diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9658 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999584648612726,
+  "eval_steps": 500,
+  "global_step": 8025,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995210837798e-05,
+      "loss": 1.2227,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999808433695394e-05,
+      "loss": 1.1204,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999956897650272e-05,
+      "loss": 1.1247,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999233737717387e-05,
+      "loss": 1.2112,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999880271862381e-05,
+      "loss": 1.1884,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999827592087336e-05,
+      "loss": 1.0598,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997653346484386e-05,
+      "loss": 0.9276,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999693499784216e-05,
+      "loss": 0.9748,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999612087769892e-05,
+      "loss": 1.1462,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995210989173844e-05,
+      "loss": 1.0131,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999420533575299e-05,
+      "loss": 0.911,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999310392128935e-05,
+      "loss": 1.1607,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999190675000282e-05,
+      "loss": 0.7777,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999061382648015e-05,
+      "loss": 0.8488,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998922515567496e-05,
+      "loss": 0.9665,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998774074290771e-05,
+      "loss": 1.1715,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998616059386566e-05,
+      "loss": 0.9061,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998448471460289e-05,
+      "loss": 0.701,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998271311154025e-05,
+      "loss": 0.7553,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9980845791465334e-05,
+      "loss": 0.9345,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997888276153245e-05,
+      "loss": 1.0435,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976824029262635e-05,
+      "loss": 0.7442,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997466960254355e-05,
+      "loss": 0.9111,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9972419489629524e-05,
+      "loss": 1.067,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997007369914149e-05,
+      "loss": 0.8336,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996763224006692e-05,
+      "loss": 0.8885,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996509512175987e-05,
+      "loss": 1.0047,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996246235394087e-05,
+      "loss": 0.9941,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995973394669693e-05,
+      "loss": 1.0487,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995690991048146e-05,
+      "loss": 0.8632,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995399025611429e-05,
+      "loss": 0.8275,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9950974994781574e-05,
+      "loss": 0.7841,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994786413803577e-05,
+      "loss": 0.9355,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994465769779559e-05,
+      "loss": 0.6579,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994135568634598e-05,
+      "loss": 0.7775,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993795811633803e-05,
+      "loss": 0.6578,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9934465000788945e-05,
+      "loss": 0.6479,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9930876353082e-05,
+      "loss": 1.0595,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99271921869665e-05,
+      "loss": 1.0505,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992341251655768e-05,
+      "loss": 0.692,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991953735633673e-05,
+      "loss": 1.0227,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9915566721150645e-05,
+      "loss": 0.8125,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991150062621225e-05,
+      "loss": 0.9772,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990733908710009e-05,
+      "loss": 1.0627,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99030821197584e-05,
+      "loss": 0.9524,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9898729740497016e-05,
+      "loss": 0.8736,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989428196599135e-05,
+      "loss": 0.9929,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988973881328228e-05,
+      "loss": 0.7192,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988510029977613e-05,
+      "loss": 1.0527,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988036644324457e-05,
+      "loss": 1.1283,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987553726182457e-05,
+      "loss": 0.8064,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987061277401832e-05,
+      "loss": 0.6942,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986559299869314e-05,
+      "loss": 0.8781,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986047795508147e-05,
+      "loss": 0.8578,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9855267662780716e-05,
+      "loss": 0.6944,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9849962141753215e-05,
+      "loss": 0.8489,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984456141232618e-05,
+      "loss": 0.9016,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983906549519158e-05,
+      "loss": 0.9608,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9833474411406103e-05,
+      "loss": 0.9215,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982778818239101e-05,
+      "loss": 0.6738,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982200682993213e-05,
+      "loss": 0.7913,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981613037617975e-05,
+      "loss": 0.8519,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9810158843648466e-05,
+      "loss": 0.7425,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.980409225521721e-05,
+      "loss": 0.757,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979793063412908e-05,
+      "loss": 0.8566,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979167400399128e-05,
+      "loss": 0.8195,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9785322388775024e-05,
+      "loss": 0.7646,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9778875812815436e-05,
+      "loss": 0.8885,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977233430081148e-05,
+      "loss": 1.0574,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976569787782584e-05,
+      "loss": 0.9052,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9758966569284856e-05,
+      "loss": 1.0471,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9752140400978374e-05,
+      "loss": 0.8318,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97452193990597e-05,
+      "loss": 0.9293,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973820359004548e-05,
+      "loss": 0.8793,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97310930008156e-05,
+      "loss": 0.8759,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972388765861305e-05,
+      "loss": 0.9394,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9716587591043885e-05,
+      "loss": 1.3653,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970919282607707e-05,
+      "loss": 0.916,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970170339204438e-05,
+      "loss": 1.0014,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9694119317640324e-05,
+      "loss": 0.6814,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968644063192198e-05,
+      "loss": 0.8628,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967866736430893e-05,
+      "loss": 0.8507,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967079954458311e-05,
+      "loss": 1.0726,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9662837202888754e-05,
+      "loss": 0.9103,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9654780369732216e-05,
+      "loss": 0.7689,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.964662907598186e-05,
+      "loss": 0.654,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9638383352868e-05,
+      "loss": 1.0917,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9630043231982716e-05,
+      "loss": 0.9325,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9621608745279755e-05,
+      "loss": 0.9688,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961307992507443e-05,
+      "loss": 0.8932,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960445680404345e-05,
+      "loss": 0.8954,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9595739415224845e-05,
+      "loss": 0.9145,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.95869277920178e-05,
+      "loss": 0.9638,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9578021968182555e-05,
+      "loss": 0.8114,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.956902197784025e-05,
+      "loss": 0.9008,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.955992785547282e-05,
+      "loss": 0.8579,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.955073963592284e-05,
+      "loss": 0.9713,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.954145735439343e-05,
+      "loss": 0.7866,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.953208104644804e-05,
+      "loss": 0.8007,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.952261074801043e-05,
+      "loss": 0.7086,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9513046495364415e-05,
+      "loss": 0.9219,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950338832515381e-05,
+      "loss": 0.8546,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.949363627438224e-05,
+      "loss": 0.6465,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.948379038041304e-05,
+      "loss": 0.9602,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.947385068096907e-05,
+      "loss": 0.8576,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9463817214132604e-05,
+      "loss": 0.9854,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9453690018345144e-05,
+      "loss": 0.8981,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9443469132407334e-05,
+      "loss": 0.9761,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9433154595478745e-05,
+      "loss": 0.5922,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942274644707778e-05,
+      "loss": 0.8056,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.941224472708149e-05,
+      "loss": 0.711,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.940164947572542e-05,
+      "loss": 0.6398,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.939096073360348e-05,
+      "loss": 1.0141,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.938017854166775e-05,
+      "loss": 0.6416,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.936930294122838e-05,
+      "loss": 0.6088,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.935833397395336e-05,
+      "loss": 0.9431,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.934727168186844e-05,
+      "loss": 0.8551,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9336116107356914e-05,
+      "loss": 0.7137,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.932486729315945e-05,
+      "loss": 0.8235,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9313525282373974e-05,
+      "loss": 0.822,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.930209011845546e-05,
+      "loss": 0.9173,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.92905618452158e-05,
+      "loss": 1.1054,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927894050682361e-05,
+      "loss": 0.8443,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926722614780408e-05,
+      "loss": 0.6809,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9255418813038756e-05,
+      "loss": 0.9954,
+      "step": 625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9243518547765446e-05,
+      "loss": 0.745,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923152539757799e-05,
+      "loss": 0.9756,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.921943940842611e-05,
+      "loss": 0.606,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9207260626615204e-05,
+      "loss": 1.1881,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9194989098806206e-05,
+      "loss": 1.0505,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.918262487201538e-05,
+      "loss": 0.6874,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9170167993614156e-05,
+      "loss": 0.9532,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.915761851132896e-05,
+      "loss": 0.9158,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.914497647324098e-05,
+      "loss": 1.0318,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9132241927786035e-05,
+      "loss": 0.6542,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.911941492375437e-05,
+      "loss": 0.8846,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9106495510290474e-05,
+      "loss": 0.9366,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909348373689287e-05,
+      "loss": 0.9427,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908037965341395e-05,
+      "loss": 0.9168,
+      "step": 695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.906718331005979e-05,
+      "loss": 0.754,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.905389475738993e-05,
+      "loss": 1.0087,
+      "step": 705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9040514046317196e-05,
+      "loss": 0.767,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.90270412281075e-05,
+      "loss": 0.6931,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.901347635437966e-05,
+      "loss": 1.0641,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.899981947710518e-05,
+      "loss": 0.831,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.898607064860805e-05,
+      "loss": 0.7105,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8972229921564573e-05,
+      "loss": 0.9568,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.895829734900315e-05,
+      "loss": 1.1616,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.894427298430403e-05,
+      "loss": 0.6351,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.893015688119921e-05,
+      "loss": 0.9329,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8915949093772115e-05,
+      "loss": 0.7473,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8901649676457486e-05,
+      "loss": 0.8513,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.888725868404108e-05,
+      "loss": 0.7557,
+      "step": 765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.887277617165955e-05,
+      "loss": 1.0341,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.885820219480018e-05,
+      "loss": 0.9113,
+      "step": 775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.884353680930067e-05,
+      "loss": 0.7057,
+      "step": 780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8828780071348954e-05,
+      "loss": 1.0549,
+      "step": 785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8813932037482954e-05,
+      "loss": 1.2697,
+      "step": 790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.87989927645904e-05,
+      "loss": 0.6278,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.878396230990856e-05,
+      "loss": 0.6395,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.876884073102407e-05,
+      "loss": 1.0411,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.875362808587267e-05,
+      "loss": 0.7791,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.873832443273904e-05,
+      "loss": 1.0095,
+      "step": 815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.872292983025651e-05,
+      "loss": 0.9415,
+      "step": 820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.870744433740688e-05,
+      "loss": 0.9221,
+      "step": 825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8691868013520184e-05,
+      "loss": 0.626,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8676200918274456e-05,
+      "loss": 1.0457,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.86604431116955e-05,
+      "loss": 0.7705,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.864459465415666e-05,
+      "loss": 0.7309,
+      "step": 845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.862865560637862e-05,
+      "loss": 0.8578,
+      "step": 850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8612626029429124e-05,
+      "loss": 0.8989,
+      "step": 855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8596505984722756e-05,
+      "loss": 0.5861,
+      "step": 860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.858029553402074e-05,
+      "loss": 1.0058,
+      "step": 865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.856399473943064e-05,
+      "loss": 0.7497,
+      "step": 870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.85476036634062e-05,
+      "loss": 0.9263,
+      "step": 875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8531122368747005e-05,
+      "loss": 0.9543,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8514550918598356e-05,
+      "loss": 0.6718,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.849788937645093e-05,
+      "loss": 0.9372,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.848113780614061e-05,
+      "loss": 0.8203,
+      "step": 895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8464296271848155e-05,
+      "loss": 0.9678,
+      "step": 900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8447364838099055e-05,
+      "loss": 0.7796,
+      "step": 905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8430343569763204e-05,
+      "loss": 0.7373,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.841323253205471e-05,
+      "loss": 0.7822,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8396031790531594e-05,
+      "loss": 0.8151,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8378741411095566e-05,
+      "loss": 0.6852,
+      "step": 925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.836136145999177e-05,
+      "loss": 0.7998,
+      "step": 930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8343892003808546e-05,
+      "loss": 0.8387,
+      "step": 935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.832633310947712e-05,
+      "loss": 0.6407,
+      "step": 940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.830868484427141e-05,
+      "loss": 0.8965,
+      "step": 945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8290947275807755e-05,
+      "loss": 1.0357,
+      "step": 950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.827312047204461e-05,
+      "loss": 0.6829,
+      "step": 955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8255204501282354e-05,
+      "loss": 0.8587,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.823719943216297e-05,
+      "loss": 0.8118,
+      "step": 965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.821910533366982e-05,
+      "loss": 0.7906,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8200922275127355e-05,
+      "loss": 0.6105,
+      "step": 975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.818265032620088e-05,
+      "loss": 0.7634,
+      "step": 980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.816428955689625e-05,
+      "loss": 1.2755,
+      "step": 985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.814584003755962e-05,
+      "loss": 0.6199,
+      "step": 990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.812730183887719e-05,
+      "loss": 0.8193,
+      "step": 995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8108675031874914e-05,
+      "loss": 0.8272,
+      "step": 1000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.808995968791824e-05,
+      "loss": 0.5445,
+      "step": 1005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.80711558787118e-05,
+      "loss": 0.5907,
+      "step": 1010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8052263676299206e-05,
+      "loss": 0.7152,
+      "step": 1015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.80332831530627e-05,
+      "loss": 0.907,
+      "step": 1020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.801421438172294e-05,
+      "loss": 0.8288,
+      "step": 1025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.799505743533868e-05,
+      "loss": 1.0975,
+      "step": 1030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.797581238730649e-05,
+      "loss": 0.7785,
+      "step": 1035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7956479311360485e-05,
+      "loss": 0.7222,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793705828157207e-05,
+      "loss": 0.878,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7917549372349616e-05,
+      "loss": 0.8496,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789795265843817e-05,
+      "loss": 0.948,
+      "step": 1055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.787826821491923e-05,
+      "loss": 0.7991,
+      "step": 1060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.785849611721037e-05,
+      "loss": 0.5161,
+      "step": 1065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.783863644106502e-05,
+      "loss": 0.8561,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.781868926257216e-05,
+      "loss": 0.6655,
+      "step": 1075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.779865465815599e-05,
+      "loss": 0.7557,
+      "step": 1080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7778532704575705e-05,
+      "loss": 0.8881,
+      "step": 1085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775832347892514e-05,
+      "loss": 0.7095,
+      "step": 1090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77380270586325e-05,
+      "loss": 0.5907,
+      "step": 1095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771764352146005e-05,
+      "loss": 0.6042,
+      "step": 1100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769717294550386e-05,
+      "loss": 0.9914,
+      "step": 1105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7676615409193456e-05,
+      "loss": 0.7606,
+      "step": 1110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765597099129153e-05,
+      "loss": 0.7185,
+      "step": 1115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.763523977089366e-05,
+      "loss": 0.99,
+      "step": 1120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.761442182742799e-05,
+      "loss": 0.7481,
+      "step": 1125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7593517240654926e-05,
+      "loss": 0.9451,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.757252609066683e-05,
+      "loss": 0.9555,
+      "step": 1135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.755144845788772e-05,
+      "loss": 1.0907,
+      "step": 1140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7530284423072965e-05,
+      "loss": 0.9622,
+      "step": 1145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7509034067308946e-05,
+      "loss": 0.9392,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.74876974720128e-05,
+      "loss": 0.727,
+      "step": 1155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.746627471893205e-05,
+      "loss": 0.9508,
+      "step": 1160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744476589014433e-05,
+      "loss": 0.6585,
+      "step": 1165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7423171068057055e-05,
+      "loss": 0.6734,
+      "step": 1170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7401490335407106e-05,
+      "loss": 0.8456,
+      "step": 1175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.737972377526053e-05,
+      "loss": 0.9227,
+      "step": 1180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7357871471012196e-05,
+      "loss": 0.7514,
+      "step": 1185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7335933506385467e-05,
+      "loss": 1.1014,
+      "step": 1190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7313909965431944e-05,
+      "loss": 0.7133,
+      "step": 1195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7291800932531064e-05,
+      "loss": 0.9482,
+      "step": 1200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.726960649238983e-05,
+      "loss": 0.8665,
+      "step": 1205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7247326730042446e-05,
+      "loss": 0.6634,
+      "step": 1210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7224961730850035e-05,
+      "loss": 0.9787,
+      "step": 1215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.72025115805003e-05,
+      "loss": 0.7553,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.717997636500715e-05,
+      "loss": 0.9283,
+      "step": 1225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7157356170710434e-05,
+      "loss": 0.8903,
+      "step": 1230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7134651084275584e-05,
+      "loss": 1.0312,
+      "step": 1235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.711186119269325e-05,
+      "loss": 0.8682,
+      "step": 1240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.708898658327905e-05,
+      "loss": 0.5847,
+      "step": 1245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7066027343673136e-05,
+      "loss": 0.8099,
+      "step": 1250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.704298356183994e-05,
+      "loss": 0.7777,
+      "step": 1255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7019855326067776e-05,
+      "loss": 0.989,
+      "step": 1260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6996642724968545e-05,
+      "loss": 0.8598,
+      "step": 1265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.697334584747739e-05,
+      "loss": 0.7962,
+      "step": 1270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.694996478285231e-05,
+      "loss": 0.6869,
+      "step": 1275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.692649962067389e-05,
+      "loss": 0.7791,
+      "step": 1280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.69029504508449e-05,
+      "loss": 0.5155,
+      "step": 1285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.687931736358997e-05,
+      "loss": 0.889,
+      "step": 1290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.685560044945526e-05,
+      "loss": 0.6497,
+      "step": 1295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6831799799308074e-05,
+      "loss": 1.0463,
+      "step": 1300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.680791550433657e-05,
+      "loss": 0.7195,
+      "step": 1305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6783947656049345e-05,
+      "loss": 0.9367,
+      "step": 1310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.675989634627513e-05,
+      "loss": 0.869,
+      "step": 1315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6735761667162425e-05,
+      "loss": 0.9682,
+      "step": 1320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6711543711179155e-05,
+      "loss": 0.8139,
+      "step": 1325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.668724257111229e-05,
+      "loss": 1.0568,
+      "step": 1330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6662858340067494e-05,
+      "loss": 0.707,
+      "step": 1335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6638391111468824e-05,
+      "loss": 0.9196,
+      "step": 1340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6613840979058286e-05,
+      "loss": 0.745,
+      "step": 1345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6589208036895535e-05,
+      "loss": 0.783,
+      "step": 1350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.656449237935749e-05,
+      "loss": 0.8856,
+      "step": 1355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6539694101138e-05,
+      "loss": 0.8025,
+      "step": 1360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.651481329724743e-05,
+      "loss": 0.9026,
+      "step": 1365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.648985006301236e-05,
+      "loss": 0.8886,
+      "step": 1370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.646480449407516e-05,
+      "loss": 0.6863,
+      "step": 1375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.643967668639367e-05,
+      "loss": 0.8155,
+      "step": 1380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.64144667362408e-05,
+      "loss": 0.9133,
+      "step": 1385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.638917474020419e-05,
+      "loss": 0.6654,
+      "step": 1390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.636380079518582e-05,
+      "loss": 1.2317,
+      "step": 1395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6338344998401635e-05,
+      "loss": 0.7528,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.631280744738118e-05,
+      "loss": 0.7146,
+      "step": 1405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.628718823996725e-05,
+      "loss": 0.8925,
+      "step": 1410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.626148747431547e-05,
+      "loss": 0.9016,
+      "step": 1415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.623570524889393e-05,
+      "loss": 0.801,
+      "step": 1420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.620984166248288e-05,
+      "loss": 0.7566,
+      "step": 1425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.618389681417421e-05,
+      "loss": 0.8144,
+      "step": 1430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6157870803371196e-05,
+      "loss": 0.6723,
+      "step": 1435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.613176372978808e-05,
+      "loss": 0.9678,
+      "step": 1440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6105575693449665e-05,
+      "loss": 0.565,
+      "step": 1445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.607930679469096e-05,
+      "loss": 1.0188,
+      "step": 1450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.605295713415676e-05,
+      "loss": 0.9695,
+      "step": 1455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.602652681280132e-05,
+      "loss": 0.912,
+      "step": 1460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6000015931887916e-05,
+      "loss": 0.5926,
+      "step": 1465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.597342459298847e-05,
+      "loss": 0.8277,
+      "step": 1470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.594675289798317e-05,
+      "loss": 0.7899,
+      "step": 1475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.592000094906007e-05,
+      "loss": 0.8279,
+      "step": 1480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.589316884871472e-05,
+      "loss": 0.7527,
+      "step": 1485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5866256699749745e-05,
+      "loss": 0.7847,
+      "step": 1490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.583926460527445e-05,
+      "loss": 0.8171,
+      "step": 1495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.581219266870446e-05,
+      "loss": 0.6312,
+      "step": 1500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5785040993761286e-05,
+      "loss": 0.5843,
+      "step": 1505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5757809684471945e-05,
+      "loss": 0.7983,
+      "step": 1510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.573049884516857e-05,
+      "loss": 0.7322,
+      "step": 1515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5703108580488e-05,
+      "loss": 0.7182,
+      "step": 1520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.567563899537135e-05,
+      "loss": 0.6021,
+      "step": 1525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.564809019506368e-05,
+      "loss": 0.9239,
+      "step": 1530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.562046228511351e-05,
+      "loss": 0.8548,
+      "step": 1535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.559275537137248e-05,
+      "loss": 0.8533,
+      "step": 1540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.556496955999492e-05,
+      "loss": 0.9148,
+      "step": 1545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5537104957437436e-05,
+      "loss": 0.7418,
+      "step": 1550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.55091616704585e-05,
+      "loss": 0.8202,
+      "step": 1555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5481139806118065e-05,
+      "loss": 1.0894,
+      "step": 1560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.545303947177714e-05,
+      "loss": 1.0099,
+      "step": 1565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.542486077509735e-05,
+      "loss": 0.5793,
+      "step": 1570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.53966038240406e-05,
+      "loss": 0.9161,
+      "step": 1575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5368268726868574e-05,
+      "loss": 0.5936,
+      "step": 1580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5339855592142376e-05,
+      "loss": 0.7161,
+      "step": 1585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.53113645287221e-05,
+      "loss": 0.6385,
+      "step": 1590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.528279564576639e-05,
+      "loss": 0.8465,
+      "step": 1595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5254149052732075e-05,
+      "loss": 0.8975,
+      "step": 1600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.522542485937369e-05,
+      "loss": 0.5722,
+      "step": 1605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.519662317574309e-05,
+      "loss": 1.3745,
+      "step": 1610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.516774411218904e-05,
+      "loss": 0.9947,
+      "step": 1615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.513878777935672e-05,
+      "loss": 0.4494,
+      "step": 1620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.510975428818743e-05,
+      "loss": 0.6444,
+      "step": 1625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5080643749918026e-05,
+      "loss": 0.6877,
+      "step": 1630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5051456276080584e-05,
+      "loss": 0.8109,
+      "step": 1635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.502219197850194e-05,
+      "loss": 0.8133,
+      "step": 1640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.499285096930327e-05,
+      "loss": 0.5652,
+      "step": 1645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.496343336089965e-05,
+      "loss": 0.947,
+      "step": 1650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.493393926599966e-05,
+      "loss": 0.9642,
+      "step": 1655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.490436879760487e-05,
+      "loss": 0.7962,
+      "step": 1660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.487472206900953e-05,
+      "loss": 0.6196,
+      "step": 1665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4844999193800006e-05,
+      "loss": 0.874,
+      "step": 1670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4815200285854454e-05,
+      "loss": 0.7627,
+      "step": 1675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4785325459342296e-05,
+      "loss": 0.8158,
+      "step": 1680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4755374828723867e-05,
+      "loss": 0.7862,
+      "step": 1685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.47253485087499e-05,
+      "loss": 0.9422,
+      "step": 1690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4695246614461116e-05,
+      "loss": 1.0298,
+      "step": 1695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.466506926118782e-05,
+      "loss": 0.7137,
+      "step": 1700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.463481656454939e-05,
+      "loss": 0.8188,
+      "step": 1705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.460448864045389e-05,
+      "loss": 1.1301,
+      "step": 1710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4574085605097586e-05,
+      "loss": 0.7368,
+      "step": 1715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4543607574964546e-05,
+      "loss": 1.1013,
+      "step": 1720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4513054666826146e-05,
+      "loss": 0.7518,
+      "step": 1725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.448242699774066e-05,
+      "loss": 0.7476,
+      "step": 1730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.445172468505279e-05,
+      "loss": 0.9361,
+      "step": 1735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.44209478463932e-05,
+      "loss": 0.5236,
+      "step": 1740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.439009659967812e-05,
+      "loss": 0.7931,
+      "step": 1745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.435917106310886e-05,
+      "loss": 0.9965,
+      "step": 1750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.432817135517134e-05,
+      "loss": 0.8096,
+      "step": 1755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.429709759463566e-05,
+      "loss": 0.9408,
+      "step": 1760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4265949900555644e-05,
+      "loss": 0.8117,
+      "step": 1765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4234728392268386e-05,
+      "loss": 0.8783,
+      "step": 1770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4203433189393776e-05,
+      "loss": 0.9271,
+      "step": 1775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.417206441183406e-05,
+      "loss": 0.79,
+      "step": 1780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.414062217977336e-05,
+      "loss": 0.798,
+      "step": 1785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4109106613677245e-05,
+      "loss": 0.8537,
+      "step": 1790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4077517834292236e-05,
+      "loss": 0.8744,
+      "step": 1795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4045855962645363e-05,
+      "loss": 0.8867,
+      "step": 1800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.40141211200437e-05,
+      "loss": 0.8562,
+      "step": 1805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.398231342807389e-05,
+      "loss": 0.9521,
+      "step": 1810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3950433008601696e-05,
+      "loss": 0.7434,
+      "step": 1815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.391847998377152e-05,
+      "loss": 0.6395,
+      "step": 1820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.388645447600593e-05,
+      "loss": 0.7408,
+      "step": 1825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.385435660800521e-05,
+      "loss": 0.699,
+      "step": 1830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.382218650274686e-05,
+      "loss": 0.702,
+      "step": 1835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3789944283485196e-05,
+      "loss": 0.7046,
+      "step": 1840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3757630073750764e-05,
+      "loss": 0.4303,
+      "step": 1845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.372524399734997e-05,
+      "loss": 0.8465,
+      "step": 1850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.369278617836455e-05,
+      "loss": 0.8183,
+      "step": 1855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.366025674115112e-05,
+      "loss": 0.8503,
+      "step": 1860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3627655810340665e-05,
+      "loss": 0.8829,
+      "step": 1865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.359498351083811e-05,
+      "loss": 0.8953,
+      "step": 1870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3562239967821805e-05,
+      "loss": 0.7919,
+      "step": 1875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.352942530674307e-05,
+      "loss": 0.9352,
+      "step": 1880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.349653965332567e-05,
+      "loss": 1.0616,
+      "step": 1885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.346358313356542e-05,
+      "loss": 0.6374,
+      "step": 1890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.34305558737296e-05,
+      "loss": 0.9775,
+      "step": 1895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.339745800035652e-05,
+      "loss": 0.8867,
+      "step": 1900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3364289640255065e-05,
+      "loss": 0.7813,
+      "step": 1905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.333105092050416e-05,
+      "loss": 0.5893,
+      "step": 1910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.329774196845229e-05,
+      "loss": 0.7871,
+      "step": 1915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.326436291171704e-05,
+      "loss": 0.757,
+      "step": 1920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3230913878184585e-05,
+      "loss": 0.9673,
+      "step": 1925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.31973949960092e-05,
+      "loss": 0.5539,
+      "step": 1930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.316380639361278e-05,
+      "loss": 0.7614,
+      "step": 1935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.313014819968433e-05,
+      "loss": 0.827,
+      "step": 1940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.30964205431795e-05,
+      "loss": 0.82,
+      "step": 1945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.306262355332006e-05,
+      "loss": 0.8691,
+      "step": 1950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.302875735959342e-05,
+      "loss": 0.6096,
+      "step": 1955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.299482209175214e-05,
+      "loss": 0.6282,
+      "step": 1960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.296081787981341e-05,
+      "loss": 0.955,
+      "step": 1965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2926744854058595e-05,
+      "loss": 1.0863,
+      "step": 1970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2892603145032686e-05,
+      "loss": 0.6423,
+      "step": 1975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.285839288354383e-05,
+      "loss": 1.0148,
+      "step": 1980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.282411420066282e-05,
+      "loss": 0.5349,
+      "step": 1985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.278976722772259e-05,
+      "loss": 0.764,
+      "step": 1990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2755352096317725e-05,
+      "loss": 0.7753,
+      "step": 1995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.272086893830394e-05,
+      "loss": 0.7138,
+      "step": 2000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.268631788579759e-05,
+      "loss": 0.8892,
+      "step": 2005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2651699071175136e-05,
+      "loss": 0.6824,
+      "step": 2010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.261701262707268e-05,
+      "loss": 0.576,
+      "step": 2015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.258225868638544e-05,
+      "loss": 0.8506,
+      "step": 2020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.254743738226721e-05,
+      "loss": 0.7759,
+      "step": 2025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.251254884812988e-05,
+      "loss": 0.9231,
+      "step": 2030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.247759321764296e-05,
+      "loss": 0.7997,
+      "step": 2035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.244257062473297e-05,
+      "loss": 0.7364,
+      "step": 2040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.240748120358302e-05,
+      "loss": 0.7245,
+      "step": 2045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2372325088632256e-05,
+      "loss": 0.8293,
+      "step": 2050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.233710241457535e-05,
+      "loss": 0.9198,
+      "step": 2055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.230181331636197e-05,
+      "loss": 0.8513,
+      "step": 2060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2266457929196304e-05,
+      "loss": 0.745,
+      "step": 2065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.223103638853648e-05,
+      "loss": 0.6924,
+      "step": 2070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.219554883009412e-05,
+      "loss": 1.2518,
+      "step": 2075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.215999538983376e-05,
+      "loss": 0.9245,
+      "step": 2080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.212437620397234e-05,
+      "loss": 0.6159,
+      "step": 2085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2088691408978716e-05,
+      "loss": 0.6667,
+      "step": 2090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.205294114157311e-05,
+      "loss": 0.6892,
+      "step": 2095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.201712553872657e-05,
+      "loss": 0.8092,
+      "step": 2100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.19812447376605e-05,
+      "loss": 0.4467,
+      "step": 2105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.194529887584607e-05,
+      "loss": 0.8628,
+      "step": 2110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.190928809100373e-05,
+      "loss": 0.6929,
+      "step": 2115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1873212521102676e-05,
+      "loss": 0.77,
+      "step": 2120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.183707230436031e-05,
+      "loss": 0.7152,
+      "step": 2125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.180086757924172e-05,
+      "loss": 0.9606,
+      "step": 2130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.176459848445915e-05,
+      "loss": 0.8166,
+      "step": 2135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.172826515897146e-05,
+      "loss": 0.8131,
+      "step": 2140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.16918677419836e-05,
+      "loss": 0.86,
+      "step": 2145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1655406372946076e-05,
+      "loss": 0.9263,
+      "step": 2150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.161888119155442e-05,
+      "loss": 0.8999,
+      "step": 2155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.158229233774864e-05,
+      "loss": 0.7593,
+      "step": 2160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.154563995171271e-05,
+      "loss": 0.9816,
+      "step": 2165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.150892417387401e-05,
+      "loss": 0.8458,
+      "step": 2170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.147214514490278e-05,
+      "loss": 1.0097,
+      "step": 2175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.143530300571161e-05,
+      "loss": 0.8787,
+      "step": 2180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.139839789745489e-05,
+      "loss": 0.9728,
+      "step": 2185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.136142996152827e-05,
+      "loss": 0.727,
+      "step": 2190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.132439933956808e-05,
+      "loss": 1.3063,
+      "step": 2195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.128730617345084e-05,
+      "loss": 0.6355,
+      "step": 2200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1250150605292736e-05,
+      "loss": 0.8068,
+      "step": 2205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1212932777448974e-05,
+      "loss": 0.5114,
+      "step": 2210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.117565283251335e-05,
+      "loss": 0.8544,
+      "step": 2215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.113831091331759e-05,
+      "loss": 0.5678,
+      "step": 2220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.110090716293093e-05,
+      "loss": 0.8468,
+      "step": 2225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.106344172465946e-05,
+      "loss": 0.964,
+      "step": 2230
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.102591474204563e-05,
+      "loss": 0.8267,
+      "step": 2235
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.098832635886769e-05,
+      "loss": 0.8758,
+      "step": 2240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0950676719139126e-05,
+      "loss": 1.008,
+      "step": 2245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0912965967108125e-05,
+      "loss": 0.9074,
+      "step": 2250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.087519424725701e-05,
+      "loss": 0.9586,
+      "step": 2255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0837361704301705e-05,
+      "loss": 0.8597,
+      "step": 2260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.079946848319116e-05,
+      "loss": 0.8619,
+      "step": 2265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.076151472910678e-05,
+      "loss": 0.9419,
+      "step": 2270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.072350058746193e-05,
+      "loss": 0.9813,
+      "step": 2275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.068542620390131e-05,
+      "loss": 0.7419,
+      "step": 2280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.064729172430046e-05,
+      "loss": 1.1764,
+      "step": 2285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.060909729476513e-05,
+      "loss": 0.8532,
+      "step": 2290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.057084306163077e-05,
+      "loss": 1.1432,
+      "step": 2295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.053252917146198e-05,
+      "loss": 0.6805,
+      "step": 2300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.049415577105189e-05,
+      "loss": 0.8166,
+      "step": 2305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.045572300742166e-05,
+      "loss": 0.7244,
+      "step": 2310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.041723102781988e-05,
+      "loss": 0.7924,
+      "step": 2315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0378679979722016e-05,
+      "loss": 0.8019,
+      "step": 2320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.034007001082985e-05,
+      "loss": 0.8452,
+      "step": 2325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0301401269070894e-05,
+      "loss": 0.6971,
+      "step": 2330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.026267390259787e-05,
+      "loss": 0.7494,
+      "step": 2335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0223888059788065e-05,
+      "loss": 0.9382,
+      "step": 2340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.018504388924285e-05,
+      "loss": 0.789,
+      "step": 2345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0146141539787044e-05,
+      "loss": 0.8008,
+      "step": 2350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0107181160468375e-05,
+      "loss": 0.8221,
+      "step": 2355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.006816290055692e-05,
+      "loss": 1.0263,
+      "step": 2360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0029086909544474e-05,
+      "loss": 0.7425,
+      "step": 2365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9989953337144056e-05,
+      "loss": 0.8,
+      "step": 2370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.995076233328929e-05,
+      "loss": 0.6524,
+      "step": 2375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.991151404813384e-05,
+      "loss": 0.81,
+      "step": 2380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.987220863205082e-05,
+      "loss": 0.784,
+      "step": 2385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9832846235632235e-05,
+      "loss": 0.8586,
+      "step": 2390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.979342700968842e-05,
+      "loss": 0.7298,
+      "step": 2395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.975395110524742e-05,
+      "loss": 0.7542,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.971441867355444e-05,
+      "loss": 0.734,
+      "step": 2405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.967482986607128e-05,
+      "loss": 0.5972,
+      "step": 2410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.963518483447569e-05,
+      "loss": 0.971,
+      "step": 2415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9595483730660884e-05,
+      "loss": 1.1673,
+      "step": 2420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9555726706734864e-05,
+      "loss": 0.9685,
+      "step": 2425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.95159139150199e-05,
+      "loss": 0.8432,
+      "step": 2430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9476045508051936e-05,
+      "loss": 0.7548,
+      "step": 2435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.943612163857999e-05,
+      "loss": 0.612,
+      "step": 2440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9396142459565554e-05,
+      "loss": 0.7485,
+      "step": 2445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9356108124182067e-05,
+      "loss": 0.5646,
+      "step": 2450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.931601878581425e-05,
+      "loss": 1.0206,
+      "step": 2455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9275874598057584e-05,
+      "loss": 0.6728,
+      "step": 2460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.92356757147177e-05,
+      "loss": 0.8359,
+      "step": 2465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.919542228980977e-05,
+      "loss": 0.674,
+      "step": 2470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9155114477557933e-05,
+      "loss": 0.7835,
+      "step": 2475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.911475243239471e-05,
+      "loss": 0.6586,
+      "step": 2480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9074336308960416e-05,
+      "loss": 0.8773,
+      "step": 2485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.903386626210255e-05,
+      "loss": 0.7754,
+      "step": 2490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.899334244687519e-05,
+      "loss": 0.7351,
+      "step": 2495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.895276501853844e-05,
+      "loss": 0.7563,
+      "step": 2500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.891213413255782e-05,
+      "loss": 0.6732,
+      "step": 2505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.887144994460365e-05,
+      "loss": 1.0051,
+      "step": 2510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.883071261055047e-05,
+      "loss": 0.9529,
+      "step": 2515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.878992228647642e-05,
+      "loss": 0.7171,
+      "step": 2520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.874907912866271e-05,
+      "loss": 0.6398,
+      "step": 2525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.870818329359294e-05,
+      "loss": 0.9811,
+      "step": 2530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.866723493795252e-05,
+      "loss": 0.8369,
+      "step": 2535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8626234218628134e-05,
+      "loss": 0.8481,
+      "step": 2540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.858518129270704e-05,
+      "loss": 0.7327,
+      "step": 2545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.854407631747654e-05,
+      "loss": 0.8802,
+      "step": 2550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8502919450423335e-05,
+      "loss": 0.8541,
+      "step": 2555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.846171084923298e-05,
+      "loss": 0.8537,
+      "step": 2560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8420450671789195e-05,
+      "loss": 0.7178,
+      "step": 2565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8379139076173334e-05,
+      "loss": 0.6856,
+      "step": 2570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.833777622066374e-05,
+      "loss": 0.7565,
+      "step": 2575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.829636226373515e-05,
+      "loss": 0.8922,
+      "step": 2580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8254897364058105e-05,
+      "loss": 0.8434,
+      "step": 2585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.82133816804983e-05,
+      "loss": 1.0232,
+      "step": 2590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8171815372115994e-05,
+      "loss": 0.7439,
+      "step": 2595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.813019859816544e-05,
+      "loss": 0.8613,
+      "step": 2600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.808853151809422e-05,
+      "loss": 0.633,
+      "step": 2605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.804681429154266e-05,
+      "loss": 0.9164,
+      "step": 2610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.800504707834319e-05,
+      "loss": 0.7818,
+      "step": 2615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7963230038519804e-05,
+      "loss": 0.9187,
+      "step": 2620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.792136333228735e-05,
+      "loss": 0.7655,
+      "step": 2625
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7879447120051013e-05,
+      "loss": 1.0656,
+      "step": 2630
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.78374815624056e-05,
+      "loss": 0.7042,
+      "step": 2635
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7795466820135004e-05,
+      "loss": 0.759,
+      "step": 2640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7753403054211564e-05,
+      "loss": 0.7199,
+      "step": 2645
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.771129042579545e-05,
+      "loss": 0.7872,
+      "step": 2650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7669129096234e-05,
+      "loss": 0.8701,
+      "step": 2655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.762691922706119e-05,
+      "loss": 0.7967,
+      "step": 2660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7584660979996944e-05,
+      "loss": 0.824,
+      "step": 2665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7542354516946555e-05,
+      "loss": 0.645,
+      "step": 2670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.9053,
+      "step": 2675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.745759759143143e-05,
+      "loss": 0.7221,
+      "step": 2680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.741514745369845e-05,
+      "loss": 0.8607,
+      "step": 2685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.737264974944153e-05,
+      "loss": 0.8016,
+      "step": 2690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.733010464148339e-05,
+      "loss": 0.6189,
+      "step": 2695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.728751229282836e-05,
+      "loss": 0.9712,
+      "step": 2700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7244872866661794e-05,
+      "loss": 0.671,
+      "step": 2705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.720218652634938e-05,
+      "loss": 0.8797,
+      "step": 2710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7159453435436565e-05,
+      "loss": 0.8484,
+      "step": 2715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.711667375764792e-05,
+      "loss": 0.7092,
+      "step": 2720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.707384765688649e-05,
+      "loss": 0.9097,
+      "step": 2725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.703097529723318e-05,
+      "loss": 1.0472,
+      "step": 2730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.698805684294616e-05,
+      "loss": 0.5841,
+      "step": 2735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.694509245846017e-05,
+      "loss": 0.8608,
+      "step": 2740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.690208230838594e-05,
+      "loss": 1.0081,
+      "step": 2745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.685902655750952e-05,
+      "loss": 0.736,
+      "step": 2750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.681592537079171e-05,
+      "loss": 0.8658,
+      "step": 2755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6772778913367365e-05,
+      "loss": 1.0521,
+      "step": 2760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.672958735054479e-05,
+      "loss": 0.7203,
+      "step": 2765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6686350847805094e-05,
+      "loss": 0.7263,
+      "step": 2770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.664306957080159e-05,
+      "loss": 0.7969,
+      "step": 2775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6599743685359124e-05,
+      "loss": 0.8942,
+      "step": 2780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.655637335747344e-05,
+      "loss": 0.8132,
+      "step": 2785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.651295875331058e-05,
+      "loss": 0.6737,
+      "step": 2790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.646950003920619e-05,
+      "loss": 0.71,
+      "step": 2795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6425997381664953e-05,
+      "loss": 0.775,
+      "step": 2800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.638245094735988e-05,
+      "loss": 0.668,
+      "step": 2805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.633886090313174e-05,
+      "loss": 0.8109,
+      "step": 2810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6295227415988355e-05,
+      "loss": 0.6153,
+      "step": 2815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6251550653103995e-05,
+      "loss": 0.6732,
+      "step": 2820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6207830781818754e-05,
+      "loss": 0.9431,
+      "step": 2825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6164067969637874e-05,
+      "loss": 0.6841,
+      "step": 2830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.612026238423111e-05,
+      "loss": 0.619,
+      "step": 2835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.607641419343212e-05,
+      "loss": 0.7571,
+      "step": 2840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.603252356523777e-05,
+      "loss": 0.9541,
+      "step": 2845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.598859066780754e-05,
+      "loss": 0.9421,
+      "step": 2850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5944615669462846e-05,
+      "loss": 0.87,
+      "step": 2855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5900598738686396e-05,
+      "loss": 0.6066,
+      "step": 2860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.585654004412158e-05,
+      "loss": 0.8069,
+      "step": 2865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.581243975457178e-05,
+      "loss": 1.0488,
+      "step": 2870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.576829803899976e-05,
+      "loss": 0.8078,
+      "step": 2875
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.572411506652697e-05,
+      "loss": 0.6125,
+      "step": 2880
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.567989100643295e-05,
+      "loss": 0.8246,
+      "step": 2885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.563562602815467e-05,
+      "loss": 0.6961,
+      "step": 2890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5591320301285846e-05,
+      "loss": 0.8546,
+      "step": 2895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.554697399557634e-05,
+      "loss": 0.6219,
+      "step": 2900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5502587280931466e-05,
+      "loss": 0.8551,
+      "step": 2905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.545816032741136e-05,
+      "loss": 1.2027,
+      "step": 2910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.541369330523034e-05,
+      "loss": 0.9114,
+      "step": 2915
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.536918638475623e-05,
+      "loss": 0.9212,
+      "step": 2920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5324639736509714e-05,
+      "loss": 0.8154,
+      "step": 2925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.52800535311637e-05,
+      "loss": 0.6818,
+      "step": 2930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5235427939542625e-05,
+      "loss": 0.6854,
+      "step": 2935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.519076313262187e-05,
+      "loss": 0.8709,
+      "step": 2940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.514605928152702e-05,
+      "loss": 0.4597,
+      "step": 2945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5101316557533294e-05,
+      "loss": 0.9278,
+      "step": 2950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.505653513206479e-05,
+      "loss": 0.8354,
+      "step": 2955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.501171517669394e-05,
+      "loss": 0.9724,
+      "step": 2960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4966856863140775e-05,
+      "loss": 0.7887,
+      "step": 2965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4921960363272264e-05,
+      "loss": 0.8948,
+      "step": 2970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.487702584910172e-05,
+      "loss": 0.868,
+      "step": 2975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.483205349278808e-05,
+      "loss": 0.6986,
+      "step": 2980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.478704346663528e-05,
+      "loss": 0.814,
+      "step": 2985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.474199594309155e-05,
+      "loss": 0.8457,
+      "step": 2990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4696911094748834e-05,
+      "loss": 0.6997,
+      "step": 2995
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4651789094342044e-05,
+      "loss": 0.933,
+      "step": 3000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.460663011474844e-05,
+      "loss": 0.7821,
+      "step": 3005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.456143432898696e-05,
+      "loss": 0.7243,
+      "step": 3010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.451620191021757e-05,
+      "loss": 0.6778,
+      "step": 3015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.447093303174058e-05,
+      "loss": 0.8237,
+      "step": 3020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4425627866996e-05,
+      "loss": 0.6878,
+      "step": 3025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4380286589562845e-05,
+      "loss": 0.7593,
+      "step": 3030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4334909373158495e-05,
+      "loss": 0.8578,
+      "step": 3035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.428949639163805e-05,
+      "loss": 0.6964,
+      "step": 3040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4244047818993603e-05,
+      "loss": 0.8752,
+      "step": 3045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.419856382935362e-05,
+      "loss": 0.6775,
+      "step": 3050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.415304459698228e-05,
+      "loss": 0.6432,
+      "step": 3055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.410749029627876e-05,
+      "loss": 0.8955,
+      "step": 3060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.406190110177661e-05,
+      "loss": 0.6664,
+      "step": 3065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.401627718814307e-05,
+      "loss": 0.7942,
+      "step": 3070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.39706187301784e-05,
+      "loss": 0.6578,
+      "step": 3075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.392492590281521e-05,
+      "loss": 0.7593,
+      "step": 3080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.387919888111778e-05,
+      "loss": 0.6513,
+      "step": 3085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3833437840281405e-05,
+      "loss": 0.8237,
+      "step": 3090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.378764295563174e-05,
+      "loss": 0.6959,
+      "step": 3095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3741814402624095e-05,
+      "loss": 0.6274,
+      "step": 3100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.369595235684274e-05,
+      "loss": 1.111,
+      "step": 3105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.365005699400031e-05,
+      "loss": 0.7869,
+      "step": 3110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.360412848993709e-05,
+      "loss": 0.7997,
+      "step": 3115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.355816702062031e-05,
+      "loss": 0.7065,
+      "step": 3120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.351217276214351e-05,
+      "loss": 0.8501,
+      "step": 3125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.346614589072587e-05,
+      "loss": 0.8161,
+      "step": 3130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.342008658271151e-05,
+      "loss": 0.678,
+      "step": 3135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.337399501456883e-05,
+      "loss": 0.5893,
+      "step": 3140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3327871362889817e-05,
+      "loss": 0.8223,
+      "step": 3145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3281715804389403e-05,
+      "loss": 0.6704,
+      "step": 3150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.323552851590475e-05,
+      "loss": 1.014,
+      "step": 3155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3189309674394584e-05,
+      "loss": 0.9535,
+      "step": 3160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.314305945693853e-05,
+      "loss": 1.0044,
+      "step": 3165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3096778040736435e-05,
+      "loss": 0.6495,
+      "step": 3170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3050465603107654e-05,
+      "loss": 0.5793,
+      "step": 3175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.300412232149041e-05,
+      "loss": 1.0168,
+      "step": 3180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.295774837344109e-05,
+      "loss": 1.1602,
+      "step": 3185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.29113439366336e-05,
+      "loss": 0.7706,
+      "step": 3190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.286490918885863e-05,
+      "loss": 0.7749,
+      "step": 3195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2818444308023e-05,
+      "loss": 1.0041,
+      "step": 3200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2771949472149004e-05,
+      "loss": 0.7467,
+      "step": 3205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.272542485937369e-05,
+      "loss": 0.6231,
+      "step": 3210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.267887064794818e-05,
+      "loss": 0.8587,
+      "step": 3215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2632287016237026e-05,
+      "loss": 0.805,
+      "step": 3220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.258567414271748e-05,
+      "loss": 0.7181,
+      "step": 3225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2539032205978825e-05,
+      "loss": 0.8973,
+      "step": 3230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2492361384721695e-05,
+      "loss": 0.7855,
+      "step": 3235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.244566185775741e-05,
+      "loss": 0.9966,
+      "step": 3240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2398933804007254e-05,
+      "loss": 0.9123,
+      "step": 3245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.235217740250181e-05,
+      "loss": 0.6088,
+      "step": 3250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.230539283238027e-05,
+      "loss": 0.7805,
+      "step": 3255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.225858027288973e-05,
+      "loss": 0.5555,
+      "step": 3260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.221173990338458e-05,
+      "loss": 0.7677,
+      "step": 3265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.21648719033257e-05,
+      "loss": 0.8128,
+      "step": 3270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.211797645227985e-05,
+      "loss": 0.8701,
+      "step": 3275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2071053729918984e-05,
+      "loss": 0.8031,
+      "step": 3280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2024103916019504e-05,
+      "loss": 0.7645,
+      "step": 3285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.197712719046165e-05,
+      "loss": 0.8307,
+      "step": 3290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.193012373322873e-05,
+      "loss": 0.9655,
+      "step": 3295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.18830937244065e-05,
+      "loss": 0.8069,
+      "step": 3300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.183603734418243e-05,
+      "loss": 0.8416,
+      "step": 3305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.178895477284503e-05,
+      "loss": 0.8313,
+      "step": 3310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1741846190783154e-05,
+      "loss": 0.6876,
+      "step": 3315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.169471177848531e-05,
+      "loss": 0.9535,
+      "step": 3320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1647551716539e-05,
+      "loss": 0.9347,
+      "step": 3325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.160036618562995e-05,
+      "loss": 0.6052,
+      "step": 3330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.155315536654148e-05,
+      "loss": 0.7152,
+      "step": 3335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.150591944015383e-05,
+      "loss": 1.0406,
+      "step": 3340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.14586585874434e-05,
+      "loss": 0.343,
+      "step": 3345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.14113729894821e-05,
+      "loss": 0.881,
+      "step": 3350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1364062827436646e-05,
+      "loss": 0.9875,
+      "step": 3355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.131672828256788e-05,
+      "loss": 0.7441,
+      "step": 3360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.126936953623005e-05,
+      "loss": 0.5469,
+      "step": 3365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.122198676987012e-05,
+      "loss": 0.7694,
+      "step": 3370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.117458016502711e-05,
+      "loss": 0.8247,
+      "step": 3375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.112714990333134e-05,
+      "loss": 0.8477,
+      "step": 3380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.107969616650378e-05,
+      "loss": 0.7026,
+      "step": 3385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.103221913635537e-05,
+      "loss": 0.4747,
+      "step": 3390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.098471899478624e-05,
+      "loss": 0.9731,
+      "step": 3395
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.093719592378512e-05,
+      "loss": 0.7526,
+      "step": 3400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.088965010542855e-05,
+      "loss": 0.7893,
+      "step": 3405
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.084208172188024e-05,
+      "loss": 0.7426,
+      "step": 3410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0794490955390366e-05,
+      "loss": 0.8126,
+      "step": 3415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.074687798829483e-05,
+      "loss": 0.6938,
+      "step": 3420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.069924300301463e-05,
+      "loss": 0.6317,
+      "step": 3425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.065158618205508e-05,
+      "loss": 0.7799,
+      "step": 3430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.06039077080052e-05,
+      "loss": 0.8339,
+      "step": 3435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.055620776353693e-05,
+      "loss": 0.8935,
+      "step": 3440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.050848653140449e-05,
+      "loss": 0.7641,
+      "step": 3445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.046074419444366e-05,
+      "loss": 0.8914,
+      "step": 3450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0412980935571082e-05,
+      "loss": 0.8372,
+      "step": 3455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0365196937783536e-05,
+      "loss": 0.8092,
+      "step": 3460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0317392384157282e-05,
+      "loss": 1.2656,
+      "step": 3465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.026956745784734e-05,
+      "loss": 0.8632,
+      "step": 3470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.022172234208676e-05,
+      "loss": 0.8376,
+      "step": 3475
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0173857220185962e-05,
+      "loss": 1.1093,
+      "step": 3480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.012597227553201e-05,
+      "loss": 0.8214,
+      "step": 3485
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.007806769158793e-05,
+      "loss": 0.9473,
+      "step": 3490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0030143651891962e-05,
+      "loss": 0.7844,
+      "step": 3495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9982200340056914e-05,
+      "loss": 0.62,
+      "step": 3500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.993423793976942e-05,
+      "loss": 0.8127,
+      "step": 3505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.988625663478926e-05,
+      "loss": 0.7885,
+      "step": 3510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9838256608948633e-05,
+      "loss": 0.6595,
+      "step": 3515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.979023804615146e-05,
+      "loss": 0.8267,
+      "step": 3520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9742201130372693e-05,
+      "loss": 0.8036,
+      "step": 3525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9694146045657607e-05,
+      "loss": 0.4342,
+      "step": 3530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9646072976121057e-05,
+      "loss": 0.5788,
+      "step": 3535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.959798210594684e-05,
+      "loss": 0.5321,
+      "step": 3540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9549873619386947e-05,
+      "loss": 0.7773,
+      "step": 3545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9501747700760834e-05,
+      "loss": 0.781,
+      "step": 3550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9453604534454777e-05,
+      "loss": 0.5729,
+      "step": 3555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9405444304921116e-05,
+      "loss": 0.7762,
+      "step": 3560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9357267196677578e-05,
+      "loss": 0.7822,
+      "step": 3565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9309073394306552e-05,
+      "loss": 0.8035,
+      "step": 3570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9260863082454377e-05,
+      "loss": 0.7845,
+      "step": 3575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9212636445830666e-05,
+      "loss": 0.8671,
+      "step": 3580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9164393669207564e-05,
+      "loss": 0.8715,
+      "step": 3585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9116134937419054e-05,
+      "loss": 0.813,
+      "step": 3590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9067860435360255e-05,
+      "loss": 0.9813,
+      "step": 3595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.901957034798671e-05,
+      "loss": 0.7841,
+      "step": 3600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8971264860313658e-05,
+      "loss": 0.6849,
+      "step": 3605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8922944157415345e-05,
+      "loss": 0.9006,
+      "step": 3610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8874608424424332e-05,
+      "loss": 0.8561,
+      "step": 3615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8826257846530753e-05,
+      "loss": 0.8382,
+      "step": 3620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8777892608981604e-05,
+      "loss": 0.6765,
+      "step": 3625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8729512897080068e-05,
+      "loss": 0.5332,
+      "step": 3630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8681118896184766e-05,
+      "loss": 0.5901,
+      "step": 3635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8632710791709087e-05,
+      "loss": 0.5963,
+      "step": 3640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.858428876912043e-05,
+      "loss": 0.8911,
+      "step": 3645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.853585301393954e-05,
+      "loss": 0.7424,
+      "step": 3650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8487403711739762e-05,
+      "loss": 0.7129,
+      "step": 3655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8438941048146358e-05,
+      "loss": 0.5907,
+      "step": 3660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8390465208835752e-05,
+      "loss": 0.7695,
+      "step": 3665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8341976379534896e-05,
+      "loss": 0.8346,
+      "step": 3670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8293474746020472e-05,
+      "loss": 0.4808,
+      "step": 3675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8244960494118234e-05,
+      "loss": 1.0394,
+      "step": 3680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8196433809702278e-05,
+      "loss": 1.0275,
+      "step": 3685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8147894878694326e-05,
+      "loss": 0.7949,
+      "step": 3690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.809934388706305e-05,
+      "loss": 0.8894,
+      "step": 3695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8050781020823296e-05,
+      "loss": 1.089,
+      "step": 3700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8002206466035418e-05,
+      "loss": 0.7988,
+      "step": 3705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7953620408804552e-05,
+      "loss": 0.737,
+      "step": 3710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.790502303527991e-05,
+      "loss": 0.7304,
+      "step": 3715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.785641453165405e-05,
+      "loss": 0.8633,
+      "step": 3720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.780779508416219e-05,
+      "loss": 0.8069,
+      "step": 3725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7759164879081457e-05,
+      "loss": 0.8406,
+      "step": 3730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7710524102730206e-05,
+      "loss": 0.7291,
+      "step": 3735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7661872941467282e-05,
+      "loss": 0.9062,
+      "step": 3740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.761321158169134e-05,
+      "loss": 0.7312,
+      "step": 3745
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.756454020984009e-05,
+      "loss": 0.5763,
+      "step": 3750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.751585901238961e-05,
+      "loss": 0.8981,
+      "step": 3755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.746716817585362e-05,
+      "loss": 0.8761,
+      "step": 3760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.741846788678276e-05,
+      "loss": 0.57,
+      "step": 3765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7369758331763922e-05,
+      "loss": 0.7965,
+      "step": 3770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7321039697419453e-05,
+      "loss": 0.6454,
+      "step": 3775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.727231217040651e-05,
+      "loss": 0.8082,
+      "step": 3780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7223575937416323e-05,
+      "loss": 0.7885,
+      "step": 3785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7174831185173465e-05,
+      "loss": 0.9693,
+      "step": 3790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7126078100435164e-05,
+      "loss": 0.7502,
+      "step": 3795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.707731686999056e-05,
+      "loss": 0.7998,
+      "step": 3800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.702854768066001e-05,
+      "loss": 0.8119,
+      "step": 3805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6979770719294355e-05,
+      "loss": 0.8713,
+      "step": 3810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6930986172774218e-05,
+      "loss": 0.6602,
+      "step": 3815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6882194228009287e-05,
+      "loss": 0.5356,
+      "step": 3820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.68333950719376e-05,
+      "loss": 0.5311,
+      "step": 3825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6784588891524808e-05,
+      "loss": 0.8653,
+      "step": 3830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.673577587376348e-05,
+      "loss": 0.806,
+      "step": 3835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.668695620567238e-05,
+      "loss": 0.9564,
+      "step": 3840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6638130074295774e-05,
+      "loss": 0.5714,
+      "step": 3845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6589297666702657e-05,
+      "loss": 1.1385,
+      "step": 3850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6540459169986088e-05,
+      "loss": 0.851,
+      "step": 3855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6491614771262452e-05,
+      "loss": 0.762,
+      "step": 3860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6442764657670748e-05,
+      "loss": 0.7061,
+      "step": 3865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6393909016371876e-05,
+      "loss": 0.7791,
+      "step": 3870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6345048034547896e-05,
+      "loss": 0.6108,
+      "step": 3875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.629618189940136e-05,
+      "loss": 0.8025,
+      "step": 3880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.624731079815454e-05,
+      "loss": 0.6989,
+      "step": 3885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.619843491804872e-05,
+      "loss": 0.834,
+      "step": 3890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.614955444634354e-05,
+      "loss": 0.7756,
+      "step": 3895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6100669570316195e-05,
+      "loss": 0.7541,
+      "step": 3900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6051780477260774e-05,
+      "loss": 0.7013,
+      "step": 3905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6002887354487502e-05,
+      "loss": 0.6681,
+      "step": 3910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5953990389322057e-05,
+      "loss": 0.7657,
+      "step": 3915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5905089769104853e-05,
+      "loss": 0.7957,
+      "step": 3920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.585618568119027e-05,
+      "loss": 0.6126,
+      "step": 3925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5807278312946005e-05,
+      "loss": 0.8527,
+      "step": 3930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5758367851752314e-05,
+      "loss": 0.7145,
+      "step": 3935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.57094544850013e-05,
+      "loss": 0.8151,
+      "step": 3940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5660538400096206e-05,
+      "loss": 0.6758,
+      "step": 3945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.561161978445068e-05,
+      "loss": 0.6533,
+      "step": 3950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5562698825488074e-05,
+      "loss": 0.883,
+      "step": 3955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5513775710640715e-05,
+      "loss": 0.622,
+      "step": 3960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.546485062734917e-05,
+      "loss": 0.7915,
+      "step": 3965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.541592376306159e-05,
+      "loss": 0.8109,
+      "step": 3970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5366995305232916e-05,
+      "loss": 0.8439,
+      "step": 3975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.531806544132421e-05,
+      "loss": 0.8517,
+      "step": 3980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5269134358801914e-05,
+      "loss": 0.9442,
+      "step": 3985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.522020224513713e-05,
+      "loss": 0.6621,
+      "step": 3990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5171269287804946e-05,
+      "loss": 0.8607,
+      "step": 3995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5122335674283624e-05,
+      "loss": 0.9728,
+      "step": 4000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5073401592054003e-05,
+      "loss": 0.9036,
+      "step": 4005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5024467228598675e-05,
+      "loss": 0.784,
+      "step": 4010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4975532771401334e-05,
+      "loss": 0.8246,
+      "step": 4015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4926598407946e-05,
+      "loss": 0.9255,
+      "step": 4020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.487766432571638e-05,
+      "loss": 0.9835,
+      "step": 4025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4828730712195064e-05,
+      "loss": 1.0478,
+      "step": 4030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4779797754862872e-05,
+      "loss": 0.7794,
+      "step": 4035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4730865641198098e-05,
+      "loss": 0.8264,
+      "step": 4040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4681934558675794e-05,
+      "loss": 0.529,
+      "step": 4045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.463300469476709e-05,
+      "loss": 0.5114,
+      "step": 4050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4584076236938417e-05,
+      "loss": 0.9952,
+      "step": 4055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4535149372650833e-05,
+      "loss": 0.7057,
+      "step": 4060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.44862242893593e-05,
+      "loss": 0.7126,
+      "step": 4065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4437301174511935e-05,
+      "loss": 0.6543,
+      "step": 4070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.438838021554933e-05,
+      "loss": 0.7542,
+      "step": 4075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.43394615999038e-05,
+      "loss": 0.8153,
+      "step": 4080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4290545514998702e-05,
+      "loss": 0.6987,
+      "step": 4085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.42416321482477e-05,
+      "loss": 0.7511,
+      "step": 4090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4192721687054e-05,
+      "loss": 0.5062,
+      "step": 4095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4143814318809743e-05,
+      "loss": 0.688,
+      "step": 4100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4094910230895157e-05,
+      "loss": 0.6306,
+      "step": 4105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4046009610677945e-05,
+      "loss": 0.8585,
+      "step": 4110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3997112645512507e-05,
+      "loss": 0.592,
+      "step": 4115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.394821952273923e-05,
+      "loss": 0.7932,
+      "step": 4120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3899330429683807e-05,
+      "loss": 0.9064,
+      "step": 4125
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3850445553656463e-05,
+      "loss": 0.715,
+      "step": 4130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3801565081951278e-05,
+      "loss": 0.8611,
+      "step": 4135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3752689201845475e-05,
+      "loss": 1.1142,
+      "step": 4140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3703818100598645e-05,
+      "loss": 0.7474,
+      "step": 4145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.365495196545211e-05,
+      "loss": 0.798,
+      "step": 4150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3606090983628133e-05,
+      "loss": 0.9182,
+      "step": 4155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3557235342329258e-05,
+      "loss": 1.1456,
+      "step": 4160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.350838522873756e-05,
+      "loss": 0.7397,
+      "step": 4165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.345954083001392e-05,
+      "loss": 0.9966,
+      "step": 4170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3410702333297356e-05,
+      "loss": 0.667,
+      "step": 4175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.336186992570423e-05,
+      "loss": 0.5898,
+      "step": 4180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.331304379432762e-05,
+      "loss": 0.6216,
+      "step": 4185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.326422412623653e-05,
+      "loss": 0.9163,
+      "step": 4190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3215411108475198e-05,
+      "loss": 0.8816,
+      "step": 4195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3166604928062406e-05,
+      "loss": 0.7659,
+      "step": 4200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3117805771990715e-05,
+      "loss": 0.7107,
+      "step": 4205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3069013827225784e-05,
+      "loss": 0.6851,
+      "step": 4210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.302022928070565e-05,
+      "loss": 0.8245,
+      "step": 4215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2971452319339997e-05,
+      "loss": 0.8563,
+      "step": 4220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2922683130009447e-05,
+      "loss": 0.6902,
+      "step": 4225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2873921899564842e-05,
+      "loss": 0.8544,
+      "step": 4230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2825168814826537e-05,
+      "loss": 0.5357,
+      "step": 4235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2776424062583687e-05,
+      "loss": 0.8008,
+      "step": 4240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2727687829593495e-05,
+      "loss": 0.691,
+      "step": 4245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.267896030258056e-05,
+      "loss": 0.7815,
+      "step": 4250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2630241668236084e-05,
+      "loss": 0.7295,
+      "step": 4255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2581532113217237e-05,
+      "loss": 0.9866,
+      "step": 4260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2532831824146392e-05,
+      "loss": 0.64,
+      "step": 4265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2484140987610394e-05,
+      "loss": 0.7155,
+      "step": 4270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.243545979015992e-05,
+      "loss": 0.6957,
+      "step": 4275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.238678841830867e-05,
+      "loss": 1.0126,
+      "step": 4280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.233812705853272e-05,
+      "loss": 0.6216,
+      "step": 4285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2289475897269806e-05,
+      "loss": 0.6242,
+      "step": 4290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2240835120918552e-05,
+      "loss": 0.9682,
+      "step": 4295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.219220491583782e-05,
+      "loss": 0.929,
+      "step": 4300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.214358546834595e-05,
+      "loss": 0.5408,
+      "step": 4305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2094976964720097e-05,
+      "loss": 0.8829,
+      "step": 4310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.204637959119546e-05,
+      "loss": 0.7236,
+      "step": 4315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1997793533964588e-05,
+      "loss": 0.8631,
+      "step": 4320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1949218979176717e-05,
+      "loss": 0.6539,
+      "step": 4325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.190065611293695e-05,
+      "loss": 0.8994,
+      "step": 4330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1852105121305673e-05,
+      "loss": 1.079,
+      "step": 4335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1803566190297735e-05,
+      "loss": 0.9701,
+      "step": 4340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.175503950588177e-05,
+      "loss": 0.658,
+      "step": 4345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1706525253979534e-05,
+      "loss": 0.7602,
+      "step": 4350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.165802362046511e-05,
+      "loss": 0.683,
+      "step": 4355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1609534791164247e-05,
+      "loss": 0.5466,
+      "step": 4360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.156105895185365e-05,
+      "loss": 0.5408,
+      "step": 4365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1512596288260244e-05,
+      "loss": 0.8613,
+      "step": 4370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1464146986060472e-05,
+      "loss": 0.9278,
+      "step": 4375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1415711230879574e-05,
+      "loss": 0.7148,
+      "step": 4380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.136728920829092e-05,
+      "loss": 1.0878,
+      "step": 4385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1318881103815243e-05,
+      "loss": 0.6247,
+      "step": 4390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1270487102919938e-05,
+      "loss": 0.7378,
+      "step": 4395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.12221073910184e-05,
+      "loss": 0.6627,
+      "step": 4400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.117374215346925e-05,
+      "loss": 0.8013,
+      "step": 4405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1125391575575667e-05,
+      "loss": 0.7353,
+      "step": 4410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.107705584258466e-05,
+      "loss": 0.9524,
+      "step": 4415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1028735139686348e-05,
+      "loss": 0.5429,
+      "step": 4420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0980429652013297e-05,
+      "loss": 0.5861,
+      "step": 4425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0932139564639748e-05,
+      "loss": 0.9457,
+      "step": 4430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.088386506258094e-05,
+      "loss": 0.7481,
+      "step": 4435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.083560633079244e-05,
+      "loss": 0.8164,
+      "step": 4440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.078736355416934e-05,
+      "loss": 0.7185,
+      "step": 4445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0739136917545636e-05,
+      "loss": 0.8923,
+      "step": 4450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0690926605693457e-05,
+      "loss": 0.6084,
+      "step": 4455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0642732803322428e-05,
+      "loss": 0.7326,
+      "step": 4460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0594555695078894e-05,
+      "loss": 0.8347,
+      "step": 4465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0546395465545225e-05,
+      "loss": 0.7387,
+      "step": 4470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0498252299239175e-05,
+      "loss": 0.9871,
+      "step": 4475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0450126380613056e-05,
+      "loss": 0.6296,
+      "step": 4480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0402017894053154e-05,
+      "loss": 0.7955,
+      "step": 4485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0353927023878945e-05,
+      "loss": 0.7652,
+      "step": 4490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0305853954342396e-05,
+      "loss": 0.6206,
+      "step": 4495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.025779886962731e-05,
+      "loss": 0.6128,
+      "step": 4500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0209761953848543e-05,
+      "loss": 0.7177,
+      "step": 4505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.016174339105137e-05,
+      "loss": 0.745,
+      "step": 4510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0113743365210745e-05,
+      "loss": 0.6405,
+      "step": 4515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.006576206023058e-05,
+      "loss": 0.7295,
+      "step": 4520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.00177996599431e-05,
+      "loss": 0.5965,
+      "step": 4525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9969856348108047e-05,
+      "loss": 0.7794,
+      "step": 4530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.992193230841208e-05,
+      "loss": 0.9054,
+      "step": 4535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9874027724467995e-05,
+      "loss": 0.5439,
+      "step": 4540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.982614277981404e-05,
+      "loss": 0.9104,
+      "step": 4545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9778277657913245e-05,
+      "loss": 0.6782,
+      "step": 4550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9730432542152665e-05,
+      "loss": 0.9321,
+      "step": 4555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9682607615842717e-05,
+      "loss": 0.7531,
+      "step": 4560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9634803062216473e-05,
+      "loss": 0.6802,
+      "step": 4565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9587019064428923e-05,
+      "loss": 0.8034,
+      "step": 4570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9539255805556344e-05,
+      "loss": 0.7508,
+      "step": 4575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9491513468595516e-05,
+      "loss": 0.6036,
+      "step": 4580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.944379223646307e-05,
+      "loss": 0.8243,
+      "step": 4585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.939609229199481e-05,
+      "loss": 0.8349,
+      "step": 4590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9348413817944923e-05,
+      "loss": 0.7201,
+      "step": 4595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.930075699698538e-05,
+      "loss": 0.925,
+      "step": 4600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9253122011705173e-05,
+      "loss": 0.7956,
+      "step": 4605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9205509044609636e-05,
+      "loss": 0.8727,
+      "step": 4610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9157918278119762e-05,
+      "loss": 1.0087,
+      "step": 4615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.911034989457145e-05,
+      "loss": 0.588,
+      "step": 4620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9062804076214887e-05,
+      "loss": 0.8992,
+      "step": 4625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.901528100521376e-05,
+      "loss": 0.705,
+      "step": 4630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8967780863644632e-05,
+      "loss": 0.7779,
+      "step": 4635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8920303833496223e-05,
+      "loss": 0.8984,
+      "step": 4640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8872850096668665e-05,
+      "loss": 0.7831,
+      "step": 4645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8825419834972902e-05,
+      "loss": 0.8313,
+      "step": 4650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8778013230129884e-05,
+      "loss": 0.9537,
+      "step": 4655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.873063046376995e-05,
+      "loss": 0.847,
+      "step": 4660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8683271717432124e-05,
+      "loss": 0.6222,
+      "step": 4665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8635937172563356e-05,
+      "loss": 0.8153,
+      "step": 4670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.858862701051791e-05,
+      "loss": 0.9213,
+      "step": 4675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8541341412556606e-05,
+      "loss": 0.9756,
+      "step": 4680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8494080559846174e-05,
+      "loss": 1.0557,
+      "step": 4685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8446844633458527e-05,
+      "loss": 0.8338,
+      "step": 4690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.839963381437006e-05,
+      "loss": 0.9507,
+      "step": 4695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.835244828346101e-05,
+      "loss": 0.786,
+      "step": 4700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8305288221514687e-05,
+      "loss": 0.9817,
+      "step": 4705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.825815380921685e-05,
+      "loss": 0.7714,
+      "step": 4710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.821104522715498e-05,
+      "loss": 1.0587,
+      "step": 4715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8163962655817574e-05,
+      "loss": 0.9325,
+      "step": 4720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.811690627559351e-05,
+      "loss": 0.9193,
+      "step": 4725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8069876266771277e-05,
+      "loss": 0.5623,
+      "step": 4730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8022872809538356e-05,
+      "loss": 0.835,
+      "step": 4735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.79758960839805e-05,
+      "loss": 0.8083,
+      "step": 4740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7928946270081022e-05,
+      "loss": 0.4305,
+      "step": 4745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7882023547720156e-05,
+      "loss": 0.8114,
+      "step": 4750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7835128096674304e-05,
+      "loss": 0.854,
+      "step": 4755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.778826009661542e-05,
+      "loss": 0.9665,
+      "step": 4760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7741419727110275e-05,
+      "loss": 1.0503,
+      "step": 4765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7694607167619744e-05,
+      "loss": 0.8919,
+      "step": 4770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7647822597498204e-05,
+      "loss": 0.6469,
+      "step": 4775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7601066195992755e-05,
+      "loss": 0.6893,
+      "step": 4780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7554338142242588e-05,
+      "loss": 0.6411,
+      "step": 4785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.750763861527831e-05,
+      "loss": 0.6449,
+      "step": 4790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7460967794021178e-05,
+      "loss": 0.9089,
+      "step": 4795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7414325857282526e-05,
+      "loss": 0.9293,
+      "step": 4800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7367712983762973e-05,
+      "loss": 0.8978,
+      "step": 4805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7321129352051815e-05,
+      "loss": 0.7008,
+      "step": 4810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7274575140626318e-05,
+      "loss": 0.5929,
+      "step": 4815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7228050527851002e-05,
+      "loss": 0.6436,
+      "step": 4820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7181555691977012e-05,
+      "loss": 0.8739,
+      "step": 4825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.713509081114138e-05,
+      "loss": 0.6002,
+      "step": 4830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7088656063366404e-05,
+      "loss": 0.6345,
+      "step": 4835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7042251626558915e-05,
+      "loss": 0.7693,
+      "step": 4840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6995877678509598e-05,
+      "loss": 0.8757,
+      "step": 4845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.694953439689236e-05,
+      "loss": 0.8265,
+      "step": 4850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6903221959263575e-05,
+      "loss": 0.9337,
+      "step": 4855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6856940543061466e-05,
+      "loss": 0.7439,
+      "step": 4860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6810690325605425e-05,
+      "loss": 0.7108,
+      "step": 4865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6764471484095256e-05,
+      "loss": 0.945,
+      "step": 4870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6718284195610606e-05,
+      "loss": 0.8079,
+      "step": 4875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.667212863711019e-05,
+      "loss": 0.7797,
+      "step": 4880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6626004985431176e-05,
+      "loss": 0.8916,
+      "step": 4885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6579913417288497e-05,
+      "loss": 0.6987,
+      "step": 4890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6533854109274137e-05,
+      "loss": 0.6642,
+      "step": 4895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.64878272378565e-05,
+      "loss": 0.8752,
+      "step": 4900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6441832979379697e-05,
+      "loss": 0.9805,
+      "step": 4905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.639587151006291e-05,
+      "loss": 0.9946,
+      "step": 4910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6349943005999692e-05,
+      "loss": 0.9213,
+      "step": 4915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.630404764315726e-05,
+      "loss": 0.6244,
+      "step": 4920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.625818559737592e-05,
+      "loss": 0.8286,
+      "step": 4925
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.621235704436826e-05,
+      "loss": 0.748,
+      "step": 4930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6166562159718594e-05,
+      "loss": 0.6695,
+      "step": 4935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6120801118882236e-05,
+      "loss": 0.7134,
+      "step": 4940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6075074097184793e-05,
+      "loss": 0.8118,
+      "step": 4945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6029381269821604e-05,
+      "loss": 0.6554,
+      "step": 4950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5983722811856934e-05,
+      "loss": 0.7553,
+      "step": 4955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5938098898223387e-05,
+      "loss": 0.6727,
+      "step": 4960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5892509703721246e-05,
+      "loss": 0.8901,
+      "step": 4965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5846955403017723e-05,
+      "loss": 0.8303,
+      "step": 4970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5801436170646388e-05,
+      "loss": 0.5564,
+      "step": 4975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5755952181006402e-05,
+      "loss": 0.5544,
+      "step": 4980
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5710503608361953e-05,
+      "loss": 0.9838,
+      "step": 4985
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.566509062684151e-05,
+      "loss": 0.5452,
+      "step": 4990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5619713410437165e-05,
+      "loss": 0.8746,
+      "step": 4995
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.557437213300401e-05,
+      "loss": 0.7794,
+      "step": 5000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5529066968259426e-05,
+      "loss": 1.0124,
+      "step": 5005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.548379808978243e-05,
+      "loss": 0.7641,
+      "step": 5010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.543856567101305e-05,
+      "loss": 0.5594,
+      "step": 5015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5393369885251564e-05,
+      "loss": 0.9393,
+      "step": 5020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.534821090565796e-05,
+      "loss": 0.804,
+      "step": 5025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5303088905251168e-05,
+      "loss": 0.8076,
+      "step": 5030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5258004056908445e-05,
+      "loss": 0.7854,
+      "step": 5035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.521295653336473e-05,
+      "loss": 0.8912,
+      "step": 5040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5167946507211921e-05,
+      "loss": 0.5772,
+      "step": 5045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5122974150898289e-05,
+      "loss": 0.9466,
+      "step": 5050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5078039636727742e-05,
+      "loss": 0.7179,
+      "step": 5055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5033143136859234e-05,
+      "loss": 0.8133,
+      "step": 5060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4988284823306065e-05,
+      "loss": 0.5126,
+      "step": 5065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4943464867935208e-05,
+      "loss": 0.9414,
+      "step": 5070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4898683442466715e-05,
+      "loss": 0.8202,
+      "step": 5075
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4853940718472981e-05,
+      "loss": 0.8107,
+      "step": 5080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4809236867378132e-05,
+      "loss": 0.4788,
+      "step": 5085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4764572060457379e-05,
+      "loss": 0.616,
+      "step": 5090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4719946468836308e-05,
+      "loss": 0.8458,
+      "step": 5095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4675360263490295e-05,
+      "loss": 0.6306,
+      "step": 5100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4630813615243777e-05,
+      "loss": 1.0367,
+      "step": 5105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4586306694769667e-05,
+      "loss": 0.7777,
+      "step": 5110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4541839672588645e-05,
+      "loss": 0.899,
+      "step": 5115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4497412719068543e-05,
+      "loss": 0.7755,
+      "step": 5120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4453026004423665e-05,
+      "loss": 0.7587,
+      "step": 5125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4408679698714155e-05,
+      "loss": 0.7048,
+      "step": 5130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4364373971845334e-05,
+      "loss": 0.7749,
+      "step": 5135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4320108993567052e-05,
+      "loss": 0.6198,
+      "step": 5140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4275884933473039e-05,
+      "loss": 0.6978,
+      "step": 5145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4231701961000255e-05,
+      "loss": 0.8386,
+      "step": 5150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4187560245428217e-05,
+      "loss": 1.0031,
+      "step": 5155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4143459955878418e-05,
+      "loss": 0.6284,
+      "step": 5160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4099401261313613e-05,
+      "loss": 0.5692,
+      "step": 5165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4055384330537158e-05,
+      "loss": 0.8584,
+      "step": 5170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4011409332192472e-05,
+      "loss": 1.2619,
+      "step": 5175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3967476434762228e-05,
+      "loss": 0.5772,
+      "step": 5180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3923585806567882e-05,
+      "loss": 0.8016,
+      "step": 5185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.38797376157689e-05,
+      "loss": 0.8467,
+      "step": 5190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3835932030362131e-05,
+      "loss": 0.8382,
+      "step": 5195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3792169218181259e-05,
+      "loss": 0.587,
+      "step": 5200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3748449346896014e-05,
+      "loss": 0.9576,
+      "step": 5205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.370477258401165e-05,
+      "loss": 0.9168,
+      "step": 5210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3661139096868269e-05,
+      "loss": 0.9009,
+      "step": 5215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3617549052640122e-05,
+      "loss": 0.8359,
+      "step": 5220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3574002618335056e-05,
+      "loss": 0.6666,
+      "step": 5225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3530499960793813e-05,
+      "loss": 0.8901,
+      "step": 5230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.348704124668943e-05,
+      "loss": 0.8002,
+      "step": 5235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3443626642526563e-05,
+      "loss": 0.7054,
+      "step": 5240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3400256314640882e-05,
+      "loss": 0.6015,
+      "step": 5245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3356930429198414e-05,
+      "loss": 0.7122,
+      "step": 5250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3313649152194912e-05,
+      "loss": 0.6281,
+      "step": 5255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.327041264945522e-05,
+      "loss": 0.6623,
+      "step": 5260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3227221086632641e-05,
+      "loss": 0.9232,
+      "step": 5265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.318407462920829e-05,
+      "loss": 0.9595,
+      "step": 5270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3140973442490479e-05,
+      "loss": 0.8968,
+      "step": 5275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3097917691614064e-05,
+      "loss": 0.8148,
+      "step": 5280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3054907541539829e-05,
+      "loss": 0.4773,
+      "step": 5285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3011943157053835e-05,
+      "loss": 0.8106,
+      "step": 5290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2969024702766814e-05,
+      "loss": 0.5096,
+      "step": 5295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2926152343113523e-05,
+      "loss": 0.6075,
+      "step": 5300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2883326242352085e-05,
+      "loss": 0.8905,
+      "step": 5305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2840546564563432e-05,
+      "loss": 0.9308,
+      "step": 5310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.279781347365063e-05,
+      "loss": 0.7087,
+      "step": 5315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2755127133338207e-05,
+      "loss": 0.6384,
+      "step": 5320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2712487707171646e-05,
+      "loss": 0.7954,
+      "step": 5325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2669895358516615e-05,
+      "loss": 0.6267,
+      "step": 5330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2627350250558474e-05,
+      "loss": 0.8153,
+      "step": 5335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2584852546301562e-05,
+      "loss": 0.7915,
+      "step": 5340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2542402408568566e-05,
+      "loss": 0.9214,
+      "step": 5345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2500000000000006e-05,
+      "loss": 0.6103,
+      "step": 5350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.245764548305346e-05,
+      "loss": 0.9779,
+      "step": 5355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2415339020003047e-05,
+      "loss": 0.8499,
+      "step": 5360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2373080772938814e-05,
+      "loss": 0.9021,
+      "step": 5365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2330870903766003e-05,
+      "loss": 0.7664,
+      "step": 5370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.228870957420456e-05,
+      "loss": 0.9179,
+      "step": 5375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2246596945788435e-05,
+      "loss": 0.6955,
+      "step": 5380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2204533179864998e-05,
+      "loss": 0.7456,
+      "step": 5385
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2162518437594409e-05,
+      "loss": 0.8434,
+      "step": 5390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2120552879948992e-05,
+      "loss": 0.909,
+      "step": 5395
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.207863666771265e-05,
+      "loss": 0.7264,
+      "step": 5400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2036769961480204e-05,
+      "loss": 0.7791,
+      "step": 5405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1994952921656816e-05,
+      "loss": 0.8897,
+      "step": 5410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1953185708457354e-05,
+      "loss": 0.8388,
+      "step": 5415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1911468481905785e-05,
+      "loss": 0.846,
+      "step": 5420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1869801401834564e-05,
+      "loss": 0.6913,
+      "step": 5425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1828184627884008e-05,
+      "loss": 0.6485,
+      "step": 5430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1786618319501708e-05,
+      "loss": 0.8394,
+      "step": 5435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1745102635941895e-05,
+      "loss": 0.606,
+      "step": 5440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1703637736264845e-05,
+      "loss": 0.8603,
+      "step": 5445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1662223779336271e-05,
+      "loss": 0.6309,
+      "step": 5450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1620860923826668e-05,
+      "loss": 0.917,
+      "step": 5455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1579549328210809e-05,
+      "loss": 0.7298,
+      "step": 5460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1538289150767034e-05,
+      "loss": 0.8996,
+      "step": 5465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1497080549576664e-05,
+      "loss": 0.5948,
+      "step": 5470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1455923682523475e-05,
+      "loss": 0.7254,
+      "step": 5475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1423035543431017e-05,
+      "loss": 0.6534,
+      "step": 5480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1381972195058676e-05,
+      "loss": 0.6552,
+      "step": 5485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1340961021841543e-05,
+      "loss": 0.7088,
+      "step": 5490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1300002180906985e-05,
+      "loss": 0.7999,
+      "step": 5495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.125909582918179e-05,
+      "loss": 0.581,
+      "step": 5500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.121824212339171e-05,
+      "loss": 0.7823,
+      "step": 5505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1177441220060764e-05,
+      "loss": 0.9573,
+      "step": 5510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1136693275510638e-05,
+      "loss": 0.847,
+      "step": 5515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1095998445860181e-05,
+      "loss": 0.811,
+      "step": 5520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1055356887024682e-05,
+      "loss": 0.5659,
+      "step": 5525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1014768754715349e-05,
+      "loss": 0.6783,
+      "step": 5530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0974234204438736e-05,
+      "loss": 0.7431,
+      "step": 5535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0933753391496035e-05,
+      "loss": 0.8003,
+      "step": 5540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0893326470982607e-05,
+      "loss": 0.7483,
+      "step": 5545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.085295359778731e-05,
+      "loss": 0.7725,
+      "step": 5550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0812634926591938e-05,
+      "loss": 0.8661,
+      "step": 5555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0772370611870616e-05,
+      "loss": 1.031,
+      "step": 5560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0732160807889211e-05,
+      "loss": 0.8442,
+      "step": 5565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0692005668704742e-05,
+      "loss": 0.7441,
+      "step": 5570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0651905348164784e-05,
+      "loss": 0.7389,
+      "step": 5575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0611859999906892e-05,
+      "loss": 0.7161,
+      "step": 5580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0571869777358004e-05,
+      "loss": 0.6675,
+      "step": 5585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0531934833733845e-05,
+      "loss": 0.7315,
+      "step": 5590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0492055322038358e-05,
+      "loss": 1.0652,
+      "step": 5595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0452231395063098e-05,
+      "loss": 0.8005,
+      "step": 5600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0412463205386661e-05,
+      "loss": 0.7738,
+      "step": 5605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0372750905374102e-05,
+      "loss": 0.8228,
+      "step": 5610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0333094647176331e-05,
+      "loss": 0.9948,
+      "step": 5615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0293494582729565e-05,
+      "loss": 0.4674,
+      "step": 5620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.025395086375468e-05,
+      "loss": 0.6554,
+      "step": 5625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.021446364175673e-05,
+      "loss": 0.821,
+      "step": 5630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0175033068024285e-05,
+      "loss": 0.8168,
+      "step": 5635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0135659293628846e-05,
+      "loss": 0.731,
+      "step": 5640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0096342469424364e-05,
+      "loss": 0.6751,
+      "step": 5645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0057082746046526e-05,
+      "loss": 0.7251,
+      "step": 5650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0017880273912298e-05,
+      "loss": 0.7067,
+      "step": 5655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.978735203219283e-06,
+      "loss": 0.8738,
+      "step": 5660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.93964768394513e-06,
+      "loss": 0.927,
+      "step": 5665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.900617865847048e-06,
+      "loss": 0.7999,
+      "step": 5670
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.861645898461114e-06,
+      "loss": 0.8896,
+      "step": 5675
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.822731931101784e-06,
+      "loss": 0.7213,
+      "step": 5680
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.783876112861332e-06,
+      "loss": 0.6355,
+      "step": 5685
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.74507859260918e-06,
+      "loss": 0.6453,
+      "step": 5690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.706339518991431e-06,
+      "loss": 0.8827,
+      "step": 5695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.667659040430252e-06,
+      "loss": 0.7852,
+      "step": 5700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.629037305123312e-06,
+      "loss": 0.9584,
+      "step": 5705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.590474461043214e-06,
+      "loss": 0.8373,
+      "step": 5710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.55197065593693e-06,
+      "loss": 0.6658,
+      "step": 5715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.513526037325234e-06,
+      "loss": 0.8167,
+      "step": 5720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.47514075250214e-06,
+      "loss": 0.6812,
+      "step": 5725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.436814948534327e-06,
+      "loss": 0.8114,
+      "step": 5730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.398548772260598e-06,
+      "loss": 0.8108,
+      "step": 5735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.360342370291284e-06,
+      "loss": 0.9142,
+      "step": 5740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.322195889007712e-06,
+      "loss": 0.7428,
+      "step": 5745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.284109474561633e-06,
+      "loss": 0.9208,
+      "step": 5750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.246083272874659e-06,
+      "loss": 1.0059,
+      "step": 5755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.208117429637709e-06,
+      "loss": 0.4276,
+      "step": 5760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.170212090310445e-06,
+      "loss": 0.7652,
+      "step": 5765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.132367400120736e-06,
+      "loss": 0.7476,
+      "step": 5770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.09458350406404e-06,
+      "loss": 0.872,
+      "step": 5775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.056860546902948e-06,
+      "loss": 1.2268,
+      "step": 5780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.019198673166551e-06,
+      "loss": 1.0005,
+      "step": 5785
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.98159802714988e-06,
+      "loss": 0.6815,
+      "step": 5790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.944058752913443e-06,
+      "loss": 0.9238,
+      "step": 5795
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.906580994282551e-06,
+      "loss": 0.8036,
+      "step": 5800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.869164894846859e-06,
+      "loss": 0.7558,
+      "step": 5805
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.831810597959809e-06,
+      "loss": 0.7719,
+      "step": 5810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.79451824673799e-06,
+      "loss": 0.6264,
+      "step": 5815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.757287984060738e-06,
+      "loss": 0.866,
+      "step": 5820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.72011995256944e-06,
+      "loss": 0.7488,
+      "step": 5825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.683014294667077e-06,
+      "loss": 0.6472,
+      "step": 5830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.64597115251769e-06,
+      "loss": 1.112,
+      "step": 5835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.608990668045747e-06,
+      "loss": 0.8848,
+      "step": 5840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.572072982935683e-06,
+      "loss": 0.8278,
+      "step": 5845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.535218238631326e-06,
+      "loss": 0.8953,
+      "step": 5850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.498426576335355e-06,
+      "loss": 0.7062,
+      "step": 5855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.461698137008762e-06,
+      "loss": 0.7313,
+      "step": 5860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.425033061370307e-06,
+      "loss": 0.7999,
+      "step": 5865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.388431489895987e-06,
+      "loss": 0.7563,
+      "step": 5870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.351893562818492e-06,
+      "loss": 0.677,
+      "step": 5875
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.315419420126668e-06,
+      "loss": 1.0062,
+      "step": 5880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.279009201564988e-06,
+      "loss": 0.7419,
+      "step": 5885
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.242663046633001e-06,
+      "loss": 0.9599,
+      "step": 5890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.206381094584815e-06,
+      "loss": 0.8503,
+      "step": 5895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.17016348442855e-06,
+      "loss": 0.629,
+      "step": 5900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.134010354925813e-06,
+      "loss": 0.7508,
+      "step": 5905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.097921844591171e-06,
+      "loss": 0.828,
+      "step": 5910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.061898091691606e-06,
+      "loss": 0.7767,
+      "step": 5915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.025939234245996e-06,
+      "loss": 0.828,
+      "step": 5920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.990045410024566e-06,
+      "loss": 0.9419,
+      "step": 5925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.95421675654841e-06,
+      "loss": 0.9734,
+      "step": 5930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.918453411088922e-06,
+      "loss": 1.0198,
+      "step": 5935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.882755510667242e-06,
+      "loss": 0.8168,
+      "step": 5940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.847123192053832e-06,
+      "loss": 0.7654,
+      "step": 5945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.81155659176783e-06,
+      "loss": 0.5594,
+      "step": 5950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.77605584607661e-06,
+      "loss": 0.7611,
+      "step": 5955
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.740621090995261e-06,
+      "loss": 0.6617,
+      "step": 5960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.705252462285984e-06,
+      "loss": 0.8358,
+      "step": 5965
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.66995009545769e-06,
+      "loss": 0.6516,
+      "step": 5970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.63471412576536e-06,
+      "loss": 0.6219,
+      "step": 5975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.599544688209617e-06,
+      "loss": 0.8534,
+      "step": 5980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.564441917536197e-06,
+      "loss": 0.8119,
+      "step": 5985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.529405948235369e-06,
+      "loss": 0.6123,
+      "step": 5990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.494436914541486e-06,
+      "loss": 0.8341,
+      "step": 5995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.45953495043245e-06,
+      "loss": 0.7594,
+      "step": 6000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.4247001896292e-06,
+      "loss": 0.826,
+      "step": 6005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.389932765595187e-06,
+      "loss": 0.7771,
+      "step": 6010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.355232811535878e-06,
+      "loss": 0.8805,
+      "step": 6015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.320600460398241e-06,
+      "loss": 0.807,
+      "step": 6020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.286035844870234e-06,
+      "loss": 0.6806,
+      "step": 6025
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.251539097380295e-06,
+      "loss": 0.7742,
+      "step": 6030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.2171103500968425e-06,
+      "loss": 0.9167,
+      "step": 6035
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.182749734927754e-06,
+      "loss": 0.7876,
+      "step": 6040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.1484573835198855e-06,
+      "loss": 0.8334,
+      "step": 6045
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.114233427258538e-06,
+      "loss": 0.7458,
+      "step": 6050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.080077997266976e-06,
+      "loss": 0.743,
+      "step": 6055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.045991224405912e-06,
+      "loss": 0.756,
+      "step": 6060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.011973239273021e-06,
+      "loss": 0.8207,
+      "step": 6065
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.978024172202419e-06,
+      "loss": 0.5677,
+      "step": 6070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.944144153264162e-06,
+      "loss": 0.8729,
+      "step": 6075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.910333312263795e-06,
+      "loss": 0.6822,
+      "step": 6080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.8765917787417975e-06,
+      "loss": 0.9239,
+      "step": 6085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.8429196819730925e-06,
+      "loss": 0.6211,
+      "step": 6090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.809317150966615e-06,
+      "loss": 1.0733,
+      "step": 6095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.775784314464717e-06,
+      "loss": 0.4436,
+      "step": 6100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.742321300942756e-06,
+      "loss": 0.5062,
+      "step": 6105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.7089282386085944e-06,
+      "loss": 0.8625,
+      "step": 6110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.675605255402032e-06,
+      "loss": 0.9675,
+      "step": 6115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.64235247899444e-06,
+      "loss": 0.6775,
+      "step": 6120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.609170036788148e-06,
+      "loss": 0.6108,
+      "step": 6125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.576058055916032e-06,
+      "loss": 0.645,
+      "step": 6130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.543016663241033e-06,
+      "loss": 0.7712,
+      "step": 6135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.510045985355598e-06,
+      "loss": 0.6582,
+      "step": 6140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.477146148581278e-06,
+      "loss": 0.6778,
+      "step": 6145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.4443172789681956e-06,
+      "loss": 0.7895,
+      "step": 6150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.411559502294573e-06,
+      "loss": 0.5898,
+      "step": 6155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3788729440662584e-06,
+      "loss": 0.6239,
+      "step": 6160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3462577295162335e-06,
+      "loss": 0.429,
+      "step": 6165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3137139836041384e-06,
+      "loss": 0.5971,
+      "step": 6170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.2812418310158e-06,
+      "loss": 0.6668,
+      "step": 6175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.248841396162739e-06,
+      "loss": 0.7783,
+      "step": 6180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.216512803181709e-06,
+      "loss": 0.8076,
+      "step": 6185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.184256175934208e-06,
+      "loss": 0.8674,
+      "step": 6190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.152071638006013e-06,
+      "loss": 0.6394,
+      "step": 6195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.119959312706702e-06,
+      "loss": 0.6875,
+      "step": 6200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.08791932306918e-06,
+      "loss": 0.896,
+      "step": 6205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.055951791849218e-06,
+      "loss": 0.6419,
+      "step": 6210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.0240568415249665e-06,
+      "loss": 0.6601,
+      "step": 6215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9922345942965055e-06,
+      "loss": 0.9213,
+      "step": 6220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.960485172085337e-06,
+      "loss": 0.8542,
+      "step": 6225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.928808696533988e-06,
+      "loss": 0.6801,
+      "step": 6230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.897205289005484e-06,
+      "loss": 0.746,
+      "step": 6235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.865675070582874e-06,
+      "loss": 0.7119,
+      "step": 6240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.834218162068858e-06,
+      "loss": 0.7555,
+      "step": 6245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.8028346839852e-06,
+      "loss": 0.9079,
+      "step": 6250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.7715247565723556e-06,
+      "loss": 0.949,
+      "step": 6255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.740288499788998e-06,
+      "loss": 0.6477,
+      "step": 6260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.709126033311507e-06,
+      "loss": 0.7771,
+      "step": 6265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.678037476533591e-06,
+      "loss": 0.7811,
+      "step": 6270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.647022948565742e-06,
+      "loss": 0.6757,
+      "step": 6275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.616082568234854e-06,
+      "loss": 0.6856,
+      "step": 6280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.585216454083722e-06,
+      "loss": 0.8179,
+      "step": 6285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.554424724370611e-06,
+      "loss": 0.6507,
+      "step": 6290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.5237074970687915e-06,
+      "loss": 0.7741,
+      "step": 6295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.493064889866093e-06,
+      "loss": 0.7659,
+      "step": 6300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.462497020164442e-06,
+      "loss": 0.7985,
+      "step": 6305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.432004005079433e-06,
+      "loss": 0.8658,
+      "step": 6310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.401585961439862e-06,
+      "loss": 0.7022,
+      "step": 6315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.371243005787279e-06,
+      "loss": 1.1631,
+      "step": 6320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.340975254375563e-06,
+      "loss": 0.8244,
+      "step": 6325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3107828231704485e-06,
+      "loss": 0.8285,
+      "step": 6330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.280665827849079e-06,
+      "loss": 0.7123,
+      "step": 6335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.250624383799615e-06,
+      "loss": 0.756,
+      "step": 6340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.220658606120738e-06,
+      "loss": 0.8428,
+      "step": 6345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.190768609621199e-06,
+      "loss": 0.7695,
+      "step": 6350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.16095450881946e-06,
+      "loss": 0.8063,
+      "step": 6355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.1312164179431465e-06,
+      "loss": 0.5847,
+      "step": 6360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.101554450928711e-06,
+      "loss": 0.7649,
+      "step": 6365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.071968721420931e-06,
+      "loss": 0.7881,
+      "step": 6370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.042459342772468e-06,
+      "loss": 0.7859,
+      "step": 6375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.013026428043516e-06,
+      "loss": 0.7341,
+      "step": 6380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.983670090001255e-06,
+      "loss": 0.8589,
+      "step": 6385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.954390441119494e-06,
+      "loss": 0.7823,
+      "step": 6390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.925187593578246e-06,
+      "loss": 0.92,
+      "step": 6395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.896061659263229e-06,
+      "loss": 0.7338,
+      "step": 6400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.867012749765509e-06,
+      "loss": 0.6321,
+      "step": 6405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.838040976381037e-06,
+      "loss": 0.6641,
+      "step": 6410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.80914645011023e-06,
+      "loss": 0.5284,
+      "step": 6415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.780329281657547e-06,
+      "loss": 1.1097,
+      "step": 6420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.751589581431065e-06,
+      "loss": 0.8217,
+      "step": 6425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7229274595420485e-06,
+      "loss": 0.5415,
+      "step": 6430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.694343025804543e-06,
+      "loss": 0.7325,
+      "step": 6435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6658363897349375e-06,
+      "loss": 0.8117,
+      "step": 6440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6374076605515575e-06,
+      "loss": 0.6711,
+      "step": 6445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.609056947174234e-06,
+      "loss": 0.8682,
+      "step": 6450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.580784358223905e-06,
+      "loss": 0.7096,
+      "step": 6455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.552590002022178e-06,
+      "loss": 0.8101,
+      "step": 6460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5244739865909325e-06,
+      "loss": 0.8439,
+      "step": 6465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.496436419651889e-06,
+      "loss": 0.5576,
+      "step": 6470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.468477408626221e-06,
+      "loss": 0.8018,
+      "step": 6475
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.440597060634119e-06,
+      "loss": 0.9029,
+      "step": 6480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.412795482494372e-06,
+      "loss": 0.7994,
+      "step": 6485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.385072780724017e-06,
+      "loss": 0.7351,
+      "step": 6490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.35742906153786e-06,
+      "loss": 0.9055,
+      "step": 6495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.329864430848091e-06,
+      "loss": 0.8344,
+      "step": 6500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.302378994263922e-06,
+      "loss": 0.6588,
+      "step": 6505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.274972857091103e-06,
+      "loss": 0.5764,
+      "step": 6510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2476461243316005e-06,
+      "loss": 0.8151,
+      "step": 6515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.220398900683134e-06,
+      "loss": 0.766,
+      "step": 6520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.193231290538791e-06,
+      "loss": 0.7581,
+      "step": 6525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.166143397986666e-06,
+      "loss": 0.689,
+      "step": 6530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.139135326809388e-06,
+      "loss": 0.7887,
+      "step": 6535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.112207180483785e-06,
+      "loss": 0.7882,
+      "step": 6540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.085359062180486e-06,
+      "loss": 0.934,
+      "step": 6545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0585910747634655e-06,
+      "loss": 0.5218,
+      "step": 6550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.031903320789718e-06,
+      "loss": 0.744,
+      "step": 6555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.005295902508826e-06,
+      "loss": 0.9781,
+      "step": 6560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.978768921862588e-06,
+      "loss": 0.8502,
+      "step": 6565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.95232248048461e-06,
+      "loss": 0.5664,
+      "step": 6570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.925956679699932e-06,
+      "loss": 1.1008,
+      "step": 6575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.899671620524631e-06,
+      "loss": 0.7372,
+      "step": 6580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.873467403665437e-06,
+      "loss": 0.9203,
+      "step": 6585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.847344129519345e-06,
+      "loss": 0.7854,
+      "step": 6590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.821301898173233e-06,
+      "loss": 0.8344,
+      "step": 6595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7953408094034774e-06,
+      "loss": 0.6665,
+      "step": 6600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7694609626755695e-06,
+      "loss": 0.8114,
+      "step": 6605
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.74366245714374e-06,
+      "loss": 0.7552,
+      "step": 6610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7179453916505654e-06,
+      "loss": 0.5585,
+      "step": 6615
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6923098647266065e-06,
+      "loss": 0.605,
+      "step": 6620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6667559745900194e-06,
+      "loss": 0.8993,
+      "step": 6625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6412838191461933e-06,
+      "loss": 0.6046,
+      "step": 6630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6158934959873353e-06,
+      "loss": 0.8597,
+      "step": 6635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5905851023921635e-06,
+      "loss": 0.7947,
+      "step": 6640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5653587353254776e-06,
+      "loss": 0.7831,
+      "step": 6645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.540214491437796e-06,
+      "loss": 0.7043,
+      "step": 6650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5151524670650304e-06,
+      "loss": 0.8143,
+      "step": 6655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.490172758228036e-06,
+      "loss": 0.953,
+      "step": 6660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.465275460632331e-06,
+      "loss": 0.8666,
+      "step": 6665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4404606696676723e-06,
+      "loss": 1.083,
+      "step": 6670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4157284804076957e-06,
+      "loss": 0.7016,
+      "step": 6675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3910789876095915e-06,
+      "loss": 0.7361,
+      "step": 6680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.366512285713677e-06,
+      "loss": 0.6306,
+      "step": 6685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3420284688430945e-06,
+      "loss": 0.8853,
+      "step": 6690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3176276308034326e-06,
+      "loss": 0.7744,
+      "step": 6695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2933098650823307e-06,
+      "loss": 0.632,
+      "step": 6700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2690752648491753e-06,
+      "loss": 0.8154,
+      "step": 6705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2449239229547122e-06,
+      "loss": 0.6898,
+      "step": 6710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2208559319306968e-06,
+      "loss": 0.8474,
+      "step": 6715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1968713839895385e-06,
+      "loss": 0.7164,
+      "step": 6720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.172970371023948e-06,
+      "loss": 1.2002,
+      "step": 6725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1491529846065903e-06,
+      "loss": 0.967,
+      "step": 6730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.125419315989725e-06,
+      "loss": 0.5756,
+      "step": 6735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1017694561048668e-06,
+      "loss": 0.7267,
+      "step": 6740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0782034955624207e-06,
+      "loss": 1.0507,
+      "step": 6745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0547215246513606e-06,
+      "loss": 0.7177,
+      "step": 6750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0313236333388536e-06,
+      "loss": 0.6095,
+      "step": 6755
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0080099112699404e-06,
+      "loss": 0.8825,
+      "step": 6760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.98478044776718e-06,
+      "loss": 0.7863,
+      "step": 6765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9616353318303037e-06,
+      "loss": 0.6207,
+      "step": 6770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9385746521358836e-06,
+      "loss": 0.7949,
+      "step": 6775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9155984970369976e-06,
+      "loss": 0.8328,
+      "step": 6780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8927069545628516e-06,
+      "loss": 0.8013,
+      "step": 6785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.869900112418511e-06,
+      "loss": 0.7731,
+      "step": 6790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8471780579845093e-06,
+      "loss": 0.6868,
+      "step": 6795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8245408783165147e-06,
+      "loss": 0.7658,
+      "step": 6800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8019886601450467e-06,
+      "loss": 0.7338,
+      "step": 6805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.77952148987507e-06,
+      "loss": 0.6259,
+      "step": 6810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7571394535857416e-06,
+      "loss": 0.7008,
+      "step": 6815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7348426370300163e-06,
+      "loss": 0.818,
+      "step": 6820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7126311256343407e-06,
+      "loss": 0.5421,
+      "step": 6825
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.690505004498359e-06,
+      "loss": 0.8767,
+      "step": 6830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.668464358394521e-06,
+      "loss": 0.7157,
+      "step": 6835
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6465092717678106e-06,
+      "loss": 0.745,
+      "step": 6840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.62463982873542e-06,
+      "loss": 0.6731,
+      "step": 6845
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6028561130863826e-06,
+      "loss": 0.8392,
+      "step": 6850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.581158208281301e-06,
+      "loss": 0.8001,
+      "step": 6855
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5595461974520064e-06,
+      "loss": 0.5024,
+      "step": 6860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5380201634012337e-06,
+      "loss": 0.9844,
+      "step": 6865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5165801886023187e-06,
+      "loss": 0.5677,
+      "step": 6870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.495226355198882e-06,
+      "loss": 0.6084,
+      "step": 6875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4739587450044928e-06,
+      "loss": 0.7992,
+      "step": 6880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.452777439502382e-06,
+      "loss": 0.6973,
+      "step": 6885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.431682519845116e-06,
+      "loss": 0.9436,
+      "step": 6890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4106740668542876e-06,
+      "loss": 0.7088,
+      "step": 6895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.389752161020209e-06,
+      "loss": 0.6255,
+      "step": 6900
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.368916882501601e-06,
+      "loss": 0.7974,
+      "step": 6905
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3481683111252844e-06,
+      "loss": 0.8323,
+      "step": 6910
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.327506526385881e-06,
+      "loss": 0.6618,
+      "step": 6915
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3069316074454977e-06,
+      "loss": 0.7402,
+      "step": 6920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2864436331334365e-06,
+      "loss": 0.7154,
+      "step": 6925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2660426819458903e-06,
+      "loss": 0.7087,
+      "step": 6930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.24572883204561e-06,
+      "loss": 0.6953,
+      "step": 6935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.225502161261672e-06,
+      "loss": 0.8798,
+      "step": 6940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2053627470891237e-06,
+      "loss": 0.8333,
+      "step": 6945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.185310666688681e-06,
+      "loss": 0.7683,
+      "step": 6950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1653459968865032e-06,
+      "loss": 0.6183,
+      "step": 6955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1454688141738037e-06,
+      "loss": 0.8997,
+      "step": 6960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1256791947066257e-06,
+      "loss": 0.759,
+      "step": 6965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1059772143055434e-06,
+      "loss": 0.9719,
+      "step": 6970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.08636294845532e-06,
+      "loss": 0.5012,
+      "step": 6975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.066836472304695e-06,
+      "loss": 0.6902,
+      "step": 6980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0473978606660253e-06,
+      "loss": 0.8044,
+      "step": 6985
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.028047188015039e-06,
+      "loss": 0.6967,
+      "step": 6990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.008784528490562e-06,
+      "loss": 0.7685,
+      "step": 6995
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.989609955894178e-06,
+      "loss": 0.6908,
+      "step": 7000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.970523543690003e-06,
+      "loss": 0.87,
+      "step": 7005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.951525365004378e-06,
+      "loss": 0.8509,
+      "step": 7010
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.932615492625586e-06,
+      "loss": 0.8198,
+      "step": 7015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9137939990035907e-06,
+      "loss": 0.8696,
+      "step": 7020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8950609562497334e-06,
+      "loss": 0.6632,
+      "step": 7025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8764164361364827e-06,
+      "loss": 0.7525,
+      "step": 7030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8578605100971425e-06,
+      "loss": 0.7667,
+      "step": 7035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.839393249225585e-06,
+      "loss": 0.5133,
+      "step": 7040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8210147242759767e-06,
+      "loss": 0.587,
+      "step": 7045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8027250056625033e-06,
+      "loss": 0.7528,
+      "step": 7050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7845241634591142e-06,
+      "loss": 1.1137,
+      "step": 7055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.766412267399234e-06,
+      "loss": 1.0754,
+      "step": 7060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7483893868755097e-06,
+      "loss": 0.7962,
+      "step": 7065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7304555909395392e-06,
+      "loss": 0.8363,
+      "step": 7070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.71261094830161e-06,
+      "loss": 0.7305,
+      "step": 7075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6948555273304356e-06,
+      "loss": 0.5929,
+      "step": 7080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6771893960528778e-06,
+      "loss": 0.6707,
+      "step": 7085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6596126221537223e-06,
+      "loss": 0.8178,
+      "step": 7090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6421252729753867e-06,
+      "loss": 0.9292,
+      "step": 7095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6247274155176573e-06,
+      "loss": 0.8629,
+      "step": 7100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6074191164374836e-06,
+      "loss": 0.5622,
+      "step": 7105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5902004420486533e-06,
+      "loss": 0.7807,
+      "step": 7110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5730714583215871e-06,
+      "loss": 0.8211,
+      "step": 7115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5560322308830783e-06,
+      "loss": 0.7833,
+      "step": 7120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5390828250160166e-06,
+      "loss": 0.9674,
+      "step": 7125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5222233056591733e-06,
+      "loss": 0.6603,
+      "step": 7130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.50545373740692e-06,
+      "loss": 0.7774,
+      "step": 7135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4887741845090037e-06,
+      "loss": 0.708,
+      "step": 7140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4721847108702974e-06,
+      "loss": 0.8392,
+      "step": 7145
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4556853800505393e-06,
+      "loss": 0.5191,
+      "step": 7150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4392762552641105e-06,
+      "loss": 0.7408,
+      "step": 7155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4229573993797735e-06,
+      "loss": 1.0118,
+      "step": 7160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4067288749204543e-06,
+      "loss": 0.6942,
+      "step": 7165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3905907440629752e-06,
+      "loss": 0.8461,
+      "step": 7170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3745430686378403e-06,
+      "loss": 0.6776,
+      "step": 7175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3585859101289844e-06,
+      "loss": 0.5997,
+      "step": 7180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3427193296735447e-06,
+      "loss": 0.7005,
+      "step": 7185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3269433880616244e-06,
+      "loss": 1.1284,
+      "step": 7190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3112581457360578e-06,
+      "loss": 0.9773,
+      "step": 7195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2956636627921791e-06,
+      "loss": 0.9411,
+      "step": 7200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2801599989775952e-06,
+      "loss": 1.0043,
+      "step": 7205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2647472136919552e-06,
+      "loss": 0.8397,
+      "step": 7210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2494253659867227e-06,
+      "loss": 0.7063,
+      "step": 7215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2341945145649485e-06,
+      "loss": 0.7289,
+      "step": 7220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2190547177810458e-06,
+      "loss": 0.7823,
+      "step": 7225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.204006033640573e-06,
+      "loss": 0.5584,
+      "step": 7230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1890485197999901e-06,
+      "loss": 0.8949,
+      "step": 7235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1741822335664786e-06,
+      "loss": 0.7686,
+      "step": 7240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1594072318976802e-06,
+      "loss": 0.8988,
+      "step": 7245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1447235714014876e-06,
+      "loss": 0.7989,
+      "step": 7250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1301313083358634e-06,
+      "loss": 0.7765,
+      "step": 7255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1156304986085663e-06,
+      "loss": 0.5501,
+      "step": 7260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1012211977769804e-06,
+      "loss": 1.0066,
+      "step": 7265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0869034610479e-06,
+      "loss": 0.6592,
+      "step": 7270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.072677343277284e-06,
+      "loss": 0.805,
+      "step": 7275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0585428989700912e-06,
+      "loss": 0.9323,
+      "step": 7280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.044500182280031e-06,
+      "loss": 0.8804,
+      "step": 7285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0305492470093798e-06,
+      "loss": 0.8296,
+      "step": 7290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0166901466087814e-06,
+      "loss": 0.8219,
+      "step": 7295
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0029229341770114e-06,
+      "loss": 0.9175,
+      "step": 7300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.892476624608017e-07,
+      "loss": 0.8356,
+      "step": 7305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.756643838546297e-07,
+      "loss": 0.6698,
+      "step": 7310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.62173150400514e-07,
+      "loss": 0.8244,
+      "step": 7315
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.487740137878176e-07,
+      "loss": 0.7697,
+      "step": 7320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.354670253530539e-07,
+      "loss": 0.4655,
+      "step": 7325
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.22252236079682e-07,
+      "loss": 1.0098,
+      "step": 7330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.091296965979185e-07,
+      "loss": 0.9949,
+      "step": 7335
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.960994571845404e-07,
+      "loss": 0.4978,
+      "step": 7340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.831615677626887e-07,
+      "loss": 0.8149,
+      "step": 7345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.703160779016867e-07,
+      "loss": 0.8824,
+      "step": 7350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.575630368168413e-07,
+      "loss": 0.862,
+      "step": 7355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.449024933692568e-07,
+      "loss": 0.6874,
+      "step": 7360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.323344960656537e-07,
+      "loss": 0.7851,
+      "step": 7365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.198590930581701e-07,
+      "loss": 0.7562,
+      "step": 7370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.074763321441914e-07,
+      "loss": 0.7379,
+      "step": 7375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.951862607661592e-07,
+      "loss": 0.6407,
+      "step": 7380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.829889260113827e-07,
+      "loss": 1.1205,
+      "step": 7385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.708843746118799e-07,
+      "loss": 0.8137,
+      "step": 7390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.588726529441842e-07,
+      "loss": 0.7034,
+      "step": 7395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.469538070291466e-07,
+      "loss": 0.7849,
+      "step": 7400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.351278825318136e-07,
+      "loss": 0.6706,
+      "step": 7405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.233949247611893e-07,
+      "loss": 0.8271,
+      "step": 7410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.117549786701039e-07,
+      "loss": 0.7366,
+      "step": 7415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.002080888550422e-07,
+      "loss": 1.0551,
+      "step": 7420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.887542995559299e-07,
+      "loss": 0.8294,
+      "step": 7425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.77393654656025e-07,
+      "loss": 0.6495,
+      "step": 7430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.661261976816963e-07,
+      "loss": 0.7162,
+      "step": 7435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.549519718022868e-07,
+      "loss": 0.7113,
+      "step": 7440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.438710198299474e-07,
+      "loss": 0.7258,
+      "step": 7445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.328833842194543e-07,
+      "loss": 0.69,
+      "step": 7450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.219891070680606e-07,
+      "loss": 0.9865,
+      "step": 7455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.111882301153394e-07,
+      "loss": 0.6798,
+      "step": 7460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.00480794743008e-07,
+      "loss": 0.5414,
+      "step": 7465
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.898668419747872e-07,
+      "loss": 0.6554,
+      "step": 7470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.793464124762227e-07,
+      "loss": 0.7175,
+      "step": 7475
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.689195465545582e-07,
+      "loss": 0.8669,
+      "step": 7480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.585862841585493e-07,
+      "loss": 1.0802,
+      "step": 7485
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.483466648783325e-07,
+      "loss": 0.7724,
+      "step": 7490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.38200727945265e-07,
+      "loss": 0.7484,
+      "step": 7495
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.281485122317825e-07,
+      "loss": 1.1423,
+      "step": 7500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.181900562512331e-07,
+      "loss": 0.8488,
+      "step": 7505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.083253981577441e-07,
+      "loss": 0.4918,
+      "step": 7510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.985545757460797e-07,
+      "loss": 0.9289,
+      "step": 7515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.888776264514782e-07,
+      "loss": 0.7569,
+      "step": 7520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.792945873495242e-07,
+      "loss": 0.5507,
+      "step": 7525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6980549515600326e-07,
+      "loss": 0.7713,
+      "step": 7530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6041038622675056e-07,
+      "loss": 0.6953,
+      "step": 7535
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5110929655753086e-07,
+      "loss": 0.7544,
+      "step": 7540
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.419022617838886e-07,
+      "loss": 0.9048,
+      "step": 7545
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.327893171810038e-07,
+      "loss": 0.5718,
+      "step": 7550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2377049766357815e-07,
+      "loss": 1.1327,
+      "step": 7555
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1484583778567676e-07,
+      "loss": 0.4878,
+      "step": 7560
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.060153717406201e-07,
+      "loss": 0.6308,
+      "step": 7565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9727913336083667e-07,
+      "loss": 0.6904,
+      "step": 7570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.886371561177299e-07,
+      "loss": 0.918,
+      "step": 7575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.800894731215698e-07,
+      "loss": 0.9767,
+      "step": 7580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.7163611712134616e-07,
+      "loss": 0.7588,
+      "step": 7585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.632771205046542e-07,
+      "loss": 0.7241,
+      "step": 7590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.550125152975703e-07,
+      "loss": 0.5491,
+      "step": 7595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4684233316451544e-07,
+      "loss": 0.6437,
+      "step": 7600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.387666054081556e-07,
+      "loss": 1.3491,
+      "step": 7605
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3078536296926586e-07,
+      "loss": 0.7387,
+      "step": 7610
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2289863642661623e-07,
+      "loss": 0.7446,
+      "step": 7615
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.151064559968608e-07,
+      "loss": 0.5848,
+      "step": 7620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0740885153441033e-07,
+      "loss": 0.9344,
+      "step": 7625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.998058525313263e-07,
+      "loss": 0.9517,
+      "step": 7630
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.922974881172019e-07,
+      "loss": 0.6362,
+      "step": 7635
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8488378705906204e-07,
+      "loss": 0.9986,
+      "step": 7640
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.775647777612383e-07,
+      "loss": 0.8388,
+      "step": 7645
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.703404882652694e-07,
+      "loss": 0.9996,
+      "step": 7650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6321094624978946e-07,
+      "loss": 0.9899,
+      "step": 7655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.561761790304235e-07,
+      "loss": 0.8683,
+      "step": 7660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.492362135596893e-07,
+      "loss": 0.7535,
+      "step": 7665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4239107642687885e-07,
+      "loss": 0.6757,
+      "step": 7670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.356407938579691e-07,
+      "loss": 0.5528,
+      "step": 7675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.289853917155249e-07,
+      "loss": 0.7041,
+      "step": 7680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.224248954985797e-07,
+      "loss": 0.5845,
+      "step": 7685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1595933034256055e-07,
+      "loss": 0.6428,
+      "step": 7690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.095887210191827e-07,
+      "loss": 0.7689,
+      "step": 7695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0331309193634408e-07,
+      "loss": 0.814,
+      "step": 7700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9713246713805588e-07,
+      "loss": 0.6538,
+      "step": 7705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.910468703043261e-07,
+      "loss": 0.6958,
+      "step": 7710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.850563247510817e-07,
+      "loss": 0.8858,
+      "step": 7715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7916085343008537e-07,
+      "loss": 0.7462,
+      "step": 7720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7336047892882456e-07,
+      "loss": 0.7871,
+      "step": 7725
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6765522347045037e-07,
+      "loss": 0.6162,
+      "step": 7730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6204510891367487e-07,
+      "loss": 0.9041,
+      "step": 7735
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5653015675269888e-07,
+      "loss": 0.5819,
+      "step": 7740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.511103881171261e-07,
+      "loss": 0.8548,
+      "step": 7745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.457858237718712e-07,
+      "loss": 0.7745,
+      "step": 7750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.405564841170992e-07,
+      "loss": 0.8332,
+      "step": 7755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3542238918813344e-07,
+      "loss": 0.7863,
+      "step": 7760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3038355865538642e-07,
+      "loss": 0.9428,
+      "step": 7765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2544001182427934e-07,
+      "loss": 0.7225,
+      "step": 7770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.2059176763516978e-07,
+      "loss": 0.8049,
+      "step": 7775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1583884466327688e-07,
+      "loss": 0.6862,
+      "step": 7780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1118126111862026e-07,
+      "loss": 1.0844,
+      "step": 7785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.066190348459395e-07,
+      "loss": 0.867,
+      "step": 7790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0215218332462195e-07,
+      "loss": 0.9057,
+      "step": 7795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.778072366865287e-08,
+      "loss": 0.9782,
+      "step": 7800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.350467262653761e-08,
+      "loss": 0.7718,
+      "step": 7805
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.932404658123228e-08,
+      "loss": 0.655,
+      "step": 7810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.523886155009653e-08,
+      "loss": 0.8366,
+      "step": 7815
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.124913318482419e-08,
+      "loss": 1.1631,
+      "step": 7820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.735487677137665e-08,
+      "loss": 0.7809,
+      "step": 7825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.355610722993566e-08,
+      "loss": 0.5667,
+      "step": 7830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.985283911483675e-08,
+      "loss": 0.6048,
+      "step": 7835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.624508661452755e-08,
+      "loss": 0.9556,
+      "step": 7840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.273286355149566e-08,
+      "loss": 0.6761,
+      "step": 7845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.931618338222145e-08,
+      "loss": 0.6436,
+      "step": 7850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.599505919713921e-08,
+      "loss": 0.9335,
+      "step": 7855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.276950372056499e-08,
+      "loss": 0.6817,
+      "step": 7860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.963952931066884e-08,
+      "loss": 0.7737,
+      "step": 7865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.660514795941651e-08,
+      "loss": 0.6928,
+      "step": 7870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.366637129252227e-08,
+      "loss": 0.7731,
+      "step": 7875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.08232105694073e-08,
+      "loss": 0.8099,
+      "step": 7880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.807567668315804e-08,
+      "loss": 0.7373,
+      "step": 7885
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.542378016048453e-08,
+      "loss": 0.7933,
+      "step": 7890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.286753116167329e-08,
+      "loss": 1.0595,
+      "step": 7895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.040693948056228e-08,
+      "loss": 0.6221,
+      "step": 7900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8042014544485406e-08,
+      "loss": 0.675,
+      "step": 7905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.577276541425311e-08,
+      "loss": 0.7004,
+      "step": 7910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3599200784105158e-08,
+      "loss": 0.8356,
+      "step": 7915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.152132898168291e-08,
+      "loss": 0.9071,
+      "step": 7920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.953915796800154e-08,
+      "loss": 0.6645,
+      "step": 7925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.765269533741121e-08,
+      "loss": 0.7965,
+      "step": 7930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.586194831757204e-08,
+      "loss": 0.8691,
+      "step": 7935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.416692376942641e-08,
+      "loss": 0.815,
+      "step": 7940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2567628187171166e-08,
+      "loss": 0.619,
+      "step": 7945
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.10640676982382e-08,
+      "loss": 0.6708,
+      "step": 7950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.656248063258378e-09,
+      "loss": 1.104,
+      "step": 7955
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.344174676055971e-09,
+      "loss": 0.8751,
+      "step": 7960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.127852563618143e-09,
+      "loss": 0.9489,
+      "step": 7965
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.007286386075505e-09,
+      "loss": 0.8496,
+      "step": 7970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.982480436682702e-09,
+      "loss": 0.7485,
+      "step": 7975
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.053438641815621e-09,
+      "loss": 0.6365,
+      "step": 7980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2201645609353235e-09,
+      "loss": 0.7357,
+      "step": 7985
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4826613865880367e-09,
+      "loss": 0.6877,
+      "step": 7990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8409319443968286e-09,
+      "loss": 0.8213,
+      "step": 7995
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2949786930366302e-09,
+      "loss": 0.9413,
+      "step": 8000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.448037242342333e-10,
+      "loss": 0.7706,
+      "step": 8005
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.904087627571885e-10,
+      "loss": 0.8707,
+      "step": 8010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3179516640825515e-10,
+      "loss": 0.678,
+      "step": 8015
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.896392602540047e-11,
+      "loss": 0.8151,
+      "step": 8020
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.91566546792199e-12,
+      "loss": 0.6908,
+      "step": 8025
+    },
+    {
+      "epoch": 1.0,
+      "step": 8025,
+      "total_flos": 1.387965865402368e+17,
+      "train_loss": 0.800769867302844,
+      "train_runtime": 34838.427,
+      "train_samples_per_second": 2.073,
+      "train_steps_per_second": 0.23
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 8025,
+  "num_train_epochs": 1,
+  "save_steps": 5000,
+  "total_flos": 1.387965865402368e+17,
+  "trial_name": null,
+  "trial_params": null
+}