diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4878 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.6209975369458128,
+  "eval_steps": 500,
+  "global_step": 4034,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999998830560636e-05,
+      "loss": 2.5898,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999997076402955e-05,
+      "loss": 2.4912,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999883056289132e-05,
+      "loss": 2.3227,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999736877291603e-05,
+      "loss": 2.2148,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999532227891695e-05,
+      "loss": 2.1625,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999269109286036e-05,
+      "loss": 2.1785,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9998947523013125e-05,
+      "loss": 2.1352,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998567470953346e-05,
+      "loss": 2.143,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998128955328933e-05,
+      "loss": 2.1098,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999763197870397e-05,
+      "loss": 2.1105,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9997076543984382e-05,
+      "loss": 2.0875,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99964626544179e-05,
+      "loss": 2.084,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995790313594056e-05,
+      "loss": 2.0598,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9995059525444162e-05,
+      "loss": 2.0734,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9994270294241266e-05,
+      "loss": 2.0902,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999342262460017e-05,
+      "loss": 2.0461,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999251652147735e-05,
+      "loss": 2.0652,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9991551990170978e-05,
+      "loss": 2.0473,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999052903632085e-05,
+      "loss": 2.0266,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9989447665908376e-05,
+      "loss": 2.023,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998830788525654e-05,
+      "loss": 2.075,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998710970102986e-05,
+      "loss": 1.9648,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9985853120234354e-05,
+      "loss": 1.9633,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.998453815021749e-05,
+      "loss": 2.0285,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9983164798668152e-05,
+      "loss": 1.9754,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9981733073616598e-05,
+      "loss": 2.0637,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99802429834344e-05,
+      "loss": 2.0363,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9978694536834407e-05,
+      "loss": 2.0332,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9977087742870683e-05,
+      "loss": 2.0195,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9975422610938463e-05,
+      "loss": 2.007,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99736991507741e-05,
+      "loss": 2.0363,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9971917372454997e-05,
+      "loss": 2.0152,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.997007728639956e-05,
+      "loss": 2.023,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9968178903367136e-05,
+      "loss": 1.9738,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966222234457922e-05,
+      "loss": 2.0262,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.996420729111296e-05,
+      "loss": 2.023,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9962134085114e-05,
+      "loss": 1.998,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960002628583486e-05,
+      "loss": 1.9816,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995781293398446e-05,
+      "loss": 2.0059,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9955565014120487e-05,
+      "loss": 2.0125,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9953258882135594e-05,
+      "loss": 1.9742,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9950894551514183e-05,
+      "loss": 2.0105,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994847203608095e-05,
+      "loss": 1.9895,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9945991350000814e-05,
+      "loss": 1.9766,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9943452507778832e-05,
+      "loss": 2.0281,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99408555242601e-05,
+      "loss": 2.0082,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.993820041462969e-05,
+      "loss": 1.968,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9935487194412547e-05,
+      "loss": 2.0312,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9932715879473385e-05,
+      "loss": 1.9996,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.992988648601663e-05,
+      "loss": 1.9934,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9926999030586293e-05,
+      "loss": 2.0039,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9924053530065882e-05,
+      "loss": 1.9926,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9921050001678316e-05,
+      "loss": 2.0473,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9917988462985805e-05,
+      "loss": 2.0082,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.991486893188976e-05,
+      "loss": 2.0148,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9911691426630685e-05,
+      "loss": 2.0199,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.990845596578807e-05,
+      "loss": 2.0039,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9905162568280283e-05,
+      "loss": 2.0012,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9901811253364458e-05,
+      "loss": 2.0234,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9898402040636383e-05,
+      "loss": 2.0258,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.989493495003039e-05,
+      "loss": 2.0055,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9891410001819228e-05,
+      "loss": 1.9863,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.988782721661395e-05,
+      "loss": 2.0074,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9884186615363802e-05,
+      "loss": 1.9902,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9880488219356086e-05,
+      "loss": 1.9859,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.987673205021604e-05,
+      "loss": 1.9738,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9872918129906712e-05,
+      "loss": 1.9441,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9869046480728834e-05,
+      "loss": 1.9602,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9865117125320693e-05,
+      "loss": 1.925,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9861130086657993e-05,
+      "loss": 2.007,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9857085388053724e-05,
+      "loss": 1.9777,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9852983053158014e-05,
+      "loss": 1.9953,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9848823105958023e-05,
+      "loss": 2.0117,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9844605570777766e-05,
+      "loss": 2.007,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9840330472277988e-05,
+      "loss": 1.9582,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9835997835456022e-05,
+      "loss": 1.9898,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9831607685645636e-05,
+      "loss": 1.9539,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9827160048516884e-05,
+      "loss": 1.9609,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.982265495007597e-05,
+      "loss": 1.9719,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9818092416665076e-05,
+      "loss": 1.9668,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.981347247496222e-05,
+      "loss": 2.0043,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9808795151981094e-05,
+      "loss": 1.9805,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.980406047507092e-05,
+      "loss": 1.9336,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.979926847191627e-05,
+      "loss": 1.9832,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9794419170536918e-05,
+      "loss": 1.9371,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.978951259928767e-05,
+      "loss": 1.9176,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97845487868582e-05,
+      "loss": 1.9609,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9779527762272877e-05,
+      "loss": 1.9219,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9774449554890615e-05,
+      "loss": 1.9543,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9769314194404675e-05,
+      "loss": 2.0285,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9764121710842508e-05,
+      "loss": 1.9438,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9758872134565568e-05,
+      "loss": 1.9477,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.975356549626915e-05,
+      "loss": 1.9926,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97482018269822e-05,
+      "loss": 1.9047,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.974278115806713e-05,
+      "loss": 1.9543,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9737303521219644e-05,
+      "loss": 1.941,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973176894846855e-05,
+      "loss": 1.9324,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9726177472175563e-05,
+      "loss": 1.948,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9720529125035143e-05,
+      "loss": 1.9906,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9714823940074257e-05,
+      "loss": 1.9484,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9709061950652237e-05,
+      "loss": 2.0168,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.970324319046055e-05,
+      "loss": 1.9805,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9697367693522624e-05,
+      "loss": 1.9488,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9691435494193616e-05,
+      "loss": 1.9934,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.968544662716026e-05,
+      "loss": 1.9469,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967940112744062e-05,
+      "loss": 1.9504,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.96732990303839e-05,
+      "loss": 1.9574,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9667140371670254e-05,
+      "loss": 1.9652,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966092518731055e-05,
+      "loss": 1.9258,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.965465351364618e-05,
+      "loss": 1.9754,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9648325387348836e-05,
+      "loss": 1.9598,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96419408454203e-05,
+      "loss": 1.9633,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963549992519223e-05,
+      "loss": 1.9184,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9629002664325938e-05,
+      "loss": 1.9316,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9622449100812165e-05,
+      "loss": 1.9582,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9615839272970868e-05,
+      "loss": 1.9004,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9609173219450998e-05,
+      "loss": 1.9352,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9602450979230258e-05,
+      "loss": 1.966,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.95956725916149e-05,
+      "loss": 1.9582,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9588838096239465e-05,
+      "loss": 1.927,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9581947533066577e-05,
+      "loss": 1.9566,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9575000942386693e-05,
+      "loss": 1.9395,
+      "step": 605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956799836481788e-05,
+      "loss": 2.0352,
+      "step": 610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956093984130557e-05,
+      "loss": 1.9344,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9553825413122313e-05,
+      "loss": 1.9039,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.954665512186756e-05,
+      "loss": 1.9918,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953942900946739e-05,
+      "loss": 1.991,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9532147118174284e-05,
+      "loss": 1.891,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9524809490566878e-05,
+      "loss": 1.9031,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9517416169549702e-05,
+      "loss": 1.9484,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9509967198352942e-05,
+      "loss": 1.8957,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.950246262053217e-05,
+      "loss": 1.8719,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.949490247996812e-05,
+      "loss": 1.973,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9487286820866396e-05,
+      "loss": 1.9586,
+      "step": 665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9479615687757234e-05,
+      "loss": 1.9395,
+      "step": 670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.947188912549524e-05,
+      "loss": 1.9574,
+      "step": 675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9464107179259126e-05,
+      "loss": 1.9906,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9456269894551432e-05,
+      "loss": 1.927,
+      "step": 685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9448377317198293e-05,
+      "loss": 1.932,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9440429493349136e-05,
+      "loss": 1.9324,
+      "step": 695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9432426469476432e-05,
+      "loss": 1.925,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9424368292375413e-05,
+      "loss": 1.9,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9416255009163805e-05,
+      "loss": 1.9699,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.940808666728155e-05,
+      "loss": 1.9215,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.939986331449053e-05,
+      "loss": 1.9695,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.939158499887428e-05,
+      "loss": 1.9352,
+      "step": 725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9383251768837722e-05,
+      "loss": 1.9164,
+      "step": 730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.937486367310687e-05,
+      "loss": 1.9543,
+      "step": 735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9366420760728546e-05,
+      "loss": 1.9219,
+      "step": 740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9357923081070095e-05,
+      "loss": 1.9211,
+      "step": 745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9349370683819107e-05,
+      "loss": 1.9047,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9340763618983104e-05,
+      "loss": 1.9582,
+      "step": 755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9332101936889262e-05,
+      "loss": 1.9391,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9323385688184116e-05,
+      "loss": 1.982,
+      "step": 765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9314614923833273e-05,
+      "loss": 1.9227,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9305789695121085e-05,
+      "loss": 1.9359,
+      "step": 775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9296910053650388e-05,
+      "loss": 1.9422,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9287976051342157e-05,
+      "loss": 1.8836,
+      "step": 785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9278987740435244e-05,
+      "loss": 1.9504,
+      "step": 790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.926994517348605e-05,
+      "loss": 1.9832,
+      "step": 795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.926084840336821e-05,
+      "loss": 1.9246,
+      "step": 800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9251697483272313e-05,
+      "loss": 1.934,
+      "step": 805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.924249246670556e-05,
+      "loss": 1.9301,
+      "step": 810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9233233407491475e-05,
+      "loss": 1.9402,
+      "step": 815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.922392035976957e-05,
+      "loss": 1.909,
+      "step": 820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.921455337799504e-05,
+      "loss": 1.9051,
+      "step": 825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.920513251693845e-05,
+      "loss": 1.9184,
+      "step": 830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.91956578316854e-05,
+      "loss": 1.8781,
+      "step": 835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.918612937763622e-05,
+      "loss": 1.9203,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.917654721050562e-05,
+      "loss": 1.9418,
+      "step": 845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.91669113863224e-05,
+      "loss": 1.9211,
+      "step": 850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9157221961429095e-05,
+      "loss": 1.8941,
+      "step": 855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9147478992481642e-05,
+      "loss": 1.8973,
+      "step": 860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.913768253644908e-05,
+      "loss": 1.9164,
+      "step": 865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.912783265061319e-05,
+      "loss": 1.9371,
+      "step": 870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.911792939256816e-05,
+      "loss": 1.9379,
+      "step": 875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.910797282022027e-05,
+      "loss": 1.9609,
+      "step": 880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9097962991787522e-05,
+      "loss": 1.8887,
+      "step": 885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.908789996579933e-05,
+      "loss": 1.9371,
+      "step": 890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9077783801096164e-05,
+      "loss": 1.8719,
+      "step": 895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9067614556829195e-05,
+      "loss": 1.9141,
+      "step": 900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9057392292459975e-05,
+      "loss": 1.9152,
+      "step": 905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904711706776006e-05,
+      "loss": 1.9086,
+      "step": 910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.903678894281069e-05,
+      "loss": 1.9277,
+      "step": 915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9026407978002415e-05,
+      "loss": 1.9461,
+      "step": 920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9015974234034747e-05,
+      "loss": 1.9277,
+      "step": 925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9005487771915816e-05,
+      "loss": 1.8914,
+      "step": 930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8994948652962e-05,
+      "loss": 1.9133,
+      "step": 935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8984356938797573e-05,
+      "loss": 1.9266,
+      "step": 940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8973712691354347e-05,
+      "loss": 1.8992,
+      "step": 945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8963015972871295e-05,
+      "loss": 1.9215,
+      "step": 950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8952266845894215e-05,
+      "loss": 1.9348,
+      "step": 955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.894146537327533e-05,
+      "loss": 1.9258,
+      "step": 960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.893061161817295e-05,
+      "loss": 1.8957,
+      "step": 965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8919705644051095e-05,
+      "loss": 1.8852,
+      "step": 970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.89087475146791e-05,
+      "loss": 1.8816,
+      "step": 975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8897737294131288e-05,
+      "loss": 1.916,
+      "step": 980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.888667504678654e-05,
+      "loss": 1.9281,
+      "step": 985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.887556083732798e-05,
+      "loss": 1.9051,
+      "step": 990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.886439473074253e-05,
+      "loss": 1.959,
+      "step": 995
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8853176792320598e-05,
+      "loss": 1.9551,
+      "step": 1000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.884190708765564e-05,
+      "loss": 1.8703,
+      "step": 1005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8830585682643806e-05,
+      "loss": 1.8895,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.881921264348355e-05,
+      "loss": 1.9199,
+      "step": 1015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8807788036675243e-05,
+      "loss": 1.9086,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.879631192902077e-05,
+      "loss": 1.8441,
+      "step": 1025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.878478438762317e-05,
+      "loss": 1.8941,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8773205479886207e-05,
+      "loss": 1.9066,
+      "step": 1035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8761575273514005e-05,
+      "loss": 1.9234,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8749893836510635e-05,
+      "loss": 1.9281,
+      "step": 1045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.873816123717973e-05,
+      "loss": 1.8965,
+      "step": 1050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8726377544124077e-05,
+      "loss": 1.9078,
+      "step": 1055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8714542826245214e-05,
+      "loss": 1.9254,
+      "step": 1060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8702657152743028e-05,
+      "loss": 1.9176,
+      "step": 1065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.869072059311537e-05,
+      "loss": 1.8797,
+      "step": 1070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8678733217157613e-05,
+      "loss": 1.8523,
+      "step": 1075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.866669509496227e-05,
+      "loss": 1.9449,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8654606296918586e-05,
+      "loss": 1.9113,
+      "step": 1085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.86424668937121e-05,
+      "loss": 1.8875,
+      "step": 1090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8630276956324265e-05,
+      "loss": 1.9047,
+      "step": 1095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.861803655603201e-05,
+      "loss": 1.8965,
+      "step": 1100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.860574576440733e-05,
+      "loss": 1.9004,
+      "step": 1105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8593404653316866e-05,
+      "loss": 1.8813,
+      "step": 1110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8581013294921496e-05,
+      "loss": 1.9023,
+      "step": 1115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8568571761675893e-05,
+      "loss": 1.9207,
+      "step": 1120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.855608012632812e-05,
+      "loss": 1.9461,
+      "step": 1125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8543538461919193e-05,
+      "loss": 1.8973,
+      "step": 1130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8530946841782656e-05,
+      "loss": 1.9004,
+      "step": 1135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8518305339544157e-05,
+      "loss": 1.8988,
+      "step": 1140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.850561402912101e-05,
+      "loss": 1.9578,
+      "step": 1145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.849287298472178e-05,
+      "loss": 1.9426,
+      "step": 1150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.848008228084581e-05,
+      "loss": 1.9039,
+      "step": 1155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8467241992282842e-05,
+      "loss": 1.9301,
+      "step": 1160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8454352194112534e-05,
+      "loss": 1.9211,
+      "step": 1165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8441412961704033e-05,
+      "loss": 1.8934,
+      "step": 1170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8428424370715543e-05,
+      "loss": 1.9039,
+      "step": 1175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.841538649709388e-05,
+      "loss": 1.9043,
+      "step": 1180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8402299417074018e-05,
+      "loss": 1.9227,
+      "step": 1185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8389163207178654e-05,
+      "loss": 1.8609,
+      "step": 1190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8375977944217762e-05,
+      "loss": 1.8926,
+      "step": 1195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8362743705288127e-05,
+      "loss": 1.9027,
+      "step": 1200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8349460567772915e-05,
+      "loss": 1.8988,
+      "step": 1205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.833612860934121e-05,
+      "loss": 1.8777,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8322747907947564e-05,
+      "loss": 1.9566,
+      "step": 1215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.830931854183153e-05,
+      "loss": 1.9574,
+      "step": 1220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8295840589517216e-05,
+      "loss": 1.8809,
+      "step": 1225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.828231412981283e-05,
+      "loss": 1.9371,
+      "step": 1230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8268739241810207e-05,
+      "loss": 1.8832,
+      "step": 1235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.825511600488435e-05,
+      "loss": 1.9004,
+      "step": 1240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.824144449869297e-05,
+      "loss": 1.9293,
+      "step": 1245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8227724803176018e-05,
+      "loss": 1.9438,
+      "step": 1250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8213956998555217e-05,
+      "loss": 1.9281,
+      "step": 1255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8200141165333592e-05,
+      "loss": 1.9148,
+      "step": 1260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8186277384295e-05,
+      "loss": 1.9016,
+      "step": 1265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8172365736503662e-05,
+      "loss": 1.868,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.815840630330368e-05,
+      "loss": 1.852,
+      "step": 1275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.814439916631857e-05,
+      "loss": 1.909,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8130344407450785e-05,
+      "loss": 1.8773,
+      "step": 1285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8116242108881223e-05,
+      "loss": 1.9145,
+      "step": 1290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8102092353068763e-05,
+      "loss": 1.9199,
+      "step": 1295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8087895222749773e-05,
+      "loss": 1.893,
+      "step": 1300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8073650800937627e-05,
+      "loss": 1.8715,
+      "step": 1305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8059359170922226e-05,
+      "loss": 1.8789,
+      "step": 1310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8045020416269506e-05,
+      "loss": 1.8844,
+      "step": 1315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.803063462082095e-05,
+      "loss": 1.8883,
+      "step": 1320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8016201868693092e-05,
+      "loss": 1.8605,
+      "step": 1325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8001722244277035e-05,
+      "loss": 1.8918,
+      "step": 1330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7987195832237957e-05,
+      "loss": 1.9223,
+      "step": 1335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7972622717514608e-05,
+      "loss": 1.9004,
+      "step": 1340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7958002985318813e-05,
+      "loss": 1.9164,
+      "step": 1345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.794333672113499e-05,
+      "loss": 1.8598,
+      "step": 1350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7928624010719634e-05,
+      "loss": 1.8852,
+      "step": 1355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.791386494010081e-05,
+      "loss": 1.8793,
+      "step": 1360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7899059595577672e-05,
+      "loss": 1.9047,
+      "step": 1365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7884208063719947e-05,
+      "loss": 1.8902,
+      "step": 1370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7869310431367415e-05,
+      "loss": 1.9273,
+      "step": 1375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7854366785629435e-05,
+      "loss": 1.8375,
+      "step": 1380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7839377213884396e-05,
+      "loss": 1.8949,
+      "step": 1385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.782434180377924e-05,
+      "loss": 1.8664,
+      "step": 1390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.7809260643228922e-05,
+      "loss": 1.9133,
+      "step": 1395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7794133820415916e-05,
+      "loss": 1.8309,
+      "step": 1400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7778961423789696e-05,
+      "loss": 1.8379,
+      "step": 1405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7763743542066204e-05,
+      "loss": 1.8891,
+      "step": 1410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.774848026422735e-05,
+      "loss": 1.9281,
+      "step": 1415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7733171679520485e-05,
+      "loss": 1.9125,
+      "step": 1420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7717817877457867e-05,
+      "loss": 1.8664,
+      "step": 1425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7702418947816167e-05,
+      "loss": 1.8996,
+      "step": 1430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.768697498063591e-05,
+      "loss": 1.9141,
+      "step": 1435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7671486066220965e-05,
+      "loss": 1.9203,
+      "step": 1440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.765595229513803e-05,
+      "loss": 1.8859,
+      "step": 1445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7640373758216075e-05,
+      "loss": 1.9449,
+      "step": 1450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.762475054654583e-05,
+      "loss": 1.9117,
+      "step": 1455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.7609082751479246e-05,
+      "loss": 1.9289,
+      "step": 1460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7593370464628957e-05,
+      "loss": 1.9289,
+      "step": 1465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7577613777867762e-05,
+      "loss": 1.8465,
+      "step": 1470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.756181278332806e-05,
+      "loss": 1.875,
+      "step": 1475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7545967573401336e-05,
+      "loss": 1.8871,
+      "step": 1480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.753007824073761e-05,
+      "loss": 1.8598,
+      "step": 1485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.751414487824489e-05,
+      "loss": 1.8867,
+      "step": 1490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7498167579088638e-05,
+      "loss": 1.9262,
+      "step": 1495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7482146436691227e-05,
+      "loss": 1.9203,
+      "step": 1500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7466081544731384e-05,
+      "loss": 1.941,
+      "step": 1505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7449972997143656e-05,
+      "loss": 1.8879,
+      "step": 1510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7433820888117837e-05,
+      "loss": 1.9273,
+      "step": 1515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7417625312098453e-05,
+      "loss": 1.9152,
+      "step": 1520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7401386363784172e-05,
+      "loss": 1.9129,
+      "step": 1525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7385104138127284e-05,
+      "loss": 1.909,
+      "step": 1530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7368778730333117e-05,
+      "loss": 1.8633,
+      "step": 1535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7352410235859503e-05,
+      "loss": 1.9238,
+      "step": 1540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7335998750416204e-05,
+      "loss": 1.8977,
+      "step": 1545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.731954436996436e-05,
+      "loss": 1.923,
+      "step": 1550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7303047190715935e-05,
+      "loss": 1.9258,
+      "step": 1555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.728650730913312e-05,
+      "loss": 1.8992,
+      "step": 1560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7269924821927837e-05,
+      "loss": 1.8902,
+      "step": 1565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.725329982606109e-05,
+      "loss": 1.8801,
+      "step": 1570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7236632418742458e-05,
+      "loss": 1.8848,
+      "step": 1575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.721992269742951e-05,
+      "loss": 1.877,
+      "step": 1580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7203170759827232e-05,
+      "loss": 1.8801,
+      "step": 1585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7186376703887452e-05,
+      "loss": 1.902,
+      "step": 1590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7169540627808276e-05,
+      "loss": 1.8844,
+      "step": 1595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7152662630033506e-05,
+      "loss": 1.8887,
+      "step": 1600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.713574280925207e-05,
+      "loss": 1.8984,
+      "step": 1605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7118781264397446e-05,
+      "loss": 1.857,
+      "step": 1610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.710177809464708e-05,
+      "loss": 1.9281,
+      "step": 1615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7084733399421808e-05,
+      "loss": 1.873,
+      "step": 1620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7067647278385263e-05,
+      "loss": 1.8996,
+      "step": 1625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.705051983144332e-05,
+      "loss": 1.8766,
+      "step": 1630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.703335115874348e-05,
+      "loss": 1.852,
+      "step": 1635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7016141360674307e-05,
+      "loss": 1.8871,
+      "step": 1640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6998890537864827e-05,
+      "loss": 1.8676,
+      "step": 1645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6981598791183957e-05,
+      "loss": 1.8727,
+      "step": 1650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.6964266221739883e-05,
+      "loss": 1.9418,
+      "step": 1655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6946892930879516e-05,
+      "loss": 1.8547,
+      "step": 1660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.692947902018785e-05,
+      "loss": 1.8535,
+      "step": 1665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.691202459148741e-05,
+      "loss": 1.8445,
+      "step": 1670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.689452974683762e-05,
+      "loss": 1.9195,
+      "step": 1675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6876994588534234e-05,
+      "loss": 1.8617,
+      "step": 1680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.685941921910873e-05,
+      "loss": 1.8926,
+      "step": 1685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.68418037413277e-05,
+      "loss": 1.8645,
+      "step": 1690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6824148258192266e-05,
+      "loss": 1.8625,
+      "step": 1695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.680645287293746e-05,
+      "loss": 1.893,
+      "step": 1700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6788717689031637e-05,
+      "loss": 1.9156,
+      "step": 1705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6770942810175854e-05,
+      "loss": 1.8918,
+      "step": 1710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6753128340303283e-05,
+      "loss": 1.8848,
+      "step": 1715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.6735274383578593e-05,
+      "loss": 1.9152,
+      "step": 1720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.671738104439732e-05,
+      "loss": 1.8973,
+      "step": 1725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6699448427385303e-05,
+      "loss": 1.8633,
+      "step": 1730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.668147663739803e-05,
+      "loss": 1.891,
+      "step": 1735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6663465779520042e-05,
+      "loss": 1.925,
+      "step": 1740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6645415959064323e-05,
+      "loss": 1.859,
+      "step": 1745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6627327281571678e-05,
+      "loss": 1.8668,
+      "step": 1750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.660919985281012e-05,
+      "loss": 1.8441,
+      "step": 1755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.659103377877423e-05,
+      "loss": 1.8395,
+      "step": 1760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6572829165684582e-05,
+      "loss": 1.8941,
+      "step": 1765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6554586119987073e-05,
+      "loss": 1.907,
+      "step": 1770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6536304748352335e-05,
+      "loss": 1.8738,
+      "step": 1775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6517985157675097e-05,
+      "loss": 1.8594,
+      "step": 1780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6499627455073565e-05,
+      "loss": 1.8676,
+      "step": 1785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6481231747888783e-05,
+      "loss": 1.8398,
+      "step": 1790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.646279814368402e-05,
+      "loss": 1.8695,
+      "step": 1795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6444326750244145e-05,
+      "loss": 1.884,
+      "step": 1800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6425817675574974e-05,
+      "loss": 1.816,
+      "step": 1805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6407271027902663e-05,
+      "loss": 1.8945,
+      "step": 1810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.638868691567306e-05,
+      "loss": 1.891,
+      "step": 1815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.637006544755108e-05,
+      "loss": 1.9117,
+      "step": 1820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.635140673242005e-05,
+      "loss": 1.8953,
+      "step": 1825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.633271087938111e-05,
+      "loss": 1.8617,
+      "step": 1830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6313977997752533e-05,
+      "loss": 1.8891,
+      "step": 1835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.629520819706912e-05,
+      "loss": 1.8766,
+      "step": 1840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.627640158708154e-05,
+      "loss": 1.8793,
+      "step": 1845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6257558277755687e-05,
+      "loss": 1.8406,
+      "step": 1850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6238678379272047e-05,
+      "loss": 1.9152,
+      "step": 1855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6219762002025055e-05,
+      "loss": 1.8902,
+      "step": 1860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6200809256622435e-05,
+      "loss": 1.8699,
+      "step": 1865
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6181820253884576e-05,
+      "loss": 1.8918,
+      "step": 1870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6162795104843858e-05,
+      "loss": 1.8691,
+      "step": 1875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.614373392074402e-05,
+      "loss": 1.85,
+      "step": 1880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6124636813039502e-05,
+      "loss": 1.8859,
+      "step": 1885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6105503893394806e-05,
+      "loss": 1.907,
+      "step": 1890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.608633527368382e-05,
+      "loss": 1.9059,
+      "step": 1895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6067131065989188e-05,
+      "loss": 1.9016,
+      "step": 1900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6047891382601643e-05,
+      "loss": 1.859,
+      "step": 1905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6028616336019337e-05,
+      "loss": 1.8859,
+      "step": 1910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6009306038947218e-05,
+      "loss": 1.8672,
+      "step": 1915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.598996060429634e-05,
+      "loss": 1.8906,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.597058014518321e-05,
+      "loss": 1.9066,
+      "step": 1925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5951164774929138e-05,
+      "loss": 1.8867,
+      "step": 1930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.593171460705956e-05,
+      "loss": 1.8762,
+      "step": 1935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.591222975530338e-05,
+      "loss": 1.8953,
+      "step": 1940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5892710333592317e-05,
+      "loss": 1.9187,
+      "step": 1945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.587315645606021e-05,
+      "loss": 1.9031,
+      "step": 1950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.585356823704238e-05,
+      "loss": 1.8879,
+      "step": 1955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5833945791074943e-05,
+      "loss": 1.9262,
+      "step": 1960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.581428923289415e-05,
+      "loss": 1.8516,
+      "step": 1965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.579459867743571e-05,
+      "loss": 1.9012,
+      "step": 1970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.577487423983413e-05,
+      "loss": 1.8746,
+      "step": 1975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5755116035422015e-05,
+      "loss": 1.9012,
+      "step": 1980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5735324179729428e-05,
+      "loss": 1.8566,
+      "step": 1985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5715498788483187e-05,
+      "loss": 1.893,
+      "step": 1990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5695639977606206e-05,
+      "loss": 1.8648,
+      "step": 1995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.56757478632168e-05,
+      "loss": 1.8738,
+      "step": 2000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5655822561628032e-05,
+      "loss": 1.875,
+      "step": 2005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5635864189347004e-05,
+      "loss": 1.8375,
+      "step": 2010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5615872863074187e-05,
+      "loss": 1.8953,
+      "step": 2015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5595848699702752e-05,
+      "loss": 1.8699,
+      "step": 2020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.557579181631787e-05,
+      "loss": 1.8258,
+      "step": 2025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5555702330196024e-05,
+      "loss": 1.8707,
+      "step": 2030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5535580358804345e-05,
+      "loss": 1.8949,
+      "step": 2035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.5515426019799898e-05,
+      "loss": 1.8504,
+      "step": 2040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.549523943102903e-05,
+      "loss": 1.8313,
+      "step": 2045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5475020710526623e-05,
+      "loss": 1.9023,
+      "step": 2050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5454769976515474e-05,
+      "loss": 1.8719,
+      "step": 2055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5434487347405553e-05,
+      "loss": 1.8934,
+      "step": 2060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5414172941793327e-05,
+      "loss": 1.8113,
+      "step": 2065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5393826878461067e-05,
+      "loss": 1.8695,
+      "step": 2070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.537344927637616e-05,
+      "loss": 1.9148,
+      "step": 2075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5353040254690396e-05,
+      "loss": 1.8875,
+      "step": 2080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.533259993273928e-05,
+      "loss": 1.8848,
+      "step": 2085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.531212843004136e-05,
+      "loss": 1.8707,
+      "step": 2090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5291625866297464e-05,
+      "loss": 1.8359,
+      "step": 2095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5271092361390078e-05,
+      "loss": 1.884,
+      "step": 2100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5250528035382583e-05,
+      "loss": 1.8422,
+      "step": 2105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5229933008518587e-05,
+      "loss": 1.8844,
+      "step": 2110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5209307401221207e-05,
+      "loss": 1.8742,
+      "step": 2115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5188651334092373e-05,
+      "loss": 1.9062,
+      "step": 2120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5167964927912123e-05,
+      "loss": 1.843,
+      "step": 2125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5147248303637878e-05,
+      "loss": 1.8656,
+      "step": 2130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5126501582403772e-05,
+      "loss": 1.8938,
+      "step": 2135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5105724885519908e-05,
+      "loss": 1.8531,
+      "step": 2140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5084918334471662e-05,
+      "loss": 1.8613,
+      "step": 2145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5064082050918978e-05,
+      "loss": 1.8605,
+      "step": 2150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.504321615669565e-05,
+      "loss": 1.8375,
+      "step": 2155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5022320773808612e-05,
+      "loss": 1.8789,
+      "step": 2160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5001396024437225e-05,
+      "loss": 1.8785,
+      "step": 2165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4980442030932559e-05,
+      "loss": 1.8898,
+      "step": 2170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4959458915816681e-05,
+      "loss": 1.852,
+      "step": 2175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4938446801781932e-05,
+      "loss": 1.8684,
+      "step": 2180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4917405811690229e-05,
+      "loss": 1.8992,
+      "step": 2185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4896336068572321e-05,
+      "loss": 1.8418,
+      "step": 2190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4875237695627084e-05,
+      "loss": 1.8438,
+      "step": 2195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4854110816220805e-05,
+      "loss": 1.8246,
+      "step": 2200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4832955553886446e-05,
+      "loss": 1.8605,
+      "step": 2205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4811772032322931e-05,
+      "loss": 1.8777,
+      "step": 2210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.479056037539442e-05,
+      "loss": 1.8516,
+      "step": 2215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4769320707129588e-05,
+      "loss": 1.8863,
+      "step": 2220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4748053151720907e-05,
+      "loss": 1.909,
+      "step": 2225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.472675783352389e-05,
+      "loss": 1.8426,
+      "step": 2230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4705434877056399e-05,
+      "loss": 1.8559,
+      "step": 2235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4684084406997903e-05,
+      "loss": 1.7957,
+      "step": 2240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4662706548188742e-05,
+      "loss": 1.8723,
+      "step": 2245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4641301425629402e-05,
+      "loss": 1.8559,
+      "step": 2250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4619869164479794e-05,
+      "loss": 1.8902,
+      "step": 2255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.459840989005851e-05,
+      "loss": 1.9121,
+      "step": 2260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4576923727842085e-05,
+      "loss": 1.8652,
+      "step": 2265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4555410803464287e-05,
+      "loss": 1.8582,
+      "step": 2270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4533871242715355e-05,
+      "loss": 1.8672,
+      "step": 2275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4512305171541287e-05,
+      "loss": 1.8355,
+      "step": 2280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4490712716043084e-05,
+      "loss": 1.8254,
+      "step": 2285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4469094002476025e-05,
+      "loss": 1.8363,
+      "step": 2290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4447449157248923e-05,
+      "loss": 1.8641,
+      "step": 2295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4425778306923388e-05,
+      "loss": 1.8461,
+      "step": 2300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4404081578213089e-05,
+      "loss": 1.8332,
+      "step": 2305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4382359097983014e-05,
+      "loss": 1.8383,
+      "step": 2310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4360610993248713e-05,
+      "loss": 1.8484,
+      "step": 2315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 1.8246,
+      "step": 2320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4317038419078098e-05,
+      "loss": 1.8758,
+      "step": 2325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4295214204419077e-05,
+      "loss": 1.841,
+      "step": 2330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4273364874808944e-05,
+      "loss": 1.857,
+      "step": 2335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4251490558004965e-05,
+      "loss": 1.8762,
+      "step": 2340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.422959138191052e-05,
+      "loss": 1.8297,
+      "step": 2345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4207667474574338e-05,
+      "loss": 1.8223,
+      "step": 2350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4185718964189765e-05,
+      "loss": 1.8578,
+      "step": 2355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4163745979093997e-05,
+      "loss": 1.8742,
+      "step": 2360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4141748647767344e-05,
+      "loss": 1.8594,
+      "step": 2365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4119727098832474e-05,
+      "loss": 1.8551,
+      "step": 2370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4097681461053654e-05,
+      "loss": 1.7965,
+      "step": 2375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4075611863336012e-05,
+      "loss": 1.8715,
+      "step": 2380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4053518434724764e-05,
+      "loss": 1.8602,
+      "step": 2385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4031401304404475e-05,
+      "loss": 1.8633,
+      "step": 2390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4009260601698306e-05,
+      "loss": 1.8785,
+      "step": 2395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3987096456067236e-05,
+      "loss": 1.9105,
+      "step": 2400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.396490899710933e-05,
+      "loss": 1.8703,
+      "step": 2405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3942698354558963e-05,
+      "loss": 1.9207,
+      "step": 2410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3920464658286076e-05,
+      "loss": 1.8531,
+      "step": 2415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3898208038295407e-05,
+      "loss": 1.8871,
+      "step": 2420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.387592862472573e-05,
+      "loss": 1.8602,
+      "step": 2425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3853626547849106e-05,
+      "loss": 1.8301,
+      "step": 2430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.38313019380701e-05,
+      "loss": 1.8434,
+      "step": 2435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3808954925925042e-05,
+      "loss": 1.8375,
+      "step": 2440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3786585642081249e-05,
+      "loss": 1.8285,
+      "step": 2445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3764194217336264e-05,
+      "loss": 1.8273,
+      "step": 2450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3741780782617098e-05,
+      "loss": 1.8934,
+      "step": 2455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3719345468979448e-05,
+      "loss": 1.8809,
+      "step": 2460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3696888407606952e-05,
+      "loss": 1.841,
+      "step": 2465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3674409729810403e-05,
+      "loss": 1.8977,
+      "step": 2470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3651909567026996e-05,
+      "loss": 1.8617,
+      "step": 2475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3629388050819547e-05,
+      "loss": 1.8402,
+      "step": 2480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3606845312875735e-05,
+      "loss": 1.8965,
+      "step": 2485
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3584281485007323e-05,
+      "loss": 1.8738,
+      "step": 2490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.3561696699149392e-05,
+      "loss": 1.8438,
+      "step": 2495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.353909108735957e-05,
+      "loss": 1.8273,
+      "step": 2500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3516464781817251e-05,
+      "loss": 1.866,
+      "step": 2505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3493817914822842e-05,
+      "loss": 1.8938,
+      "step": 2510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.347115061879697e-05,
+      "loss": 1.8707,
+      "step": 2515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3448463026279706e-05,
+      "loss": 1.8719,
+      "step": 2520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3425755269929815e-05,
+      "loss": 1.8828,
+      "step": 2525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.340302748252395e-05,
+      "loss": 1.8516,
+      "step": 2530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3380279796955904e-05,
+      "loss": 1.8398,
+      "step": 2535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.33575123462358e-05,
+      "loss": 1.8379,
+      "step": 2540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3334725263489347e-05,
+      "loss": 1.7961,
+      "step": 2545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3311918681957039e-05,
+      "loss": 1.8301,
+      "step": 2550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3289092734993383e-05,
+      "loss": 1.8578,
+      "step": 2555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3266247556066122e-05,
+      "loss": 1.8805,
+      "step": 2560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3243383278755456e-05,
+      "loss": 1.8656,
+      "step": 2565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3220500036753246e-05,
+      "loss": 1.8711,
+      "step": 2570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3197597963862249e-05,
+      "loss": 1.8555,
+      "step": 2575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.317467719399534e-05,
+      "loss": 1.8461,
+      "step": 2580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3151737861174695e-05,
+      "loss": 1.857,
+      "step": 2585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3128780099531056e-05,
+      "loss": 1.8625,
+      "step": 2590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3105804043302906e-05,
+      "loss": 1.8645,
+      "step": 2595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3082809826835707e-05,
+      "loss": 1.8754,
+      "step": 2600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3059797584581107e-05,
+      "loss": 1.7832,
+      "step": 2605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3036767451096148e-05,
+      "loss": 1.8918,
+      "step": 2610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3013719561042497e-05,
+      "loss": 1.8262,
+      "step": 2615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2990654049185636e-05,
+      "loss": 1.8594,
+      "step": 2620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.296757105039409e-05,
+      "loss": 1.9031,
+      "step": 2625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2944470699638636e-05,
+      "loss": 1.8621,
+      "step": 2630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2921353131991507e-05,
+      "loss": 1.8305,
+      "step": 2635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2898218482625606e-05,
+      "loss": 1.8445,
+      "step": 2640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.287506688681372e-05,
+      "loss": 1.8398,
+      "step": 2645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.285189847992772e-05,
+      "loss": 1.8926,
+      "step": 2650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2828713397437785e-05,
+      "loss": 1.884,
+      "step": 2655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2805511774911585e-05,
+      "loss": 1.907,
+      "step": 2660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2782293748013514e-05,
+      "loss": 1.8539,
+      "step": 2665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2759059452503882e-05,
+      "loss": 1.8465,
+      "step": 2670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.273580902423812e-05,
+      "loss": 1.8309,
+      "step": 2675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2712542599165998e-05,
+      "loss": 1.8531,
+      "step": 2680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2689260313330824e-05,
+      "loss": 1.8684,
+      "step": 2685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2665962302868632e-05,
+      "loss": 1.8535,
+      "step": 2690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.264264870400742e-05,
+      "loss": 1.8496,
+      "step": 2695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.261931965306632e-05,
+      "loss": 1.8457,
+      "step": 2700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2595975286454822e-05,
+      "loss": 1.8426,
+      "step": 2705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.257261574067197e-05,
+      "loss": 1.8559,
+      "step": 2710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.254924115230556e-05,
+      "loss": 1.8844,
+      "step": 2715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.252585165803135e-05,
+      "loss": 1.8625,
+      "step": 2720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2502447394612251e-05,
+      "loss": 1.8988,
+      "step": 2725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2479028498897534e-05,
+      "loss": 1.8109,
+      "step": 2730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2455595107822025e-05,
+      "loss": 1.8715,
+      "step": 2735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2432147358405313e-05,
+      "loss": 1.8562,
+      "step": 2740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.240868538775094e-05,
+      "loss": 1.8934,
+      "step": 2745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2385209333045602e-05,
+      "loss": 1.8352,
+      "step": 2750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2361719331558346e-05,
+      "loss": 1.8867,
+      "step": 2755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.233821552063977e-05,
+      "loss": 1.8832,
+      "step": 2760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2314698037721213e-05,
+      "loss": 1.7937,
+      "step": 2765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2291167020313976e-05,
+      "loss": 1.8508,
+      "step": 2770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2267622606008473e-05,
+      "loss": 1.891,
+      "step": 2775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2244064932473467e-05,
+      "loss": 1.8707,
+      "step": 2780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2220494137455253e-05,
+      "loss": 1.8609,
+      "step": 2785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2196910358776835e-05,
+      "loss": 1.8242,
+      "step": 2790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2173313734337151e-05,
+      "loss": 1.8313,
+      "step": 2795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2149704402110243e-05,
+      "loss": 1.8926,
+      "step": 2800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2126082500144456e-05,
+      "loss": 1.8586,
+      "step": 2805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.210244816656164e-05,
+      "loss": 1.9055,
+      "step": 2810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2078801539556326e-05,
+      "loss": 1.8445,
+      "step": 2815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2055142757394936e-05,
+      "loss": 1.8613,
+      "step": 2820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2031471958414955e-05,
+      "loss": 1.852,
+      "step": 2825
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2007789281024146e-05,
+      "loss": 1.8387,
+      "step": 2830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.198409486369972e-05,
+      "loss": 1.8547,
+      "step": 2835
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1960388844987529e-05,
+      "loss": 1.8676,
+      "step": 2840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1936671363501266e-05,
+      "loss": 1.8418,
+      "step": 2845
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1912942557921651e-05,
+      "loss": 1.9207,
+      "step": 2850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1889202566995618e-05,
+      "loss": 1.8258,
+      "step": 2855
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1865451529535496e-05,
+      "loss": 1.873,
+      "step": 2860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1841689584418217e-05,
+      "loss": 1.8898,
+      "step": 2865
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1817916870584483e-05,
+      "loss": 1.8578,
+      "step": 2870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.179413352703796e-05,
+      "loss": 1.8547,
+      "step": 2875
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1770339692844484e-05,
+      "loss": 1.8133,
+      "step": 2880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1746535507131218e-05,
+      "loss": 1.7902,
+      "step": 2885
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1722721109085854e-05,
+      "loss": 1.8973,
+      "step": 2890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1698896637955802e-05,
+      "loss": 1.8512,
+      "step": 2895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1675062233047365e-05,
+      "loss": 1.8355,
+      "step": 2900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1651218033724937e-05,
+      "loss": 1.8313,
+      "step": 2905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1627364179410184e-05,
+      "loss": 1.8418,
+      "step": 2910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1603500809581214e-05,
+      "loss": 1.8227,
+      "step": 2915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.157962806377179e-05,
+      "loss": 1.8645,
+      "step": 2920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.155574608157048e-05,
+      "loss": 1.8609,
+      "step": 2925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1531855002619875e-05,
+      "loss": 1.8406,
+      "step": 2930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1507954966615752e-05,
+      "loss": 1.8137,
+      "step": 2935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1484046113306262e-05,
+      "loss": 1.8957,
+      "step": 2940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1460128582491106e-05,
+      "loss": 1.8273,
+      "step": 2945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.143620251402073e-05,
+      "loss": 1.8398,
+      "step": 2950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1412268047795503e-05,
+      "loss": 1.8965,
+      "step": 2955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1388325323764889e-05,
+      "loss": 1.8652,
+      "step": 2960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1364374481926647e-05,
+      "loss": 1.8387,
+      "step": 2965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1340415662325999e-05,
+      "loss": 1.8918,
+      "step": 2970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1316449005054808e-05,
+      "loss": 1.7996,
+      "step": 2975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1292474650250776e-05,
+      "loss": 1.8379,
+      "step": 2980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1268492738096604e-05,
+      "loss": 1.8355,
+      "step": 2985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.124450340881919e-05,
+      "loss": 1.827,
+      "step": 2990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1220506802688797e-05,
+      "loss": 1.8508,
+      "step": 2995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1196503060018243e-05,
+      "loss": 1.8781,
+      "step": 3000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1172492321162068e-05,
+      "loss": 1.809,
+      "step": 3005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1148474726515716e-05,
+      "loss": 1.8578,
+      "step": 3010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1124450416514731e-05,
+      "loss": 1.8727,
+      "step": 3015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1100419531633917e-05,
+      "loss": 1.8715,
+      "step": 3020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.107638221238652e-05,
+      "loss": 1.8211,
+      "step": 3025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1052338599323412e-05,
+      "loss": 1.8605,
+      "step": 3030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1028288833032261e-05,
+      "loss": 1.8559,
+      "step": 3035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1004233054136726e-05,
+      "loss": 1.8289,
+      "step": 3040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.098017140329561e-05,
+      "loss": 1.8332,
+      "step": 3045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0956104021202052e-05,
+      "loss": 1.8469,
+      "step": 3050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0932031048582716e-05,
+      "loss": 1.8113,
+      "step": 3055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.090795262619694e-05,
+      "loss": 1.857,
+      "step": 3060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0883868894835931e-05,
+      "loss": 1.893,
+      "step": 3065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0859779995321945e-05,
+      "loss": 1.8262,
+      "step": 3070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0835686068507449e-05,
+      "loss": 1.8484,
+      "step": 3075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0811587255274313e-05,
+      "loss": 1.8898,
+      "step": 3080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0787483696532975e-05,
+      "loss": 1.8203,
+      "step": 3085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.076337553322162e-05,
+      "loss": 1.8195,
+      "step": 3090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0739262906305357e-05,
+      "loss": 1.8562,
+      "step": 3095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0715145956775395e-05,
+      "loss": 1.8625,
+      "step": 3100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0691024825648222e-05,
+      "loss": 1.8664,
+      "step": 3105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.066689965396477e-05,
+      "loss": 1.868,
+      "step": 3110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0642770582789605e-05,
+      "loss": 1.8977,
+      "step": 3115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0618637753210086e-05,
+      "loss": 1.8945,
+      "step": 3120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0594501306335551e-05,
+      "loss": 1.8363,
+      "step": 3125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0570361383296493e-05,
+      "loss": 1.8469,
+      "step": 3130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0546218125243721e-05,
+      "loss": 1.8566,
+      "step": 3135
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0522071673347558e-05,
+      "loss": 1.9008,
+      "step": 3140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.049792216879699e-05,
+      "loss": 1.8352,
+      "step": 3145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.047376975279886e-05,
+      "loss": 1.8617,
+      "step": 3150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0449614566577034e-05,
+      "loss": 1.8285,
+      "step": 3155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0425456751371564e-05,
+      "loss": 1.8336,
+      "step": 3160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0401296448437895e-05,
+      "loss": 1.8645,
+      "step": 3165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0377133799046005e-05,
+      "loss": 1.8746,
+      "step": 3170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0352968944479593e-05,
+      "loss": 1.8672,
+      "step": 3175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0328802026035254e-05,
+      "loss": 1.8777,
+      "step": 3180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0304633185021654e-05,
+      "loss": 1.8359,
+      "step": 3185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.028046256275869e-05,
+      "loss": 1.9297,
+      "step": 3190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.025629030057669e-05,
+      "loss": 1.8879,
+      "step": 3195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0232116539815558e-05,
+      "loss": 1.8141,
+      "step": 3200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0207941421823966e-05,
+      "loss": 1.898,
+      "step": 3205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0183765087958522e-05,
+      "loss": 1.8684,
+      "step": 3210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.015958767958294e-05,
+      "loss": 1.8238,
+      "step": 3215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.013540933806722e-05,
+      "loss": 1.7922,
+      "step": 3220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0111230204786821e-05,
+      "loss": 1.8602,
+      "step": 3225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0087050421121825e-05,
+      "loss": 1.8328,
+      "step": 3230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0062870128456122e-05,
+      "loss": 1.8723,
+      "step": 3235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0038689468176575e-05,
+      "loss": 1.8508,
+      "step": 3240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0014508581672196e-05,
+      "loss": 1.7988,
+      "step": 3245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.990327610333324e-06,
+      "loss": 1.7996,
+      "step": 3250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.966146695550795e-06,
+      "loss": 1.8434,
+      "step": 3255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.941965978715105e-06,
+      "loss": 1.825,
+      "step": 3260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.917785601215608e-06,
+      "loss": 1.8621,
+      "step": 3265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.893605704439648e-06,
+      "loss": 1.8645,
+      "step": 3270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.869426429771788e-06,
+      "loss": 1.85,
+      "step": 3275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.845247918592937e-06,
+      "loss": 1.8496,
+      "step": 3280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.821070312279541e-06,
+      "loss": 1.8742,
+      "step": 3285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.796893752202759e-06,
+      "loss": 1.8531,
+      "step": 3290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.772718379727628e-06,
+      "loss": 1.7797,
+      "step": 3295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.748544336212246e-06,
+      "loss": 1.8344,
+      "step": 3300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.724371763006934e-06,
+      "loss": 1.8703,
+      "step": 3305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.70020080145342e-06,
+      "loss": 1.891,
+      "step": 3310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.676031592884008e-06,
+      "loss": 1.8309,
+      "step": 3315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.651864278620752e-06,
+      "loss": 1.875,
+      "step": 3320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.627698999974628e-06,
+      "loss": 1.8305,
+      "step": 3325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.603535898244713e-06,
+      "loss": 1.8523,
+      "step": 3330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.579375114717352e-06,
+      "loss": 1.8469,
+      "step": 3335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.555216790665334e-06,
+      "loss": 1.8332,
+      "step": 3340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.531061067347067e-06,
+      "loss": 1.8715,
+      "step": 3345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.506908086005758e-06,
+      "loss": 1.8391,
+      "step": 3350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.482757987868572e-06,
+      "loss": 1.8223,
+      "step": 3355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.458610914145826e-06,
+      "loss": 1.8008,
+      "step": 3360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.434467006030142e-06,
+      "loss": 1.8586,
+      "step": 3365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.410326404695642e-06,
+      "loss": 1.8391,
+      "step": 3370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.3861892512971e-06,
+      "loss": 1.8363,
+      "step": 3375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.362055686969141e-06,
+      "loss": 1.8496,
+      "step": 3380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.3379258528254e-06,
+      "loss": 1.8738,
+      "step": 3385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.313799889957702e-06,
+      "loss": 1.7855,
+      "step": 3390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.289677939435227e-06,
+      "loss": 1.8883,
+      "step": 3395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.26556014230371e-06,
+      "loss": 1.841,
+      "step": 3400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.241446639584595e-06,
+      "loss": 1.8367,
+      "step": 3405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.2173375722742e-06,
+      "loss": 1.8586,
+      "step": 3410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.19323308134293e-06,
+      "loss": 1.8453,
+      "step": 3415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.169133307734415e-06,
+      "loss": 1.8121,
+      "step": 3420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.145038392364716e-06,
+      "loss": 1.852,
+      "step": 3425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.12094847612148e-06,
+      "loss": 1.8926,
+      "step": 3430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.09686369986312e-06,
+      "loss": 1.8027,
+      "step": 3435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.072784204417995e-06,
+      "loss": 1.8336,
+      "step": 3440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.048710130583587e-06,
+      "loss": 1.8316,
+      "step": 3445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.024641619125686e-06,
+      "loss": 1.7977,
+      "step": 3450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.000578810777546e-06,
+      "loss": 1.8398,
+      "step": 3455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.976521846239076e-06,
+      "loss": 1.8707,
+      "step": 3460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.952470866176021e-06,
+      "loss": 1.8699,
+      "step": 3465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.92842601121913e-06,
+      "loss": 1.8477,
+      "step": 3470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.90438742196333e-06,
+      "loss": 1.8152,
+      "step": 3475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.880355238966923e-06,
+      "loss": 1.8336,
+      "step": 3480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.856329602750745e-06,
+      "loss": 1.8633,
+      "step": 3485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.832310653797358e-06,
+      "loss": 1.8289,
+      "step": 3490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.808298532550215e-06,
+      "loss": 1.8238,
+      "step": 3495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.784293379412851e-06,
+      "loss": 1.8645,
+      "step": 3500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.760295334748055e-06,
+      "loss": 1.8246,
+      "step": 3505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.736304538877052e-06,
+      "loss": 1.8078,
+      "step": 3510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.712321132078681e-06,
+      "loss": 1.8301,
+      "step": 3515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.688345254588579e-06,
+      "loss": 1.8199,
+      "step": 3520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.66437704659835e-06,
+      "loss": 1.8492,
+      "step": 3525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.640416648254765e-06,
+      "loss": 1.8578,
+      "step": 3530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.61646419965892e-06,
+      "loss": 1.875,
+      "step": 3535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.592519840865432e-06,
+      "loss": 1.8703,
+      "step": 3540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.568583711881612e-06,
+      "loss": 1.8609,
+      "step": 3545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.544655952666653e-06,
+      "loss": 1.8371,
+      "step": 3550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.520736703130808e-06,
+      "loss": 1.8473,
+      "step": 3555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.496826103134572e-06,
+      "loss": 1.8336,
+      "step": 3560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.472924292487858e-06,
+      "loss": 1.8562,
+      "step": 3565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.449031410949206e-06,
+      "loss": 1.8715,
+      "step": 3570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.425147598224919e-06,
+      "loss": 1.8234,
+      "step": 3575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.401272993968288e-06,
+      "loss": 1.8098,
+      "step": 3580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.377407737778761e-06,
+      "loss": 1.8313,
+      "step": 3585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.35355196920112e-06,
+      "loss": 1.8742,
+      "step": 3590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.329705827724675e-06,
+      "loss": 1.834,
+      "step": 3595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.305869452782446e-06,
+      "loss": 1.8805,
+      "step": 3600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.282042983750344e-06,
+      "loss": 1.818,
+      "step": 3605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.258226559946354e-06,
+      "loss": 1.8094,
+      "step": 3610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.234420320629726e-06,
+      "loss": 1.8445,
+      "step": 3615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.210624405000169e-06,
+      "loss": 1.8637,
+      "step": 3620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.186838952197019e-06,
+      "loss": 1.8492,
+      "step": 3625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.163064101298432e-06,
+      "loss": 1.8484,
+      "step": 3630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.139299991320576e-06,
+      "loss": 1.843,
+      "step": 3635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.115546761216822e-06,
+      "loss": 1.823,
+      "step": 3640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.091804549876902e-06,
+      "loss": 1.8543,
+      "step": 3645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.068073496126145e-06,
+      "loss": 1.8672,
+      "step": 3650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.044353738724623e-06,
+      "loss": 1.8789,
+      "step": 3655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.020645416366367e-06,
+      "loss": 1.8461,
+      "step": 3660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.996948667678534e-06,
+      "loss": 1.7867,
+      "step": 3665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.973263631220615e-06,
+      "loss": 1.8305,
+      "step": 3670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.94959044548362e-06,
+      "loss": 1.8418,
+      "step": 3675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.92592924888925e-06,
+      "loss": 1.8441,
+      "step": 3680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.902280179789122e-06,
+      "loss": 1.8629,
+      "step": 3685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.878643376463931e-06,
+      "loss": 1.8051,
+      "step": 3690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.855018977122652e-06,
+      "loss": 1.827,
+      "step": 3695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.831407119901737e-06,
+      "loss": 1.8574,
+      "step": 3700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.807807942864297e-06,
+      "loss": 1.8219,
+      "step": 3705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.784221583999298e-06,
+      "loss": 1.8328,
+      "step": 3710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.760648181220757e-06,
+      "loss": 1.8082,
+      "step": 3715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.73708787236694e-06,
+      "loss": 1.8543,
+      "step": 3720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.713540795199545e-06,
+      "loss": 1.8109,
+      "step": 3725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.6900070874029e-06,
+      "loss": 1.8277,
+      "step": 3730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 7.666486886583162e-06,
+      "loss": 1.7719,
+      "step": 3735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.642980330267516e-06,
+      "loss": 1.8289,
+      "step": 3740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.619487555903351e-06,
+      "loss": 1.8992,
+      "step": 3745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.596008700857482e-06,
+      "loss": 1.8113,
+      "step": 3750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.572543902415333e-06,
+      "loss": 1.8219,
+      "step": 3755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.549093297780133e-06,
+      "loss": 1.8102,
+      "step": 3760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.525657024072117e-06,
+      "loss": 1.8398,
+      "step": 3765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.50223521832773e-06,
+      "loss": 1.8523,
+      "step": 3770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.47882801749882e-06,
+      "loss": 1.8371,
+      "step": 3775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.455435558451824e-06,
+      "loss": 1.907,
+      "step": 3780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.432057977966993e-06,
+      "loss": 1.8527,
+      "step": 3785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.4086954127375774e-06,
+      "loss": 1.841,
+      "step": 3790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.385347999369035e-06,
+      "loss": 1.8402,
+      "step": 3795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.362015874378219e-06,
+      "loss": 1.8461,
+      "step": 3800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.338699174192594e-06,
+      "loss": 1.8148,
+      "step": 3805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.3153980351494345e-06,
+      "loss": 1.8535,
+      "step": 3810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.292112593495016e-06,
+      "loss": 1.791,
+      "step": 3815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.268842985383841e-06,
+      "loss": 1.8148,
+      "step": 3820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.2455893468778216e-06,
+      "loss": 1.8105,
+      "step": 3825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.222351813945498e-06,
+      "loss": 1.8844,
+      "step": 3830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.1991305224612355e-06,
+      "loss": 1.8742,
+      "step": 3835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.175925608204428e-06,
+      "loss": 1.8496,
+      "step": 3840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.152737206858721e-06,
+      "loss": 1.8777,
+      "step": 3845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.129565454011189e-06,
+      "loss": 1.8273,
+      "step": 3850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.106410485151573e-06,
+      "loss": 1.825,
+      "step": 3855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.0832724356714715e-06,
+      "loss": 1.8219,
+      "step": 3860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.0601514408635465e-06,
+      "loss": 1.8441,
+      "step": 3865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.037047635920744e-06,
+      "loss": 1.8234,
+      "step": 3870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.0139611559355e-06,
+      "loss": 1.8348,
+      "step": 3875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.990892135898937e-06,
+      "loss": 1.7973,
+      "step": 3880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.967840710700095e-06,
+      "loss": 1.8117,
+      "step": 3885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.944807015125133e-06,
+      "loss": 1.8633,
+      "step": 3890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.921791183856537e-06,
+      "loss": 1.857,
+      "step": 3895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.898793351472341e-06,
+      "loss": 1.7801,
+      "step": 3900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.8758136524453355e-06,
+      "loss": 1.8223,
+      "step": 3905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.852852221142282e-06,
+      "loss": 1.8617,
+      "step": 3910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.829909191823121e-06,
+      "loss": 1.8484,
+      "step": 3915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.806984698640202e-06,
+      "loss": 1.8879,
+      "step": 3920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.784078875637485e-06,
+      "loss": 1.8449,
+      "step": 3925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.761191856749763e-06,
+      "loss": 1.8734,
+      "step": 3930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.738323775801876e-06,
+      "loss": 1.8012,
+      "step": 3935
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.715474766507932e-06,
+      "loss": 1.9027,
+      "step": 3940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.692644962470531e-06,
+      "loss": 1.8098,
+      "step": 3945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.669834497179957e-06,
+      "loss": 1.8309,
+      "step": 3950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.6470435040134326e-06,
+      "loss": 1.8656,
+      "step": 3955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.624272116234319e-06,
+      "loss": 1.8434,
+      "step": 3960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.601520466991338e-06,
+      "loss": 1.8066,
+      "step": 3965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.578788689317801e-06,
+      "loss": 1.8625,
+      "step": 3970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.556076916130822e-06,
+      "loss": 1.8395,
+      "step": 3975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.5333852802305556e-06,
+      "loss": 1.8301,
+      "step": 3980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.510713914299387e-06,
+      "loss": 1.8047,
+      "step": 3985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.488062950901198e-06,
+      "loss": 1.8117,
+      "step": 3990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.465432522480573e-06,
+      "loss": 1.8406,
+      "step": 3995
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.442822761362015e-06,
+      "loss": 1.8008,
+      "step": 4000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.420233799749187e-06,
+      "loss": 1.8313,
+      "step": 4005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.397665769724137e-06,
+      "loss": 1.8418,
+      "step": 4010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.375118803246504e-06,
+      "loss": 1.8219,
+      "step": 4015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.352593032152787e-06,
+      "loss": 1.7992,
+      "step": 4020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.330088588155538e-06,
+      "loss": 1.8555,
+      "step": 4025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.307605602842612e-06,
+      "loss": 1.8664,
+      "step": 4030
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 1.7902071475982666,
+      "eval_runtime": 616.4318,
+      "eval_samples_per_second": 37.49,
+      "eval_steps_per_second": 9.373,
+      "step": 4034
+    },
+    {
+      "epoch": 0.62,
+      "step": 4034,
+      "total_flos": 192478407819264.0,
+      "train_loss": 1.8920561942240952,
+      "train_runtime": 23913.723,
+      "train_samples_per_second": 8.692,
+      "train_steps_per_second": 0.272
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 6496,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 192478407819264.0,
+  "trial_name": null,
+  "trial_params": null
+}