diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7219 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.04590885180048778,
+  "eval_steps": 500,
+  "global_step": 6000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 1.2156,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 1.2101,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 1.1858,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.199,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 1.1992,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 1.1734,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5e-05,
+      "loss": 1.1799,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 1.1723,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-05,
+      "loss": 1.1838,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 1.1698,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 1.1688,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 1.1672,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 1.1656,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7e-05,
+      "loss": 1.1483,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.1688,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 1.1456,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.5e-05,
+      "loss": 1.1409,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-05,
+      "loss": 1.1599,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.5e-05,
+      "loss": 1.1257,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 1.1539,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999963830712e-05,
+      "loss": 1.1499,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999855322848e-05,
+      "loss": 1.1355,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999674476408e-05,
+      "loss": 1.1335,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999421291396e-05,
+      "loss": 1.1193,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999095767815e-05,
+      "loss": 1.1306,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999998697905673e-05,
+      "loss": 1.148,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99999822770497e-05,
+      "loss": 1.1319,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999997685165716e-05,
+      "loss": 1.1169,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999997070287919e-05,
+      "loss": 1.1197,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999996383071588e-05,
+      "loss": 1.1411,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999995623516733e-05,
+      "loss": 1.1143,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999994791623364e-05,
+      "loss": 1.111,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999993887391494e-05,
+      "loss": 1.1167,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999992910821135e-05,
+      "loss": 1.1393,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.9999918619123e-05,
+      "loss": 1.1051,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999990740665008e-05,
+      "loss": 1.117,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999989547079272e-05,
+      "loss": 1.0956,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999988281155112e-05,
+      "loss": 1.1334,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999986942892544e-05,
+      "loss": 1.111,
+      "step": 195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999985532291588e-05,
+      "loss": 1.112,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999984049352263e-05,
+      "loss": 1.1184,
+      "step": 205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999982494074592e-05,
+      "loss": 1.1257,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999980866458598e-05,
+      "loss": 1.104,
+      "step": 215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999979166504305e-05,
+      "loss": 1.0967,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999977394211736e-05,
+      "loss": 1.0979,
+      "step": 225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999975549580917e-05,
+      "loss": 1.0952,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999973632611875e-05,
+      "loss": 1.102,
+      "step": 235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999971643304638e-05,
+      "loss": 1.1008,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999969581659234e-05,
+      "loss": 1.0884,
+      "step": 245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999967447675692e-05,
+      "loss": 1.0899,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999965241354046e-05,
+      "loss": 1.078,
+      "step": 255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999962962694325e-05,
+      "loss": 1.0792,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999960611696562e-05,
+      "loss": 1.0639,
+      "step": 265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999958188360794e-05,
+      "loss": 1.0653,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999955692687054e-05,
+      "loss": 1.0866,
+      "step": 275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999953124675376e-05,
+      "loss": 1.0832,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999950484325802e-05,
+      "loss": 1.0877,
+      "step": 285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999947771638365e-05,
+      "loss": 1.0937,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999944986613108e-05,
+      "loss": 1.109,
+      "step": 295
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999942129250071e-05,
+      "loss": 1.0861,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999939199549293e-05,
+      "loss": 1.0899,
+      "step": 305
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999936197510817e-05,
+      "loss": 1.0721,
+      "step": 310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999933123134689e-05,
+      "loss": 1.0772,
+      "step": 315
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999929976420952e-05,
+      "loss": 1.0577,
+      "step": 320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999926757369649e-05,
+      "loss": 1.06,
+      "step": 325
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999923465980831e-05,
+      "loss": 1.0483,
+      "step": 330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999920102254541e-05,
+      "loss": 1.0811,
+      "step": 335
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999916666190833e-05,
+      "loss": 1.093,
+      "step": 340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999913157789752e-05,
+      "loss": 1.0754,
+      "step": 345
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999909577051351e-05,
+      "loss": 1.0525,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99990592397568e-05,
+      "loss": 1.0802,
+      "step": 355
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999902198562796e-05,
+      "loss": 1.0531,
+      "step": 360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999898400812748e-05,
+      "loss": 1.0758,
+      "step": 365
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999894530725594e-05,
+      "loss": 1.0635,
+      "step": 370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999890588301388e-05,
+      "loss": 1.0869,
+      "step": 375
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999886573540189e-05,
+      "loss": 1.0866,
+      "step": 380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999882486442054e-05,
+      "loss": 1.0687,
+      "step": 385
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999878327007041e-05,
+      "loss": 1.0733,
+      "step": 390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999874095235213e-05,
+      "loss": 1.0445,
+      "step": 395
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99986979112663e-05,
+      "loss": 1.0558,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999865414681351e-05,
+      "loss": 1.0751,
+      "step": 405
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999860965899446e-05,
+      "loss": 1.0535,
+      "step": 410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999856444780974e-05,
+      "loss": 1.0522,
+      "step": 415
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999851851326003e-05,
+      "loss": 1.0484,
+      "step": 420
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999847185534596e-05,
+      "loss": 1.0533,
+      "step": 425
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999842447406825e-05,
+      "loss": 1.0743,
+      "step": 430
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999837636942755e-05,
+      "loss": 1.069,
+      "step": 435
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999832754142459e-05,
+      "loss": 1.0605,
+      "step": 440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999827799006005e-05,
+      "loss": 1.0507,
+      "step": 445
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999822771533467e-05,
+      "loss": 1.0606,
+      "step": 450
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999817671724914e-05,
+      "loss": 1.0516,
+      "step": 455
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999812499580423e-05,
+      "loss": 1.0773,
+      "step": 460
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999807255100068e-05,
+      "loss": 1.0455,
+      "step": 465
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999801938283923e-05,
+      "loss": 1.0471,
+      "step": 470
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99979654913207e-05,
+      "loss": 1.0401,
+      "step": 475
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999791087644581e-05,
+      "loss": 1.0668,
+      "step": 480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999785553821538e-05,
+      "loss": 1.0649,
+      "step": 485
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999779947663021e-05,
+      "loss": 1.029,
+      "step": 490
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99977426916911e-05,
+      "loss": 1.0506,
+      "step": 495
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99976851833989e-05,
+      "loss": 1.0443,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999762695175441e-05,
+      "loss": 1.0662,
+      "step": 505
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999756799675848e-05,
+      "loss": 1.0568,
+      "step": 510
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999750831841196e-05,
+      "loss": 1.0364,
+      "step": 515
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999744791671572e-05,
+      "loss": 1.0481,
+      "step": 520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999738679167066e-05,
+      "loss": 1.0359,
+      "step": 525
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999732494327762e-05,
+      "loss": 1.0338,
+      "step": 530
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99972623715375e-05,
+      "loss": 1.0625,
+      "step": 535
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999719907645124e-05,
+      "loss": 1.0407,
+      "step": 540
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999713505801972e-05,
+      "loss": 1.0485,
+      "step": 545
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99970703162439e-05,
+      "loss": 1.0579,
+      "step": 550
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999700485112468e-05,
+      "loss": 1.0435,
+      "step": 555
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999693866266303e-05,
+      "loss": 1.0294,
+      "step": 560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99968717508599e-05,
+      "loss": 1.0234,
+      "step": 565
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999680411571626e-05,
+      "loss": 1.0656,
+      "step": 570
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999673575723309e-05,
+      "loss": 1.0504,
+      "step": 575
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999666667541139e-05,
+      "loss": 1.0584,
+      "step": 580
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999659687025214e-05,
+      "loss": 1.0431,
+      "step": 585
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999652634175634e-05,
+      "loss": 1.0673,
+      "step": 590
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999645508992504e-05,
+      "loss": 1.0417,
+      "step": 595
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999638311475925e-05,
+      "loss": 1.0322,
+      "step": 600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999631041626003e-05,
+      "loss": 1.0344,
+      "step": 605
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999623699442842e-05,
+      "loss": 1.0323,
+      "step": 610
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999616284926547e-05,
+      "loss": 1.0191,
+      "step": 615
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999608798077228e-05,
+      "loss": 1.0051,
+      "step": 620
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999601238894992e-05,
+      "loss": 1.0618,
+      "step": 625
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999593607379947e-05,
+      "loss": 1.0371,
+      "step": 630
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999585903532206e-05,
+      "loss": 1.0368,
+      "step": 635
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999578127351879e-05,
+      "loss": 1.0598,
+      "step": 640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999570278839079e-05,
+      "loss": 1.018,
+      "step": 645
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999562357993918e-05,
+      "loss": 1.0128,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999554364816512e-05,
+      "loss": 1.0099,
+      "step": 655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999546299306977e-05,
+      "loss": 1.0185,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99953816146543e-05,
+      "loss": 1.0499,
+      "step": 665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999529951291986e-05,
+      "loss": 1.0211,
+      "step": 670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999521668786766e-05,
+      "loss": 1.0019,
+      "step": 675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99951331394989e-05,
+      "loss": 1.0278,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999504886781479e-05,
+      "loss": 1.0327,
+      "step": 685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999496387281654e-05,
+      "loss": 1.039,
+      "step": 690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999487815450538e-05,
+      "loss": 1.0241,
+      "step": 695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999479171288255e-05,
+      "loss": 1.0455,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99947045479493e-05,
+      "loss": 1.0378,
+      "step": 705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99946166597069e-05,
+      "loss": 1.0352,
+      "step": 710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999452804815662e-05,
+      "loss": 1.0161,
+      "step": 715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999443871329973e-05,
+      "loss": 1.0189,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999434865513754e-05,
+      "loss": 1.0051,
+      "step": 725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999425787367134e-05,
+      "loss": 1.0396,
+      "step": 730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999416636890244e-05,
+      "loss": 1.0058,
+      "step": 735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999407414083218e-05,
+      "loss": 1.0184,
+      "step": 740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999398118946188e-05,
+      "loss": 1.036,
+      "step": 745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999388751479289e-05,
+      "loss": 1.0354,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999379311682656e-05,
+      "loss": 1.0348,
+      "step": 755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999369799556426e-05,
+      "loss": 1.0484,
+      "step": 760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999360215100738e-05,
+      "loss": 1.0447,
+      "step": 765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999350558315728e-05,
+      "loss": 0.9954,
+      "step": 770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999340829201538e-05,
+      "loss": 0.9995,
+      "step": 775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999331027758308e-05,
+      "loss": 1.0248,
+      "step": 780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999321153986179e-05,
+      "loss": 1.0149,
+      "step": 785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999311207885293e-05,
+      "loss": 1.0189,
+      "step": 790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999301189455799e-05,
+      "loss": 0.9989,
+      "step": 795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999291098697836e-05,
+      "loss": 1.0212,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999280935611552e-05,
+      "loss": 1.0188,
+      "step": 805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999270700197095e-05,
+      "loss": 1.0343,
+      "step": 810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999260392454613e-05,
+      "loss": 1.0164,
+      "step": 815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999250012384254e-05,
+      "loss": 1.0091,
+      "step": 820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99923955998617e-05,
+      "loss": 1.0407,
+      "step": 825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999229035260508e-05,
+      "loss": 1.0381,
+      "step": 830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999218438207425e-05,
+      "loss": 1.0027,
+      "step": 835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999207768827072e-05,
+      "loss": 1.0019,
+      "step": 840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999197027119604e-05,
+      "loss": 1.0071,
+      "step": 845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999186213085177e-05,
+      "loss": 1.016,
+      "step": 850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999175326723946e-05,
+      "loss": 1.0218,
+      "step": 855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999164368036072e-05,
+      "loss": 1.0197,
+      "step": 860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999153337021708e-05,
+      "loss": 1.0132,
+      "step": 865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999142233681016e-05,
+      "loss": 1.0108,
+      "step": 870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999131058014157e-05,
+      "loss": 1.0106,
+      "step": 875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999119810021295e-05,
+      "loss": 0.9738,
+      "step": 880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999108489702589e-05,
+      "loss": 1.0508,
+      "step": 885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999097097058204e-05,
+      "loss": 1.0135,
+      "step": 890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999085632088305e-05,
+      "loss": 1.0175,
+      "step": 895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99907409479306e-05,
+      "loss": 0.9804,
+      "step": 900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999062485172632e-05,
+      "loss": 1.012,
+      "step": 905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99905080322719e-05,
+      "loss": 1.0255,
+      "step": 910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999039048956906e-05,
+      "loss": 1.0153,
+      "step": 915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999027222361947e-05,
+      "loss": 1.0228,
+      "step": 920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999015323442485e-05,
+      "loss": 1.0188,
+      "step": 925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999003352198691e-05,
+      "loss": 1.0076,
+      "step": 930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99899130863074e-05,
+      "loss": 1.0255,
+      "step": 935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998979192738807e-05,
+      "loss": 1.0145,
+      "step": 940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998967004523065e-05,
+      "loss": 1.0349,
+      "step": 945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99895474398369e-05,
+      "loss": 1.0195,
+      "step": 950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998942411120862e-05,
+      "loss": 1.0115,
+      "step": 955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998930005934759e-05,
+      "loss": 1.0336,
+      "step": 960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998917528425557e-05,
+      "loss": 1.0095,
+      "step": 965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99890497859344e-05,
+      "loss": 1.0031,
+      "step": 970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998892356438589e-05,
+      "loss": 1.0132,
+      "step": 975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998879661961187e-05,
+      "loss": 0.989,
+      "step": 980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998866895161416e-05,
+      "loss": 1.0273,
+      "step": 985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998854056039462e-05,
+      "loss": 1.0165,
+      "step": 990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998841144595508e-05,
+      "loss": 1.0053,
+      "step": 995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998828160829745e-05,
+      "loss": 1.0298,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99881510474236e-05,
+      "loss": 1.0059,
+      "step": 1005
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99880197633354e-05,
+      "loss": 1.0117,
+      "step": 1010
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998788775603474e-05,
+      "loss": 0.9917,
+      "step": 1015
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998775502552358e-05,
+      "loss": 0.9973,
+      "step": 1020
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998762157180378e-05,
+      "loss": 0.9982,
+      "step": 1025
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998748739487731e-05,
+      "loss": 0.9968,
+      "step": 1030
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99873524947461e-05,
+      "loss": 1.0177,
+      "step": 1035
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998721687141211e-05,
+      "loss": 1.0194,
+      "step": 1040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998708052487728e-05,
+      "loss": 0.9801,
+      "step": 1045
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99869434551436e-05,
+      "loss": 0.9892,
+      "step": 1050
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998680566221305e-05,
+      "loss": 1.0166,
+      "step": 1055
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998666714608763e-05,
+      "loss": 1.0162,
+      "step": 1060
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998652790676934e-05,
+      "loss": 1.0096,
+      "step": 1065
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998638794426019e-05,
+      "loss": 0.9999,
+      "step": 1070
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998624725856218e-05,
+      "loss": 1.0033,
+      "step": 1075
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99861058496774e-05,
+      "loss": 1.0171,
+      "step": 1080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998596371760787e-05,
+      "loss": 1.0059,
+      "step": 1085
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998582086235563e-05,
+      "loss": 1.0121,
+      "step": 1090
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998567728392276e-05,
+      "loss": 0.9755,
+      "step": 1095
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998553298231135e-05,
+      "loss": 1.0027,
+      "step": 1100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998538795752347e-05,
+      "loss": 0.9963,
+      "step": 1105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998524220956123e-05,
+      "loss": 0.9887,
+      "step": 1110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998509573842672e-05,
+      "loss": 0.995,
+      "step": 1115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998494854412208e-05,
+      "loss": 1.0047,
+      "step": 1120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998480062664943e-05,
+      "loss": 0.99,
+      "step": 1125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998465198601093e-05,
+      "loss": 0.9865,
+      "step": 1130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998450262220869e-05,
+      "loss": 0.9726,
+      "step": 1135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998435253524489e-05,
+      "loss": 0.9944,
+      "step": 1140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998420172512171e-05,
+      "loss": 0.9954,
+      "step": 1145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998405019184132e-05,
+      "loss": 1.0172,
+      "step": 1150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998389793540593e-05,
+      "loss": 1.0165,
+      "step": 1155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998374495581773e-05,
+      "loss": 1.0005,
+      "step": 1160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998359125307895e-05,
+      "loss": 0.9804,
+      "step": 1165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998343682719176e-05,
+      "loss": 1.0045,
+      "step": 1170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998328167815847e-05,
+      "loss": 0.9797,
+      "step": 1175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998312580598125e-05,
+      "loss": 1.005,
+      "step": 1180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998296921066243e-05,
+      "loss": 0.9875,
+      "step": 1185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998281189220422e-05,
+      "loss": 0.9963,
+      "step": 1190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998265385060892e-05,
+      "loss": 1.0019,
+      "step": 1195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998249508587881e-05,
+      "loss": 1.0036,
+      "step": 1200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998233559801618e-05,
+      "loss": 1.0108,
+      "step": 1205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998217538702335e-05,
+      "loss": 0.9918,
+      "step": 1210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998201445290263e-05,
+      "loss": 1.0265,
+      "step": 1215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998185279565635e-05,
+      "loss": 0.9929,
+      "step": 1220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998169041528685e-05,
+      "loss": 1.0016,
+      "step": 1225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998152731179647e-05,
+      "loss": 1.0004,
+      "step": 1230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998136348518759e-05,
+      "loss": 0.9969,
+      "step": 1235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998119893546259e-05,
+      "loss": 0.9858,
+      "step": 1240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998103366262379e-05,
+      "loss": 0.9962,
+      "step": 1245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998086766667365e-05,
+      "loss": 0.9962,
+      "step": 1250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998070094761453e-05,
+      "loss": 1.0064,
+      "step": 1255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998053350544885e-05,
+      "loss": 1.0161,
+      "step": 1260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998036534017905e-05,
+      "loss": 0.9969,
+      "step": 1265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998019645180754e-05,
+      "loss": 0.9815,
+      "step": 1270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998002684033679e-05,
+      "loss": 1.0107,
+      "step": 1275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997985650576922e-05,
+      "loss": 1.0042,
+      "step": 1280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99796854481073e-05,
+      "loss": 1.0125,
+      "step": 1285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997951366735354e-05,
+      "loss": 0.9897,
+      "step": 1290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997934116351039e-05,
+      "loss": 0.987,
+      "step": 1295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997916793658037e-05,
+      "loss": 1.0036,
+      "step": 1300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997899398656596e-05,
+      "loss": 0.975,
+      "step": 1305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997881931346971e-05,
+      "loss": 1.0073,
+      "step": 1310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99786439172941e-05,
+      "loss": 0.9911,
+      "step": 1315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997846779804171e-05,
+      "loss": 1.0022,
+      "step": 1320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997829095571507e-05,
+      "loss": 1.0017,
+      "step": 1325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997811339031675e-05,
+      "loss": 0.9843,
+      "step": 1330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99779351018493e-05,
+      "loss": 1.0268,
+      "step": 1335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997775609031531e-05,
+      "loss": 0.999,
+      "step": 1340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997757635571738e-05,
+      "loss": 1.0134,
+      "step": 1345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997739589805809e-05,
+      "loss": 0.9971,
+      "step": 1350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997721471734007e-05,
+      "loss": 1.0123,
+      "step": 1355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997703281356592e-05,
+      "loss": 1.0082,
+      "step": 1360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997685018673829e-05,
+      "loss": 0.9792,
+      "step": 1365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997666683685983e-05,
+      "loss": 1.0079,
+      "step": 1370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997648276393315e-05,
+      "loss": 1.0031,
+      "step": 1375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997629796796097e-05,
+      "loss": 0.9783,
+      "step": 1380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997611244894591e-05,
+      "loss": 0.9834,
+      "step": 1385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99759262068907e-05,
+      "loss": 0.9937,
+      "step": 1390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9975739241798e-05,
+      "loss": 1.0115,
+      "step": 1395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997555155367054e-05,
+      "loss": 0.9858,
+      "step": 1400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997536314251102e-05,
+      "loss": 0.9898,
+      "step": 1405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997517400832217e-05,
+      "loss": 0.9791,
+      "step": 1410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997498415110673e-05,
+      "loss": 0.9797,
+      "step": 1415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997479357086744e-05,
+      "loss": 0.99,
+      "step": 1420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997460226760706e-05,
+      "loss": 0.9908,
+      "step": 1425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997441024132836e-05,
+      "loss": 1.0023,
+      "step": 1430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997421749203412e-05,
+      "loss": 0.9882,
+      "step": 1435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997402401972713e-05,
+      "loss": 0.9883,
+      "step": 1440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997382982441017e-05,
+      "loss": 0.978,
+      "step": 1445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997363490608608e-05,
+      "loss": 1.0027,
+      "step": 1450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997343926475765e-05,
+      "loss": 0.9826,
+      "step": 1455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997324290042773e-05,
+      "loss": 0.9806,
+      "step": 1460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997304581309915e-05,
+      "loss": 0.9916,
+      "step": 1465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997284800277478e-05,
+      "loss": 0.9776,
+      "step": 1470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997264946945746e-05,
+      "loss": 1.0079,
+      "step": 1475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997245021315007e-05,
+      "loss": 0.9773,
+      "step": 1480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997225023385548e-05,
+      "loss": 0.9809,
+      "step": 1485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997204953157662e-05,
+      "loss": 0.9851,
+      "step": 1490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997184810631637e-05,
+      "loss": 0.9815,
+      "step": 1495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997164595807765e-05,
+      "loss": 0.9682,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997144308686335e-05,
+      "loss": 0.9945,
+      "step": 1505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997123949267646e-05,
+      "loss": 0.9747,
+      "step": 1510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99710351755199e-05,
+      "loss": 0.9982,
+      "step": 1515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997083013539661e-05,
+      "loss": 0.9502,
+      "step": 1520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997062437230957e-05,
+      "loss": 0.9955,
+      "step": 1525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997041788626177e-05,
+      "loss": 0.9922,
+      "step": 1530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997021067725619e-05,
+      "loss": 0.9834,
+      "step": 1535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997000274529582e-05,
+      "loss": 0.9851,
+      "step": 1540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996979409038368e-05,
+      "loss": 0.9955,
+      "step": 1545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996958471252278e-05,
+      "loss": 1.0148,
+      "step": 1550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996937461171614e-05,
+      "loss": 0.9884,
+      "step": 1555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996916378796682e-05,
+      "loss": 0.9882,
+      "step": 1560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996895224127787e-05,
+      "loss": 0.9972,
+      "step": 1565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996873997165233e-05,
+      "loss": 0.9859,
+      "step": 1570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996852697909329e-05,
+      "loss": 0.9798,
+      "step": 1575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996831326360381e-05,
+      "loss": 0.9785,
+      "step": 1580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996809882518701e-05,
+      "loss": 0.9972,
+      "step": 1585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996788366384597e-05,
+      "loss": 1.0099,
+      "step": 1590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996766777958381e-05,
+      "loss": 0.9895,
+      "step": 1595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996745117240365e-05,
+      "loss": 1.0025,
+      "step": 1600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996723384230864e-05,
+      "loss": 0.9829,
+      "step": 1605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996701578930191e-05,
+      "loss": 0.9969,
+      "step": 1610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996679701338662e-05,
+      "loss": 0.9879,
+      "step": 1615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996657751456593e-05,
+      "loss": 1.0106,
+      "step": 1620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9966357292843e-05,
+      "loss": 0.9736,
+      "step": 1625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996613634822107e-05,
+      "loss": 0.9745,
+      "step": 1630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996591468070327e-05,
+      "loss": 0.9935,
+      "step": 1635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996569229029285e-05,
+      "loss": 0.9822,
+      "step": 1640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996546917699303e-05,
+      "loss": 0.9723,
+      "step": 1645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9965245340807e-05,
+      "loss": 0.9888,
+      "step": 1650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996502078173803e-05,
+      "loss": 1.0103,
+      "step": 1655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996479549978937e-05,
+      "loss": 0.9863,
+      "step": 1660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996456949496427e-05,
+      "loss": 0.9913,
+      "step": 1665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996434276726598e-05,
+      "loss": 0.9909,
+      "step": 1670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996411531669782e-05,
+      "loss": 0.9572,
+      "step": 1675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996388714326304e-05,
+      "loss": 0.979,
+      "step": 1680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996365824696498e-05,
+      "loss": 0.9792,
+      "step": 1685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996342862780693e-05,
+      "loss": 0.9762,
+      "step": 1690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996319828579221e-05,
+      "loss": 0.9975,
+      "step": 1695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996296722092418e-05,
+      "loss": 0.9759,
+      "step": 1700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996273543320613e-05,
+      "loss": 1.0028,
+      "step": 1705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996250292264147e-05,
+      "loss": 0.961,
+      "step": 1710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996226968923351e-05,
+      "loss": 0.9954,
+      "step": 1715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996203573298566e-05,
+      "loss": 0.971,
+      "step": 1720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996180105390131e-05,
+      "loss": 0.9661,
+      "step": 1725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996156565198384e-05,
+      "loss": 0.973,
+      "step": 1730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996132952723668e-05,
+      "loss": 0.9879,
+      "step": 1735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996109267966318e-05,
+      "loss": 0.9583,
+      "step": 1740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996085510926684e-05,
+      "loss": 0.9823,
+      "step": 1745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996061681605106e-05,
+      "loss": 0.9788,
+      "step": 1750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99603778000193e-05,
+      "loss": 0.9906,
+      "step": 1755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996013806117501e-05,
+      "loss": 1.0164,
+      "step": 1760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995989759952167e-05,
+      "loss": 0.976,
+      "step": 1765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995965641506274e-05,
+      "loss": 0.9911,
+      "step": 1770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995941450780174e-05,
+      "loss": 0.9788,
+      "step": 1775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995917187774212e-05,
+      "loss": 0.9941,
+      "step": 1780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995892852488744e-05,
+      "loss": 0.9978,
+      "step": 1785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99586844492412e-05,
+      "loss": 0.9619,
+      "step": 1790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995843965080694e-05,
+      "loss": 0.9713,
+      "step": 1795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995819412958817e-05,
+      "loss": 0.9513,
+      "step": 1800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99579478855885e-05,
+      "loss": 0.9792,
+      "step": 1805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995770091881145e-05,
+      "loss": 0.987,
+      "step": 1810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995745322926059e-05,
+      "loss": 0.9924,
+      "step": 1815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995720481693952e-05,
+      "loss": 0.9475,
+      "step": 1820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995695568185184e-05,
+      "loss": 0.9948,
+      "step": 1825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995670582400112e-05,
+      "loss": 0.9876,
+      "step": 1830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995645524339103e-05,
+      "loss": 0.9999,
+      "step": 1835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995620394002514e-05,
+      "loss": 0.9576,
+      "step": 1840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995595191390712e-05,
+      "loss": 1.007,
+      "step": 1845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99556991650406e-05,
+      "loss": 0.9608,
+      "step": 1850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995544569342925e-05,
+      "loss": 0.9729,
+      "step": 1855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995519149907673e-05,
+      "loss": 0.9983,
+      "step": 1860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995493658198673e-05,
+      "loss": 0.9481,
+      "step": 1865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995468094216291e-05,
+      "loss": 1.0112,
+      "step": 1870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995442457960899e-05,
+      "loss": 0.9601,
+      "step": 1875
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995416749432867e-05,
+      "loss": 1.0077,
+      "step": 1880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995390968632568e-05,
+      "loss": 0.9687,
+      "step": 1885
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995365115560373e-05,
+      "loss": 0.974,
+      "step": 1890
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995339190216659e-05,
+      "loss": 0.9889,
+      "step": 1895
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995313192601797e-05,
+      "loss": 0.9794,
+      "step": 1900
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995287122716168e-05,
+      "loss": 1.0031,
+      "step": 1905
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995260980560146e-05,
+      "loss": 0.978,
+      "step": 1910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995234766134109e-05,
+      "loss": 0.9547,
+      "step": 1915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995208479438439e-05,
+      "loss": 0.9583,
+      "step": 1920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995182120473512e-05,
+      "loss": 0.9523,
+      "step": 1925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995155689239712e-05,
+      "loss": 0.9884,
+      "step": 1930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995129185737424e-05,
+      "loss": 0.9968,
+      "step": 1935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995102609967027e-05,
+      "loss": 0.9701,
+      "step": 1940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995075961928906e-05,
+      "loss": 0.9818,
+      "step": 1945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99504924162345e-05,
+      "loss": 0.9735,
+      "step": 1950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995022449051041e-05,
+      "loss": 0.9534,
+      "step": 1955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.994995584212072e-05,
+      "loss": 0.9737,
+      "step": 1960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994968647106925e-05,
+      "loss": 0.9737,
+      "step": 1965
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994941637735996e-05,
+      "loss": 0.9822,
+      "step": 1970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994914556099673e-05,
+      "loss": 0.9647,
+      "step": 1975
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994887402198346e-05,
+      "loss": 0.9656,
+      "step": 1980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99486017603241e-05,
+      "loss": 0.9679,
+      "step": 1985
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99483287760226e-05,
+      "loss": 0.9828,
+      "step": 1990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994805506908289e-05,
+      "loss": 0.9764,
+      "step": 1995
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994778063950894e-05,
+      "loss": 0.9729,
+      "step": 2000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994750548730472e-05,
+      "loss": 0.9665,
+      "step": 2005
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994722961247419e-05,
+      "loss": 0.9837,
+      "step": 2010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994695301502137e-05,
+      "loss": 0.9853,
+      "step": 2015
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994667569495024e-05,
+      "loss": 0.9661,
+      "step": 2020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994639765226483e-05,
+      "loss": 0.9548,
+      "step": 2025
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994611888696917e-05,
+      "loss": 0.959,
+      "step": 2030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994583939906726e-05,
+      "loss": 0.976,
+      "step": 2035
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994555918856315e-05,
+      "loss": 0.963,
+      "step": 2040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994527825546092e-05,
+      "loss": 0.9783,
+      "step": 2045
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994499659976462e-05,
+      "loss": 0.989,
+      "step": 2050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994471422147834e-05,
+      "loss": 1.0184,
+      "step": 2055
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994443112060612e-05,
+      "loss": 0.9556,
+      "step": 2060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99441472971521e-05,
+      "loss": 0.9432,
+      "step": 2065
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994386275112037e-05,
+      "loss": 0.9639,
+      "step": 2070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994357748251506e-05,
+      "loss": 0.9822,
+      "step": 2075
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994329149134028e-05,
+      "loss": 0.9976,
+      "step": 2080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994300477760017e-05,
+      "loss": 0.9629,
+      "step": 2085
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994271734129887e-05,
+      "loss": 0.9943,
+      "step": 2090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994242918244056e-05,
+      "loss": 0.9754,
+      "step": 2095
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994214030102942e-05,
+      "loss": 0.9866,
+      "step": 2100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994185069706959e-05,
+      "loss": 0.9668,
+      "step": 2105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994156037056528e-05,
+      "loss": 0.9659,
+      "step": 2110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994126932152069e-05,
+      "loss": 0.9673,
+      "step": 2115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994097754994003e-05,
+      "loss": 0.9898,
+      "step": 2120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994068505582752e-05,
+      "loss": 0.9604,
+      "step": 2125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99403918391874e-05,
+      "loss": 0.9824,
+      "step": 2130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99400979000239e-05,
+      "loss": 0.9712,
+      "step": 2135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993980323834128e-05,
+      "loss": 0.9682,
+      "step": 2140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99395078541438e-05,
+      "loss": 0.9704,
+      "step": 2145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993921174743573e-05,
+      "loss": 0.9754,
+      "step": 2150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993891491822136e-05,
+      "loss": 0.9678,
+      "step": 2155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993861736650498e-05,
+      "loss": 0.9509,
+      "step": 2160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993831909229091e-05,
+      "loss": 0.9623,
+      "step": 2165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993802009558345e-05,
+      "loss": 0.9708,
+      "step": 2170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993772037638692e-05,
+      "loss": 0.961,
+      "step": 2175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993741993470566e-05,
+      "loss": 0.9876,
+      "step": 2180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993711877054403e-05,
+      "loss": 0.9532,
+      "step": 2185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99368168839064e-05,
+      "loss": 0.9477,
+      "step": 2190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993651427479709e-05,
+      "loss": 0.9883,
+      "step": 2195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993621094322052e-05,
+      "loss": 0.9824,
+      "step": 2200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993590688918105e-05,
+      "loss": 0.9878,
+      "step": 2205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993560211268309e-05,
+      "loss": 0.97,
+      "step": 2210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993529661373107e-05,
+      "loss": 0.9718,
+      "step": 2215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993499039232938e-05,
+      "loss": 0.9773,
+      "step": 2220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993468344848246e-05,
+      "loss": 0.981,
+      "step": 2225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993437578219476e-05,
+      "loss": 0.9746,
+      "step": 2230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993406739347071e-05,
+      "loss": 0.995,
+      "step": 2235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99337582823148e-05,
+      "loss": 0.9771,
+      "step": 2240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993344844873148e-05,
+      "loss": 0.9787,
+      "step": 2245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993313789272525e-05,
+      "loss": 0.9839,
+      "step": 2250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993282661430057e-05,
+      "loss": 0.9711,
+      "step": 2255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9932514613462e-05,
+      "loss": 0.9781,
+      "step": 2260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9932201890214e-05,
+      "loss": 0.9503,
+      "step": 2265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993188844456111e-05,
+      "loss": 0.9762,
+      "step": 2270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993157427650787e-05,
+      "loss": 0.9794,
+      "step": 2275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993125938605884e-05,
+      "loss": 0.9797,
+      "step": 2280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993094377321857e-05,
+      "loss": 0.9566,
+      "step": 2285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993062743799159e-05,
+      "loss": 1.0076,
+      "step": 2290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993031038038252e-05,
+      "loss": 0.9476,
+      "step": 2295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992999260039593e-05,
+      "loss": 0.9654,
+      "step": 2300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992967409803641e-05,
+      "loss": 0.945,
+      "step": 2305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992935487330858e-05,
+      "loss": 0.9683,
+      "step": 2310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992903492621705e-05,
+      "loss": 0.9812,
+      "step": 2315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992871425676647e-05,
+      "loss": 0.9598,
+      "step": 2320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992839286496145e-05,
+      "loss": 0.9797,
+      "step": 2325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992807075080666e-05,
+      "loss": 0.9865,
+      "step": 2330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992774791430675e-05,
+      "loss": 0.9623,
+      "step": 2335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992742435546638e-05,
+      "loss": 0.9494,
+      "step": 2340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992710007429025e-05,
+      "loss": 0.9831,
+      "step": 2345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992677507078306e-05,
+      "loss": 0.9661,
+      "step": 2350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99264493449495e-05,
+      "loss": 0.9619,
+      "step": 2355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992612289679427e-05,
+      "loss": 0.9859,
+      "step": 2360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992579572632212e-05,
+      "loss": 0.9567,
+      "step": 2365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992546783353775e-05,
+      "loss": 0.9657,
+      "step": 2370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992513921844592e-05,
+      "loss": 0.9724,
+      "step": 2375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99248098810514e-05,
+      "loss": 0.9664,
+      "step": 2380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992447982135893e-05,
+      "loss": 0.9954,
+      "step": 2385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99241490393733e-05,
+      "loss": 0.9603,
+      "step": 2390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99238175350993e-05,
+      "loss": 0.9825,
+      "step": 2395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992348530854171e-05,
+      "loss": 0.9424,
+      "step": 2400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992315235970534e-05,
+      "loss": 1.0038,
+      "step": 2405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992281868859501e-05,
+      "loss": 0.9786,
+      "step": 2410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992248429521555e-05,
+      "loss": 1.0031,
+      "step": 2415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992214917957181e-05,
+      "loss": 0.9717,
+      "step": 2420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992181334166861e-05,
+      "loss": 0.9554,
+      "step": 2425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992147678151083e-05,
+      "loss": 0.957,
+      "step": 2430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992113949910333e-05,
+      "loss": 0.9814,
+      "step": 2435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9920801494451e-05,
+      "loss": 0.9767,
+      "step": 2440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992046276755871e-05,
+      "loss": 0.974,
+      "step": 2445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992012331843137e-05,
+      "loss": 0.9591,
+      "step": 2450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991978314707391e-05,
+      "loss": 0.9512,
+      "step": 2455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991944225349124e-05,
+      "loss": 0.9513,
+      "step": 2460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991910063768826e-05,
+      "loss": 0.9754,
+      "step": 2465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991875829966996e-05,
+      "loss": 0.9741,
+      "step": 2470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991841523944128e-05,
+      "loss": 0.9948,
+      "step": 2475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991807145700716e-05,
+      "loss": 0.9715,
+      "step": 2480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99177269523726e-05,
+      "loss": 0.9411,
+      "step": 2485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991738172554257e-05,
+      "loss": 0.9612,
+      "step": 2490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991703577652207e-05,
+      "loss": 0.971,
+      "step": 2495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991668910531611e-05,
+      "loss": 0.9582,
+      "step": 2500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991634171192969e-05,
+      "loss": 0.9719,
+      "step": 2505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991599359636786e-05,
+      "loss": 0.974,
+      "step": 2510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991564475863562e-05,
+      "loss": 0.9601,
+      "step": 2515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991529519873806e-05,
+      "loss": 0.9992,
+      "step": 2520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991494491668021e-05,
+      "loss": 0.9718,
+      "step": 2525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991459391246713e-05,
+      "loss": 0.9492,
+      "step": 2530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991424218610393e-05,
+      "loss": 0.9697,
+      "step": 2535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991388973759568e-05,
+      "loss": 0.9408,
+      "step": 2540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991353656694747e-05,
+      "loss": 0.9519,
+      "step": 2545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991318267416443e-05,
+      "loss": 0.9695,
+      "step": 2550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991282805925166e-05,
+      "loss": 0.953,
+      "step": 2555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991247272221431e-05,
+      "loss": 0.977,
+      "step": 2560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99121166630575e-05,
+      "loss": 0.9597,
+      "step": 2565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99117598817864e-05,
+      "loss": 0.9684,
+      "step": 2570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991140237840615e-05,
+      "loss": 0.9752,
+      "step": 2575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991104415292196e-05,
+      "loss": 0.953,
+      "step": 2580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991068520533897e-05,
+      "loss": 0.9544,
+      "step": 2585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99103255356624e-05,
+      "loss": 0.9618,
+      "step": 2590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990996514389746e-05,
+      "loss": 0.9606,
+      "step": 2595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990960403004932e-05,
+      "loss": 0.9705,
+      "step": 2600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990924219412326e-05,
+      "loss": 0.981,
+      "step": 2605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990887963612447e-05,
+      "loss": 0.9922,
+      "step": 2610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990851635605822e-05,
+      "loss": 0.9775,
+      "step": 2615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990815235392977e-05,
+      "loss": 0.9797,
+      "step": 2620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990778762974437e-05,
+      "loss": 0.9252,
+      "step": 2625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990742218350731e-05,
+      "loss": 0.9712,
+      "step": 2630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990705601522388e-05,
+      "loss": 0.9577,
+      "step": 2635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990668912489934e-05,
+      "loss": 0.9824,
+      "step": 2640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990632151253905e-05,
+      "loss": 0.9481,
+      "step": 2645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99059531781483e-05,
+      "loss": 0.9409,
+      "step": 2650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990558412173241e-05,
+      "loss": 0.9221,
+      "step": 2655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990521434329675e-05,
+      "loss": 0.9454,
+      "step": 2660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990484384284664e-05,
+      "loss": 0.9753,
+      "step": 2665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990447262038747e-05,
+      "loss": 0.9698,
+      "step": 2670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990410067592457e-05,
+      "loss": 0.9887,
+      "step": 2675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990372800946338e-05,
+      "loss": 0.9918,
+      "step": 2680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990335462100923e-05,
+      "loss": 0.946,
+      "step": 2685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990298051056756e-05,
+      "loss": 0.953,
+      "step": 2690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990260567814375e-05,
+      "loss": 0.9579,
+      "step": 2695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990223012374327e-05,
+      "loss": 0.9718,
+      "step": 2700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990185384737151e-05,
+      "loss": 0.9649,
+      "step": 2705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990147684903396e-05,
+      "loss": 0.9648,
+      "step": 2710
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990109912873601e-05,
+      "loss": 0.9684,
+      "step": 2715
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990072068648319e-05,
+      "loss": 0.974,
+      "step": 2720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990034152228094e-05,
+      "loss": 0.9631,
+      "step": 2725
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989996163613475e-05,
+      "loss": 0.924,
+      "step": 2730
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989958102805011e-05,
+      "loss": 0.9324,
+      "step": 2735
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989919969803254e-05,
+      "loss": 0.9585,
+      "step": 2740
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989881764608757e-05,
+      "loss": 0.9447,
+      "step": 2745
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989843487222068e-05,
+      "loss": 0.9407,
+      "step": 2750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989805137643744e-05,
+      "loss": 0.9678,
+      "step": 2755
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989766715874341e-05,
+      "loss": 0.9445,
+      "step": 2760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989728221914411e-05,
+      "loss": 0.9441,
+      "step": 2765
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989689655764516e-05,
+      "loss": 0.9379,
+      "step": 2770
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98965101742521e-05,
+      "loss": 0.989,
+      "step": 2775
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989612306897054e-05,
+      "loss": 0.9502,
+      "step": 2780
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989573524180607e-05,
+      "loss": 0.9626,
+      "step": 2785
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98953466927643e-05,
+      "loss": 0.9555,
+      "step": 2790
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989495742185086e-05,
+      "loss": 0.9532,
+      "step": 2795
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989456742907137e-05,
+      "loss": 0.9806,
+      "step": 2800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989417671443149e-05,
+      "loss": 0.9624,
+      "step": 2805
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989378527793685e-05,
+      "loss": 0.9542,
+      "step": 2810
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989339311959314e-05,
+      "loss": 0.9597,
+      "step": 2815
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9893000239406e-05,
+      "loss": 0.9742,
+      "step": 2820
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989260663738114e-05,
+      "loss": 0.9583,
+      "step": 2825
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989221231352426e-05,
+      "loss": 0.9653,
+      "step": 2830
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989181726784104e-05,
+      "loss": 0.9584,
+      "step": 2835
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989142150033722e-05,
+      "loss": 0.9441,
+      "step": 2840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989102501101851e-05,
+      "loss": 0.9607,
+      "step": 2845
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989062779989064e-05,
+      "loss": 0.949,
+      "step": 2850
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989022986695938e-05,
+      "loss": 0.9403,
+      "step": 2855
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988983121223048e-05,
+      "loss": 0.9837,
+      "step": 2860
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988943183570969e-05,
+      "loss": 0.9608,
+      "step": 2865
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98890317374028e-05,
+      "loss": 0.9494,
+      "step": 2870
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98886309173156e-05,
+      "loss": 0.9331,
+      "step": 2875
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98882293754539e-05,
+      "loss": 0.9601,
+      "step": 2880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988782711182348e-05,
+      "loss": 0.9762,
+      "step": 2885
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988742412643017e-05,
+      "loss": 0.9627,
+      "step": 2890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988702041927982e-05,
+      "loss": 0.9483,
+      "step": 2895
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988661599037827e-05,
+      "loss": 0.9433,
+      "step": 2900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988621083973133e-05,
+      "loss": 0.9397,
+      "step": 2905
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988580496734492e-05,
+      "loss": 0.9371,
+      "step": 2910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988539837322485e-05,
+      "loss": 0.9729,
+      "step": 2915
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988499105737707e-05,
+      "loss": 0.9568,
+      "step": 2920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988458301980741e-05,
+      "loss": 0.9745,
+      "step": 2925
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988417426052182e-05,
+      "loss": 0.9363,
+      "step": 2930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988376477952619e-05,
+      "loss": 0.9477,
+      "step": 2935
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988335457682644e-05,
+      "loss": 0.9821,
+      "step": 2940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988294365242853e-05,
+      "loss": 0.9804,
+      "step": 2945
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988253200633837e-05,
+      "loss": 0.9717,
+      "step": 2950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988211963856196e-05,
+      "loss": 0.9571,
+      "step": 2955
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988170654910522e-05,
+      "loss": 0.9724,
+      "step": 2960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988129273797415e-05,
+      "loss": 0.9604,
+      "step": 2965
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988087820517475e-05,
+      "loss": 0.9234,
+      "step": 2970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988046295071299e-05,
+      "loss": 0.9695,
+      "step": 2975
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988004697459488e-05,
+      "loss": 0.9592,
+      "step": 2980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987963027682645e-05,
+      "loss": 0.9661,
+      "step": 2985
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987921285741374e-05,
+      "loss": 0.9454,
+      "step": 2990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987879471636277e-05,
+      "loss": 0.9358,
+      "step": 2995
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987837585367958e-05,
+      "loss": 0.9475,
+      "step": 3000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987795626937027e-05,
+      "loss": 0.933,
+      "step": 3005
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987753596344087e-05,
+      "loss": 0.9583,
+      "step": 3010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987711493589749e-05,
+      "loss": 0.9376,
+      "step": 3015
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987669318674618e-05,
+      "loss": 0.9416,
+      "step": 3020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98762707159931e-05,
+      "loss": 0.9595,
+      "step": 3025
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987584752364431e-05,
+      "loss": 0.9855,
+      "step": 3030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987542360970596e-05,
+      "loss": 0.9597,
+      "step": 3035
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987499897418418e-05,
+      "loss": 0.9726,
+      "step": 3040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98745736170851e-05,
+      "loss": 0.9685,
+      "step": 3045
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98741475384149e-05,
+      "loss": 0.9664,
+      "step": 3050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987372073817972e-05,
+      "loss": 0.9555,
+      "step": 3055
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987329321638574e-05,
+      "loss": 0.943,
+      "step": 3060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987286497303917e-05,
+      "loss": 0.9623,
+      "step": 3065
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987243600814615e-05,
+      "loss": 0.972,
+      "step": 3070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987200632171295e-05,
+      "loss": 0.9521,
+      "step": 3075
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987157591374574e-05,
+      "loss": 0.9481,
+      "step": 3080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987114478425078e-05,
+      "loss": 0.9587,
+      "step": 3085
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987071293323429e-05,
+      "loss": 0.9447,
+      "step": 3090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987028036070251e-05,
+      "loss": 0.9517,
+      "step": 3095
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986984706666172e-05,
+      "loss": 0.9486,
+      "step": 3100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986941305111816e-05,
+      "loss": 0.9582,
+      "step": 3105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986897831407813e-05,
+      "loss": 0.9656,
+      "step": 3110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986854285554792e-05,
+      "loss": 0.9478,
+      "step": 3115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986810667553383e-05,
+      "loss": 0.9584,
+      "step": 3120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986766977404216e-05,
+      "loss": 0.9519,
+      "step": 3125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986723215107926e-05,
+      "loss": 0.9423,
+      "step": 3130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98667938066514e-05,
+      "loss": 0.9626,
+      "step": 3135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986635474076498e-05,
+      "loss": 0.9262,
+      "step": 3140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986591495342634e-05,
+      "loss": 0.9287,
+      "step": 3145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986547444464182e-05,
+      "loss": 0.958,
+      "step": 3150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986503321441782e-05,
+      "loss": 0.9664,
+      "step": 3155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986459126276071e-05,
+      "loss": 0.9388,
+      "step": 3160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986414858967689e-05,
+      "loss": 0.9616,
+      "step": 3165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986370519517277e-05,
+      "loss": 0.9589,
+      "step": 3170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986326107925472e-05,
+      "loss": 0.954,
+      "step": 3175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986281624192923e-05,
+      "loss": 0.9443,
+      "step": 3180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98623706832027e-05,
+      "loss": 0.9544,
+      "step": 3185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986192440308159e-05,
+      "loss": 0.9629,
+      "step": 3190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986147740157235e-05,
+      "loss": 0.9422,
+      "step": 3195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986102967868143e-05,
+      "loss": 0.9301,
+      "step": 3200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986058123441533e-05,
+      "loss": 0.9615,
+      "step": 3205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986013206878054e-05,
+      "loss": 0.9521,
+      "step": 3210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985968218178354e-05,
+      "loss": 0.9564,
+      "step": 3215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985923157343086e-05,
+      "loss": 0.9552,
+      "step": 3220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9858780243729e-05,
+      "loss": 0.9667,
+      "step": 3225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985832819268451e-05,
+      "loss": 0.9487,
+      "step": 3230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985787542030392e-05,
+      "loss": 0.9393,
+      "step": 3235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985742192659378e-05,
+      "loss": 0.9652,
+      "step": 3240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985696771156064e-05,
+      "loss": 0.9717,
+      "step": 3245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985651277521107e-05,
+      "loss": 0.923,
+      "step": 3250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985605711755168e-05,
+      "loss": 0.9487,
+      "step": 3255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985560073858905e-05,
+      "loss": 0.9355,
+      "step": 3260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.985514363832977e-05,
+      "loss": 0.9257,
+      "step": 3265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985468581678046e-05,
+      "loss": 0.9471,
+      "step": 3270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985422727394776e-05,
+      "loss": 0.9547,
+      "step": 3275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985376800983828e-05,
+      "loss": 0.9629,
+      "step": 3280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985330802445866e-05,
+      "loss": 0.9455,
+      "step": 3285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98528473178156e-05,
+      "loss": 0.9455,
+      "step": 3290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985238588991572e-05,
+      "loss": 0.9431,
+      "step": 3295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985192374076571e-05,
+      "loss": 0.9609,
+      "step": 3300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985146087037225e-05,
+      "loss": 0.9485,
+      "step": 3305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985099727874204e-05,
+      "loss": 0.9553,
+      "step": 3310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985053296588181e-05,
+      "loss": 0.9358,
+      "step": 3315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985006793179824e-05,
+      "loss": 0.9569,
+      "step": 3320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984960217649808e-05,
+      "loss": 0.9522,
+      "step": 3325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984913569998806e-05,
+      "loss": 0.9712,
+      "step": 3330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984866850227495e-05,
+      "loss": 0.979,
+      "step": 3335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984820058336547e-05,
+      "loss": 0.9749,
+      "step": 3340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984773194326642e-05,
+      "loss": 0.9413,
+      "step": 3345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984726258198458e-05,
+      "loss": 0.9443,
+      "step": 3350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984679249952673e-05,
+      "loss": 0.9586,
+      "step": 3355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984632169589967e-05,
+      "loss": 0.9568,
+      "step": 3360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98458501711102e-05,
+      "loss": 0.9275,
+      "step": 3365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984537792516519e-05,
+      "loss": 0.9601,
+      "step": 3370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984490495807143e-05,
+      "loss": 0.9618,
+      "step": 3375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984443126983576e-05,
+      "loss": 0.9608,
+      "step": 3380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984395686046506e-05,
+      "loss": 0.9724,
+      "step": 3385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984348172996617e-05,
+      "loss": 0.9567,
+      "step": 3390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984300587834596e-05,
+      "loss": 0.9408,
+      "step": 3395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984252930561136e-05,
+      "loss": 0.9285,
+      "step": 3400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984205201176921e-05,
+      "loss": 0.951,
+      "step": 3405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984157399682644e-05,
+      "loss": 0.9324,
+      "step": 3410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984109526078995e-05,
+      "loss": 0.9259,
+      "step": 3415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98406158036667e-05,
+      "loss": 0.9536,
+      "step": 3420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98401356254636e-05,
+      "loss": 0.9704,
+      "step": 3425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983965472618762e-05,
+      "loss": 0.941,
+      "step": 3430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983917310584567e-05,
+      "loss": 0.9594,
+      "step": 3435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983869076444478e-05,
+      "loss": 0.9418,
+      "step": 3440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983820770199188e-05,
+      "loss": 0.9682,
+      "step": 3445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983772391849397e-05,
+      "loss": 0.9628,
+      "step": 3450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983723941395807e-05,
+      "loss": 0.9458,
+      "step": 3455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983675418839117e-05,
+      "loss": 0.9762,
+      "step": 3460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98362682418003e-05,
+      "loss": 0.9409,
+      "step": 3465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983578157419247e-05,
+      "loss": 0.9813,
+      "step": 3470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983529418557475e-05,
+      "loss": 0.9503,
+      "step": 3475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983480607595417e-05,
+      "loss": 0.9477,
+      "step": 3480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983431724533781e-05,
+      "loss": 0.9405,
+      "step": 3485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983382769373273e-05,
+      "loss": 0.9388,
+      "step": 3490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983333742114601e-05,
+      "loss": 0.9633,
+      "step": 3495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983284642758475e-05,
+      "loss": 0.9622,
+      "step": 3500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983235471305605e-05,
+      "loss": 0.922,
+      "step": 3505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983186227756703e-05,
+      "loss": 0.9391,
+      "step": 3510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98313691211248e-05,
+      "loss": 0.9547,
+      "step": 3515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983087524373653e-05,
+      "loss": 0.9371,
+      "step": 3520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983038064540933e-05,
+      "loss": 0.9425,
+      "step": 3525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982988532615034e-05,
+      "loss": 0.9477,
+      "step": 3530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982938928596678e-05,
+      "loss": 0.9215,
+      "step": 3535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982889252486578e-05,
+      "loss": 0.9298,
+      "step": 3540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982839504285456e-05,
+      "loss": 0.941,
+      "step": 3545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982789683994029e-05,
+      "loss": 0.9774,
+      "step": 3550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98273979161302e-05,
+      "loss": 0.9478,
+      "step": 3555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98268982714315e-05,
+      "loss": 0.9498,
+      "step": 3560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982639790585142e-05,
+      "loss": 0.9439,
+      "step": 3565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98258968193972e-05,
+      "loss": 0.9132,
+      "step": 3570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982539501207608e-05,
+      "loss": 0.9469,
+      "step": 3575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982489248389533e-05,
+      "loss": 0.933,
+      "step": 3580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982438923486221e-05,
+      "loss": 0.9278,
+      "step": 3585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982388526498401e-05,
+      "loss": 0.9245,
+      "step": 3590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9823380574268e-05,
+      "loss": 0.9437,
+      "step": 3595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982287516272153e-05,
+      "loss": 0.9146,
+      "step": 3600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982236903035188e-05,
+      "loss": 0.9315,
+      "step": 3605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982186217716636e-05,
+      "loss": 0.9393,
+      "step": 3610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982135460317235e-05,
+      "loss": 0.9539,
+      "step": 3615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982084630837714e-05,
+      "loss": 0.9791,
+      "step": 3620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98203372927881e-05,
+      "loss": 0.9368,
+      "step": 3625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98198275564126e-05,
+      "loss": 0.9365,
+      "step": 3630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981931709925803e-05,
+      "loss": 0.9814,
+      "step": 3635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981880592133174e-05,
+      "loss": 0.9837,
+      "step": 3640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981829402264115e-05,
+      "loss": 0.9369,
+      "step": 3645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981778140319369e-05,
+      "loss": 0.9369,
+      "step": 3650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981726806299671e-05,
+      "loss": 0.9518,
+      "step": 3655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98167540020577e-05,
+      "loss": 0.9459,
+      "step": 3660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981623922038407e-05,
+      "loss": 0.9269,
+      "step": 3665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981572371798327e-05,
+      "loss": 0.9352,
+      "step": 3670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981520749486276e-05,
+      "loss": 0.9423,
+      "step": 3675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981469055102999e-05,
+      "loss": 0.9563,
+      "step": 3680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981417288649247e-05,
+      "loss": 0.957,
+      "step": 3685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981365450125766e-05,
+      "loss": 0.9173,
+      "step": 3690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981313539533308e-05,
+      "loss": 0.9469,
+      "step": 3695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981261556872624e-05,
+      "loss": 0.9644,
+      "step": 3700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981209502144466e-05,
+      "loss": 0.9536,
+      "step": 3705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981157375349586e-05,
+      "loss": 0.9411,
+      "step": 3710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981105176488739e-05,
+      "loss": 0.9322,
+      "step": 3715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98105290556268e-05,
+      "loss": 0.9065,
+      "step": 3720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981000562572166e-05,
+      "loss": 0.9351,
+      "step": 3725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980948147517953e-05,
+      "loss": 0.966,
+      "step": 3730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980895660400799e-05,
+      "loss": 0.9216,
+      "step": 3735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980843101221467e-05,
+      "loss": 0.9335,
+      "step": 3740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980790469980712e-05,
+      "loss": 0.9228,
+      "step": 3745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9807377666793e-05,
+      "loss": 0.9461,
+      "step": 3750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98068499131799e-05,
+      "loss": 0.9523,
+      "step": 3755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980632143897548e-05,
+      "loss": 0.931,
+      "step": 3760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980579224418738e-05,
+      "loss": 0.9803,
+      "step": 3765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980526232882325e-05,
+      "loss": 0.9319,
+      "step": 3770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980473169289076e-05,
+      "loss": 0.9586,
+      "step": 3775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980420033639759e-05,
+      "loss": 0.9355,
+      "step": 3780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980366825935143e-05,
+      "loss": 0.9495,
+      "step": 3785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980313546175997e-05,
+      "loss": 0.9278,
+      "step": 3790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980260194363092e-05,
+      "loss": 0.9161,
+      "step": 3795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980206770497199e-05,
+      "loss": 0.9615,
+      "step": 3800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980153274579092e-05,
+      "loss": 0.9334,
+      "step": 3805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980099706609547e-05,
+      "loss": 0.9547,
+      "step": 3810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980046066589334e-05,
+      "loss": 0.945,
+      "step": 3815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979992354519235e-05,
+      "loss": 0.938,
+      "step": 3820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979938570400023e-05,
+      "loss": 0.9603,
+      "step": 3825
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979884714232476e-05,
+      "loss": 0.9362,
+      "step": 3830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979830786017375e-05,
+      "loss": 0.9541,
+      "step": 3835
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9797767857555e-05,
+      "loss": 0.9351,
+      "step": 3840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979722713447633e-05,
+      "loss": 0.9355,
+      "step": 3845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979668569094553e-05,
+      "loss": 0.9527,
+      "step": 3850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979614352697046e-05,
+      "loss": 0.9584,
+      "step": 3855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979560064255897e-05,
+      "loss": 0.9346,
+      "step": 3860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979505703771888e-05,
+      "loss": 0.9661,
+      "step": 3865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97945127124581e-05,
+      "loss": 0.9625,
+      "step": 3870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979396766678447e-05,
+      "loss": 0.967,
+      "step": 3875
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97934219007059e-05,
+      "loss": 0.9148,
+      "step": 3880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979287541423026e-05,
+      "loss": 0.9511,
+      "step": 3885
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979232820736548e-05,
+      "loss": 0.9432,
+      "step": 3890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979178028011947e-05,
+      "loss": 0.9645,
+      "step": 3895
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979123163250015e-05,
+      "loss": 0.9383,
+      "step": 3900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979068226451546e-05,
+      "loss": 0.9474,
+      "step": 3905
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979013217617336e-05,
+      "loss": 0.9216,
+      "step": 3910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97895813674818e-05,
+      "loss": 0.9336,
+      "step": 3915
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978902983844873e-05,
+      "loss": 0.9506,
+      "step": 3920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978847758908216e-05,
+      "loss": 0.9319,
+      "step": 3925
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978792461939007e-05,
+      "loss": 0.9224,
+      "step": 3930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978737092938046e-05,
+      "loss": 0.9442,
+      "step": 3935
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978681651906132e-05,
+      "loss": 0.9486,
+      "step": 3940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978626138844071e-05,
+      "loss": 0.9363,
+      "step": 3945
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978570553752664e-05,
+      "loss": 0.9428,
+      "step": 3950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978514896632714e-05,
+      "loss": 0.9187,
+      "step": 3955
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978459167485029e-05,
+      "loss": 0.9456,
+      "step": 3960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978403366310412e-05,
+      "loss": 0.9465,
+      "step": 3965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978347493109672e-05,
+      "loss": 0.9612,
+      "step": 3970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978291547883619e-05,
+      "loss": 0.9245,
+      "step": 3975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97823553063306e-05,
+      "loss": 0.9592,
+      "step": 3980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978179441358805e-05,
+      "loss": 0.9433,
+      "step": 3985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97812328006167e-05,
+      "loss": 0.9517,
+      "step": 3990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978067046742462e-05,
+      "loss": 0.9348,
+      "step": 3995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978010741401998e-05,
+      "loss": 0.9406,
+      "step": 4000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977954364041091e-05,
+      "loss": 0.9331,
+      "step": 4005
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977897914660557e-05,
+      "loss": 0.94,
+      "step": 4010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977841393261211e-05,
+      "loss": 0.9364,
+      "step": 4015
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977784799843875e-05,
+      "loss": 0.9489,
+      "step": 4020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977728134409366e-05,
+      "loss": 0.9312,
+      "step": 4025
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9776713969585e-05,
+      "loss": 0.9616,
+      "step": 4030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977614587492103e-05,
+      "loss": 0.984,
+      "step": 4035
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977557706010996e-05,
+      "loss": 0.9375,
+      "step": 4040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977500752515998e-05,
+      "loss": 0.9393,
+      "step": 4045
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977443727007938e-05,
+      "loss": 0.9727,
+      "step": 4050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977386629487638e-05,
+      "loss": 0.9621,
+      "step": 4055
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977329459955925e-05,
+      "loss": 0.9513,
+      "step": 4060
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977272218413627e-05,
+      "loss": 0.944,
+      "step": 4065
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977214904861569e-05,
+      "loss": 0.9381,
+      "step": 4070
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977157519300585e-05,
+      "loss": 0.9394,
+      "step": 4075
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977100061731501e-05,
+      "loss": 0.941,
+      "step": 4080
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977042532155151e-05,
+      "loss": 0.9302,
+      "step": 4085
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976984930572366e-05,
+      "loss": 0.9338,
+      "step": 4090
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976927256983979e-05,
+      "loss": 0.9403,
+      "step": 4095
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976869511390824e-05,
+      "loss": 0.9695,
+      "step": 4100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97681169379374e-05,
+      "loss": 0.9358,
+      "step": 4105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97675380419356e-05,
+      "loss": 0.9589,
+      "step": 4110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976695842591122e-05,
+      "loss": 0.9639,
+      "step": 4115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976637808987265e-05,
+      "loss": 0.9569,
+      "step": 4120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976579703382829e-05,
+      "loss": 0.9129,
+      "step": 4125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976521525778653e-05,
+      "loss": 0.9538,
+      "step": 4130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976463276175582e-05,
+      "loss": 0.9478,
+      "step": 4135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976404954574457e-05,
+      "loss": 0.9529,
+      "step": 4140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97634656097612e-05,
+      "loss": 0.9185,
+      "step": 4145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976288095381417e-05,
+      "loss": 0.936,
+      "step": 4150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976229557791194e-05,
+      "loss": 0.9493,
+      "step": 4155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9761709482063e-05,
+      "loss": 0.9666,
+      "step": 4160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97611226662758e-05,
+      "loss": 0.9473,
+      "step": 4165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976053513055885e-05,
+      "loss": 0.9225,
+      "step": 4170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975994687492063e-05,
+      "loss": 0.9266,
+      "step": 4175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975935789936966e-05,
+      "loss": 0.9468,
+      "step": 4180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975876820391448e-05,
+      "loss": 0.9177,
+      "step": 4185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975817778856358e-05,
+      "loss": 0.9347,
+      "step": 4190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975758665332556e-05,
+      "loss": 0.9192,
+      "step": 4195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975699479820891e-05,
+      "loss": 0.9571,
+      "step": 4200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975640222322223e-05,
+      "loss": 0.9516,
+      "step": 4205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975580892837407e-05,
+      "loss": 0.9655,
+      "step": 4210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975521491367305e-05,
+      "loss": 0.9416,
+      "step": 4215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975462017912773e-05,
+      "loss": 0.9295,
+      "step": 4220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975402472474673e-05,
+      "loss": 0.9416,
+      "step": 4225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975342855053866e-05,
+      "loss": 0.9501,
+      "step": 4230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975283165651215e-05,
+      "loss": 0.9416,
+      "step": 4235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975223404267583e-05,
+      "loss": 0.9365,
+      "step": 4240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975163570903835e-05,
+      "loss": 0.9446,
+      "step": 4245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975103665560836e-05,
+      "loss": 0.9358,
+      "step": 4250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975043688239453e-05,
+      "loss": 0.9352,
+      "step": 4255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974983638940557e-05,
+      "loss": 0.9532,
+      "step": 4260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97492351766501e-05,
+      "loss": 0.94,
+      "step": 4265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974863324413688e-05,
+      "loss": 0.9413,
+      "step": 4270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974803059187457e-05,
+      "loss": 0.9475,
+      "step": 4275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974742721987194e-05,
+      "loss": 0.9411,
+      "step": 4280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974682312813768e-05,
+      "loss": 0.9447,
+      "step": 4285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974621831668053e-05,
+      "loss": 0.9318,
+      "step": 4290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974561278550927e-05,
+      "loss": 0.9379,
+      "step": 4295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974500653463263e-05,
+      "loss": 0.955,
+      "step": 4300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974439956405941e-05,
+      "loss": 0.9581,
+      "step": 4305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974379187379837e-05,
+      "loss": 0.9625,
+      "step": 4310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974318346385829e-05,
+      "loss": 0.9549,
+      "step": 4315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9742574334248e-05,
+      "loss": 0.9448,
+      "step": 4320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974196448497632e-05,
+      "loss": 0.9332,
+      "step": 4325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974135391605204e-05,
+      "loss": 0.9563,
+      "step": 4330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9740742627484e-05,
+      "loss": 0.9404,
+      "step": 4335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974013061928107e-05,
+      "loss": 0.9479,
+      "step": 4340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973951789145208e-05,
+      "loss": 0.9543,
+      "step": 4345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973890444400591e-05,
+      "loss": 0.9317,
+      "step": 4350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973829027695143e-05,
+      "loss": 0.9366,
+      "step": 4355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97376753902975e-05,
+      "loss": 0.922,
+      "step": 4360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973705978405307e-05,
+      "loss": 0.9289,
+      "step": 4365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973644345822699e-05,
+      "loss": 0.947,
+      "step": 4370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973582641282822e-05,
+      "loss": 0.9216,
+      "step": 4375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973520864786565e-05,
+      "loss": 0.934,
+      "step": 4380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973459016334825e-05,
+      "loss": 0.9737,
+      "step": 4385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973397095928497e-05,
+      "loss": 0.953,
+      "step": 4390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973335103568473e-05,
+      "loss": 0.9317,
+      "step": 4395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973273039255652e-05,
+      "loss": 0.9299,
+      "step": 4400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973210902990935e-05,
+      "loss": 0.9472,
+      "step": 4405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973148694775217e-05,
+      "loss": 0.929,
+      "step": 4410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973086414609398e-05,
+      "loss": 0.9237,
+      "step": 4415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973024062494381e-05,
+      "loss": 0.9542,
+      "step": 4420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972961638431066e-05,
+      "loss": 0.9552,
+      "step": 4425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972899142420359e-05,
+      "loss": 0.9095,
+      "step": 4430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972836574463162e-05,
+      "loss": 0.976,
+      "step": 4435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972773934560381e-05,
+      "loss": 0.9407,
+      "step": 4440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972711222712923e-05,
+      "loss": 0.9582,
+      "step": 4445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972648438921693e-05,
+      "loss": 0.9533,
+      "step": 4450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972585583187601e-05,
+      "loss": 0.9241,
+      "step": 4455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972522655511555e-05,
+      "loss": 0.9258,
+      "step": 4460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97245965589447e-05,
+      "loss": 0.9254,
+      "step": 4465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972396584337252e-05,
+      "loss": 0.9212,
+      "step": 4470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972333440840816e-05,
+      "loss": 0.9383,
+      "step": 4475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972270225406073e-05,
+      "loss": 0.9481,
+      "step": 4480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972206938033941e-05,
+      "loss": 0.9321,
+      "step": 4485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972143578725335e-05,
+      "loss": 0.9289,
+      "step": 4490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97208014748117e-05,
+      "loss": 0.9415,
+      "step": 4495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972016644302364e-05,
+      "loss": 0.9151,
+      "step": 4500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.971953069189838e-05,
+      "loss": 0.9308,
+      "step": 4505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97188942214451e-05,
+      "loss": 0.9534,
+      "step": 4510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.971825703167299e-05,
+      "loss": 0.9283,
+      "step": 4515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97176191225913e-05,
+      "loss": 0.9422,
+      "step": 4520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.971698049420925e-05,
+      "loss": 0.9372,
+      "step": 4525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.971634114653607e-05,
+      "loss": 0.9204,
+      "step": 4530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.971570107958103e-05,
+      "loss": 0.9344,
+      "step": 4535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.971506029335336e-05,
+      "loss": 0.95,
+      "step": 4540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.971441878786237e-05,
+      "loss": 0.9331,
+      "step": 4545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97137765631173e-05,
+      "loss": 0.9323,
+      "step": 4550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.971313361912747e-05,
+      "loss": 0.9243,
+      "step": 4555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.971248995590217e-05,
+      "loss": 0.9388,
+      "step": 4560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.971184557345072e-05,
+      "loss": 0.9362,
+      "step": 4565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.971120047178241e-05,
+      "loss": 0.9231,
+      "step": 4570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971055465090664e-05,
+      "loss": 0.9329,
+      "step": 4575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97099081108327e-05,
+      "loss": 0.9493,
+      "step": 4580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970926085156995e-05,
+      "loss": 0.9146,
+      "step": 4585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970861287312776e-05,
+      "loss": 0.9172,
+      "step": 4590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970796417551552e-05,
+      "loss": 0.9232,
+      "step": 4595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97073147587426e-05,
+      "loss": 0.9277,
+      "step": 4600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970666462281839e-05,
+      "loss": 0.9501,
+      "step": 4605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970601376775232e-05,
+      "loss": 0.9218,
+      "step": 4610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970536219355378e-05,
+      "loss": 0.95,
+      "step": 4615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970470990023219e-05,
+      "loss": 0.9373,
+      "step": 4620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970405688779704e-05,
+      "loss": 0.9438,
+      "step": 4625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970340315625773e-05,
+      "loss": 0.9105,
+      "step": 4630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97027487056237e-05,
+      "loss": 0.9557,
+      "step": 4635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970209353590449e-05,
+      "loss": 0.9651,
+      "step": 4640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970143764710951e-05,
+      "loss": 0.9444,
+      "step": 4645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970078103924829e-05,
+      "loss": 0.9251,
+      "step": 4650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970012371233031e-05,
+      "loss": 0.9406,
+      "step": 4655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969946566636508e-05,
+      "loss": 0.9339,
+      "step": 4660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969880690136215e-05,
+      "loss": 0.911,
+      "step": 4665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969814741733101e-05,
+      "loss": 0.9401,
+      "step": 4670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969748721428121e-05,
+      "loss": 0.9821,
+      "step": 4675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969682629222232e-05,
+      "loss": 0.9349,
+      "step": 4680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96961646511639e-05,
+      "loss": 0.935,
+      "step": 4685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969550229111551e-05,
+      "loss": 0.9612,
+      "step": 4690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969483921208673e-05,
+      "loss": 0.9449,
+      "step": 4695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969417541408717e-05,
+      "loss": 0.939,
+      "step": 4700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969351089712641e-05,
+      "loss": 0.9319,
+      "step": 4705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969284566121409e-05,
+      "loss": 0.9199,
+      "step": 4710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969217970635983e-05,
+      "loss": 0.9717,
+      "step": 4715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969151303257324e-05,
+      "loss": 0.936,
+      "step": 4720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969084563986399e-05,
+      "loss": 0.9181,
+      "step": 4725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969017752824174e-05,
+      "loss": 0.917,
+      "step": 4730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968950869771614e-05,
+      "loss": 0.9386,
+      "step": 4735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968883914829687e-05,
+      "loss": 0.927,
+      "step": 4740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968816887999362e-05,
+      "loss": 0.9303,
+      "step": 4745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968749789281606e-05,
+      "loss": 0.9265,
+      "step": 4750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968682618677395e-05,
+      "loss": 0.9245,
+      "step": 4755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968615376187698e-05,
+      "loss": 0.9446,
+      "step": 4760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968548061813487e-05,
+      "loss": 0.9715,
+      "step": 4765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968480675555735e-05,
+      "loss": 0.9349,
+      "step": 4770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968413217415422e-05,
+      "loss": 0.9053,
+      "step": 4775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96834568739352e-05,
+      "loss": 0.9249,
+      "step": 4780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968278085491007e-05,
+      "loss": 0.9401,
+      "step": 4785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968210411708859e-05,
+      "loss": 0.9353,
+      "step": 4790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968142666048058e-05,
+      "loss": 0.9099,
+      "step": 4795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968074848509584e-05,
+      "loss": 0.9265,
+      "step": 4800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968006959094415e-05,
+      "loss": 0.9315,
+      "step": 4805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967938997803537e-05,
+      "loss": 0.9112,
+      "step": 4810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967870964637932e-05,
+      "loss": 0.9523,
+      "step": 4815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967802859598581e-05,
+      "loss": 0.9404,
+      "step": 4820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967734682686474e-05,
+      "loss": 0.9359,
+      "step": 4825
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967666433902596e-05,
+      "loss": 0.951,
+      "step": 4830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967598113247933e-05,
+      "loss": 0.9389,
+      "step": 4835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967529720723476e-05,
+      "loss": 0.9253,
+      "step": 4840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96746125633021e-05,
+      "loss": 0.9557,
+      "step": 4845
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967392720069131e-05,
+      "loss": 0.9504,
+      "step": 4850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967324111941227e-05,
+      "loss": 0.9518,
+      "step": 4855
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967255431947491e-05,
+      "loss": 0.944,
+      "step": 4860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967186680088919e-05,
+      "loss": 0.9572,
+      "step": 4865
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967117856366504e-05,
+      "loss": 0.9466,
+      "step": 4870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96704896078124e-05,
+      "loss": 0.9244,
+      "step": 4875
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966979993334124e-05,
+      "loss": 0.9302,
+      "step": 4880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966910954026158e-05,
+      "loss": 0.9302,
+      "step": 4885
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966841842858338e-05,
+      "loss": 0.921,
+      "step": 4890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966772659831662e-05,
+      "loss": 0.9107,
+      "step": 4895
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966703404947134e-05,
+      "loss": 0.9257,
+      "step": 4900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966634078205754e-05,
+      "loss": 0.9343,
+      "step": 4905
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966564679608528e-05,
+      "loss": 0.9528,
+      "step": 4910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966495209156455e-05,
+      "loss": 0.9429,
+      "step": 4915
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966425666850545e-05,
+      "loss": 0.9489,
+      "step": 4920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966356052691801e-05,
+      "loss": 0.909,
+      "step": 4925
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966286366681233e-05,
+      "loss": 0.9333,
+      "step": 4930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966216608819845e-05,
+      "loss": 0.9314,
+      "step": 4935
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96614677910865e-05,
+      "loss": 0.914,
+      "step": 4940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966076877548655e-05,
+      "loss": 0.932,
+      "step": 4945
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966006904140874e-05,
+      "loss": 0.9178,
+      "step": 4950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96593685888632e-05,
+      "loss": 0.9269,
+      "step": 4955
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965866741786003e-05,
+      "loss": 0.9247,
+      "step": 4960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965796552840942e-05,
+      "loss": 0.9152,
+      "step": 4965
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965726292052148e-05,
+      "loss": 0.9394,
+      "step": 4970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965655959420639e-05,
+      "loss": 0.9551,
+      "step": 4975
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965585554947434e-05,
+      "loss": 0.9218,
+      "step": 4980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965515078633549e-05,
+      "loss": 0.9457,
+      "step": 4985
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965444530480006e-05,
+      "loss": 0.9091,
+      "step": 4990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965373910487827e-05,
+      "loss": 0.9216,
+      "step": 4995
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965303218658028e-05,
+      "loss": 0.9171,
+      "step": 5000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965232454991637e-05,
+      "loss": 0.942,
+      "step": 5005
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965161619489677e-05,
+      "loss": 0.9308,
+      "step": 5010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965090712153171e-05,
+      "loss": 0.9286,
+      "step": 5015
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965019732983145e-05,
+      "loss": 0.9377,
+      "step": 5020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964948681980628e-05,
+      "loss": 0.9247,
+      "step": 5025
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964877559146646e-05,
+      "loss": 0.9532,
+      "step": 5030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964806364482229e-05,
+      "loss": 0.9557,
+      "step": 5035
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964735097988408e-05,
+      "loss": 0.9133,
+      "step": 5040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964663759666209e-05,
+      "loss": 0.9528,
+      "step": 5045
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96459234951667e-05,
+      "loss": 0.956,
+      "step": 5050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964520867540821e-05,
+      "loss": 0.9457,
+      "step": 5055
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964449313739699e-05,
+      "loss": 0.9258,
+      "step": 5060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964377688114334e-05,
+      "loss": 0.9532,
+      "step": 5065
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96430599066577e-05,
+      "loss": 0.9487,
+      "step": 5070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964234221395037e-05,
+      "loss": 0.9323,
+      "step": 5075
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964162380303178e-05,
+      "loss": 0.924,
+      "step": 5080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964090467391228e-05,
+      "loss": 0.9415,
+      "step": 5085
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964018482660231e-05,
+      "loss": 0.913,
+      "step": 5090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963946426111229e-05,
+      "loss": 0.9457,
+      "step": 5095
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963874297745262e-05,
+      "loss": 0.9442,
+      "step": 5100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963802097563376e-05,
+      "loss": 0.9316,
+      "step": 5105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963729825566612e-05,
+      "loss": 0.93,
+      "step": 5110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963657481756017e-05,
+      "loss": 0.9235,
+      "step": 5115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96358506613264e-05,
+      "loss": 0.9344,
+      "step": 5120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963512578697527e-05,
+      "loss": 0.9254,
+      "step": 5125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963440019451726e-05,
+      "loss": 0.9348,
+      "step": 5130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963367388396287e-05,
+      "loss": 0.9321,
+      "step": 5135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963294685532263e-05,
+      "loss": 0.9702,
+      "step": 5140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963221910860704e-05,
+      "loss": 0.9461,
+      "step": 5145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963149064382662e-05,
+      "loss": 0.937,
+      "step": 5150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963076146099192e-05,
+      "loss": 0.9238,
+      "step": 5155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96300315601135e-05,
+      "loss": 0.9382,
+      "step": 5160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962930094120191e-05,
+      "loss": 0.9298,
+      "step": 5165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962856960426771e-05,
+      "loss": 0.9425,
+      "step": 5170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96278375493215e-05,
+      "loss": 0.9437,
+      "step": 5175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962710477637384e-05,
+      "loss": 0.9287,
+      "step": 5180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962637128543538e-05,
+      "loss": 0.9242,
+      "step": 5185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962563707651669e-05,
+      "loss": 0.9407,
+      "step": 5190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96249021496284e-05,
+      "loss": 0.9465,
+      "step": 5195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962416650478117e-05,
+      "loss": 0.9092,
+      "step": 5200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962343014198562e-05,
+      "loss": 0.947,
+      "step": 5205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96226930612524e-05,
+      "loss": 0.9164,
+      "step": 5210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962195526259219e-05,
+      "loss": 0.9327,
+      "step": 5215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962121674601563e-05,
+      "loss": 0.9242,
+      "step": 5220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962047751153346e-05,
+      "loss": 0.9204,
+      "step": 5225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961973755915632e-05,
+      "loss": 0.9433,
+      "step": 5230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961899688889495e-05,
+      "loss": 0.9638,
+      "step": 5235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961825550076007e-05,
+      "loss": 0.9443,
+      "step": 5240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961751339476238e-05,
+      "loss": 0.924,
+      "step": 5245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961677057091264e-05,
+      "loss": 0.9251,
+      "step": 5250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961602702922157e-05,
+      "loss": 0.9336,
+      "step": 5255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961528276969994e-05,
+      "loss": 0.9146,
+      "step": 5260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961453779235854e-05,
+      "loss": 0.9203,
+      "step": 5265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961379209720812e-05,
+      "loss": 0.9191,
+      "step": 5270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961304568425946e-05,
+      "loss": 0.942,
+      "step": 5275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961229855352338e-05,
+      "loss": 0.9305,
+      "step": 5280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961155070501072e-05,
+      "loss": 0.9366,
+      "step": 5285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961080213873223e-05,
+      "loss": 0.9374,
+      "step": 5290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961005285469877e-05,
+      "loss": 0.9277,
+      "step": 5295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960930285292121e-05,
+      "loss": 0.914,
+      "step": 5300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960855213341039e-05,
+      "loss": 0.9343,
+      "step": 5305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960780069617712e-05,
+      "loss": 0.9318,
+      "step": 5310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960704854123232e-05,
+      "loss": 0.9076,
+      "step": 5315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960629566858688e-05,
+      "loss": 0.9397,
+      "step": 5320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960554207825166e-05,
+      "loss": 0.925,
+      "step": 5325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960478777023756e-05,
+      "loss": 0.9469,
+      "step": 5330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960403274455552e-05,
+      "loss": 0.9249,
+      "step": 5335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960327700121648e-05,
+      "loss": 0.9221,
+      "step": 5340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960252054023132e-05,
+      "loss": 0.9128,
+      "step": 5345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960176336161101e-05,
+      "loss": 0.9429,
+      "step": 5350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960100546536652e-05,
+      "loss": 0.9444,
+      "step": 5355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960024685150879e-05,
+      "loss": 0.9399,
+      "step": 5360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95994875200488e-05,
+      "loss": 0.9396,
+      "step": 5365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959872747099755e-05,
+      "loss": 0.9582,
+      "step": 5370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959796670436603e-05,
+      "loss": 0.9432,
+      "step": 5375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959720522016524e-05,
+      "loss": 0.9199,
+      "step": 5380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959644301840621e-05,
+      "loss": 0.9337,
+      "step": 5385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959568009909996e-05,
+      "loss": 0.9333,
+      "step": 5390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959491646225752e-05,
+      "loss": 0.9414,
+      "step": 5395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959415210788993e-05,
+      "loss": 0.919,
+      "step": 5400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959338703600828e-05,
+      "loss": 0.9364,
+      "step": 5405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959262124662363e-05,
+      "loss": 0.9445,
+      "step": 5410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959185473974704e-05,
+      "loss": 0.9235,
+      "step": 5415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959108751538962e-05,
+      "loss": 0.9262,
+      "step": 5420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959031957356244e-05,
+      "loss": 0.9392,
+      "step": 5425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958955091427664e-05,
+      "loss": 0.9408,
+      "step": 5430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958878153754334e-05,
+      "loss": 0.9302,
+      "step": 5435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958801144337367e-05,
+      "loss": 0.9265,
+      "step": 5440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958724063177874e-05,
+      "loss": 0.9292,
+      "step": 5445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958646910276974e-05,
+      "loss": 0.955,
+      "step": 5450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958569685635782e-05,
+      "loss": 0.9467,
+      "step": 5455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958492389255415e-05,
+      "loss": 0.9141,
+      "step": 5460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958415021136991e-05,
+      "loss": 0.9203,
+      "step": 5465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95833758128163e-05,
+      "loss": 0.9401,
+      "step": 5470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958260069690453e-05,
+      "loss": 0.9308,
+      "step": 5475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95818248636458e-05,
+      "loss": 0.9298,
+      "step": 5480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958104831305133e-05,
+      "loss": 0.9264,
+      "step": 5485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958027104513236e-05,
+      "loss": 0.9474,
+      "step": 5490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957949305990018e-05,
+      "loss": 0.9436,
+      "step": 5495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957871435736596e-05,
+      "loss": 0.9315,
+      "step": 5500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957793493754104e-05,
+      "loss": 0.9285,
+      "step": 5505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957715480043666e-05,
+      "loss": 0.9173,
+      "step": 5510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957637394606412e-05,
+      "loss": 0.9135,
+      "step": 5515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95755923744347e-05,
+      "loss": 0.9513,
+      "step": 5520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957481008555973e-05,
+      "loss": 0.9129,
+      "step": 5525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957402707945052e-05,
+      "loss": 0.9407,
+      "step": 5530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95732433561184e-05,
+      "loss": 0.9286,
+      "step": 5535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957245891557469e-05,
+      "loss": 0.9137,
+      "step": 5540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957167375783076e-05,
+      "loss": 0.9439,
+      "step": 5545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957088788289797e-05,
+      "loss": 0.9363,
+      "step": 5550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957010129078767e-05,
+      "loss": 0.9624,
+      "step": 5555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956931398151126e-05,
+      "loss": 0.9336,
+      "step": 5560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956852595508013e-05,
+      "loss": 0.939,
+      "step": 5565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956773721150567e-05,
+      "loss": 0.9347,
+      "step": 5570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95669477507993e-05,
+      "loss": 0.9275,
+      "step": 5575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956615757297243e-05,
+      "loss": 0.9163,
+      "step": 5580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956536667803651e-05,
+      "loss": 0.8917,
+      "step": 5585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956457506600297e-05,
+      "loss": 0.937,
+      "step": 5590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956378273688327e-05,
+      "loss": 0.9149,
+      "step": 5595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956298969068885e-05,
+      "loss": 0.9508,
+      "step": 5600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956219592743121e-05,
+      "loss": 0.913,
+      "step": 5605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956140144712183e-05,
+      "loss": 0.9154,
+      "step": 5610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95606062497722e-05,
+      "loss": 0.9406,
+      "step": 5615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955981033539384e-05,
+      "loss": 0.932,
+      "step": 5620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955901370399822e-05,
+      "loss": 0.9144,
+      "step": 5625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955821635559693e-05,
+      "loss": 0.9145,
+      "step": 5630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955741829020145e-05,
+      "loss": 0.9292,
+      "step": 5635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955661950782333e-05,
+      "loss": 0.9316,
+      "step": 5640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955582000847416e-05,
+      "loss": 0.9481,
+      "step": 5645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955501979216551e-05,
+      "loss": 0.9235,
+      "step": 5650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955421885890892e-05,
+      "loss": 0.9247,
+      "step": 5655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955341720871601e-05,
+      "loss": 0.9322,
+      "step": 5660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955261484159834e-05,
+      "loss": 0.9257,
+      "step": 5665
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955181175756756e-05,
+      "loss": 0.9269,
+      "step": 5670
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955100795663525e-05,
+      "loss": 0.9187,
+      "step": 5675
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95502034388131e-05,
+      "loss": 0.921,
+      "step": 5680
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954939820411267e-05,
+      "loss": 0.9108,
+      "step": 5685
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954859225254567e-05,
+      "loss": 0.9278,
+      "step": 5690
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954778558412374e-05,
+      "loss": 0.9421,
+      "step": 5695
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954697819885854e-05,
+      "loss": 0.9385,
+      "step": 5700
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954617009676177e-05,
+      "loss": 0.9331,
+      "step": 5705
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954536127784512e-05,
+      "loss": 0.9232,
+      "step": 5710
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954455174212027e-05,
+      "loss": 0.9252,
+      "step": 5715
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954374148959896e-05,
+      "loss": 0.9198,
+      "step": 5720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954293052029289e-05,
+      "loss": 0.9284,
+      "step": 5725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954211883421381e-05,
+      "loss": 0.9364,
+      "step": 5730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954130643137345e-05,
+      "loss": 0.9114,
+      "step": 5735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954049331178358e-05,
+      "loss": 0.9284,
+      "step": 5740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953967947545595e-05,
+      "loss": 0.939,
+      "step": 5745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953886492240232e-05,
+      "loss": 0.9337,
+      "step": 5750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953804965263451e-05,
+      "loss": 0.9216,
+      "step": 5755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95372336661643e-05,
+      "loss": 0.9241,
+      "step": 5760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953641696300348e-05,
+      "loss": 0.9221,
+      "step": 5765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953559954316388e-05,
+      "loss": 0.9284,
+      "step": 5770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953478140665733e-05,
+      "loss": 0.9314,
+      "step": 5775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953396255349565e-05,
+      "loss": 0.9259,
+      "step": 5780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953314298369072e-05,
+      "loss": 0.9478,
+      "step": 5785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953232269725437e-05,
+      "loss": 0.9467,
+      "step": 5790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953150169419847e-05,
+      "loss": 0.9474,
+      "step": 5795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95306799745349e-05,
+      "loss": 0.9368,
+      "step": 5800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.952985753827554e-05,
+      "loss": 0.9341,
+      "step": 5805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95290343854323e-05,
+      "loss": 0.953,
+      "step": 5810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.952821051601711e-05,
+      "loss": 0.9359,
+      "step": 5815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.952738593004186e-05,
+      "loss": 0.9254,
+      "step": 5820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.952656062751846e-05,
+      "loss": 0.9051,
+      "step": 5825
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.952573460845891e-05,
+      "loss": 0.9393,
+      "step": 5830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.952490787287512e-05,
+      "loss": 0.9143,
+      "step": 5835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.952408042077905e-05,
+      "loss": 0.9244,
+      "step": 5840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.952325225218269e-05,
+      "loss": 0.9331,
+      "step": 5845
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.952242336709801e-05,
+      "loss": 0.9405,
+      "step": 5850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9521593765537e-05,
+      "loss": 0.9254,
+      "step": 5855
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.952076344751165e-05,
+      "loss": 0.9253,
+      "step": 5860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.951993241303401e-05,
+      "loss": 0.9386,
+      "step": 5865
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.951910066211609e-05,
+      "loss": 0.9477,
+      "step": 5870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95182681947699e-05,
+      "loss": 0.9356,
+      "step": 5875
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95174350110075e-05,
+      "loss": 0.9165,
+      "step": 5880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951660111084095e-05,
+      "loss": 0.9424,
+      "step": 5885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951576649428231e-05,
+      "loss": 0.9325,
+      "step": 5890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951493116134365e-05,
+      "loss": 0.9383,
+      "step": 5895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951409511203707e-05,
+      "loss": 0.9191,
+      "step": 5900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951325834637464e-05,
+      "loss": 0.9369,
+      "step": 5905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95124208643685e-05,
+      "loss": 0.9138,
+      "step": 5910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951158266603073e-05,
+      "loss": 0.9141,
+      "step": 5915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951074375137348e-05,
+      "loss": 0.939,
+      "step": 5920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950990412040888e-05,
+      "loss": 0.9267,
+      "step": 5925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950906377314909e-05,
+      "loss": 0.9165,
+      "step": 5930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950822270960623e-05,
+      "loss": 0.9244,
+      "step": 5935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950738092979253e-05,
+      "loss": 0.9393,
+      "step": 5940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950653843372011e-05,
+      "loss": 0.9403,
+      "step": 5945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950569522140119e-05,
+      "loss": 0.9453,
+      "step": 5950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950485129284796e-05,
+      "loss": 0.956,
+      "step": 5955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950400664807262e-05,
+      "loss": 0.9113,
+      "step": 5960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950316128708742e-05,
+      "loss": 0.9242,
+      "step": 5965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950231520990456e-05,
+      "loss": 0.9326,
+      "step": 5970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950146841653632e-05,
+      "loss": 0.9406,
+      "step": 5975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950062090699489e-05,
+      "loss": 0.9147,
+      "step": 5980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949977268129258e-05,
+      "loss": 0.9357,
+      "step": 5985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949892373944166e-05,
+      "loss": 0.9194,
+      "step": 5990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949807408145438e-05,
+      "loss": 0.9144,
+      "step": 5995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949722370734307e-05,
+      "loss": 0.946,
+      "step": 6000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 130693,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.6825629945102336e+19,
+  "trial_name": null,
+  "trial_params": null
+}