317438965096e-07,
+      "loss": 0.9795,
+      "step": 192475
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.691190124612253e-07,
+      "loss": 1.0903,
+      "step": 192480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6894490610136383e-07,
+      "loss": 1.1024,
+      "step": 192485
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.687708553110624e-07,
+      "loss": 1.2698,
+      "step": 192490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6859686009131137e-07,
+      "loss": 1.2356,
+      "step": 192495
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.684229204431077e-07,
+      "loss": 1.314,
+      "step": 192500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.682490363674417e-07,
+      "loss": 1.2354,
+      "step": 192505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6807520786530814e-07,
+      "loss": 1.1014,
+      "step": 192510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.679014349377007e-07,
+      "loss": 1.2629,
+      "step": 192515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.677277175856108e-07,
+      "loss": 1.2243,
+      "step": 192520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.675540558100287e-07,
+      "loss": 1.1602,
+      "step": 192525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.673804496119481e-07,
+      "loss": 1.3873,
+      "step": 192530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.672068989923604e-07,
+      "loss": 1.3239,
+      "step": 192535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6703340395225486e-07,
+      "loss": 1.1426,
+      "step": 192540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6685996449262507e-07,
+      "loss": 1.453,
+      "step": 192545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.666865806144592e-07,
+      "loss": 1.2698,
+      "step": 192550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6651325231874746e-07,
+      "loss": 1.1444,
+      "step": 192555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6633997960648027e-07,
+      "loss": 1.131,
+      "step": 192560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.661667624786479e-07,
+      "loss": 1.2184,
+      "step": 192565
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.659936009362396e-07,
+      "loss": 1.3331,
+      "step": 192570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.658204949802412e-07,
+      "loss": 0.9014,
+      "step": 192575
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6564744461164307e-07,
+      "loss": 1.0831,
+      "step": 192580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.654744498314354e-07,
+      "loss": 1.2636,
+      "step": 192585
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.653015106406032e-07,
+      "loss": 1.0956,
+      "step": 192590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6512862704013654e-07,
+      "loss": 0.9781,
+      "step": 192595
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.649557990310203e-07,
+      "loss": 1.562,
+      "step": 192600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.647830266142415e-07,
+      "loss": 1.4069,
+      "step": 192605
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6461030979078927e-07,
+      "loss": 1.5956,
+      "step": 192610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6443764856164843e-07,
+      "loss": 1.329,
+      "step": 192615
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6426504292780377e-07,
+      "loss": 1.4232,
+      "step": 192620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.640924928902433e-07,
+      "loss": 1.2958,
+      "step": 192625
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6391999844995077e-07,
+      "loss": 1.2045,
+      "step": 192630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.63747559607912e-07,
+      "loss": 1.0913,
+      "step": 192635
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6357517636511065e-07,
+      "loss": 1.0413,
+      "step": 192640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6340284872253265e-07,
+      "loss": 1.0329,
+      "step": 192645
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6323057668115937e-07,
+      "loss": 1.2753,
+      "step": 192650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6305836024197784e-07,
+      "loss": 1.3365,
+      "step": 192655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6288619940596947e-07,
+      "loss": 1.3787,
+      "step": 192660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.627140941741191e-07,
+      "loss": 1.1165,
+      "step": 192665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6254204454740804e-07,
+      "loss": 1.16,
+      "step": 192670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6237005052681787e-07,
+      "loss": 1.1617,
+      "step": 192675
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6219811211333437e-07,
+      "loss": 1.1962,
+      "step": 192680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6202622930793566e-07,
+      "loss": 1.0839,
+      "step": 192685
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.618544021116054e-07,
+      "loss": 1.4707,
+      "step": 192690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6168263052532397e-07,
+      "loss": 1.5225,
+      "step": 192695
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.615109145500716e-07,
+      "loss": 1.1039,
+      "step": 192700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6133925418683093e-07,
+      "loss": 1.2112,
+      "step": 192705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.611676494365811e-07,
+      "loss": 1.0958,
+      "step": 192710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.609961003003025e-07,
+      "loss": 1.1744,
+      "step": 192715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6082460677897216e-07,
+      "loss": 1.2091,
+      "step": 192720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.606531688735736e-07,
+      "loss": 1.2747,
+      "step": 192725
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6048178658508397e-07,
+      "loss": 1.0482,
+      "step": 192730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.603104599144801e-07,
+      "loss": 1.3944,
+      "step": 192735
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6013918886274357e-07,
+      "loss": 1.107,
+      "step": 192740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.599679734308491e-07,
+      "loss": 1.3857,
+      "step": 192745
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5979681361977704e-07,
+      "loss": 1.1726,
+      "step": 192750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5962570943050434e-07,
+      "loss": 1.1573,
+      "step": 192755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5945466086400804e-07,
+      "loss": 1.5248,
+      "step": 192760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5928366792126405e-07,
+      "loss": 1.2449,
+      "step": 192765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5911273060324924e-07,
+      "loss": 1.094,
+      "step": 192770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5894184891093967e-07,
+      "loss": 1.0754,
+      "step": 192775
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.587710228453133e-07,
+      "loss": 0.9986,
+      "step": 192780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5860025240734165e-07,
+      "loss": 1.0954,
+      "step": 192785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.584295375980017e-07,
+      "loss": 1.1568,
+      "step": 192790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5825887841827047e-07,
+      "loss": 1.2106,
+      "step": 192795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.580882748691194e-07,
+      "loss": 1.2284,
+      "step": 192800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.579177269515243e-07,
+      "loss": 1.1082,
+      "step": 192805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5774723466645777e-07,
+      "loss": 1.0342,
+      "step": 192810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5757679801489575e-07,
+      "loss": 1.1405,
+      "step": 192815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.574064169978097e-07,
+      "loss": 1.3308,
+      "step": 192820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.572360916161742e-07,
+      "loss": 1.3738,
+      "step": 192825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.570658218709599e-07,
+      "loss": 1.1168,
+      "step": 192830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5689560776314013e-07,
+      "loss": 1.3962,
+      "step": 192835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.567254492936855e-07,
+      "loss": 1.3416,
+      "step": 192840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.565553464635706e-07,
+      "loss": 1.0199,
+      "step": 192845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.563852992737648e-07,
+      "loss": 1.1945,
+      "step": 192850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.562153077252394e-07,
+      "loss": 1.1925,
+      "step": 192855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5604537181896594e-07,
+      "loss": 1.2068,
+      "step": 192860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5587549155591253e-07,
+      "loss": 1.2492,
+      "step": 192865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.557056669370528e-07,
+      "loss": 1.3065,
+      "step": 192870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.555358979633549e-07,
+      "loss": 1.2177,
+      "step": 192875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5536618463578687e-07,
+      "loss": 1.1703,
+      "step": 192880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5519652695531807e-07,
+      "loss": 1.1432,
+      "step": 192885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.550269249229209e-07,
+      "loss": 1.2148,
+      "step": 192890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5485737853955915e-07,
+      "loss": 1.4779,
+      "step": 192895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5468788780620535e-07,
+      "loss": 1.12,
+      "step": 192900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5451845272382314e-07,
+      "loss": 0.9231,
+      "step": 192905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5434907329338285e-07,
+      "loss": 1.5412,
+      "step": 192910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5417974951585154e-07,
+      "loss": 1.0695,
+      "step": 192915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.540104813921962e-07,
+      "loss": 1.0414,
+      "step": 192920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.538412689233827e-07,
+      "loss": 1.2931,
+      "step": 192925
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.536721121103758e-07,
+      "loss": 1.1194,
+      "step": 192930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.535030109541459e-07,
+      "loss": 1.4824,
+      "step": 192935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.533339654556544e-07,
+      "loss": 1.1243,
+      "step": 192940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.531649756158683e-07,
+      "loss": 1.2354,
+      "step": 192945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.529960414357535e-07,
+      "loss": 0.9109,
+      "step": 192950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5282716291627265e-07,
+      "loss": 1.3024,
+      "step": 192955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5265834005839043e-07,
+      "loss": 1.1997,
+      "step": 192960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5248957286307383e-07,
+      "loss": 1.2367,
+      "step": 192965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.523208613312833e-07,
+      "loss": 1.4719,
+      "step": 192970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.521522054639836e-07,
+      "loss": 1.4277,
+      "step": 192975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.519836052621372e-07,
+      "loss": 1.4429,
+      "step": 192980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.518150607267089e-07,
+      "loss": 1.1423,
+      "step": 192985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5164657185865806e-07,
+      "loss": 1.1502,
+      "step": 192990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5147813865894933e-07,
+      "loss": 1.2735,
+      "step": 192995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.513097611285431e-07,
+      "loss": 1.3036,
+      "step": 193000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.511414392684031e-07,
+      "loss": 1.0821,
+      "step": 193005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.509731730794862e-07,
+      "loss": 1.2652,
+      "step": 193010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.508049625627584e-07,
+      "loss": 0.8419,
+      "step": 193015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5063680771917786e-07,
+      "loss": 0.865,
+      "step": 193020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5046870854970264e-07,
+      "loss": 0.944,
+      "step": 193025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5030066505529636e-07,
+      "loss": 1.2026,
+      "step": 193030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5013267723691834e-07,
+      "loss": 1.0158,
+      "step": 193035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4996474509552446e-07,
+      "loss": 1.186,
+      "step": 193040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.497968686320773e-07,
+      "loss": 1.3477,
+      "step": 193045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4962904784753384e-07,
+      "loss": 1.2828,
+      "step": 193050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.494612827428522e-07,
+      "loss": 1.152,
+      "step": 193055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4929357331899274e-07,
+      "loss": 1.6097,
+      "step": 193060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4912591957691025e-07,
+      "loss": 1.2001,
+      "step": 193065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4895832151756283e-07,
+      "loss": 1.0816,
+      "step": 193070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4879077914190863e-07,
+      "loss": 1.2734,
+      "step": 193075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.486232924509035e-07,
+      "loss": 1.0617,
+      "step": 193080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.484558614455057e-07,
+      "loss": 1.1326,
+      "step": 193085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4828848612666766e-07,
+      "loss": 1.2927,
+      "step": 193090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.481211664953476e-07,
+      "loss": 1.3808,
+      "step": 193095
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.479539025525013e-07,
+      "loss": 0.9492,
+      "step": 193100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.477866942990836e-07,
+      "loss": 1.3234,
+      "step": 193105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.476195417360494e-07,
+      "loss": 1.0408,
+      "step": 193110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.474524448643534e-07,
+      "loss": 1.2302,
+      "step": 193115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4728540368494593e-07,
+      "loss": 1.4959,
+      "step": 193120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4711841819878734e-07,
+      "loss": 1.1942,
+      "step": 193125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4695148840682915e-07,
+      "loss": 1.325,
+      "step": 193130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4678461431002163e-07,
+      "loss": 1.3779,
+      "step": 193135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.466177959093208e-07,
+      "loss": 1.1471,
+      "step": 193140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.464510332056769e-07,
+      "loss": 1.1464,
+      "step": 193145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.462843262000447e-07,
+      "loss": 1.3402,
+      "step": 193150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4611767489337467e-07,
+      "loss": 1.2926,
+      "step": 193155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.459510792866193e-07,
+      "loss": 1.2665,
+      "step": 193160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.45784539380729e-07,
+      "loss": 1.3056,
+      "step": 193165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4561805517665516e-07,
+      "loss": 1.1796,
+      "step": 193170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4545162667534926e-07,
+      "loss": 0.9913,
+      "step": 193175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.452852538777617e-07,
+      "loss": 1.5626,
+      "step": 193180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.451189367848406e-07,
+      "loss": 1.0473,
+      "step": 193185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.449526753975362e-07,
+      "loss": 1.0692,
+      "step": 193190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.447864697168012e-07,
+      "loss": 1.4713,
+      "step": 193195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4462031974358034e-07,
+      "loss": 1.1374,
+      "step": 193200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4445422547882626e-07,
+      "loss": 1.4659,
+      "step": 193205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4428818692348366e-07,
+      "loss": 1.2235,
+      "step": 193210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4412220407850407e-07,
+      "loss": 1.1692,
+      "step": 193215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.439562769448334e-07,
+      "loss": 1.1268,
+      "step": 193220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.437904055234197e-07,
+      "loss": 1.2527,
+      "step": 193225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.436245898152112e-07,
+      "loss": 1.2518,
+      "step": 193230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.434588298211527e-07,
+      "loss": 1.5359,
+      "step": 193235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.432931255421911e-07,
+      "loss": 1.6519,
+      "step": 193240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.431274769792746e-07,
+      "loss": 1.2136,
+      "step": 193245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4296188413334697e-07,
+      "loss": 1.2128,
+      "step": 193250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.427963470053563e-07,
+      "loss": 1.2467,
+      "step": 193255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4263086559624503e-07,
+      "loss": 1.3081,
+      "step": 193260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4246543990695926e-07,
+      "loss": 1.3398,
+      "step": 193265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.423000699384459e-07,
+      "loss": 1.1889,
+      "step": 193270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4213475569164645e-07,
+      "loss": 1.0772,
+      "step": 193275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.419694971675057e-07,
+      "loss": 1.1054,
+      "step": 193280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.418042943669663e-07,
+      "loss": 1.3381,
+      "step": 193285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4163914729097515e-07,
+      "loss": 0.9799,
+      "step": 193290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.414740559404716e-07,
+      "loss": 1.451,
+      "step": 193295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4130902031640034e-07,
+      "loss": 1.6751,
+      "step": 193300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4114404041970185e-07,
+      "loss": 1.2504,
+      "step": 193305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.40979116251322e-07,
+      "loss": 1.103,
+      "step": 193310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.408142478121989e-07,
+      "loss": 1.2182,
+      "step": 193315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4064943510327623e-07,
+      "loss": 1.207,
+      "step": 193320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4048467812549436e-07,
+      "loss": 1.2869,
+      "step": 193325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4031997687979257e-07,
+      "loss": 1.2177,
+      "step": 193330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4015533136711457e-07,
+      "loss": 0.9175,
+      "step": 193335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3999074158839843e-07,
+      "loss": 1.2134,
+      "step": 193340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3982620754458454e-07,
+      "loss": 1.2442,
+      "step": 193345
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3966172923661324e-07,
+      "loss": 1.3103,
+      "step": 193350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.394973066654216e-07,
+      "loss": 0.8053,
+      "step": 193355
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3933293983195214e-07,
+      "loss": 1.1502,
+      "step": 193360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3916862873713973e-07,
+      "loss": 1.2002,
+      "step": 193365
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.390043733819247e-07,
+      "loss": 1.28,
+      "step": 193370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3884017376724523e-07,
+      "loss": 1.2415,
+      "step": 193375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3867602989403715e-07,
+      "loss": 1.3367,
+      "step": 193380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.385119417632409e-07,
+      "loss": 1.2556,
+      "step": 193385
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3834790937579122e-07,
+      "loss": 1.1811,
+      "step": 193390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3818393273262408e-07,
+      "loss": 1.3328,
+      "step": 193395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.380200118346776e-07,
+      "loss": 1.2432,
+      "step": 193400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.378561466828877e-07,
+      "loss": 1.2338,
+      "step": 193405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.376923372781892e-07,
+      "loss": 1.3646,
+      "step": 193410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3752858362151909e-07,
+      "loss": 1.3668,
+      "step": 193415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3736488571381223e-07,
+      "loss": 1.2848,
+      "step": 193420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3720124355600004e-07,
+      "loss": 0.9263,
+      "step": 193425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3703765714902182e-07,
+      "loss": 1.3801,
+      "step": 193430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3687412649380903e-07,
+      "loss": 1.1233,
+      "step": 193435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3671065159129648e-07,
+      "loss": 2.2454,
+      "step": 193440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3654723244241672e-07,
+      "loss": 1.235,
+      "step": 193445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.363838690481035e-07,
+      "loss": 1.1498,
+      "step": 193450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.362205614092905e-07,
+      "loss": 1.2358,
+      "step": 193455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.360573095269092e-07,
+      "loss": 1.1484,
+      "step": 193460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3589411340189217e-07,
+      "loss": 1.5523,
+      "step": 193465
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.357309730351709e-07,
+      "loss": 1.0513,
+      "step": 193470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3556788842767797e-07,
+      "loss": 1.1332,
+      "step": 193475
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3540485958034375e-07,
+      "loss": 1.4724,
+      "step": 193480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3524188649410083e-07,
+      "loss": 1.4696,
+      "step": 193485
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.350789691698785e-07,
+      "loss": 1.2969,
+      "step": 193490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3491610760860596e-07,
+      "loss": 1.4303,
+      "step": 193495
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.347533018112158e-07,
+      "loss": 1.4104,
+      "step": 193500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.345905517786373e-07,
+      "loss": 1.1352,
+      "step": 193505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3442785751179753e-07,
+      "loss": 1.3536,
+      "step": 193510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.342652190116268e-07,
+      "loss": 1.3689,
+      "step": 193515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.341026362790555e-07,
+      "loss": 1.4261,
+      "step": 193520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3394010931501067e-07,
+      "loss": 1.3054,
+      "step": 193525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3377763812042043e-07,
+      "loss": 1.0796,
+      "step": 193530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3361522269621296e-07,
+      "loss": 1.2586,
+      "step": 193535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3345286304331417e-07,
+      "loss": 1.2969,
+      "step": 193540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.332905591626533e-07,
+      "loss": 1.2525,
+      "step": 193545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.331283110551563e-07,
+      "loss": 1.194,
+      "step": 193550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3296611872174913e-07,
+      "loss": 1.2075,
+      "step": 193555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3280398216335987e-07,
+      "loss": 1.2932,
+      "step": 193560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3264190138091226e-07,
+      "loss": 1.0538,
+      "step": 193565
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3247987637533332e-07,
+      "loss": 1.0221,
+      "step": 193570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3231790714754788e-07,
+      "loss": 1.1427,
+      "step": 193575
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3215599369848074e-07,
+      "loss": 1.016,
+      "step": 193580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3199413602905564e-07,
+      "loss": 1.1107,
+      "step": 193585
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3183233414019846e-07,
+      "loss": 1.2723,
+      "step": 193590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3167058803283293e-07,
+      "loss": 1.1971,
+      "step": 193595
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3150889770788164e-07,
+      "loss": 1.1665,
+      "step": 193600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3134726316626832e-07,
+      "loss": 1.2121,
+      "step": 193605
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3118568440891665e-07,
+      "loss": 1.3795,
+      "step": 193610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.310241614367492e-07,
+      "loss": 1.2603,
+      "step": 193615
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3086269425068752e-07,
+      "loss": 1.3575,
+      "step": 193620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3070128285165416e-07,
+      "loss": 1.1926,
+      "step": 193625
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3053992724057173e-07,
+      "loss": 1.5638,
+      "step": 193630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3037862741835947e-07,
+      "loss": 1.3449,
+      "step": 193635
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3021738338594113e-07,
+      "loss": 1.0499,
+      "step": 193640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3005619514423594e-07,
+      "loss": 1.155,
+      "step": 193645
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2989506269416427e-07,
+      "loss": 1.3507,
+      "step": 193650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2973398603664655e-07,
+      "loss": 1.1679,
+      "step": 193655
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2957296517260307e-07,
+      "loss": 1.0368,
+      "step": 193660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2941200010295317e-07,
+      "loss": 1.3857,
+      "step": 193665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2925109082861496e-07,
+      "loss": 0.921,
+      "step": 193670
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2909023735050994e-07,
+      "loss": 0.9977,
+      "step": 193675
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2892943966955295e-07,
+      "loss": 1.0742,
+      "step": 193680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2876869778666543e-07,
+      "loss": 1.1107,
+      "step": 193685
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.286080117027645e-07,
+      "loss": 1.0676,
+      "step": 193690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.284473814187671e-07,
+      "loss": 1.1038,
+      "step": 193695
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2828680693559034e-07,
+      "loss": 1.2778,
+      "step": 193700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2812628825415128e-07,
+      "loss": 1.5028,
+      "step": 193705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2796582537536804e-07,
+      "loss": 1.1879,
+      "step": 193710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2780541830015436e-07,
+      "loss": 1.1228,
+      "step": 193715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2764506702942945e-07,
+      "loss": 1.3539,
+      "step": 193720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.274847715641071e-07,
+      "loss": 1.5059,
+      "step": 193725
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2732453190510206e-07,
+      "loss": 1.5056,
+      "step": 193730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.271643480533303e-07,
+      "loss": 0.9342,
+      "step": 193735
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2700422000970667e-07,
+      "loss": 1.1589,
+      "step": 193740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2684414777514597e-07,
+      "loss": 1.2344,
+      "step": 193745
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2668413135056078e-07,
+      "loss": 1.0555,
+      "step": 193750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2652417073686594e-07,
+      "loss": 1.3033,
+      "step": 193755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2636426593497517e-07,
+      "loss": 1.1543,
+      "step": 193760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2620441694579999e-07,
+      "loss": 1.1587,
+      "step": 193765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2604462377025626e-07,
+      "loss": 1.1068,
+      "step": 193770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2588488640925334e-07,
+      "loss": 1.2996,
+      "step": 193775
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2572520486370485e-07,
+      "loss": 1.27,
+      "step": 193780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2556557913452347e-07,
+      "loss": 1.1222,
+      "step": 193785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2540600922261956e-07,
+      "loss": 1.2451,
+      "step": 193790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2524649512890462e-07,
+      "loss": 1.2605,
+      "step": 193795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.250870368542879e-07,
+      "loss": 1.2572,
+      "step": 193800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2492763439968423e-07,
+      "loss": 1.1315,
+      "step": 193805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.247682877660007e-07,
+      "loss": 1.3252,
+      "step": 193810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2460899695414652e-07,
+      "loss": 1.5339,
+      "step": 193815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.244497619650343e-07,
+      "loss": 1.4943,
+      "step": 193820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2429058279957006e-07,
+      "loss": 1.2192,
+      "step": 193825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2413145945866523e-07,
+      "loss": 1.5071,
+      "step": 193830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2397239194322794e-07,
+      "loss": 1.213,
+      "step": 193835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2381338025416756e-07,
+      "loss": 1.0739,
+      "step": 193840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2365442439238883e-07,
+      "loss": 1.3856,
+      "step": 193845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2349552435880218e-07,
+      "loss": 1.1179,
+      "step": 193850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2333668015431465e-07,
+      "loss": 1.4104,
+      "step": 193855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2317789177983218e-07,
+      "loss": 1.5462,
+      "step": 193860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2301915923626405e-07,
+      "loss": 1.3321,
+      "step": 193865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2286048252451398e-07,
+      "loss": 1.1084,
+      "step": 193870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2270186164549012e-07,
+      "loss": 1.1402,
+      "step": 193875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.225432966000962e-07,
+      "loss": 1.1307,
+      "step": 193880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.223847873892404e-07,
+      "loss": 1.2849,
+      "step": 193885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.222263340138253e-07,
+      "loss": 1.045,
+      "step": 193890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2206793647475578e-07,
+      "loss": 1.1373,
+      "step": 193895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2190959477293995e-07,
+      "loss": 1.6034,
+      "step": 193900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.217513089092782e-07,
+      "loss": 1.2006,
+      "step": 193905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2159307888467542e-07,
+      "loss": 1.1822,
+      "step": 193910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2143490470003527e-07,
+      "loss": 1.1383,
+      "step": 193915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.212767863562626e-07,
+      "loss": 1.6517,
+      "step": 193920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.211187238542578e-07,
+      "loss": 1.6818,
+      "step": 193925
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2096071719492462e-07,
+      "loss": 1.2098,
+      "step": 193930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2080276637916675e-07,
+      "loss": 1.2118,
+      "step": 193935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2064487140788348e-07,
+      "loss": 1.5992,
+      "step": 193940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2048703228197742e-07,
+      "loss": 1.1599,
+      "step": 193945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.203292490023523e-07,
+      "loss": 1.1105,
+      "step": 193950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.201715215699063e-07,
+      "loss": 1.2604,
+      "step": 193955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2001384998554087e-07,
+      "loss": 1.3026,
+      "step": 193960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1985623425015533e-07,
+      "loss": 1.1921,
+      "step": 193965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1969867436465342e-07,
+      "loss": 1.2934,
+      "step": 193970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1954117032992995e-07,
+      "loss": 1.3424,
+      "step": 193975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1938372214688864e-07,
+      "loss": 1.2936,
+      "step": 193980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1922632981642656e-07,
+      "loss": 0.9734,
+      "step": 193985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1906899333944188e-07,
+      "loss": 1.0364,
+      "step": 193990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1891171271683497e-07,
+      "loss": 1.0245,
+      "step": 193995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1875448794950294e-07,
+      "loss": 1.1225,
+      "step": 194000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.185973190383417e-07,
+      "loss": 1.3584,
+      "step": 194005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1844020598425275e-07,
+      "loss": 1.2546,
+      "step": 194010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1828314878813094e-07,
+      "loss": 1.3894,
+      "step": 194015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1812614745087334e-07,
+      "loss": 1.3382,
+      "step": 194020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.179692019733759e-07,
+      "loss": 1.1203,
+      "step": 194025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1781231235653677e-07,
+      "loss": 0.9996,
+      "step": 194030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.176554786012497e-07,
+      "loss": 1.3883,
+      "step": 194035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1749870070841172e-07,
+      "loss": 1.3492,
+      "step": 194040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.173419786789177e-07,
+      "loss": 1.1683,
+      "step": 194045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1718531251366359e-07,
+      "loss": 1.0848,
+      "step": 194050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.17028702213542e-07,
+      "loss": 1.3568,
+      "step": 194055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1687214777944775e-07,
+      "loss": 1.2703,
+      "step": 194060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1671564921227682e-07,
+      "loss": 0.9879,
+      "step": 194065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1655920651291962e-07,
+      "loss": 1.3742,
+      "step": 194070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1640281968227317e-07,
+      "loss": 1.3738,
+      "step": 194075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.162464887212279e-07,
+      "loss": 1.2362,
+      "step": 194080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.160902136306775e-07,
+      "loss": 1.1085,
+      "step": 194085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1593399441151464e-07,
+      "loss": 1.4009,
+      "step": 194090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.157778310646308e-07,
+      "loss": 1.0922,
+      "step": 194095
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.156217235909175e-07,
+      "loss": 1.1981,
+      "step": 194100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1546567199126734e-07,
+      "loss": 1.3235,
+      "step": 194105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1530967626657072e-07,
+      "loss": 1.246,
+      "step": 194110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1515373641771919e-07,
+      "loss": 1.1489,
+      "step": 194115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.14997852445602e-07,
+      "loss": 1.1968,
+      "step": 194120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1484202435110958e-07,
+      "loss": 1.4494,
+      "step": 194125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1468625213513228e-07,
+      "loss": 1.1294,
+      "step": 194130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1453053579855943e-07,
+      "loss": 1.2352,
+      "step": 194135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1437487534228142e-07,
+      "loss": 1.1509,
+      "step": 194140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.142192707671864e-07,
+      "loss": 1.2158,
+      "step": 194145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1406372207416038e-07,
+      "loss": 1.3263,
+      "step": 194150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1390822926409484e-07,
+      "loss": 1.3993,
+      "step": 194155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1375279233787793e-07,
+      "loss": 1.5638,
+      "step": 194160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1359741129639566e-07,
+      "loss": 1.1408,
+      "step": 194165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1344208614053508e-07,
+      "loss": 1.6105,
+      "step": 194170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1328681687118436e-07,
+      "loss": 0.9245,
+      "step": 194175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1313160348922944e-07,
+      "loss": 1.2496,
+      "step": 194180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1297644599555632e-07,
+      "loss": 1.3245,
+      "step": 194185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1282134439105317e-07,
+      "loss": 1.1379,
+      "step": 194190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.126662986766037e-07,
+      "loss": 1.1538,
+      "step": 194195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.125113088530928e-07,
+      "loss": 1.6095,
+      "step": 194200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1235637492140748e-07,
+      "loss": 1.2564,
+      "step": 194205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1220149688243263e-07,
+      "loss": 1.4097,
+      "step": 194210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1204667473704977e-07,
+      "loss": 1.4174,
+      "step": 194215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1189190848614483e-07,
+      "loss": 1.1077,
+      "step": 194220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.117371981306038e-07,
+      "loss": 1.1939,
+      "step": 194225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1158254367130705e-07,
+      "loss": 1.3835,
+      "step": 194230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1142794510913722e-07,
+      "loss": 1.3279,
+      "step": 194235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1127340244498028e-07,
+      "loss": 1.2237,
+      "step": 194240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1111891567971555e-07,
+      "loss": 1.9354,
+      "step": 194245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1096448481422783e-07,
+      "loss": 1.2218,
+      "step": 194250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1081010984939865e-07,
+      "loss": 1.2477,
+      "step": 194255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1065579078610842e-07,
+      "loss": 1.3013,
+      "step": 194260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1050152762523757e-07,
+      "loss": 1.0461,
+      "step": 194265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.103473203676687e-07,
+      "loss": 1.2596,
+      "step": 194270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1019316901428332e-07,
+      "loss": 1.2225,
+      "step": 194275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1003907356595964e-07,
+      "loss": 1.3478,
+      "step": 194280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0988503402357808e-07,
+      "loss": 1.3859,
+      "step": 194285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.09731050388019e-07,
+      "loss": 1.0777,
+      "step": 194290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.095771226601606e-07,
+      "loss": 1.3809,
+      "step": 194295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0942325084088332e-07,
+      "loss": 1.1474,
+      "step": 194300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0926943493106643e-07,
+      "loss": 1.3072,
+      "step": 194305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.091156749315859e-07,
+      "loss": 1.5382,
+      "step": 194310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0896197084331994e-07,
+      "loss": 1.3139,
+      "step": 194315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.088083226671478e-07,
+      "loss": 1.0644,
+      "step": 194320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0865473040394767e-07,
+      "loss": 1.3749,
+      "step": 194325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0850119405459335e-07,
+      "loss": 1.255,
+      "step": 194330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.083477136199652e-07,
+      "loss": 1.1969,
+      "step": 194335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0819428910093698e-07,
+      "loss": 1.0708,
+      "step": 194340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0804092049838575e-07,
+      "loss": 1.2319,
+      "step": 194345
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0788760781318863e-07,
+      "loss": 1.1925,
+      "step": 194350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0773435104622042e-07,
+      "loss": 1.0918,
+      "step": 194355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0758115019835378e-07,
+      "loss": 1.3358,
+      "step": 194360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0742800527046693e-07,
+      "loss": 1.2189,
+      "step": 194365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0727491626343243e-07,
+      "loss": 1.3662,
+      "step": 194370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.071218831781263e-07,
+      "loss": 1.1799,
+      "step": 194375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.069689060154212e-07,
+      "loss": 1.1669,
+      "step": 194380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.068159847761897e-07,
+      "loss": 1.1029,
+      "step": 194385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0666311946130667e-07,
+      "loss": 1.0669,
+      "step": 194390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.065103100716448e-07,
+      "loss": 1.2667,
+      "step": 194395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0635755660807665e-07,
+      "loss": 1.0459,
+      "step": 194400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0620485907147492e-07,
+      "loss": 1.1209,
+      "step": 194405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0605221746270886e-07,
+      "loss": 0.9433,
+      "step": 194410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.058996317826545e-07,
+      "loss": 1.2914,
+      "step": 194415
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0574710203218108e-07,
+      "loss": 1.1736,
+      "step": 194420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0559462821215903e-07,
+      "loss": 1.2066,
+      "step": 194425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.054422103234599e-07,
+      "loss": 1.7396,
+      "step": 194430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0528984836695298e-07,
+      "loss": 1.0308,
+      "step": 194435
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0513754234350868e-07,
+      "loss": 1.271,
+      "step": 194440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0498529225399856e-07,
+      "loss": 1.0438,
+      "step": 194445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0483309809929074e-07,
+      "loss": 1.2261,
+      "step": 194450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.046809598802535e-07,
+      "loss": 0.9978,
+      "step": 194455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0452887759775607e-07,
+      "loss": 1.0264,
+      "step": 194460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0437685125266783e-07,
+      "loss": 1.1246,
+      "step": 194465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.042248808458569e-07,
+      "loss": 1.2176,
+      "step": 194470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.040729663781893e-07,
+      "loss": 1.7677,
+      "step": 194475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0392110785053432e-07,
+      "loss": 1.2315,
+      "step": 194480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0376930526375904e-07,
+      "loss": 1.1169,
+      "step": 194485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0361755861872946e-07,
+      "loss": 1.0431,
+      "step": 194490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0346586791631373e-07,
+      "loss": 1.302,
+      "step": 194495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0331423315737564e-07,
+      "loss": 1.1369,
+      "step": 194500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0316265434278226e-07,
+      "loss": 1.1853,
+      "step": 194505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0301113147340067e-07,
+      "loss": 1.2453,
+      "step": 194510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0285966455009465e-07,
+      "loss": 1.2236,
+      "step": 194515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0270825357373015e-07,
+      "loss": 1.5585,
+      "step": 194520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.025568985451687e-07,
+      "loss": 1.2972,
+      "step": 194525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.024055994652785e-07,
+      "loss": 1.5299,
+      "step": 194530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0225435633492217e-07,
+      "loss": 1.5125,
+      "step": 194535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0210316915496131e-07,
+      "loss": 1.2381,
+      "step": 194540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0195203792626294e-07,
+      "loss": 1.4855,
+      "step": 194545
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0180096264968752e-07,
+      "loss": 1.1379,
+      "step": 194550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0164994332609878e-07,
+      "loss": 1.2262,
+      "step": 194555
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0149897995635936e-07,
+      "loss": 1.3624,
+      "step": 194560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0134807254133193e-07,
+      "loss": 1.0193,
+      "step": 194565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0119722108187577e-07,
+      "loss": 0.9439,
+      "step": 194570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0104642557885357e-07,
+      "loss": 1.2725,
+      "step": 194575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0089568603312682e-07,
+      "loss": 1.1275,
+      "step": 194580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0074500244555595e-07,
+      "loss": 1.1001,
+      "step": 194585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.005943748170025e-07,
+      "loss": 1.3956,
+      "step": 194590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0044380314832578e-07,
+      "loss": 1.3056,
+      "step": 194595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0029328744038512e-07,
+      "loss": 1.2159,
+      "step": 194600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0014282769404093e-07,
+      "loss": 1.1114,
+      "step": 194605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.999924239101525e-07,
+      "loss": 1.2283,
+      "step": 194610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9984207608957806e-07,
+      "loss": 1.4789,
+      "step": 194615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.996917842331747e-07,
+      "loss": 1.3445,
+      "step": 194620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9954154834180395e-07,
+      "loss": 1.1688,
+      "step": 194625
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.993913684163229e-07,
+      "loss": 0.9659,
+      "step": 194630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.992412444575864e-07,
+      "loss": 1.1665,
+      "step": 194635
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9909117646645604e-07,
+      "loss": 1.2411,
+      "step": 194640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9894116444378443e-07,
+      "loss": 1.2295,
+      "step": 194645
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9879120839043087e-07,
+      "loss": 0.85,
+      "step": 194650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9864130830725137e-07,
+      "loss": 1.5522,
+      "step": 194655
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9849146419510302e-07,
+      "loss": 1.2036,
+      "step": 194660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9834167605483845e-07,
+      "loss": 1.2878,
+      "step": 194665
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9819194388731366e-07,
+      "loss": 1.092,
+      "step": 194670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9804226769338684e-07,
+      "loss": 1.2185,
+      "step": 194675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9789264747391067e-07,
+      "loss": 1.2477,
+      "step": 194680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.977430832297389e-07,
+      "loss": 1.2611,
+      "step": 194685
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9759357496172638e-07,
+      "loss": 0.958,
+      "step": 194690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9744412267072687e-07,
+      "loss": 1.291,
+      "step": 194695
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9729472635759307e-07,
+      "loss": 1.3395,
+      "step": 194700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9714538602317978e-07,
+      "loss": 1.442,
+      "step": 194705
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9699610166833972e-07,
+      "loss": 1.2696,
+      "step": 194710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9684687329392216e-07,
+      "loss": 1.1969,
+      "step": 194715
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.966977009007831e-07,
+      "loss": 0.756,
+      "step": 194720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9654858448977297e-07,
+      "loss": 1.7847,
+      "step": 194725
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9639952406174222e-07,
+      "loss": 1.0737,
+      "step": 194730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9625051961754348e-07,
+      "loss": 1.1933,
+      "step": 194735
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.961015711580272e-07,
+      "loss": 1.4429,
+      "step": 194740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9595267868404489e-07,
+      "loss": 1.3175,
+      "step": 194745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9580384219644476e-07,
+      "loss": 1.309,
+      "step": 194750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9565506169607839e-07,
+      "loss": 1.0788,
+      "step": 194755
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9550633718379398e-07,
+      "loss": 1.1093,
+      "step": 194760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9535766866044303e-07,
+      "loss": 1.2787,
+      "step": 194765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9520905612687267e-07,
+      "loss": 1.2008,
+      "step": 194770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9506049958393335e-07,
+      "loss": 1.4335,
+      "step": 194775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9491199903246994e-07,
+      "loss": 1.0244,
+      "step": 194780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9476355447333396e-07,
+      "loss": 1.1634,
+      "step": 194785
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.946151659073725e-07,
+      "loss": 1.1683,
+      "step": 194790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.944668333354305e-07,
+      "loss": 1.1137,
+      "step": 194795
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9431855675835832e-07,
+      "loss": 1.4734,
+      "step": 194800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9417033617700199e-07,
+      "loss": 1.2473,
+      "step": 194805
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9402217159220526e-07,
+      "loss": 1.3611,
+      "step": 194810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9387406300481636e-07,
+      "loss": 1.4829,
+      "step": 194815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.937260104156813e-07,
+      "loss": 1.2992,
+      "step": 194820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.935780138256449e-07,
+      "loss": 1.1594,
+      "step": 194825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9343007323555209e-07,
+      "loss": 1.0516,
+      "step": 194830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.932821886462488e-07,
+      "loss": 1.0977,
+      "step": 194835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.931343600585789e-07,
+      "loss": 1.0389,
+      "step": 194840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9298658747338495e-07,
+      "loss": 1.1581,
+      "step": 194845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.928388708915141e-07,
+      "loss": 1.4633,
+      "step": 194850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9269121031380676e-07,
+      "loss": 0.9103,
+      "step": 194855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9254360574110785e-07,
+      "loss": 1.4783,
+      "step": 194860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9239605717426003e-07,
+      "loss": 1.132,
+      "step": 194865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.92248564614107e-07,
+      "loss": 1.1544,
+      "step": 194870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9210112806148818e-07,
+      "loss": 1.1479,
+      "step": 194875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9195374751724727e-07,
+      "loss": 1.2936,
+      "step": 194880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9180642298222585e-07,
+      "loss": 1.0639,
+      "step": 194885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9165915445726545e-07,
+      "loss": 1.0914,
+      "step": 194890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9151194194320655e-07,
+      "loss": 1.1974,
+      "step": 194895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9136478544088955e-07,
+      "loss": 1.1031,
+      "step": 194900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.91217684951156e-07,
+      "loss": 1.0942,
+      "step": 194905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9107064047484524e-07,
+      "loss": 1.3034,
+      "step": 194910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.909236520127966e-07,
+      "loss": 1.6464,
+      "step": 194915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9077671956584943e-07,
+      "loss": 1.3329,
+      "step": 194920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.90629843134843e-07,
+      "loss": 1.1321,
+      "step": 194925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.904830227206167e-07,
+      "loss": 1.2334,
+      "step": 194930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.903362583240087e-07,
+      "loss": 1.3306,
+      "step": 194935
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9018954994585503e-07,
+      "loss": 1.1894,
+      "step": 194940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9004289758699722e-07,
+      "loss": 1.329,
+      "step": 194945
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8989630124827018e-07,
+      "loss": 1.731,
+      "step": 194950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.89749760930511e-07,
+      "loss": 1.1608,
+      "step": 194955
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.896032766345579e-07,
+      "loss": 1.2977,
+      "step": 194960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.894568483612469e-07,
+      "loss": 1.3779,
+      "step": 194965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8931047611141284e-07,
+      "loss": 1.206,
+      "step": 194970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8916415988589287e-07,
+      "loss": 1.4216,
+      "step": 194975
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8901789968552296e-07,
+      "loss": 1.3379,
+      "step": 194980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8887169551113804e-07,
+      "loss": 1.3362,
+      "step": 194985
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8872554736357184e-07,
+      "loss": 1.1401,
+      "step": 194990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.885794552436593e-07,
+      "loss": 1.7043,
+      "step": 194995
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8843341915223635e-07,
+      "loss": 1.1267,
+      "step": 195000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8828743909013458e-07,
+      "loss": 1.239,
+      "step": 195005
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8814151505819e-07,
+      "loss": 1.2373,
+      "step": 195010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.879956470572353e-07,
+      "loss": 1.2561,
+      "step": 195015
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8784983508809973e-07,
+      "loss": 1.1404,
+      "step": 195020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.877040791516216e-07,
+      "loss": 1.2828,
+      "step": 195025
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.875583792486313e-07,
+      "loss": 1.4358,
+      "step": 195030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.874127353799593e-07,
+      "loss": 1.1574,
+      "step": 195035
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8726714754643827e-07,
+      "loss": 0.9335,
+      "step": 195040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8712161574889865e-07,
+      "loss": 1.0428,
+      "step": 195045
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.869761399881742e-07,
+      "loss": 1.0194,
+      "step": 195050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8683072026509207e-07,
+      "loss": 1.1838,
+      "step": 195055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8668535658048714e-07,
+      "loss": 1.083,
+      "step": 195060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.865400489351843e-07,
+      "loss": 1.4202,
+      "step": 195065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.863947973300173e-07,
+      "loss": 1.2168,
+      "step": 195070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8624960176581442e-07,
+      "loss": 1.1111,
+      "step": 195075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8610446224340494e-07,
+      "loss": 1.0171,
+      "step": 195080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8595937876361603e-07,
+      "loss": 1.4762,
+      "step": 195085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.858143513272781e-07,
+      "loss": 1.1399,
+      "step": 195090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.856693799352194e-07,
+      "loss": 0.9851,
+      "step": 195095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.85524464588267e-07,
+      "loss": 1.349,
+      "step": 195100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8537960528724807e-07,
+      "loss": 1.3094,
+      "step": 195105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8523480203299193e-07,
+      "loss": 1.2847,
+      "step": 195110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8509005482632237e-07,
+      "loss": 1.2431,
+      "step": 195115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8494536366806758e-07,
+      "loss": 1.5581,
+      "step": 195120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8480072855905472e-07,
+      "loss": 1.3417,
+      "step": 195125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8465614950010868e-07,
+      "loss": 1.047,
+      "step": 195130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8451162649205544e-07,
+      "loss": 1.0097,
+      "step": 195135
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8436715953571882e-07,
+      "loss": 1.2413,
+      "step": 195140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8422274863192592e-07,
+      "loss": 1.0197,
+      "step": 195145
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.840783937815005e-07,
+      "loss": 1.0843,
+      "step": 195150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.839340949852675e-07,
+      "loss": 1.4583,
+      "step": 195155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8378985224404954e-07,
+      "loss": 1.067,
+      "step": 195160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8364566555867157e-07,
+      "loss": 1.1136,
+      "step": 195165
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8350153492995626e-07,
+      "loss": 1.2058,
+      "step": 195170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8335746035872848e-07,
+      "loss": 1.7128,
+      "step": 195175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8321344184580758e-07,
+      "loss": 1.316,
+      "step": 195180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8306947939201847e-07,
+      "loss": 1.211,
+      "step": 195185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8292557299818382e-07,
+      "loss": 1.1509,
+      "step": 195190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8278172266512296e-07,
+      "loss": 1.2495,
+      "step": 195195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8263792839365967e-07,
+      "loss": 1.2135,
+      "step": 195200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8249419018461223e-07,
+      "loss": 1.5504,
+      "step": 195205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.823505080388055e-07,
+      "loss": 1.1458,
+      "step": 195210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.822068819570566e-07,
+      "loss": 1.101,
+      "step": 195215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8206331194018822e-07,
+      "loss": 1.116,
+      "step": 195220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.819197979890175e-07,
+      "loss": 1.1806,
+      "step": 195225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8177634010436596e-07,
+      "loss": 1.2974,
+      "step": 195230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8163293828705185e-07,
+      "loss": 1.3144,
+      "step": 195235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8148959253789567e-07,
+      "loss": 1.0004,
+      "step": 195240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8134630285771337e-07,
+      "loss": 1.2466,
+      "step": 195245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8120306924732544e-07,
+      "loss": 1.1928,
+      "step": 195250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.81059891707549e-07,
+      "loss": 1.3909,
+      "step": 195255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8091677023920228e-07,
+      "loss": 1.6426,
+      "step": 195260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8077370484310019e-07,
+      "loss": 1.2867,
+      "step": 195265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8063069552006317e-07,
+      "loss": 1.182,
+      "step": 195270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8048774227090616e-07,
+      "loss": 1.1872,
+      "step": 195275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.80344845096444e-07,
+      "loss": 1.5553,
+      "step": 195280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.802020039974961e-07,
+      "loss": 1.1703,
+      "step": 195285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8005921897487622e-07,
+      "loss": 1.3503,
+      "step": 195290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7991649002939925e-07,
+      "loss": 1.2985,
+      "step": 195295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7977381716188123e-07,
+      "loss": 1.2445,
+      "step": 195300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7963120037313597e-07,
+      "loss": 1.7259,
+      "step": 195305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7948863966397834e-07,
+      "loss": 1.208,
+      "step": 195310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7934613503522435e-07,
+      "loss": 0.9414,
+      "step": 195315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7920368648768448e-07,
+      "loss": 1.247,
+      "step": 195320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7906129402217365e-07,
+      "loss": 1.4412,
+      "step": 195325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7891895763950563e-07,
+      "loss": 1.1177,
+      "step": 195330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7877667734049198e-07,
+      "loss": 1.3702,
+      "step": 195335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7863445312594762e-07,
+      "loss": 1.2102,
+      "step": 195340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.784922849966808e-07,
+      "loss": 1.3052,
+      "step": 195345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7835017295350644e-07,
+      "loss": 1.1995,
+      "step": 195350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7820811699723494e-07,
+      "loss": 1.1476,
+      "step": 195355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7806611712867794e-07,
+      "loss": 1.3583,
+      "step": 195360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7792417334864698e-07,
+      "loss": 1.3982,
+      "step": 195365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.777822856579503e-07,
+      "loss": 1.187,
+      "step": 195370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7764045405739948e-07,
+      "loss": 1.1122,
+      "step": 195375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.774986785478061e-07,
+      "loss": 1.5936,
+      "step": 195380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7735695912997842e-07,
+      "loss": 1.0619,
+      "step": 195385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7721529580472463e-07,
+      "loss": 1.4124,
+      "step": 195390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7707368857285412e-07,
+      "loss": 1.2253,
+      "step": 195395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7693213743517846e-07,
+      "loss": 1.1081,
+      "step": 195400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7679064239250253e-07,
+      "loss": 1.0528,
+      "step": 195405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7664920344563463e-07,
+      "loss": 1.2228,
+      "step": 195410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7650782059538408e-07,
+      "loss": 1.3581,
+      "step": 195415
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7636649384255688e-07,
+      "loss": 1.1775,
+      "step": 195420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.762252231879602e-07,
+      "loss": 1.1636,
+      "step": 195425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7608400863240338e-07,
+      "loss": 1.1424,
+      "step": 195430
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.759428501766891e-07,
+      "loss": 1.0833,
+      "step": 195435
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7580174782162562e-07,
+      "loss": 0.9727,
+      "step": 195440
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7566070156801672e-07,
+      "loss": 1.6377,
+      "step": 195445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7551971141667068e-07,
+      "loss": 1.2165,
+      "step": 195450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7537877736839126e-07,
+      "loss": 1.2971,
+      "step": 195455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.752378994239834e-07,
+      "loss": 1.4352,
+      "step": 195460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7509707758424977e-07,
+      "loss": 1.2748,
+      "step": 195465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.749563118499975e-07,
+      "loss": 1.1851,
+      "step": 195470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.748156022220293e-07,
+      "loss": 1.0994,
+      "step": 195475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7467494870114787e-07,
+      "loss": 1.1377,
+      "step": 195480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7453435128815698e-07,
+      "loss": 1.306,
+      "step": 195485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7439380998385936e-07,
+      "loss": 1.2802,
+      "step": 195490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7425332478905876e-07,
+      "loss": 1.1054,
+      "step": 195495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.741128957045568e-07,
+      "loss": 1.3156,
+      "step": 195500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7397252273115395e-07,
+      "loss": 1.2217,
+      "step": 195505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.73832205869654e-07,
+      "loss": 0.9909,
+      "step": 195510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.736919451208552e-07,
+      "loss": 1.4006,
+      "step": 195515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7355174048556135e-07,
+      "loss": 1.0919,
+      "step": 195520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7341159196457403e-07,
+      "loss": 1.1472,
+      "step": 195525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7327149955869039e-07,
+      "loss": 1.2122,
+      "step": 195530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.73131463268712e-07,
+      "loss": 1.1,
+      "step": 195535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7299148309543713e-07,
+      "loss": 1.466,
+      "step": 195540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7285155903966844e-07,
+      "loss": 1.2007,
+      "step": 195545
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7271169110220197e-07,
+      "loss": 1.0143,
+      "step": 195550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7257187928383712e-07,
+      "loss": 1.094,
+      "step": 195555
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.724321235853721e-07,
+      "loss": 1.4099,
+      "step": 195560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7229242400760627e-07,
+      "loss": 1.1888,
+      "step": 195565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7215278055133567e-07,
+      "loss": 1.3183,
+      "step": 195570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7201319321736076e-07,
+      "loss": 1.5866,
+      "step": 195575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7187366200647426e-07,
+      "loss": 1.0726,
+      "step": 195580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7173418691947553e-07,
+      "loss": 1.2116,
+      "step": 195585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7159476795716168e-07,
+      "loss": 1.125,
+      "step": 195590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7145540512032766e-07,
+      "loss": 1.09,
+      "step": 195595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7131609840976948e-07,
+      "loss": 1.4642,
+      "step": 195600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7117684782628207e-07,
+      "loss": 1.4633,
+      "step": 195605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7103765337066258e-07,
+      "loss": 1.3511,
+      "step": 195610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.708985150437048e-07,
+      "loss": 1.2366,
+      "step": 195615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7075943284620146e-07,
+      "loss": 1.5418,
+      "step": 195620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7062040677894965e-07,
+      "loss": 1.3368,
+      "step": 195625
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.70481436842741e-07,
+      "loss": 1.2344,
+      "step": 195630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.703425230383704e-07,
+      "loss": 0.8928,
+      "step": 195635
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7020366536663168e-07,
+      "loss": 0.9017,
+      "step": 195640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7006486382831756e-07,
+      "loss": 1.4304,
+      "step": 195645
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6992611842421847e-07,
+      "loss": 1.2697,
+      "step": 195650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6978742915512935e-07,
+      "loss": 1.3889,
+      "step": 195655
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6964879602184182e-07,
+      "loss": 1.1497,
+      "step": 195660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.695102190251452e-07,
+      "loss": 1.3915,
+      "step": 195665
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6937169816583442e-07,
+      "loss": 1.0875,
+      "step": 195670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6923323344469665e-07,
+      "loss": 1.0713,
+      "step": 195675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.690948248625257e-07,
+      "loss": 0.9401,
+      "step": 195680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6895647242011094e-07,
+      "loss": 1.143,
+      "step": 195685
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.688181761182428e-07,
+      "loss": 1.2503,
+      "step": 195690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.686799359577107e-07,
+      "loss": 1.0022,
+      "step": 195695
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.685417519393029e-07,
+      "loss": 1.0765,
+      "step": 195700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6840362406380984e-07,
+      "loss": 1.406,
+      "step": 195705
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.682655523320209e-07,
+      "loss": 1.4422,
+      "step": 195710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6812753674472216e-07,
+      "loss": 1.1303,
+      "step": 195715
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6798957730270516e-07,
+      "loss": 1.4404,
+      "step": 195720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6785167400675484e-07,
+      "loss": 1.3509,
+      "step": 195725
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6771382685765836e-07,
+      "loss": 1.1223,
+      "step": 195730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.675760358562062e-07,
+      "loss": 1.2684,
+      "step": 195735
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6743830100318326e-07,
+      "loss": 1.3456,
+      "step": 195740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.673006222993756e-07,
+      "loss": 1.111,
+      "step": 195745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6716299974556928e-07,
+      "loss": 1.0985,
+      "step": 195750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6702543334255138e-07,
+      "loss": 1.0844,
+      "step": 195755
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6688792309110802e-07,
+      "loss": 1.5372,
+      "step": 195760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6675046899202186e-07,
+      "loss": 1.0773,
+      "step": 195765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6661307104607894e-07,
+      "loss": 0.9906,
+      "step": 195770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.664757292540653e-07,
+      "loss": 1.1138,
+      "step": 195775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6633844361676366e-07,
+      "loss": 1.0431,
+      "step": 195780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6620121413496004e-07,
+      "loss": 1.2793,
+      "step": 195785
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6606404080943494e-07,
+      "loss": 0.9195,
+      "step": 195790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.659269236409733e-07,
+      "loss": 1.2036,
+      "step": 195795
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6578986263036002e-07,
+      "loss": 1.1366,
+      "step": 195800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.656528577783756e-07,
+      "loss": 1.2004,
+      "step": 195805
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6551590908580163e-07,
+      "loss": 1.2474,
+      "step": 195810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.653790165534219e-07,
+      "loss": 1.1093,
+      "step": 195815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.65242180182017e-07,
+      "loss": 1.0624,
+      "step": 195820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.651053999723684e-07,
+      "loss": 1.311,
+      "step": 195825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.649686759252589e-07,
+      "loss": 1.5786,
+      "step": 195830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6483200804146894e-07,
+      "loss": 1.0432,
+      "step": 195835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6469539632177568e-07,
+      "loss": 1.3098,
+      "step": 195840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6455884076696294e-07,
+      "loss": 1.0649,
+      "step": 195845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.64422341377809e-07,
+      "loss": 1.2653,
+      "step": 195850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6428589815509432e-07,
+      "loss": 1.1015,
+      "step": 195855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.641495110995972e-07,
+      "loss": 1.1932,
+      "step": 195860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.640131802120948e-07,
+      "loss": 1.3408,
+      "step": 195865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6387690549336978e-07,
+      "loss": 1.4762,
+      "step": 195870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6374068694419597e-07,
+      "loss": 0.9348,
+      "step": 195875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6360452456535503e-07,
+      "loss": 1.6263,
+      "step": 195880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6346841835762184e-07,
+      "loss": 1.2385,
+      "step": 195885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6333236832177467e-07,
+      "loss": 1.2389,
+      "step": 195890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.631963744585907e-07,
+      "loss": 1.3195,
+      "step": 195895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.630604367688471e-07,
+      "loss": 1.6306,
+      "step": 195900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6292455525331763e-07,
+      "loss": 1.473,
+      "step": 195905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.627887299127795e-07,
+      "loss": 1.5485,
+      "step": 195910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6265296074800985e-07,
+      "loss": 1.439,
+      "step": 195915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6251724775978251e-07,
+      "loss": 1.3302,
+      "step": 195920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.623815909488724e-07,
+      "loss": 1.1676,
+      "step": 195925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.622459903160556e-07,
+      "loss": 1.1617,
+      "step": 195930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6211044586210477e-07,
+      "loss": 1.0439,
+      "step": 195935
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.619749575877938e-07,
+      "loss": 1.2102,
+      "step": 195940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.618395254938987e-07,
+      "loss": 1.084,
+      "step": 195945
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6170414958119106e-07,
+      "loss": 1.7029,
+      "step": 195950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6156882985044366e-07,
+      "loss": 0.9755,
+      "step": 195955
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6143356630243023e-07,
+      "loss": 1.169,
+      "step": 195960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6129835893792355e-07,
+      "loss": 1.585,
+      "step": 195965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6116320775769411e-07,
+      "loss": 1.0123,
+      "step": 195970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610281127625146e-07,
+      "loss": 1.5682,
+      "step": 195975
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6089307395315668e-07,
+      "loss": 1.1386,
+      "step": 195980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6075809133039187e-07,
+      "loss": 0.9802,
+      "step": 195985
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6062316489498852e-07,
+      "loss": 1.5902,
+      "step": 195990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6048829464772152e-07,
+      "loss": 1.1749,
+      "step": 195995
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6035348058935695e-07,
+      "loss": 1.3814,
+      "step": 196000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6021872272066642e-07,
+      "loss": 1.5044,
+      "step": 196005
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6008402104241927e-07,
+      "loss": 1.2811,
+      "step": 196010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5994937555538604e-07,
+      "loss": 0.92,
+      "step": 196015
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5981478626033277e-07,
+      "loss": 1.3225,
+      "step": 196020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5968025315803105e-07,
+      "loss": 1.1705,
+      "step": 196025
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5954577624924582e-07,
+      "loss": 1.1556,
+      "step": 196030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5941135553474872e-07,
+      "loss": 1.4436,
+      "step": 196035
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.592769910153047e-07,
+      "loss": 1.0197,
+      "step": 196040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.591426826916831e-07,
+      "loss": 1.4492,
+      "step": 196045
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5900843056464777e-07,
+      "loss": 1.1317,
+      "step": 196050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5887423463496808e-07,
+      "loss": 1.2984,
+      "step": 196055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5874009490341014e-07,
+      "loss": 1.0369,
+      "step": 196060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5860601137073994e-07,
+      "loss": 1.0611,
+      "step": 196065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5847198403772135e-07,
+      "loss": 1.2215,
+      "step": 196070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5833801290512042e-07,
+      "loss": 1.4795,
+      "step": 196075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5820409797370428e-07,
+      "loss": 1.152,
+      "step": 196080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.580702392442357e-07,
+      "loss": 1.3447,
+      "step": 196085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5793643671747961e-07,
+      "loss": 1.8064,
+      "step": 196090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5780269039420094e-07,
+      "loss": 1.0768,
+      "step": 196095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.576690002751602e-07,
+      "loss": 1.1295,
+      "step": 196100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5753536636112455e-07,
+      "loss": 1.4599,
+      "step": 196105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.574017886528567e-07,
+      "loss": 1.3734,
+      "step": 196110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.572682671511172e-07,
+      "loss": 1.1969,
+      "step": 196115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5713480185666986e-07,
+      "loss": 1.3613,
+      "step": 196120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5700139277027737e-07,
+      "loss": 1.28,
+      "step": 196125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5686803989270137e-07,
+      "loss": 1.1683,
+      "step": 196130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5673474322470127e-07,
+      "loss": 1.2861,
+      "step": 196135
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.56601502767042e-07,
+      "loss": 1.0394,
+      "step": 196140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5646831852048182e-07,
+      "loss": 1.1602,
+      "step": 196145
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5633519048578126e-07,
+      "loss": 1.1623,
+      "step": 196150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5620211866370195e-07,
+      "loss": 0.9651,
+      "step": 196155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5606910305500322e-07,
+      "loss": 1.1005,
+      "step": 196160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.559361436604434e-07,
+      "loss": 1.3946,
+      "step": 196165
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.55803240480783e-07,
+      "loss": 1.5634,
+      "step": 196170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5567039351678138e-07,
+      "loss": 1.014,
+      "step": 196175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5553760276919682e-07,
+      "loss": 1.1423,
+      "step": 196180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5540486823878653e-07,
+      "loss": 1.3302,
+      "step": 196185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5527218992630987e-07,
+      "loss": 1.4277,
+      "step": 196190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5513956783252514e-07,
+      "loss": 1.3652,
+      "step": 196195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.550070019581873e-07,
+      "loss": 1.33,
+      "step": 196200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.548744923040546e-07,
+      "loss": 1.0074,
+      "step": 196205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5474203887088423e-07,
+      "loss": 1.511,
+      "step": 196210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5460964165943228e-07,
+      "loss": 1.1658,
+      "step": 196215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5447730067045365e-07,
+      "loss": 1.2575,
+      "step": 196220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.543450159047055e-07,
+      "loss": 1.0314,
+      "step": 196225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5421278736294287e-07,
+      "loss": 1.3,
+      "step": 196230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5408061504592175e-07,
+      "loss": 0.9959,
+      "step": 196235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.539484989543949e-07,
+      "loss": 1.162,
+      "step": 196240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5381643908911725e-07,
+      "loss": 1.4027,
+      "step": 196245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5368443545084488e-07,
+      "loss": 1.3229,
+      "step": 196250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5355248804033052e-07,
+      "loss": 1.3158,
+      "step": 196255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5342059685832577e-07,
+      "loss": 1.7433,
+      "step": 196260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.532887619055856e-07,
+      "loss": 1.1109,
+      "step": 196265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5315698318286387e-07,
+      "loss": 1.2766,
+      "step": 196270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5302526069091217e-07,
+      "loss": 1.1149,
+      "step": 196275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5289359443048102e-07,
+      "loss": 1.2718,
+      "step": 196280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5276198440232426e-07,
+      "loss": 1.4569,
+      "step": 196285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.526304306071935e-07,
+      "loss": 1.0858,
+      "step": 196290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5249893304583818e-07,
+      "loss": 1.3399,
+      "step": 196295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.523674917190121e-07,
+      "loss": 1.2611,
+      "step": 196300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5223610662746359e-07,
+      "loss": 1.1111,
+      "step": 196305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.52104777771942e-07,
+      "loss": 1.2616,
+      "step": 196310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.519735051532012e-07,
+      "loss": 1.0657,
+      "step": 196315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5184228877198726e-07,
+      "loss": 1.195,
+      "step": 196320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5171112862905067e-07,
+      "loss": 1.1806,
+      "step": 196325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5158002472514088e-07,
+      "loss": 1.0679,
+      "step": 196330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5144897706100502e-07,
+      "loss": 1.4354,
+      "step": 196335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5131798563739476e-07,
+      "loss": 1.1163,
+      "step": 196340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5118705045505388e-07,
+      "loss": 1.2254,
+      "step": 196345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5105617151473296e-07,
+      "loss": 1.1174,
+      "step": 196350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5092534881717803e-07,
+      "loss": 1.4862,
+      "step": 196355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5079458236313738e-07,
+      "loss": 1.0739,
+      "step": 196360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.506638721533571e-07,
+      "loss": 1.2923,
+      "step": 196365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5053321818858435e-07,
+      "loss": 1.084,
+      "step": 196370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.50402620469563e-07,
+      "loss": 1.4597,
+      "step": 196375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.502720789970402e-07,
+      "loss": 1.2438,
+      "step": 196380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5014159377176208e-07,
+      "loss": 1.0794,
+      "step": 196385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5001116479447354e-07,
+      "loss": 1.1339,
+      "step": 196390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4988079206591955e-07,
+      "loss": 1.3065,
+      "step": 196395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4975047558684396e-07,
+      "loss": 1.2644,
+      "step": 196400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4962021535798955e-07,
+      "loss": 1.1567,
+      "step": 196405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4949001138010234e-07,
+      "loss": 0.979,
+      "step": 196410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4935986365392618e-07,
+      "loss": 1.2846,
+      "step": 196415
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4922977218020384e-07,
+      "loss": 1.7249,
+      "step": 196420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4909973695967695e-07,
+      "loss": 1.3285,
+      "step": 196425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4896975799308822e-07,
+      "loss": 1.6326,
+      "step": 196430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4883983528118152e-07,
+      "loss": 0.9424,
+      "step": 196435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4870996882469735e-07,
+      "loss": 1.1078,
+      "step": 196440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4858015862437847e-07,
+      "loss": 1.2761,
+      "step": 196445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4845040468096427e-07,
+      "loss": 1.122,
+      "step": 196450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4832070699519753e-07,
+      "loss": 1.0967,
+      "step": 196455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.481910655678187e-07,
+      "loss": 1.3295,
+      "step": 196460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4806148039956724e-07,
+      "loss": 0.9015,
+      "step": 196465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4793195149118367e-07,
+      "loss": 1.3284,
+      "step": 196470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4780247884340738e-07,
+      "loss": 1.4236,
+      "step": 196475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4767306245697887e-07,
+      "loss": 0.9231,
+      "step": 196480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4754370233263647e-07,
+      "loss": 0.9577,
+      "step": 196485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4741439847111738e-07,
+      "loss": 1.3211,
+      "step": 196490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.472851508731621e-07,
+      "loss": 1.1019,
+      "step": 196495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.471559595395078e-07,
+      "loss": 1.1524,
+      "step": 196500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4702682447089167e-07,
+      "loss": 1.0107,
+      "step": 196505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4689774566805314e-07,
+      "loss": 1.1994,
+      "step": 196510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4676872313172719e-07,
+      "loss": 1.2088,
+      "step": 196515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4663975686265097e-07,
+      "loss": 1.5886,
+      "step": 196520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4651084686156168e-07,
+      "loss": 1.1406,
+      "step": 196525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4638199312919544e-07,
+      "loss": 0.9871,
+      "step": 196530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4625319566628826e-07,
+      "loss": 1.1415,
+      "step": 196535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4612445447357514e-07,
+      "loss": 1.2584,
+      "step": 196540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4599576955178997e-07,
+      "loss": 0.876,
+      "step": 196545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4586714090166988e-07,
+      "loss": 1.1522,
+      "step": 196550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4573856852394875e-07,
+      "loss": 0.9242,
+      "step": 196555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4561005241936045e-07,
+      "loss": 1.027,
+      "step": 196560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.454815925886388e-07,
+      "loss": 1.145,
+      "step": 196565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.453531890325166e-07,
+      "loss": 1.1247,
+      "step": 196570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.452248417517299e-07,
+      "loss": 1.3759,
+      "step": 196575
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4509655074700924e-07,
+      "loss": 1.2224,
+      "step": 196580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4496831601908623e-07,
+      "loss": 1.4595,
+      "step": 196585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4484013756869587e-07,
+      "loss": 1.3671,
+      "step": 196590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4471201539656866e-07,
+      "loss": 1.0048,
+      "step": 196595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4458394950343625e-07,
+      "loss": 1.0951,
+      "step": 196600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4445593989003138e-07,
+      "loss": 1.3172,
+      "step": 196605
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4432798655708345e-07,
+      "loss": 1.2327,
+      "step": 196610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4420008950532306e-07,
+      "loss": 1.3674,
+      "step": 196615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4407224873548175e-07,
+      "loss": 1.2581,
+      "step": 196620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4394446424828902e-07,
+      "loss": 1.0857,
+      "step": 196625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4381673604447422e-07,
+      "loss": 1.1502,
+      "step": 196630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.436890641247668e-07,
+      "loss": 1.5371,
+      "step": 196635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4356144848989507e-07,
+      "loss": 1.1444,
+      "step": 196640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4343388914059064e-07,
+      "loss": 1.2613,
+      "step": 196645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4330638607757964e-07,
+      "loss": 1.1275,
+      "step": 196650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4317893930159032e-07,
+      "loss": 1.1294,
+      "step": 196655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4305154881334992e-07,
+      "loss": 1.016,
+      "step": 196660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4292421461358786e-07,
+      "loss": 1.0664,
+      "step": 196665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4279693670303018e-07,
+      "loss": 1.333,
+      "step": 196670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4266971508240412e-07,
+      "loss": 1.7327,
+      "step": 196675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4254254975243465e-07,
+      "loss": 1.0736,
+      "step": 196680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4241544071384893e-07,
+      "loss": 1.1686,
+      "step": 196685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.422883879673742e-07,
+      "loss": 1.0859,
+      "step": 196690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.421613915137332e-07,
+      "loss": 1.2091,
+      "step": 196695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4203445135365424e-07,
+      "loss": 1.2481,
+      "step": 196700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4190756748786117e-07,
+      "loss": 1.3243,
+      "step": 196705
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4178073991707675e-07,
+      "loss": 1.3188,
+      "step": 196710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4165396864202596e-07,
+      "loss": 1.4278,
+      "step": 196715
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.415272536634349e-07,
+      "loss": 1.6671,
+      "step": 196720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.414005949820252e-07,
+      "loss": 1.2892,
+      "step": 196725
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.412739925985196e-07,
+      "loss": 1.4092,
+      "step": 196730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.411474465136431e-07,
+      "loss": 1.2149,
+      "step": 196735
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4102095672811733e-07,
+      "loss": 1.4776,
+      "step": 196740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4089452324266394e-07,
+      "loss": 1.1418,
+      "step": 196745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.407681460580057e-07,
+      "loss": 1.1394,
+      "step": 196750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4064182517486423e-07,
+      "loss": 1.1287,
+      "step": 196755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.405155605939601e-07,
+      "loss": 1.3451,
+      "step": 196760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4038935231601604e-07,
+      "loss": 1.0801,
+      "step": 196765
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4026320034175257e-07,
+      "loss": 1.2895,
+      "step": 196770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4013710467188801e-07,
+      "loss": 1.2884,
+      "step": 196775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4001106530714404e-07,
+      "loss": 1.1103,
+      "step": 196780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3988508224824004e-07,
+      "loss": 1.0732,
+      "step": 196785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3975915549589547e-07,
+      "loss": 1.2119,
+      "step": 196790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3963328505082972e-07,
+      "loss": 1.3121,
+      "step": 196795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3950747091376116e-07,
+      "loss": 1.2015,
+      "step": 196800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3938171308540804e-07,
+      "loss": 1.2745,
+      "step": 196805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.392560115664898e-07,
+      "loss": 1.0126,
+      "step": 196810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.391303663577226e-07,
+      "loss": 1.1079,
+      "step": 196815
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3900477745982576e-07,
+      "loss": 1.0686,
+      "step": 196820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3887924487351434e-07,
+      "loss": 1.2776,
+      "step": 196825
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3875376859950552e-07,
+      "loss": 0.9935,
+      "step": 196830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.386283486385187e-07,
+      "loss": 1.5397,
+      "step": 196835
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.385029849912667e-07,
+      "loss": 1.5136,
+      "step": 196840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3837767765846665e-07,
+      "loss": 1.1819,
+      "step": 196845
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.382524266408347e-07,
+      "loss": 1.163,
+      "step": 196850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3812723193908472e-07,
+      "loss": 1.1227,
+      "step": 196855
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.380020935539328e-07,
+      "loss": 1.2204,
+      "step": 196860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.378770114860939e-07,
+      "loss": 1.2304,
+      "step": 196865
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.377519857362808e-07,
+      "loss": 1.1773,
+      "step": 196870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3762701630520847e-07,
+      "loss": 0.8904,
+      "step": 196875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.375021031935908e-07,
+      "loss": 1.2235,
+      "step": 196880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3737724640214056e-07,
+      "loss": 0.9382,
+      "step": 196885
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.372524459315705e-07,
+      "loss": 1.2607,
+      "step": 196890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.371277017825934e-07,
+      "loss": 1.3091,
+      "step": 196895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3700301395592308e-07,
+      "loss": 1.1435,
+      "step": 196900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3687838245226903e-07,
+      "loss": 1.0021,
+      "step": 196905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.367538072723451e-07,
+      "loss": 1.3034,
+      "step": 196910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3662928841686296e-07,
+      "loss": 1.0416,
+      "step": 196915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.365048258865309e-07,
+      "loss": 1.4373,
+      "step": 196920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3638041968206173e-07,
+      "loss": 1.1413,
+      "step": 196925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3625606980416594e-07,
+      "loss": 1.2392,
+      "step": 196930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.36131776253553e-07,
+      "loss": 1.4665,
+      "step": 196935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3600753903093344e-07,
+      "loss": 1.1847,
+      "step": 196940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3588335813701446e-07,
+      "loss": 1.1902,
+      "step": 196945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3575923357250887e-07,
+      "loss": 1.2818,
+      "step": 196950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3563516533812271e-07,
+      "loss": 1.0765,
+      "step": 196955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3551115343456544e-07,
+      "loss": 1.1145,
+      "step": 196960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.353871978625454e-07,
+      "loss": 1.1691,
+      "step": 196965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3526329862276976e-07,
+      "loss": 1.2466,
+      "step": 196970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3513945571594688e-07,
+      "loss": 1.301,
+      "step": 196975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3501566914278396e-07,
+      "loss": 1.102,
+      "step": 196980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.34891938903986e-07,
+      "loss": 1.3678,
+      "step": 196985
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.347682650002613e-07,
+      "loss": 1.2885,
+      "step": 196990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3464464743231708e-07,
+      "loss": 1.3167,
+      "step": 196995
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3452108620085725e-07,
+      "loss": 1.1609,
+      "step": 197000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3439758130658897e-07,
+      "loss": 1.4302,
+      "step": 197005
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3427413275021616e-07,
+      "loss": 1.2358,
+      "step": 197010
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.341507405324438e-07,
+      "loss": 1.2589,
+      "step": 197015
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.340274046539769e-07,
+      "loss": 1.0976,
+      "step": 197020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3390412511552154e-07,
+      "loss": 1.1766,
+      "step": 197025
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3378090191777937e-07,
+      "loss": 1.228,
+      "step": 197030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.336577350614543e-07,
+      "loss": 1.204,
+      "step": 197035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.335346245472502e-07,
+      "loss": 1.268,
+      "step": 197040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3341157037587093e-07,
+      "loss": 1.1115,
+      "step": 197045
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.332885725480171e-07,
+      "loss": 1.2363,
+      "step": 197050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3316563106439362e-07,
+      "loss": 1.0819,
+      "step": 197055
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3304274592569998e-07,
+      "loss": 1.5067,
+      "step": 197060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.329199171326412e-07,
+      "loss": 1.1299,
+      "step": 197065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3279714468591443e-07,
+      "loss": 1.4146,
+      "step": 197070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.326744285862247e-07,
+      "loss": 1.0787,
+      "step": 197075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3255176883427146e-07,
+      "loss": 1.1697,
+      "step": 197080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.32429165430753e-07,
+      "loss": 1.2751,
+      "step": 197085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3230661837637214e-07,
+      "loss": 1.4242,
+      "step": 197090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.321841276718283e-07,
+      "loss": 1.2868,
+      "step": 197095
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3206169331781982e-07,
+      "loss": 0.9926,
+      "step": 197100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3193931531504721e-07,
+      "loss": 1.0218,
+      "step": 197105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3181699366420885e-07,
+      "loss": 1.5409,
+      "step": 197110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3169472836600194e-07,
+      "loss": 1.1364,
+      "step": 197115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3157251942112703e-07,
+      "loss": 1.2618,
+      "step": 197120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3145036683028023e-07,
+      "loss": 1.3914,
+      "step": 197125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3132827059415986e-07,
+      "loss": 1.0501,
+      "step": 197130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3120623071346205e-07,
+      "loss": 1.0909,
+      "step": 197135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3108424718888624e-07,
+      "loss": 1.2766,
+      "step": 197140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.309623200211274e-07,
+      "loss": 1.058,
+      "step": 197145
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3084044921088056e-07,
+      "loss": 0.9948,
+      "step": 197150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3071863475884294e-07,
+      "loss": 1.4299,
+      "step": 197155
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3059687666571063e-07,
+      "loss": 1.115,
+      "step": 197160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3047517493217755e-07,
+      "loss": 1.3072,
+      "step": 197165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3035352955894088e-07,
+      "loss": 1.2561,
+      "step": 197170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3023194054669343e-07,
+      "loss": 1.2687,
+      "step": 197175
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3011040789612905e-07,
+      "loss": 1.2614,
+      "step": 197180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.29988931607945e-07,
+      "loss": 1.3076,
+      "step": 197185
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2986751168283186e-07,
+      "loss": 1.3698,
+      "step": 197190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2974614812148235e-07,
+      "loss": 0.9596,
+      "step": 197195
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2962484092459372e-07,
+      "loss": 1.3574,
+      "step": 197200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2950359009285428e-07,
+      "loss": 1.1998,
+      "step": 197205
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2938239562695798e-07,
+      "loss": 1.5151,
+      "step": 197210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2926125752759754e-07,
+      "loss": 1.5619,
+      "step": 197215
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2914017579546468e-07,
+      "loss": 1.185,
+      "step": 197220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.29019150431251e-07,
+      "loss": 1.2543,
+      "step": 197225
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2889818143564603e-07,
+      "loss": 1.2989,
+      "step": 197230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.287772688093414e-07,
+      "loss": 1.1227,
+      "step": 197235
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2865641255302873e-07,
+      "loss": 1.2291,
+      "step": 197240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2853561266739756e-07,
+      "loss": 1.4549,
+      "step": 197245
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2841486915313618e-07,
+      "loss": 1.1374,
+      "step": 197250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2829418201093625e-07,
+      "loss": 1.1878,
+      "step": 197255
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.281735512414861e-07,
+      "loss": 0.9594,
+      "step": 197260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2805297684547412e-07,
+      "loss": 1.2963,
+      "step": 197265
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2793245882358861e-07,
+      "loss": 1.4465,
+      "step": 197270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2781199717651904e-07,
+      "loss": 1.5493,
+      "step": 197275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2769159190495372e-07,
+      "loss": 1.0936,
+      "step": 197280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2757124300957878e-07,
+      "loss": 1.009,
+      "step": 197285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2745095049108257e-07,
+      "loss": 1.2543,
+      "step": 197290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2733071435014898e-07,
+      "loss": 1.3214,
+      "step": 197295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2721053458746967e-07,
+      "loss": 1.085,
+      "step": 197300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2709041120372746e-07,
+      "loss": 1.304,
+      "step": 197305
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.269703441996084e-07,
+      "loss": 1.3918,
+      "step": 197310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2685033357579978e-07,
+      "loss": 1.1365,
+      "step": 197315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2673037933298548e-07,
+      "loss": 1.1604,
+      "step": 197320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2661048147185163e-07,
+      "loss": 1.3334,
+      "step": 197325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2649063999308208e-07,
+      "loss": 1.323,
+      "step": 197330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2637085489736188e-07,
+      "loss": 1.3842,
+      "step": 197335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2625112618537384e-07,
+      "loss": 1.2052,
+      "step": 197340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.261314538578029e-07,
+      "loss": 1.2648,
+      "step": 197345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2601183791533188e-07,
+      "loss": 1.4454,
+      "step": 197350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.258922783586436e-07,
+      "loss": 1.3561,
+      "step": 197355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.25772775188423e-07,
+      "loss": 1.4186,
+      "step": 197360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2565332840534848e-07,
+      "loss": 1.0868,
+      "step": 197365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2553393801010617e-07,
+      "loss": 1.2292,
+      "step": 197370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2541460400337435e-07,
+      "loss": 1.1005,
+      "step": 197375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2529532638583807e-07,
+      "loss": 1.1287,
+      "step": 197380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.251761051581757e-07,
+      "loss": 1.344,
+      "step": 197385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2505694032106884e-07,
+      "loss": 1.6586,
+      "step": 197390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2493783187519814e-07,
+      "loss": 0.8866,
+      "step": 197395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2481877982124412e-07,
+      "loss": 1.2695,
+      "step": 197400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2469978415988627e-07,
+      "loss": 1.1673,
+      "step": 197405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.24580844891804e-07,
+      "loss": 1.194,
+      "step": 197410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2446196201767569e-07,
+      "loss": 1.8744,
+      "step": 197415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.243431355381819e-07,
+      "loss": 1.0164,
+      "step": 197420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2422436545400097e-07,
+      "loss": 1.142,
+      "step": 197425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2410565176581013e-07,
+      "loss": 1.1255,
+      "step": 197430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2398699447428775e-07,
+      "loss": 1.4423,
+      "step": 197435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2386839358011106e-07,
+      "loss": 0.9777,
+      "step": 197440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2374984908395837e-07,
+      "loss": 1.3289,
+      "step": 197445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2363136098650585e-07,
+      "loss": 1.196,
+      "step": 197450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2351292928843072e-07,
+      "loss": 1.0649,
+      "step": 197455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.23394553990408e-07,
+      "loss": 1.1036,
+      "step": 197460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2327623509311492e-07,
+      "loss": 1.1389,
+      "step": 197465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.231579725972265e-07,
+      "loss": 1.1613,
+      "step": 197470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2303976650341886e-07,
+      "loss": 1.4795,
+      "step": 197475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2292161681236593e-07,
+      "loss": 1.4356,
+      "step": 197480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2280352352474266e-07,
+      "loss": 1.7361,
+      "step": 197485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2268548664122526e-07,
+      "loss": 1.3008,
+      "step": 197490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2256750616248535e-07,
+      "loss": 1.279,
+      "step": 197495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.22449582089198e-07,
+      "loss": 1.0897,
+      "step": 197500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2233171442203595e-07,
+      "loss": 1.0472,
+      "step": 197505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2221390316167204e-07,
+      "loss": 1.2933,
+      "step": 197510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2209614830878013e-07,
+      "loss": 1.1398,
+      "step": 197515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2197844986403197e-07,
+      "loss": 1.2264,
+      "step": 197520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2186080782809916e-07,
+      "loss": 1.29,
+      "step": 197525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2174322220165458e-07,
+      "loss": 1.2006,
+      "step": 197530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2162569298536875e-07,
+      "loss": 1.3503,
+      "step": 197535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.215082201799145e-07,
+      "loss": 1.2956,
+      "step": 197540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2139080378596124e-07,
+      "loss": 1.2664,
+      "step": 197545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.212734438041785e-07,
+      "loss": 1.2916,
+      "step": 197550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2115614023523903e-07,
+      "loss": 1.526,
+      "step": 197555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2103889307981121e-07,
+      "loss": 1.0682,
+      "step": 197560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2092170233856336e-07,
+      "loss": 1.3989,
+      "step": 197565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2080456801216723e-07,
+      "loss": 0.9954,
+      "step": 197570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2068749010129e-07,
+      "loss": 1.7426,
+      "step": 197575
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2057046860660225e-07,
+      "loss": 1.3413,
+      "step": 197580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2045350352876905e-07,
+      "loss": 1.0116,
+      "step": 197585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2033659486846093e-07,
+      "loss": 1.1677,
+      "step": 197590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2021974262634517e-07,
+      "loss": 1.1561,
+      "step": 197595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2010294680308675e-07,
+      "loss": 1.3071,
+      "step": 197600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.199862073993563e-07,
+      "loss": 1.3921,
+      "step": 197605
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1986952441581768e-07,
+      "loss": 1.3503,
+      "step": 197610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1975289785313815e-07,
+      "loss": 1.0842,
+      "step": 197615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1963632771198386e-07,
+      "loss": 1.1697,
+      "step": 197620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1951981399302094e-07,
+      "loss": 1.0,
+      "step": 197625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1940335669691215e-07,
+      "loss": 1.1065,
+      "step": 197630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1928695582432703e-07,
+      "loss": 1.3842,
+      "step": 197635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1917061137592612e-07,
+      "loss": 1.2808,
+      "step": 197640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1905432335237555e-07,
+      "loss": 1.594,
+      "step": 197645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1893809175433924e-07,
+      "loss": 1.1772,
+      "step": 197650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1882191658248221e-07,
+      "loss": 1.0766,
+      "step": 197655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1870579783746617e-07,
+      "loss": 1.2898,
+      "step": 197660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.185897355199539e-07,
+      "loss": 1.0786,
+      "step": 197665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1847372963060932e-07,
+      "loss": 1.2517,
+      "step": 197670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1835778017009414e-07,
+      "loss": 1.6705,
+      "step": 197675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1824188713907114e-07,
+      "loss": 1.2474,
+      "step": 197680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1812605053820314e-07,
+      "loss": 1.2917,
+      "step": 197685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.180102703681496e-07,
+      "loss": 1.1591,
+      "step": 197690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1789454662957111e-07,
+      "loss": 1.3063,
+      "step": 197695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1777887932313159e-07,
+      "loss": 1.077,
+      "step": 197700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1766326844948938e-07,
+      "loss": 1.124,
+      "step": 197705
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1754771400930509e-07,
+      "loss": 1.556,
+      "step": 197710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1743221600323817e-07,
+      "loss": 1.0269,
+      "step": 197715
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1731677443194922e-07,
+      "loss": 1.1797,
+      "step": 197720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1720138929609659e-07,
+      "loss": 1.1412,
+      "step": 197725
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1708606059633976e-07,
+      "loss": 1.2489,
+      "step": 197730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1697078833333708e-07,
+      "loss": 1.4101,
+      "step": 197735
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1685557250774693e-07,
+      "loss": 1.2058,
+      "step": 197740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1674041312022655e-07,
+      "loss": 0.9808,
+      "step": 197745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1662531017143431e-07,
+      "loss": 1.6134,
+      "step": 197750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1651026366202855e-07,
+      "loss": 1.2097,
+      "step": 197755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1639527359266322e-07,
+      "loss": 1.3302,
+      "step": 197760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1628033996399779e-07,
+      "loss": 1.4857,
+      "step": 197765
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1616546277668727e-07,
+      "loss": 1.4844,
+      "step": 197770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1605064203138782e-07,
+      "loss": 1.4554,
+      "step": 197775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1593587772875559e-07,
+      "loss": 0.9968,
+      "step": 197780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1582116986944669e-07,
+      "loss": 1.2335,
+      "step": 197785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1570651845411284e-07,
+      "loss": 1.2593,
+      "step": 197790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1559192348341242e-07,
+      "loss": 1.3295,
+      "step": 197795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1547738495799932e-07,
+      "loss": 1.5473,
+      "step": 197800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1536290287852636e-07,
+      "loss": 1.1401,
+      "step": 197805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1524847724564637e-07,
+      "loss": 1.5148,
+      "step": 197810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1513410806001546e-07,
+      "loss": 1.1302,
+      "step": 197815
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1501979532228425e-07,
+      "loss": 1.4431,
+      "step": 197820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1490553903310774e-07,
+      "loss": 1.109,
+      "step": 197825
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1479133919313656e-07,
+      "loss": 1.8076,
+      "step": 197830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.146771958030235e-07,
+      "loss": 1.4233,
+      "step": 197835
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1456310886342137e-07,
+      "loss": 1.1358,
+      "step": 197840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1444907837498076e-07,
+      "loss": 1.3185,
+      "step": 197845
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1433510433835227e-07,
+      "loss": 1.428,
+      "step": 197850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1422118675418759e-07,
+      "loss": 1.1143,
+      "step": 197855
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.141073256231362e-07,
+      "loss": 1.0721,
+      "step": 197860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1399352094585092e-07,
+      "loss": 1.0154,
+      "step": 197865
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1387977272297901e-07,
+      "loss": 1.1808,
+      "step": 197870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1376608095516994e-07,
+      "loss": 1.6427,
+      "step": 197875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1365244564307431e-07,
+      "loss": 1.2539,
+      "step": 197880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1353886678734161e-07,
+      "loss": 1.0364,
+      "step": 197885
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1342534438861908e-07,
+      "loss": 1.3354,
+      "step": 197890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1331187844755509e-07,
+      "loss": 0.9705,
+      "step": 197895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1319846896479802e-07,
+      "loss": 1.1747,
+      "step": 197900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1308511594099514e-07,
+      "loss": 1.3623,
+      "step": 197905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1297181937679369e-07,
+      "loss": 1.1793,
+      "step": 197910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1285857927284095e-07,
+      "loss": 1.1484,
+      "step": 197915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1274539562978415e-07,
+      "loss": 1.4753,
+      "step": 197920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1263226844826836e-07,
+      "loss": 1.291,
+      "step": 197925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1251919772894082e-07,
+      "loss": 1.158,
+      "step": 197930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1240618347244547e-07,
+      "loss": 1.3334,
+      "step": 197935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1229322567943068e-07,
+      "loss": 1.413,
+      "step": 197940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1218032435053817e-07,
+      "loss": 0.8617,
+      "step": 197945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1206747948641517e-07,
+      "loss": 1.4152,
+      "step": 197950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1195469108770452e-07,
+      "loss": 1.1225,
+      "step": 197955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1184195915505125e-07,
+      "loss": 1.2735,
+      "step": 197960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.117292836890993e-07,
+      "loss": 1.0973,
+      "step": 197965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1161666469049038e-07,
+      "loss": 1.1378,
+      "step": 197970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1150410215986839e-07,
+      "loss": 1.2543,
+      "step": 197975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1139159609787842e-07,
+      "loss": 1.4738,
+      "step": 197980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1127914650515992e-07,
+      "loss": 1.2727,
+      "step": 197985
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1116675338235682e-07,
+      "loss": 1.2377,
+      "step": 197990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1105441673010975e-07,
+      "loss": 1.2823,
+      "step": 197995
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1094213654906038e-07,
+      "loss": 0.7698,
+      "step": 198000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1082991283985046e-07,
+      "loss": 1.3271,
+      "step": 198005
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1071774560312164e-07,
+      "loss": 1.5744,
+      "step": 198010
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1060563483951237e-07,
+      "loss": 1.1761,
+      "step": 198015
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.104935805496643e-07,
+      "loss": 1.3613,
+      "step": 198020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1038158273421695e-07,
+      "loss": 1.5687,
+      "step": 198025
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.102696413938098e-07,
+      "loss": 1.2383,
+      "step": 198030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1015775652908234e-07,
+      "loss": 1.3164,
+      "step": 198035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1004592814067405e-07,
+      "loss": 1.1432,
+      "step": 198040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.099341562292211e-07,
+      "loss": 1.3118,
+      "step": 198045
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0982244079536408e-07,
+      "loss": 1.5294,
+      "step": 198050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0971078183974138e-07,
+      "loss": 1.4494,
+      "step": 198055
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0959917936299025e-07,
+      "loss": 1.2405,
+      "step": 198060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0948763336574574e-07,
+      "loss": 1.1617,
+      "step": 198065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0937614384864626e-07,
+      "loss": 1.2289,
+      "step": 198070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0926471081233014e-07,
+      "loss": 0.9621,
+      "step": 198075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0915333425743136e-07,
+      "loss": 1.2251,
+      "step": 198080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0904201418458716e-07,
+      "loss": 0.9997,
+      "step": 198085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.089307505944337e-07,
+      "loss": 1.287,
+      "step": 198090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0881954348760382e-07,
+      "loss": 1.1964,
+      "step": 198095
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0870839286473477e-07,
+      "loss": 1.0566,
+      "step": 198100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0859729872646163e-07,
+      "loss": 1.149,
+      "step": 198105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.084862610734183e-07,
+      "loss": 1.1582,
+      "step": 198110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0837527990623763e-07,
+      "loss": 1.2715,
+      "step": 198115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0826435522555467e-07,
+      "loss": 1.035,
+      "step": 198120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0815348703200335e-07,
+      "loss": 1.1454,
+      "step": 198125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0804267532621426e-07,
+      "loss": 1.4226,
+      "step": 198130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0793192010882358e-07,
+      "loss": 1.4088,
+      "step": 198135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.078212213804608e-07,
+      "loss": 1.34,
+      "step": 198140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0771057914175987e-07,
+      "loss": 1.7312,
+      "step": 198145
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0759999339335248e-07,
+      "loss": 1.2331,
+      "step": 198150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0748946413586925e-07,
+      "loss": 1.2618,
+      "step": 198155
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.073789913699419e-07,
+      "loss": 1.2894,
+      "step": 198160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0726857509620214e-07,
+      "loss": 1.1787,
+      "step": 198165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0715821531527948e-07,
+      "loss": 1.2997,
+      "step": 198170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.070479120278034e-07,
+      "loss": 1.3882,
+      "step": 198175
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0693766523440562e-07,
+      "loss": 1.2944,
+      "step": 198180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0682747493571344e-07,
+      "loss": 1.3658,
+      "step": 198185
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0671734113235854e-07,
+      "loss": 1.0083,
+      "step": 198190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0660726382496823e-07,
+      "loss": 1.1584,
+      "step": 198195
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0649724301417197e-07,
+      "loss": 1.1764,
+      "step": 198200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0638727870059817e-07,
+      "loss": 1.2167,
+      "step": 198205
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.062773708848741e-07,
+      "loss": 1.3002,
+      "step": 198210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0616751956762705e-07,
+      "loss": 1.3145,
+      "step": 198215
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0605772474948539e-07,
+      "loss": 1.2439,
+      "step": 198220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0594798643107418e-07,
+      "loss": 1.2713,
+      "step": 198225
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0583830461302292e-07,
+      "loss": 1.0955,
+      "step": 198230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0572867929595665e-07,
+      "loss": 1.2245,
+      "step": 198235
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0561911048050156e-07,
+      "loss": 0.9223,
+      "step": 198240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0550959816728156e-07,
+      "loss": 1.2303,
+      "step": 198245
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0540014235692509e-07,
+      "loss": 1.5392,
+      "step": 198250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0529074305005382e-07,
+      "loss": 1.2084,
+      "step": 198255
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0518140024729506e-07,
+      "loss": 1.258,
+      "step": 198260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0507211394927386e-07,
+      "loss": 1.1132,
+      "step": 198265
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0496288415661193e-07,
+      "loss": 1.3907,
+      "step": 198270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0485371086993324e-07,
+      "loss": 1.2715,
+      "step": 198275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0474459408986282e-07,
+      "loss": 1.3151,
+      "step": 198280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0463553381702352e-07,
+      "loss": 1.2441,
+      "step": 198285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0452653005203594e-07,
+      "loss": 1.2387,
+      "step": 198290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0441758279552516e-07,
+      "loss": 1.158,
+      "step": 198295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0430869204811289e-07,
+      "loss": 1.0765,
+      "step": 198300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0419985781041974e-07,
+      "loss": 1.353,
+      "step": 198305
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0409108008306745e-07,
+      "loss": 1.0093,
+      "step": 198310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0398235886667884e-07,
+      "loss": 1.386,
+      "step": 198315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0387369416187232e-07,
+      "loss": 1.3583,
+      "step": 198320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.037650859692696e-07,
+      "loss": 1.5385,
+      "step": 198325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.036565342894913e-07,
+      "loss": 1.2083,
+      "step": 198330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0354803912315803e-07,
+      "loss": 1.3479,
+      "step": 198335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0343960047088708e-07,
+      "loss": 1.0631,
+      "step": 198340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0333121833329907e-07,
+      "loss": 0.9442,
+      "step": 198345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0322289271101349e-07,
+      "loss": 1.3533,
+      "step": 198350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0311462360464652e-07,
+      "loss": 1.1335,
+      "step": 198355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.030064110148199e-07,
+      "loss": 1.5662,
+      "step": 198360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0289825494214978e-07,
+      "loss": 1.2542,
+      "step": 198365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0279015538725235e-07,
+      "loss": 1.1974,
+      "step": 198370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.026821123507471e-07,
+      "loss": 1.1383,
+      "step": 198375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0257412583325022e-07,
+      "loss": 1.3235,
+      "step": 198380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0246619583537898e-07,
+      "loss": 1.2009,
+      "step": 198385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0235832235774845e-07,
+      "loss": 1.1019,
+      "step": 198390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0225050540097592e-07,
+      "loss": 1.1709,
+      "step": 198395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0214274496567644e-07,
+      "loss": 1.3436,
+      "step": 198400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0203504105246509e-07,
+      "loss": 1.1336,
+      "step": 198405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0192739366195803e-07,
+      "loss": 1.2873,
+      "step": 198410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0181980279476811e-07,
+      "loss": 1.3994,
+      "step": 198415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0171226845151149e-07,
+      "loss": 1.2379,
+      "step": 198420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0160479063280104e-07,
+      "loss": 1.2094,
+      "step": 198425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0149736933925292e-07,
+      "loss": 1.2536,
+      "step": 198430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0139000457147662e-07,
+      "loss": 1.0674,
+      "step": 198435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0128269633008836e-07,
+      "loss": 1.2312,
+      "step": 198440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0117544461569983e-07,
+      "loss": 1.3476,
+      "step": 198445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.010682494289239e-07,
+      "loss": 1.5499,
+      "step": 198450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.009611107703723e-07,
+      "loss": 1.0009,
+      "step": 198455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0085402864065674e-07,
+      "loss": 1.3402,
+      "step": 198460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0074700304038899e-07,
+      "loss": 1.3298,
+      "step": 198465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0064003397018075e-07,
+      "loss": 1.4451,
+      "step": 198470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0053312143064154e-07,
+      "loss": 1.7974,
+      "step": 198475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.004262654223831e-07,
+      "loss": 1.643,
+      "step": 198480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0031946594601493e-07,
+      "loss": 1.3437,
+      "step": 198485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0021272300214768e-07,
+      "loss": 1.3493,
+      "step": 198490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0010603659138973e-07,
+      "loss": 1.1279,
+      "step": 198495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.99994067143517e-08,
+      "loss": 1.141,
+      "step": 198500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.98928333716409e-08,
+      "loss": 1.0977,
+      "step": 198505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.978631656386795e-08,
+      "loss": 1.4082,
+      "step": 198510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.967985629163901e-08,
+      "loss": 1.4359,
+      "step": 198515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.957345255556362e-08,
+      "loss": 1.1259,
+      "step": 198520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.946710535624793e-08,
+      "loss": 1.2817,
+      "step": 198525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.936081469430147e-08,
+      "loss": 1.2211,
+      "step": 198530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.925458057032933e-08,
+      "loss": 1.2015,
+      "step": 198535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.914840298493766e-08,
+      "loss": 1.1013,
+      "step": 198540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.904228193873489e-08,
+      "loss": 1.4427,
+      "step": 198545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.893621743232495e-08,
+      "loss": 1.1739,
+      "step": 198550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.883020946631517e-08,
+      "loss": 1.3734,
+      "step": 198555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.872425804131058e-08,
+      "loss": 0.9423,
+      "step": 198560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.861836315791518e-08,
+      "loss": 1.8839,
+      "step": 198565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.851252481673512e-08,
+      "loss": 1.1637,
+      "step": 198570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.840674301837438e-08,
+      "loss": 1.2387,
+      "step": 198575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.830101776343581e-08,
+      "loss": 1.1165,
+      "step": 198580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.819534905252337e-08,
+      "loss": 1.3261,
+      "step": 198585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.808973688624324e-08,
+      "loss": 1.1181,
+      "step": 198590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.798418126519493e-08,
+      "loss": 1.2262,
+      "step": 198595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.787868218998353e-08,
+      "loss": 1.482,
+      "step": 198600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.777323966121077e-08,
+      "loss": 1.5498,
+      "step": 198605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.76678536794784e-08,
+      "loss": 1.2514,
+      "step": 198610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.756252424538815e-08,
+      "loss": 1.1634,
+      "step": 198615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.745725135954288e-08,
+      "loss": 1.281,
+      "step": 198620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.735203502254209e-08,
+      "loss": 1.1341,
+      "step": 198625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.724687523498755e-08,
+      "loss": 0.8446,
+      "step": 198630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.714177199747987e-08,
+      "loss": 1.3632,
+      "step": 198635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.703672531061858e-08,
+      "loss": 1.1461,
+      "step": 198640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.69317351750032e-08,
+      "loss": 1.1683,
+      "step": 198645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.682680159123547e-08,
+      "loss": 0.9008,
+      "step": 198650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.672192455991269e-08,
+      "loss": 1.1303,
+      "step": 198655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.661710408163327e-08,
+      "loss": 1.2784,
+      "step": 198660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.651234015699783e-08,
+      "loss": 1.0295,
+      "step": 198665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.64076327866037e-08,
+      "loss": 0.9995,
+      "step": 198670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.630298197104815e-08,
+      "loss": 1.3194,
+      "step": 198675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.619838771092849e-08,
+      "loss": 1.2746,
+      "step": 198680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.609385000684424e-08,
+      "loss": 1.1326,
+      "step": 198685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.598936885939048e-08,
+      "loss": 1.4753,
+      "step": 198690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.58849442691645e-08,
+      "loss": 1.293,
+      "step": 198695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.578057623676252e-08,
+      "loss": 1.2579,
+      "step": 198700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.567626476277958e-08,
+      "loss": 1.3249,
+      "step": 198705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.55720098478119e-08,
+      "loss": 1.0977,
+      "step": 198710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.546781149245676e-08,
+      "loss": 1.4281,
+      "step": 198715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.536366969730703e-08,
+      "loss": 1.3659,
+      "step": 198720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.52595844629578e-08,
+      "loss": 1.3537,
+      "step": 198725
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.515555579000302e-08,
+      "loss": 0.9498,
+      "step": 198730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.505158367903777e-08,
+      "loss": 1.3188,
+      "step": 198735
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.494766813065603e-08,
+      "loss": 1.6729,
+      "step": 198740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.484380914544955e-08,
+      "loss": 1.1687,
+      "step": 198745
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.47400067240134e-08,
+      "loss": 1.4168,
+      "step": 198750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.463626086693934e-08,
+      "loss": 1.0719,
+      "step": 198755
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.453257157481909e-08,
+      "loss": 1.1304,
+      "step": 198760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.442893884824556e-08,
+      "loss": 1.3859,
+      "step": 198765
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.432536268781157e-08,
+      "loss": 0.9884,
+      "step": 198770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.422184309410554e-08,
+      "loss": 1.0037,
+      "step": 198775
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.411838006772256e-08,
+      "loss": 1.6351,
+      "step": 198780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.401497360925105e-08,
+      "loss": 1.2941,
+      "step": 198785
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.391162371928165e-08,
+      "loss": 1.1536,
+      "step": 198790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.380833039840608e-08,
+      "loss": 1.3026,
+      "step": 198795
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.370509364721169e-08,
+      "loss": 1.0153,
+      "step": 198800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.36019134662891e-08,
+      "loss": 1.2461,
+      "step": 198805
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.349878985622896e-08,
+      "loss": 1.1898,
+      "step": 198810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.339572281761855e-08,
+      "loss": 1.0188,
+      "step": 198815
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.329271235104631e-08,
+      "loss": 1.441,
+      "step": 198820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.318975845710065e-08,
+      "loss": 1.2566,
+      "step": 198825
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.30868611363711e-08,
+      "loss": 1.2205,
+      "step": 198830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.298402038944387e-08,
+      "loss": 0.9768,
+      "step": 198835
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.288123621690515e-08,
+      "loss": 1.2935,
+      "step": 198840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.277850861934446e-08,
+      "loss": 1.3696,
+      "step": 198845
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.267583759734688e-08,
+      "loss": 1.3785,
+      "step": 198850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.257322315149864e-08,
+      "loss": 1.0929,
+      "step": 198855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.24706652823859e-08,
+      "loss": 1.1651,
+      "step": 198860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.23681639905949e-08,
+      "loss": 1.2618,
+      "step": 198865
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.226571927670957e-08,
+      "loss": 1.1949,
+      "step": 198870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.216333114131726e-08,
+      "loss": 1.3006,
+      "step": 198875
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.20609995850008e-08,
+      "loss": 1.267,
+      "step": 198880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.195872460834532e-08,
+      "loss": 1.0612,
+      "step": 198885
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.185650621193365e-08,
+      "loss": 1.2776,
+      "step": 198890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.175434439635089e-08,
+      "loss": 1.3735,
+      "step": 198895
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.165223916217992e-08,
+      "loss": 0.9882,
+      "step": 198900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.155019051000358e-08,
+      "loss": 1.3738,
+      "step": 198905
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.144819844040586e-08,
+      "loss": 1.2193,
+      "step": 198910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.134626295396743e-08,
+      "loss": 1.2052,
+      "step": 198915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.124438405127111e-08,
+      "loss": 1.1932,
+      "step": 198920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.114256173289982e-08,
+      "loss": 1.0964,
+      "step": 198925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.104079599943305e-08,
+      "loss": 1.1866,
+      "step": 198930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.093908685145369e-08,
+      "loss": 1.324,
+      "step": 198935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.083743428954128e-08,
+      "loss": 1.3722,
+      "step": 198940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.073583831427645e-08,
+      "loss": 1.4614,
+      "step": 198945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.063429892623987e-08,
+      "loss": 1.1695,
+      "step": 198950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.053281612601217e-08,
+      "loss": 1.2289,
+      "step": 198955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.043138991417066e-08,
+      "loss": 1.2949,
+      "step": 198960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.033002029129711e-08,
+      "loss": 0.9955,
+      "step": 198965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.022870725796772e-08,
+      "loss": 1.0832,
+      "step": 198970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.012745081476315e-08,
+      "loss": 1.2032,
+      "step": 198975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.00262509622607e-08,
+      "loss": 1.3796,
+      "step": 198980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.992510770103768e-08,
+      "loss": 1.1484,
+      "step": 198985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.982402103167365e-08,
+      "loss": 1.2552,
+      "step": 198990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.97229909547448e-08,
+      "loss": 1.3139,
+      "step": 198995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.962201747082733e-08,
+      "loss": 1.2624,
+      "step": 199000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.952110058049856e-08,
+      "loss": 1.114,
+      "step": 199005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.942024028433472e-08,
+      "loss": 1.2996,
+      "step": 199010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.931943658291087e-08,
+      "loss": 1.2286,
+      "step": 199015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.921868947680546e-08,
+      "loss": 1.3138,
+      "step": 199020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.911799896659024e-08,
+      "loss": 1.3376,
+      "step": 199025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.901736505284253e-08,
+      "loss": 1.4009,
+      "step": 199030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.891678773613521e-08,
+      "loss": 1.2738,
+      "step": 199035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.88162670170456e-08,
+      "loss": 1.0985,
+      "step": 199040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.871580289614545e-08,
+      "loss": 1.1722,
+      "step": 199045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.861539537400766e-08,
+      "loss": 1.2288,
+      "step": 199050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.851504445120729e-08,
+      "loss": 1.2324,
+      "step": 199055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.841475012831724e-08,
+      "loss": 1.5039,
+      "step": 199060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.831451240590928e-08,
+      "loss": 1.4271,
+      "step": 199065
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.821433128455736e-08,
+      "loss": 1.0213,
+      "step": 199070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.81142067648333e-08,
+      "loss": 0.9964,
+      "step": 199075
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.801413884730659e-08,
+      "loss": 0.9442,
+      "step": 199080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.791412753255124e-08,
+      "loss": 1.4913,
+      "step": 199085
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.78141728211368e-08,
+      "loss": 1.2039,
+      "step": 199090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.771427471363503e-08,
+      "loss": 0.995,
+      "step": 199095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.761443321061658e-08,
+      "loss": 1.5397,
+      "step": 199100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.7514648312651e-08,
+      "loss": 1.2099,
+      "step": 199105
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.741492002030782e-08,
+      "loss": 1.0842,
+      "step": 199110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.731524833415661e-08,
+      "loss": 1.251,
+      "step": 199115
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.721563325476801e-08,
+      "loss": 1.3257,
+      "step": 199120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.711607478270933e-08,
+      "loss": 1.6197,
+      "step": 199125
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.701657291854904e-08,
+      "loss": 1.3378,
+      "step": 199130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.691712766285665e-08,
+      "loss": 1.36,
+      "step": 199135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.681773901619839e-08,
+      "loss": 1.2154,
+      "step": 199140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.671840697914269e-08,
+      "loss": 1.454,
+      "step": 199145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.661913155225688e-08,
+      "loss": 1.2112,
+      "step": 199150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.651991273610827e-08,
+      "loss": 1.8273,
+      "step": 199155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.642075053126308e-08,
+      "loss": 1.1855,
+      "step": 199160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.632164493828643e-08,
+      "loss": 1.2299,
+      "step": 199165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.622259595774562e-08,
+      "loss": 1.2001,
+      "step": 199170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.612360359020688e-08,
+      "loss": 1.2502,
+      "step": 199175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.60246678362342e-08,
+      "loss": 1.1296,
+      "step": 199180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.592578869639267e-08,
+      "loss": 1.2651,
+      "step": 199185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.582696617124742e-08,
+      "loss": 1.3363,
+      "step": 199190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.57282002613613e-08,
+      "loss": 1.2062,
+      "step": 199195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.562949096730056e-08,
+      "loss": 1.1118,
+      "step": 199200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.553083828962805e-08,
+      "loss": 1.0574,
+      "step": 199205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.543224222890667e-08,
+      "loss": 1.1758,
+      "step": 199210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.533370278570041e-08,
+      "loss": 0.9122,
+      "step": 199215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.523521996056994e-08,
+      "loss": 1.2846,
+      "step": 199220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.513679375408035e-08,
+      "loss": 1.3994,
+      "step": 199225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.503842416679231e-08,
+      "loss": 1.3558,
+      "step": 199230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.49401111992676e-08,
+      "loss": 1.3716,
+      "step": 199235
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.484185485206686e-08,
+      "loss": 0.9451,
+      "step": 199240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.474365512575189e-08,
+      "loss": 1.2095,
+      "step": 199245
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.464551202088445e-08,
+      "loss": 1.3186,
+      "step": 199250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.454742553802519e-08,
+      "loss": 1.0611,
+      "step": 199255
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.444939567773148e-08,
+      "loss": 1.3144,
+      "step": 199260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.435142244056504e-08,
+      "loss": 1.2186,
+      "step": 199265
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.425350582708547e-08,
+      "loss": 1.7741,
+      "step": 199270
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.415564583785119e-08,
+      "loss": 1.2452,
+      "step": 199275
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.405784247342174e-08,
+      "loss": 1.2976,
+      "step": 199280
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.39600957343556e-08,
+      "loss": 1.6048,
+      "step": 199285
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.386240562121006e-08,
+      "loss": 1.3648,
+      "step": 199290
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.376477213454359e-08,
+      "loss": 1.2121,
+      "step": 199295
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.366719527491462e-08,
+      "loss": 1.0296,
+      "step": 199300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.356967504287828e-08,
+      "loss": 1.4403,
+      "step": 199305
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.347221143899409e-08,
+      "loss": 1.4017,
+      "step": 199310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.337480446381607e-08,
+      "loss": 1.3069,
+      "step": 199315
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.327745411790266e-08,
+      "loss": 1.307,
+      "step": 199320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.318016040180787e-08,
+      "loss": 1.2126,
+      "step": 199325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.308292331608791e-08,
+      "loss": 1.0953,
+      "step": 199330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.298574286129901e-08,
+      "loss": 1.424,
+      "step": 199335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.288861903799405e-08,
+      "loss": 1.5143,
+      "step": 199340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.279155184673037e-08,
+      "loss": 1.3115,
+      "step": 199345
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.269454128806087e-08,
+      "loss": 1.2888,
+      "step": 199350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.259758736253953e-08,
+      "loss": 1.2013,
+      "step": 199355
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.250069007071926e-08,
+      "loss": 1.4032,
+      "step": 199360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.240384941315515e-08,
+      "loss": 1.1043,
+      "step": 199365
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.230706539039902e-08,
+      "loss": 1.152,
+      "step": 199370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.221033800300371e-08,
+      "loss": 1.127,
+      "step": 199375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.211366725152215e-08,
+      "loss": 1.0955,
+      "step": 199380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.201705313650388e-08,
+      "loss": 1.0307,
+      "step": 199385
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.192049565850512e-08,
+      "loss": 1.2056,
+      "step": 199390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.182399481807434e-08,
+      "loss": 1.0227,
+      "step": 199395
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.17275506157622e-08,
+      "loss": 1.0878,
+      "step": 199400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.163116305212049e-08,
+      "loss": 1.5892,
+      "step": 199405
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.153483212769986e-08,
+      "loss": 1.1164,
+      "step": 199410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.143855784304988e-08,
+      "loss": 1.6185,
+      "step": 199415
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.134234019872011e-08,
+      "loss": 1.032,
+      "step": 199420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.124617919526124e-08,
+      "loss": 1.1163,
+      "step": 199425
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.115007483322057e-08,
+      "loss": 1.5644,
+      "step": 199430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.105402711314769e-08,
+      "loss": 1.2501,
+      "step": 199435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.095803603559215e-08,
+      "loss": 1.1368,
+      "step": 199440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.08621016011002e-08,
+      "loss": 1.1901,
+      "step": 199445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.076622381022137e-08,
+      "loss": 1.2346,
+      "step": 199450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.067040266350191e-08,
+      "loss": 1.3028,
+      "step": 199455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.057463816149025e-08,
+      "loss": 0.9935,
+      "step": 199460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.047893030473264e-08,
+      "loss": 1.2886,
+      "step": 199465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.03832790937753e-08,
+      "loss": 1.0907,
+      "step": 199470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.028768452916557e-08,
+      "loss": 1.1325,
+      "step": 199475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.019214661144747e-08,
+      "loss": 1.2507,
+      "step": 199480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.009666534116722e-08,
+      "loss": 1.2972,
+      "step": 199485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.000124071887105e-08,
+      "loss": 1.2848,
+      "step": 199490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.990587274510409e-08,
+      "loss": 1.1938,
+      "step": 199495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.981056142040922e-08,
+      "loss": 1.1857,
+      "step": 199500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.971530674533045e-08,
+      "loss": 1.0214,
+      "step": 199505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.962010872041514e-08,
+      "loss": 1.3555,
+      "step": 199510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.952496734620286e-08,
+      "loss": 1.2621,
+      "step": 199515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.942988262324091e-08,
+      "loss": 1.3208,
+      "step": 199520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.93348545520678e-08,
+      "loss": 1.2418,
+      "step": 199525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.923988313322972e-08,
+      "loss": 1.389,
+      "step": 199530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.914496836726848e-08,
+      "loss": 1.4046,
+      "step": 199535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.905011025472475e-08,
+      "loss": 1.2428,
+      "step": 199540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.895530879614032e-08,
+      "loss": 1.6407,
+      "step": 199545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.886056399205699e-08,
+      "loss": 1.1771,
+      "step": 199550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.876587584301764e-08,
+      "loss": 1.0211,
+      "step": 199555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.867124434956075e-08,
+      "loss": 1.2452,
+      "step": 199560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.857666951222587e-08,
+      "loss": 1.2121,
+      "step": 199565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.848215133155479e-08,
+      "loss": 1.011,
+      "step": 199570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.83876898080882e-08,
+      "loss": 1.1678,
+      "step": 199575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.829328494236343e-08,
+      "loss": 1.0796,
+      "step": 199580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.819893673492118e-08,
+      "loss": 1.1389,
+      "step": 199585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.810464518629879e-08,
+      "loss": 1.4819,
+      "step": 199590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.801041029703471e-08,
+      "loss": 1.0849,
+      "step": 199595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.791623206766851e-08,
+      "loss": 1.4041,
+      "step": 199600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.782211049873756e-08,
+      "loss": 1.0191,
+      "step": 199605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.772804559077807e-08,
+      "loss": 1.1849,
+      "step": 199610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.763403734432962e-08,
+      "loss": 1.0051,
+      "step": 199615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.754008575992622e-08,
+      "loss": 1.4945,
+      "step": 199620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.744619083810634e-08,
+      "loss": 1.4144,
+      "step": 199625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.73523525794051e-08,
+      "loss": 1.0555,
+      "step": 199630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.725857098435985e-08,
+      "loss": 1.1712,
+      "step": 199635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.716484605350461e-08,
+      "loss": 1.1254,
+      "step": 199640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.70711777873745e-08,
+      "loss": 0.9125,
+      "step": 199645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.697756618650575e-08,
+      "loss": 0.9784,
+      "step": 199650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.688401125143242e-08,
+      "loss": 1.0826,
+      "step": 199655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.679051298268847e-08,
+      "loss": 0.9575,
+      "step": 199660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.669707138080684e-08,
+      "loss": 1.3564,
+      "step": 199665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.660368644632376e-08,
+      "loss": 1.237,
+      "step": 199670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.651035817976992e-08,
+      "loss": 1.305,
+      "step": 199675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.641708658167935e-08,
+      "loss": 1.2022,
+      "step": 199680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.632387165258492e-08,
+      "loss": 1.3545,
+      "step": 199685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.623071339301846e-08,
+      "loss": 1.1445,
+      "step": 199690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.613761180351175e-08,
+      "loss": 1.4787,
+      "step": 199695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.604456688459771e-08,
+      "loss": 1.3386,
+      "step": 199700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.595157863680703e-08,
+      "loss": 1.4398,
+      "step": 199705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.585864706066926e-08,
+      "loss": 1.559,
+      "step": 199710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.576577215671732e-08,
+      "loss": 1.2922,
+      "step": 199715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.567295392548079e-08,
+      "loss": 1.4259,
+      "step": 199720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.558019236748814e-08,
+      "loss": 1.1702,
+      "step": 199725
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.548748748327228e-08,
+      "loss": 1.2703,
+      "step": 199730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.539483927335944e-08,
+      "loss": 1.1135,
+      "step": 199735
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.530224773828033e-08,
+      "loss": 1.3041,
+      "step": 199740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.520971287856338e-08,
+      "loss": 1.3099,
+      "step": 199745
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.511723469473708e-08,
+      "loss": 1.4727,
+      "step": 199750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.502481318732991e-08,
+      "loss": 1.5976,
+      "step": 199755
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.493244835686808e-08,
+      "loss": 1.1494,
+      "step": 199760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.484014020388119e-08,
+      "loss": 1.1242,
+      "step": 199765
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.47478887288955e-08,
+      "loss": 1.418,
+      "step": 199770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.465569393243722e-08,
+      "loss": 1.1653,
+      "step": 199775
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.456355581503372e-08,
+      "loss": 1.3588,
+      "step": 199780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.447147437721014e-08,
+      "loss": 1.4781,
+      "step": 199785
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.437944961949383e-08,
+      "loss": 1.3347,
+      "step": 199790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.428748154240883e-08,
+      "loss": 1.3566,
+      "step": 199795
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.419557014648138e-08,
+      "loss": 1.1896,
+      "step": 199800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.410371543223439e-08,
+      "loss": 1.2317,
+      "step": 199805
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.40119174001952e-08,
+      "loss": 1.2647,
+      "step": 199810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.392017605088564e-08,
+      "loss": 1.2342,
+      "step": 199815
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.382849138483195e-08,
+      "loss": 1.2383,
+      "step": 199820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.373686340255593e-08,
+      "loss": 1.2456,
+      "step": 199825
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.36452921045805e-08,
+      "loss": 1.3127,
+      "step": 199830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.355377749142967e-08,
+      "loss": 1.233,
+      "step": 199835
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.346231956362637e-08,
+      "loss": 1.3221,
+      "step": 199840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.33709183216913e-08,
+      "loss": 1.3148,
+      "step": 199845
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.327957376614847e-08,
+      "loss": 1.0952,
+      "step": 199850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.318828589751747e-08,
+      "loss": 1.0587,
+      "step": 199855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.309705471632122e-08,
+      "loss": 1.1154,
+      "step": 199860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.300588022307931e-08,
+      "loss": 1.3949,
+      "step": 199865
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.291476241831352e-08,
+      "loss": 1.2351,
+      "step": 199870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.282370130254346e-08,
+      "loss": 1.3096,
+      "step": 199875
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.273269687628981e-08,
+      "loss": 1.1029,
+      "step": 199880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.264174914007104e-08,
+      "loss": 1.4766,
+      "step": 199885
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.255085809440787e-08,
+      "loss": 1.2674,
+      "step": 199890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.246002373981986e-08,
+      "loss": 1.1747,
+      "step": 199895
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.236924607682328e-08,
+      "loss": 1.2835,
+      "step": 199900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.227852510593769e-08,
+      "loss": 1.3027,
+      "step": 199905
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.21878608276827e-08,
+      "loss": 1.3174,
+      "step": 199910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.209725324257345e-08,
+      "loss": 1.3203,
+      "step": 199915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.20067023511295e-08,
+      "loss": 1.2485,
+      "step": 199920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.191620815386602e-08,
+      "loss": 1.1234,
+      "step": 199925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.182577065130258e-08,
+      "loss": 1.2601,
+      "step": 199930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.173538984395212e-08,
+      "loss": 1.3632,
+      "step": 199935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.16450657323331e-08,
+      "loss": 1.1055,
+      "step": 199940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.155479831696066e-08,
+      "loss": 1.1817,
+      "step": 199945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.146458759835107e-08,
+      "loss": 1.0344,
+      "step": 199950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.137443357701835e-08,
+      "loss": 1.0905,
+      "step": 199955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.128433625347763e-08,
+      "loss": 1.0548,
+      "step": 199960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.119429562824298e-08,
+      "loss": 1.2883,
+      "step": 199965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.110431170182952e-08,
+      "loss": 1.3669,
+      "step": 199970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.101438447475128e-08,
+      "loss": 1.1046,
+      "step": 199975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.09245139475212e-08,
+      "loss": 1.1306,
+      "step": 199980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.083470012065219e-08,
+      "loss": 1.1421,
+      "step": 199985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.074494299465829e-08,
+      "loss": 1.1921,
+      "step": 199990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.065524257005019e-08,
+      "loss": 1.2693,
+      "step": 199995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.056559884734083e-08,
+      "loss": 1.1807,
+      "step": 200000
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 1.2209757566452026,
+      "eval_runtime": 6466.1972,
+      "eval_samples_per_second": 3.574,
+      "eval_steps_per_second": 1.787,
+      "step": 200000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.047601182704312e-08,
+      "loss": 1.2831,
+      "step": 200005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.038648150966887e-08,
+      "loss": 1.3454,
+      "step": 200010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.029700789572658e-08,
+      "loss": 1.2234,
+      "step": 200015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.020759098573026e-08,
+      "loss": 1.256,
+      "step": 200020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.011823078018953e-08,
+      "loss": 1.4308,
+      "step": 200025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.002892727961396e-08,
+      "loss": 1.1336,
+      "step": 200030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.993968048451427e-08,
+      "loss": 1.2462,
+      "step": 200035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.985049039539893e-08,
+      "loss": 1.28,
+      "step": 200040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.976135701277975e-08,
+      "loss": 1.257,
+      "step": 200045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.967228033716411e-08,
+      "loss": 1.2234,
+      "step": 200050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.958326036906049e-08,
+      "loss": 1.394,
+      "step": 200055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.949429710897737e-08,
+      "loss": 0.9321,
+      "step": 200060
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.940539055742213e-08,
+      "loss": 1.635,
+      "step": 200065
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.931654071490434e-08,
+      "loss": 1.2952,
+      "step": 200070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.92277475819303e-08,
+      "loss": 1.0814,
+      "step": 200075
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.913901115900734e-08,
+      "loss": 1.0379,
+      "step": 200080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.905033144664175e-08,
+      "loss": 1.046,
+      "step": 200085
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.896170844533978e-08,
+      "loss": 1.3149,
+      "step": 200090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.887314215560769e-08,
+      "loss": 1.109,
+      "step": 200095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.878463257795176e-08,
+      "loss": 1.1823,
+      "step": 200100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.869617971287822e-08,
+      "loss": 1.6766,
+      "step": 200105
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.860778356089005e-08,
+      "loss": 1.3666,
+      "step": 200110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.851944412249234e-08,
+      "loss": 1.1455,
+      "step": 200115
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.84311613981914e-08,
+      "loss": 1.3546,
+      "step": 200120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.834293538849013e-08,
+      "loss": 1.2632,
+      "step": 200125
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.825476609389148e-08,
+      "loss": 1.2815,
+      "step": 200130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.81666535149006e-08,
+      "loss": 1.3043,
+      "step": 200135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.807859765202041e-08,
+      "loss": 1.2397,
+      "step": 200140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.799059850575273e-08,
+      "loss": 1.1884,
+      "step": 200145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.790265607660163e-08,
+      "loss": 1.2207,
+      "step": 200150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.781477036506668e-08,
+      "loss": 1.4216,
+      "step": 200155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.772694137165192e-08,
+      "loss": 1.2127,
+      "step": 200160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.763916909685919e-08,
+      "loss": 1.2274,
+      "step": 200165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.75514535411892e-08,
+      "loss": 1.5233,
+      "step": 200170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.746379470514153e-08,
+      "loss": 1.1487,
+      "step": 200175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.737619258921912e-08,
+      "loss": 1.2422,
+      "step": 200180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.728864719391937e-08,
+      "loss": 1.5794,
+      "step": 200185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.720115851974518e-08,
+      "loss": 1.0141,
+      "step": 200190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.711372656719395e-08,
+      "loss": 1.0607,
+      "step": 200195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.702635133676639e-08,
+      "loss": 1.2486,
+      "step": 200200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.693903282895986e-08,
+      "loss": 0.9635,
+      "step": 200205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.68517710442751e-08,
+      "loss": 0.9866,
+      "step": 200210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.676456598320836e-08,
+      "loss": 1.2226,
+      "step": 200215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.667741764625923e-08,
+      "loss": 1.0678,
+      "step": 200220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.659032603392401e-08,
+      "loss": 1.1545,
+      "step": 200225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.650329114670118e-08,
+      "loss": 1.1667,
+      "step": 200230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.6416312985087e-08,
+      "loss": 1.4932,
+      "step": 200235
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.632939154957885e-08,
+      "loss": 2.1665,
+      "step": 200240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.624252684067301e-08,
+      "loss": 1.0012,
+      "step": 200245
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.615571885886463e-08,
+      "loss": 1.0562,
+      "step": 200250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.606896760464998e-08,
+      "loss": 1.5289,
+      "step": 200255
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.598227307852534e-08,
+      "loss": 1.0022,
+      "step": 200260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.589563528098364e-08,
+      "loss": 1.0451,
+      "step": 200265
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.580905421252226e-08,
+      "loss": 1.3588,
+      "step": 200270
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.572252987363304e-08,
+      "loss": 1.1187,
+      "step": 200275
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.563606226481223e-08,
+      "loss": 0.8701,
+      "step": 200280
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.554965138655278e-08,
+      "loss": 0.9917,
+      "step": 200285
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.546329723934764e-08,
+      "loss": 1.2332,
+      "step": 200290
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.537699982369084e-08,
+      "loss": 1.1055,
+      "step": 200295
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.529075914007421e-08,
+      "loss": 1.4061,
+      "step": 200300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.52045751889907e-08,
+      "loss": 1.3636,
+      "step": 200305
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.511844797093326e-08,
+      "loss": 1.6758,
+      "step": 200310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.503237748639258e-08,
+      "loss": 1.344,
+      "step": 200315
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.494636373586049e-08,
+      "loss": 1.4663,
+      "step": 200320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.486040671982774e-08,
+      "loss": 1.4612,
+      "step": 200325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.477450643878724e-08,
+      "loss": 1.1613,
+      "step": 200330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.46886628932264e-08,
+      "loss": 1.1294,
+      "step": 200335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.460287608363814e-08,
+      "loss": 1.2926,
+      "step": 200340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.451714601051096e-08,
+      "loss": 1.2536,
+      "step": 200345
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.443147267433448e-08,
+      "loss": 1.6525,
+      "step": 200350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.434585607559718e-08,
+      "loss": 1.0865,
+      "step": 200355
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.42602962147909e-08,
+      "loss": 1.3876,
+      "step": 200360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.417479309240082e-08,
+      "loss": 1.4776,
+      "step": 200365
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.408934670891653e-08,
+      "loss": 1.3944,
+      "step": 200370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.400395706482765e-08,
+      "loss": 1.1114,
+      "step": 200375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.391862416061934e-08,
+      "loss": 1.502,
+      "step": 200380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.383334799678009e-08,
+      "loss": 1.1971,
+      "step": 200385
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.37481285737962e-08,
+      "loss": 1.1804,
+      "step": 200390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.366296589215504e-08,
+      "loss": 1.1448,
+      "step": 200395
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.357785995234178e-08,
+      "loss": 1.3366,
+      "step": 200400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.349281075484492e-08,
+      "loss": 1.3664,
+      "step": 200405
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.340781830014742e-08,
+      "loss": 1.1087,
+      "step": 200410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.332288258873553e-08,
+      "loss": 1.4147,
+      "step": 200415
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.323800362109445e-08,
+      "loss": 1.3499,
+      "step": 200420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.315318139770932e-08,
+      "loss": 1.2042,
+      "step": 200425
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.306841591906421e-08,
+      "loss": 1.5211,
+      "step": 200430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.298370718564318e-08,
+      "loss": 1.1203,
+      "step": 200435
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.289905519792915e-08,
+      "loss": 1.3608,
+      "step": 200440
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.281445995640623e-08,
+      "loss": 1.4686,
+      "step": 200445
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.272992146155843e-08,
+      "loss": 1.3175,
+      "step": 200450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.26454397138665e-08,
+      "loss": 1.2308,
+      "step": 200455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.256101471381449e-08,
+      "loss": 1.1627,
+      "step": 200460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.247664646188312e-08,
+      "loss": 1.1994,
+      "step": 200465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.239233495855646e-08,
+      "loss": 1.1765,
+      "step": 200470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.230808020431412e-08,
+      "loss": 1.3654,
+      "step": 200475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.222388219963682e-08,
+      "loss": 1.417,
+      "step": 200480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.213974094500641e-08,
+      "loss": 1.2775,
+      "step": 200485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.205565644090362e-08,
+      "loss": 1.5482,
+      "step": 200490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.197162868780803e-08,
+      "loss": 1.5708,
+      "step": 200495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.188765768619932e-08,
+      "loss": 1.3402,
+      "step": 200500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.180374343655814e-08,
+      "loss": 1.4322,
+      "step": 200505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.171988593936195e-08,
+      "loss": 1.2297,
+      "step": 200510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.163608519509035e-08,
+      "loss": 1.0905,
+      "step": 200515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.155234120422293e-08,
+      "loss": 1.3362,
+      "step": 200520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.146865396723712e-08,
+      "loss": 1.0774,
+      "step": 200525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.138502348461028e-08,
+      "loss": 1.2286,
+      "step": 200530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.130144975682095e-08,
+      "loss": 1.197,
+      "step": 200535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.12179327843454e-08,
+      "loss": 1.3337,
+      "step": 200540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.113447256766103e-08,
+      "loss": 1.1859,
+      "step": 200545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.105106910724523e-08,
+      "loss": 1.0105,
+      "step": 200550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.096772240357207e-08,
+      "loss": 1.1885,
+      "step": 200555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.088443245712005e-08,
+      "loss": 1.5796,
+      "step": 200560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.080119926836325e-08,
+      "loss": 0.9431,
+      "step": 200565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.071802283777795e-08,
+      "loss": 1.0613,
+      "step": 200570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.06349031658382e-08,
+      "loss": 1.6048,
+      "step": 200575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.055184025301918e-08,
+      "loss": 1.3182,
+      "step": 200580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.046883409979498e-08,
+      "loss": 1.1993,
+      "step": 200585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.038588470663964e-08,
+      "loss": 1.2324,
+      "step": 200590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.030299207402723e-08,
+      "loss": 1.0981,
+      "step": 200595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.022015620243071e-08,
+      "loss": 1.1367,
+      "step": 200600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.013737709232304e-08,
+      "loss": 0.9115,
+      "step": 200605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.005465474417716e-08,
+      "loss": 1.3447,
+      "step": 200610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.997198915846602e-08,
+      "loss": 1.3001,
+      "step": 200615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.988938033566149e-08,
+      "loss": 1.4763,
+      "step": 200620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.980682827623319e-08,
+      "loss": 1.1725,
+      "step": 200625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.972433298065516e-08,
+      "loss": 1.247,
+      "step": 200630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.964189444939816e-08,
+      "loss": 1.0889,
+      "step": 200635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.955951268293181e-08,
+      "loss": 0.9618,
+      "step": 200640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.947718768172683e-08,
+      "loss": 1.233,
+      "step": 200645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9394919446253975e-08,
+      "loss": 1.619,
+      "step": 200650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9312707976982855e-08,
+      "loss": 1.6678,
+      "step": 200655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9230553274381986e-08,
+      "loss": 1.1803,
+      "step": 200660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9148455338922106e-08,
+      "loss": 1.4335,
+      "step": 200665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.906641417107173e-08,
+      "loss": 0.9693,
+      "step": 200670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8984429771297146e-08,
+      "loss": 1.4698,
+      "step": 200675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8902502140069095e-08,
+      "loss": 1.7819,
+      "step": 200680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.882063127785609e-08,
+      "loss": 1.5372,
+      "step": 200685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.87388171851222e-08,
+      "loss": 1.6802,
+      "step": 200690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.865705986233705e-08,
+      "loss": 1.1307,
+      "step": 200695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8575359309968046e-08,
+      "loss": 1.3971,
+      "step": 200700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.849371552848038e-08,
+      "loss": 0.8972,
+      "step": 200705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.841212851833922e-08,
+      "loss": 1.2601,
+      "step": 200710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8330598280014195e-08,
+      "loss": 1.0449,
+      "step": 200715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8249124813967164e-08,
+      "loss": 1.3058,
+      "step": 200720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.816770812066441e-08,
+      "loss": 1.1361,
+      "step": 200725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.808634820057224e-08,
+      "loss": 1.1931,
+      "step": 200730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.800504505415472e-08,
+      "loss": 1.1854,
+      "step": 200735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.792379868187481e-08,
+      "loss": 1.3081,
+      "step": 200740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7842609084196585e-08,
+      "loss": 1.4561,
+      "step": 200745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.776147626158635e-08,
+      "loss": 1.5782,
+      "step": 200750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7680400214503715e-08,
+      "loss": 1.2074,
+      "step": 200755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7599380943414994e-08,
+      "loss": 1.7547,
+      "step": 200760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.751841844878092e-08,
+      "loss": 1.3051,
+      "step": 200765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.743751273106335e-08,
+      "loss": 1.5599,
+      "step": 200770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.735666379072635e-08,
+      "loss": 1.0695,
+      "step": 200775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.727587162823067e-08,
+      "loss": 1.2473,
+      "step": 200780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7195136244037055e-08,
+      "loss": 1.0845,
+      "step": 200785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.711445763860624e-08,
+      "loss": 1.2975,
+      "step": 200790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7033835812400076e-08,
+      "loss": 1.3486,
+      "step": 200795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.695327076587931e-08,
+      "loss": 1.4314,
+      "step": 200800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.687276249950246e-08,
+      "loss": 1.3679,
+      "step": 200805
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6792311013730285e-08,
+      "loss": 1.1968,
+      "step": 200810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.67119163090224e-08,
+      "loss": 1.1346,
+      "step": 200815
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.663157838583622e-08,
+      "loss": 1.2058,
+      "step": 200820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.65512972446336e-08,
+      "loss": 1.5411,
+      "step": 200825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.647107288587084e-08,
+      "loss": 1.1533,
+      "step": 200830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.639090531000535e-08,
+      "loss": 1.2194,
+      "step": 200835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.631079451749566e-08,
+      "loss": 1.1091,
+      "step": 200840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6230740508801394e-08,
+      "loss": 1.1551,
+      "step": 200845
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.615074328437664e-08,
+      "loss": 1.0635,
+      "step": 200850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.607080284467992e-08,
+      "loss": 1.3612,
+      "step": 200855
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.599091919016641e-08,
+      "loss": 1.2229,
+      "step": 200860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.591109232129466e-08,
+      "loss": 1.471,
+      "step": 200865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.583132223851762e-08,
+      "loss": 1.1113,
+      "step": 200870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5751608942292704e-08,
+      "loss": 1.3262,
+      "step": 200875
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.56719524330751e-08,
+      "loss": 1.1658,
+      "step": 200880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.559235271131891e-08,
+      "loss": 1.2593,
+      "step": 200885
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.551280977747819e-08,
+      "loss": 1.2333,
+      "step": 200890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5433323632009264e-08,
+      "loss": 1.2959,
+      "step": 200895
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.535389427536397e-08,
+      "loss": 1.0105,
+      "step": 200900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.527452170799752e-08,
+      "loss": 1.5764,
+      "step": 200905
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.519520593036065e-08,
+      "loss": 1.2296,
+      "step": 200910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.511594694290967e-08,
+      "loss": 1.3802,
+      "step": 200915
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.503674474609533e-08,
+      "loss": 1.1879,
+      "step": 200920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4957599340369486e-08,
+      "loss": 1.2328,
+      "step": 200925
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.487851072618511e-08,
+      "loss": 1.3662,
+      "step": 200930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4799478903992955e-08,
+      "loss": 1.0597,
+      "step": 200935
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.472050387424488e-08,
+      "loss": 1.3244,
+      "step": 200940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4641585637392747e-08,
+      "loss": 1.2465,
+      "step": 200945
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.456272419388509e-08,
+      "loss": 1.3251,
+      "step": 200950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.448391954417376e-08,
+      "loss": 1.2052,
+      "step": 200955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4405171688709514e-08,
+      "loss": 1.0287,
+      "step": 200960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.432648062793977e-08,
+      "loss": 1.3824,
+      "step": 200965
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.424784636231639e-08,
+      "loss": 1.0351,
+      "step": 200970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.416926889228569e-08,
+      "loss": 1.1989,
+      "step": 200975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4090748218298407e-08,
+      "loss": 1.268,
+      "step": 200980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.401228434080197e-08,
+      "loss": 1.3229,
+      "step": 200985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.393387726024602e-08,
+      "loss": 1.3009,
+      "step": 200990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3855526977076854e-08,
+      "loss": 1.3287,
+      "step": 200995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.377723349174191e-08,
+      "loss": 1.2209,
+      "step": 201000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.369899680468749e-08,
+      "loss": 1.6261,
+      "step": 201005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3620816916363224e-08,
+      "loss": 1.0529,
+      "step": 201010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.354269382721211e-08,
+      "loss": 1.0751,
+      "step": 201015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3464627537683776e-08,
+      "loss": 1.0056,
+      "step": 201020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3386618048220094e-08,
+      "loss": 1.0814,
+      "step": 201025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3308665359269594e-08,
+      "loss": 1.1961,
+      "step": 201030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3230769471277476e-08,
+      "loss": 0.9491,
+      "step": 201035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.315293038468672e-08,
+      "loss": 1.2352,
+      "step": 201040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.307514809994252e-08,
+      "loss": 1.2038,
+      "step": 201045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.299742261748897e-08,
+      "loss": 1.2614,
+      "step": 201050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.291975393777127e-08,
+      "loss": 1.1659,
+      "step": 201055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2842142061230175e-08,
+      "loss": 1.5198,
+      "step": 201060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.276458698831199e-08,
+      "loss": 1.4102,
+      "step": 201065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.268708871945749e-08,
+      "loss": 1.1251,
+      "step": 201070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2609647255109644e-08,
+      "loss": 1.1128,
+      "step": 201075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2532262595711424e-08,
+      "loss": 1.7313,
+      "step": 201080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.245493474170471e-08,
+      "loss": 1.1608,
+      "step": 201085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.237766369353026e-08,
+      "loss": 1.2327,
+      "step": 201090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.230044945162882e-08,
+      "loss": 1.436,
+      "step": 201095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.222329201644338e-08,
+      "loss": 1.1269,
+      "step": 201100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2146191388412484e-08,
+      "loss": 1.3175,
+      "step": 201105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.206914756797799e-08,
+      "loss": 1.3389,
+      "step": 201110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.199216055557843e-08,
+      "loss": 1.1867,
+      "step": 201115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1915230351654575e-08,
+      "loss": 1.0458,
+      "step": 201120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1838356956644965e-08,
+      "loss": 1.3136,
+      "step": 201125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.176154037099035e-08,
+      "loss": 1.3251,
+      "step": 201130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.168478059512705e-08,
+      "loss": 1.0941,
+      "step": 201135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.160807762949471e-08,
+      "loss": 1.23,
+      "step": 201140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1531431474529657e-08,
+      "loss": 1.2442,
+      "step": 201145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.145484213067264e-08,
+      "loss": 1.153,
+      "step": 201150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.137830959835888e-08,
+      "loss": 1.0093,
+      "step": 201155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1301833878025785e-08,
+      "loss": 1.1654,
+      "step": 201160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.122541497010969e-08,
+      "loss": 1.3007,
+      "step": 201165
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.114905287504801e-08,
+      "loss": 0.9805,
+      "step": 201170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1072747593275964e-08,
+      "loss": 1.2005,
+      "step": 201175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0996499125229856e-08,
+      "loss": 1.2778,
+      "step": 201180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.092030747134491e-08,
+      "loss": 1.2062,
+      "step": 201185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.084417263205521e-08,
+      "loss": 1.2947,
+      "step": 201190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.076809460779708e-08,
+      "loss": 1.0523,
+      "step": 201195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.069207339900461e-08,
+      "loss": 1.0164,
+      "step": 201200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.061610900611191e-08,
+      "loss": 0.9461,
+      "step": 201205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.054020142955196e-08,
+      "loss": 1.1235,
+      "step": 201210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.046435066975886e-08,
+      "loss": 1.3529,
+      "step": 201215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.038855672716558e-08,
+      "loss": 1.3848,
+      "step": 201220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0312819602205135e-08,
+      "loss": 1.2865,
+      "step": 201225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.023713929531049e-08,
+      "loss": 1.0552,
+      "step": 201230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.016151580691242e-08,
+      "loss": 1.1034,
+      "step": 201235
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.008594913744391e-08,
+      "loss": 1.0487,
+      "step": 201240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.001043928733684e-08,
+      "loss": 1.3885,
+      "step": 201245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.993498625702087e-08,
+      "loss": 1.2927,
+      "step": 201250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.985959004692897e-08,
+      "loss": 2.0326,
+      "step": 201255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.97842506574886e-08,
+      "loss": 1.4525,
+      "step": 201260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9708968089133835e-08,
+      "loss": 1.1896,
+      "step": 201265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9633742342292125e-08,
+      "loss": 1.261,
+      "step": 201270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.955857341739312e-08,
+      "loss": 1.2889,
+      "step": 201275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.948346131486648e-08,
+      "loss": 0.9843,
+      "step": 201280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9408406035141854e-08,
+      "loss": 1.3883,
+      "step": 201285
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.933340757864668e-08,
+      "loss": 1.3879,
+      "step": 201290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.925846594580952e-08,
+      "loss": 0.9884,
+      "step": 201295
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.918358113706001e-08,
+      "loss": 1.1203,
+      "step": 201300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9108753152822264e-08,
+      "loss": 1.4061,
+      "step": 201305
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.903398199352705e-08,
+      "loss": 1.1585,
+      "step": 201310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8959267659599575e-08,
+      "loss": 1.2933,
+      "step": 201315
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.888461015146617e-08,
+      "loss": 1.0543,
+      "step": 201320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8810009469554274e-08,
+      "loss": 1.4566,
+      "step": 201325
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.873546561429021e-08,
+      "loss": 0.95,
+      "step": 201330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.866097858609808e-08,
+      "loss": 1.8121,
+      "step": 201335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.858654838540422e-08,
+      "loss": 1.3471,
+      "step": 201340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.851217501263383e-08,
+      "loss": 1.1143,
+      "step": 201345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.843785846821103e-08,
+      "loss": 1.5961,
+      "step": 201350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8363598752561026e-08,
+      "loss": 1.1974,
+      "step": 201355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8289395866107924e-08,
+      "loss": 1.2916,
+      "step": 201360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.821524980927361e-08,
+      "loss": 1.2569,
+      "step": 201365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.81411605824833e-08,
+      "loss": 1.5669,
+      "step": 201370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.806712818615999e-08,
+      "loss": 1.3458,
+      "step": 201375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.799315262072557e-08,
+      "loss": 1.4594,
+      "step": 201380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.791923388660302e-08,
+      "loss": 1.4594,
+      "step": 201385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.784537198421424e-08,
+      "loss": 1.3169,
+      "step": 201390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.777156691398221e-08,
+      "loss": 1.4565,
+      "step": 201395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7697818676326615e-08,
+      "loss": 0.9622,
+      "step": 201400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.762412727166821e-08,
+      "loss": 1.0992,
+      "step": 201405
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.755049270043111e-08,
+      "loss": 1.0317,
+      "step": 201410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.747691496303164e-08,
+      "loss": 1.1546,
+      "step": 201415
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7403394059892804e-08,
+      "loss": 1.3263,
+      "step": 201420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7329929991434265e-08,
+      "loss": 1.0042,
+      "step": 201425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7256522758074573e-08,
+      "loss": 1.3155,
+      "step": 201430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.718317236023229e-08,
+      "loss": 1.2086,
+      "step": 201435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.710987879832818e-08,
+      "loss": 1.0374,
+      "step": 201440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.703664207277969e-08,
+      "loss": 1.2458,
+      "step": 201445
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6963462184005384e-08,
+      "loss": 1.4103,
+      "step": 201450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.689033913242269e-08,
+      "loss": 1.0977,
+      "step": 201455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6817272918449063e-08,
+      "loss": 1.1981,
+      "step": 201460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6744263542503056e-08,
+      "loss": 1.0248,
+      "step": 201465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.667131100499989e-08,
+      "loss": 1.2315,
+      "step": 201470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6598415306357e-08,
+      "loss": 1.4352,
+      "step": 201475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.652557644699074e-08,
+      "loss": 1.1069,
+      "step": 201480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.645279442731743e-08,
+      "loss": 0.9825,
+      "step": 201485
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.638006924775229e-08,
+      "loss": 1.7713,
+      "step": 201490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.630740090870944e-08,
+      "loss": 1.4651,
+      "step": 201495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6234789410605216e-08,
+      "loss": 1.54,
+      "step": 201500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.616223475385484e-08,
+      "loss": 1.5422,
+      "step": 201505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.608973693887242e-08,
+      "loss": 1.0558,
+      "step": 201510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6017295966069855e-08,
+      "loss": 1.1587,
+      "step": 201515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.594491183586236e-08,
+      "loss": 1.1618,
+      "step": 201520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.587258454866406e-08,
+      "loss": 1.3939,
+      "step": 201525
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.580031410488572e-08,
+      "loss": 1.3658,
+      "step": 201530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.572810050494259e-08,
+      "loss": 1.3864,
+      "step": 201535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.565594374924542e-08,
+      "loss": 1.0973,
+      "step": 201540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.558384383820724e-08,
+      "loss": 0.9532,
+      "step": 201545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.551180077223993e-08,
+      "loss": 0.9883,
+      "step": 201550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.543981455175317e-08,
+      "loss": 1.3166,
+      "step": 201555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.536788517715995e-08,
+      "loss": 1.2781,
+      "step": 201560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.529601264886885e-08,
+      "loss": 1.142,
+      "step": 201565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.522419696729286e-08,
+      "loss": 1.2698,
+      "step": 201570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5152438132841646e-08,
+      "loss": 1.1303,
+      "step": 201575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5080736145922674e-08,
+      "loss": 1.3124,
+      "step": 201580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5009091006947836e-08,
+      "loss": 1.5552,
+      "step": 201585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4937502716324575e-08,
+      "loss": 1.0241,
+      "step": 201590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.486597127446368e-08,
+      "loss": 1.222,
+      "step": 201595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4794496681771495e-08,
+      "loss": 1.5548,
+      "step": 201600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.472307893865768e-08,
+      "loss": 1.2121,
+      "step": 201605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.46517180455297e-08,
+      "loss": 1.388,
+      "step": 201610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4580414002795e-08,
+      "loss": 1.355,
+      "step": 201615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.450916681085993e-08,
+      "loss": 1.2768,
+      "step": 201620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.443797647013193e-08,
+      "loss": 0.9556,
+      "step": 201625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.436684298101956e-08,
+      "loss": 1.1467,
+      "step": 201630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.429576634392474e-08,
+      "loss": 1.124,
+      "step": 201635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.422474655925713e-08,
+      "loss": 1.6171,
+      "step": 201640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.415378362742084e-08,
+      "loss": 1.4329,
+      "step": 201645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4082877548821124e-08,
+      "loss": 1.2122,
+      "step": 201650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.401202832386431e-08,
+      "loss": 1.1756,
+      "step": 201655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.394123595295119e-08,
+      "loss": 1.1441,
+      "step": 201660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.387050043649033e-08,
+      "loss": 1.7693,
+      "step": 201665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.379982177488251e-08,
+      "loss": 1.3042,
+      "step": 201670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.372919996853409e-08,
+      "loss": 0.947,
+      "step": 201675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.365863501784584e-08,
+      "loss": 1.1616,
+      "step": 201680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3588126923221893e-08,
+      "loss": 1.4244,
+      "step": 201685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3517675685065264e-08,
+      "loss": 1.3121,
+      "step": 201690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3447281303777844e-08,
+      "loss": 1.1876,
+      "step": 201695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3376943779761535e-08,
+      "loss": 1.0019,
+      "step": 201700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.330666311341825e-08,
+      "loss": 1.4094,
+      "step": 201705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3236439305148756e-08,
+      "loss": 1.1435,
+      "step": 201710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3166272355354976e-08,
+      "loss": 1.0404,
+      "step": 201715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.309616226443658e-08,
+      "loss": 1.152,
+      "step": 201720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.302610903279436e-08,
+      "loss": 1.2798,
+      "step": 201725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2956112660828e-08,
+      "loss": 1.1467,
+      "step": 201730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.288617314893939e-08,
+      "loss": 1.021,
+      "step": 201735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.281629049752489e-08,
+      "loss": 1.1956,
+      "step": 201740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2746464706985294e-08,
+      "loss": 1.2621,
+      "step": 201745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.267669577771805e-08,
+      "loss": 1.2654,
+      "step": 201750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.260698371012284e-08,
+      "loss": 0.9943,
+      "step": 201755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2537328504598244e-08,
+      "loss": 1.343,
+      "step": 201760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.24677301615406e-08,
+      "loss": 1.3522,
+      "step": 201765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2398188681347374e-08,
+      "loss": 1.1348,
+      "step": 201770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.232870406441713e-08,
+      "loss": 1.1776,
+      "step": 201775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.225927631114512e-08,
+      "loss": 1.0593,
+      "step": 201780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2189905421927693e-08,
+      "loss": 1.2185,
+      "step": 201785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.21205913971634e-08,
+      "loss": 1.2104,
+      "step": 201790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.205133423724528e-08,
+      "loss": 1.2228,
+      "step": 201795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.198213394257078e-08,
+      "loss": 1.0477,
+      "step": 201800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1912990513532924e-08,
+      "loss": 1.272,
+      "step": 201805
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1843903950529176e-08,
+      "loss": 1.2983,
+      "step": 201810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.177487425395255e-08,
+      "loss": 1.3685,
+      "step": 201815
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.170590142419717e-08,
+      "loss": 1.1724,
+      "step": 201820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.163698546165607e-08,
+      "loss": 1.5486,
+      "step": 201825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.156812636672558e-08,
+      "loss": 1.1282,
+      "step": 201830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.149932413979651e-08,
+      "loss": 1.2676,
+      "step": 201835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1430578781262996e-08,
+      "loss": 1.1,
+      "step": 201840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.136189029151694e-08,
+      "loss": 1.337,
+      "step": 201845
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1293258670950243e-08,
+      "loss": 1.6327,
+      "step": 201850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1224683919955934e-08,
+      "loss": 1.1455,
+      "step": 201855
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1156166038924805e-08,
+      "loss": 1.1644,
+      "step": 201860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.108770502824877e-08,
+      "loss": 0.9628,
+      "step": 201865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.101930088831751e-08,
+      "loss": 1.2071,
+      "step": 201870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.095095361952295e-08,
+      "loss": 1.4869,
+      "step": 201875
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.088266322225476e-08,
+      "loss": 1.4788,
+      "step": 201880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.081442969690375e-08,
+      "loss": 1.1354,
+      "step": 201885
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0746253043858484e-08,
+      "loss": 1.3042,
+      "step": 201890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.067813326350867e-08,
+      "loss": 1.121,
+      "step": 201895
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.061007035624287e-08,
+      "loss": 1.1292,
+      "step": 201900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.054206432245078e-08,
+      "loss": 1.1222,
+      "step": 201905
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0474115162520975e-08,
+      "loss": 1.295,
+      "step": 201910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.040622287683982e-08,
+      "loss": 1.1889,
+      "step": 201915
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.033838746579588e-08,
+      "loss": 1.3661,
+      "step": 201920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0270608929776636e-08,
+      "loss": 1.2407,
+      "step": 201925
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.020288726916954e-08,
+      "loss": 1.2905,
+      "step": 201930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.013522248435986e-08,
+      "loss": 1.382,
+      "step": 201935
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.006761457573616e-08,
+      "loss": 1.1356,
+      "step": 201940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0000063543682574e-08,
+      "loss": 1.2729,
+      "step": 201945
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.993256938858547e-08,
+      "loss": 1.1471,
+      "step": 201950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9865132110831206e-08,
+      "loss": 1.1853,
+      "step": 201955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.979775171080391e-08,
+      "loss": 1.1653,
+      "step": 201960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.973042818888773e-08,
+      "loss": 1.256,
+      "step": 201965
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9663161545469006e-08,
+      "loss": 1.1755,
+      "step": 201970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.959595178093079e-08,
+      "loss": 1.2952,
+      "step": 201975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9528798895657196e-08,
+      "loss": 1.2507,
+      "step": 201980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.946170289003015e-08,
+      "loss": 1.069,
+      "step": 201985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9394663764434905e-08,
+      "loss": 1.562,
+      "step": 201990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9327681519253366e-08,
+      "loss": 1.1336,
+      "step": 201995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.926075615486857e-08,
+      "loss": 1.3866,
+      "step": 202000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.919388767166132e-08,
+      "loss": 1.1187,
+      "step": 202005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9127076070014646e-08,
+      "loss": 1.0843,
+      "step": 202010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.906032135031046e-08,
+      "loss": 1.0422,
+      "step": 202015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.899362351292846e-08,
+      "loss": 1.7957,
+      "step": 202020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.892698255825167e-08,
+      "loss": 1.1125,
+      "step": 202025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.886039848665868e-08,
+      "loss": 1.1816,
+      "step": 202030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8793871298530295e-08,
+      "loss": 1.2205,
+      "step": 202035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.872740099424621e-08,
+      "loss": 1.7213,
+      "step": 202040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8660987574187234e-08,
+      "loss": 1.2349,
+      "step": 202045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.859463103873085e-08,
+      "loss": 1.2757,
+      "step": 202050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.852833138825784e-08,
+      "loss": 1.0238,
+      "step": 202055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.84620886231446e-08,
+      "loss": 1.1937,
+      "step": 202060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.839590274377192e-08,
+      "loss": 1.1324,
+      "step": 202065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.832977375051617e-08,
+      "loss": 1.7013,
+      "step": 202070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.826370164375481e-08,
+      "loss": 2.3499,
+      "step": 202075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.819768642386645e-08,
+      "loss": 1.1072,
+      "step": 202080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8131728091227446e-08,
+      "loss": 1.0925,
+      "step": 202085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.806582664621306e-08,
+      "loss": 1.2264,
+      "step": 202090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.799998208920186e-08,
+      "loss": 1.1616,
+      "step": 202095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.793419442056801e-08,
+      "loss": 1.093,
+      "step": 202100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.786846364068897e-08,
+      "loss": 1.2358,
+      "step": 202105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.78027897499389e-08,
+      "loss": 1.2205,
+      "step": 202110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.773717274869304e-08,
+      "loss": 1.0076,
+      "step": 202115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.767161263732555e-08,
+      "loss": 1.2742,
+      "step": 202120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.760610941621168e-08,
+      "loss": 1.3501,
+      "step": 202125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.754066308572557e-08,
+      "loss": 1.1911,
+      "step": 202130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.747527364624026e-08,
+      "loss": 1.4786,
+      "step": 202135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.74099410981299e-08,
+      "loss": 1.1603,
+      "step": 202140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7344665441767516e-08,
+      "loss": 1.0324,
+      "step": 202145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.727944667752392e-08,
+      "loss": 1.1853,
+      "step": 202150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.721428480577438e-08,
+      "loss": 0.9369,
+      "step": 202155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.71491798268897e-08,
+      "loss": 1.0937,
+      "step": 202160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7084131741241816e-08,
+      "loss": 1.4427,
+      "step": 202165
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.701914054920264e-08,
+      "loss": 1.5794,
+      "step": 202170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6954206251141875e-08,
+      "loss": 1.1742,
+      "step": 202175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.688932884743146e-08,
+      "loss": 1.5695,
+      "step": 202180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.682450833844109e-08,
+      "loss": 1.2451,
+      "step": 202185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.675974472454269e-08,
+      "loss": 1.1297,
+      "step": 202190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6695038006103743e-08,
+      "loss": 1.139,
+      "step": 202195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.663038818349618e-08,
+      "loss": 1.3627,
+      "step": 202200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.656579525708637e-08,
+      "loss": 1.3701,
+      "step": 202205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.650125922724623e-08,
+      "loss": 1.5837,
+      "step": 202210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.643678009434215e-08,
+      "loss": 1.1756,
+      "step": 202215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6372357858742716e-08,
+      "loss": 1.124,
+      "step": 202220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.630799252081651e-08,
+      "loss": 0.9761,
+      "step": 202225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.624368408092993e-08,
+      "loss": 1.1866,
+      "step": 202230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6179432539451555e-08,
+      "loss": 1.1879,
+      "step": 202235
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6115237896747755e-08,
+      "loss": 1.4173,
+      "step": 202240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.605110015318492e-08,
+      "loss": 1.136,
+      "step": 202245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.59870193091294e-08,
+      "loss": 1.66,
+      "step": 202250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5922995364947585e-08,
+      "loss": 1.1244,
+      "step": 202255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.585902832100363e-08,
+      "loss": 1.0888,
+      "step": 202260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.579511817766501e-08,
+      "loss": 1.5808,
+      "step": 202265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5731264935294775e-08,
+      "loss": 1.3757,
+      "step": 202270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.566746859425818e-08,
+      "loss": 1.2615,
+      "step": 202275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5603729154920495e-08,
+      "loss": 1.574,
+      "step": 202280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5540046617644766e-08,
+      "loss": 1.092,
+      "step": 202285
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5476420982794026e-08,
+      "loss": 1.1295,
+      "step": 202290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.541285225073243e-08,
+      "loss": 1.2267,
+      "step": 202295
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5349340421823033e-08,
+      "loss": 1.1767,
+      "step": 202300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.528588549642886e-08,
+      "loss": 1.0801,
+      "step": 202305
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.522248747491186e-08,
+      "loss": 1.276,
+      "step": 202310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.515914635763396e-08,
+      "loss": 1.2387,
+      "step": 202315
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5095862144957084e-08,
+      "loss": 1.0126,
+      "step": 202320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.503263483724206e-08,
+      "loss": 1.0215,
+      "step": 202325
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.496946443485083e-08,
+      "loss": 1.0632,
+      "step": 202330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4906350938143094e-08,
+      "loss": 1.4984,
+      "step": 202335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4843294347479687e-08,
+      "loss": 1.2541,
+      "step": 202340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4780294663221417e-08,
+      "loss": 1.3155,
+      "step": 202345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.471735188572689e-08,
+      "loss": 1.6023,
+      "step": 202350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4654466015356936e-08,
+      "loss": 1.2473,
+      "step": 202355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4591637052470154e-08,
+      "loss": 1.1036,
+      "step": 202360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4528864997424025e-08,
+      "loss": 1.0692,
+      "step": 202365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.446614985057828e-08,
+      "loss": 1.4909,
+      "step": 202370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4403491612291505e-08,
+      "loss": 1.1292,
+      "step": 202375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.434089028292009e-08,
+      "loss": 1.3271,
+      "step": 202380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.427834586282264e-08,
+      "loss": 1.1172,
+      "step": 202385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.421585835235441e-08,
+      "loss": 1.327,
+      "step": 202390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.415342775187624e-08,
+      "loss": 1.5389,
+      "step": 202395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.409105406174007e-08,
+      "loss": 1.1603,
+      "step": 202400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4028737282305605e-08,
+      "loss": 0.9143,
+      "step": 202405
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3966477413925895e-08,
+      "loss": 1.4177,
+      "step": 202410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390427445695843e-08,
+      "loss": 1.3373,
+      "step": 202415
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.384212841175849e-08,
+      "loss": 1.136,
+      "step": 202420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.378003927868023e-08,
+      "loss": 1.3276,
+      "step": 202425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.371800705807782e-08,
+      "loss": 1.2212,
+      "step": 202430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3656031750305405e-08,
+      "loss": 1.3866,
+      "step": 202435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.359411335571827e-08,
+      "loss": 1.0435,
+      "step": 202440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3532251874669464e-08,
+      "loss": 1.2981,
+      "step": 202445
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.347044730751203e-08,
+      "loss": 1.0045,
+      "step": 202450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.340869965459792e-08,
+      "loss": 1.1782,
+      "step": 202455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.334700891628129e-08,
+      "loss": 1.1551,
+      "step": 202460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3285375092914075e-08,
+      "loss": 1.2725,
+      "step": 202465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.322379818484711e-08,
+      "loss": 1.2619,
+      "step": 202470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.316227819243456e-08,
+      "loss": 1.4516,
+      "step": 202475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.310081511602392e-08,
+      "loss": 1.5772,
+      "step": 202480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.303940895596935e-08,
+      "loss": 1.1604,
+      "step": 202485
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.297805971262169e-08,
+      "loss": 1.2533,
+      "step": 202490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.291676738632843e-08,
+      "loss": 1.317,
+      "step": 202495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.285553197744151e-08,
+      "loss": 1.1439,
+      "step": 202500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.279435348631066e-08,
+      "loss": 1.4452,
+      "step": 202505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.273323191328448e-08,
+      "loss": 0.9558,
+      "step": 202510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2672167258713806e-08,
+      "loss": 1.5191,
+      "step": 202515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.261115952294503e-08,
+      "loss": 1.311,
+      "step": 202520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2550208706327855e-08,
+      "loss": 1.3709,
+      "step": 202525
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.248931480920869e-08,
+      "loss": 1.2077,
+      "step": 202530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2428477831938364e-08,
+      "loss": 1.4383,
+      "step": 202535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.236769777486215e-08,
+      "loss": 1.3101,
+      "step": 202540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2306974638326436e-08,
+      "loss": 1.0526,
+      "step": 202545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.224630842267984e-08,
+      "loss": 1.2709,
+      "step": 202550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2185699128267636e-08,
+      "loss": 1.0206,
+      "step": 202555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.212514675543732e-08,
+      "loss": 1.1041,
+      "step": 202560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.206465130453307e-08,
+      "loss": 1.2527,
+      "step": 202565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2004212775900154e-08,
+      "loss": 1.1764,
+      "step": 202570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.194383116988498e-08,
+      "loss": 1.2215,
+      "step": 202575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1883506486831694e-08,
+      "loss": 1.5424,
+      "step": 202580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1823238727085594e-08,
+      "loss": 1.1857,
+      "step": 202585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1763027890989725e-08,
+      "loss": 1.1045,
+      "step": 202590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.170287397888938e-08,
+      "loss": 1.2496,
+      "step": 202595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.164277699112539e-08,
+      "loss": 1.2691,
+      "step": 202600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1582736928044145e-08,
+      "loss": 1.1534,
+      "step": 202605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1522753789986484e-08,
+      "loss": 1.3669,
+      "step": 202610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.146282757729546e-08,
+      "loss": 1.1255,
+      "step": 202615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.140295829031303e-08,
+      "loss": 1.1855,
+      "step": 202620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.134314592938115e-08,
+      "loss": 1.0873,
+      "step": 202625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1283390494841745e-08,
+      "loss": 1.3405,
+      "step": 202630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1223691987036786e-08,
+      "loss": 1.4929,
+      "step": 202635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.116405040630488e-08,
+      "loss": 1.4174,
+      "step": 202640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.110446575298909e-08,
+      "loss": 1.1155,
+      "step": 202645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.104493802742803e-08,
+      "loss": 1.4098,
+      "step": 202650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.098546722996254e-08,
+      "loss": 1.0095,
+      "step": 202655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.092605336093235e-08,
+      "loss": 1.1613,
+      "step": 202660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0866696420674966e-08,
+      "loss": 1.1139,
+      "step": 202665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.080739640953123e-08,
+      "loss": 1.1731,
+      "step": 202670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.074815332783976e-08,
+      "loss": 1.3212,
+      "step": 202675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.068896717593806e-08,
+      "loss": 1.3596,
+      "step": 202680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.062983795416363e-08,
+      "loss": 1.0786,
+      "step": 202685
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.057076566285622e-08,
+      "loss": 1.5724,
+      "step": 202690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.051175030234999e-08,
+      "loss": 1.4901,
+      "step": 202695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0452791872985776e-08,
+      "loss": 1.2845,
+      "step": 202700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0393890375096654e-08,
+      "loss": 1.423,
+      "step": 202705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.033504580902124e-08,
+      "loss": 1.2544,
+      "step": 202710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.027625817509483e-08,
+      "loss": 1.2656,
+      "step": 202715
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.021752747365381e-08,
+      "loss": 1.1362,
+      "step": 202720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0158853705032355e-08,
+      "loss": 1.4553,
+      "step": 202725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.010023686956687e-08,
+      "loss": 1.91,
+      "step": 202730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0041676967590413e-08,
+      "loss": 1.2109,
+      "step": 202735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9983173999438285e-08,
+      "loss": 1.075,
+      "step": 202740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9924727965445765e-08,
+      "loss": 1.2245,
+      "step": 202745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.986633886594481e-08,
+      "loss": 1.1365,
+      "step": 202750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.98080067012696e-08,
+      "loss": 1.394,
+      "step": 202755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9749731471753195e-08,
+      "loss": 1.2724,
+      "step": 202760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.969151317772756e-08,
+      "loss": 1.2186,
+      "step": 202765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9633351819526867e-08,
+      "loss": 0.9377,
+      "step": 202770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9575247397481964e-08,
+      "loss": 1.0729,
+      "step": 202775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.951719991192481e-08,
+      "loss": 1.4007,
+      "step": 202780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.945920936318736e-08,
+      "loss": 1.5089,
+      "step": 202785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9401275751600456e-08,
+      "loss": 0.9161,
+      "step": 202790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.934339907749495e-08,
+      "loss": 1.3418,
+      "step": 202795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9285579341201686e-08,
+      "loss": 1.2244,
+      "step": 202800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.92278165430504e-08,
+      "loss": 1.1856,
+      "step": 202805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9170110683370834e-08,
+      "loss": 1.4654,
+      "step": 202810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9112461762492717e-08,
+      "loss": 1.0721,
+      "step": 202815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9054869780745786e-08,
+      "loss": 1.4273,
+      "step": 202820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.899733473845867e-08,
+      "loss": 1.1411,
+      "step": 202825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8939856635958885e-08,
+      "loss": 1.3182,
+      "step": 202830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8882435473575055e-08,
+      "loss": 1.2748,
+      "step": 202835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8825071251636916e-08,
+      "loss": 1.2443,
+      "step": 202840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8767763970469765e-08,
+      "loss": 1.4807,
+      "step": 202845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8710513630402225e-08,
+      "loss": 1.2419,
+      "step": 202850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8653320231760706e-08,
+      "loss": 1.0917,
+      "step": 202855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8596183774871612e-08,
+      "loss": 0.9567,
+      "step": 202860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8539104260061345e-08,
+      "loss": 1.031,
+      "step": 202865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8482081687655204e-08,
+      "loss": 1.1221,
+      "step": 202870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8425116057980708e-08,
+      "loss": 1.7137,
+      "step": 202875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8368207371362034e-08,
+      "loss": 1.3111,
+      "step": 202880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8311355628123372e-08,
+      "loss": 0.881,
+      "step": 202885
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8254560828591127e-08,
+      "loss": 1.1206,
+      "step": 202890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8197822973088372e-08,
+      "loss": 1.2682,
+      "step": 202895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8141142061939297e-08,
+      "loss": 1.1725,
+      "step": 202900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8084518095466974e-08,
+      "loss": 1.3432,
+      "step": 202905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8027951073996695e-08,
+      "loss": 1.0034,
+      "step": 202910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7971440997849318e-08,
+      "loss": 1.2592,
+      "step": 202915
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.791498786734903e-08,
+      "loss": 1.4553,
+      "step": 202920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7858591682817793e-08,
+      "loss": 1.2237,
+      "step": 202925
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.780225244457646e-08,
+      "loss": 1.219,
+      "step": 202930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.774597015294922e-08,
+      "loss": 1.5351,
+      "step": 202935
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7689744808255815e-08,
+      "loss": 1.273,
+      "step": 202940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.76335764108171e-08,
+      "loss": 0.8929,
+      "step": 202945
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7577464960953927e-08,
+      "loss": 0.9859,
+      "step": 202950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7521410458988262e-08,
+      "loss": 1.2858,
+      "step": 202955
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7465412905238742e-08,
+      "loss": 1.2485,
+      "step": 202960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7409472300025107e-08,
+      "loss": 1.1953,
+      "step": 202965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7353588643667105e-08,
+      "loss": 1.2203,
+      "step": 202970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.729776193648337e-08,
+      "loss": 1.198,
+      "step": 202975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7241992178794753e-08,
+      "loss": 1.4393,
+      "step": 202980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.718627937091767e-08,
+      "loss": 1.1374,
+      "step": 202985
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.713062351317075e-08,
+      "loss": 1.8586,
+      "step": 202990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7075024605871524e-08,
+      "loss": 1.2352,
+      "step": 202995
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7019482649337515e-08,
+      "loss": 1.2832,
+      "step": 203000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6963997643887353e-08,
+      "loss": 1.2282,
+      "step": 203005
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.690856958983634e-08,
+      "loss": 1.2701,
+      "step": 203010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6853198487500898e-08,
+      "loss": 1.3626,
+      "step": 203015
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6797884337198544e-08,
+      "loss": 1.3112,
+      "step": 203020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6742627139243472e-08,
+      "loss": 1.2967,
+      "step": 203025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6687426893953205e-08,
+      "loss": 1.3972,
+      "step": 203030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6632283601641938e-08,
+      "loss": 1.0301,
+      "step": 203035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6577197262623866e-08,
+      "loss": 1.163,
+      "step": 203040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6522167877214287e-08,
+      "loss": 1.1826,
+      "step": 203045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6467195445727402e-08,
+      "loss": 1.1091,
+      "step": 203050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6412279968477396e-08,
+      "loss": 1.1358,
+      "step": 203055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6357421445777354e-08,
+      "loss": 1.3502,
+      "step": 203060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.630261987794036e-08,
+      "loss": 1.1756,
+      "step": 203065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.624787526528061e-08,
+      "loss": 1.4205,
+      "step": 203070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.619318760810896e-08,
+      "loss": 1.1049,
+      "step": 203075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.613855690673961e-08,
+      "loss": 1.247,
+      "step": 203080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.608398316148342e-08,
+      "loss": 1.3369,
+      "step": 203085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6029466372652358e-08,
+      "loss": 1.2157,
+      "step": 203090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.597500654055618e-08,
+      "loss": 1.3365,
+      "step": 203095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5920603665509082e-08,
+      "loss": 1.2798,
+      "step": 203100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.586625774781859e-08,
+      "loss": 1.1446,
+      "step": 203105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.581196878779668e-08,
+      "loss": 1.2074,
+      "step": 203110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5757736785753108e-08,
+      "loss": 1.2719,
+      "step": 203115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.570356174199873e-08,
+      "loss": 1.1246,
+      "step": 203120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5649443656839967e-08,
+      "loss": 1.1057,
+      "step": 203125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5595382530588798e-08,
+      "loss": 1.349,
+      "step": 203130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5541378363551638e-08,
+      "loss": 1.1568,
+      "step": 203135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5487431156038246e-08,
+      "loss": 1.2481,
+      "step": 203140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5433540908357256e-08,
+      "loss": 1.2688,
+      "step": 203145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5379707620815096e-08,
+      "loss": 1.1865,
+      "step": 203150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5325931293719298e-08,
+      "loss": 1.223,
+      "step": 203155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.52722119273785e-08,
+      "loss": 0.9197,
+      "step": 203160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5218549522096902e-08,
+      "loss": 1.3798,
+      "step": 203165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5164944078183152e-08,
+      "loss": 1.0546,
+      "step": 203170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5111395595942556e-08,
+      "loss": 1.2625,
+      "step": 203175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5057904075681538e-08,
+      "loss": 1.4005,
+      "step": 203180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.500446951770319e-08,
+      "loss": 1.1066,
+      "step": 203185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.495109192231504e-08,
+      "loss": 1.8193,
+      "step": 203190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4897771289822403e-08,
+      "loss": 1.2814,
+      "step": 203195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4844507620527258e-08,
+      "loss": 1.0169,
+      "step": 203200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4791300914736026e-08,
+      "loss": 1.0681,
+      "step": 203205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.473815117275069e-08,
+      "loss": 1.0412,
+      "step": 203210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.468505839487656e-08,
+      "loss": 1.5011,
+      "step": 203215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4632022581415616e-08,
+      "loss": 1.4262,
+      "step": 203220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.457904373267206e-08,
+      "loss": 1.3295,
+      "step": 203225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.452612184894565e-08,
+      "loss": 1.7029,
+      "step": 203230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.44732569305417e-08,
+      "loss": 1.3531,
+      "step": 203235
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.442044897776108e-08,
+      "loss": 1.2236,
+      "step": 203240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.436769799090466e-08,
+      "loss": 1.4149,
+      "step": 203245
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.431500397027442e-08,
+      "loss": 1.8285,
+      "step": 203250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.426236691616901e-08,
+      "loss": 1.1919,
+      "step": 203255
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4209786828892633e-08,
+      "loss": 1.5541,
+      "step": 203260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.415726370874394e-08,
+      "loss": 1.3819,
+      "step": 203265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4104797556022686e-08,
+      "loss": 1.1558,
+      "step": 203270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4052388371028635e-08,
+      "loss": 1.3244,
+      "step": 203275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4000036154060435e-08,
+      "loss": 0.9111,
+      "step": 203280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.394774090541785e-08,
+      "loss": 1.1897,
+      "step": 203285
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3895502625400633e-08,
+      "loss": 1.1002,
+      "step": 203290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3843321314304112e-08,
+      "loss": 1.0668,
+      "step": 203295
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3791196972429155e-08,
+      "loss": 1.3582,
+      "step": 203300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3739129600071075e-08,
+      "loss": 1.0647,
+      "step": 203305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.368711919752853e-08,
+      "loss": 1.162,
+      "step": 203310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3635165765099056e-08,
+      "loss": 1.5059,
+      "step": 203315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.358326930307797e-08,
+      "loss": 1.0687,
+      "step": 203320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3531429811761706e-08,
+      "loss": 1.7282,
+      "step": 203325
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3479647291447804e-08,
+      "loss": 1.2061,
+      "step": 203330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.342792174243047e-08,
+      "loss": 1.3852,
+      "step": 203335
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3376253165006136e-08,
+      "loss": 1.3808,
+      "step": 203340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3324641559469007e-08,
+      "loss": 1.0833,
+      "step": 203345
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.327308692611441e-08,
+      "loss": 1.2142,
+      "step": 203350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3221589265236544e-08,
+      "loss": 1.5789,
+      "step": 203355
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3170148577128515e-08,
+      "loss": 1.0035,
+      "step": 203360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.311876486208564e-08,
+      "loss": 1.2843,
+      "step": 203365
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.306743812040102e-08,
+      "loss": 1.0981,
+      "step": 203370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.301616835236775e-08,
+      "loss": 1.1362,
+      "step": 203375
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2964955558276712e-08,
+      "loss": 1.6689,
+      "step": 203380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2913799738423224e-08,
+      "loss": 1.4734,
+      "step": 203385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2862700893098167e-08,
+      "loss": 0.9589,
+      "step": 203390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2811659022593525e-08,
+      "loss": 1.1699,
+      "step": 203395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2760674127199067e-08,
+      "loss": 1.2455,
+      "step": 203400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2709746207209006e-08,
+      "loss": 1.4663,
+      "step": 203405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2658875262911994e-08,
+      "loss": 1.3241,
+      "step": 203410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2608061294598915e-08,
+      "loss": 1.2635,
+      "step": 203415
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2557304302560644e-08,
+      "loss": 1.2415,
+      "step": 203420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2506604287086952e-08,
+      "loss": 1.3972,
+      "step": 203425
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2455961248465386e-08,
+      "loss": 1.2947,
+      "step": 203430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.240537518698793e-08,
+      "loss": 1.2359,
+      "step": 203435
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2354846102942142e-08,
+      "loss": 1.0728,
+      "step": 203440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.230437399661667e-08,
+      "loss": 1.1354,
+      "step": 203445
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2253958868299064e-08,
+      "loss": 1.1108,
+      "step": 203450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2203600718279095e-08,
+      "loss": 1.5896,
+      "step": 203455
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2153299546842088e-08,
+      "loss": 1.184,
+      "step": 203460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2103055354277814e-08,
+      "loss": 1.3709,
+      "step": 203465
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2052868140870486e-08,
+      "loss": 1.4265,
+      "step": 203470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2002737906907656e-08,
+      "loss": 1.2922,
+      "step": 203475
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.195266465267687e-08,
+      "loss": 1.2328,
+      "step": 203480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1902648378462342e-08,
+      "loss": 0.9632,
+      "step": 203485
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1852689084550515e-08,
+      "loss": 1.152,
+      "step": 203490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1802786771226713e-08,
+      "loss": 1.264,
+      "step": 203495
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1752941438776266e-08,
+      "loss": 1.0847,
+      "step": 203500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1703153087483388e-08,
+      "loss": 1.1301,
+      "step": 203505
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1653421717631185e-08,
+      "loss": 1.1384,
+      "step": 203510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1603747329504986e-08,
+      "loss": 1.1779,
+      "step": 203515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.155412992338901e-08,
+      "loss": 1.098,
+      "step": 203520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1504569499566362e-08,
+      "loss": 1.3636,
+      "step": 203525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.145506605831904e-08,
+      "loss": 1.49,
+      "step": 203530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.140561959993015e-08,
+      "loss": 1.1051,
+      "step": 203535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1356230124681688e-08,
+      "loss": 1.3215,
+      "step": 203540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.130689763285676e-08,
+      "loss": 1.0973,
+      "step": 203545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1257622124736255e-08,
+      "loss": 1.1761,
+      "step": 203550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1208403600602168e-08,
+      "loss": 1.103,
+      "step": 203555
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1159242060735385e-08,
+      "loss": 1.2406,
+      "step": 203560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1110137505415685e-08,
+      "loss": 1.1528,
+      "step": 203565
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.106108993492506e-08,
+      "loss": 1.1353,
+      "step": 203570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.101209934954218e-08,
+      "loss": 1.1421,
+      "step": 203575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0963165749547932e-08,
+      "loss": 1.0055,
+      "step": 203580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.091428913522098e-08,
+      "loss": 1.2118,
+      "step": 203585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0865469506841096e-08,
+      "loss": 0.9955,
+      "step": 203590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0816706864685844e-08,
+      "loss": 1.1385,
+      "step": 203595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0768001209036103e-08,
+      "loss": 1.1041,
+      "step": 203600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0719352540167216e-08,
+      "loss": 1.2286,
+      "step": 203605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.067076085835784e-08,
+      "loss": 1.0312,
+      "step": 203610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0622226163886648e-08,
+      "loss": 1.3193,
+      "step": 203615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0573748457030085e-08,
+      "loss": 1.3293,
+      "step": 203620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0525327738063484e-08,
+      "loss": 1.242,
+      "step": 203625
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.047696400726662e-08,
+      "loss": 1.2463,
+      "step": 203630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0428657264912612e-08,
+      "loss": 1.1695,
+      "step": 203635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0380407511279012e-08,
+      "loss": 1.1821,
+      "step": 203640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0332214746640043e-08,
+      "loss": 1.0753,
+      "step": 203645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0284078971273268e-08,
+      "loss": 1.1417,
+      "step": 203650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0236000185450687e-08,
+      "loss": 1.0978,
+      "step": 203655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0187978389448747e-08,
+      "loss": 1.4276,
+      "step": 203660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0140013583540563e-08,
+      "loss": 1.0693,
+      "step": 203665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.009210576800036e-08,
+      "loss": 1.6026,
+      "step": 203670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.004425494310347e-08,
+      "loss": 1.0474,
+      "step": 203675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.999646110912079e-08,
+      "loss": 1.3916,
+      "step": 203680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9948724266325436e-08,
+      "loss": 1.4531,
+      "step": 203685
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.990104441499163e-08,
+      "loss": 1.3602,
+      "step": 203690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9853421555390273e-08,
+      "loss": 1.4197,
+      "step": 203695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9805855687793363e-08,
+      "loss": 1.0423,
+      "step": 203700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9758346812472906e-08,
+      "loss": 1.0617,
+      "step": 203705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9710894929700907e-08,
+      "loss": 1.0794,
+      "step": 203710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9663500039747153e-08,
+      "loss": 1.1037,
+      "step": 203715
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9616162142882534e-08,
+      "loss": 0.8551,
+      "step": 203720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9568881239377946e-08,
+      "loss": 1.0541,
+      "step": 203725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9521657329503173e-08,
+      "loss": 1.4609,
+      "step": 203730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9474490413528003e-08,
+      "loss": 0.8737,
+      "step": 203735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9427380491721103e-08,
+      "loss": 1.2147,
+      "step": 203740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9380327564352265e-08,
+      "loss": 1.4481,
+      "step": 203745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9333331631689046e-08,
+      "loss": 1.1265,
+      "step": 203750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9286392694001232e-08,
+      "loss": 1.0255,
+      "step": 203755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.923951075155639e-08,
+      "loss": 1.0939,
+      "step": 203760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.919268580462208e-08,
+      "loss": 1.131,
+      "step": 203765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9145917853464758e-08,
+      "loss": 1.0169,
+      "step": 203770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9099206898353096e-08,
+      "loss": 1.0226,
+      "step": 203775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9052552939553547e-08,
+      "loss": 1.0355,
+      "step": 203780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.900595597733257e-08,
+      "loss": 1.1723,
+      "step": 203785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.89594160119555e-08,
+      "loss": 1.1887,
+      "step": 203790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8912933043687687e-08,
+      "loss": 0.973,
+      "step": 203795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.886650707279558e-08,
+      "loss": 1.4733,
+      "step": 203800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8820138099545638e-08,
+      "loss": 1.0893,
+      "step": 203805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8773826124199866e-08,
+      "loss": 1.0851,
+      "step": 203810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8727571147023614e-08,
+      "loss": 1.4349,
+      "step": 203815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8681373168282223e-08,
+      "loss": 1.4218,
+      "step": 203820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8635232188237706e-08,
+      "loss": 1.2597,
+      "step": 203825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8589148207155406e-08,
+      "loss": 1.248,
+      "step": 203830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8543121225297332e-08,
+      "loss": 1.4819,
+      "step": 203835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8497151242926613e-08,
+      "loss": 1.1463,
+      "step": 203840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.845123826030637e-08,
+      "loss": 0.9479,
+      "step": 203845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8405382277696394e-08,
+      "loss": 1.2669,
+      "step": 203850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8359583295362026e-08,
+      "loss": 1.5684,
+      "step": 203855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8313841313563064e-08,
+      "loss": 1.4131,
+      "step": 203860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8268156332559295e-08,
+      "loss": 1.1271,
+      "step": 203865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8222528352613843e-08,
+      "loss": 1.1086,
+      "step": 203870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8176957373986504e-08,
+      "loss": 1.2057,
+      "step": 203875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8131443396937064e-08,
+      "loss": 0.9754,
+      "step": 203880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.808598642172643e-08,
+      "loss": 1.3451,
+      "step": 203885
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8040586448614394e-08,
+      "loss": 1.0844,
+      "step": 203890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7995243477857417e-08,
+      "loss": 1.2176,
+      "step": 203895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7949957509718618e-08,
+      "loss": 1.2262,
+      "step": 203900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7904728544453353e-08,
+      "loss": 1.4621,
+      "step": 203905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.78595565823203e-08,
+      "loss": 1.2908,
+      "step": 203910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7814441623579258e-08,
+      "loss": 1.2443,
+      "step": 203915
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7769383668485573e-08,
+      "loss": 1.4158,
+      "step": 203920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7724382717299038e-08,
+      "loss": 1.671,
+      "step": 203925
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7679438770273895e-08,
+      "loss": 1.2804,
+      "step": 203930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7634551827668822e-08,
+      "loss": 1.238,
+      "step": 203935
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7589721889739175e-08,
+      "loss": 1.4553,
+      "step": 203940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7544948956741415e-08,
+      "loss": 1.2151,
+      "step": 203945
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7500233028932e-08,
+      "loss": 1.1217,
+      "step": 203950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7455574106565176e-08,
+      "loss": 1.2575,
+      "step": 203955
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7410972189895182e-08,
+      "loss": 1.3357,
+      "step": 203960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7366427279177366e-08,
+      "loss": 1.3886,
+      "step": 203965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7321939374668194e-08,
+      "loss": 0.9325,
+      "step": 203970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7277508476618577e-08,
+      "loss": 1.0988,
+      "step": 203975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.723313458528386e-08,
+      "loss": 1.4039,
+      "step": 203980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7188817700917184e-08,
+      "loss": 1.0628,
+      "step": 203985
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.714455782377167e-08,
+      "loss": 1.0924,
+      "step": 203990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.710035495410045e-08,
+      "loss": 1.2816,
+      "step": 203995
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.705620909215555e-08,
+      "loss": 1.2759,
+      "step": 204000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.701212023818788e-08,
+      "loss": 1.2357,
+      "step": 204005
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6968088392451674e-08,
+      "loss": 1.6642,
+      "step": 204010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.692411355519674e-08,
+      "loss": 1.457,
+      "step": 204015
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6880195726673988e-08,
+      "loss": 1.2208,
+      "step": 204020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.683633490713543e-08,
+      "loss": 1.1561,
+      "step": 204025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6792531096830877e-08,
+      "loss": 1.1262,
+      "step": 204030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6748784296010123e-08,
+      "loss": 1.2536,
+      "step": 204035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.670509450492297e-08,
+      "loss": 1.5128,
+      "step": 204040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6661461723820326e-08,
+      "loss": 1.0239,
+      "step": 204045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6617885952949775e-08,
+      "loss": 1.1155,
+      "step": 204050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6574367192560005e-08,
+      "loss": 1.3675,
+      "step": 204055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6530905442900813e-08,
+      "loss": 1.2961,
+      "step": 204060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6487500704219782e-08,
+      "loss": 1.2058,
+      "step": 204065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6444152976764494e-08,
+      "loss": 1.0143,
+      "step": 204070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6400862260783635e-08,
+      "loss": 1.1525,
+      "step": 204075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6357628556522564e-08,
+      "loss": 1.1006,
+      "step": 204080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6314451864228865e-08,
+      "loss": 1.7698,
+      "step": 204085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.627133218415011e-08,
+      "loss": 1.2963,
+      "step": 204090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.622826951653278e-08,
+      "loss": 1.0209,
+      "step": 204095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6185263861620003e-08,
+      "loss": 1.3024,
+      "step": 204100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6142315219660476e-08,
+      "loss": 0.978,
+      "step": 204105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6099423590897334e-08,
+      "loss": 1.3581,
+      "step": 204110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6056588975577047e-08,
+      "loss": 1.3791,
+      "step": 204115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6013811373942755e-08,
+      "loss": 1.0521,
+      "step": 204120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5971090786239818e-08,
+      "loss": 1.3418,
+      "step": 204125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.592842721271248e-08,
+      "loss": 1.0855,
+      "step": 204130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.588582065360278e-08,
+      "loss": 1.3905,
+      "step": 204135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5843271109156066e-08,
+      "loss": 1.0264,
+      "step": 204140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5800778579613263e-08,
+      "loss": 1.0687,
+      "step": 204145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5758343065218617e-08,
+      "loss": 1.1343,
+      "step": 204150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571596456621416e-08,
+      "loss": 1.3135,
+      "step": 204155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5673643082840806e-08,
+      "loss": 1.2034,
+      "step": 204160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5631378615341697e-08,
+      "loss": 1.3026,
+      "step": 204165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.558917116395775e-08,
+      "loss": 1.1268,
+      "step": 204170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5547020728928774e-08,
+      "loss": 1.3249,
+      "step": 204175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5504927310496797e-08,
+      "loss": 1.3874,
+      "step": 204180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.546289090890274e-08,
+      "loss": 0.971,
+      "step": 204185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5420911524385297e-08,
+      "loss": 1.3098,
+      "step": 204190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5378989157184276e-08,
+      "loss": 1.3043,
+      "step": 204195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.533712380753949e-08,
+      "loss": 0.9595,
+      "step": 204200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5295315475690742e-08,
+      "loss": 1.0068,
+      "step": 204205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.525356416187562e-08,
+      "loss": 1.0116,
+      "step": 204210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.521186986633283e-08,
+      "loss": 1.0482,
+      "step": 204215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.517023258929995e-08,
+      "loss": 1.1693,
+      "step": 204220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.512865233101679e-08,
+      "loss": 1.0293,
+      "step": 204225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5087129091718722e-08,
+      "loss": 1.2301,
+      "step": 204230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.504566287164333e-08,
+      "loss": 1.3329,
+      "step": 204235
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.50042536710282e-08,
+      "loss": 1.3043,
+      "step": 204240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4962901490108706e-08,
+      "loss": 1.2412,
+      "step": 204245
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.492160632912132e-08,
+      "loss": 1.3268,
+      "step": 204250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4880368188302518e-08,
+      "loss": 1.4082,
+      "step": 204255
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4839187067887672e-08,
+      "loss": 1.2527,
+      "step": 204260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4798062968111037e-08,
+      "loss": 1.6728,
+      "step": 204265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4756995889207981e-08,
+      "loss": 1.4936,
+      "step": 204270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4715985831413871e-08,
+      "loss": 1.0191,
+      "step": 204275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4675032794960743e-08,
+      "loss": 1.3906,
+      "step": 204280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4634136780085074e-08,
+      "loss": 1.205,
+      "step": 204285
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.45932977870189e-08,
+      "loss": 1.4625,
+      "step": 204290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4552515815995371e-08,
+      "loss": 0.9563,
+      "step": 204295
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4511790867247632e-08,
+      "loss": 1.1337,
+      "step": 204300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4471122941007721e-08,
+      "loss": 1.0688,
+      "step": 204305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4430512037508781e-08,
+      "loss": 1.0628,
+      "step": 204310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4389958156982853e-08,
+      "loss": 1.1472,
+      "step": 204315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.434946129966086e-08,
+      "loss": 1.053,
+      "step": 204320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4309021465774841e-08,
+      "loss": 1.7179,
+      "step": 204325
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4268638655554612e-08,
+      "loss": 1.1078,
+      "step": 204330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4228312869231099e-08,
+      "loss": 1.2755,
+      "step": 204335
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4188044107035226e-08,
+      "loss": 1.4746,
+      "step": 204340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4147832369195701e-08,
+      "loss": 1.3384,
+      "step": 204345
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4107677655943452e-08,
+      "loss": 1.1062,
+      "step": 204350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4067579967507183e-08,
+      "loss": 1.1848,
+      "step": 204355
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4027539304116711e-08,
+      "loss": 1.4377,
+      "step": 204360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.398755566599852e-08,
+      "loss": 1.0729,
+      "step": 204365
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3947629053383538e-08,
+      "loss": 1.0211,
+      "step": 204370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.390775946649825e-08,
+      "loss": 1.3123,
+      "step": 204375
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3867946905571361e-08,
+      "loss": 1.1063,
+      "step": 204380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3828191370828248e-08,
+      "loss": 1.1118,
+      "step": 204385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3788492862498725e-08,
+      "loss": 1.343,
+      "step": 204390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.374885138080706e-08,
+      "loss": 1.1995,
+      "step": 204395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3709266925980846e-08,
+      "loss": 1.1659,
+      "step": 204400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.366973949824546e-08,
+      "loss": 1.0692,
+      "step": 204405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3630269097827387e-08,
+      "loss": 1.1168,
+      "step": 204410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3590855724952002e-08,
+      "loss": 1.079,
+      "step": 204415
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.355149937984357e-08,
+      "loss": 1.0265,
+      "step": 204420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3512200062727465e-08,
+      "loss": 1.0502,
+      "step": 204425
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3472957773827955e-08,
+      "loss": 1.3554,
+      "step": 204430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3433772513369303e-08,
+      "loss": 1.2404,
+      "step": 204435
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3394644281574665e-08,
+      "loss": 1.1833,
+      "step": 204440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3355573078668304e-08,
+      "loss": 0.9112,
+      "step": 204445
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3316558904872267e-08,
+      "loss": 1.0705,
+      "step": 204450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3277601760410819e-08,
+      "loss": 1.0603,
+      "step": 204455
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3238701645506002e-08,
+      "loss": 1.1861,
+      "step": 204460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3199858560378754e-08,
+      "loss": 1.1126,
+      "step": 204465
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3161072505252226e-08,
+      "loss": 1.5172,
+      "step": 204470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3122343480347355e-08,
+      "loss": 1.2222,
+      "step": 204475
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3083671485885074e-08,
+      "loss": 1.4305,
+      "step": 204480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.304505652208632e-08,
+      "loss": 1.5333,
+      "step": 204485
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3006498589172023e-08,
+      "loss": 1.0658,
+      "step": 204490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.296799768736201e-08,
+      "loss": 0.9774,
+      "step": 204495
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2929553816877217e-08,
+      "loss": 1.1236,
+      "step": 204500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2891166977935243e-08,
+      "loss": 1.3028,
+      "step": 204505
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2852837170757027e-08,
+      "loss": 1.0974,
+      "step": 204510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.281456439556017e-08,
+      "loss": 1.1362,
+      "step": 204515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2776348652563386e-08,
+      "loss": 1.3061,
+      "step": 204520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.27381899419865e-08,
+      "loss": 1.1468,
+      "step": 204525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2700088264044897e-08,
+      "loss": 1.3973,
+      "step": 204530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2662043618958397e-08,
+      "loss": 1.076,
+      "step": 204535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2624056006943497e-08,
+      "loss": 1.1307,
+      "step": 204540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2586125428216689e-08,
+      "loss": 1.8181,
+      "step": 204545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2548251882994467e-08,
+      "loss": 1.6839,
+      "step": 204550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2510435371493324e-08,
+      "loss": 1.177,
+      "step": 204555
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2472675893929753e-08,
+      "loss": 1.2376,
+      "step": 204560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.243497345051914e-08,
+      "loss": 1.1348,
+      "step": 204565
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2397328041476863e-08,
+      "loss": 1.2224,
+      "step": 204570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2359739667017201e-08,
+      "loss": 1.1912,
+      "step": 204575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2322208327355534e-08,
+      "loss": 1.4011,
+      "step": 204580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2284734022706134e-08,
+      "loss": 1.2268,
+      "step": 204585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2247316753282168e-08,
+      "loss": 1.0417,
+      "step": 204590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2209956519299016e-08,
+      "loss": 1.0841,
+      "step": 204595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.217265332096873e-08,
+      "loss": 1.5986,
+      "step": 204600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2135407158503364e-08,
+      "loss": 1.3967,
+      "step": 204605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2098218032118303e-08,
+      "loss": 1.2835,
+      "step": 204610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2061085942023376e-08,
+      "loss": 1.219,
+      "step": 204615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2024010888431748e-08,
+      "loss": 1.3853,
+      "step": 204620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.198699287155547e-08,
+      "loss": 1.3474,
+      "step": 204625
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1950031891605485e-08,
+      "loss": 1.2282,
+      "step": 204630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1913127948793846e-08,
+      "loss": 1.2477,
+      "step": 204635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1876281043329274e-08,
+      "loss": 1.1702,
+      "step": 204640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1839491175422712e-08,
+      "loss": 1.4268,
+      "step": 204645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1802758345286213e-08,
+      "loss": 1.2036,
+      "step": 204650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.176608255312739e-08,
+      "loss": 1.3653,
+      "step": 204655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1729463799156072e-08,
+      "loss": 1.2805,
+      "step": 204660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1692902083582092e-08,
+      "loss": 1.1524,
+      "step": 204665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1656397406614172e-08,
+      "loss": 1.0918,
+      "step": 204670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1619949768461036e-08,
+      "loss": 1.1282,
+      "step": 204675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1583559169329183e-08,
+      "loss": 1.0821,
+      "step": 204680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1547225609427338e-08,
+      "loss": 1.2531,
+      "step": 204685
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1510949088964218e-08,
+      "loss": 1.3226,
+      "step": 204690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.147472960814522e-08,
+      "loss": 1.4742,
+      "step": 204695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.143856716717795e-08,
+      "loss": 1.3509,
+      "step": 204700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1402461766268913e-08,
+      "loss": 1.0608,
+      "step": 204705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.13664134056235e-08,
+      "loss": 1.2736,
+      "step": 204710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1330422085449322e-08,
+      "loss": 1.1637,
+      "step": 204715
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.129448780595066e-08,
+      "loss": 1.0376,
+      "step": 204720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1258610567332906e-08,
+      "loss": 1.2043,
+      "step": 204725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.122279036980034e-08,
+      "loss": 1.0615,
+      "step": 204730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1187027213558355e-08,
+      "loss": 0.9753,
+      "step": 204735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.115132109881123e-08,
+      "loss": 1.6164,
+      "step": 204740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1115672025762136e-08,
+      "loss": 1.2649,
+      "step": 204745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1080079994615357e-08,
+      "loss": 1.4257,
+      "step": 204750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1044545005574059e-08,
+      "loss": 1.6353,
+      "step": 204755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1009067058841416e-08,
+      "loss": 1.2069,
+      "step": 204760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0973646154618378e-08,
+      "loss": 1.2714,
+      "step": 204765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0938282293109226e-08,
+      "loss": 0.9993,
+      "step": 204770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.090297547451602e-08,
+      "loss": 1.0355,
+      "step": 204775
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.086772569903749e-08,
+      "loss": 1.4914,
+      "step": 204780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0832532966877918e-08,
+      "loss": 1.4034,
+      "step": 204785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0797397278238253e-08,
+      "loss": 0.9522,
+      "step": 204790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0762318633317225e-08,
+      "loss": 1.0115,
+      "step": 204795
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0727297032315787e-08,
+      "loss": 1.0697,
+      "step": 204800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0692332475434886e-08,
+      "loss": 1.4517,
+      "step": 204805
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0657424962873253e-08,
+      "loss": 1.1319,
+      "step": 204810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0622574494829618e-08,
+      "loss": 1.174,
+      "step": 204815
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0587781071504931e-08,
+      "loss": 1.1205,
+      "step": 204820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0553044693096815e-08,
+      "loss": 1.159,
+      "step": 204825
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0518365359802884e-08,
+      "loss": 1.4591,
+      "step": 204830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0483743071822983e-08,
+      "loss": 1.0533,
+      "step": 204835
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0449177829353618e-08,
+      "loss": 1.301,
+      "step": 204840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0414669632592412e-08,
+      "loss": 0.9176,
+      "step": 204845
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0380218481735871e-08,
+      "loss": 1.601,
+      "step": 204850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0345824376981617e-08,
+      "loss": 1.0972,
+      "step": 204855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.031148731852505e-08,
+      "loss": 1.0733,
+      "step": 204860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0277207306562676e-08,
+      "loss": 1.4148,
+      "step": 204865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0242984341292118e-08,
+      "loss": 1.3112,
+      "step": 204870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0208818422905442e-08,
+      "loss": 1.3549,
+      "step": 204875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.017470955160027e-08,
+      "loss": 1.2977,
+      "step": 204880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.014065772756978e-08,
+      "loss": 1.1685,
+      "step": 204885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0106662951010482e-08,
+      "loss": 1.2491,
+      "step": 204890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0072725222114444e-08,
+      "loss": 1.1636,
+      "step": 204895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0038844541075954e-08,
+      "loss": 1.016,
+      "step": 204900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0005020908089303e-08,
+      "loss": 1.0297,
+      "step": 204905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.971254323346558e-09,
+      "loss": 1.1823,
+      "step": 204910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.937544787042009e-09,
+      "loss": 1.2611,
+      "step": 204915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.903892299366614e-09,
+      "loss": 1.1214,
+      "step": 204920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.870296860513551e-09,
+      "loss": 1.252,
+      "step": 204925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.83675847067378e-09,
+      "loss": 0.9902,
+      "step": 204930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.803277130039368e-09,
+      "loss": 1.1035,
+      "step": 204935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.769852838802384e-09,
+      "loss": 1.0044,
+      "step": 204940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.736485597151568e-09,
+      "loss": 1.2785,
+      "step": 204945
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.703175405280096e-09,
+      "loss": 1.5006,
+      "step": 204950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.669922263375597e-09,
+      "loss": 1.0272,
+      "step": 204955
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.63672617163014e-09,
+      "loss": 1.4471,
+      "step": 204960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.60358713023135e-09,
+      "loss": 1.1658,
+      "step": 204965
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.57050513937019e-09,
+      "loss": 1.1513,
+      "step": 204970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.537480199234284e-09,
+      "loss": 1.2547,
+      "step": 204975
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.504512310013481e-09,
+      "loss": 1.1506,
+      "step": 204980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.471601471894298e-09,
+      "loss": 1.0995,
+      "step": 204985
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.438747685066584e-09,
+      "loss": 1.3116,
+      "step": 204990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.405950949716858e-09,
+      "loss": 1.2461,
+      "step": 204995
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.373211266031635e-09,
+      "loss": 1.4545,
+      "step": 205000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.340528634199653e-09,
+      "loss": 1.0806,
+      "step": 205005
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.30790305440743e-09,
+      "loss": 1.3175,
+      "step": 205010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.275334526839264e-09,
+      "loss": 1.0071,
+      "step": 205015
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.242823051683891e-09,
+      "loss": 1.3443,
+      "step": 205020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.2103686291245e-09,
+      "loss": 1.2049,
+      "step": 205025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.177971259347607e-09,
+      "loss": 1.2583,
+      "step": 205030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.14563094253862e-09,
+      "loss": 1.2915,
+      "step": 205035
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.113347678880724e-09,
+      "loss": 1.1753,
+      "step": 205040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.08112146856044e-09,
+      "loss": 1.4041,
+      "step": 205045
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.04895231175984e-09,
+      "loss": 1.4552,
+      "step": 205050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.016840208663224e-09,
+      "loss": 1.1891,
+      "step": 205055
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.98478515945489e-09,
+      "loss": 1.0172,
+      "step": 205060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.952787164316912e-09,
+      "loss": 1.4582,
+      "step": 205065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.920846223431368e-09,
+      "loss": 1.3438,
+      "step": 205070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.888962336982554e-09,
+      "loss": 1.2518,
+      "step": 205075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.857135505151437e-09,
+      "loss": 0.9827,
+      "step": 205080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.825365728118985e-09,
+      "loss": 1.2681,
+      "step": 205085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.793653006068382e-09,
+      "loss": 1.3858,
+      "step": 205090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.761997339179484e-09,
+      "loss": 1.2471,
+      "step": 205095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.73039872763326e-09,
+      "loss": 0.9895,
+      "step": 205100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.698857171609565e-09,
+      "loss": 1.2373,
+      "step": 205105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.667372671290476e-09,
+      "loss": 1.4401,
+      "step": 205110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.635945226853626e-09,
+      "loss": 1.1224,
+      "step": 205115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.604574838479984e-09,
+      "loss": 1.2222,
+      "step": 205120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.573261506348296e-09,
+      "loss": 1.1807,
+      "step": 205125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.542005230637307e-09,
+      "loss": 1.1326,
+      "step": 205130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.510806011524653e-09,
+      "loss": 0.9936,
+      "step": 205135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.479663849189078e-09,
+      "loss": 1.0626,
+      "step": 205140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.448578743809333e-09,
+      "loss": 1.1465,
+      "step": 205145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.417550695561938e-09,
+      "loss": 1.1917,
+      "step": 205150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.386579704624531e-09,
+      "loss": 1.2879,
+      "step": 205155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.355665771173637e-09,
+      "loss": 1.6959,
+      "step": 205160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.324808895385783e-09,
+      "loss": 1.1333,
+      "step": 205165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.294009077437493e-09,
+      "loss": 1.266,
+      "step": 205170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.263266317504182e-09,
+      "loss": 1.356,
+      "step": 205175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.232580615762375e-09,
+      "loss": 1.2562,
+      "step": 205180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.20195197238527e-09,
+      "loss": 1.1239,
+      "step": 205185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.171380387550498e-09,
+      "loss": 1.1749,
+      "step": 205190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.140865861430146e-09,
+      "loss": 1.1747,
+      "step": 205195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.110408394200742e-09,
+      "loss": 1.2348,
+      "step": 205200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.080007986034367e-09,
+      "loss": 1.3078,
+      "step": 205205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.049664637105325e-09,
+      "loss": 1.4459,
+      "step": 205210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.019378347586814e-09,
+      "loss": 1.3085,
+      "step": 205215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.989149117652028e-09,
+      "loss": 1.1923,
+      "step": 205220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.958976947474161e-09,
+      "loss": 1.4099,
+      "step": 205225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.928861837224188e-09,
+      "loss": 1.2437,
+      "step": 205230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.898803787074193e-09,
+      "loss": 1.2425,
+      "step": 205235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.868802797197372e-09,
+      "loss": 1.1686,
+      "step": 205240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.838858867762478e-09,
+      "loss": 1.2143,
+      "step": 205245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.808971998942704e-09,
+      "loss": 1.5848,
+      "step": 205250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.779142190909027e-09,
+      "loss": 1.0915,
+      "step": 205255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.749369443829092e-09,
+      "loss": 1.1829,
+      "step": 205260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.719653757876088e-09,
+      "loss": 1.0846,
+      "step": 205265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.689995133217664e-09,
+      "loss": 1.6155,
+      "step": 205270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.660393570023683e-09,
+      "loss": 0.9205,
+      "step": 205275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.630849068462898e-09,
+      "loss": 1.3005,
+      "step": 205280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.601361628705173e-09,
+      "loss": 1.1643,
+      "step": 205285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.571931250917043e-09,
+      "loss": 1.1354,
+      "step": 205290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.542557935268369e-09,
+      "loss": 1.2215,
+      "step": 205295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.513241681925687e-09,
+      "loss": 1.3796,
+      "step": 205300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.483982491057751e-09,
+      "loss": 1.0026,
+      "step": 205305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.454780362829983e-09,
+      "loss": 1.1391,
+      "step": 205310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.425635297410027e-09,
+      "loss": 1.1092,
+      "step": 205315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.396547294964418e-09,
+      "loss": 1.2496,
+      "step": 205320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.367516355659687e-09,
+      "loss": 0.9982,
+      "step": 205325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.338542479660149e-09,
+      "loss": 1.1292,
+      "step": 205330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.309625667132336e-09,
+      "loss": 1.2868,
+      "step": 205335
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.280765918241672e-09,
+      "loss": 1.0861,
+      "step": 205340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.25196323315136e-09,
+      "loss": 1.239,
+      "step": 205345
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.223217612027933e-09,
+      "loss": 1.6217,
+      "step": 205350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.194529055033483e-09,
+      "loss": 1.2156,
+      "step": 205355
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.165897562333435e-09,
+      "loss": 1.3055,
+      "step": 205360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.13732313409099e-09,
+      "loss": 0.9348,
+      "step": 205365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.108805770469351e-09,
+      "loss": 1.4717,
+      "step": 205370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.080345471630612e-09,
+      "loss": 1.1941,
+      "step": 205375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.0519422377379745e-09,
+      "loss": 1.2239,
+      "step": 205380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.023596068953531e-09,
+      "loss": 1.2091,
+      "step": 205385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.995306965439374e-09,
+      "loss": 1.066,
+      "step": 205390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.967074927356488e-09,
+      "loss": 1.4242,
+      "step": 205395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.9388999548669625e-09,
+      "loss": 0.975,
+      "step": 205400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.910782048130671e-09,
+      "loss": 1.2117,
+      "step": 205405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.8827212073074855e-09,
+      "loss": 1.176,
+      "step": 205410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.854717432560609e-09,
+      "loss": 1.1713,
+      "step": 205415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.826770724046583e-09,
+      "loss": 1.1507,
+      "step": 205420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.7988810819274994e-09,
+      "loss": 1.5023,
+      "step": 205425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.771048506362121e-09,
+      "loss": 1.109,
+      "step": 205430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.743272997508099e-09,
+      "loss": 1.3335,
+      "step": 205435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.715554555524195e-09,
+      "loss": 1.1831,
+      "step": 205440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.687893180570282e-09,
+      "loss": 1.0092,
+      "step": 205445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.6602888728040104e-09,
+      "loss": 0.9957,
+      "step": 205450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.6327416323819224e-09,
+      "loss": 1.2282,
+      "step": 205455
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.60525145946167e-09,
+      "loss": 1.5487,
+      "step": 205460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.577818354199794e-09,
+      "loss": 1.4937,
+      "step": 205465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.5504423167550565e-09,
+      "loss": 1.1476,
+      "step": 205470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.523123347280669e-09,
+      "loss": 1.013,
+      "step": 205475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.495861445935392e-09,
+      "loss": 0.9534,
+      "step": 205480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.468656612873547e-09,
+      "loss": 1.2063,
+      "step": 205485
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.4415088482505664e-09,
+      "loss": 1.1479,
+      "step": 205490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.41441815222188e-09,
+      "loss": 1.4334,
+      "step": 205495
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.387384524941809e-09,
+      "loss": 1.3436,
+      "step": 205500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3604079665646745e-09,
+      "loss": 1.2471,
+      "step": 205505
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.333488477244798e-09,
+      "loss": 1.1656,
+      "step": 205510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3066260571353896e-09,
+      "loss": 1.4998,
+      "step": 205515
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.2798207063907715e-09,
+      "loss": 1.3855,
+      "step": 205520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.2530724251641525e-09,
+      "loss": 1.2705,
+      "step": 205525
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.226381213606525e-09,
+      "loss": 1.2358,
+      "step": 205530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.1997470718722085e-09,
+      "loss": 1.3658,
+      "step": 205535
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.173170000112194e-09,
+      "loss": 1.1708,
+      "step": 205540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.146649998478582e-09,
+      "loss": 0.991,
+      "step": 205545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.120187067122363e-09,
+      "loss": 1.103,
+      "step": 205550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.093781206195637e-09,
+      "loss": 1.3153,
+      "step": 205555
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.0674324158482846e-09,
+      "loss": 1.0792,
+      "step": 205560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.041140696231296e-09,
+      "loss": 1.1427,
+      "step": 205565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.014906047495661e-09,
+      "loss": 1.322,
+      "step": 205570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.98872846978904e-09,
+      "loss": 1.2327,
+      "step": 205575
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9626079632624235e-09,
+      "loss": 1.3874,
+      "step": 205580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9365445280645806e-09,
+      "loss": 1.3202,
+      "step": 205585
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.910538164344281e-09,
+      "loss": 1.1365,
+      "step": 205590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8845888722502964e-09,
+      "loss": 0.9731,
+      "step": 205595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8586966519302845e-09,
+      "loss": 1.1339,
+      "step": 205600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.832861503533016e-09,
+      "loss": 1.2299,
+      "step": 205605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.80708342720615e-09,
+      "loss": 1.2427,
+      "step": 205610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.781362423095127e-09,
+      "loss": 1.1887,
+      "step": 205615
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.755698491347606e-09,
+      "loss": 1.2233,
+      "step": 205620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.730091632111246e-09,
+      "loss": 1.2883,
+      "step": 205625
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.704541845531486e-09,
+      "loss": 1.1513,
+      "step": 205630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.679049131753766e-09,
+      "loss": 1.1192,
+      "step": 205635
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6536134909246365e-09,
+      "loss": 1.2425,
+      "step": 205640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.628234923189535e-09,
+      "loss": 1.0287,
+      "step": 205645
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6029134286916805e-09,
+      "loss": 1.1781,
+      "step": 205650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.577649007576513e-09,
+      "loss": 1.393,
+      "step": 205655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5524416599894715e-09,
+      "loss": 1.2749,
+      "step": 205660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.527291386072664e-09,
+      "loss": 1.304,
+      "step": 205665
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.50219818597042e-09,
+      "loss": 1.2564,
+      "step": 205670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.477162059827068e-09,
+      "loss": 1.0303,
+      "step": 205675
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.452183007784717e-09,
+      "loss": 0.9555,
+      "step": 205680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.427261029985476e-09,
+      "loss": 1.1488,
+      "step": 205685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.4023961265714516e-09,
+      "loss": 1.2354,
+      "step": 205690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.377588297686975e-09,
+      "loss": 1.0765,
+      "step": 205695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.352837543470824e-09,
+      "loss": 1.1823,
+      "step": 205700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3281438640662156e-09,
+      "loss": 1.2746,
+      "step": 205705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.303507259613039e-09,
+      "loss": 1.2829,
+      "step": 205710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.278927730253403e-09,
+      "loss": 1.5892,
+      "step": 205715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.254405276126085e-09,
+      "loss": 1.4387,
+      "step": 205720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.229939897372083e-09,
+      "loss": 1.6399,
+      "step": 205725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.205531594130175e-09,
+      "loss": 1.2051,
+      "step": 205730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1811803665413605e-09,
+      "loss": 1.1794,
+      "step": 205735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.156886214743306e-09,
+      "loss": 1.2443,
+      "step": 205740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.13264913887479e-09,
+      "loss": 1.1685,
+      "step": 205745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1084691390757e-09,
+      "loss": 1.188,
+      "step": 205750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.084346215482594e-09,
+      "loss": 1.0796,
+      "step": 205755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.060280368233139e-09,
+      "loss": 1.0082,
+      "step": 205760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.036271597465003e-09,
+      "loss": 1.1874,
+      "step": 205765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.012319903315854e-09,
+      "loss": 1.2241,
+      "step": 205770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.988425285923359e-09,
+      "loss": 1.0603,
+      "step": 205775
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.964587745421856e-09,
+      "loss": 1.0956,
+      "step": 205780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9408072819479015e-09,
+      "loss": 1.1046,
+      "step": 205785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.917083895639163e-09,
+      "loss": 1.301,
+      "step": 205790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.893417586628868e-09,
+      "loss": 1.3377,
+      "step": 205795
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.869808355052463e-09,
+      "loss": 1.2901,
+      "step": 205800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8462562010465065e-09,
+      "loss": 1.0572,
+      "step": 205805
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.822761124743114e-09,
+      "loss": 1.3321,
+      "step": 205810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.799323126278843e-09,
+      "loss": 1.0873,
+      "step": 205815
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7759422057858106e-09,
+      "loss": 1.2988,
+      "step": 205820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7526183633983535e-09,
+      "loss": 1.0359,
+      "step": 205825
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7293515992496985e-09,
+      "loss": 1.1338,
+      "step": 205830
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.706141913471962e-09,
+      "loss": 1.1978,
+      "step": 205835
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.682989306198371e-09,
+      "loss": 1.562,
+      "step": 205840
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.659893777559932e-09,
+      "loss": 1.124,
+      "step": 205845
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.636855327690981e-09,
+      "loss": 1.4749,
+      "step": 205850
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.613873956721415e-09,
+      "loss": 1.1075,
+      "step": 205855
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.59094966478224e-09,
+      "loss": 1.3425,
+      "step": 205860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.568082452004463e-09,
+      "loss": 1.1574,
+      "step": 205865
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.545272318519089e-09,
+      "loss": 1.3417,
+      "step": 205870
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.522519264457126e-09,
+      "loss": 1.3096,
+      "step": 205875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.499823289946248e-09,
+      "loss": 1.2196,
+      "step": 205880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4771843951185725e-09,
+      "loss": 1.518,
+      "step": 205885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.454602580101775e-09,
+      "loss": 1.1918,
+      "step": 205890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.432077845025751e-09,
+      "loss": 1.2059,
+      "step": 205895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.409610190018177e-09,
+      "loss": 1.1007,
+      "step": 205900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.387199615208948e-09,
+      "loss": 1.4052,
+      "step": 205905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.364846120723521e-09,
+      "loss": 1.4761,
+      "step": 205910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.34254970669179e-09,
+      "loss": 1.1402,
+      "step": 205915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.320310373240321e-09,
+      "loss": 1.3567,
+      "step": 205920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.298128120496792e-09,
+      "loss": 1.2413,
+      "step": 205925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.276002948585545e-09,
+      "loss": 1.0822,
+      "step": 205930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.253934857635367e-09,
+      "loss": 0.9666,
+      "step": 205935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.231923847771713e-09,
+      "loss": 1.3175,
+      "step": 205940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.209969919120038e-09,
+      "loss": 1.8993,
+      "step": 205945
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.188073071805798e-09,
+      "loss": 1.7953,
+      "step": 205950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.166233305953338e-09,
+      "loss": 1.1993,
+      "step": 205955
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.144450621688112e-09,
+      "loss": 0.9005,
+      "step": 205960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.122725019134466e-09,
+      "loss": 1.3585,
+      "step": 205965
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.101056498415634e-09,
+      "loss": 1.1267,
+      "step": 205970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.079445059657072e-09,
+      "loss": 1.2052,
+      "step": 205975
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.057890702980905e-09,
+      "loss": 1.153,
+      "step": 205980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.036393428510366e-09,
+      "loss": 1.4762,
+      "step": 205985
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.01495323636869e-09,
+      "loss": 1.2492,
+      "step": 205990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.993570126678004e-09,
+      "loss": 1.3234,
+      "step": 205995
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.97224409956043e-09,
+      "loss": 1.3017,
+      "step": 206000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.950975155136982e-09,
+      "loss": 1.2175,
+      "step": 206005
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.929763293530897e-09,
+      "loss": 1.1644,
+      "step": 206010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.908608514862078e-09,
+      "loss": 1.2124,
+      "step": 206015
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.887510819250429e-09,
+      "loss": 1.1592,
+      "step": 206020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.866470206818074e-09,
+      "loss": 1.0242,
+      "step": 206025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.845486677684917e-09,
+      "loss": 1.4088,
+      "step": 206030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.824560231970864e-09,
+      "loss": 1.1345,
+      "step": 206035
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.803690869794707e-09,
+      "loss": 1.233,
+      "step": 206040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.782878591276351e-09,
+      "loss": 1.0488,
+      "step": 206045
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.762123396533479e-09,
+      "loss": 1.0931,
+      "step": 206050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.741425285685995e-09,
+      "loss": 1.0309,
+      "step": 206055
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7207842588504737e-09,
+      "loss": 1.1457,
+      "step": 206060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.700200316146818e-09,
+      "loss": 1.3309,
+      "step": 206065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6796734576916016e-09,
+      "loss": 1.2665,
+      "step": 206070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6592036836013978e-09,
+      "loss": 1.1586,
+      "step": 206075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6387909939938903e-09,
+      "loss": 1.2987,
+      "step": 206080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6184353889856525e-09,
+      "loss": 1.3874,
+      "step": 206085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.598136868692148e-09,
+      "loss": 1.0098,
+      "step": 206090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.57789543322995e-09,
+      "loss": 1.5775,
+      "step": 206095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5577110827156314e-09,
+      "loss": 1.0479,
+      "step": 206100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5375838172624356e-09,
+      "loss": 1.0283,
+      "step": 206105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5175136369858253e-09,
+      "loss": 1.3224,
+      "step": 206110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.497500542001264e-09,
+      "loss": 0.9829,
+      "step": 206115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.477544532423105e-09,
+      "loss": 1.1989,
+      "step": 206120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4576456083645903e-09,
+      "loss": 1.4655,
+      "step": 206125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4378037699400736e-09,
+      "loss": 1.1379,
+      "step": 206130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4180190172616868e-09,
+      "loss": 0.9303,
+      "step": 206135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.398291350442673e-09,
+      "loss": 1.4748,
+      "step": 206140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.378620769597385e-09,
+      "loss": 1.0954,
+      "step": 206145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359007274835735e-09,
+      "loss": 1.2082,
+      "step": 206150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3394508662709656e-09,
+      "loss": 1.0797,
+      "step": 206155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3199515440152097e-09,
+      "loss": 1.0859,
+      "step": 206160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.300509308179489e-09,
+      "loss": 1.1437,
+      "step": 206165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2811241588737165e-09,
+      "loss": 1.1749,
+      "step": 206170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2617960962100235e-09,
+      "loss": 1.2753,
+      "step": 206175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2425251202983233e-09,
+      "loss": 1.0222,
+      "step": 206180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.223311231248527e-09,
+      "loss": 1.1598,
+      "step": 206185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.204154429170547e-09,
+      "loss": 1.1132,
+      "step": 206190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1850547141731857e-09,
+      "loss": 1.2767,
+      "step": 206195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1660120863663547e-09,
+      "loss": 1.2662,
+      "step": 206200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.147026545857745e-09,
+      "loss": 1.4264,
+      "step": 206205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.12809809275727e-09,
+      "loss": 1.0946,
+      "step": 206210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1092267271715102e-09,
+      "loss": 1.4023,
+      "step": 206215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.090412449209268e-09,
+      "loss": 1.0788,
+      "step": 206220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0716552589782343e-09,
+      "loss": 0.9386,
+      "step": 206225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.052955156583881e-09,
+      "loss": 1.2932,
+      "step": 206230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.03431214213501e-09,
+      "loss": 1.1928,
+      "step": 206235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.015726215735981e-09,
+      "loss": 1.2813,
+      "step": 206240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.997197377494487e-09,
+      "loss": 1.2857,
+      "step": 206245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9787256275171093e-09,
+      "loss": 1.1175,
+      "step": 206250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9603109659070984e-09,
+      "loss": 1.4991,
+      "step": 206255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9419533927699253e-09,
+      "loss": 1.3159,
+      "step": 206260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.923652908212171e-09,
+      "loss": 1.2186,
+      "step": 206265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9054095123370874e-09,
+      "loss": 1.3477,
+      "step": 206270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8872232052490347e-09,
+      "loss": 1.4841,
+      "step": 206275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8690939870512635e-09,
+      "loss": 1.3682,
+      "step": 206280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8510218578492456e-09,
+      "loss": 1.0821,
+      "step": 206285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8330068177440107e-09,
+      "loss": 1.0377,
+      "step": 206290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.81504886683881e-09,
+      "loss": 1.2575,
+      "step": 206295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.797148005238004e-09,
+      "loss": 1.2158,
+      "step": 206300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.779304233041513e-09,
+      "loss": 1.2168,
+      "step": 206305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.761517550352588e-09,
+      "loss": 1.2147,
+      "step": 206310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.743787957272259e-09,
+      "loss": 1.3363,
+      "step": 206315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.726115453902667e-09,
+      "loss": 1.1725,
+      "step": 206320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7085000403426208e-09,
+      "loss": 1.2629,
+      "step": 206325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.690941716695372e-09,
+      "loss": 1.2732,
+      "step": 206330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6734404830597306e-09,
+      "loss": 1.3109,
+      "step": 206335
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.655996339535616e-09,
+      "loss": 1.6448,
+      "step": 206340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6386092862229484e-09,
+      "loss": 1.1984,
+      "step": 206345
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.621279323220538e-09,
+      "loss": 1.3288,
+      "step": 206350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6040064506283046e-09,
+      "loss": 1.3144,
+      "step": 206355
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.586790668542838e-09,
+      "loss": 1.1782,
+      "step": 206360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.569631977065168e-09,
+      "loss": 1.1043,
+      "step": 206365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5525303762929943e-09,
+      "loss": 1.14,
+      "step": 206370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.535485866321796e-09,
+      "loss": 1.5701,
+      "step": 206375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5184984472503837e-09,
+      "loss": 1.2053,
+      "step": 206380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.501568119175346e-09,
+      "loss": 1.3595,
+      "step": 206385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4846948821932726e-09,
+      "loss": 1.428,
+      "step": 206390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.467878736400753e-09,
+      "loss": 1.136,
+      "step": 206395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4511196818943763e-09,
+      "loss": 1.1962,
+      "step": 206400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4344177187696215e-09,
+      "loss": 1.4271,
+      "step": 206405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.417772847120858e-09,
+      "loss": 1.2984,
+      "step": 206410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.401185067044676e-09,
+      "loss": 1.5375,
+      "step": 206415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3846543786343324e-09,
+      "loss": 1.1207,
+      "step": 206420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.368180781984197e-09,
+      "loss": 1.3614,
+      "step": 206425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3517642771897496e-09,
+      "loss": 1.1198,
+      "step": 206430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3354048643442485e-09,
+      "loss": 1.0383,
+      "step": 206435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.319102543539842e-09,
+      "loss": 1.0913,
+      "step": 206440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.302857314872009e-09,
+      "loss": 1.2956,
+      "step": 206445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2866691784306783e-09,
+      "loss": 1.1581,
+      "step": 206450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2705381343113287e-09,
+      "loss": 1.2956,
+      "step": 206455
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.254464182603888e-09,
+      "loss": 1.1002,
+      "step": 206460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2384473234005055e-09,
+      "loss": 1.3563,
+      "step": 206465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.222487556792219e-09,
+      "loss": 1.1454,
+      "step": 206470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.206584882872287e-09,
+      "loss": 1.346,
+      "step": 206475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1907393017295276e-09,
+      "loss": 0.9816,
+      "step": 206480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.174950813454979e-09,
+      "loss": 1.1176,
+      "step": 206485
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1592194181385696e-09,
+      "loss": 1.1665,
+      "step": 206490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.143545115870227e-09,
+      "loss": 0.9421,
+      "step": 206495
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1279279067398795e-09,
+      "loss": 1.0976,
+      "step": 206500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.112367790836345e-09,
+      "loss": 1.3575,
+      "step": 206505
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.096864768248441e-09,
+      "loss": 1.2888,
+      "step": 206510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.081418839066096e-09,
+      "loss": 1.5373,
+      "step": 206515
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.066030003374797e-09,
+      "loss": 1.1478,
+      "step": 206520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.050698261264472e-09,
+      "loss": 1.3034,
+      "step": 206525
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0354236128217188e-09,
+      "loss": 1.1937,
+      "step": 206530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0202060581342443e-09,
+      "loss": 1.2264,
+      "step": 206535
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0050455972897564e-09,
+      "loss": 1.3141,
+      "step": 206540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9899422303726325e-09,
+      "loss": 0.9878,
+      "step": 206545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9748959574705794e-09,
+      "loss": 1.353,
+      "step": 206550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9599067786701953e-09,
+      "loss": 1.1472,
+      "step": 206555
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9449746940547465e-09,
+      "loss": 0.9985,
+      "step": 206560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.930099703711941e-09,
+      "loss": 1.2973,
+      "step": 206565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.915281807726155e-09,
+      "loss": 1.3532,
+      "step": 206570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9005210061806558e-09,
+      "loss": 1.0409,
+      "step": 206575
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8858172991609303e-09,
+      "loss": 1.0704,
+      "step": 206580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.871170686751356e-09,
+      "loss": 1.054,
+      "step": 206585
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8565811690340884e-09,
+      "loss": 1.4252,
+      "step": 206590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8420487460935054e-09,
+      "loss": 1.1474,
+      "step": 206595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8275734180128734e-09,
+      "loss": 1.1237,
+      "step": 206600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.813155184874349e-09,
+      "loss": 1.2555,
+      "step": 206605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7987940467600884e-09,
+      "loss": 0.951,
+      "step": 206610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.784490003751138e-09,
+      "loss": 1.6515,
+      "step": 206615
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7702430559318751e-09,
+      "loss": 1.5169,
+      "step": 206620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.756053203381125e-09,
+      "loss": 1.3007,
+      "step": 206625
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7419204461810447e-09,
+      "loss": 1.4949,
+      "step": 206630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.72784478441268e-09,
+      "loss": 0.9435,
+      "step": 206635
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.713826218154857e-09,
+      "loss": 1.3376,
+      "step": 206640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6998647474897323e-09,
+      "loss": 1.3287,
+      "step": 206645
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6859603724961315e-09,
+      "loss": 0.6999,
+      "step": 206650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6721130932528806e-09,
+      "loss": 1.3351,
+      "step": 206655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6583229098388053e-09,
+      "loss": 1.0363,
+      "step": 206660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.644589822333842e-09,
+      "loss": 1.3039,
+      "step": 206665
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.630913830816816e-09,
+      "loss": 1.3697,
+      "step": 206670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.617294935364333e-09,
+      "loss": 1.3184,
+      "step": 206675
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6037331360541087e-09,
+      "loss": 1.208,
+      "step": 206680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5902284329649686e-09,
+      "loss": 1.1206,
+      "step": 206685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5767808261724082e-09,
+      "loss": 1.0913,
+      "step": 206690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5633903157552533e-09,
+      "loss": 1.2186,
+      "step": 206695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5500569017878886e-09,
+      "loss": 1.1132,
+      "step": 206700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5367805843480299e-09,
+      "loss": 1.0874,
+      "step": 206705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.523561363510062e-09,
+      "loss": 1.4499,
+      "step": 206710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5103992393505906e-09,
+      "loss": 1.1394,
+      "step": 206715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4972942119451105e-09,
+      "loss": 1.2635,
+      "step": 206720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4842462813668968e-09,
+      "loss": 1.7627,
+      "step": 206725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4712554476925544e-09,
+      "loss": 1.1976,
+      "step": 206730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4583217109942483e-09,
+      "loss": 1.5855,
+      "step": 206735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.445445071346363e-09,
+      "loss": 1.6473,
+      "step": 206740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4326255288232837e-09,
+      "loss": 1.3951,
+      "step": 206745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.419863083497175e-09,
+      "loss": 1.2142,
+      "step": 206750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4071577354424216e-09,
+      "loss": 1.2267,
+      "step": 206755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3945094847289675e-09,
+      "loss": 1.2216,
+      "step": 206760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3819183314323082e-09,
+      "loss": 1.2243,
+      "step": 206765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3693842756212772e-09,
+      "loss": 1.4662,
+      "step": 206770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3569073173691495e-09,
+      "loss": 1.1851,
+      "step": 206775
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3444874567480892e-09,
+      "loss": 1.0289,
+      "step": 206780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3321246938269306e-09,
+      "loss": 1.6796,
+      "step": 206785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3198190286767277e-09,
+      "loss": 1.2666,
+      "step": 206790
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3075704613685348e-09,
+      "loss": 1.0727,
+      "step": 206795
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.295378991972296e-09,
+      "loss": 1.3694,
+      "step": 206800
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2832446205568449e-09,
+      "loss": 1.2144,
+      "step": 206805
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.271167347192126e-09,
+      "loss": 1.0537,
+      "step": 206810
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2591471719458625e-09,
+      "loss": 1.0511,
+      "step": 206815
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.247184094889109e-09,
+      "loss": 1.0666,
+      "step": 206820
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2352781160884786e-09,
+      "loss": 1.0693,
+      "step": 206825
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.223429235611695e-09,
+      "loss": 1.089,
+      "step": 206830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2116374535275921e-09,
+      "loss": 1.6049,
+      "step": 206835
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1999027699027832e-09,
+      "loss": 1.5052,
+      "step": 206840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1882251848038818e-09,
+      "loss": 1.0807,
+      "step": 206845
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1766046982986112e-09,
+      "loss": 1.1937,
+      "step": 206850
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.165041310452475e-09,
+      "loss": 1.1824,
+      "step": 206855
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1535350213320862e-09,
+      "loss": 1.3009,
+      "step": 206860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1420858310029482e-09,
+      "loss": 1.3415,
+      "step": 206865
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.130693739530564e-09,
+      "loss": 1.2457,
+      "step": 206870
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1193587469793266e-09,
+      "loss": 1.1346,
+      "step": 206875
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1080808534158494e-09,
+      "loss": 1.4657,
+      "step": 206880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0968600589011947e-09,
+      "loss": 1.4526,
+      "step": 206885
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.085696363503086e-09,
+      "loss": 1.2094,
+      "step": 206890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0745897672825857e-09,
+      "loss": 1.2309,
+      "step": 206895
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0635402703040865e-09,
+      "loss": 1.1197,
+      "step": 206900
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0525478726308714e-09,
+      "loss": 1.0972,
+      "step": 206905
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0416125743262228e-09,
+      "loss": 1.0548,
+      "step": 206910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0307343754512034e-09,
+      "loss": 1.3568,
+      "step": 206915
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0199132760690954e-09,
+      "loss": 0.9414,
+      "step": 206920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0091492762409615e-09,
+      "loss": 1.2319,
+      "step": 206925
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.984423760278638e-10,
+      "loss": 1.2466,
+      "step": 206930
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.87792575493085e-10,
+      "loss": 1.1457,
+      "step": 206935
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.77199874695467e-10,
+      "loss": 1.199,
+      "step": 206940
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.66664273696072e-10,
+      "loss": 1.5101,
+      "step": 206945
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.561857725548517e-10,
+      "loss": 1.5137,
+      "step": 206950
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.457643713317588e-10,
+      "loss": 1.3501,
+      "step": 206955
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.354000700856347e-10,
+      "loss": 1.2362,
+      "step": 206960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.250928688775418e-10,
+      "loss": 1.2333,
+      "step": 206965
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.148427677652117e-10,
+      "loss": 1.2292,
+      "step": 206970
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.046497668063759e-10,
+      "loss": 1.0828,
+      "step": 206975
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.945138660598762e-10,
+      "loss": 1.2222,
+      "step": 206980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.844350655845546e-10,
+      "loss": 1.1672,
+      "step": 206985
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.744133654370324e-10,
+      "loss": 1.2207,
+      "step": 206990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.644487656750411e-10,
+      "loss": 1.1265,
+      "step": 206995
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.54541266354092e-10,
+      "loss": 0.9876,
+      "step": 207000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.446908675319166e-10,
+      "loss": 1.2115,
+      "step": 207005
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.348975692651362e-10,
+      "loss": 1.3508,
+      "step": 207010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.251613716092622e-10,
+      "loss": 1.3837,
+      "step": 207015
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.154822746198054e-10,
+      "loss": 1.0992,
+      "step": 207020
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.058602783511671e-10,
+      "loss": 1.1745,
+      "step": 207025
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.962953828599684e-10,
+      "loss": 1.279,
+      "step": 207030
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.867875881995002e-10,
+      "loss": 1.2745,
+      "step": 207035
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.773368944252735e-10,
+      "loss": 0.9948,
+      "step": 207040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.67943301590579e-10,
+      "loss": 1.4963,
+      "step": 207045
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.586068097487076e-10,
+      "loss": 1.5524,
+      "step": 207050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.493274189529498e-10,
+      "loss": 1.4826,
+      "step": 207055
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.401051292565964e-10,
+      "loss": 1.0848,
+      "step": 207060
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.309399407129381e-10,
+      "loss": 1.4026,
+      "step": 207065
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.218318533730451e-10,
+      "loss": 1.2402,
+      "step": 207070
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.127808672902082e-10,
+      "loss": 1.1483,
+      "step": 207075
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.037869825154975e-10,
+      "loss": 1.2877,
+      "step": 207080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.948501990999834e-10,
+      "loss": 1.1826,
+      "step": 207085
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.859705170958464e-10,
+      "loss": 0.9742,
+      "step": 207090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.771479365519362e-10,
+      "loss": 1.2472,
+      "step": 207095
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.683824575204334e-10,
+      "loss": 1.1836,
+      "step": 207100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.596740800501878e-10,
+      "loss": 1.184,
+      "step": 207105
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.510228041911593e-10,
+      "loss": 1.3473,
+      "step": 207110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.42428629993308e-10,
+      "loss": 1.1385,
+      "step": 207115
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.338915575054838e-10,
+      "loss": 1.1943,
+      "step": 207120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.254115867765365e-10,
+      "loss": 1.0466,
+      "step": 207125
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.169887178542055e-10,
+      "loss": 1.3532,
+      "step": 207130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.086229507873409e-10,
+      "loss": 1.2478,
+      "step": 207135
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.003142856236821e-10,
+      "loss": 1.1445,
+      "step": 207140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.920627224098585e-10,
+      "loss": 1.7006,
+      "step": 207145
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.838682611936097e-10,
+      "loss": 1.2497,
+      "step": 207150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.757309020215651e-10,
+      "loss": 1.7915,
+      "step": 207155
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.67650644940354e-10,
+      "loss": 1.2492,
+      "step": 207160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.596274899966059e-10,
+      "loss": 1.5989,
+      "step": 207165
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.516614372347296e-10,
+      "loss": 1.3543,
+      "step": 207170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.437524867024646e-10,
+      "loss": 1.2876,
+      "step": 207175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.359006384419996e-10,
+      "loss": 1.2566,
+      "step": 207180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.28105892501074e-10,
+      "loss": 1.0163,
+      "step": 207185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.203682489218765e-10,
+      "loss": 1.4427,
+      "step": 207190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.126877077499259e-10,
+      "loss": 1.3484,
+      "step": 207195
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.050642690296314e-10,
+      "loss": 1.4657,
+      "step": 207200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.974979328031815e-10,
+      "loss": 1.3212,
+      "step": 207205
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.899886991138747e-10,
+      "loss": 1.2005,
+      "step": 207210
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.825365680050098e-10,
+      "loss": 1.105,
+      "step": 207215
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.751415395198855e-10,
+      "loss": 1.0841,
+      "step": 207220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.678036136984698e-10,
+      "loss": 1.3158,
+      "step": 207225
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.605227905851717e-10,
+      "loss": 1.4065,
+      "step": 207230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.532990702210693e-10,
+      "loss": 1.2152,
+      "step": 207235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.461324526461308e-10,
+      "loss": 1.2922,
+      "step": 207240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3902293790254456e-10,
+      "loss": 1.4343,
+      "step": 207245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3197052603027865e-10,
+      "loss": 1.3205,
+      "step": 207250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2497521706930113e-10,
+      "loss": 0.9957,
+      "step": 207255
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.180370110606902e-10,
+      "loss": 1.5313,
+      "step": 207260
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.111559080421934e-10,
+      "loss": 0.96,
+      "step": 207265
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0433190805599933e-10,
+      "loss": 1.3801,
+      "step": 207270
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.97565011137635e-10,
+      "loss": 1.2153,
+      "step": 207275
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.90855217329289e-10,
+      "loss": 1.2668,
+      "step": 207280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8420252666648837e-10,
+      "loss": 1.2529,
+      "step": 207285
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7760693918809097e-10,
+      "loss": 1.2987,
+      "step": 207290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.710684549318444e-10,
+      "loss": 1.316,
+      "step": 207295
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.645870739354962e-10,
+      "loss": 1.1709,
+      "step": 207300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5816279623568373e-10,
+      "loss": 1.2626,
+      "step": 207305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5179562186793414e-10,
+      "loss": 1.3376,
+      "step": 207310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4548555087110524e-10,
+      "loss": 1.1205,
+      "step": 207315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3923258327961396e-10,
+      "loss": 1.5028,
+      "step": 207320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.330367191289874e-10,
+      "loss": 1.1426,
+      "step": 207325
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.268979584558629e-10,
+      "loss": 1.0244,
+      "step": 207330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2081630129465744e-10,
+      "loss": 1.1955,
+      "step": 207335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.147917476786777e-10,
+      "loss": 1.762,
+      "step": 207340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.08824297644561e-10,
+      "loss": 1.2423,
+      "step": 207345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0291395122450385e-10,
+      "loss": 1.1783,
+      "step": 207350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9706070845403336e-10,
+      "loss": 1.1058,
+      "step": 207355
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9126456936534596e-10,
+      "loss": 1.2812,
+      "step": 207360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8552553399285865e-10,
+      "loss": 1.4118,
+      "step": 207365
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.798436023665474e-10,
+      "loss": 1.1409,
+      "step": 207370
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.742187745219393e-10,
+      "loss": 1.0804,
+      "step": 207375
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6865105049012077e-10,
+      "loss": 1.2141,
+      "step": 207380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6314043030217785e-10,
+      "loss": 1.2449,
+      "step": 207385
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.576869139903071e-10,
+      "loss": 1.1922,
+      "step": 207390
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5229050158559477e-10,
+      "loss": 1.3055,
+      "step": 207395
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.469511931180169e-10,
+      "loss": 1.0529,
+      "step": 207400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.416689886186596e-10,
+      "loss": 1.1796,
+      "step": 207405
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.364438881186093e-10,
+      "loss": 1.4629,
+      "step": 207410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.312758916467317e-10,
+      "loss": 1.1119,
+      "step": 207415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.261649992318926e-10,
+      "loss": 1.3924,
+      "step": 207420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2111121090406807e-10,
+      "loss": 1.1321,
+      "step": 207425
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1611452669323407e-10,
+      "loss": 1.1445,
+      "step": 207430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1117494662603599e-10,
+      "loss": 1.1889,
+      "step": 207435
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0629247073133963e-10,
+      "loss": 1.082,
+      "step": 207440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0146709903801075e-10,
+      "loss": 1.5404,
+      "step": 207445
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9669883157158452e-10,
+      "loss": 1.4808,
+      "step": 207450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9198766836092674e-10,
+      "loss": 1.4694,
+      "step": 207455
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8733360943268274e-10,
+      "loss": 1.2256,
+      "step": 207460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8273665481238768e-10,
+      "loss": 1.2436,
+      "step": 207465
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.781968045277971e-10,
+      "loss": 1.3001,
+      "step": 207470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.737140586033359e-10,
+      "loss": 1.4242,
+      "step": 207475
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6928841706675968e-10,
+      "loss": 1.0306,
+      "step": 207480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6491987994138313e-10,
+      "loss": 1.4684,
+      "step": 207485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6060844725163115e-10,
+      "loss": 1.3649,
+      "step": 207490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5635411902414909e-10,
+      "loss": 1.0161,
+      "step": 207495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5215689528336185e-10,
+      "loss": 1.2857,
+      "step": 207500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.480167760503637e-10,
+      "loss": 1.2614,
+      "step": 207505
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4393376135179992e-10,
+      "loss": 1.4143,
+      "step": 207510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3990785120987505e-10,
+      "loss": 1.1881,
+      "step": 207515
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.359390456467935e-10,
+      "loss": 1.5888,
+      "step": 207520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3202734468586996e-10,
+      "loss": 1.6428,
+      "step": 207525
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2817274835041916e-10,
+      "loss": 1.3586,
+      "step": 207530
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2437525666042505e-10,
+      "loss": 1.2252,
+      "step": 207535
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2063486964031256e-10,
+      "loss": 1.2644,
+      "step": 207540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.169515873089555e-10,
+      "loss": 1.1994,
+      "step": 207545
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.133254096874481e-10,
+      "loss": 1.4219,
+      "step": 207550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.097563367979948e-10,
+      "loss": 1.1325,
+      "step": 207555
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0624436866057963e-10,
+      "loss": 1.8244,
+      "step": 207560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0278950529518661e-10,
+      "loss": 1.2167,
+      "step": 207565
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.939174672068951e-11,
+      "loss": 1.5265,
+      "step": 207570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.605109295707237e-11,
+      "loss": 1.4024,
+      "step": 207575
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.27675440243192e-11,
+      "loss": 1.2125,
+      "step": 207580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.954109993908333e-11,
+      "loss": 1.3079,
+      "step": 207585
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.637176072134879e-11,
+      "loss": 1.2097,
+      "step": 207590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.325952638998936e-11,
+      "loss": 1.1825,
+      "step": 207595
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.020439696054815e-11,
+      "loss": 1.7065,
+      "step": 207600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.720637245189899e-11,
+      "loss": 1.1599,
+      "step": 207605
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.426545288180543e-11,
+      "loss": 1.2854,
+      "step": 207610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.138163826470035e-11,
+      "loss": 1.2403,
+      "step": 207615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.855492861945756e-11,
+      "loss": 1.3886,
+      "step": 207620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.578532396050996e-11,
+      "loss": 1.3028,
+      "step": 207625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.307282430562112e-11,
+      "loss": 0.9075,
+      "step": 207630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.041742966811371e-11,
+      "loss": 1.2429,
+      "step": 207635
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.7819140064641065e-11,
+      "loss": 1.2202,
+      "step": 207640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.527795550852588e-11,
+      "loss": 1.6076,
+      "step": 207645
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.2793876016421494e-11,
+      "loss": 1.2766,
+      "step": 207650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.036690160054036e-11,
+      "loss": 1.2463,
+      "step": 207655
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.79970322764256e-11,
+      "loss": 1.4653,
+      "step": 207660
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.568426805628967e-11,
+      "loss": 1.0409,
+      "step": 207665
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.342860895345524e-11,
+      "loss": 1.1999,
+      "step": 207670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1230054981244994e-11,
+      "loss": 1.0179,
+      "step": 207675
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9088606151871376e-11,
+      "loss": 1.0538,
+      "step": 207680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.700426247865707e-11,
+      "loss": 1.4769,
+      "step": 207685
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.497702397159408e-11,
+      "loss": 1.307,
+      "step": 207690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3006890644005084e-11,
+      "loss": 1.0895,
+      "step": 207695
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.109386250699231e-11,
+      "loss": 1.707,
+      "step": 207700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9237939569437544e-11,
+      "loss": 1.2972,
+      "step": 207705
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7439121844663463e-11,
+      "loss": 0.9232,
+      "step": 207710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5697409341551847e-11,
+      "loss": 1.0497,
+      "step": 207715
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4012802070094708e-11,
+      "loss": 1.1025,
+      "step": 207720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2385300040284052e-11,
+      "loss": 1.416,
+      "step": 207725
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0814903261001662e-11,
+      "loss": 1.1898,
+      "step": 207730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9301611741129324e-11,
+      "loss": 1.1878,
+      "step": 207735
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7845425489548816e-11,
+      "loss": 1.1746,
+      "step": 207740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.644634451403171e-11,
+      "loss": 1.2324,
+      "step": 207745
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5104368824570004e-11,
+      "loss": 1.2073,
+      "step": 207750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3819498425604594e-11,
+      "loss": 1.6117,
+      "step": 207755
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.259173332823771e-11,
+      "loss": 1.2236,
+      "step": 207760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.142107353580002e-11,
+      "loss": 1.0858,
+      "step": 207765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0307519057173309e-11,
+      "loss": 1.3195,
+      "step": 207770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.251069899018916e-12,
+      "loss": 1.1027,
+      "step": 207775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.251726065777732e-12,
+      "loss": 1.7533,
+      "step": 207780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.309487564111095e-12,
+      "loss": 1.1439,
+      "step": 207785
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.4243543984598974e-12,
+      "loss": 1.3977,
+      "step": 207790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.596326575485478e-12,
+      "loss": 1.4657,
+      "step": 207795
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8254040996287275e-12,
+      "loss": 1.006,
+      "step": 207800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.111586974220316e-12,
+      "loss": 0.9912,
+      "step": 207805
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4548752037011356e-12,
+      "loss": 0.9285,
+      "step": 207810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.855268791401855e-12,
+      "loss": 1.0488,
+      "step": 207815
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3127677417633664e-12,
+      "loss": 1.0731,
+      "step": 207820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.827372058116339e-12,
+      "loss": 1.218,
+      "step": 207825
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3990817415709956e-12,
+      "loss": 1.1464,
+      "step": 207830
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0278967965682284e-12,
+      "loss": 1.202,
+      "step": 207835
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.138172231080376e-13,
+      "loss": 1.202,
+      "step": 207840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.568430245210919e-13,
+      "loss": 1.2406,
+      "step": 207845
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.569742019176147e-13,
+      "loss": 1.1284,
+      "step": 207850
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1421075640782876e-13,
+      "loss": 0.9122,
+      "step": 207855
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.855268910195719e-14,
+      "loss": 1.1604,
+      "step": 207860
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.1307,
+      "step": 207865
+    },
+    {
+      "epoch": 1.0,
+      "step": 207865,
+      "total_flos": 3.402689247346426e+18,
+      "train_loss": 1.2418477715456762,
+      "train_runtime": 258582.6373,
+      "train_samples_per_second": 0.804,
+      "train_steps_per_second": 0.804
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 207865,
+  "num_train_epochs": 1,
+  "save_steps": 10000,
+  "total_flos": 3.402689247346426e+18,
+  "trial_name": null,
+  "trial_params": null
+}