diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.0,
-  "global_step": 59,
+  "epoch": 101.0,
+  "global_step": 5959,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -72,11 +72,7091 @@
       "learning_rate": 1.550127458494346e-06,
       "loss": 1.7807,
       "step": 55
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.722707657413062e-08,
+      "loss": 2.2416,
+      "step": 60
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.4713240590690364e-06,
+      "loss": 2.036,
+      "step": 65
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.1434665150939459e-05,
+      "loss": 1.9217,
+      "step": 70
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.3426119396138327e-05,
+      "loss": 1.8372,
+      "step": 75
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.8600717816522507e-05,
+      "loss": 1.784,
+      "step": 80
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.5889193527623905e-05,
+      "loss": 1.8846,
+      "step": 85
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.407332684094147e-05,
+      "loss": 1.8889,
+      "step": 90
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.187178621569686e-05,
+      "loss": 1.8718,
+      "step": 95
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00010803041634070827,
+      "loss": 1.8726,
+      "step": 100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012141061126329156,
+      "loss": 1.8162,
+      "step": 105
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00013106954541784705,
+      "loss": 1.7323,
+      "step": 110
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00013632660913388424,
+      "loss": 1.8048,
+      "step": 115
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001368113672944154,
+      "loss": 1.9135,
+      "step": 120
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00013248966177323044,
+      "loss": 1.726,
+      "step": 125
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00012366601836206413,
+      "loss": 1.6579,
+      "step": 130
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00011096218858530879,
+      "loss": 1.6411,
+      "step": 135
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.527333843746984e-05,
+      "loss": 1.8306,
+      "step": 140
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 7.77049711716633e-05,
+      "loss": 1.8107,
+      "step": 145
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.949502882833675e-05,
+      "loss": 1.7814,
+      "step": 150
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.192666156253025e-05,
+      "loss": 1.654,
+      "step": 155
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.6237811414691256e-05,
+      "loss": 1.7283,
+      "step": 160
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3533981637935892e-05,
+      "loss": 1.8214,
+      "step": 165
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.710338226769622e-06,
+      "loss": 1.7687,
+      "step": 170
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.886327055845878e-07,
+      "loss": 1.7183,
+      "step": 175
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 8.733908661157559e-07,
+      "loss": 1.6793,
+      "step": 180
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.130454582152937e-06,
+      "loss": 1.6161,
+      "step": 185
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.5789388736708423e-05,
+      "loss": 1.6901,
+      "step": 190
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.9169583659291692e-05,
+      "loss": 1.7063,
+      "step": 195
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.5328213784303035e-05,
+      "loss": 1.7222,
+      "step": 200
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.312667315905842e-05,
+      "loss": 1.7548,
+      "step": 205
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 8.1310806472376e-05,
+      "loss": 1.7133,
+      "step": 210
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.859928218347747e-05,
+      "loss": 1.5574,
+      "step": 215
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00011377388060386165,
+      "loss": 1.6825,
+      "step": 220
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012576533484906052,
+      "loss": 1.6971,
+      "step": 225
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00013372867594093092,
+      "loss": 1.6711,
+      "step": 230
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00013710277292342587,
+      "loss": 1.6385,
+      "step": 235
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00013564987254150566,
+      "loss": 1.6699,
+      "step": 240
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00012947235234888086,
+      "loss": 1.7898,
+      "step": 245
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011900550674920642,
+      "loss": 1.578,
+      "step": 250
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00010498687429701432,
+      "loss": 1.6301,
+      "step": 255
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 8.840426758749807e-05,
+      "loss": 1.5338,
+      "step": 260
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 7.042616777063153e-05,
+      "loss": 1.6108,
+      "step": 265
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 5.231938838884156e-05,
+      "loss": 1.5748,
+      "step": 270
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.535981029518021e-05,
+      "loss": 1.647,
+      "step": 275
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.0742477650140126e-05,
+      "loss": 1.5056,
+      "step": 280
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.497390016384942e-06,
+      "loss": 1.5544,
+      "step": 285
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.416924199324192e-06,
+      "loss": 1.6663,
+      "step": 290
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 1.5545,
+      "step": 295
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.416924199324169e-06,
+      "loss": 1.5904,
+      "step": 300
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.497390016384903e-06,
+      "loss": 1.4718,
+      "step": 305
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.074247765013998e-05,
+      "loss": 1.6686,
+      "step": 310
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.535981029518024e-05,
+      "loss": 1.4802,
+      "step": 315
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 5.231938838884147e-05,
+      "loss": 1.5853,
+      "step": 320
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 7.042616777063145e-05,
+      "loss": 1.4866,
+      "step": 325
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.8404267587498e-05,
+      "loss": 1.5326,
+      "step": 330
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00010498687429701424,
+      "loss": 1.5453,
+      "step": 335
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00011900550674920627,
+      "loss": 1.5737,
+      "step": 340
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00012947235234888078,
+      "loss": 1.5864,
+      "step": 345
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.00013564987254150568,
+      "loss": 1.5359,
+      "step": 350
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.00013710277292342587,
+      "loss": 1.5031,
+      "step": 355
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.000133728675940931,
+      "loss": 1.5171,
+      "step": 360
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00012576533484906052,
+      "loss": 1.4829,
+      "step": 365
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00011377388060386172,
+      "loss": 1.535,
+      "step": 370
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 9.859928218347764e-05,
+      "loss": 1.6041,
+      "step": 375
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 8.131080647237608e-05,
+      "loss": 1.5994,
+      "step": 380
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 6.31266731590584e-05,
+      "loss": 1.4118,
+      "step": 385
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 4.532821378430311e-05,
+      "loss": 1.5561,
+      "step": 390
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.916958365929176e-05,
+      "loss": 1.469,
+      "step": 395
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.578938873670855e-05,
+      "loss": 1.522,
+      "step": 400
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 6.130454582152975e-06,
+      "loss": 1.4166,
+      "step": 405
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 8.733908661157864e-07,
+      "loss": 1.365,
+      "step": 410
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 3.8863270558459543e-07,
+      "loss": 1.4054,
+      "step": 415
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.7103382267695455e-06,
+      "loss": 1.4166,
+      "step": 420
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.3533981637935914e-05,
+      "loss": 1.4485,
+      "step": 425
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.6237811414691188e-05,
+      "loss": 1.4459,
+      "step": 430
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 4.192666156253006e-05,
+      "loss": 1.4346,
+      "step": 435
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 5.9495028828336655e-05,
+      "loss": 1.3644,
+      "step": 440
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 7.770497117166322e-05,
+      "loss": 1.5092,
+      "step": 445
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 9.527333843746983e-05,
+      "loss": 1.4805,
+      "step": 450
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.00011096218858530872,
+      "loss": 1.4888,
+      "step": 455
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.00012366601836206402,
+      "loss": 1.3819,
+      "step": 460
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.0001324896617732304,
+      "loss": 1.4309,
+      "step": 465
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.0001368113672944154,
+      "loss": 1.4564,
+      "step": 470
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 0.00013632660913388424,
+      "loss": 1.4574,
+      "step": 475
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 0.00013106954541784708,
+      "loss": 1.449,
+      "step": 480
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 0.00012141061126329153,
+      "loss": 1.3888,
+      "step": 485
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 0.00010803041634070854,
+      "loss": 1.3585,
+      "step": 490
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 9.1871786215697e-05,
+      "loss": 1.3667,
+      "step": 495
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 7.40733268409415e-05,
+      "loss": 1.4703,
+      "step": 500
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 5.588919352762405e-05,
+      "loss": 1.4113,
+      "step": 505
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.860071781652248e-05,
+      "loss": 1.447,
+      "step": 510
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.3426119396138205e-05,
+      "loss": 1.4628,
+      "step": 515
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.1434665150939559e-05,
+      "loss": 1.3236,
+      "step": 520
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 3.471324059069044e-06,
+      "loss": 1.3998,
+      "step": 525
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 9.722707657414585e-08,
+      "loss": 1.3596,
+      "step": 530
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.550127458494308e-06,
+      "loss": 1.4176,
+      "step": 535
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 7.727647651119162e-06,
+      "loss": 1.3178,
+      "step": 540
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.8194493250793477e-05,
+      "loss": 1.2094,
+      "step": 545
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.221312570298565e-05,
+      "loss": 1.2475,
+      "step": 550
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 4.879573241250212e-05,
+      "loss": 1.3111,
+      "step": 555
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 6.677383222936843e-05,
+      "loss": 1.3479,
+      "step": 560
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 8.488061161115841e-05,
+      "loss": 1.3577,
+      "step": 565
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 0.00010184018970481966,
+      "loss": 1.3625,
+      "step": 570
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 0.00011645752234985993,
+      "loss": 1.4285,
+      "step": 575
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 0.00012770260998361515,
+      "loss": 1.3736,
+      "step": 580
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 0.00013478307580067574,
+      "loss": 1.3182,
+      "step": 585
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 0.0001372,
+      "loss": 1.3859,
+      "step": 590
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 0.00013478307580067577,
+      "loss": 1.3133,
+      "step": 595
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 0.00012770260998361523,
+      "loss": 1.2928,
+      "step": 600
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 0.00011645752234986005,
+      "loss": 1.3042,
+      "step": 605
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 0.00010184018970481979,
+      "loss": 1.2487,
+      "step": 610
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 8.488061161115881e-05,
+      "loss": 1.3463,
+      "step": 615
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 6.677383222936884e-05,
+      "loss": 1.3703,
+      "step": 620
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.879573241250204e-05,
+      "loss": 1.3632,
+      "step": 625
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.221312570298559e-05,
+      "loss": 1.3018,
+      "step": 630
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.8194493250793422e-05,
+      "loss": 1.309,
+      "step": 635
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 7.727647651119238e-06,
+      "loss": 1.2896,
+      "step": 640
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.5501274584943384e-06,
+      "loss": 1.2617,
+      "step": 645
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.722707657413823e-08,
+      "loss": 1.2505,
+      "step": 650
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.4713240590689907e-06,
+      "loss": 1.2368,
+      "step": 655
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.1434665150939459e-05,
+      "loss": 1.2241,
+      "step": 660
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.3426119396138076e-05,
+      "loss": 1.2006,
+      "step": 665
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.8600717816522324e-05,
+      "loss": 1.2346,
+      "step": 670
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 5.5889193527623885e-05,
+      "loss": 1.2934,
+      "step": 675
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 7.407332684094109e-05,
+      "loss": 1.2511,
+      "step": 680
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 9.187178621569662e-05,
+      "loss": 1.291,
+      "step": 685
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 0.0001080304163407082,
+      "loss": 1.2377,
+      "step": 690
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 0.00012141061126329158,
+      "loss": 1.3313,
+      "step": 695
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 0.0001310695454178471,
+      "loss": 1.2796,
+      "step": 700
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 0.0001363266091338842,
+      "loss": 1.172,
+      "step": 705
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 0.0001368113672944154,
+      "loss": 1.1964,
+      "step": 710
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 0.00013248966177323038,
+      "loss": 1.1786,
+      "step": 715
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 0.00012366601836206426,
+      "loss": 1.2633,
+      "step": 720
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 0.00011096218858530884,
+      "loss": 1.3484,
+      "step": 725
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 9.527333843746975e-05,
+      "loss": 1.2342,
+      "step": 730
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 7.770497117166362e-05,
+      "loss": 1.2577,
+      "step": 735
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 5.9495028828336825e-05,
+      "loss": 1.1856,
+      "step": 740
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.192666156253067e-05,
+      "loss": 1.1636,
+      "step": 745
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.6237811414691517e-05,
+      "loss": 1.2436,
+      "step": 750
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 1.3533981637936021e-05,
+      "loss": 1.1413,
+      "step": 755
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.7103382267696065e-06,
+      "loss": 1.2202,
+      "step": 760
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 3.886327055845878e-07,
+      "loss": 1.1975,
+      "step": 765
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 8.73390866115794e-07,
+      "loss": 1.2147,
+      "step": 770
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 6.130454582152906e-06,
+      "loss": 1.1828,
+      "step": 775
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.5789388736708446e-05,
+      "loss": 1.1669,
+      "step": 780
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.9169583659291425e-05,
+      "loss": 1.0396,
+      "step": 785
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.5328213784302954e-05,
+      "loss": 1.144,
+      "step": 790
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 6.312667315905846e-05,
+      "loss": 1.1023,
+      "step": 795
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 8.131080647237566e-05,
+      "loss": 1.159,
+      "step": 800
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 9.859928218347728e-05,
+      "loss": 1.1336,
+      "step": 805
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 0.00011377388060386158,
+      "loss": 1.1809,
+      "step": 810
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 0.00012576533484906028,
+      "loss": 1.1879,
+      "step": 815
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 0.00013372867594093086,
+      "loss": 1.2067,
+      "step": 820
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 0.00013710277292342587,
+      "loss": 1.1995,
+      "step": 825
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 0.00013564987254150566,
+      "loss": 1.0728,
+      "step": 830
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 0.00012947235234888075,
+      "loss": 1.1226,
+      "step": 835
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 0.00011900550674920657,
+      "loss": 1.0724,
+      "step": 840
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 0.00010498687429701438,
+      "loss": 1.2117,
+      "step": 845
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 8.840426758749793e-05,
+      "loss": 1.1809,
+      "step": 850
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 7.042616777063161e-05,
+      "loss": 1.1632,
+      "step": 855
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 5.2319388388841635e-05,
+      "loss": 1.1441,
+      "step": 860
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 3.53598102951806e-05,
+      "loss": 1.0918,
+      "step": 865
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 2.0742477650140278e-05,
+      "loss": 1.1435,
+      "step": 870
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 9.497390016384988e-06,
+      "loss": 1.1574,
+      "step": 875
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 2.416924199324215e-06,
+      "loss": 1.137,
+      "step": 880
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 0.0,
+      "loss": 1.1283,
+      "step": 885
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 2.4169241993241463e-06,
+      "loss": 1.1263,
+      "step": 890
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 9.497390016384857e-06,
+      "loss": 1.0131,
+      "step": 895
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 2.0742477650140095e-05,
+      "loss": 1.0513,
+      "step": 900
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 3.535981029518038e-05,
+      "loss": 0.9842,
+      "step": 905
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 5.231938838884139e-05,
+      "loss": 1.0041,
+      "step": 910
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 7.042616777063135e-05,
+      "loss": 1.0598,
+      "step": 915
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 8.840426758749768e-05,
+      "loss": 0.9953,
+      "step": 920
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 0.00010498687429701418,
+      "loss": 1.1262,
+      "step": 925
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 0.00011900550674920638,
+      "loss": 1.1427,
+      "step": 930
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 0.00012947235234888062,
+      "loss": 1.0945,
+      "step": 935
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 0.0001356498725415056,
+      "loss": 1.1189,
+      "step": 940
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 0.00013710277292342587,
+      "loss": 1.1328,
+      "step": 945
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 0.00013372867594093094,
+      "loss": 0.9658,
+      "step": 950
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 0.00012576533484906068,
+      "loss": 1.0325,
+      "step": 955
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 0.00011377388060386179,
+      "loss": 1.0981,
+      "step": 960
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 9.85992821834775e-05,
+      "loss": 1.0437,
+      "step": 965
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 8.131080647237592e-05,
+      "loss": 1.0232,
+      "step": 970
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 6.312667315905823e-05,
+      "loss": 1.0634,
+      "step": 975
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 4.532821378430365e-05,
+      "loss": 1.17,
+      "step": 980
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 2.9169583659292028e-05,
+      "loss": 1.0492,
+      "step": 985
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 1.5789388736708606e-05,
+      "loss": 1.0114,
+      "step": 990
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 6.130454582153005e-06,
+      "loss": 1.0472,
+      "step": 995
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 8.733908661157635e-07,
+      "loss": 0.9773,
+      "step": 1000
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 3.886327055846107e-07,
+      "loss": 1.0456,
+      "step": 1005
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 4.710338226769515e-06,
+      "loss": 1.0436,
+      "step": 1010
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 1.3533981637935869e-05,
+      "loss": 0.916,
+      "step": 1015
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 2.6237811414691317e-05,
+      "loss": 0.9274,
+      "step": 1020
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 4.1926661562530436e-05,
+      "loss": 1.0476,
+      "step": 1025
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 5.949502882833706e-05,
+      "loss": 1.0242,
+      "step": 1030
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 7.770497117166289e-05,
+      "loss": 1.001,
+      "step": 1035
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 9.527333843746952e-05,
+      "loss": 0.9579,
+      "step": 1040
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 0.00011096218858530865,
+      "loss": 0.9389,
+      "step": 1045
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 0.0001236660183620641,
+      "loss": 0.9494,
+      "step": 1050
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 0.00013248966177323046,
+      "loss": 1.0333,
+      "step": 1055
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 0.00013681136729441534,
+      "loss": 0.991,
+      "step": 1060
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 0.00013632660913388432,
+      "loss": 1.0615,
+      "step": 1065
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 0.0001310695454178472,
+      "loss": 1.0309,
+      "step": 1070
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 0.00012141061126329175,
+      "loss": 0.9265,
+      "step": 1075
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 0.00010803041634070842,
+      "loss": 1.0156,
+      "step": 1080
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 9.187178621569685e-05,
+      "loss": 0.8855,
+      "step": 1085
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 7.407332684094182e-05,
+      "loss": 0.9234,
+      "step": 1090
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 5.588919352762413e-05,
+      "loss": 1.0271,
+      "step": 1095
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 3.8600717816522554e-05,
+      "loss": 0.8368,
+      "step": 1100
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 2.3426119396138266e-05,
+      "loss": 0.9153,
+      "step": 1105
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 1.143466515093933e-05,
+      "loss": 0.9869,
+      "step": 1110
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 3.471324059069227e-06,
+      "loss": 0.9458,
+      "step": 1115
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 9.722707657415346e-08,
+      "loss": 0.9596,
+      "step": 1120
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 1.5501274584942853e-06,
+      "loss": 0.9193,
+      "step": 1125
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 7.727647651119123e-06,
+      "loss": 0.8474,
+      "step": 1130
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 1.8194493250793585e-05,
+      "loss": 0.9306,
+      "step": 1135
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 3.221312570298537e-05,
+      "loss": 0.8185,
+      "step": 1140
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 4.879573241250181e-05,
+      "loss": 0.8653,
+      "step": 1145
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 6.67738322293686e-05,
+      "loss": 0.8867,
+      "step": 1150
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 8.488061161115857e-05,
+      "loss": 0.9633,
+      "step": 1155
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 0.00010184018970482001,
+      "loss": 0.9252,
+      "step": 1160
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 0.00011645752234986022,
+      "loss": 0.8119,
+      "step": 1165
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 0.00012770260998361488,
+      "loss": 0.9919,
+      "step": 1170
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 0.00013478307580067571,
+      "loss": 1.0657,
+      "step": 1175
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 0.0001372,
+      "loss": 0.8867,
+      "step": 1180
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 0.0001347830758006758,
+      "loss": 0.8722,
+      "step": 1185
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 0.00012770260998361504,
+      "loss": 0.9027,
+      "step": 1190
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 0.00011645752234986047,
+      "loss": 0.9142,
+      "step": 1195
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 0.00010184018970482031,
+      "loss": 0.9368,
+      "step": 1200
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 8.488061161115841e-05,
+      "loss": 0.9328,
+      "step": 1205
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 6.677383222936893e-05,
+      "loss": 0.9455,
+      "step": 1210
+    },
+    {
+      "epoch": 20.59,
+      "learning_rate": 4.879573241250166e-05,
+      "loss": 0.8107,
+      "step": 1215
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 3.221312570298566e-05,
+      "loss": 0.8213,
+      "step": 1220
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 1.8194493250793812e-05,
+      "loss": 0.8629,
+      "step": 1225
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 7.727647651119504e-06,
+      "loss": 0.8399,
+      "step": 1230
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 1.5501274584943613e-06,
+      "loss": 0.8685,
+      "step": 1235
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 9.722707657410776e-08,
+      "loss": 0.8757,
+      "step": 1240
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 3.4713240590691203e-06,
+      "loss": 0.7704,
+      "step": 1245
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 1.1434665150939413e-05,
+      "loss": 0.773,
+      "step": 1250
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 2.3426119396138015e-05,
+      "loss": 0.768,
+      "step": 1255
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 3.860071781652269e-05,
+      "loss": 0.8515,
+      "step": 1260
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 5.5889193527623803e-05,
+      "loss": 0.7945,
+      "step": 1265
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 7.407332684094197e-05,
+      "loss": 0.8015,
+      "step": 1270
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 9.187178621569608e-05,
+      "loss": 0.8248,
+      "step": 1275
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 0.00010803041634070815,
+      "loss": 0.8062,
+      "step": 1280
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 0.00012141061126329121,
+      "loss": 0.85,
+      "step": 1285
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 0.00013106954541784708,
+      "loss": 0.8513,
+      "step": 1290
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 0.00013632660913388418,
+      "loss": 0.9707,
+      "step": 1295
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 0.00013681136729441537,
+      "loss": 0.8592,
+      "step": 1300
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 0.0001324896617732304,
+      "loss": 0.8644,
+      "step": 1305
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 0.00012366601836206432,
+      "loss": 0.8634,
+      "step": 1310
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 0.00011096218858530853,
+      "loss": 0.8125,
+      "step": 1315
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 9.527333843746984e-05,
+      "loss": 0.864,
+      "step": 1320
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 7.770497117166273e-05,
+      "loss": 0.8875,
+      "step": 1325
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 5.9495028828337394e-05,
+      "loss": 0.7048,
+      "step": 1330
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 4.1926661562530294e-05,
+      "loss": 0.8041,
+      "step": 1335
+    },
+    {
+      "epoch": 22.71,
+      "learning_rate": 2.6237811414691578e-05,
+      "loss": 0.8016,
+      "step": 1340
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 1.3533981637935777e-05,
+      "loss": 0.7686,
+      "step": 1345
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 4.710338226769637e-06,
+      "loss": 0.7585,
+      "step": 1350
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 3.8863270558464874e-07,
+      "loss": 0.837,
+      "step": 1355
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 8.733908661156265e-07,
+      "loss": 0.8028,
+      "step": 1360
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 6.130454582152869e-06,
+      "loss": 0.6584,
+      "step": 1365
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 1.578938873670808e-05,
+      "loss": 0.6132,
+      "step": 1370
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 2.9169583659291753e-05,
+      "loss": 0.693,
+      "step": 1375
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 4.532821378430287e-05,
+      "loss": 0.7615,
+      "step": 1380
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 6.31266731590579e-05,
+      "loss": 0.7913,
+      "step": 1385
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 8.131080647237607e-05,
+      "loss": 0.7448,
+      "step": 1390
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 9.859928218347718e-05,
+      "loss": 0.7617,
+      "step": 1395
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 0.0001137738806038619,
+      "loss": 0.719,
+      "step": 1400
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 0.00012576533484906052,
+      "loss": 0.8524,
+      "step": 1405
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 0.00013372867594093084,
+      "loss": 0.8465,
+      "step": 1410
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 0.00013710277292342584,
+      "loss": 0.8268,
+      "step": 1415
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 0.00013564987254150568,
+      "loss": 0.6952,
+      "step": 1420
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 0.000129472352348881,
+      "loss": 0.7045,
+      "step": 1425
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 0.0001190055067492063,
+      "loss": 0.8174,
+      "step": 1430
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 0.00010498687429701445,
+      "loss": 0.8089,
+      "step": 1435
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 8.840426758749845e-05,
+      "loss": 0.732,
+      "step": 1440
+    },
+    {
+      "epoch": 24.49,
+      "learning_rate": 7.04261677706312e-05,
+      "loss": 0.7616,
+      "step": 1445
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 5.2319388388841716e-05,
+      "loss": 0.7403,
+      "step": 1450
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 3.5359810295179815e-05,
+      "loss": 0.7289,
+      "step": 1455
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 2.0742477650139987e-05,
+      "loss": 0.7526,
+      "step": 1460
+    },
+    {
+      "epoch": 24.83,
+      "learning_rate": 9.497390016385033e-06,
+      "loss": 0.6211,
+      "step": 1465
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 2.4169241993243597e-06,
+      "loss": 0.7174,
+      "step": 1470
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 0.0,
+      "loss": 0.7111,
+      "step": 1475
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 2.4169241993241234e-06,
+      "loss": 0.6603,
+      "step": 1480
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 9.497390016385064e-06,
+      "loss": 0.6645,
+      "step": 1485
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 2.0742477650139333e-05,
+      "loss": 0.6387,
+      "step": 1490
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 3.5359810295179876e-05,
+      "loss": 0.6582,
+      "step": 1495
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 5.2319388388840835e-05,
+      "loss": 0.7156,
+      "step": 1500
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 7.042616777063128e-05,
+      "loss": 0.6633,
+      "step": 1505
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 8.84042675874976e-05,
+      "loss": 0.6631,
+      "step": 1510
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 0.0001049868742970145,
+      "loss": 0.7089,
+      "step": 1515
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 0.00011900550674920632,
+      "loss": 0.6112,
+      "step": 1520
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 0.0001294723523488806,
+      "loss": 0.7087,
+      "step": 1525
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 0.00013564987254150568,
+      "loss": 0.7079,
+      "step": 1530
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 0.00013710277292342587,
+      "loss": 0.7044,
+      "step": 1535
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 0.0001337286759409308,
+      "loss": 0.6848,
+      "step": 1540
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 0.000125765334849061,
+      "loss": 0.6899,
+      "step": 1545
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 0.00011377388060386184,
+      "loss": 0.6688,
+      "step": 1550
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.859928218347801e-05,
+      "loss": 0.6701,
+      "step": 1555
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.1310806472376e-05,
+      "loss": 0.6325,
+      "step": 1560
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 6.31266731590588e-05,
+      "loss": 0.723,
+      "step": 1565
+    },
+    {
+      "epoch": 26.61,
+      "learning_rate": 4.5328213784303727e-05,
+      "loss": 0.6405,
+      "step": 1570
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 2.91695836592917e-05,
+      "loss": 0.693,
+      "step": 1575
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 1.5789388736708656e-05,
+      "loss": 0.7138,
+      "step": 1580
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 6.130454582152846e-06,
+      "loss": 0.6322,
+      "step": 1585
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 8.733908661157712e-07,
+      "loss": 0.6351,
+      "step": 1590
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 3.8863270558464874e-07,
+      "loss": 0.7019,
+      "step": 1595
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 4.710338226769309e-06,
+      "loss": 0.5841,
+      "step": 1600
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 1.3533981637935816e-05,
+      "loss": 0.6115,
+      "step": 1605
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 2.623781141469087e-05,
+      "loss": 0.6035,
+      "step": 1610
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 4.1926661562530355e-05,
+      "loss": 0.563,
+      "step": 1615
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 5.949502882833649e-05,
+      "loss": 0.5205,
+      "step": 1620
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 7.770497117166281e-05,
+      "loss": 0.6091,
+      "step": 1625
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 9.5273338437469e-05,
+      "loss": 0.5453,
+      "step": 1630
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 0.00011096218858530857,
+      "loss": 0.6827,
+      "step": 1635
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 0.00012366601836206377,
+      "loss": 0.6791,
+      "step": 1640
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 0.00013248966177323044,
+      "loss": 0.6586,
+      "step": 1645
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 0.00013681136729441537,
+      "loss": 0.6288,
+      "step": 1650
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 0.00013632660913388432,
+      "loss": 0.6334,
+      "step": 1655
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 0.00013106954541784705,
+      "loss": 0.5685,
+      "step": 1660
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 0.00012141061126329179,
+      "loss": 0.6338,
+      "step": 1665
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 0.00010803041634070808,
+      "loss": 0.6357,
+      "step": 1670
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 9.187178621569693e-05,
+      "loss": 0.6479,
+      "step": 1675
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 7.407332684094189e-05,
+      "loss": 0.6516,
+      "step": 1680
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 5.588919352762469e-05,
+      "loss": 0.5449,
+      "step": 1685
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 3.860071781652263e-05,
+      "loss": 0.5642,
+      "step": 1690
+    },
+    {
+      "epoch": 28.73,
+      "learning_rate": 2.34261193961387e-05,
+      "loss": 0.5598,
+      "step": 1695
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 1.1434665150939376e-05,
+      "loss": 0.5501,
+      "step": 1700
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 3.4713240590690974e-06,
+      "loss": 0.6496,
+      "step": 1705
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 9.722707657416107e-08,
+      "loss": 0.6054,
+      "step": 1710
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 1.550127458494369e-06,
+      "loss": 0.5238,
+      "step": 1715
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 7.727647651119086e-06,
+      "loss": 0.5338,
+      "step": 1720
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 1.8194493250793856e-05,
+      "loss": 0.5257,
+      "step": 1725
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 3.2213125702984884e-05,
+      "loss": 0.4891,
+      "step": 1730
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 4.8795732412501724e-05,
+      "loss": 0.5434,
+      "step": 1735
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 6.677383222936803e-05,
+      "loss": 0.4987,
+      "step": 1740
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 8.488061161115847e-05,
+      "loss": 0.5118,
+      "step": 1745
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 0.0001018401897048195,
+      "loss": 0.5495,
+      "step": 1750
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 0.00011645752234986016,
+      "loss": 0.5927,
+      "step": 1755
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 0.00012770260998361507,
+      "loss": 0.6367,
+      "step": 1760
+    },
+    {
+      "epoch": 29.92,
+      "learning_rate": 0.0001347830758006757,
+      "loss": 0.6231,
+      "step": 1765
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 0.0001372,
+      "loss": 0.5553,
+      "step": 1770
+    },
+    {
+      "epoch": 30.08,
+      "learning_rate": 0.00013478307580067582,
+      "loss": 0.6387,
+      "step": 1775
+    },
+    {
+      "epoch": 30.17,
+      "learning_rate": 0.0001277026099836153,
+      "loss": 0.5631,
+      "step": 1780
+    },
+    {
+      "epoch": 30.25,
+      "learning_rate": 0.00011645752234986052,
+      "loss": 0.5864,
+      "step": 1785
+    },
+    {
+      "epoch": 30.34,
+      "learning_rate": 0.00010184018970481994,
+      "loss": 0.5652,
+      "step": 1790
+    },
+    {
+      "epoch": 30.42,
+      "learning_rate": 8.488061161115898e-05,
+      "loss": 0.5867,
+      "step": 1795
+    },
+    {
+      "epoch": 30.51,
+      "learning_rate": 6.677383222936853e-05,
+      "loss": 0.5299,
+      "step": 1800
+    },
+    {
+      "epoch": 30.59,
+      "learning_rate": 4.879573241250221e-05,
+      "loss": 0.4777,
+      "step": 1805
+    },
+    {
+      "epoch": 30.68,
+      "learning_rate": 3.221312570298531e-05,
+      "loss": 0.5385,
+      "step": 1810
+    },
+    {
+      "epoch": 30.76,
+      "learning_rate": 1.8194493250794202e-05,
+      "loss": 0.5095,
+      "step": 1815
+    },
+    {
+      "epoch": 30.85,
+      "learning_rate": 7.727647651119314e-06,
+      "loss": 0.5411,
+      "step": 1820
+    },
+    {
+      "epoch": 30.93,
+      "learning_rate": 1.5501274584944757e-06,
+      "loss": 0.4905,
+      "step": 1825
+    },
+    {
+      "epoch": 31.02,
+      "learning_rate": 9.722707657413062e-08,
+      "loss": 0.4989,
+      "step": 1830
+    },
+    {
+      "epoch": 31.1,
+      "learning_rate": 3.4713240590689373e-06,
+      "loss": 0.4341,
+      "step": 1835
+    },
+    {
+      "epoch": 31.19,
+      "learning_rate": 1.1434665150939101e-05,
+      "loss": 0.4691,
+      "step": 1840
+    },
+    {
+      "epoch": 31.27,
+      "learning_rate": 2.3426119396138317e-05,
+      "loss": 0.4898,
+      "step": 1845
+    },
+    {
+      "epoch": 31.36,
+      "learning_rate": 3.8600717816522174e-05,
+      "loss": 0.4434,
+      "step": 1850
+    },
+    {
+      "epoch": 31.44,
+      "learning_rate": 5.5889193527624196e-05,
+      "loss": 0.4916,
+      "step": 1855
+    },
+    {
+      "epoch": 31.53,
+      "learning_rate": 7.40733268409414e-05,
+      "loss": 0.4605,
+      "step": 1860
+    },
+    {
+      "epoch": 31.61,
+      "learning_rate": 9.187178621569647e-05,
+      "loss": 0.4772,
+      "step": 1865
+    },
+    {
+      "epoch": 31.69,
+      "learning_rate": 0.00010803041634070767,
+      "loss": 0.4936,
+      "step": 1870
+    },
+    {
+      "epoch": 31.78,
+      "learning_rate": 0.00012141061126329147,
+      "loss": 0.5163,
+      "step": 1875
+    },
+    {
+      "epoch": 31.86,
+      "learning_rate": 0.00013106954541784683,
+      "loss": 0.494,
+      "step": 1880
+    },
+    {
+      "epoch": 31.95,
+      "learning_rate": 0.00013632660913388427,
+      "loss": 0.5058,
+      "step": 1885
+    },
+    {
+      "epoch": 32.03,
+      "learning_rate": 0.00013681136729441542,
+      "loss": 0.5373,
+      "step": 1890
+    },
+    {
+      "epoch": 32.12,
+      "learning_rate": 0.00013248966177323063,
+      "loss": 0.532,
+      "step": 1895
+    },
+    {
+      "epoch": 32.2,
+      "learning_rate": 0.00012366601836206407,
+      "loss": 0.4622,
+      "step": 1900
+    },
+    {
+      "epoch": 32.29,
+      "learning_rate": 0.00011096218858530896,
+      "loss": 0.5211,
+      "step": 1905
+    },
+    {
+      "epoch": 32.37,
+      "learning_rate": 9.527333843747036e-05,
+      "loss": 0.4867,
+      "step": 1910
+    },
+    {
+      "epoch": 32.46,
+      "learning_rate": 7.770497117166331e-05,
+      "loss": 0.5372,
+      "step": 1915
+    },
+    {
+      "epoch": 32.54,
+      "learning_rate": 5.9495028828336994e-05,
+      "loss": 0.4491,
+      "step": 1920
+    },
+    {
+      "epoch": 32.63,
+      "learning_rate": 4.1926661562529915e-05,
+      "loss": 0.514,
+      "step": 1925
+    },
+    {
+      "epoch": 32.71,
+      "learning_rate": 2.6237811414691263e-05,
+      "loss": 0.4827,
+      "step": 1930
+    },
+    {
+      "epoch": 32.8,
+      "learning_rate": 1.3533981637936699e-05,
+      "loss": 0.5054,
+      "step": 1935
+    },
+    {
+      "epoch": 32.88,
+      "learning_rate": 4.710338226769492e-06,
+      "loss": 0.5075,
+      "step": 1940
+    },
+    {
+      "epoch": 32.97,
+      "learning_rate": 3.886327055847097e-07,
+      "loss": 0.4416,
+      "step": 1945
+    },
+    {
+      "epoch": 33.05,
+      "learning_rate": 8.733908661158474e-07,
+      "loss": 0.3942,
+      "step": 1950
+    },
+    {
+      "epoch": 33.14,
+      "learning_rate": 6.130454582152633e-06,
+      "loss": 0.4056,
+      "step": 1955
+    },
+    {
+      "epoch": 33.22,
+      "learning_rate": 1.5789388736707714e-05,
+      "loss": 0.4238,
+      "step": 1960
+    },
+    {
+      "epoch": 33.31,
+      "learning_rate": 2.916958365929129e-05,
+      "loss": 0.4191,
+      "step": 1965
+    },
+    {
+      "epoch": 33.39,
+      "learning_rate": 4.5328213784302344e-05,
+      "loss": 0.4243,
+      "step": 1970
+    },
+    {
+      "epoch": 33.47,
+      "learning_rate": 6.31266731590583e-05,
+      "loss": 0.4473,
+      "step": 1975
+    },
+    {
+      "epoch": 33.56,
+      "learning_rate": 8.131080647237551e-05,
+      "loss": 0.4325,
+      "step": 1980
+    },
+    {
+      "epoch": 33.64,
+      "learning_rate": 9.859928218347668e-05,
+      "loss": 0.4047,
+      "step": 1985
+    },
+    {
+      "epoch": 33.73,
+      "learning_rate": 0.00011377388060386145,
+      "loss": 0.4184,
+      "step": 1990
+    },
+    {
+      "epoch": 33.81,
+      "learning_rate": 0.0001257653348490602,
+      "loss": 0.4285,
+      "step": 1995
+    },
+    {
+      "epoch": 33.9,
+      "learning_rate": 0.00013372867594093094,
+      "loss": 0.5034,
+      "step": 2000
+    },
+    {
+      "epoch": 33.98,
+      "learning_rate": 0.00013710277292342587,
+      "loss": 0.5398,
+      "step": 2005
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 0.0001356498725415056,
+      "loss": 0.4968,
+      "step": 2010
+    },
+    {
+      "epoch": 34.15,
+      "learning_rate": 0.00012947235234888084,
+      "loss": 0.4584,
+      "step": 2015
+    },
+    {
+      "epoch": 34.24,
+      "learning_rate": 0.00011900550674920667,
+      "loss": 0.4254,
+      "step": 2020
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 0.00010498687429701411,
+      "loss": 0.472,
+      "step": 2025
+    },
+    {
+      "epoch": 34.41,
+      "learning_rate": 8.840426758749807e-05,
+      "loss": 0.4033,
+      "step": 2030
+    },
+    {
+      "epoch": 34.49,
+      "learning_rate": 7.042616777063081e-05,
+      "loss": 0.4878,
+      "step": 2035
+    },
+    {
+      "epoch": 34.58,
+      "learning_rate": 5.231938838884132e-05,
+      "loss": 0.3808,
+      "step": 2040
+    },
+    {
+      "epoch": 34.66,
+      "learning_rate": 3.535981029518117e-05,
+      "loss": 0.4661,
+      "step": 2045
+    },
+    {
+      "epoch": 34.75,
+      "learning_rate": 2.07424776501397e-05,
+      "loss": 0.4579,
+      "step": 2050
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 9.497390016385323e-06,
+      "loss": 0.4363,
+      "step": 2055
+    },
+    {
+      "epoch": 34.92,
+      "learning_rate": 2.416924199324002e-06,
+      "loss": 0.3991,
+      "step": 2060
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 0.0,
+      "loss": 0.3743,
+      "step": 2065
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 2.4169241993239714e-06,
+      "loss": 0.3955,
+      "step": 2070
+    },
+    {
+      "epoch": 35.17,
+      "learning_rate": 9.497390016384774e-06,
+      "loss": 0.3526,
+      "step": 2075
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 2.074247765013962e-05,
+      "loss": 0.3186,
+      "step": 2080
+    },
+    {
+      "epoch": 35.34,
+      "learning_rate": 3.535981029518023e-05,
+      "loss": 0.361,
+      "step": 2085
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 5.231938838884122e-05,
+      "loss": 0.3574,
+      "step": 2090
+    },
+    {
+      "epoch": 35.51,
+      "learning_rate": 7.04261677706307e-05,
+      "loss": 0.3685,
+      "step": 2095
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 8.840426758749799e-05,
+      "loss": 0.3853,
+      "step": 2100
+    },
+    {
+      "epoch": 35.68,
+      "learning_rate": 0.00010498687429701402,
+      "loss": 0.3843,
+      "step": 2105
+    },
+    {
+      "epoch": 35.76,
+      "learning_rate": 0.00011900550674920659,
+      "loss": 0.4264,
+      "step": 2110
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 0.00012947235234888078,
+      "loss": 0.439,
+      "step": 2115
+    },
+    {
+      "epoch": 35.93,
+      "learning_rate": 0.00013564987254150536,
+      "loss": 0.4012,
+      "step": 2120
+    },
+    {
+      "epoch": 36.02,
+      "learning_rate": 0.00013710277292342587,
+      "loss": 0.4393,
+      "step": 2125
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 0.0001337286759409313,
+      "loss": 0.3709,
+      "step": 2130
+    },
+    {
+      "epoch": 36.19,
+      "learning_rate": 0.00012576533484906025,
+      "loss": 0.3955,
+      "step": 2135
+    },
+    {
+      "epoch": 36.27,
+      "learning_rate": 0.00011377388060386227,
+      "loss": 0.4227,
+      "step": 2140
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 9.859928218347676e-05,
+      "loss": 0.4189,
+      "step": 2145
+    },
+    {
+      "epoch": 36.44,
+      "learning_rate": 8.131080647237657e-05,
+      "loss": 0.362,
+      "step": 2150
+    },
+    {
+      "epoch": 36.53,
+      "learning_rate": 6.312667315905937e-05,
+      "loss": 0.4009,
+      "step": 2155
+    },
+    {
+      "epoch": 36.61,
+      "learning_rate": 4.532821378430335e-05,
+      "loss": 0.3754,
+      "step": 2160
+    },
+    {
+      "epoch": 36.69,
+      "learning_rate": 2.9169583659292163e-05,
+      "loss": 0.4087,
+      "step": 2165
+    },
+    {
+      "epoch": 36.78,
+      "learning_rate": 1.57893887367084e-05,
+      "loss": 0.3784,
+      "step": 2170
+    },
+    {
+      "epoch": 36.86,
+      "learning_rate": 6.1304545821530815e-06,
+      "loss": 0.3498,
+      "step": 2175
+    },
+    {
+      "epoch": 36.95,
+      "learning_rate": 8.733908661158625e-07,
+      "loss": 0.3977,
+      "step": 2180
+    },
+    {
+      "epoch": 37.03,
+      "learning_rate": 3.8863270558459543e-07,
+      "loss": 0.3301,
+      "step": 2185
+    },
+    {
+      "epoch": 37.12,
+      "learning_rate": 4.710338226769454e-06,
+      "loss": 0.3393,
+      "step": 2190
+    },
+    {
+      "epoch": 37.2,
+      "learning_rate": 1.353398163793606e-05,
+      "loss": 0.336,
+      "step": 2195
+    },
+    {
+      "epoch": 37.29,
+      "learning_rate": 2.623781141469118e-05,
+      "loss": 0.304,
+      "step": 2200
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 4.1926661562528925e-05,
+      "loss": 0.3332,
+      "step": 2205
+    },
+    {
+      "epoch": 37.46,
+      "learning_rate": 5.949502882833689e-05,
+      "loss": 0.3265,
+      "step": 2210
+    },
+    {
+      "epoch": 37.54,
+      "learning_rate": 7.770497117166225e-05,
+      "loss": 0.2978,
+      "step": 2215
+    },
+    {
+      "epoch": 37.63,
+      "learning_rate": 9.527333843747026e-05,
+      "loss": 0.3156,
+      "step": 2220
+    },
+    {
+      "epoch": 37.71,
+      "learning_rate": 0.00011096218858530812,
+      "loss": 0.3648,
+      "step": 2225
+    },
+    {
+      "epoch": 37.8,
+      "learning_rate": 0.00012366601836206342,
+      "loss": 0.3573,
+      "step": 2230
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 0.00013248966177323022,
+      "loss": 0.3774,
+      "step": 2235
+    },
+    {
+      "epoch": 37.97,
+      "learning_rate": 0.00013681136729441532,
+      "loss": 0.3772,
+      "step": 2240
+    },
+    {
+      "epoch": 38.05,
+      "learning_rate": 0.00013632660913388427,
+      "loss": 0.3823,
+      "step": 2245
+    },
+    {
+      "epoch": 38.14,
+      "learning_rate": 0.0001310695454178473,
+      "loss": 0.328,
+      "step": 2250
+    },
+    {
+      "epoch": 38.22,
+      "learning_rate": 0.00012141061126329216,
+      "loss": 0.3486,
+      "step": 2255
+    },
+    {
+      "epoch": 38.31,
+      "learning_rate": 0.00010803041634070855,
+      "loss": 0.3668,
+      "step": 2260
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 9.187178621569747e-05,
+      "loss": 0.3798,
+      "step": 2265
+    },
+    {
+      "epoch": 38.47,
+      "learning_rate": 7.40733268409415e-05,
+      "loss": 0.4158,
+      "step": 2270
+    },
+    {
+      "epoch": 38.56,
+      "learning_rate": 5.58891935276243e-05,
+      "loss": 0.3535,
+      "step": 2275
+    },
+    {
+      "epoch": 38.64,
+      "learning_rate": 3.8600717816523143e-05,
+      "loss": 0.3979,
+      "step": 2280
+    },
+    {
+      "epoch": 38.73,
+      "learning_rate": 2.3426119396138395e-05,
+      "loss": 0.3366,
+      "step": 2285
+    },
+    {
+      "epoch": 38.81,
+      "learning_rate": 1.1434665150939696e-05,
+      "loss": 0.3252,
+      "step": 2290
+    },
+    {
+      "epoch": 38.9,
+      "learning_rate": 3.471324059068968e-06,
+      "loss": 0.2936,
+      "step": 2295
+    },
+    {
+      "epoch": 38.98,
+      "learning_rate": 9.722707657413823e-08,
+      "loss": 0.3053,
+      "step": 2300
+    },
+    {
+      "epoch": 39.07,
+      "learning_rate": 1.5501274584944604e-06,
+      "loss": 0.2954,
+      "step": 2305
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 7.727647651119269e-06,
+      "loss": 0.3342,
+      "step": 2310
+    },
+    {
+      "epoch": 39.24,
+      "learning_rate": 1.8194493250792806e-05,
+      "loss": 0.2785,
+      "step": 2315
+    },
+    {
+      "epoch": 39.32,
+      "learning_rate": 3.221312570298606e-05,
+      "loss": 0.2683,
+      "step": 2320
+    },
+    {
+      "epoch": 39.41,
+      "learning_rate": 4.8795732412501175e-05,
+      "loss": 0.2651,
+      "step": 2325
+    },
+    {
+      "epoch": 39.49,
+      "learning_rate": 6.67738322293694e-05,
+      "loss": 0.2762,
+      "step": 2330
+    },
+    {
+      "epoch": 39.58,
+      "learning_rate": 8.488061161115792e-05,
+      "loss": 0.3293,
+      "step": 2335
+    },
+    {
+      "epoch": 39.66,
+      "learning_rate": 0.00010184018970481899,
+      "loss": 0.33,
+      "step": 2340
+    },
+    {
+      "epoch": 39.75,
+      "learning_rate": 0.00011645752234985975,
+      "loss": 0.3273,
+      "step": 2345
+    },
+    {
+      "epoch": 39.83,
+      "learning_rate": 0.00012770260998361477,
+      "loss": 0.3042,
+      "step": 2350
+    },
+    {
+      "epoch": 39.92,
+      "learning_rate": 0.0001347830758006758,
+      "loss": 0.3306,
+      "step": 2355
+    },
+    {
+      "epoch": 40.0,
+      "learning_rate": 0.0001372,
+      "loss": 0.3255,
+      "step": 2360
+    },
+    {
+      "epoch": 40.08,
+      "learning_rate": 0.00013478307580067598,
+      "loss": 0.3174,
+      "step": 2365
+    },
+    {
+      "epoch": 40.17,
+      "learning_rate": 0.00012770260998361512,
+      "loss": 0.3026,
+      "step": 2370
+    },
+    {
+      "epoch": 40.25,
+      "learning_rate": 0.00011645752234986024,
+      "loss": 0.3501,
+      "step": 2375
+    },
+    {
+      "epoch": 40.34,
+      "learning_rate": 0.00010184018970481959,
+      "loss": 0.3238,
+      "step": 2380
+    },
+    {
+      "epoch": 40.42,
+      "learning_rate": 8.488061161115858e-05,
+      "loss": 0.308,
+      "step": 2385
+    },
+    {
+      "epoch": 40.51,
+      "learning_rate": 6.677383222937007e-05,
+      "loss": 0.3262,
+      "step": 2390
+    },
+    {
+      "epoch": 40.59,
+      "learning_rate": 4.8795732412501825e-05,
+      "loss": 0.3656,
+      "step": 2395
+    },
+    {
+      "epoch": 40.68,
+      "learning_rate": 3.221312570298663e-05,
+      "loss": 0.2899,
+      "step": 2400
+    },
+    {
+      "epoch": 40.76,
+      "learning_rate": 1.8194493250793927e-05,
+      "loss": 0.3111,
+      "step": 2405
+    },
+    {
+      "epoch": 40.85,
+      "learning_rate": 7.727647651119131e-06,
+      "loss": 0.3155,
+      "step": 2410
+    },
+    {
+      "epoch": 40.93,
+      "learning_rate": 1.5501274584945974e-06,
+      "loss": 0.245,
+      "step": 2415
+    },
+    {
+      "epoch": 41.02,
+      "learning_rate": 9.722707657410015e-08,
+      "loss": 0.2667,
+      "step": 2420
+    },
+    {
+      "epoch": 41.1,
+      "learning_rate": 3.471324059069067e-06,
+      "loss": 0.2208,
+      "step": 2425
+    },
+    {
+      "epoch": 41.19,
+      "learning_rate": 1.1434665150939863e-05,
+      "loss": 0.2605,
+      "step": 2430
+    },
+    {
+      "epoch": 41.27,
+      "learning_rate": 2.3426119396137883e-05,
+      "loss": 0.2173,
+      "step": 2435
+    },
+    {
+      "epoch": 41.36,
+      "learning_rate": 3.860071781652254e-05,
+      "loss": 0.2611,
+      "step": 2440
+    },
+    {
+      "epoch": 41.44,
+      "learning_rate": 5.5889193527624596e-05,
+      "loss": 0.235,
+      "step": 2445
+    },
+    {
+      "epoch": 41.53,
+      "learning_rate": 7.407332684094083e-05,
+      "loss": 0.2731,
+      "step": 2450
+    },
+    {
+      "epoch": 41.61,
+      "learning_rate": 9.187178621569684e-05,
+      "loss": 0.2705,
+      "step": 2455
+    },
+    {
+      "epoch": 41.69,
+      "learning_rate": 0.0001080304163407072,
+      "loss": 0.2805,
+      "step": 2460
+    },
+    {
+      "epoch": 41.78,
+      "learning_rate": 0.00012141061126329236,
+      "loss": 0.2565,
+      "step": 2465
+    },
+    {
+      "epoch": 41.86,
+      "learning_rate": 0.000131069545417847,
+      "loss": 0.3185,
+      "step": 2470
+    },
+    {
+      "epoch": 41.95,
+      "learning_rate": 0.00013632660913388402,
+      "loss": 0.3083,
+      "step": 2475
+    },
+    {
+      "epoch": 42.03,
+      "learning_rate": 0.0001368113672944155,
+      "loss": 0.3283,
+      "step": 2480
+    },
+    {
+      "epoch": 42.12,
+      "learning_rate": 0.00013248966177323046,
+      "loss": 0.282,
+      "step": 2485
+    },
+    {
+      "epoch": 42.2,
+      "learning_rate": 0.00012366601836206383,
+      "loss": 0.2761,
+      "step": 2490
+    },
+    {
+      "epoch": 42.29,
+      "learning_rate": 0.00011096218858530942,
+      "loss": 0.2853,
+      "step": 2495
+    },
+    {
+      "epoch": 42.37,
+      "learning_rate": 9.527333843746999e-05,
+      "loss": 0.3187,
+      "step": 2500
+    },
+    {
+      "epoch": 42.46,
+      "learning_rate": 7.77049711716629e-05,
+      "loss": 0.3139,
+      "step": 2505
+    },
+    {
+      "epoch": 42.54,
+      "learning_rate": 5.949502882833756e-05,
+      "loss": 0.2639,
+      "step": 2510
+    },
+    {
+      "epoch": 42.63,
+      "learning_rate": 4.192666156253045e-05,
+      "loss": 0.2431,
+      "step": 2515
+    },
+    {
+      "epoch": 42.71,
+      "learning_rate": 2.6237811414690944e-05,
+      "loss": 0.259,
+      "step": 2520
+    },
+    {
+      "epoch": 42.8,
+      "learning_rate": 1.3533981637936456e-05,
+      "loss": 0.276,
+      "step": 2525
+    },
+    {
+      "epoch": 42.88,
+      "learning_rate": 4.710338226769698e-06,
+      "loss": 0.2709,
+      "step": 2530
+    },
+    {
+      "epoch": 42.97,
+      "learning_rate": 3.886327055847706e-07,
+      "loss": 0.2713,
+      "step": 2535
+    },
+    {
+      "epoch": 43.05,
+      "learning_rate": 8.733908661159159e-07,
+      "loss": 0.2457,
+      "step": 2540
+    },
+    {
+      "epoch": 43.14,
+      "learning_rate": 6.1304545821528e-06,
+      "loss": 0.2204,
+      "step": 2545
+    },
+    {
+      "epoch": 43.22,
+      "learning_rate": 1.578938873670735e-05,
+      "loss": 0.2285,
+      "step": 2550
+    },
+    {
+      "epoch": 43.31,
+      "learning_rate": 2.9169583659290815e-05,
+      "loss": 0.2334,
+      "step": 2555
+    },
+    {
+      "epoch": 43.39,
+      "learning_rate": 4.532821378430272e-05,
+      "loss": 0.2263,
+      "step": 2560
+    },
+    {
+      "epoch": 43.47,
+      "learning_rate": 6.31266731590587e-05,
+      "loss": 0.2353,
+      "step": 2565
+    },
+    {
+      "epoch": 43.56,
+      "learning_rate": 8.131080647237496e-05,
+      "loss": 0.2352,
+      "step": 2570
+    },
+    {
+      "epoch": 43.64,
+      "learning_rate": 9.859928218347703e-05,
+      "loss": 0.2328,
+      "step": 2575
+    },
+    {
+      "epoch": 43.73,
+      "learning_rate": 0.00011377388060386175,
+      "loss": 0.2444,
+      "step": 2580
+    },
+    {
+      "epoch": 43.81,
+      "learning_rate": 0.0001257653348490599,
+      "loss": 0.2268,
+      "step": 2585
+    },
+    {
+      "epoch": 43.9,
+      "learning_rate": 0.00013372867594093078,
+      "loss": 0.2535,
+      "step": 2590
+    },
+    {
+      "epoch": 43.98,
+      "learning_rate": 0.00013710277292342587,
+      "loss": 0.2902,
+      "step": 2595
+    },
+    {
+      "epoch": 44.07,
+      "learning_rate": 0.0001356498725415055,
+      "loss": 0.2786,
+      "step": 2600
+    },
+    {
+      "epoch": 44.15,
+      "learning_rate": 0.00012947235234888108,
+      "loss": 0.2634,
+      "step": 2605
+    },
+    {
+      "epoch": 44.24,
+      "learning_rate": 0.00011900550674920639,
+      "loss": 0.2493,
+      "step": 2610
+    },
+    {
+      "epoch": 44.32,
+      "learning_rate": 0.00010498687429701377,
+      "loss": 0.2681,
+      "step": 2615
+    },
+    {
+      "epoch": 44.41,
+      "learning_rate": 8.840426758749863e-05,
+      "loss": 0.2863,
+      "step": 2620
+    },
+    {
+      "epoch": 44.49,
+      "learning_rate": 7.042616777063136e-05,
+      "loss": 0.2442,
+      "step": 2625
+    },
+    {
+      "epoch": 44.58,
+      "learning_rate": 5.231938838884094e-05,
+      "loss": 0.2379,
+      "step": 2630
+    },
+    {
+      "epoch": 44.66,
+      "learning_rate": 3.535981029518082e-05,
+      "loss": 0.2382,
+      "step": 2635
+    },
+    {
+      "epoch": 44.75,
+      "learning_rate": 2.074247765014011e-05,
+      "loss": 0.24,
+      "step": 2640
+    },
+    {
+      "epoch": 44.83,
+      "learning_rate": 9.497390016385613e-06,
+      "loss": 0.217,
+      "step": 2645
+    },
+    {
+      "epoch": 44.92,
+      "learning_rate": 2.416924199323895e-06,
+      "loss": 0.2417,
+      "step": 2650
+    },
+    {
+      "epoch": 45.0,
+      "learning_rate": 0.0,
+      "loss": 0.2366,
+      "step": 2655
+    },
+    {
+      "epoch": 45.08,
+      "learning_rate": 2.4169241993238265e-06,
+      "loss": 0.207,
+      "step": 2660
+    },
+    {
+      "epoch": 45.17,
+      "learning_rate": 9.497390016384484e-06,
+      "loss": 0.1959,
+      "step": 2665
+    },
+    {
+      "epoch": 45.25,
+      "learning_rate": 2.0742477650139912e-05,
+      "loss": 0.2013,
+      "step": 2670
+    },
+    {
+      "epoch": 45.34,
+      "learning_rate": 3.535981029518058e-05,
+      "loss": 0.1915,
+      "step": 2675
+    },
+    {
+      "epoch": 45.42,
+      "learning_rate": 5.231938838884067e-05,
+      "loss": 0.2045,
+      "step": 2680
+    },
+    {
+      "epoch": 45.51,
+      "learning_rate": 7.042616777063111e-05,
+      "loss": 0.2009,
+      "step": 2685
+    },
+    {
+      "epoch": 45.59,
+      "learning_rate": 8.840426758749837e-05,
+      "loss": 0.2131,
+      "step": 2690
+    },
+    {
+      "epoch": 45.68,
+      "learning_rate": 0.00010498687429701354,
+      "loss": 0.2117,
+      "step": 2695
+    },
+    {
+      "epoch": 45.76,
+      "learning_rate": 0.00011900550674920621,
+      "loss": 0.2232,
+      "step": 2700
+    },
+    {
+      "epoch": 45.85,
+      "learning_rate": 0.00012947235234888097,
+      "loss": 0.2327,
+      "step": 2705
+    },
+    {
+      "epoch": 45.93,
+      "learning_rate": 0.00013564987254150544,
+      "loss": 0.2621,
+      "step": 2710
+    },
+    {
+      "epoch": 46.02,
+      "learning_rate": 0.0001371027729234259,
+      "loss": 0.2101,
+      "step": 2715
+    },
+    {
+      "epoch": 46.1,
+      "learning_rate": 0.0001337286759409315,
+      "loss": 0.2214,
+      "step": 2720
+    },
+    {
+      "epoch": 46.19,
+      "learning_rate": 0.00012576533484906003,
+      "loss": 0.2503,
+      "step": 2725
+    },
+    {
+      "epoch": 46.27,
+      "learning_rate": 0.00011377388060386196,
+      "loss": 0.216,
+      "step": 2730
+    },
+    {
+      "epoch": 46.36,
+      "learning_rate": 9.859928218347728e-05,
+      "loss": 0.2487,
+      "step": 2735
+    },
+    {
+      "epoch": 46.44,
+      "learning_rate": 8.131080647237713e-05,
+      "loss": 0.2208,
+      "step": 2740
+    },
+    {
+      "epoch": 46.53,
+      "learning_rate": 6.312667315905896e-05,
+      "loss": 0.2208,
+      "step": 2745
+    },
+    {
+      "epoch": 46.61,
+      "learning_rate": 4.532821378430297e-05,
+      "loss": 0.2331,
+      "step": 2750
+    },
+    {
+      "epoch": 46.69,
+      "learning_rate": 2.9169583659292638e-05,
+      "loss": 0.2307,
+      "step": 2755
+    },
+    {
+      "epoch": 46.78,
+      "learning_rate": 1.5789388736708765e-05,
+      "loss": 0.2043,
+      "step": 2760
+    },
+    {
+      "epoch": 46.86,
+      "learning_rate": 6.130454582152914e-06,
+      "loss": 0.2309,
+      "step": 2765
+    },
+    {
+      "epoch": 46.95,
+      "learning_rate": 8.73390866115954e-07,
+      "loss": 0.211,
+      "step": 2770
+    },
+    {
+      "epoch": 47.03,
+      "learning_rate": 3.886327055845345e-07,
+      "loss": 0.2177,
+      "step": 2775
+    },
+    {
+      "epoch": 47.12,
+      "learning_rate": 4.710338226769599e-06,
+      "loss": 0.1786,
+      "step": 2780
+    },
+    {
+      "epoch": 47.2,
+      "learning_rate": 1.3533981637936296e-05,
+      "loss": 0.1993,
+      "step": 2785
+    },
+    {
+      "epoch": 47.29,
+      "learning_rate": 2.623781141469073e-05,
+      "loss": 0.1859,
+      "step": 2790
+    },
+    {
+      "epoch": 47.37,
+      "learning_rate": 4.1926661562528403e-05,
+      "loss": 0.1549,
+      "step": 2795
+    },
+    {
+      "epoch": 47.46,
+      "learning_rate": 5.94950288283373e-05,
+      "loss": 0.1598,
+      "step": 2800
+    },
+    {
+      "epoch": 47.54,
+      "learning_rate": 7.770497117166265e-05,
+      "loss": 0.1624,
+      "step": 2805
+    },
+    {
+      "epoch": 47.63,
+      "learning_rate": 9.527333843746975e-05,
+      "loss": 0.1806,
+      "step": 2810
+    },
+    {
+      "epoch": 47.71,
+      "learning_rate": 0.00011096218858530769,
+      "loss": 0.219,
+      "step": 2815
+    },
+    {
+      "epoch": 47.8,
+      "learning_rate": 0.00012366601836206366,
+      "loss": 0.2425,
+      "step": 2820
+    },
+    {
+      "epoch": 47.88,
+      "learning_rate": 0.00013248966177323036,
+      "loss": 0.1936,
+      "step": 2825
+    },
+    {
+      "epoch": 47.97,
+      "learning_rate": 0.00013681136729441526,
+      "loss": 0.2217,
+      "step": 2830
+    },
+    {
+      "epoch": 48.05,
+      "learning_rate": 0.00013632660913388435,
+      "loss": 0.2092,
+      "step": 2835
+    },
+    {
+      "epoch": 48.14,
+      "learning_rate": 0.0001310695454178471,
+      "loss": 0.2336,
+      "step": 2840
+    },
+    {
+      "epoch": 48.22,
+      "learning_rate": 0.00012141061126329252,
+      "loss": 0.225,
+      "step": 2845
+    },
+    {
+      "epoch": 48.31,
+      "learning_rate": 0.00010803041634070903,
+      "loss": 0.2351,
+      "step": 2850
+    },
+    {
+      "epoch": 48.39,
+      "learning_rate": 9.18717862156971e-05,
+      "loss": 0.1936,
+      "step": 2855
+    },
+    {
+      "epoch": 48.47,
+      "learning_rate": 7.40733268409411e-05,
+      "loss": 0.2027,
+      "step": 2860
+    },
+    {
+      "epoch": 48.56,
+      "learning_rate": 5.588919352762486e-05,
+      "loss": 0.1996,
+      "step": 2865
+    },
+    {
+      "epoch": 48.64,
+      "learning_rate": 3.860071781652278e-05,
+      "loss": 0.2031,
+      "step": 2870
+    },
+    {
+      "epoch": 48.73,
+      "learning_rate": 2.342611939613809e-05,
+      "loss": 0.1813,
+      "step": 2875
+    },
+    {
+      "epoch": 48.81,
+      "learning_rate": 1.1434665150940007e-05,
+      "loss": 0.1967,
+      "step": 2880
+    },
+    {
+      "epoch": 48.9,
+      "learning_rate": 3.4713240590691508e-06,
+      "loss": 0.1893,
+      "step": 2885
+    },
+    {
+      "epoch": 48.98,
+      "learning_rate": 9.722707657411538e-08,
+      "loss": 0.1778,
+      "step": 2890
+    },
+    {
+      "epoch": 49.07,
+      "learning_rate": 1.550127458494544e-06,
+      "loss": 0.1728,
+      "step": 2895
+    },
+    {
+      "epoch": 49.15,
+      "learning_rate": 7.72764765111901e-06,
+      "loss": 0.1562,
+      "step": 2900
+    },
+    {
+      "epoch": 49.24,
+      "learning_rate": 1.8194493250792426e-05,
+      "loss": 0.1671,
+      "step": 2905
+    },
+    {
+      "epoch": 49.32,
+      "learning_rate": 3.22131257029864e-05,
+      "loss": 0.162,
+      "step": 2910
+    },
+    {
+      "epoch": 49.41,
+      "learning_rate": 4.879573241250157e-05,
+      "loss": 0.1621,
+      "step": 2915
+    },
+    {
+      "epoch": 49.49,
+      "learning_rate": 6.677383222936883e-05,
+      "loss": 0.1595,
+      "step": 2920
+    },
+    {
+      "epoch": 49.58,
+      "learning_rate": 8.488061161115738e-05,
+      "loss": 0.1635,
+      "step": 2925
+    },
+    {
+      "epoch": 49.66,
+      "learning_rate": 0.00010184018970481936,
+      "loss": 0.1674,
+      "step": 2930
+    },
+    {
+      "epoch": 49.75,
+      "learning_rate": 0.00011645752234986003,
+      "loss": 0.1613,
+      "step": 2935
+    },
+    {
+      "epoch": 49.83,
+      "learning_rate": 0.0001277026099836145,
+      "loss": 0.188,
+      "step": 2940
+    },
+    {
+      "epoch": 49.92,
+      "learning_rate": 0.00013478307580067566,
+      "loss": 0.1844,
+      "step": 2945
+    },
+    {
+      "epoch": 50.0,
+      "learning_rate": 0.0001372,
+      "loss": 0.1913,
+      "step": 2950
+    },
+    {
+      "epoch": 50.08,
+      "learning_rate": 0.00013478307580067612,
+      "loss": 0.1985,
+      "step": 2955
+    },
+    {
+      "epoch": 50.17,
+      "learning_rate": 0.00012770260998361542,
+      "loss": 0.1961,
+      "step": 2960
+    },
+    {
+      "epoch": 50.25,
+      "learning_rate": 0.00011645752234985994,
+      "loss": 0.2094,
+      "step": 2965
+    },
+    {
+      "epoch": 50.34,
+      "learning_rate": 0.00010184018970481924,
+      "loss": 0.1942,
+      "step": 2970
+    },
+    {
+      "epoch": 50.42,
+      "learning_rate": 8.488061161115913e-05,
+      "loss": 0.1878,
+      "step": 2975
+    },
+    {
+      "epoch": 50.51,
+      "learning_rate": 6.677383222937064e-05,
+      "loss": 0.1811,
+      "step": 2980
+    },
+    {
+      "epoch": 50.59,
+      "learning_rate": 4.879573241250144e-05,
+      "loss": 0.1911,
+      "step": 2985
+    },
+    {
+      "epoch": 50.68,
+      "learning_rate": 3.221312570298629e-05,
+      "loss": 0.1697,
+      "step": 2990
+    },
+    {
+      "epoch": 50.76,
+      "learning_rate": 1.8194493250793653e-05,
+      "loss": 0.1849,
+      "step": 2995
+    },
+    {
+      "epoch": 50.85,
+      "learning_rate": 7.727647651119846e-06,
+      "loss": 0.1565,
+      "step": 3000
+    },
+    {
+      "epoch": 50.93,
+      "learning_rate": 1.5501274584945138e-06,
+      "loss": 0.1677,
+      "step": 3005
+    },
+    {
+      "epoch": 51.02,
+      "learning_rate": 9.7227076574123e-08,
+      "loss": 0.1654,
+      "step": 3010
+    },
+    {
+      "epoch": 51.1,
+      "learning_rate": 3.4713240590685794e-06,
+      "loss": 0.1566,
+      "step": 3015
+    },
+    {
+      "epoch": 51.19,
+      "learning_rate": 1.143466515093901e-05,
+      "loss": 0.1451,
+      "step": 3020
+    },
+    {
+      "epoch": 51.27,
+      "learning_rate": 2.3426119396138188e-05,
+      "loss": 0.1363,
+      "step": 3025
+    },
+    {
+      "epoch": 51.36,
+      "learning_rate": 3.86007178165229e-05,
+      "loss": 0.1452,
+      "step": 3030
+    },
+    {
+      "epoch": 51.44,
+      "learning_rate": 5.588919352762307e-05,
+      "loss": 0.1347,
+      "step": 3035
+    },
+    {
+      "epoch": 51.53,
+      "learning_rate": 7.407332684094123e-05,
+      "loss": 0.1572,
+      "step": 3040
+    },
+    {
+      "epoch": 51.61,
+      "learning_rate": 9.187178621569722e-05,
+      "loss": 0.1435,
+      "step": 3045
+    },
+    {
+      "epoch": 51.69,
+      "learning_rate": 0.00010803041634070754,
+      "loss": 0.1511,
+      "step": 3050
+    },
+    {
+      "epoch": 51.78,
+      "learning_rate": 0.00012141061126329137,
+      "loss": 0.1698,
+      "step": 3055
+    },
+    {
+      "epoch": 51.86,
+      "learning_rate": 0.00013106954541784719,
+      "loss": 0.1696,
+      "step": 3060
+    },
+    {
+      "epoch": 51.95,
+      "learning_rate": 0.00013632660913388408,
+      "loss": 0.1841,
+      "step": 3065
+    },
+    {
+      "epoch": 52.03,
+      "learning_rate": 0.00013681136729441545,
+      "loss": 0.1958,
+      "step": 3070
+    },
+    {
+      "epoch": 52.12,
+      "learning_rate": 0.00013248966177323033,
+      "loss": 0.1647,
+      "step": 3075
+    },
+    {
+      "epoch": 52.2,
+      "learning_rate": 0.00012366601836206358,
+      "loss": 0.1537,
+      "step": 3080
+    },
+    {
+      "epoch": 52.29,
+      "learning_rate": 0.00011096218858530911,
+      "loss": 0.1952,
+      "step": 3085
+    },
+    {
+      "epoch": 52.37,
+      "learning_rate": 9.527333843747141e-05,
+      "loss": 0.1742,
+      "step": 3090
+    },
+    {
+      "epoch": 52.46,
+      "learning_rate": 7.770497117166251e-05,
+      "loss": 0.1563,
+      "step": 3095
+    },
+    {
+      "epoch": 52.54,
+      "learning_rate": 5.949502882833716e-05,
+      "loss": 0.1734,
+      "step": 3100
+    },
+    {
+      "epoch": 52.63,
+      "learning_rate": 4.192666156253007e-05,
+      "loss": 0.1661,
+      "step": 3105
+    },
+    {
+      "epoch": 52.71,
+      "learning_rate": 2.6237811414692164e-05,
+      "loss": 0.1631,
+      "step": 3110
+    },
+    {
+      "epoch": 52.8,
+      "learning_rate": 1.353398163793622e-05,
+      "loss": 0.1648,
+      "step": 3115
+    },
+    {
+      "epoch": 52.88,
+      "learning_rate": 4.710338226769553e-06,
+      "loss": 0.1587,
+      "step": 3120
+    },
+    {
+      "epoch": 52.97,
+      "learning_rate": 3.8863270558472487e-07,
+      "loss": 0.1612,
+      "step": 3125
+    },
+    {
+      "epoch": 53.05,
+      "learning_rate": 8.733908661156646e-07,
+      "loss": 0.1572,
+      "step": 3130
+    },
+    {
+      "epoch": 53.14,
+      "learning_rate": 6.130454582152967e-06,
+      "loss": 0.1406,
+      "step": 3135
+    },
+    {
+      "epoch": 53.22,
+      "learning_rate": 1.5789388736707606e-05,
+      "loss": 0.1475,
+      "step": 3140
+    },
+    {
+      "epoch": 53.31,
+      "learning_rate": 2.916958365929115e-05,
+      "loss": 0.1196,
+      "step": 3145
+    },
+    {
+      "epoch": 53.39,
+      "learning_rate": 4.5328213784303096e-05,
+      "loss": 0.1415,
+      "step": 3150
+    },
+    {
+      "epoch": 53.47,
+      "learning_rate": 6.31266731590591e-05,
+      "loss": 0.1302,
+      "step": 3155
+    },
+    {
+      "epoch": 53.56,
+      "learning_rate": 8.131080647237535e-05,
+      "loss": 0.1331,
+      "step": 3160
+    },
+    {
+      "epoch": 53.64,
+      "learning_rate": 9.859928218347565e-05,
+      "loss": 0.1277,
+      "step": 3165
+    },
+    {
+      "epoch": 53.73,
+      "learning_rate": 0.00011377388060386206,
+      "loss": 0.1383,
+      "step": 3170
+    },
+    {
+      "epoch": 53.81,
+      "learning_rate": 0.0001257653348490601,
+      "loss": 0.1486,
+      "step": 3175
+    },
+    {
+      "epoch": 53.9,
+      "learning_rate": 0.00013372867594093092,
+      "loss": 0.1657,
+      "step": 3180
+    },
+    {
+      "epoch": 53.98,
+      "learning_rate": 0.0001371027729234258,
+      "loss": 0.1587,
+      "step": 3185
+    },
+    {
+      "epoch": 54.07,
+      "learning_rate": 0.0001356498725415054,
+      "loss": 0.1642,
+      "step": 3190
+    },
+    {
+      "epoch": 54.15,
+      "learning_rate": 0.0001294723523488809,
+      "loss": 0.1701,
+      "step": 3195
+    },
+    {
+      "epoch": 54.24,
+      "learning_rate": 0.00011900550674920745,
+      "loss": 0.1712,
+      "step": 3200
+    },
+    {
+      "epoch": 54.32,
+      "learning_rate": 0.00010498687429701509,
+      "loss": 0.1495,
+      "step": 3205
+    },
+    {
+      "epoch": 54.41,
+      "learning_rate": 8.840426758749825e-05,
+      "loss": 0.1551,
+      "step": 3210
+    },
+    {
+      "epoch": 54.49,
+      "learning_rate": 7.042616777063097e-05,
+      "loss": 0.1707,
+      "step": 3215
+    },
+    {
+      "epoch": 54.58,
+      "learning_rate": 5.2319388388842434e-05,
+      "loss": 0.1731,
+      "step": 3220
+    },
+    {
+      "epoch": 54.66,
+      "learning_rate": 3.535981029518047e-05,
+      "loss": 0.1576,
+      "step": 3225
+    },
+    {
+      "epoch": 54.75,
+      "learning_rate": 2.074247765013982e-05,
+      "loss": 0.1289,
+      "step": 3230
+    },
+    {
+      "epoch": 54.83,
+      "learning_rate": 9.497390016385406e-06,
+      "loss": 0.1336,
+      "step": 3235
+    },
+    {
+      "epoch": 54.92,
+      "learning_rate": 2.4169241993242987e-06,
+      "loss": 0.1407,
+      "step": 3240
+    },
+    {
+      "epoch": 55.0,
+      "learning_rate": 0.0,
+      "loss": 0.1442,
+      "step": 3245
+    },
+    {
+      "epoch": 55.08,
+      "learning_rate": 2.4169241993239332e-06,
+      "loss": 0.1158,
+      "step": 3250
+    },
+    {
+      "epoch": 55.17,
+      "learning_rate": 9.497390016384691e-06,
+      "loss": 0.1171,
+      "step": 3255
+    },
+    {
+      "epoch": 55.25,
+      "learning_rate": 2.0742477650138808e-05,
+      "loss": 0.1272,
+      "step": 3260
+    },
+    {
+      "epoch": 55.34,
+      "learning_rate": 3.535981029518094e-05,
+      "loss": 0.1239,
+      "step": 3265
+    },
+    {
+      "epoch": 55.42,
+      "learning_rate": 5.2319388388841065e-05,
+      "loss": 0.1224,
+      "step": 3270
+    },
+    {
+      "epoch": 55.51,
+      "learning_rate": 7.042616777062956e-05,
+      "loss": 0.1238,
+      "step": 3275
+    },
+    {
+      "epoch": 55.59,
+      "learning_rate": 8.84042675874969e-05,
+      "loss": 0.1305,
+      "step": 3280
+    },
+    {
+      "epoch": 55.68,
+      "learning_rate": 0.00010498687429701388,
+      "loss": 0.1418,
+      "step": 3285
+    },
+    {
+      "epoch": 55.76,
+      "learning_rate": 0.00011900550674920648,
+      "loss": 0.1285,
+      "step": 3290
+    },
+    {
+      "epoch": 55.85,
+      "learning_rate": 0.00012947235234888027,
+      "loss": 0.1404,
+      "step": 3295
+    },
+    {
+      "epoch": 55.93,
+      "learning_rate": 0.00013564987254150552,
+      "loss": 0.1524,
+      "step": 3300
+    },
+    {
+      "epoch": 56.02,
+      "learning_rate": 0.00013710277292342587,
+      "loss": 0.1522,
+      "step": 3305
+    },
+    {
+      "epoch": 56.1,
+      "learning_rate": 0.00013372867594093135,
+      "loss": 0.1521,
+      "step": 3310
+    },
+    {
+      "epoch": 56.19,
+      "learning_rate": 0.0001257653348490609,
+      "loss": 0.1473,
+      "step": 3315
+    },
+    {
+      "epoch": 56.27,
+      "learning_rate": 0.00011377388060386166,
+      "loss": 0.152,
+      "step": 3320
+    },
+    {
+      "epoch": 56.36,
+      "learning_rate": 9.859928218347691e-05,
+      "loss": 0.1528,
+      "step": 3325
+    },
+    {
+      "epoch": 56.44,
+      "learning_rate": 8.131080647237673e-05,
+      "loss": 0.1561,
+      "step": 3330
+    },
+    {
+      "epoch": 56.53,
+      "learning_rate": 6.312667315905857e-05,
+      "loss": 0.1366,
+      "step": 3335
+    },
+    {
+      "epoch": 56.61,
+      "learning_rate": 4.532821378430259e-05,
+      "loss": 0.1394,
+      "step": 3340
+    },
+    {
+      "epoch": 56.69,
+      "learning_rate": 2.9169583659292302e-05,
+      "loss": 0.1275,
+      "step": 3345
+    },
+    {
+      "epoch": 56.78,
+      "learning_rate": 1.5789388736708507e-05,
+      "loss": 0.1409,
+      "step": 3350
+    },
+    {
+      "epoch": 56.86,
+      "learning_rate": 6.130454582152747e-06,
+      "loss": 0.1339,
+      "step": 3355
+    },
+    {
+      "epoch": 56.95,
+      "learning_rate": 8.73390866115893e-07,
+      "loss": 0.1218,
+      "step": 3360
+    },
+    {
+      "epoch": 57.03,
+      "learning_rate": 3.8863270558457256e-07,
+      "loss": 0.1235,
+      "step": 3365
+    },
+    {
+      "epoch": 57.12,
+      "learning_rate": 4.710338226769036e-06,
+      "loss": 0.1012,
+      "step": 3370
+    },
+    {
+      "epoch": 57.2,
+      "learning_rate": 1.353398163793654e-05,
+      "loss": 0.1109,
+      "step": 3375
+    },
+    {
+      "epoch": 57.29,
+      "learning_rate": 2.6237811414691053e-05,
+      "loss": 0.117,
+      "step": 3380
+    },
+    {
+      "epoch": 57.37,
+      "learning_rate": 4.1926661562528776e-05,
+      "loss": 0.1091,
+      "step": 3385
+    },
+    {
+      "epoch": 57.46,
+      "learning_rate": 5.9495028828335754e-05,
+      "loss": 0.1172,
+      "step": 3390
+    },
+    {
+      "epoch": 57.54,
+      "learning_rate": 7.770497117166304e-05,
+      "loss": 0.1116,
+      "step": 3395
+    },
+    {
+      "epoch": 57.63,
+      "learning_rate": 9.527333843747011e-05,
+      "loss": 0.1244,
+      "step": 3400
+    },
+    {
+      "epoch": 57.71,
+      "learning_rate": 0.00011096218858530799,
+      "loss": 0.1134,
+      "step": 3405
+    },
+    {
+      "epoch": 57.8,
+      "learning_rate": 0.0001236660183620639,
+      "loss": 0.1238,
+      "step": 3410
+    },
+    {
+      "epoch": 57.88,
+      "learning_rate": 0.00013248966177323052,
+      "loss": 0.1272,
+      "step": 3415
+    },
+    {
+      "epoch": 57.97,
+      "learning_rate": 0.0001368113672944153,
+      "loss": 0.1467,
+      "step": 3420
+    },
+    {
+      "epoch": 58.05,
+      "learning_rate": 0.0001363266091338843,
+      "loss": 0.1349,
+      "step": 3425
+    },
+    {
+      "epoch": 58.14,
+      "learning_rate": 0.00013106954541784694,
+      "loss": 0.1379,
+      "step": 3430
+    },
+    {
+      "epoch": 58.22,
+      "learning_rate": 0.00012141061126329227,
+      "loss": 0.1474,
+      "step": 3435
+    },
+    {
+      "epoch": 58.31,
+      "learning_rate": 0.00010803041634070869,
+      "loss": 0.1547,
+      "step": 3440
+    },
+    {
+      "epoch": 58.39,
+      "learning_rate": 9.187178621569854e-05,
+      "loss": 0.1392,
+      "step": 3445
+    },
+    {
+      "epoch": 58.47,
+      "learning_rate": 7.40733268409407e-05,
+      "loss": 0.1571,
+      "step": 3450
+    },
+    {
+      "epoch": 58.56,
+      "learning_rate": 5.588919352762446e-05,
+      "loss": 0.1301,
+      "step": 3455
+    },
+    {
+      "epoch": 58.64,
+      "learning_rate": 3.860071781652417e-05,
+      "loss": 0.1265,
+      "step": 3460
+    },
+    {
+      "epoch": 58.73,
+      "learning_rate": 2.3426119396139255e-05,
+      "loss": 0.1262,
+      "step": 3465
+    },
+    {
+      "epoch": 58.81,
+      "learning_rate": 1.1434665150939787e-05,
+      "loss": 0.1344,
+      "step": 3470
+    },
+    {
+      "epoch": 58.9,
+      "learning_rate": 3.471324059069021e-06,
+      "loss": 0.1164,
+      "step": 3475
+    },
+    {
+      "epoch": 58.98,
+      "learning_rate": 9.722707657419916e-08,
+      "loss": 0.118,
+      "step": 3480
+    },
+    {
+      "epoch": 59.07,
+      "learning_rate": 1.5501274584942167e-06,
+      "loss": 0.117,
+      "step": 3485
+    },
+    {
+      "epoch": 59.15,
+      "learning_rate": 7.727647651119192e-06,
+      "loss": 0.1019,
+      "step": 3490
+    },
+    {
+      "epoch": 59.24,
+      "learning_rate": 1.8194493250792694e-05,
+      "loss": 0.1007,
+      "step": 3495
+    },
+    {
+      "epoch": 59.32,
+      "learning_rate": 3.2213125702985094e-05,
+      "loss": 0.1053,
+      "step": 3500
+    },
+    {
+      "epoch": 59.41,
+      "learning_rate": 4.8795732412501954e-05,
+      "loss": 0.0923,
+      "step": 3505
+    },
+    {
+      "epoch": 59.49,
+      "learning_rate": 6.677383222936923e-05,
+      "loss": 0.0997,
+      "step": 3510
+    },
+    {
+      "epoch": 59.58,
+      "learning_rate": 8.488061161115776e-05,
+      "loss": 0.091,
+      "step": 3515
+    },
+    {
+      "epoch": 59.66,
+      "learning_rate": 0.00010184018970481971,
+      "loss": 0.106,
+      "step": 3520
+    },
+    {
+      "epoch": 59.75,
+      "learning_rate": 0.00011645752234986032,
+      "loss": 0.12,
+      "step": 3525
+    },
+    {
+      "epoch": 59.83,
+      "learning_rate": 0.0001277026099836147,
+      "loss": 0.1184,
+      "step": 3530
+    },
+    {
+      "epoch": 59.92,
+      "learning_rate": 0.00013478307580067577,
+      "loss": 0.1445,
+      "step": 3535
+    },
+    {
+      "epoch": 60.0,
+      "learning_rate": 0.0001372,
+      "loss": 0.1241,
+      "step": 3540
+    },
+    {
+      "epoch": 60.08,
+      "learning_rate": 0.000134783075800676,
+      "loss": 0.1203,
+      "step": 3545
+    },
+    {
+      "epoch": 60.17,
+      "learning_rate": 0.0001277026099836152,
+      "loss": 0.1347,
+      "step": 3550
+    },
+    {
+      "epoch": 60.25,
+      "learning_rate": 0.00011645752234986105,
+      "loss": 0.1409,
+      "step": 3555
+    },
+    {
+      "epoch": 60.34,
+      "learning_rate": 0.00010184018970482059,
+      "loss": 0.125,
+      "step": 3560
+    },
+    {
+      "epoch": 60.42,
+      "learning_rate": 8.488061161115875e-05,
+      "loss": 0.128,
+      "step": 3565
+    },
+    {
+      "epoch": 60.51,
+      "learning_rate": 6.677383222937024e-05,
+      "loss": 0.131,
+      "step": 3570
+    },
+    {
+      "epoch": 60.59,
+      "learning_rate": 4.8795732412502916e-05,
+      "loss": 0.1158,
+      "step": 3575
+    },
+    {
+      "epoch": 60.68,
+      "learning_rate": 3.221312570298595e-05,
+      "loss": 0.1166,
+      "step": 3580
+    },
+    {
+      "epoch": 60.76,
+      "learning_rate": 1.819449325079338e-05,
+      "loss": 0.1199,
+      "step": 3585
+    },
+    {
+      "epoch": 60.85,
+      "learning_rate": 7.727647651119657e-06,
+      "loss": 0.1088,
+      "step": 3590
+    },
+    {
+      "epoch": 60.93,
+      "learning_rate": 1.55012745849443e-06,
+      "loss": 0.1035,
+      "step": 3595
+    },
+    {
+      "epoch": 61.02,
+      "learning_rate": 9.722707657414585e-08,
+      "loss": 0.1119,
+      "step": 3600
+    },
+    {
+      "epoch": 61.1,
+      "learning_rate": 3.471324059068709e-06,
+      "loss": 0.1008,
+      "step": 3605
+    },
+    {
+      "epoch": 61.19,
+      "learning_rate": 1.143466515093923e-05,
+      "loss": 0.0991,
+      "step": 3610
+    },
+    {
+      "epoch": 61.27,
+      "learning_rate": 2.3426119396138493e-05,
+      "loss": 0.0946,
+      "step": 3615
+    },
+    {
+      "epoch": 61.36,
+      "learning_rate": 3.860071781652325e-05,
+      "loss": 0.0931,
+      "step": 3620
+    },
+    {
+      "epoch": 61.44,
+      "learning_rate": 5.588919352762347e-05,
+      "loss": 0.0919,
+      "step": 3625
+    },
+    {
+      "epoch": 61.53,
+      "learning_rate": 7.407332684093968e-05,
+      "loss": 0.0903,
+      "step": 3630
+    },
+    {
+      "epoch": 61.61,
+      "learning_rate": 9.18717862156976e-05,
+      "loss": 0.0997,
+      "step": 3635
+    },
+    {
+      "epoch": 61.69,
+      "learning_rate": 0.00010803041634070786,
+      "loss": 0.0884,
+      "step": 3640
+    },
+    {
+      "epoch": 61.78,
+      "learning_rate": 0.00012141061126329163,
+      "loss": 0.1136,
+      "step": 3645
+    },
+    {
+      "epoch": 61.86,
+      "learning_rate": 0.00013106954541784654,
+      "loss": 0.1208,
+      "step": 3650
+    },
+    {
+      "epoch": 61.95,
+      "learning_rate": 0.00013632660913388416,
+      "loss": 0.1254,
+      "step": 3655
+    },
+    {
+      "epoch": 62.03,
+      "learning_rate": 0.0001368113672944154,
+      "loss": 0.1188,
+      "step": 3660
+    },
+    {
+      "epoch": 62.12,
+      "learning_rate": 0.00013248966177323087,
+      "loss": 0.1197,
+      "step": 3665
+    },
+    {
+      "epoch": 62.2,
+      "learning_rate": 0.0001236660183620645,
+      "loss": 0.1203,
+      "step": 3670
+    },
+    {
+      "epoch": 62.29,
+      "learning_rate": 0.00011096218858530879,
+      "loss": 0.1246,
+      "step": 3675
+    },
+    {
+      "epoch": 62.37,
+      "learning_rate": 9.527333843747105e-05,
+      "loss": 0.1079,
+      "step": 3680
+    },
+    {
+      "epoch": 62.46,
+      "learning_rate": 7.770497117166404e-05,
+      "loss": 0.1164,
+      "step": 3685
+    },
+    {
+      "epoch": 62.54,
+      "learning_rate": 5.9495028828336764e-05,
+      "loss": 0.1245,
+      "step": 3690
+    },
+    {
+      "epoch": 62.63,
+      "learning_rate": 4.1926661562529704e-05,
+      "loss": 0.1111,
+      "step": 3695
+    },
+    {
+      "epoch": 62.71,
+      "learning_rate": 2.6237811414691842e-05,
+      "loss": 0.1016,
+      "step": 3700
+    },
+    {
+      "epoch": 62.8,
+      "learning_rate": 1.3533981637937141e-05,
+      "loss": 0.102,
+      "step": 3705
+    },
+    {
+      "epoch": 62.88,
+      "learning_rate": 4.710338226769408e-06,
+      "loss": 0.1049,
+      "step": 3710
+    },
+    {
+      "epoch": 62.97,
+      "learning_rate": 3.886327055846792e-07,
+      "loss": 0.1034,
+      "step": 3715
+    },
+    {
+      "epoch": 63.05,
+      "learning_rate": 8.733908661157332e-07,
+      "loss": 0.0972,
+      "step": 3720
+    },
+    {
+      "epoch": 63.14,
+      "learning_rate": 6.130454582152328e-06,
+      "loss": 0.1074,
+      "step": 3725
+    },
+    {
+      "epoch": 63.22,
+      "learning_rate": 1.5789388736707867e-05,
+      "loss": 0.0942,
+      "step": 3730
+    },
+    {
+      "epoch": 63.31,
+      "learning_rate": 2.916958365929148e-05,
+      "loss": 0.099,
+      "step": 3735
+    },
+    {
+      "epoch": 63.39,
+      "learning_rate": 4.532821378430164e-05,
+      "loss": 0.0895,
+      "step": 3740
+    },
+    {
+      "epoch": 63.47,
+      "learning_rate": 6.312667315905757e-05,
+      "loss": 0.0808,
+      "step": 3745
+    },
+    {
+      "epoch": 63.56,
+      "learning_rate": 8.131080647237574e-05,
+      "loss": 0.0872,
+      "step": 3750
+    },
+    {
+      "epoch": 63.64,
+      "learning_rate": 9.859928218347602e-05,
+      "loss": 0.0956,
+      "step": 3755
+    },
+    {
+      "epoch": 63.73,
+      "learning_rate": 0.00011377388060386089,
+      "loss": 0.0938,
+      "step": 3760
+    },
+    {
+      "epoch": 63.81,
+      "learning_rate": 0.00012576533484906033,
+      "loss": 0.0916,
+      "step": 3765
+    },
+    {
+      "epoch": 63.9,
+      "learning_rate": 0.00013372867594093105,
+      "loss": 0.1239,
+      "step": 3770
+    },
+    {
+      "epoch": 63.98,
+      "learning_rate": 0.0001371027729234258,
+      "loss": 0.1043,
+      "step": 3775
+    },
+    {
+      "epoch": 64.07,
+      "learning_rate": 0.00013564987254150574,
+      "loss": 0.122,
+      "step": 3780
+    },
+    {
+      "epoch": 64.15,
+      "learning_rate": 0.00012947235234888073,
+      "loss": 0.1096,
+      "step": 3785
+    },
+    {
+      "epoch": 64.24,
+      "learning_rate": 0.00011900550674920717,
+      "loss": 0.1126,
+      "step": 3790
+    },
+    {
+      "epoch": 64.32,
+      "learning_rate": 0.0001049868742970164,
+      "loss": 0.1065,
+      "step": 3795
+    },
+    {
+      "epoch": 64.41,
+      "learning_rate": 8.840426758749784e-05,
+      "loss": 0.1101,
+      "step": 3800
+    },
+    {
+      "epoch": 64.49,
+      "learning_rate": 7.042616777063057e-05,
+      "loss": 0.1059,
+      "step": 3805
+    },
+    {
+      "epoch": 64.58,
+      "learning_rate": 5.231938838884205e-05,
+      "loss": 0.1075,
+      "step": 3810
+    },
+    {
+      "epoch": 64.66,
+      "learning_rate": 3.535981029518182e-05,
+      "loss": 0.1053,
+      "step": 3815
+    },
+    {
+      "epoch": 64.75,
+      "learning_rate": 2.0742477650140925e-05,
+      "loss": 0.1078,
+      "step": 3820
+    },
+    {
+      "epoch": 64.83,
+      "learning_rate": 9.49739001638421e-06,
+      "loss": 0.1014,
+      "step": 3825
+    },
+    {
+      "epoch": 64.92,
+      "learning_rate": 2.416924199324192e-06,
+      "loss": 0.1026,
+      "step": 3830
+    },
+    {
+      "epoch": 65.0,
+      "learning_rate": 0.0,
+      "loss": 0.0956,
+      "step": 3835
+    },
+    {
+      "epoch": 65.08,
+      "learning_rate": 2.4169241993240324e-06,
+      "loss": 0.0804,
+      "step": 3840
+    },
+    {
+      "epoch": 65.17,
+      "learning_rate": 9.497390016383905e-06,
+      "loss": 0.0904,
+      "step": 3845
+    },
+    {
+      "epoch": 65.25,
+      "learning_rate": 2.074247765014049e-05,
+      "loss": 0.098,
+      "step": 3850
+    },
+    {
+      "epoch": 65.34,
+      "learning_rate": 3.5359810295179585e-05,
+      "loss": 0.0852,
+      "step": 3855
+    },
+    {
+      "epoch": 65.42,
+      "learning_rate": 5.231938838884145e-05,
+      "loss": 0.0784,
+      "step": 3860
+    },
+    {
+      "epoch": 65.51,
+      "learning_rate": 7.042616777062997e-05,
+      "loss": 0.077,
+      "step": 3865
+    },
+    {
+      "epoch": 65.59,
+      "learning_rate": 8.840426758749542e-05,
+      "loss": 0.0785,
+      "step": 3870
+    },
+    {
+      "epoch": 65.68,
+      "learning_rate": 0.00010498687429701424,
+      "loss": 0.0817,
+      "step": 3875
+    },
+    {
+      "epoch": 65.76,
+      "learning_rate": 0.00011900550674920675,
+      "loss": 0.098,
+      "step": 3880
+    },
+    {
+      "epoch": 65.85,
+      "learning_rate": 0.00012947235234888046,
+      "loss": 0.0944,
+      "step": 3885
+    },
+    {
+      "epoch": 65.93,
+      "learning_rate": 0.0001356498725415052,
+      "loss": 0.1064,
+      "step": 3890
+    },
+    {
+      "epoch": 66.02,
+      "learning_rate": 0.00013710277292342595,
+      "loss": 0.1177,
+      "step": 3895
+    },
+    {
+      "epoch": 66.1,
+      "learning_rate": 0.00013372867594093062,
+      "loss": 0.0901,
+      "step": 3900
+    },
+    {
+      "epoch": 66.19,
+      "learning_rate": 0.00012576533484906066,
+      "loss": 0.1034,
+      "step": 3905
+    },
+    {
+      "epoch": 66.27,
+      "learning_rate": 0.00011377388060386282,
+      "loss": 0.1063,
+      "step": 3910
+    },
+    {
+      "epoch": 66.36,
+      "learning_rate": 9.859928218347832e-05,
+      "loss": 0.1063,
+      "step": 3915
+    },
+    {
+      "epoch": 66.44,
+      "learning_rate": 8.131080647237825e-05,
+      "loss": 0.1231,
+      "step": 3920
+    },
+    {
+      "epoch": 66.53,
+      "learning_rate": 6.312667315905816e-05,
+      "loss": 0.1059,
+      "step": 3925
+    },
+    {
+      "epoch": 66.61,
+      "learning_rate": 4.532821378430405e-05,
+      "loss": 0.102,
+      "step": 3930
+    },
+    {
+      "epoch": 66.69,
+      "learning_rate": 2.9169583659291973e-05,
+      "loss": 0.0961,
+      "step": 3935
+    },
+    {
+      "epoch": 66.78,
+      "learning_rate": 1.5789388736709497e-05,
+      "loss": 0.0941,
+      "step": 3940
+    },
+    {
+      "epoch": 66.86,
+      "learning_rate": 6.130454582154194e-06,
+      "loss": 0.0914,
+      "step": 3945
+    },
+    {
+      "epoch": 66.95,
+      "learning_rate": 8.733908661158245e-07,
+      "loss": 0.0927,
+      "step": 3950
+    },
+    {
+      "epoch": 67.03,
+      "learning_rate": 3.8863270558461825e-07,
+      "loss": 0.096,
+      "step": 3955
+    },
+    {
+      "epoch": 67.12,
+      "learning_rate": 4.710338226769187e-06,
+      "loss": 0.0744,
+      "step": 3960
+    },
+    {
+      "epoch": 67.2,
+      "learning_rate": 1.3533981637934452e-05,
+      "loss": 0.0924,
+      "step": 3965
+    },
+    {
+      "epoch": 67.29,
+      "learning_rate": 2.6237811414689833e-05,
+      "loss": 0.0853,
+      "step": 3970
+    },
+    {
+      "epoch": 67.37,
+      "learning_rate": 4.192666156253094e-05,
+      "loss": 0.072,
+      "step": 3975
+    },
+    {
+      "epoch": 67.46,
+      "learning_rate": 5.949502882833616e-05,
+      "loss": 0.0702,
+      "step": 3980
+    },
+    {
+      "epoch": 67.54,
+      "learning_rate": 7.77049711716615e-05,
+      "loss": 0.0825,
+      "step": 3985
+    },
+    {
+      "epoch": 67.63,
+      "learning_rate": 9.527333843746868e-05,
+      "loss": 0.0771,
+      "step": 3990
+    },
+    {
+      "epoch": 67.71,
+      "learning_rate": 0.00011096218858530984,
+      "loss": 0.0807,
+      "step": 3995
+    },
+    {
+      "epoch": 67.8,
+      "learning_rate": 0.00012366601836206413,
+      "loss": 0.0884,
+      "step": 4000
+    },
+    {
+      "epoch": 67.88,
+      "learning_rate": 0.00013248966177322998,
+      "loss": 0.087,
+      "step": 4005
+    },
+    {
+      "epoch": 67.97,
+      "learning_rate": 0.00013681136729441534,
+      "loss": 0.0841,
+      "step": 4010
+    },
+    {
+      "epoch": 68.05,
+      "learning_rate": 0.00013632660913388454,
+      "loss": 0.106,
+      "step": 4015
+    },
+    {
+      "epoch": 68.14,
+      "learning_rate": 0.00013106954541784678,
+      "loss": 0.0954,
+      "step": 4020
+    },
+    {
+      "epoch": 68.22,
+      "learning_rate": 0.00012141061126329201,
+      "loss": 0.0902,
+      "step": 4025
+    },
+    {
+      "epoch": 68.31,
+      "learning_rate": 0.00010803041634070835,
+      "loss": 0.1001,
+      "step": 4030
+    },
+    {
+      "epoch": 68.39,
+      "learning_rate": 9.187178621569818e-05,
+      "loss": 0.0991,
+      "step": 4035
+    },
+    {
+      "epoch": 68.47,
+      "learning_rate": 7.407332684094223e-05,
+      "loss": 0.0972,
+      "step": 4040
+    },
+    {
+      "epoch": 68.56,
+      "learning_rate": 5.588919352762407e-05,
+      "loss": 0.0961,
+      "step": 4045
+    },
+    {
+      "epoch": 68.64,
+      "learning_rate": 3.860071781652205e-05,
+      "loss": 0.0918,
+      "step": 4050
+    },
+    {
+      "epoch": 68.73,
+      "learning_rate": 2.342611939613895e-05,
+      "loss": 0.0951,
+      "step": 4055
+    },
+    {
+      "epoch": 68.81,
+      "learning_rate": 1.1434665150939565e-05,
+      "loss": 0.0848,
+      "step": 4060
+    },
+    {
+      "epoch": 68.9,
+      "learning_rate": 3.4713240590695086e-06,
+      "loss": 0.0889,
+      "step": 4065
+    },
+    {
+      "epoch": 68.98,
+      "learning_rate": 9.722707657406968e-08,
+      "loss": 0.0915,
+      "step": 4070
+    },
+    {
+      "epoch": 69.07,
+      "learning_rate": 1.5501274584943003e-06,
+      "loss": 0.0866,
+      "step": 4075
+    },
+    {
+      "epoch": 69.15,
+      "learning_rate": 7.727647651119382e-06,
+      "loss": 0.0738,
+      "step": 4080
+    },
+    {
+      "epoch": 69.24,
+      "learning_rate": 1.8194493250792965e-05,
+      "loss": 0.0928,
+      "step": 4085
+    },
+    {
+      "epoch": 69.32,
+      "learning_rate": 3.221312570298377e-05,
+      "loss": 0.0694,
+      "step": 4090
+    },
+    {
+      "epoch": 69.41,
+      "learning_rate": 4.879573241250233e-05,
+      "loss": 0.0678,
+      "step": 4095
+    },
+    {
+      "epoch": 69.49,
+      "learning_rate": 6.677383222936964e-05,
+      "loss": 0.0659,
+      "step": 4100
+    },
+    {
+      "epoch": 69.58,
+      "learning_rate": 8.488061161115815e-05,
+      "loss": 0.0785,
+      "step": 4105
+    },
+    {
+      "epoch": 69.66,
+      "learning_rate": 0.00010184018970481836,
+      "loss": 0.0811,
+      "step": 4110
+    },
+    {
+      "epoch": 69.75,
+      "learning_rate": 0.00011645752234985924,
+      "loss": 0.0769,
+      "step": 4115
+    },
+    {
+      "epoch": 69.83,
+      "learning_rate": 0.00012770260998361588,
+      "loss": 0.0816,
+      "step": 4120
+    },
+    {
+      "epoch": 69.92,
+      "learning_rate": 0.00013478307580067585,
+      "loss": 0.0849,
+      "step": 4125
+    },
+    {
+      "epoch": 70.0,
+      "learning_rate": 0.0001372,
+      "loss": 0.0908,
+      "step": 4130
+    },
+    {
+      "epoch": 70.08,
+      "learning_rate": 0.00013478307580067593,
+      "loss": 0.0923,
+      "step": 4135
+    },
+    {
+      "epoch": 70.17,
+      "learning_rate": 0.000127702609983616,
+      "loss": 0.0909,
+      "step": 4140
+    },
+    {
+      "epoch": 70.25,
+      "learning_rate": 0.00011645752234985937,
+      "loss": 0.102,
+      "step": 4145
+    },
+    {
+      "epoch": 70.34,
+      "learning_rate": 0.00010184018970482025,
+      "loss": 0.0877,
+      "step": 4150
+    },
+    {
+      "epoch": 70.42,
+      "learning_rate": 8.488061161115835e-05,
+      "loss": 0.0968,
+      "step": 4155
+    },
+    {
+      "epoch": 70.51,
+      "learning_rate": 6.677383222936983e-05,
+      "loss": 0.0911,
+      "step": 4160
+    },
+    {
+      "epoch": 70.59,
+      "learning_rate": 4.87957324125044e-05,
+      "loss": 0.0904,
+      "step": 4165
+    },
+    {
+      "epoch": 70.68,
+      "learning_rate": 3.22131257029856e-05,
+      "loss": 0.0836,
+      "step": 4170
+    },
+    {
+      "epoch": 70.76,
+      "learning_rate": 1.8194493250793104e-05,
+      "loss": 0.0777,
+      "step": 4175
+    },
+    {
+      "epoch": 70.85,
+      "learning_rate": 7.727647651119474e-06,
+      "loss": 0.0776,
+      "step": 4180
+    },
+    {
+      "epoch": 70.93,
+      "learning_rate": 1.5501274584947573e-06,
+      "loss": 0.0893,
+      "step": 4185
+    },
+    {
+      "epoch": 71.02,
+      "learning_rate": 9.722707657406206e-08,
+      "loss": 0.0805,
+      "step": 4190
+    },
+    {
+      "epoch": 71.1,
+      "learning_rate": 3.4713240590694476e-06,
+      "loss": 0.0848,
+      "step": 4195
+    },
+    {
+      "epoch": 71.19,
+      "learning_rate": 1.1434665150939452e-05,
+      "loss": 0.0684,
+      "step": 4200
+    },
+    {
+      "epoch": 71.27,
+      "learning_rate": 2.3426119396137327e-05,
+      "loss": 0.0728,
+      "step": 4205
+    },
+    {
+      "epoch": 71.36,
+      "learning_rate": 3.860071781652187e-05,
+      "loss": 0.067,
+      "step": 4210
+    },
+    {
+      "epoch": 71.44,
+      "learning_rate": 5.5889193527621953e-05,
+      "loss": 0.0669,
+      "step": 4215
+    },
+    {
+      "epoch": 71.53,
+      "learning_rate": 7.407332684094203e-05,
+      "loss": 0.0603,
+      "step": 4220
+    },
+    {
+      "epoch": 71.61,
+      "learning_rate": 9.187178621569615e-05,
+      "loss": 0.0772,
+      "step": 4225
+    },
+    {
+      "epoch": 71.69,
+      "learning_rate": 0.0001080304163407082,
+      "loss": 0.0701,
+      "step": 4230
+    },
+    {
+      "epoch": 71.78,
+      "learning_rate": 0.00012141061126329064,
+      "loss": 0.0767,
+      "step": 4235
+    },
+    {
+      "epoch": 71.86,
+      "learning_rate": 0.00013106954541784589,
+      "loss": 0.0867,
+      "step": 4240
+    },
+    {
+      "epoch": 71.95,
+      "learning_rate": 0.0001363266091338842,
+      "loss": 0.0768,
+      "step": 4245
+    },
+    {
+      "epoch": 72.03,
+      "learning_rate": 0.00013681136729441537,
+      "loss": 0.0877,
+      "step": 4250
+    },
+    {
+      "epoch": 72.12,
+      "learning_rate": 0.00013248966177323074,
+      "loss": 0.0935,
+      "step": 4255
+    },
+    {
+      "epoch": 72.2,
+      "learning_rate": 0.00012366601836206543,
+      "loss": 0.0872,
+      "step": 4260
+    },
+    {
+      "epoch": 72.29,
+      "learning_rate": 0.00011096218858531,
+      "loss": 0.0952,
+      "step": 4265
+    },
+    {
+      "epoch": 72.37,
+      "learning_rate": 9.527333843746887e-05,
+      "loss": 0.092,
+      "step": 4270
+    },
+    {
+      "epoch": 72.46,
+      "learning_rate": 7.770497117166364e-05,
+      "loss": 0.0833,
+      "step": 4275
+    },
+    {
+      "epoch": 72.54,
+      "learning_rate": 5.9495028828338295e-05,
+      "loss": 0.0857,
+      "step": 4280
+    },
+    {
+      "epoch": 72.63,
+      "learning_rate": 4.192666156253113e-05,
+      "loss": 0.0812,
+      "step": 4285
+    },
+    {
+      "epoch": 72.71,
+      "learning_rate": 2.6237811414689992e-05,
+      "loss": 0.0871,
+      "step": 4290
+    },
+    {
+      "epoch": 72.8,
+      "learning_rate": 1.353398163793574e-05,
+      "loss": 0.078,
+      "step": 4295
+    },
+    {
+      "epoch": 72.88,
+      "learning_rate": 4.7103382267699724e-06,
+      "loss": 0.0784,
+      "step": 4300
+    },
+    {
+      "epoch": 72.97,
+      "learning_rate": 3.886327055846411e-07,
+      "loss": 0.0798,
+      "step": 4305
+    },
+    {
+      "epoch": 73.05,
+      "learning_rate": 8.733908661154818e-07,
+      "loss": 0.0737,
+      "step": 4310
+    },
+    {
+      "epoch": 73.14,
+      "learning_rate": 6.130454582153303e-06,
+      "loss": 0.071,
+      "step": 4315
+    },
+    {
+      "epoch": 73.22,
+      "learning_rate": 1.5789388736708124e-05,
+      "loss": 0.0664,
+      "step": 4320
+    },
+    {
+      "epoch": 73.31,
+      "learning_rate": 2.9169583659291807e-05,
+      "loss": 0.0646,
+      "step": 4325
+    },
+    {
+      "epoch": 73.39,
+      "learning_rate": 4.5328213784302026e-05,
+      "loss": 0.0692,
+      "step": 4330
+    },
+    {
+      "epoch": 73.47,
+      "learning_rate": 6.312667315905602e-05,
+      "loss": 0.0647,
+      "step": 4335
+    },
+    {
+      "epoch": 73.56,
+      "learning_rate": 8.131080647237614e-05,
+      "loss": 0.0666,
+      "step": 4340
+    },
+    {
+      "epoch": 73.64,
+      "learning_rate": 9.859928218347813e-05,
+      "loss": 0.074,
+      "step": 4345
+    },
+    {
+      "epoch": 73.73,
+      "learning_rate": 0.0001137738806038612,
+      "loss": 0.0691,
+      "step": 4350
+    },
+    {
+      "epoch": 73.81,
+      "learning_rate": 0.00012576533484905946,
+      "loss": 0.0743,
+      "step": 4355
+    },
+    {
+      "epoch": 73.9,
+      "learning_rate": 0.00013372867594093057,
+      "loss": 0.0751,
+      "step": 4360
+    },
+    {
+      "epoch": 73.98,
+      "learning_rate": 0.00013710277292342595,
+      "loss": 0.0803,
+      "step": 4365
+    },
+    {
+      "epoch": 74.07,
+      "learning_rate": 0.00013564987254150566,
+      "loss": 0.0772,
+      "step": 4370
+    },
+    {
+      "epoch": 74.15,
+      "learning_rate": 0.0001294723523488814,
+      "loss": 0.0829,
+      "step": 4375
+    },
+    {
+      "epoch": 74.24,
+      "learning_rate": 0.0001190055067492069,
+      "loss": 0.086,
+      "step": 4380
+    },
+    {
+      "epoch": 74.32,
+      "learning_rate": 0.00010498687429701605,
+      "loss": 0.0936,
+      "step": 4385
+    },
+    {
+      "epoch": 74.41,
+      "learning_rate": 8.840426758749746e-05,
+      "loss": 0.0903,
+      "step": 4390
+    },
+    {
+      "epoch": 74.49,
+      "learning_rate": 7.042616777063211e-05,
+      "loss": 0.0929,
+      "step": 4395
+    },
+    {
+      "epoch": 74.58,
+      "learning_rate": 5.231938838884165e-05,
+      "loss": 0.0843,
+      "step": 4400
+    },
+    {
+      "epoch": 74.66,
+      "learning_rate": 3.535981029518147e-05,
+      "loss": 0.0722,
+      "step": 4405
+    },
+    {
+      "epoch": 74.75,
+      "learning_rate": 2.074247765014203e-05,
+      "loss": 0.0749,
+      "step": 4410
+    },
+    {
+      "epoch": 74.83,
+      "learning_rate": 9.497390016384994e-06,
+      "loss": 0.0694,
+      "step": 4415
+    },
+    {
+      "epoch": 74.92,
+      "learning_rate": 2.4169241993240857e-06,
+      "loss": 0.0763,
+      "step": 4420
+    },
+    {
+      "epoch": 75.0,
+      "learning_rate": 0.0,
+      "loss": 0.0681,
+      "step": 4425
+    },
+    {
+      "epoch": 75.08,
+      "learning_rate": 2.4169241993236287e-06,
+      "loss": 0.0652,
+      "step": 4430
+    },
+    {
+      "epoch": 75.17,
+      "learning_rate": 9.497390016384112e-06,
+      "loss": 0.0767,
+      "step": 4435
+    },
+    {
+      "epoch": 75.25,
+      "learning_rate": 2.074247765014078e-05,
+      "loss": 0.0677,
+      "step": 4440
+    },
+    {
+      "epoch": 75.34,
+      "learning_rate": 3.535981029517994e-05,
+      "loss": 0.0722,
+      "step": 4445
+    },
+    {
+      "epoch": 75.42,
+      "learning_rate": 5.2319388388839954e-05,
+      "loss": 0.058,
+      "step": 4450
+    },
+    {
+      "epoch": 75.51,
+      "learning_rate": 7.042616777063036e-05,
+      "loss": 0.0625,
+      "step": 4455
+    },
+    {
+      "epoch": 75.59,
+      "learning_rate": 8.84042675874958e-05,
+      "loss": 0.0646,
+      "step": 4460
+    },
+    {
+      "epoch": 75.68,
+      "learning_rate": 0.00010498687429701457,
+      "loss": 0.0574,
+      "step": 4465
+    },
+    {
+      "epoch": 75.76,
+      "learning_rate": 0.00011900550674920571,
+      "loss": 0.0658,
+      "step": 4470
+    },
+    {
+      "epoch": 75.85,
+      "learning_rate": 0.00012947235234888062,
+      "loss": 0.0753,
+      "step": 4475
+    },
+    {
+      "epoch": 75.93,
+      "learning_rate": 0.00013564987254150528,
+      "loss": 0.0723,
+      "step": 4480
+    },
+    {
+      "epoch": 76.02,
+      "learning_rate": 0.00013710277292342592,
+      "loss": 0.0786,
+      "step": 4485
+    },
+    {
+      "epoch": 76.1,
+      "learning_rate": 0.0001337286759409311,
+      "loss": 0.068,
+      "step": 4490
+    },
+    {
+      "epoch": 76.19,
+      "learning_rate": 0.00012576533484906044,
+      "loss": 0.0919,
+      "step": 4495
+    },
+    {
+      "epoch": 76.27,
+      "learning_rate": 0.00011377388060386252,
+      "loss": 0.0829,
+      "step": 4500
+    },
+    {
+      "epoch": 76.36,
+      "learning_rate": 9.859928218347796e-05,
+      "loss": 0.0836,
+      "step": 4505
+    },
+    {
+      "epoch": 76.44,
+      "learning_rate": 8.131080647237784e-05,
+      "loss": 0.081,
+      "step": 4510
+    },
+    {
+      "epoch": 76.53,
+      "learning_rate": 6.312667315905776e-05,
+      "loss": 0.0725,
+      "step": 4515
+    },
+    {
+      "epoch": 76.61,
+      "learning_rate": 4.5328213784303666e-05,
+      "loss": 0.0789,
+      "step": 4520
+    },
+    {
+      "epoch": 76.69,
+      "learning_rate": 2.9169583659291645e-05,
+      "loss": 0.0796,
+      "step": 4525
+    },
+    {
+      "epoch": 76.78,
+      "learning_rate": 1.5789388736709236e-05,
+      "loss": 0.075,
+      "step": 4530
+    },
+    {
+      "epoch": 76.86,
+      "learning_rate": 6.130454582154026e-06,
+      "loss": 0.0759,
+      "step": 4535
+    },
+    {
+      "epoch": 76.95,
+      "learning_rate": 8.733908661157635e-07,
+      "loss": 0.0688,
+      "step": 4540
+    },
+    {
+      "epoch": 77.03,
+      "learning_rate": 3.8863270558465636e-07,
+      "loss": 0.0699,
+      "step": 4545
+    },
+    {
+      "epoch": 77.12,
+      "learning_rate": 4.710338226769332e-06,
+      "loss": 0.0637,
+      "step": 4550
+    },
+    {
+      "epoch": 77.2,
+      "learning_rate": 1.3533981637934696e-05,
+      "loss": 0.0541,
+      "step": 4555
+    },
+    {
+      "epoch": 77.29,
+      "learning_rate": 2.623781141469015e-05,
+      "loss": 0.058,
+      "step": 4560
+    },
+    {
+      "epoch": 77.37,
+      "learning_rate": 4.192666156253132e-05,
+      "loss": 0.0671,
+      "step": 4565
+    },
+    {
+      "epoch": 77.46,
+      "learning_rate": 5.949502882833657e-05,
+      "loss": 0.0629,
+      "step": 4570
+    },
+    {
+      "epoch": 77.54,
+      "learning_rate": 7.770497117166192e-05,
+      "loss": 0.0566,
+      "step": 4575
+    },
+    {
+      "epoch": 77.63,
+      "learning_rate": 9.527333843746906e-05,
+      "loss": 0.059,
+      "step": 4580
+    },
+    {
+      "epoch": 77.71,
+      "learning_rate": 0.00011096218858531017,
+      "loss": 0.0611,
+      "step": 4585
+    },
+    {
+      "epoch": 77.8,
+      "learning_rate": 0.00012366601836206437,
+      "loss": 0.067,
+      "step": 4590
+    },
+    {
+      "epoch": 77.88,
+      "learning_rate": 0.0001324896617732301,
+      "loss": 0.0716,
+      "step": 4595
+    },
+    {
+      "epoch": 77.97,
+      "learning_rate": 0.00013681136729441537,
+      "loss": 0.0714,
+      "step": 4600
+    },
+    {
+      "epoch": 78.05,
+      "learning_rate": 0.00013632660913388448,
+      "loss": 0.0818,
+      "step": 4605
+    },
+    {
+      "epoch": 78.14,
+      "learning_rate": 0.00013106954541784662,
+      "loss": 0.0696,
+      "step": 4610
+    },
+    {
+      "epoch": 78.22,
+      "learning_rate": 0.00012141061126329175,
+      "loss": 0.0852,
+      "step": 4615
+    },
+    {
+      "epoch": 78.31,
+      "learning_rate": 0.00010803041634070802,
+      "loss": 0.0832,
+      "step": 4620
+    },
+    {
+      "epoch": 78.39,
+      "learning_rate": 9.187178621569779e-05,
+      "loss": 0.0826,
+      "step": 4625
+    },
+    {
+      "epoch": 78.47,
+      "learning_rate": 7.407332684094379e-05,
+      "loss": 0.0803,
+      "step": 4630
+    },
+    {
+      "epoch": 78.56,
+      "learning_rate": 5.588919352762366e-05,
+      "loss": 0.0655,
+      "step": 4635
+    },
+    {
+      "epoch": 78.64,
+      "learning_rate": 3.8600717816521687e-05,
+      "loss": 0.0778,
+      "step": 4640
+    },
+    {
+      "epoch": 78.73,
+      "learning_rate": 2.3426119396138645e-05,
+      "loss": 0.0721,
+      "step": 4645
+    },
+    {
+      "epoch": 78.81,
+      "learning_rate": 1.1434665150940419e-05,
+      "loss": 0.0758,
+      "step": 4650
+    },
+    {
+      "epoch": 78.9,
+      "learning_rate": 3.471324059069379e-06,
+      "loss": 0.0676,
+      "step": 4655
+    },
+    {
+      "epoch": 78.98,
+      "learning_rate": 9.722707657404684e-08,
+      "loss": 0.0666,
+      "step": 4660
+    },
+    {
+      "epoch": 79.07,
+      "learning_rate": 1.5501274584943842e-06,
+      "loss": 0.072,
+      "step": 4665
+    },
+    {
+      "epoch": 79.15,
+      "learning_rate": 7.727647651118667e-06,
+      "loss": 0.0587,
+      "step": 4670
+    },
+    {
+      "epoch": 79.24,
+      "learning_rate": 1.819449325079324e-05,
+      "loss": 0.0528,
+      "step": 4675
+    },
+    {
+      "epoch": 79.32,
+      "learning_rate": 3.221312570298412e-05,
+      "loss": 0.0603,
+      "step": 4680
+    },
+    {
+      "epoch": 79.41,
+      "learning_rate": 4.8795732412502726e-05,
+      "loss": 0.0598,
+      "step": 4685
+    },
+    {
+      "epoch": 79.49,
+      "learning_rate": 6.67738322293681e-05,
+      "loss": 0.0569,
+      "step": 4690
+    },
+    {
+      "epoch": 79.58,
+      "learning_rate": 8.488061161115856e-05,
+      "loss": 0.0499,
+      "step": 4695
+    },
+    {
+      "epoch": 79.66,
+      "learning_rate": 0.00010184018970481871,
+      "loss": 0.056,
+      "step": 4700
+    },
+    {
+      "epoch": 79.75,
+      "learning_rate": 0.0001164575223498581,
+      "loss": 0.0578,
+      "step": 4705
+    },
+    {
+      "epoch": 79.83,
+      "learning_rate": 0.00012770260998361512,
+      "loss": 0.0639,
+      "step": 4710
+    },
+    {
+      "epoch": 79.92,
+      "learning_rate": 0.00013478307580067596,
+      "loss": 0.0654,
+      "step": 4715
+    },
+    {
+      "epoch": 80.0,
+      "learning_rate": 0.0001372,
+      "loss": 0.0753,
+      "step": 4720
+    },
+    {
+      "epoch": 80.08,
+      "learning_rate": 0.0001347830758006763,
+      "loss": 0.068,
+      "step": 4725
+    },
+    {
+      "epoch": 80.17,
+      "learning_rate": 0.00012770260998361577,
+      "loss": 0.0654,
+      "step": 4730
+    },
+    {
+      "epoch": 80.25,
+      "learning_rate": 0.00011645752234985907,
+      "loss": 0.0748,
+      "step": 4735
+    },
+    {
+      "epoch": 80.34,
+      "learning_rate": 0.00010184018970481989,
+      "loss": 0.0768,
+      "step": 4740
+    },
+    {
+      "epoch": 80.42,
+      "learning_rate": 8.488061161115984e-05,
+      "loss": 0.081,
+      "step": 4745
+    },
+    {
+      "epoch": 80.51,
+      "learning_rate": 6.677383222936944e-05,
+      "loss": 0.0816,
+      "step": 4750
+    },
+    {
+      "epoch": 80.59,
+      "learning_rate": 4.879573241250401e-05,
+      "loss": 0.0761,
+      "step": 4755
+    },
+    {
+      "epoch": 80.68,
+      "learning_rate": 3.221312570298526e-05,
+      "loss": 0.0779,
+      "step": 4760
+    },
+    {
+      "epoch": 80.76,
+      "learning_rate": 1.8194493250794154e-05,
+      "loss": 0.0677,
+      "step": 4765
+    },
+    {
+      "epoch": 80.85,
+      "learning_rate": 7.727647651119284e-06,
+      "loss": 0.0545,
+      "step": 4770
+    },
+    {
+      "epoch": 80.93,
+      "learning_rate": 1.5501274584946737e-06,
+      "loss": 0.0604,
+      "step": 4775
+    },
+    {
+      "epoch": 81.02,
+      "learning_rate": 9.722707657397828e-08,
+      "loss": 0.0588,
+      "step": 4780
+    },
+    {
+      "epoch": 81.1,
+      "learning_rate": 3.47132405906896e-06,
+      "loss": 0.0608,
+      "step": 4785
+    },
+    {
+      "epoch": 81.19,
+      "learning_rate": 1.1434665150939672e-05,
+      "loss": 0.0524,
+      "step": 4790
+    },
+    {
+      "epoch": 81.27,
+      "learning_rate": 2.3426119396137632e-05,
+      "loss": 0.0672,
+      "step": 4795
+    },
+    {
+      "epoch": 81.36,
+      "learning_rate": 3.860071781652048e-05,
+      "loss": 0.0474,
+      "step": 4800
+    },
+    {
+      "epoch": 81.44,
+      "learning_rate": 5.5889193527622347e-05,
+      "loss": 0.0541,
+      "step": 4805
+    },
+    {
+      "epoch": 81.53,
+      "learning_rate": 7.40733268409405e-05,
+      "loss": 0.0498,
+      "step": 4810
+    },
+    {
+      "epoch": 81.61,
+      "learning_rate": 9.187178621569835e-05,
+      "loss": 0.0455,
+      "step": 4815
+    },
+    {
+      "epoch": 81.69,
+      "learning_rate": 0.00010803041634070853,
+      "loss": 0.0539,
+      "step": 4820
+    },
+    {
+      "epoch": 81.78,
+      "learning_rate": 0.0001214106112632909,
+      "loss": 0.0573,
+      "step": 4825
+    },
+    {
+      "epoch": 81.86,
+      "learning_rate": 0.00013106954541784605,
+      "loss": 0.0609,
+      "step": 4830
+    },
+    {
+      "epoch": 81.95,
+      "learning_rate": 0.00013632660913388427,
+      "loss": 0.0603,
+      "step": 4835
+    },
+    {
+      "epoch": 82.03,
+      "learning_rate": 0.00013681136729441553,
+      "loss": 0.0709,
+      "step": 4840
+    },
+    {
+      "epoch": 82.12,
+      "learning_rate": 0.0001324896617732313,
+      "loss": 0.0654,
+      "step": 4845
+    },
+    {
+      "epoch": 82.2,
+      "learning_rate": 0.00012366601836206402,
+      "loss": 0.0706,
+      "step": 4850
+    },
+    {
+      "epoch": 82.29,
+      "learning_rate": 0.0001109621885853097,
+      "loss": 0.0702,
+      "step": 4855
+    },
+    {
+      "epoch": 82.37,
+      "learning_rate": 9.52733384374685e-05,
+      "loss": 0.0743,
+      "step": 4860
+    },
+    {
+      "epoch": 82.46,
+      "learning_rate": 7.770497117166324e-05,
+      "loss": 0.0766,
+      "step": 4865
+    },
+    {
+      "epoch": 82.54,
+      "learning_rate": 5.949502882833789e-05,
+      "loss": 0.0691,
+      "step": 4870
+    },
+    {
+      "epoch": 82.63,
+      "learning_rate": 4.192666156253255e-05,
+      "loss": 0.072,
+      "step": 4875
+    },
+    {
+      "epoch": 82.71,
+      "learning_rate": 2.6237811414691212e-05,
+      "loss": 0.0551,
+      "step": 4880
+    },
+    {
+      "epoch": 82.8,
+      "learning_rate": 1.3533981637936661e-05,
+      "loss": 0.0676,
+      "step": 4885
+    },
+    {
+      "epoch": 82.88,
+      "learning_rate": 4.710338226769111e-06,
+      "loss": 0.0629,
+      "step": 4890
+    },
+    {
+      "epoch": 82.97,
+      "learning_rate": 3.8863270558459543e-07,
+      "loss": 0.0614,
+      "step": 4895
+    },
+    {
+      "epoch": 83.05,
+      "learning_rate": 8.733908661155503e-07,
+      "loss": 0.057,
+      "step": 4900
+    },
+    {
+      "epoch": 83.14,
+      "learning_rate": 6.13045458215347e-06,
+      "loss": 0.0538,
+      "step": 4905
+    },
+    {
+      "epoch": 83.22,
+      "learning_rate": 1.5789388736708385e-05,
+      "loss": 0.0509,
+      "step": 4910
+    },
+    {
+      "epoch": 83.31,
+      "learning_rate": 2.9169583659290544e-05,
+      "loss": 0.0513,
+      "step": 4915
+    },
+    {
+      "epoch": 83.39,
+      "learning_rate": 4.532821378430057e-05,
+      "loss": 0.0465,
+      "step": 4920
+    },
+    {
+      "epoch": 83.47,
+      "learning_rate": 6.312667315905837e-05,
+      "loss": 0.0522,
+      "step": 4925
+    },
+    {
+      "epoch": 83.56,
+      "learning_rate": 8.131080647237844e-05,
+      "loss": 0.0574,
+      "step": 4930
+    },
+    {
+      "epoch": 83.64,
+      "learning_rate": 9.85992821834785e-05,
+      "loss": 0.0581,
+      "step": 4935
+    },
+    {
+      "epoch": 83.73,
+      "learning_rate": 0.0001137738806038615,
+      "loss": 0.0563,
+      "step": 4940
+    },
+    {
+      "epoch": 83.81,
+      "learning_rate": 0.0001257653348490597,
+      "loss": 0.0587,
+      "step": 4945
+    },
+    {
+      "epoch": 83.9,
+      "learning_rate": 0.00013372867594093008,
+      "loss": 0.0569,
+      "step": 4950
+    },
+    {
+      "epoch": 83.98,
+      "learning_rate": 0.00013710277292342587,
+      "loss": 0.0614,
+      "step": 4955
+    },
+    {
+      "epoch": 84.07,
+      "learning_rate": 0.00013564987254150598,
+      "loss": 0.0651,
+      "step": 4960
+    },
+    {
+      "epoch": 84.15,
+      "learning_rate": 0.00012947235234888035,
+      "loss": 0.0607,
+      "step": 4965
+    },
+    {
+      "epoch": 84.24,
+      "learning_rate": 0.00011900550674920662,
+      "loss": 0.0656,
+      "step": 4970
+    },
+    {
+      "epoch": 84.32,
+      "learning_rate": 0.0001049868742970157,
+      "loss": 0.0584,
+      "step": 4975
+    },
+    {
+      "epoch": 84.41,
+      "learning_rate": 8.840426758749708e-05,
+      "loss": 0.0668,
+      "step": 4980
+    },
+    {
+      "epoch": 84.49,
+      "learning_rate": 7.04261677706317e-05,
+      "loss": 0.0659,
+      "step": 4985
+    },
+    {
+      "epoch": 84.58,
+      "learning_rate": 5.231938838884316e-05,
+      "loss": 0.066,
+      "step": 4990
+    },
+    {
+      "epoch": 84.66,
+      "learning_rate": 3.535981029518282e-05,
+      "loss": 0.0632,
+      "step": 4995
+    },
+    {
+      "epoch": 84.75,
+      "learning_rate": 2.0742477650140346e-05,
+      "loss": 0.0742,
+      "step": 5000
+    },
+    {
+      "epoch": 84.83,
+      "learning_rate": 9.4973900163838e-06,
+      "loss": 0.0586,
+      "step": 5005
+    },
+    {
+      "epoch": 84.92,
+      "learning_rate": 2.4169241993239866e-06,
+      "loss": 0.0588,
+      "step": 5010
+    },
+    {
+      "epoch": 85.0,
+      "learning_rate": 0.0,
+      "loss": 0.0531,
+      "step": 5015
+    },
+    {
+      "epoch": 85.08,
+      "learning_rate": 2.416924199323735e-06,
+      "loss": 0.0476,
+      "step": 5020
+    },
+    {
+      "epoch": 85.17,
+      "learning_rate": 9.497390016383327e-06,
+      "loss": 0.0496,
+      "step": 5025
+    },
+    {
+      "epoch": 85.25,
+      "learning_rate": 2.0742477650139675e-05,
+      "loss": 0.0555,
+      "step": 5030
+    },
+    {
+      "epoch": 85.34,
+      "learning_rate": 3.535981029517858e-05,
+      "loss": 0.0468,
+      "step": 5035
+    },
+    {
+      "epoch": 85.42,
+      "learning_rate": 5.2319388388842245e-05,
+      "loss": 0.0464,
+      "step": 5040
+    },
+    {
+      "epoch": 85.51,
+      "learning_rate": 7.042616777063077e-05,
+      "loss": 0.0492,
+      "step": 5045
+    },
+    {
+      "epoch": 85.59,
+      "learning_rate": 8.840426758749618e-05,
+      "loss": 0.0536,
+      "step": 5050
+    },
+    {
+      "epoch": 85.68,
+      "learning_rate": 0.00010498687429701491,
+      "loss": 0.0514,
+      "step": 5055
+    },
+    {
+      "epoch": 85.76,
+      "learning_rate": 0.00011900550674920598,
+      "loss": 0.057,
+      "step": 5060
+    },
+    {
+      "epoch": 85.85,
+      "learning_rate": 0.00012947235234887991,
+      "loss": 0.0542,
+      "step": 5065
+    },
+    {
+      "epoch": 85.93,
+      "learning_rate": 0.00013564987254150495,
+      "loss": 0.0564,
+      "step": 5070
+    },
+    {
+      "epoch": 86.02,
+      "learning_rate": 0.00013710277292342592,
+      "loss": 0.064,
+      "step": 5075
+    },
+    {
+      "epoch": 86.1,
+      "learning_rate": 0.00013372867594093038,
+      "loss": 0.0597,
+      "step": 5080
+    },
+    {
+      "epoch": 86.19,
+      "learning_rate": 0.00012576533484906022,
+      "loss": 0.0623,
+      "step": 5085
+    },
+    {
+      "epoch": 86.27,
+      "learning_rate": 0.0001137738806038622,
+      "loss": 0.0575,
+      "step": 5090
+    },
+    {
+      "epoch": 86.36,
+      "learning_rate": 9.859928218347934e-05,
+      "loss": 0.0666,
+      "step": 5095
+    },
+    {
+      "epoch": 86.44,
+      "learning_rate": 8.131080647237936e-05,
+      "loss": 0.0666,
+      "step": 5100
+    },
+    {
+      "epoch": 86.53,
+      "learning_rate": 6.31266731590593e-05,
+      "loss": 0.0662,
+      "step": 5105
+    },
+    {
+      "epoch": 86.61,
+      "learning_rate": 4.532821378430512e-05,
+      "loss": 0.0634,
+      "step": 5110
+    },
+    {
+      "epoch": 86.69,
+      "learning_rate": 2.9169583659291313e-05,
+      "loss": 0.0579,
+      "step": 5115
+    },
+    {
+      "epoch": 86.78,
+      "learning_rate": 1.5789388736708978e-05,
+      "loss": 0.0519,
+      "step": 5120
+    },
+    {
+      "epoch": 86.86,
+      "learning_rate": 6.130454582153858e-06,
+      "loss": 0.0604,
+      "step": 5125
+    },
+    {
+      "epoch": 86.95,
+      "learning_rate": 8.73390866115695e-07,
+      "loss": 0.0576,
+      "step": 5130
+    },
+    {
+      "epoch": 87.03,
+      "learning_rate": 3.8863270558449643e-07,
+      "loss": 0.0524,
+      "step": 5135
+    },
+    {
+      "epoch": 87.12,
+      "learning_rate": 4.710338226768769e-06,
+      "loss": 0.0541,
+      "step": 5140
+    },
+    {
+      "epoch": 87.2,
+      "learning_rate": 1.3533981637933775e-05,
+      "loss": 0.0473,
+      "step": 5145
+    },
+    {
+      "epoch": 87.29,
+      "learning_rate": 2.6237811414690473e-05,
+      "loss": 0.049,
+      "step": 5150
+    },
+    {
+      "epoch": 87.37,
+      "learning_rate": 4.192666156253168e-05,
+      "loss": 0.0497,
+      "step": 5155
+    },
+    {
+      "epoch": 87.46,
+      "learning_rate": 5.949502882833696e-05,
+      "loss": 0.0499,
+      "step": 5160
+    },
+    {
+      "epoch": 87.54,
+      "learning_rate": 7.770497117166231e-05,
+      "loss": 0.0495,
+      "step": 5165
+    },
+    {
+      "epoch": 87.63,
+      "learning_rate": 9.527333843746763e-05,
+      "loss": 0.0543,
+      "step": 5170
+    },
+    {
+      "epoch": 87.71,
+      "learning_rate": 0.00011096218858530895,
+      "loss": 0.0479,
+      "step": 5175
+    },
+    {
+      "epoch": 87.8,
+      "learning_rate": 0.00012366601836206345,
+      "loss": 0.0551,
+      "step": 5180
+    },
+    {
+      "epoch": 87.88,
+      "learning_rate": 0.00013248966177322954,
+      "loss": 0.0545,
+      "step": 5185
+    },
+    {
+      "epoch": 87.97,
+      "learning_rate": 0.00013681136729441542,
+      "loss": 0.0589,
+      "step": 5190
+    },
+    {
+      "epoch": 88.05,
+      "learning_rate": 0.00013632660913388443,
+      "loss": 0.0645,
+      "step": 5195
+    },
+    {
+      "epoch": 88.14,
+      "learning_rate": 0.00013106954541784645,
+      "loss": 0.0602,
+      "step": 5200
+    },
+    {
+      "epoch": 88.22,
+      "learning_rate": 0.0001214106112632915,
+      "loss": 0.0542,
+      "step": 5205
+    },
+    {
+      "epoch": 88.31,
+      "learning_rate": 0.0001080304163407093,
+      "loss": 0.0617,
+      "step": 5210
+    },
+    {
+      "epoch": 88.39,
+      "learning_rate": 9.187178621569925e-05,
+      "loss": 0.063,
+      "step": 5215
+    },
+    {
+      "epoch": 88.47,
+      "learning_rate": 7.407332684094144e-05,
+      "loss": 0.0639,
+      "step": 5220
+    },
+    {
+      "epoch": 88.56,
+      "learning_rate": 5.5889193527625186e-05,
+      "loss": 0.0587,
+      "step": 5225
+    },
+    {
+      "epoch": 88.64,
+      "learning_rate": 3.860071781652132e-05,
+      "loss": 0.0594,
+      "step": 5230
+    },
+    {
+      "epoch": 88.73,
+      "learning_rate": 2.342611939613834e-05,
+      "loss": 0.0555,
+      "step": 5235
+    },
+    {
+      "epoch": 88.81,
+      "learning_rate": 1.1434665150940197e-05,
+      "loss": 0.0594,
+      "step": 5240
+    },
+    {
+      "epoch": 88.9,
+      "learning_rate": 3.4713240590698665e-06,
+      "loss": 0.0553,
+      "step": 5245
+    },
+    {
+      "epoch": 88.98,
+      "learning_rate": 9.722707657413062e-08,
+      "loss": 0.0614,
+      "step": 5250
+    },
+    {
+      "epoch": 89.07,
+      "learning_rate": 1.5501274584940568e-06,
+      "loss": 0.0495,
+      "step": 5255
+    },
+    {
+      "epoch": 89.15,
+      "learning_rate": 7.727647651119748e-06,
+      "loss": 0.0453,
+      "step": 5260
+    },
+    {
+      "epoch": 89.24,
+      "learning_rate": 1.8194493250793514e-05,
+      "loss": 0.049,
+      "step": 5265
+    },
+    {
+      "epoch": 89.32,
+      "learning_rate": 3.221312570298446e-05,
+      "loss": 0.0523,
+      "step": 5270
+    },
+    {
+      "epoch": 89.41,
+      "learning_rate": 4.879573241250311e-05,
+      "loss": 0.0466,
+      "step": 5275
+    },
+    {
+      "epoch": 89.49,
+      "learning_rate": 6.67738322293685e-05,
+      "loss": 0.0471,
+      "step": 5280
+    },
+    {
+      "epoch": 89.58,
+      "learning_rate": 8.488061161115704e-05,
+      "loss": 0.0472,
+      "step": 5285
+    },
+    {
+      "epoch": 89.66,
+      "learning_rate": 0.00010184018970481735,
+      "loss": 0.0432,
+      "step": 5290
+    },
+    {
+      "epoch": 89.75,
+      "learning_rate": 0.0001164575223498598,
+      "loss": 0.0476,
+      "step": 5295
+    },
+    {
+      "epoch": 89.83,
+      "learning_rate": 0.0001277026099836163,
+      "loss": 0.0497,
+      "step": 5300
+    },
+    {
+      "epoch": 89.92,
+      "learning_rate": 0.00013478307580067607,
+      "loss": 0.0568,
+      "step": 5305
+    },
+    {
+      "epoch": 90.0,
+      "learning_rate": 0.0001372,
+      "loss": 0.0585,
+      "step": 5310
+    },
+    {
+      "epoch": 90.08,
+      "learning_rate": 0.0001347830758006762,
+      "loss": 0.058,
+      "step": 5315
+    },
+    {
+      "epoch": 90.17,
+      "learning_rate": 0.0001277026099836166,
+      "loss": 0.0557,
+      "step": 5320
+    },
+    {
+      "epoch": 90.25,
+      "learning_rate": 0.00011645752234986017,
+      "loss": 0.0601,
+      "step": 5325
+    },
+    {
+      "epoch": 90.34,
+      "learning_rate": 0.00010184018970482124,
+      "loss": 0.0574,
+      "step": 5330
+    },
+    {
+      "epoch": 90.42,
+      "learning_rate": 8.488061161115755e-05,
+      "loss": 0.0634,
+      "step": 5335
+    },
+    {
+      "epoch": 90.51,
+      "learning_rate": 6.677383222936903e-05,
+      "loss": 0.059,
+      "step": 5340
+    },
+    {
+      "epoch": 90.59,
+      "learning_rate": 4.879573241250363e-05,
+      "loss": 0.0638,
+      "step": 5345
+    },
+    {
+      "epoch": 90.68,
+      "learning_rate": 3.221312570298492e-05,
+      "loss": 0.0514,
+      "step": 5350
+    },
+    {
+      "epoch": 90.76,
+      "learning_rate": 1.819449325079388e-05,
+      "loss": 0.0537,
+      "step": 5355
+    },
+    {
+      "epoch": 90.85,
+      "learning_rate": 7.727647651119999e-06,
+      "loss": 0.048,
+      "step": 5360
+    },
+    {
+      "epoch": 90.93,
+      "learning_rate": 1.550127458495001e-06,
+      "loss": 0.0544,
+      "step": 5365
+    },
+    {
+      "epoch": 91.02,
+      "learning_rate": 9.722707657410776e-08,
+      "loss": 0.0456,
+      "step": 5370
+    },
+    {
+      "epoch": 91.1,
+      "learning_rate": 3.471324059069699e-06,
+      "loss": 0.0485,
+      "step": 5375
+    },
+    {
+      "epoch": 91.19,
+      "learning_rate": 1.14346651509399e-05,
+      "loss": 0.047,
+      "step": 5380
+    },
+    {
+      "epoch": 91.27,
+      "learning_rate": 2.3426119396137937e-05,
+      "loss": 0.0489,
+      "step": 5385
+    },
+    {
+      "epoch": 91.36,
+      "learning_rate": 3.860071781652084e-05,
+      "loss": 0.0459,
+      "step": 5390
+    },
+    {
+      "epoch": 91.44,
+      "learning_rate": 5.588919352762083e-05,
+      "loss": 0.0471,
+      "step": 5395
+    },
+    {
+      "epoch": 91.53,
+      "learning_rate": 7.407332684094089e-05,
+      "loss": 0.0428,
+      "step": 5400
+    },
+    {
+      "epoch": 91.61,
+      "learning_rate": 9.187178621569508e-05,
+      "loss": 0.0452,
+      "step": 5405
+    },
+    {
+      "epoch": 91.69,
+      "learning_rate": 0.00010803041634070884,
+      "loss": 0.0513,
+      "step": 5410
+    },
+    {
+      "epoch": 91.78,
+      "learning_rate": 0.00012141061126329114,
+      "loss": 0.0464,
+      "step": 5415
+    },
+    {
+      "epoch": 91.86,
+      "learning_rate": 0.00013106954541784624,
+      "loss": 0.0518,
+      "step": 5420
+    },
+    {
+      "epoch": 91.95,
+      "learning_rate": 0.00013632660913388432,
+      "loss": 0.0564,
+      "step": 5425
+    },
+    {
+      "epoch": 92.03,
+      "learning_rate": 0.00013681136729441548,
+      "loss": 0.051,
+      "step": 5430
+    },
+    {
+      "epoch": 92.12,
+      "learning_rate": 0.00013248966177323117,
+      "loss": 0.0527,
+      "step": 5435
+    },
+    {
+      "epoch": 92.2,
+      "learning_rate": 0.0001236660183620661,
+      "loss": 0.0553,
+      "step": 5440
+    },
+    {
+      "epoch": 92.29,
+      "learning_rate": 0.00011096218858530938,
+      "loss": 0.0554,
+      "step": 5445
+    },
+    {
+      "epoch": 92.37,
+      "learning_rate": 9.527333843746813e-05,
+      "loss": 0.0556,
+      "step": 5450
+    },
+    {
+      "epoch": 92.46,
+      "learning_rate": 7.770497117166284e-05,
+      "loss": 0.0588,
+      "step": 5455
+    },
+    {
+      "epoch": 92.54,
+      "learning_rate": 5.9495028828337496e-05,
+      "loss": 0.0569,
+      "step": 5460
+    },
+    {
+      "epoch": 92.63,
+      "learning_rate": 4.192666156253218e-05,
+      "loss": 0.0547,
+      "step": 5465
+    },
+    {
+      "epoch": 92.71,
+      "learning_rate": 2.623781141469089e-05,
+      "loss": 0.0554,
+      "step": 5470
+    },
+    {
+      "epoch": 92.8,
+      "learning_rate": 1.3533981637936417e-05,
+      "loss": 0.0541,
+      "step": 5475
+    },
+    {
+      "epoch": 92.88,
+      "learning_rate": 4.710338226770383e-06,
+      "loss": 0.0486,
+      "step": 5480
+    },
+    {
+      "epoch": 92.97,
+      "learning_rate": 3.8863270558454974e-07,
+      "loss": 0.0512,
+      "step": 5485
+    },
+    {
+      "epoch": 93.05,
+      "learning_rate": 8.733908661156113e-07,
+      "loss": 0.047,
+      "step": 5490
+    },
+    {
+      "epoch": 93.14,
+      "learning_rate": 6.130454582153638e-06,
+      "loss": 0.0441,
+      "step": 5495
+    },
+    {
+      "epoch": 93.22,
+      "learning_rate": 1.5789388736708636e-05,
+      "loss": 0.0469,
+      "step": 5500
+    },
+    {
+      "epoch": 93.31,
+      "learning_rate": 2.9169583659290876e-05,
+      "loss": 0.0444,
+      "step": 5505
+    },
+    {
+      "epoch": 93.39,
+      "learning_rate": 4.532821378430094e-05,
+      "loss": 0.0462,
+      "step": 5510
+    },
+    {
+      "epoch": 93.47,
+      "learning_rate": 6.312667315905488e-05,
+      "loss": 0.0411,
+      "step": 5515
+    },
+    {
+      "epoch": 93.56,
+      "learning_rate": 8.131080647237501e-05,
+      "loss": 0.0466,
+      "step": 5520
+    },
+    {
+      "epoch": 93.64,
+      "learning_rate": 9.859928218347886e-05,
+      "loss": 0.0455,
+      "step": 5525
+    },
+    {
+      "epoch": 93.73,
+      "learning_rate": 0.00011377388060386181,
+      "loss": 0.0421,
+      "step": 5530
+    },
+    {
+      "epoch": 93.81,
+      "learning_rate": 0.00012576533484905992,
+      "loss": 0.049,
+      "step": 5535
+    },
+    {
+      "epoch": 93.9,
+      "learning_rate": 0.00013372867594093019,
+      "loss": 0.0533,
+      "step": 5540
+    },
+    {
+      "epoch": 93.98,
+      "learning_rate": 0.00013710277292342587,
+      "loss": 0.06,
+      "step": 5545
+    },
+    {
+      "epoch": 94.07,
+      "learning_rate": 0.0001356498725415059,
+      "loss": 0.0563,
+      "step": 5550
+    },
+    {
+      "epoch": 94.15,
+      "learning_rate": 0.00012947235234888197,
+      "loss": 0.0559,
+      "step": 5555
+    },
+    {
+      "epoch": 94.24,
+      "learning_rate": 0.00011900550674920635,
+      "loss": 0.065,
+      "step": 5560
+    },
+    {
+      "epoch": 94.32,
+      "learning_rate": 0.00010498687429701536,
+      "loss": 0.0568,
+      "step": 5565
+    },
+    {
+      "epoch": 94.41,
+      "learning_rate": 8.840426758749669e-05,
+      "loss": 0.0619,
+      "step": 5570
+    },
+    {
+      "epoch": 94.49,
+      "learning_rate": 7.042616777063131e-05,
+      "loss": 0.0625,
+      "step": 5575
+    },
+    {
+      "epoch": 94.58,
+      "learning_rate": 5.231938838884276e-05,
+      "loss": 0.0563,
+      "step": 5580
+    },
+    {
+      "epoch": 94.66,
+      "learning_rate": 3.5359810295182465e-05,
+      "loss": 0.0581,
+      "step": 5585
+    },
+    {
+      "epoch": 94.75,
+      "learning_rate": 2.074247765014285e-05,
+      "loss": 0.0551,
+      "step": 5590
+    },
+    {
+      "epoch": 94.83,
+      "learning_rate": 9.497390016385574e-06,
+      "loss": 0.0524,
+      "step": 5595
+    },
+    {
+      "epoch": 94.92,
+      "learning_rate": 2.41692419932388e-06,
+      "loss": 0.0431,
+      "step": 5600
+    },
+    {
+      "epoch": 95.0,
+      "learning_rate": 0.0,
+      "loss": 0.0466,
+      "step": 5605
+    },
+    {
+      "epoch": 95.08,
+      "learning_rate": 2.4169241993238418e-06,
+      "loss": 0.0458,
+      "step": 5610
+    },
+    {
+      "epoch": 95.17,
+      "learning_rate": 9.497390016383526e-06,
+      "loss": 0.0504,
+      "step": 5615
+    },
+    {
+      "epoch": 95.25,
+      "learning_rate": 2.0742477650139963e-05,
+      "loss": 0.0462,
+      "step": 5620
+    },
+    {
+      "epoch": 95.34,
+      "learning_rate": 3.535981029517894e-05,
+      "loss": 0.0432,
+      "step": 5625
+    },
+    {
+      "epoch": 95.42,
+      "learning_rate": 5.231938838883885e-05,
+      "loss": 0.0389,
+      "step": 5630
+    },
+    {
+      "epoch": 95.51,
+      "learning_rate": 7.042616777063117e-05,
+      "loss": 0.0444,
+      "step": 5635
+    },
+    {
+      "epoch": 95.59,
+      "learning_rate": 8.840426758749657e-05,
+      "loss": 0.0406,
+      "step": 5640
+    },
+    {
+      "epoch": 95.68,
+      "learning_rate": 0.00010498687429701524,
+      "loss": 0.0471,
+      "step": 5645
+    },
+    {
+      "epoch": 95.76,
+      "learning_rate": 0.00011900550674920625,
+      "loss": 0.0442,
+      "step": 5650
+    },
+    {
+      "epoch": 95.85,
+      "learning_rate": 0.0001294723523488801,
+      "loss": 0.0465,
+      "step": 5655
+    },
+    {
+      "epoch": 95.93,
+      "learning_rate": 0.00013564987254150503,
+      "loss": 0.0506,
+      "step": 5660
+    },
+    {
+      "epoch": 96.02,
+      "learning_rate": 0.0001371027729234259,
+      "loss": 0.0551,
+      "step": 5665
+    },
+    {
+      "epoch": 96.1,
+      "learning_rate": 0.00013372867594093146,
+      "loss": 0.0517,
+      "step": 5670
+    },
+    {
+      "epoch": 96.19,
+      "learning_rate": 0.00012576533484906,
+      "loss": 0.0512,
+      "step": 5675
+    },
+    {
+      "epoch": 96.27,
+      "learning_rate": 0.00011377388060386191,
+      "loss": 0.057,
+      "step": 5680
+    },
+    {
+      "epoch": 96.36,
+      "learning_rate": 9.859928218347899e-05,
+      "loss": 0.0536,
+      "step": 5685
+    },
+    {
+      "epoch": 96.44,
+      "learning_rate": 8.131080647237897e-05,
+      "loss": 0.0631,
+      "step": 5690
+    },
+    {
+      "epoch": 96.53,
+      "learning_rate": 6.31266731590589e-05,
+      "loss": 0.0536,
+      "step": 5695
+    },
+    {
+      "epoch": 96.61,
+      "learning_rate": 4.532821378430474e-05,
+      "loss": 0.0553,
+      "step": 5700
+    },
+    {
+      "epoch": 96.69,
+      "learning_rate": 2.9169583659290984e-05,
+      "loss": 0.0547,
+      "step": 5705
+    },
+    {
+      "epoch": 96.78,
+      "learning_rate": 1.5789388736708727e-05,
+      "loss": 0.0504,
+      "step": 5710
+    },
+    {
+      "epoch": 96.86,
+      "learning_rate": 6.130454582153691e-06,
+      "loss": 0.051,
+      "step": 5715
+    },
+    {
+      "epoch": 96.95,
+      "learning_rate": 8.733908661156341e-07,
+      "loss": 0.0544,
+      "step": 5720
+    },
+    {
+      "epoch": 97.03,
+      "learning_rate": 3.886327055845345e-07,
+      "loss": 0.0491,
+      "step": 5725
+    },
+    {
+      "epoch": 97.12,
+      "learning_rate": 4.710338226768914e-06,
+      "loss": 0.0393,
+      "step": 5730
+    },
+    {
+      "epoch": 97.2,
+      "learning_rate": 1.353398163793401e-05,
+      "loss": 0.0491,
+      "step": 5735
+    },
+    {
+      "epoch": 97.29,
+      "learning_rate": 2.6237811414690785e-05,
+      "loss": 0.0451,
+      "step": 5740
+    },
+    {
+      "epoch": 97.37,
+      "learning_rate": 4.1926661562532056e-05,
+      "loss": 0.0457,
+      "step": 5745
+    },
+    {
+      "epoch": 97.46,
+      "learning_rate": 5.949502882833735e-05,
+      "loss": 0.0365,
+      "step": 5750
+    },
+    {
+      "epoch": 97.54,
+      "learning_rate": 7.77049711716627e-05,
+      "loss": 0.0395,
+      "step": 5755
+    },
+    {
+      "epoch": 97.63,
+      "learning_rate": 9.527333843746801e-05,
+      "loss": 0.0429,
+      "step": 5760
+    },
+    {
+      "epoch": 97.71,
+      "learning_rate": 0.00011096218858530927,
+      "loss": 0.0457,
+      "step": 5765
+    },
+    {
+      "epoch": 97.8,
+      "learning_rate": 0.0001236660183620637,
+      "loss": 0.0419,
+      "step": 5770
+    },
+    {
+      "epoch": 97.88,
+      "learning_rate": 0.0001324896617732297,
+      "loss": 0.0472,
+      "step": 5775
+    },
+    {
+      "epoch": 97.97,
+      "learning_rate": 0.00013681136729441548,
+      "loss": 0.0512,
+      "step": 5780
+    },
+    {
+      "epoch": 98.05,
+      "learning_rate": 0.00013632660913388435,
+      "loss": 0.0487,
+      "step": 5785
+    },
+    {
+      "epoch": 98.14,
+      "learning_rate": 0.0001310695454178463,
+      "loss": 0.0505,
+      "step": 5790
+    },
+    {
+      "epoch": 98.22,
+      "learning_rate": 0.00012141061126329122,
+      "loss": 0.0493,
+      "step": 5795
+    },
+    {
+      "epoch": 98.31,
+      "learning_rate": 0.00010803041634070896,
+      "loss": 0.0533,
+      "step": 5800
+    },
+    {
+      "epoch": 98.39,
+      "learning_rate": 9.187178621569886e-05,
+      "loss": 0.0511,
+      "step": 5805
+    },
+    {
+      "epoch": 98.47,
+      "learning_rate": 7.407332684094491e-05,
+      "loss": 0.0488,
+      "step": 5810
+    },
+    {
+      "epoch": 98.56,
+      "learning_rate": 5.588919352762479e-05,
+      "loss": 0.057,
+      "step": 5815
+    },
+    {
+      "epoch": 98.64,
+      "learning_rate": 3.860071781652096e-05,
+      "loss": 0.0504,
+      "step": 5820
+    },
+    {
+      "epoch": 98.73,
+      "learning_rate": 2.3426119396138036e-05,
+      "loss": 0.051,
+      "step": 5825
+    },
+    {
+      "epoch": 98.81,
+      "learning_rate": 1.1434665150939969e-05,
+      "loss": 0.0492,
+      "step": 5830
+    },
+    {
+      "epoch": 98.9,
+      "learning_rate": 3.471324059069745e-06,
+      "loss": 0.045,
+      "step": 5835
+    },
+    {
+      "epoch": 98.98,
+      "learning_rate": 9.722707657410776e-08,
+      "loss": 0.0443,
+      "step": 5840
+    },
+    {
+      "epoch": 99.07,
+      "learning_rate": 1.550127458494148e-06,
+      "loss": 0.0471,
+      "step": 5845
+    },
+    {
+      "epoch": 99.15,
+      "learning_rate": 7.72764765111814e-06,
+      "loss": 0.0412,
+      "step": 5850
+    },
+    {
+      "epoch": 99.24,
+      "learning_rate": 1.819449325079379e-05,
+      "loss": 0.036,
+      "step": 5855
+    },
+    {
+      "epoch": 99.32,
+      "learning_rate": 3.22131257029848e-05,
+      "loss": 0.0413,
+      "step": 5860
+    },
+    {
+      "epoch": 99.41,
+      "learning_rate": 4.879573241250349e-05,
+      "loss": 0.0414,
+      "step": 5865
+    },
+    {
+      "epoch": 99.49,
+      "learning_rate": 6.67738322293689e-05,
+      "loss": 0.0373,
+      "step": 5870
+    },
+    {
+      "epoch": 99.58,
+      "learning_rate": 8.488061161115743e-05,
+      "loss": 0.0373,
+      "step": 5875
+    },
+    {
+      "epoch": 99.66,
+      "learning_rate": 0.00010184018970481772,
+      "loss": 0.0382,
+      "step": 5880
+    },
+    {
+      "epoch": 99.75,
+      "learning_rate": 0.0001164575223498573,
+      "loss": 0.0412,
+      "step": 5885
+    },
+    {
+      "epoch": 99.83,
+      "learning_rate": 0.00012770260998361453,
+      "loss": 0.0452,
+      "step": 5890
+    },
+    {
+      "epoch": 99.92,
+      "learning_rate": 0.00013478307580067617,
+      "loss": 0.0459,
+      "step": 5895
+    },
+    {
+      "epoch": 100.0,
+      "learning_rate": 0.0001372,
+      "loss": 0.0481,
+      "step": 5900
+    },
+    {
+      "epoch": 100.08,
+      "learning_rate": 0.0001347830758006761,
+      "loss": 0.0498,
+      "step": 5905
+    },
+    {
+      "epoch": 100.17,
+      "learning_rate": 0.00012770260998361637,
+      "loss": 0.0453,
+      "step": 5910
+    },
+    {
+      "epoch": 100.25,
+      "learning_rate": 0.0001164575223498599,
+      "loss": 0.0556,
+      "step": 5915
+    },
+    {
+      "epoch": 100.34,
+      "learning_rate": 0.00010184018970482089,
+      "loss": 0.0514,
+      "step": 5920
+    },
+    {
+      "epoch": 100.42,
+      "learning_rate": 8.488061161116095e-05,
+      "loss": 0.0529,
+      "step": 5925
+    },
+    {
+      "epoch": 100.51,
+      "learning_rate": 6.677383222936864e-05,
+      "loss": 0.0467,
+      "step": 5930
+    },
+    {
+      "epoch": 100.59,
+      "learning_rate": 4.879573241250324e-05,
+      "loss": 0.0495,
+      "step": 5935
+    },
+    {
+      "epoch": 100.68,
+      "learning_rate": 3.221312570298457e-05,
+      "loss": 0.0482,
+      "step": 5940
+    },
+    {
+      "epoch": 100.76,
+      "learning_rate": 1.8194493250793605e-05,
+      "loss": 0.0465,
+      "step": 5945
+    },
+    {
+      "epoch": 100.85,
+      "learning_rate": 7.727647651119809e-06,
+      "loss": 0.0495,
+      "step": 5950
+    },
+    {
+      "epoch": 100.93,
+      "learning_rate": 1.5501274584949098e-06,
+      "loss": 0.0458,
+      "step": 5955
     }
   ],
-  "max_steps": 59,
-  "num_train_epochs": 1,
-  "total_flos": 61403627520000.0,
+  "max_steps": 5959,
+  "num_train_epochs": 101,
+  "total_flos": 6201766379520000.0,
   "trial_name": null,
   "trial_params": null
 }