diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,36214 @@
+{
+  "best_metric": 0.5343035343035343,
+  "best_model_checkpoint": "./final/question/text2sql-t5-base-schema-generator/checkpoint-24000",
+  "epoch": 102.56410256410257,
+  "global_step": 24000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.337783711615487e-08,
+      "loss": 2.3603,
+      "step": 1
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.335113484646195e-07,
+      "loss": 2.6238,
+      "step": 4
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.67022696929239e-07,
+      "loss": 2.5591,
+      "step": 8
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.005340453938585e-07,
+      "loss": 2.563,
+      "step": 12
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.34045393858478e-07,
+      "loss": 2.7499,
+      "step": 16
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.675567423230975e-07,
+      "loss": 2.4741,
+      "step": 20
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.01068090787717e-07,
+      "loss": 2.3916,
+      "step": 24
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.345794392523364e-07,
+      "loss": 2.568,
+      "step": 28
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.068090787716956e-06,
+      "loss": 2.5685,
+      "step": 32
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.2016021361815755e-06,
+      "loss": 2.4451,
+      "step": 36
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.335113484646195e-06,
+      "loss": 2.6357,
+      "step": 40
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4686248331108143e-06,
+      "loss": 2.4322,
+      "step": 44
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.602136181575434e-06,
+      "loss": 2.4776,
+      "step": 48
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7356475300400534e-06,
+      "loss": 2.1708,
+      "step": 52
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8691588785046728e-06,
+      "loss": 2.3881,
+      "step": 56
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.0026702269692925e-06,
+      "loss": 2.181,
+      "step": 60
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.136181575433912e-06,
+      "loss": 2.2601,
+      "step": 64
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.2696929238985316e-06,
+      "loss": 2.234,
+      "step": 68
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.403204272363151e-06,
+      "loss": 2.3423,
+      "step": 72
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.5367156208277703e-06,
+      "loss": 2.2052,
+      "step": 76
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.67022696929239e-06,
+      "loss": 2.3827,
+      "step": 80
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.8037383177570094e-06,
+      "loss": 1.9671,
+      "step": 84
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9372496662216287e-06,
+      "loss": 1.9526,
+      "step": 88
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0707610146862484e-06,
+      "loss": 1.9553,
+      "step": 92
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.204272363150868e-06,
+      "loss": 2.0414,
+      "step": 96
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.337783711615487e-06,
+      "loss": 2.3561,
+      "step": 100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.471295060080107e-06,
+      "loss": 1.9045,
+      "step": 104
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6048064085447266e-06,
+      "loss": 1.841,
+      "step": 108
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7383177570093455e-06,
+      "loss": 1.7683,
+      "step": 112
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.871829105473966e-06,
+      "loss": 1.8482,
+      "step": 116
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.005340453938585e-06,
+      "loss": 1.7921,
+      "step": 120
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.138851802403204e-06,
+      "loss": 1.7806,
+      "step": 124
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.272363150867824e-06,
+      "loss": 1.6736,
+      "step": 128
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.405874499332444e-06,
+      "loss": 1.8067,
+      "step": 132
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.539385847797063e-06,
+      "loss": 1.7831,
+      "step": 136
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.6728971962616825e-06,
+      "loss": 1.7054,
+      "step": 140
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.806408544726302e-06,
+      "loss": 1.63,
+      "step": 144
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.939919893190922e-06,
+      "loss": 1.5281,
+      "step": 148
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.0734312416555405e-06,
+      "loss": 1.5909,
+      "step": 152
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.206942590120161e-06,
+      "loss": 1.6147,
+      "step": 156
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.34045393858478e-06,
+      "loss": 1.4966,
+      "step": 160
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.473965287049399e-06,
+      "loss": 1.3448,
+      "step": 164
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.607476635514019e-06,
+      "loss": 1.449,
+      "step": 168
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.740987983978639e-06,
+      "loss": 1.388,
+      "step": 172
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.874499332443257e-06,
+      "loss": 1.454,
+      "step": 176
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.0080106809078775e-06,
+      "loss": 1.3239,
+      "step": 180
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.141522029372497e-06,
+      "loss": 1.3004,
+      "step": 184
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 6.275033377837117e-06,
+      "loss": 1.3924,
+      "step": 188
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 6.408544726301736e-06,
+      "loss": 1.3636,
+      "step": 192
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.542056074766355e-06,
+      "loss": 1.2384,
+      "step": 196
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 6.675567423230974e-06,
+      "loss": 1.2257,
+      "step": 200
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 6.809078771695594e-06,
+      "loss": 1.2446,
+      "step": 204
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.942590120160214e-06,
+      "loss": 1.4018,
+      "step": 208
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.076101468624834e-06,
+      "loss": 1.2668,
+      "step": 212
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.209612817089453e-06,
+      "loss": 1.2525,
+      "step": 216
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.343124165554073e-06,
+      "loss": 1.1977,
+      "step": 220
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.476635514018691e-06,
+      "loss": 1.1998,
+      "step": 224
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.610146862483311e-06,
+      "loss": 1.1087,
+      "step": 228
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.743658210947931e-06,
+      "loss": 1.0731,
+      "step": 232
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.87716955941255e-06,
+      "loss": 1.0327,
+      "step": 236
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 8.01068090787717e-06,
+      "loss": 1.0915,
+      "step": 240
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 8.14419225634179e-06,
+      "loss": 1.0304,
+      "step": 244
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.277703604806409e-06,
+      "loss": 1.0082,
+      "step": 248
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.411214953271028e-06,
+      "loss": 1.0408,
+      "step": 252
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.544726301735647e-06,
+      "loss": 0.983,
+      "step": 256
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.678237650200267e-06,
+      "loss": 0.9362,
+      "step": 260
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.811748998664888e-06,
+      "loss": 0.9181,
+      "step": 264
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.945260347129507e-06,
+      "loss": 0.8616,
+      "step": 268
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.078771695594126e-06,
+      "loss": 0.8992,
+      "step": 272
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.212283044058744e-06,
+      "loss": 0.8689,
+      "step": 276
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.345794392523365e-06,
+      "loss": 0.8462,
+      "step": 280
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.479305740987984e-06,
+      "loss": 0.9342,
+      "step": 284
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.612817089452604e-06,
+      "loss": 0.8598,
+      "step": 288
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.746328437917223e-06,
+      "loss": 0.8383,
+      "step": 292
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.879839786381844e-06,
+      "loss": 0.7935,
+      "step": 296
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0013351134846462e-05,
+      "loss": 0.7795,
+      "step": 300
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.0146862483311081e-05,
+      "loss": 0.8016,
+      "step": 304
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.02803738317757e-05,
+      "loss": 0.743,
+      "step": 308
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.0413885180240321e-05,
+      "loss": 0.7807,
+      "step": 312
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.054739652870494e-05,
+      "loss": 0.7376,
+      "step": 316
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.068090787716956e-05,
+      "loss": 0.7415,
+      "step": 320
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.081441922563418e-05,
+      "loss": 0.7283,
+      "step": 324
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.0947930574098799e-05,
+      "loss": 0.6967,
+      "step": 328
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1081441922563418e-05,
+      "loss": 0.6988,
+      "step": 332
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1214953271028037e-05,
+      "loss": 0.7329,
+      "step": 336
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1348464619492657e-05,
+      "loss": 0.6674,
+      "step": 340
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.1481975967957278e-05,
+      "loss": 0.6287,
+      "step": 344
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1615487316421897e-05,
+      "loss": 0.7225,
+      "step": 348
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.1748998664886515e-05,
+      "loss": 0.6345,
+      "step": 352
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.1882510013351136e-05,
+      "loss": 0.6217,
+      "step": 356
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2016021361815755e-05,
+      "loss": 0.6306,
+      "step": 360
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2149532710280374e-05,
+      "loss": 0.5966,
+      "step": 364
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.2283044058744994e-05,
+      "loss": 0.6166,
+      "step": 368
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.2416555407209613e-05,
+      "loss": 0.588,
+      "step": 372
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.2550066755674234e-05,
+      "loss": 0.6105,
+      "step": 376
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.2683578104138852e-05,
+      "loss": 0.5674,
+      "step": 380
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.2817089452603473e-05,
+      "loss": 0.5682,
+      "step": 384
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.295060080106809e-05,
+      "loss": 0.5839,
+      "step": 388
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.308411214953271e-05,
+      "loss": 0.5281,
+      "step": 392
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.321762349799733e-05,
+      "loss": 0.5438,
+      "step": 396
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.3351134846461948e-05,
+      "loss": 0.544,
+      "step": 400
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.348464619492657e-05,
+      "loss": 0.5287,
+      "step": 404
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.3618157543391189e-05,
+      "loss": 0.5177,
+      "step": 408
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.375166889185581e-05,
+      "loss": 0.5275,
+      "step": 412
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.3885180240320427e-05,
+      "loss": 0.5346,
+      "step": 416
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4018691588785047e-05,
+      "loss": 0.5277,
+      "step": 420
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.4152202937249668e-05,
+      "loss": 0.5483,
+      "step": 424
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.5117,
+      "step": 428
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.4419225634178906e-05,
+      "loss": 0.4839,
+      "step": 432
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.4552736982643526e-05,
+      "loss": 0.4601,
+      "step": 436
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4686248331108147e-05,
+      "loss": 0.4971,
+      "step": 440
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4819759679572764e-05,
+      "loss": 0.4743,
+      "step": 444
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4953271028037382e-05,
+      "loss": 0.4872,
+      "step": 448
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.5086782376502003e-05,
+      "loss": 0.4172,
+      "step": 452
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.5220293724966622e-05,
+      "loss": 0.4379,
+      "step": 456
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.5353805073431243e-05,
+      "loss": 0.4265,
+      "step": 460
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.5487316421895863e-05,
+      "loss": 0.4154,
+      "step": 464
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.5620827770360482e-05,
+      "loss": 0.4336,
+      "step": 468
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.57543391188251e-05,
+      "loss": 0.4155,
+      "step": 472
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.588785046728972e-05,
+      "loss": 0.469,
+      "step": 476
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.602136181575434e-05,
+      "loss": 0.443,
+      "step": 480
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.615487316421896e-05,
+      "loss": 0.4061,
+      "step": 484
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.628838451268358e-05,
+      "loss": 0.417,
+      "step": 488
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.6421895861148198e-05,
+      "loss": 0.3833,
+      "step": 492
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6555407209612817e-05,
+      "loss": 0.4274,
+      "step": 496
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.6688918558077437e-05,
+      "loss": 0.3719,
+      "step": 500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.6822429906542056e-05,
+      "loss": 0.4054,
+      "step": 504
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.6955941255006675e-05,
+      "loss": 0.3914,
+      "step": 508
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.7089452603471295e-05,
+      "loss": 0.4241,
+      "step": 512
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7222963951935918e-05,
+      "loss": 0.3763,
+      "step": 516
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.7356475300400533e-05,
+      "loss": 0.3755,
+      "step": 520
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.7489986648865153e-05,
+      "loss": 0.3663,
+      "step": 524
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.7623497997329775e-05,
+      "loss": 0.3577,
+      "step": 528
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.775700934579439e-05,
+      "loss": 0.3447,
+      "step": 532
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.7890520694259014e-05,
+      "loss": 0.3899,
+      "step": 536
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.8024032042723633e-05,
+      "loss": 0.3441,
+      "step": 540
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.8157543391188253e-05,
+      "loss": 0.3935,
+      "step": 544
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.8291054739652872e-05,
+      "loss": 0.3579,
+      "step": 548
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.8424566088117488e-05,
+      "loss": 0.3555,
+      "step": 552
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.855807743658211e-05,
+      "loss": 0.3561,
+      "step": 556
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.869158878504673e-05,
+      "loss": 0.3363,
+      "step": 560
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.882510013351135e-05,
+      "loss": 0.358,
+      "step": 564
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.895861148197597e-05,
+      "loss": 0.3316,
+      "step": 568
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.9092122830440588e-05,
+      "loss": 0.36,
+      "step": 572
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.9225634178905207e-05,
+      "loss": 0.3553,
+      "step": 576
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.9359145527369827e-05,
+      "loss": 0.3348,
+      "step": 580
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.9492656875834446e-05,
+      "loss": 0.3186,
+      "step": 584
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.9626168224299065e-05,
+      "loss": 0.2991,
+      "step": 588
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.9759679572763688e-05,
+      "loss": 0.3197,
+      "step": 592
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.9893190921228304e-05,
+      "loss": 0.3489,
+      "step": 596
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.0026702269692923e-05,
+      "loss": 0.3344,
+      "step": 600
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.0160213618157546e-05,
+      "loss": 0.3118,
+      "step": 604
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.0293724966622162e-05,
+      "loss": 0.2895,
+      "step": 608
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.0427236315086785e-05,
+      "loss": 0.3054,
+      "step": 612
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.05607476635514e-05,
+      "loss": 0.2828,
+      "step": 616
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.0694259012016024e-05,
+      "loss": 0.305,
+      "step": 620
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.0827770360480643e-05,
+      "loss": 0.3166,
+      "step": 624
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.096128170894526e-05,
+      "loss": 0.3252,
+      "step": 628
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.109479305740988e-05,
+      "loss": 0.3304,
+      "step": 632
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.12283044058745e-05,
+      "loss": 0.2743,
+      "step": 636
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.136181575433912e-05,
+      "loss": 0.2986,
+      "step": 640
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.149532710280374e-05,
+      "loss": 0.293,
+      "step": 644
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.162883845126836e-05,
+      "loss": 0.2777,
+      "step": 648
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.1762349799732978e-05,
+      "loss": 0.2572,
+      "step": 652
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.1895861148197598e-05,
+      "loss": 0.2667,
+      "step": 656
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.2029372496662217e-05,
+      "loss": 0.2817,
+      "step": 660
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.2162883845126836e-05,
+      "loss": 0.2712,
+      "step": 664
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.229639519359146e-05,
+      "loss": 0.2583,
+      "step": 668
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.2429906542056075e-05,
+      "loss": 0.2684,
+      "step": 672
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.2563417890520698e-05,
+      "loss": 0.2811,
+      "step": 676
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.2696929238985313e-05,
+      "loss": 0.2657,
+      "step": 680
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.2830440587449933e-05,
+      "loss": 0.2856,
+      "step": 684
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.2963951935914556e-05,
+      "loss": 0.2647,
+      "step": 688
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.309746328437917e-05,
+      "loss": 0.2373,
+      "step": 692
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.3230974632843794e-05,
+      "loss": 0.2789,
+      "step": 696
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.3364485981308414e-05,
+      "loss": 0.2637,
+      "step": 700
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.349799732977303e-05,
+      "loss": 0.2691,
+      "step": 704
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.3631508678237652e-05,
+      "loss": 0.2516,
+      "step": 708
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.376502002670227e-05,
+      "loss": 0.2382,
+      "step": 712
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.389853137516689e-05,
+      "loss": 0.2398,
+      "step": 716
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.403204272363151e-05,
+      "loss": 0.2353,
+      "step": 720
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.416555407209613e-05,
+      "loss": 0.2257,
+      "step": 724
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.429906542056075e-05,
+      "loss": 0.2939,
+      "step": 728
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.4432576769025368e-05,
+      "loss": 0.2483,
+      "step": 732
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.4566088117489988e-05,
+      "loss": 0.2338,
+      "step": 736
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.4699599465954607e-05,
+      "loss": 0.2811,
+      "step": 740
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.4833110814419226e-05,
+      "loss": 0.2147,
+      "step": 744
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.4966622162883846e-05,
+      "loss": 0.1995,
+      "step": 748
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 2.5100133511348468e-05,
+      "loss": 0.231,
+      "step": 752
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.5233644859813084e-05,
+      "loss": 0.2072,
+      "step": 756
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.5367156208277704e-05,
+      "loss": 0.2304,
+      "step": 760
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.5500667556742326e-05,
+      "loss": 0.2172,
+      "step": 764
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.5634178905206946e-05,
+      "loss": 0.2318,
+      "step": 768
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.576769025367156e-05,
+      "loss": 0.2285,
+      "step": 772
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.590120160213618e-05,
+      "loss": 0.217,
+      "step": 776
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.6034712950600804e-05,
+      "loss": 0.1973,
+      "step": 780
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.616822429906542e-05,
+      "loss": 0.2131,
+      "step": 784
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.630173564753004e-05,
+      "loss": 0.2343,
+      "step": 788
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.643524699599466e-05,
+      "loss": 0.2307,
+      "step": 792
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.656875834445928e-05,
+      "loss": 0.224,
+      "step": 796
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.6702269692923897e-05,
+      "loss": 0.2008,
+      "step": 800
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.683578104138852e-05,
+      "loss": 0.2024,
+      "step": 804
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.696929238985314e-05,
+      "loss": 0.2134,
+      "step": 808
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.7102803738317755e-05,
+      "loss": 0.1942,
+      "step": 812
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.7236315086782378e-05,
+      "loss": 0.2103,
+      "step": 816
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.7369826435246997e-05,
+      "loss": 0.212,
+      "step": 820
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.750333778371162e-05,
+      "loss": 0.2368,
+      "step": 824
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.7636849132176236e-05,
+      "loss": 0.1905,
+      "step": 828
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.7770360480640855e-05,
+      "loss": 0.1892,
+      "step": 832
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.7903871829105478e-05,
+      "loss": 0.2069,
+      "step": 836
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.8037383177570094e-05,
+      "loss": 0.1957,
+      "step": 840
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.8170894526034713e-05,
+      "loss": 0.2093,
+      "step": 844
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.8304405874499336e-05,
+      "loss": 0.2003,
+      "step": 848
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.8437917222963955e-05,
+      "loss": 0.2081,
+      "step": 852
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.1746,
+      "step": 856
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.8704939919893194e-05,
+      "loss": 0.1738,
+      "step": 860
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.8838451268357813e-05,
+      "loss": 0.1868,
+      "step": 864
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.897196261682243e-05,
+      "loss": 0.1829,
+      "step": 868
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.910547396528705e-05,
+      "loss": 0.1794,
+      "step": 872
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.923898531375167e-05,
+      "loss": 0.1696,
+      "step": 876
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.9372496662216294e-05,
+      "loss": 0.1907,
+      "step": 880
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.9506008010680906e-05,
+      "loss": 0.1813,
+      "step": 884
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.963951935914553e-05,
+      "loss": 0.1519,
+      "step": 888
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.977303070761015e-05,
+      "loss": 0.1903,
+      "step": 892
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.9906542056074764e-05,
+      "loss": 0.1712,
+      "step": 896
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.0040053404539387e-05,
+      "loss": 0.1914,
+      "step": 900
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.0173564753004006e-05,
+      "loss": 0.164,
+      "step": 904
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.030707610146863e-05,
+      "loss": 0.1671,
+      "step": 908
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.0440587449933245e-05,
+      "loss": 0.1688,
+      "step": 912
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 3.0574098798397864e-05,
+      "loss": 0.1588,
+      "step": 916
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.070761014686249e-05,
+      "loss": 0.1685,
+      "step": 920
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 3.08411214953271e-05,
+      "loss": 0.1704,
+      "step": 924
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.0974632843791726e-05,
+      "loss": 0.1467,
+      "step": 928
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.110814419225634e-05,
+      "loss": 0.1631,
+      "step": 932
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.1241655540720964e-05,
+      "loss": 0.1584,
+      "step": 936
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 3.137516688918558e-05,
+      "loss": 0.1595,
+      "step": 940
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 3.15086782376502e-05,
+      "loss": 0.1576,
+      "step": 944
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 3.1642189586114826e-05,
+      "loss": 0.173,
+      "step": 948
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 3.177570093457944e-05,
+      "loss": 0.1775,
+      "step": 952
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.190921228304406e-05,
+      "loss": 0.1542,
+      "step": 956
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 3.204272363150868e-05,
+      "loss": 0.129,
+      "step": 960
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.2176234979973296e-05,
+      "loss": 0.1671,
+      "step": 964
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 3.230974632843792e-05,
+      "loss": 0.1541,
+      "step": 968
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.244325767690254e-05,
+      "loss": 0.1524,
+      "step": 972
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 3.257676902536716e-05,
+      "loss": 0.1559,
+      "step": 976
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.2710280373831774e-05,
+      "loss": 0.1447,
+      "step": 980
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 3.2843791722296396e-05,
+      "loss": 0.1596,
+      "step": 984
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.297730307076102e-05,
+      "loss": 0.1287,
+      "step": 988
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.3110814419225635e-05,
+      "loss": 0.1448,
+      "step": 992
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.324432576769025e-05,
+      "loss": 0.1479,
+      "step": 996
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.3377837116154874e-05,
+      "loss": 0.1587,
+      "step": 1000
+    },
+    {
+      "epoch": 8.55,
+      "eval_exact_match": 0.35654885654885654,
+      "eval_loss": 0.4734349548816681,
+      "eval_runtime": 95.1537,
+      "eval_samples_per_second": 10.11,
+      "step": 1000
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.3511348464619496e-05,
+      "loss": 0.1445,
+      "step": 1004
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.364485981308411e-05,
+      "loss": 0.1414,
+      "step": 1008
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 3.3778371161548735e-05,
+      "loss": 0.153,
+      "step": 1012
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 3.391188251001335e-05,
+      "loss": 0.1389,
+      "step": 1016
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 3.404539385847797e-05,
+      "loss": 0.1528,
+      "step": 1020
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.417890520694259e-05,
+      "loss": 0.1281,
+      "step": 1024
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 3.431241655540721e-05,
+      "loss": 0.1358,
+      "step": 1028
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.4445927903871835e-05,
+      "loss": 0.1392,
+      "step": 1032
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 3.457943925233645e-05,
+      "loss": 0.115,
+      "step": 1036
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.471295060080107e-05,
+      "loss": 0.1376,
+      "step": 1040
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 3.484646194926569e-05,
+      "loss": 0.1347,
+      "step": 1044
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.4979973297730306e-05,
+      "loss": 0.1198,
+      "step": 1048
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 3.511348464619493e-05,
+      "loss": 0.1278,
+      "step": 1052
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 3.524699599465955e-05,
+      "loss": 0.141,
+      "step": 1056
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.538050734312417e-05,
+      "loss": 0.1191,
+      "step": 1060
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.551401869158878e-05,
+      "loss": 0.1298,
+      "step": 1064
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.5647530040053406e-05,
+      "loss": 0.1373,
+      "step": 1068
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.578104138851803e-05,
+      "loss": 0.1241,
+      "step": 1072
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.5914552736982644e-05,
+      "loss": 0.126,
+      "step": 1076
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.604806408544727e-05,
+      "loss": 0.1065,
+      "step": 1080
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.618157543391188e-05,
+      "loss": 0.1143,
+      "step": 1084
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.6315086782376506e-05,
+      "loss": 0.1323,
+      "step": 1088
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.644859813084112e-05,
+      "loss": 0.1351,
+      "step": 1092
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.6582109479305744e-05,
+      "loss": 0.1231,
+      "step": 1096
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.671562082777037e-05,
+      "loss": 0.123,
+      "step": 1100
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 3.6849132176234976e-05,
+      "loss": 0.1214,
+      "step": 1104
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 3.69826435246996e-05,
+      "loss": 0.115,
+      "step": 1108
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 3.711615487316422e-05,
+      "loss": 0.1416,
+      "step": 1112
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 3.7249666221628844e-05,
+      "loss": 0.1126,
+      "step": 1116
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 3.738317757009346e-05,
+      "loss": 0.1179,
+      "step": 1120
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 3.7516688918558076e-05,
+      "loss": 0.1122,
+      "step": 1124
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 3.76502002670227e-05,
+      "loss": 0.1116,
+      "step": 1128
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 3.7783711615487315e-05,
+      "loss": 0.1148,
+      "step": 1132
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 3.791722296395194e-05,
+      "loss": 0.1102,
+      "step": 1136
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.805073431241656e-05,
+      "loss": 0.1213,
+      "step": 1140
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 3.8184245660881176e-05,
+      "loss": 0.1105,
+      "step": 1144
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 3.831775700934579e-05,
+      "loss": 0.1012,
+      "step": 1148
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 3.8451268357810415e-05,
+      "loss": 0.128,
+      "step": 1152
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 3.858477970627504e-05,
+      "loss": 0.1074,
+      "step": 1156
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.8718291054739654e-05,
+      "loss": 0.1104,
+      "step": 1160
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 3.8851802403204276e-05,
+      "loss": 0.1231,
+      "step": 1164
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.898531375166889e-05,
+      "loss": 0.1078,
+      "step": 1168
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.9118825100133515e-05,
+      "loss": 0.1061,
+      "step": 1172
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.925233644859813e-05,
+      "loss": 0.1007,
+      "step": 1176
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.9385847797062754e-05,
+      "loss": 0.0987,
+      "step": 1180
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.9519359145527376e-05,
+      "loss": 0.121,
+      "step": 1184
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.965287049399199e-05,
+      "loss": 0.1,
+      "step": 1188
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.978638184245661e-05,
+      "loss": 0.1001,
+      "step": 1192
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.991989319092123e-05,
+      "loss": 0.1033,
+      "step": 1196
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.005340453938585e-05,
+      "loss": 0.0859,
+      "step": 1200
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.018691588785047e-05,
+      "loss": 0.1105,
+      "step": 1204
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.032042723631509e-05,
+      "loss": 0.1025,
+      "step": 1208
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.045393858477971e-05,
+      "loss": 0.0994,
+      "step": 1212
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.0587449933244324e-05,
+      "loss": 0.1133,
+      "step": 1216
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.072096128170895e-05,
+      "loss": 0.0973,
+      "step": 1220
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.085447263017357e-05,
+      "loss": 0.0917,
+      "step": 1224
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.0987983978638186e-05,
+      "loss": 0.0861,
+      "step": 1228
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.11214953271028e-05,
+      "loss": 0.0975,
+      "step": 1232
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.1255006675567424e-05,
+      "loss": 0.1049,
+      "step": 1236
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.138851802403205e-05,
+      "loss": 0.1036,
+      "step": 1240
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.152202937249666e-05,
+      "loss": 0.0842,
+      "step": 1244
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.1655540720961286e-05,
+      "loss": 0.1029,
+      "step": 1248
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.17890520694259e-05,
+      "loss": 0.0906,
+      "step": 1252
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.192256341789052e-05,
+      "loss": 0.0986,
+      "step": 1256
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.205607476635514e-05,
+      "loss": 0.0958,
+      "step": 1260
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.218958611481976e-05,
+      "loss": 0.0898,
+      "step": 1264
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.2323097463284386e-05,
+      "loss": 0.0979,
+      "step": 1268
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.2456608811749e-05,
+      "loss": 0.0911,
+      "step": 1272
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.259012016021362e-05,
+      "loss": 0.0909,
+      "step": 1276
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.272363150867824e-05,
+      "loss": 0.0909,
+      "step": 1280
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.0849,
+      "step": 1284
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.299065420560748e-05,
+      "loss": 0.095,
+      "step": 1288
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.31241655540721e-05,
+      "loss": 0.0726,
+      "step": 1292
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.325767690253672e-05,
+      "loss": 0.0996,
+      "step": 1296
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.3391188251001334e-05,
+      "loss": 0.0783,
+      "step": 1300
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.3524699599465956e-05,
+      "loss": 0.082,
+      "step": 1304
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.365821094793058e-05,
+      "loss": 0.0966,
+      "step": 1308
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.3791722296395195e-05,
+      "loss": 0.0815,
+      "step": 1312
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.392523364485982e-05,
+      "loss": 0.0783,
+      "step": 1316
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.4058744993324434e-05,
+      "loss": 0.0772,
+      "step": 1320
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.4192256341789056e-05,
+      "loss": 0.0763,
+      "step": 1324
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.432576769025367e-05,
+      "loss": 0.0859,
+      "step": 1328
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.4459279038718295e-05,
+      "loss": 0.0764,
+      "step": 1332
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.459279038718292e-05,
+      "loss": 0.0784,
+      "step": 1336
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.472630173564753e-05,
+      "loss": 0.0779,
+      "step": 1340
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.485981308411215e-05,
+      "loss": 0.0706,
+      "step": 1344
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.499332443257677e-05,
+      "loss": 0.0837,
+      "step": 1348
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.5126835781041395e-05,
+      "loss": 0.0881,
+      "step": 1352
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.526034712950601e-05,
+      "loss": 0.0766,
+      "step": 1356
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.539385847797063e-05,
+      "loss": 0.0777,
+      "step": 1360
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.552736982643525e-05,
+      "loss": 0.0893,
+      "step": 1364
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.5660881174899866e-05,
+      "loss": 0.0801,
+      "step": 1368
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.579439252336449e-05,
+      "loss": 0.0758,
+      "step": 1372
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.592790387182911e-05,
+      "loss": 0.0781,
+      "step": 1376
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.606141522029373e-05,
+      "loss": 0.0725,
+      "step": 1380
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.619492656875834e-05,
+      "loss": 0.0707,
+      "step": 1384
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.6328437917222966e-05,
+      "loss": 0.073,
+      "step": 1388
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.646194926568759e-05,
+      "loss": 0.0707,
+      "step": 1392
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.6595460614152204e-05,
+      "loss": 0.0692,
+      "step": 1396
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.672897196261683e-05,
+      "loss": 0.0793,
+      "step": 1400
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.686248331108144e-05,
+      "loss": 0.076,
+      "step": 1404
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.699599465954606e-05,
+      "loss": 0.073,
+      "step": 1408
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.712950600801068e-05,
+      "loss": 0.0662,
+      "step": 1412
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.7263017356475304e-05,
+      "loss": 0.0678,
+      "step": 1416
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.739652870493993e-05,
+      "loss": 0.0677,
+      "step": 1420
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.753004005340454e-05,
+      "loss": 0.0697,
+      "step": 1424
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.766355140186916e-05,
+      "loss": 0.0655,
+      "step": 1428
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.779706275033378e-05,
+      "loss": 0.0676,
+      "step": 1432
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.79305740987984e-05,
+      "loss": 0.0618,
+      "step": 1436
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.806408544726302e-05,
+      "loss": 0.0716,
+      "step": 1440
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.819759679572764e-05,
+      "loss": 0.0745,
+      "step": 1444
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.833110814419226e-05,
+      "loss": 0.0706,
+      "step": 1448
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.8464619492656875e-05,
+      "loss": 0.0709,
+      "step": 1452
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.85981308411215e-05,
+      "loss": 0.064,
+      "step": 1456
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.873164218958612e-05,
+      "loss": 0.0654,
+      "step": 1460
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.8865153538050736e-05,
+      "loss": 0.0687,
+      "step": 1464
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.899866488651535e-05,
+      "loss": 0.0764,
+      "step": 1468
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.9132176234979975e-05,
+      "loss": 0.0666,
+      "step": 1472
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.92656875834446e-05,
+      "loss": 0.069,
+      "step": 1476
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.9399198931909214e-05,
+      "loss": 0.0684,
+      "step": 1480
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.9532710280373836e-05,
+      "loss": 0.0606,
+      "step": 1484
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.966622162883845e-05,
+      "loss": 0.0658,
+      "step": 1488
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.979973297730307e-05,
+      "loss": 0.0677,
+      "step": 1492
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.993324432576769e-05,
+      "loss": 0.0651,
+      "step": 1496
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 5.0066755674232314e-05,
+      "loss": 0.0679,
+      "step": 1500
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 5.0200267022696936e-05,
+      "loss": 0.0582,
+      "step": 1504
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 5.033377837116155e-05,
+      "loss": 0.047,
+      "step": 1508
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 5.046728971962617e-05,
+      "loss": 0.0541,
+      "step": 1512
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 5.0600801068090784e-05,
+      "loss": 0.0634,
+      "step": 1516
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 5.073431241655541e-05,
+      "loss": 0.061,
+      "step": 1520
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 5.086782376502003e-05,
+      "loss": 0.0724,
+      "step": 1524
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 5.100133511348465e-05,
+      "loss": 0.058,
+      "step": 1528
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 5.113484646194927e-05,
+      "loss": 0.0561,
+      "step": 1532
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 5.126835781041389e-05,
+      "loss": 0.058,
+      "step": 1536
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 5.14018691588785e-05,
+      "loss": 0.0563,
+      "step": 1540
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 5.153538050734312e-05,
+      "loss": 0.06,
+      "step": 1544
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 5.1668891855807746e-05,
+      "loss": 0.0575,
+      "step": 1548
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 5.180240320427236e-05,
+      "loss": 0.0602,
+      "step": 1552
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 5.1935914552736984e-05,
+      "loss": 0.0645,
+      "step": 1556
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 5.206942590120161e-05,
+      "loss": 0.061,
+      "step": 1560
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 5.220293724966623e-05,
+      "loss": 0.0522,
+      "step": 1564
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 5.233644859813084e-05,
+      "loss": 0.0489,
+      "step": 1568
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 5.246995994659546e-05,
+      "loss": 0.0543,
+      "step": 1572
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 5.260347129506008e-05,
+      "loss": 0.0581,
+      "step": 1576
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 5.27369826435247e-05,
+      "loss": 0.0564,
+      "step": 1580
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 5.287049399198932e-05,
+      "loss": 0.0569,
+      "step": 1584
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 5.3004005340453946e-05,
+      "loss": 0.0568,
+      "step": 1588
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 5.313751668891856e-05,
+      "loss": 0.0524,
+      "step": 1592
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 5.327102803738318e-05,
+      "loss": 0.0515,
+      "step": 1596
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 5.3404539385847794e-05,
+      "loss": 0.0517,
+      "step": 1600
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 5.3538050734312416e-05,
+      "loss": 0.062,
+      "step": 1604
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 5.367156208277704e-05,
+      "loss": 0.0501,
+      "step": 1608
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 5.380507343124166e-05,
+      "loss": 0.0555,
+      "step": 1612
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 5.393858477970628e-05,
+      "loss": 0.0587,
+      "step": 1616
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 5.40720961281709e-05,
+      "loss": 0.0568,
+      "step": 1620
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 5.420560747663551e-05,
+      "loss": 0.0502,
+      "step": 1624
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 5.433911882510013e-05,
+      "loss": 0.0512,
+      "step": 1628
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 5.4472630173564755e-05,
+      "loss": 0.0551,
+      "step": 1632
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 5.460614152202938e-05,
+      "loss": 0.0509,
+      "step": 1636
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 5.4739652870493994e-05,
+      "loss": 0.0587,
+      "step": 1640
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 5.4873164218958616e-05,
+      "loss": 0.0472,
+      "step": 1644
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 5.500667556742324e-05,
+      "loss": 0.0514,
+      "step": 1648
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 5.514018691588785e-05,
+      "loss": 0.0434,
+      "step": 1652
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 5.527369826435247e-05,
+      "loss": 0.0548,
+      "step": 1656
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 5.540720961281709e-05,
+      "loss": 0.0514,
+      "step": 1660
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 5.554072096128171e-05,
+      "loss": 0.0511,
+      "step": 1664
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 5.567423230974633e-05,
+      "loss": 0.0581,
+      "step": 1668
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 5.5807743658210955e-05,
+      "loss": 0.0444,
+      "step": 1672
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 5.594125500667558e-05,
+      "loss": 0.0597,
+      "step": 1676
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 5.607476635514019e-05,
+      "loss": 0.0542,
+      "step": 1680
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 5.62082777036048e-05,
+      "loss": 0.0452,
+      "step": 1684
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 5.6341789052069426e-05,
+      "loss": 0.0511,
+      "step": 1688
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 5.647530040053405e-05,
+      "loss": 0.0468,
+      "step": 1692
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 5.660881174899867e-05,
+      "loss": 0.0488,
+      "step": 1696
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 5.674232309746329e-05,
+      "loss": 0.0429,
+      "step": 1700
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 5.687583444592791e-05,
+      "loss": 0.0441,
+      "step": 1704
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 5.700934579439252e-05,
+      "loss": 0.0515,
+      "step": 1708
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 0.0462,
+      "step": 1712
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 5.7276368491321764e-05,
+      "loss": 0.0416,
+      "step": 1716
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 5.740987983978639e-05,
+      "loss": 0.0509,
+      "step": 1720
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 5.7543391188251e-05,
+      "loss": 0.0376,
+      "step": 1724
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 5.7676902536715626e-05,
+      "loss": 0.0408,
+      "step": 1728
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 5.781041388518025e-05,
+      "loss": 0.047,
+      "step": 1732
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 5.794392523364486e-05,
+      "loss": 0.05,
+      "step": 1736
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 5.807743658210948e-05,
+      "loss": 0.0406,
+      "step": 1740
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 5.82109479305741e-05,
+      "loss": 0.0408,
+      "step": 1744
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 5.834445927903872e-05,
+      "loss": 0.0472,
+      "step": 1748
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 5.847797062750334e-05,
+      "loss": 0.0468,
+      "step": 1752
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 5.8611481975967965e-05,
+      "loss": 0.0552,
+      "step": 1756
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 5.874499332443259e-05,
+      "loss": 0.0395,
+      "step": 1760
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 5.8878504672897196e-05,
+      "loss": 0.0378,
+      "step": 1764
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 5.901201602136181e-05,
+      "loss": 0.0389,
+      "step": 1768
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 5.9145527369826435e-05,
+      "loss": 0.0375,
+      "step": 1772
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 5.927903871829106e-05,
+      "loss": 0.0478,
+      "step": 1776
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 5.941255006675568e-05,
+      "loss": 0.0348,
+      "step": 1780
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 5.95460614152203e-05,
+      "loss": 0.04,
+      "step": 1784
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 5.967957276368492e-05,
+      "loss": 0.0459,
+      "step": 1788
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 5.981308411214953e-05,
+      "loss": 0.0428,
+      "step": 1792
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 5.994659546061415e-05,
+      "loss": 0.0408,
+      "step": 1796
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 6.0080106809078774e-05,
+      "loss": 0.0428,
+      "step": 1800
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 6.0213618157543397e-05,
+      "loss": 0.0366,
+      "step": 1804
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 6.034712950600801e-05,
+      "loss": 0.0474,
+      "step": 1808
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 6.0480640854472635e-05,
+      "loss": 0.0432,
+      "step": 1812
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 6.061415220293726e-05,
+      "loss": 0.0309,
+      "step": 1816
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 6.074766355140187e-05,
+      "loss": 0.0418,
+      "step": 1820
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 6.088117489986649e-05,
+      "loss": 0.0387,
+      "step": 1824
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 6.101468624833111e-05,
+      "loss": 0.0433,
+      "step": 1828
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 6.114819759679573e-05,
+      "loss": 0.0363,
+      "step": 1832
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 6.128170894526035e-05,
+      "loss": 0.0438,
+      "step": 1836
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 6.141522029372497e-05,
+      "loss": 0.0396,
+      "step": 1840
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 6.15487316421896e-05,
+      "loss": 0.0319,
+      "step": 1844
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 6.16822429906542e-05,
+      "loss": 0.0409,
+      "step": 1848
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 6.181575433911883e-05,
+      "loss": 0.0412,
+      "step": 1852
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 6.194926568758345e-05,
+      "loss": 0.0395,
+      "step": 1856
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 6.208277703604807e-05,
+      "loss": 0.0372,
+      "step": 1860
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 6.221628838451268e-05,
+      "loss": 0.0427,
+      "step": 1864
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 6.23497997329773e-05,
+      "loss": 0.0407,
+      "step": 1868
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 6.248331108144193e-05,
+      "loss": 0.0417,
+      "step": 1872
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 6.261682242990654e-05,
+      "loss": 0.0361,
+      "step": 1876
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 6.275033377837116e-05,
+      "loss": 0.0367,
+      "step": 1880
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 6.288384512683578e-05,
+      "loss": 0.0411,
+      "step": 1884
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 6.30173564753004e-05,
+      "loss": 0.0433,
+      "step": 1888
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 6.315086782376503e-05,
+      "loss": 0.0355,
+      "step": 1892
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 6.328437917222965e-05,
+      "loss": 0.0365,
+      "step": 1896
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 6.341789052069427e-05,
+      "loss": 0.0378,
+      "step": 1900
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 6.355140186915888e-05,
+      "loss": 0.0369,
+      "step": 1904
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 6.368491321762349e-05,
+      "loss": 0.0341,
+      "step": 1908
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 6.381842456608812e-05,
+      "loss": 0.0329,
+      "step": 1912
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 6.395193591455274e-05,
+      "loss": 0.0377,
+      "step": 1916
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 6.408544726301736e-05,
+      "loss": 0.0372,
+      "step": 1920
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 6.421895861148198e-05,
+      "loss": 0.0326,
+      "step": 1924
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 6.435246995994659e-05,
+      "loss": 0.0353,
+      "step": 1928
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 6.448598130841122e-05,
+      "loss": 0.0338,
+      "step": 1932
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 6.461949265687584e-05,
+      "loss": 0.0377,
+      "step": 1936
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 6.475300400534046e-05,
+      "loss": 0.039,
+      "step": 1940
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 6.488651535380508e-05,
+      "loss": 0.0354,
+      "step": 1944
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 6.502002670226969e-05,
+      "loss": 0.0377,
+      "step": 1948
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 6.515353805073432e-05,
+      "loss": 0.0379,
+      "step": 1952
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 6.528704939919892e-05,
+      "loss": 0.034,
+      "step": 1956
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 6.542056074766355e-05,
+      "loss": 0.0309,
+      "step": 1960
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 6.555407209612817e-05,
+      "loss": 0.0367,
+      "step": 1964
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 6.568758344459279e-05,
+      "loss": 0.0303,
+      "step": 1968
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 6.582109479305742e-05,
+      "loss": 0.0269,
+      "step": 1972
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 6.595460614152204e-05,
+      "loss": 0.0352,
+      "step": 1976
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 6.608811748998666e-05,
+      "loss": 0.0372,
+      "step": 1980
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 6.622162883845127e-05,
+      "loss": 0.0355,
+      "step": 1984
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 6.635514018691589e-05,
+      "loss": 0.0359,
+      "step": 1988
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 6.64886515353805e-05,
+      "loss": 0.0377,
+      "step": 1992
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 6.662216288384512e-05,
+      "loss": 0.0301,
+      "step": 1996
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 6.675567423230975e-05,
+      "loss": 0.0362,
+      "step": 2000
+    },
+    {
+      "epoch": 17.09,
+      "eval_exact_match": 0.446985446985447,
+      "eval_loss": 0.5254557728767395,
+      "eval_runtime": 86.8762,
+      "eval_samples_per_second": 11.073,
+      "step": 2000
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 6.688918558077437e-05,
+      "loss": 0.0301,
+      "step": 2004
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 6.702269692923899e-05,
+      "loss": 0.0302,
+      "step": 2008
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 6.71562082777036e-05,
+      "loss": 0.0299,
+      "step": 2012
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 6.728971962616822e-05,
+      "loss": 0.0341,
+      "step": 2016
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 6.742323097463285e-05,
+      "loss": 0.0311,
+      "step": 2020
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 6.755674232309747e-05,
+      "loss": 0.0277,
+      "step": 2024
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 6.769025367156209e-05,
+      "loss": 0.0366,
+      "step": 2028
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 6.78237650200267e-05,
+      "loss": 0.0338,
+      "step": 2032
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 6.795727636849132e-05,
+      "loss": 0.0433,
+      "step": 2036
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 6.809078771695593e-05,
+      "loss": 0.0266,
+      "step": 2040
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 6.822429906542056e-05,
+      "loss": 0.0312,
+      "step": 2044
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 6.835781041388518e-05,
+      "loss": 0.0337,
+      "step": 2048
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 6.84913217623498e-05,
+      "loss": 0.036,
+      "step": 2052
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 6.862483311081442e-05,
+      "loss": 0.0397,
+      "step": 2056
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 6.875834445927905e-05,
+      "loss": 0.0328,
+      "step": 2060
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 6.889185580774367e-05,
+      "loss": 0.0318,
+      "step": 2064
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 6.902536715620828e-05,
+      "loss": 0.0326,
+      "step": 2068
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 6.91588785046729e-05,
+      "loss": 0.026,
+      "step": 2072
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 6.929238985313752e-05,
+      "loss": 0.0295,
+      "step": 2076
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 6.942590120160213e-05,
+      "loss": 0.0328,
+      "step": 2080
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 6.955941255006676e-05,
+      "loss": 0.0286,
+      "step": 2084
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 6.969292389853138e-05,
+      "loss": 0.0308,
+      "step": 2088
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 6.9826435246996e-05,
+      "loss": 0.0301,
+      "step": 2092
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 6.995994659546061e-05,
+      "loss": 0.0271,
+      "step": 2096
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 7.009345794392523e-05,
+      "loss": 0.0303,
+      "step": 2100
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 7.022696929238986e-05,
+      "loss": 0.0236,
+      "step": 2104
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 7.036048064085448e-05,
+      "loss": 0.0297,
+      "step": 2108
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 7.04939919893191e-05,
+      "loss": 0.0216,
+      "step": 2112
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 7.062750333778372e-05,
+      "loss": 0.026,
+      "step": 2116
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 7.076101468624833e-05,
+      "loss": 0.0299,
+      "step": 2120
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 7.089452603471294e-05,
+      "loss": 0.0303,
+      "step": 2124
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 7.102803738317757e-05,
+      "loss": 0.0256,
+      "step": 2128
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 7.116154873164219e-05,
+      "loss": 0.0304,
+      "step": 2132
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 7.129506008010681e-05,
+      "loss": 0.0241,
+      "step": 2136
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 7.142857142857143e-05,
+      "loss": 0.0294,
+      "step": 2140
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 7.156208277703606e-05,
+      "loss": 0.0279,
+      "step": 2144
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 7.169559412550068e-05,
+      "loss": 0.0332,
+      "step": 2148
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 7.182910547396529e-05,
+      "loss": 0.0315,
+      "step": 2152
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 7.196261682242991e-05,
+      "loss": 0.0215,
+      "step": 2156
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 7.209612817089453e-05,
+      "loss": 0.032,
+      "step": 2160
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 7.222963951935914e-05,
+      "loss": 0.0257,
+      "step": 2164
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 7.236315086782377e-05,
+      "loss": 0.0287,
+      "step": 2168
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 7.249666221628839e-05,
+      "loss": 0.0284,
+      "step": 2172
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 7.263017356475301e-05,
+      "loss": 0.0316,
+      "step": 2176
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 7.276368491321762e-05,
+      "loss": 0.0268,
+      "step": 2180
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 7.289719626168224e-05,
+      "loss": 0.0291,
+      "step": 2184
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 7.303070761014687e-05,
+      "loss": 0.0284,
+      "step": 2188
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 7.316421895861149e-05,
+      "loss": 0.0239,
+      "step": 2192
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 7.329773030707611e-05,
+      "loss": 0.0247,
+      "step": 2196
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 7.343124165554073e-05,
+      "loss": 0.0192,
+      "step": 2200
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 7.356475300400534e-05,
+      "loss": 0.0236,
+      "step": 2204
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 7.369826435246995e-05,
+      "loss": 0.0235,
+      "step": 2208
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 7.383177570093458e-05,
+      "loss": 0.0313,
+      "step": 2212
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 7.39652870493992e-05,
+      "loss": 0.0292,
+      "step": 2216
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 7.409879839786382e-05,
+      "loss": 0.0295,
+      "step": 2220
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 7.423230974632844e-05,
+      "loss": 0.0251,
+      "step": 2224
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 7.436582109479307e-05,
+      "loss": 0.0245,
+      "step": 2228
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 7.449933244325769e-05,
+      "loss": 0.0205,
+      "step": 2232
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 7.46328437917223e-05,
+      "loss": 0.0219,
+      "step": 2236
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 7.476635514018692e-05,
+      "loss": 0.0262,
+      "step": 2240
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 7.489986648865154e-05,
+      "loss": 0.0237,
+      "step": 2244
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 7.503337783711615e-05,
+      "loss": 0.024,
+      "step": 2248
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 7.516688918558078e-05,
+      "loss": 0.0244,
+      "step": 2252
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 7.53004005340454e-05,
+      "loss": 0.0257,
+      "step": 2256
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 7.543391188251002e-05,
+      "loss": 0.0238,
+      "step": 2260
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 7.556742323097463e-05,
+      "loss": 0.0306,
+      "step": 2264
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 7.570093457943925e-05,
+      "loss": 0.0274,
+      "step": 2268
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 7.583444592790388e-05,
+      "loss": 0.0236,
+      "step": 2272
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 7.59679572763685e-05,
+      "loss": 0.0282,
+      "step": 2276
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 7.610146862483312e-05,
+      "loss": 0.0274,
+      "step": 2280
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 7.623497997329774e-05,
+      "loss": 0.0228,
+      "step": 2284
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 7.636849132176235e-05,
+      "loss": 0.0242,
+      "step": 2288
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 7.650200267022698e-05,
+      "loss": 0.0209,
+      "step": 2292
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 7.663551401869158e-05,
+      "loss": 0.0224,
+      "step": 2296
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 7.676902536715621e-05,
+      "loss": 0.0221,
+      "step": 2300
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 7.690253671562083e-05,
+      "loss": 0.0209,
+      "step": 2304
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 7.703604806408545e-05,
+      "loss": 0.021,
+      "step": 2308
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 7.716955941255008e-05,
+      "loss": 0.023,
+      "step": 2312
+    },
+    {
+      "epoch": 19.79,
+      "learning_rate": 7.73030707610147e-05,
+      "loss": 0.0221,
+      "step": 2316
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 7.743658210947931e-05,
+      "loss": 0.0222,
+      "step": 2320
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 7.757009345794393e-05,
+      "loss": 0.0233,
+      "step": 2324
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 7.770360480640855e-05,
+      "loss": 0.0213,
+      "step": 2328
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 7.783711615487318e-05,
+      "loss": 0.0216,
+      "step": 2332
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 7.797062750333778e-05,
+      "loss": 0.0225,
+      "step": 2336
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 7.810413885180241e-05,
+      "loss": 0.0263,
+      "step": 2340
+    },
+    {
+      "epoch": 20.03,
+      "learning_rate": 7.823765020026703e-05,
+      "loss": 0.019,
+      "step": 2344
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 7.837116154873164e-05,
+      "loss": 0.0232,
+      "step": 2348
+    },
+    {
+      "epoch": 20.1,
+      "learning_rate": 7.850467289719626e-05,
+      "loss": 0.0274,
+      "step": 2352
+    },
+    {
+      "epoch": 20.14,
+      "learning_rate": 7.863818424566088e-05,
+      "loss": 0.0227,
+      "step": 2356
+    },
+    {
+      "epoch": 20.17,
+      "learning_rate": 7.877169559412551e-05,
+      "loss": 0.0229,
+      "step": 2360
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 7.890520694259013e-05,
+      "loss": 0.0222,
+      "step": 2364
+    },
+    {
+      "epoch": 20.24,
+      "learning_rate": 7.903871829105475e-05,
+      "loss": 0.0241,
+      "step": 2368
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 7.917222963951936e-05,
+      "loss": 0.0236,
+      "step": 2372
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 7.930574098798398e-05,
+      "loss": 0.0204,
+      "step": 2376
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 7.94392523364486e-05,
+      "loss": 0.0221,
+      "step": 2380
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 7.957276368491322e-05,
+      "loss": 0.0165,
+      "step": 2384
+    },
+    {
+      "epoch": 20.41,
+      "learning_rate": 7.970627503337784e-05,
+      "loss": 0.0171,
+      "step": 2388
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 7.983978638184246e-05,
+      "loss": 0.0206,
+      "step": 2392
+    },
+    {
+      "epoch": 20.48,
+      "learning_rate": 7.997329773030708e-05,
+      "loss": 0.0207,
+      "step": 2396
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 8.01068090787717e-05,
+      "loss": 0.0176,
+      "step": 2400
+    },
+    {
+      "epoch": 20.55,
+      "learning_rate": 8.024032042723632e-05,
+      "loss": 0.0198,
+      "step": 2404
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 8.037383177570094e-05,
+      "loss": 0.0249,
+      "step": 2408
+    },
+    {
+      "epoch": 20.62,
+      "learning_rate": 8.050734312416556e-05,
+      "loss": 0.02,
+      "step": 2412
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 8.064085447263018e-05,
+      "loss": 0.0216,
+      "step": 2416
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 8.07743658210948e-05,
+      "loss": 0.0259,
+      "step": 2420
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 8.090787716955942e-05,
+      "loss": 0.0226,
+      "step": 2424
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 8.104138851802403e-05,
+      "loss": 0.025,
+      "step": 2428
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 8.117489986648865e-05,
+      "loss": 0.0201,
+      "step": 2432
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 8.130841121495327e-05,
+      "loss": 0.0225,
+      "step": 2436
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 8.14419225634179e-05,
+      "loss": 0.0217,
+      "step": 2440
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 8.157543391188252e-05,
+      "loss": 0.0217,
+      "step": 2444
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 8.170894526034714e-05,
+      "loss": 0.0199,
+      "step": 2448
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 8.184245660881176e-05,
+      "loss": 0.0194,
+      "step": 2452
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 8.197596795727637e-05,
+      "loss": 0.0223,
+      "step": 2456
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 8.2109479305741e-05,
+      "loss": 0.0181,
+      "step": 2460
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 8.22429906542056e-05,
+      "loss": 0.0187,
+      "step": 2464
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 8.237650200267023e-05,
+      "loss": 0.0177,
+      "step": 2468
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 8.251001335113485e-05,
+      "loss": 0.0204,
+      "step": 2472
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 8.264352469959947e-05,
+      "loss": 0.0186,
+      "step": 2476
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 8.27770360480641e-05,
+      "loss": 0.02,
+      "step": 2480
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 8.29105473965287e-05,
+      "loss": 0.0203,
+      "step": 2484
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 8.304405874499333e-05,
+      "loss": 0.0219,
+      "step": 2488
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 8.317757009345795e-05,
+      "loss": 0.0197,
+      "step": 2492
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 8.331108144192257e-05,
+      "loss": 0.0167,
+      "step": 2496
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 8.34445927903872e-05,
+      "loss": 0.0163,
+      "step": 2500
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 8.35781041388518e-05,
+      "loss": 0.0157,
+      "step": 2504
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 8.371161548731643e-05,
+      "loss": 0.0235,
+      "step": 2508
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 8.384512683578104e-05,
+      "loss": 0.0212,
+      "step": 2512
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 8.397863818424566e-05,
+      "loss": 0.0187,
+      "step": 2516
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 8.411214953271028e-05,
+      "loss": 0.015,
+      "step": 2520
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 8.42456608811749e-05,
+      "loss": 0.0163,
+      "step": 2524
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 8.437917222963953e-05,
+      "loss": 0.0167,
+      "step": 2528
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 8.451268357810415e-05,
+      "loss": 0.019,
+      "step": 2532
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 8.464619492656877e-05,
+      "loss": 0.017,
+      "step": 2536
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 8.477970627503338e-05,
+      "loss": 0.0211,
+      "step": 2540
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 8.4913217623498e-05,
+      "loss": 0.0154,
+      "step": 2544
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 8.504672897196261e-05,
+      "loss": 0.0236,
+      "step": 2548
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 8.518024032042724e-05,
+      "loss": 0.0182,
+      "step": 2552
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 8.531375166889186e-05,
+      "loss": 0.0207,
+      "step": 2556
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 8.544726301735648e-05,
+      "loss": 0.0171,
+      "step": 2560
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 8.55807743658211e-05,
+      "loss": 0.0187,
+      "step": 2564
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 0.0175,
+      "step": 2568
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 8.584779706275034e-05,
+      "loss": 0.0176,
+      "step": 2572
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 8.598130841121496e-05,
+      "loss": 0.0159,
+      "step": 2576
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 8.611481975967958e-05,
+      "loss": 0.0172,
+      "step": 2580
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 8.62483311081442e-05,
+      "loss": 0.017,
+      "step": 2584
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 8.638184245660881e-05,
+      "loss": 0.0187,
+      "step": 2588
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 8.651535380507344e-05,
+      "loss": 0.0171,
+      "step": 2592
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 8.664886515353804e-05,
+      "loss": 0.0175,
+      "step": 2596
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 8.678237650200267e-05,
+      "loss": 0.0168,
+      "step": 2600
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 8.691588785046729e-05,
+      "loss": 0.0182,
+      "step": 2604
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 8.704939919893191e-05,
+      "loss": 0.0211,
+      "step": 2608
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 8.718291054739654e-05,
+      "loss": 0.0226,
+      "step": 2612
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 8.731642189586116e-05,
+      "loss": 0.0118,
+      "step": 2616
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 8.744993324432578e-05,
+      "loss": 0.0179,
+      "step": 2620
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 8.758344459279039e-05,
+      "loss": 0.0161,
+      "step": 2624
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 8.771695594125501e-05,
+      "loss": 0.016,
+      "step": 2628
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 8.785046728971964e-05,
+      "loss": 0.0194,
+      "step": 2632
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 8.798397863818424e-05,
+      "loss": 0.0161,
+      "step": 2636
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 8.811748998664887e-05,
+      "loss": 0.0177,
+      "step": 2640
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 8.825100133511349e-05,
+      "loss": 0.0165,
+      "step": 2644
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 8.838451268357811e-05,
+      "loss": 0.0147,
+      "step": 2648
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 8.851802403204272e-05,
+      "loss": 0.0148,
+      "step": 2652
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 8.865153538050734e-05,
+      "loss": 0.0209,
+      "step": 2656
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 8.878504672897197e-05,
+      "loss": 0.0135,
+      "step": 2660
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 8.891855807743659e-05,
+      "loss": 0.0192,
+      "step": 2664
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 8.905206942590121e-05,
+      "loss": 0.0183,
+      "step": 2668
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 8.918558077436584e-05,
+      "loss": 0.019,
+      "step": 2672
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 8.931909212283044e-05,
+      "loss": 0.0151,
+      "step": 2676
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 8.945260347129505e-05,
+      "loss": 0.0127,
+      "step": 2680
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 8.958611481975968e-05,
+      "loss": 0.0169,
+      "step": 2684
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 8.97196261682243e-05,
+      "loss": 0.0143,
+      "step": 2688
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 8.985313751668892e-05,
+      "loss": 0.0195,
+      "step": 2692
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 8.998664886515354e-05,
+      "loss": 0.0151,
+      "step": 2696
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 9.012016021361817e-05,
+      "loss": 0.0124,
+      "step": 2700
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 9.025367156208279e-05,
+      "loss": 0.0156,
+      "step": 2704
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 9.03871829105474e-05,
+      "loss": 0.0157,
+      "step": 2708
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 9.052069425901202e-05,
+      "loss": 0.0169,
+      "step": 2712
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 9.065420560747664e-05,
+      "loss": 0.018,
+      "step": 2716
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 9.078771695594125e-05,
+      "loss": 0.0156,
+      "step": 2720
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 9.092122830440588e-05,
+      "loss": 0.0147,
+      "step": 2724
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 9.10547396528705e-05,
+      "loss": 0.0145,
+      "step": 2728
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 9.118825100133512e-05,
+      "loss": 0.0122,
+      "step": 2732
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 9.132176234979973e-05,
+      "loss": 0.0126,
+      "step": 2736
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 9.145527369826435e-05,
+      "loss": 0.0147,
+      "step": 2740
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 9.158878504672898e-05,
+      "loss": 0.0209,
+      "step": 2744
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 9.17222963951936e-05,
+      "loss": 0.0159,
+      "step": 2748
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 9.185580774365822e-05,
+      "loss": 0.0116,
+      "step": 2752
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 9.198931909212284e-05,
+      "loss": 0.0136,
+      "step": 2756
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 9.212283044058745e-05,
+      "loss": 0.016,
+      "step": 2760
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 9.225634178905206e-05,
+      "loss": 0.0152,
+      "step": 2764
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 9.238985313751669e-05,
+      "loss": 0.0125,
+      "step": 2768
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 9.252336448598131e-05,
+      "loss": 0.0166,
+      "step": 2772
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 9.265687583444593e-05,
+      "loss": 0.013,
+      "step": 2776
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 9.279038718291055e-05,
+      "loss": 0.0162,
+      "step": 2780
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 9.292389853137518e-05,
+      "loss": 0.0128,
+      "step": 2784
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 9.305740987983979e-05,
+      "loss": 0.0104,
+      "step": 2788
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 9.319092122830441e-05,
+      "loss": 0.0143,
+      "step": 2792
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 9.332443257676903e-05,
+      "loss": 0.0143,
+      "step": 2796
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 9.345794392523365e-05,
+      "loss": 0.0162,
+      "step": 2800
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 9.359145527369826e-05,
+      "loss": 0.0167,
+      "step": 2804
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 9.372496662216289e-05,
+      "loss": 0.0143,
+      "step": 2808
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 9.385847797062751e-05,
+      "loss": 0.015,
+      "step": 2812
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 9.399198931909212e-05,
+      "loss": 0.012,
+      "step": 2816
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 9.412550066755674e-05,
+      "loss": 0.0107,
+      "step": 2820
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 9.425901201602136e-05,
+      "loss": 0.0114,
+      "step": 2824
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 9.439252336448599e-05,
+      "loss": 0.0133,
+      "step": 2828
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 9.452603471295061e-05,
+      "loss": 0.0135,
+      "step": 2832
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 9.465954606141523e-05,
+      "loss": 0.0157,
+      "step": 2836
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 9.479305740987985e-05,
+      "loss": 0.0134,
+      "step": 2840
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 9.492656875834446e-05,
+      "loss": 0.0155,
+      "step": 2844
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 9.506008010680909e-05,
+      "loss": 0.0098,
+      "step": 2848
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 9.51935914552737e-05,
+      "loss": 0.0121,
+      "step": 2852
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 9.532710280373832e-05,
+      "loss": 0.0147,
+      "step": 2856
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 9.546061415220294e-05,
+      "loss": 0.0128,
+      "step": 2860
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 9.559412550066756e-05,
+      "loss": 0.0153,
+      "step": 2864
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 9.572763684913219e-05,
+      "loss": 0.0118,
+      "step": 2868
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 9.58611481975968e-05,
+      "loss": 0.0139,
+      "step": 2872
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 9.599465954606142e-05,
+      "loss": 0.0115,
+      "step": 2876
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 9.612817089452604e-05,
+      "loss": 0.0127,
+      "step": 2880
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 9.626168224299066e-05,
+      "loss": 0.0127,
+      "step": 2884
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 9.639519359145529e-05,
+      "loss": 0.0121,
+      "step": 2888
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 9.65287049399199e-05,
+      "loss": 0.011,
+      "step": 2892
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 9.666221628838452e-05,
+      "loss": 0.0126,
+      "step": 2896
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 9.679572763684913e-05,
+      "loss": 0.0112,
+      "step": 2900
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 9.692923898531375e-05,
+      "loss": 0.0152,
+      "step": 2904
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 9.706275033377837e-05,
+      "loss": 0.014,
+      "step": 2908
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 9.7196261682243e-05,
+      "loss": 0.013,
+      "step": 2912
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 9.732977303070762e-05,
+      "loss": 0.0113,
+      "step": 2916
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 9.746328437917224e-05,
+      "loss": 0.0125,
+      "step": 2920
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 9.759679572763686e-05,
+      "loss": 0.012,
+      "step": 2924
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 9.773030707610147e-05,
+      "loss": 0.0127,
+      "step": 2928
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 9.78638184245661e-05,
+      "loss": 0.0107,
+      "step": 2932
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 9.79973297730307e-05,
+      "loss": 0.0094,
+      "step": 2936
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 9.813084112149533e-05,
+      "loss": 0.0148,
+      "step": 2940
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 9.826435246995995e-05,
+      "loss": 0.0133,
+      "step": 2944
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 9.839786381842457e-05,
+      "loss": 0.0105,
+      "step": 2948
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 9.85313751668892e-05,
+      "loss": 0.0114,
+      "step": 2952
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 9.86648865153538e-05,
+      "loss": 0.0133,
+      "step": 2956
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 9.879839786381843e-05,
+      "loss": 0.01,
+      "step": 2960
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 9.893190921228305e-05,
+      "loss": 0.0107,
+      "step": 2964
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 9.906542056074767e-05,
+      "loss": 0.0138,
+      "step": 2968
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 9.91989319092123e-05,
+      "loss": 0.0114,
+      "step": 2972
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 9.93324432576769e-05,
+      "loss": 0.0128,
+      "step": 2976
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 9.946595460614153e-05,
+      "loss": 0.0095,
+      "step": 2980
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 9.959946595460614e-05,
+      "loss": 0.0113,
+      "step": 2984
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 9.973297730307076e-05,
+      "loss": 0.0093,
+      "step": 2988
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 9.986648865153538e-05,
+      "loss": 0.0118,
+      "step": 2992
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 0.0001,
+      "loss": 0.0119,
+      "step": 2996
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 9.99999945668858e-05,
+      "loss": 0.0147,
+      "step": 3000
+    },
+    {
+      "epoch": 25.64,
+      "eval_exact_match": 0.4625779625779626,
+      "eval_loss": 0.6119692325592041,
+      "eval_runtime": 85.6703,
+      "eval_samples_per_second": 11.229,
+      "step": 3000
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 9.99999782675444e-05,
+      "loss": 0.0108,
+      "step": 3004
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 9.999995110197932e-05,
+      "loss": 0.0127,
+      "step": 3008
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 9.999991307019647e-05,
+      "loss": 0.0102,
+      "step": 3012
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 9.999986417220411e-05,
+      "loss": 0.0167,
+      "step": 3016
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 9.999980440801289e-05,
+      "loss": 0.0135,
+      "step": 3020
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 9.999973377763576e-05,
+      "loss": 0.0114,
+      "step": 3024
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 9.999965228108811e-05,
+      "loss": 0.0118,
+      "step": 3028
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 9.999955991838763e-05,
+      "loss": 0.0123,
+      "step": 3032
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 9.99994566895544e-05,
+      "loss": 0.0136,
+      "step": 3036
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 9.999934259461086e-05,
+      "loss": 0.0105,
+      "step": 3040
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.999921763358177e-05,
+      "loss": 0.0088,
+      "step": 3044
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.999908180649433e-05,
+      "loss": 0.0111,
+      "step": 3048
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.999893511337803e-05,
+      "loss": 0.0089,
+      "step": 3052
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.999877755426476e-05,
+      "loss": 0.0081,
+      "step": 3056
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.99986091291888e-05,
+      "loss": 0.0105,
+      "step": 3060
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.999842983818668e-05,
+      "loss": 0.0115,
+      "step": 3064
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.99982396812974e-05,
+      "loss": 0.0082,
+      "step": 3068
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.99980386585623e-05,
+      "loss": 0.0141,
+      "step": 3072
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.999782677002505e-05,
+      "loss": 0.0092,
+      "step": 3076
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.999760401573169e-05,
+      "loss": 0.0117,
+      "step": 3080
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.999737039573065e-05,
+      "loss": 0.0092,
+      "step": 3084
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.99971259100727e-05,
+      "loss": 0.011,
+      "step": 3088
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 9.999687055881095e-05,
+      "loss": 0.0115,
+      "step": 3092
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 9.999660434200093e-05,
+      "loss": 0.007,
+      "step": 3096
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 9.999632725970047e-05,
+      "loss": 0.0127,
+      "step": 3100
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 9.999603931196979e-05,
+      "loss": 0.0122,
+      "step": 3104
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 9.999574049887146e-05,
+      "loss": 0.0111,
+      "step": 3108
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 9.999543082047044e-05,
+      "loss": 0.0138,
+      "step": 3112
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 9.999511027683402e-05,
+      "loss": 0.0079,
+      "step": 3116
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 9.999477886803186e-05,
+      "loss": 0.0105,
+      "step": 3120
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 9.999443659413598e-05,
+      "loss": 0.0112,
+      "step": 3124
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 9.99940834552208e-05,
+      "loss": 0.0124,
+      "step": 3128
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 9.9993719451363e-05,
+      "loss": 0.0089,
+      "step": 3132
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 9.999334458264173e-05,
+      "loss": 0.0087,
+      "step": 3136
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 9.999295884913846e-05,
+      "loss": 0.0126,
+      "step": 3140
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 9.9992562250937e-05,
+      "loss": 0.009,
+      "step": 3144
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 9.999215478812358e-05,
+      "loss": 0.011,
+      "step": 3148
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 9.99917364607867e-05,
+      "loss": 0.0114,
+      "step": 3152
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 9.999130726901729e-05,
+      "loss": 0.0098,
+      "step": 3156
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 9.999086721290864e-05,
+      "loss": 0.0096,
+      "step": 3160
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 9.999041629255637e-05,
+      "loss": 0.0122,
+      "step": 3164
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 9.99899545080585e-05,
+      "loss": 0.0073,
+      "step": 3168
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 9.998948185951535e-05,
+      "loss": 0.0094,
+      "step": 3172
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 9.998899834702964e-05,
+      "loss": 0.008,
+      "step": 3176
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 9.99885039707065e-05,
+      "loss": 0.0086,
+      "step": 3180
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 9.99879987306533e-05,
+      "loss": 0.0122,
+      "step": 3184
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 9.99874826269799e-05,
+      "loss": 0.0103,
+      "step": 3188
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 9.998695565979844e-05,
+      "loss": 0.0085,
+      "step": 3192
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 9.998641782922342e-05,
+      "loss": 0.0083,
+      "step": 3196
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 9.998586913537177e-05,
+      "loss": 0.0111,
+      "step": 3200
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 9.99853095783627e-05,
+      "loss": 0.0077,
+      "step": 3204
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 9.99847391583178e-05,
+      "loss": 0.012,
+      "step": 3208
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 9.998415787536111e-05,
+      "loss": 0.0117,
+      "step": 3212
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 9.998356572961887e-05,
+      "loss": 0.0099,
+      "step": 3216
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 9.998296272121983e-05,
+      "loss": 0.0098,
+      "step": 3220
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 9.998234885029501e-05,
+      "loss": 0.0066,
+      "step": 3224
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 9.998172411697781e-05,
+      "loss": 0.0109,
+      "step": 3228
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 9.998108852140402e-05,
+      "loss": 0.0069,
+      "step": 3232
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 9.998044206371177e-05,
+      "loss": 0.0079,
+      "step": 3236
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 9.997978474404154e-05,
+      "loss": 0.0117,
+      "step": 3240
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 9.997911656253618e-05,
+      "loss": 0.009,
+      "step": 3244
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 9.997843751934093e-05,
+      "loss": 0.0104,
+      "step": 3248
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 9.997774761460332e-05,
+      "loss": 0.0116,
+      "step": 3252
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 9.997704684847332e-05,
+      "loss": 0.0079,
+      "step": 3256
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 9.997633522110322e-05,
+      "loss": 0.0105,
+      "step": 3260
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 9.997561273264764e-05,
+      "loss": 0.0091,
+      "step": 3264
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 9.997487938326362e-05,
+      "loss": 0.0069,
+      "step": 3268
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 9.997413517311055e-05,
+      "loss": 0.0114,
+      "step": 3272
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 9.997338010235013e-05,
+      "loss": 0.0084,
+      "step": 3276
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 9.99726141711465e-05,
+      "loss": 0.0074,
+      "step": 3280
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 9.997183737966606e-05,
+      "loss": 0.0082,
+      "step": 3284
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 9.997104972807768e-05,
+      "loss": 0.0112,
+      "step": 3288
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 9.997025121655248e-05,
+      "loss": 0.0125,
+      "step": 3292
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 9.996944184526405e-05,
+      "loss": 0.0116,
+      "step": 3296
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 9.996862161438825e-05,
+      "loss": 0.0083,
+      "step": 3300
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 9.996779052410337e-05,
+      "loss": 0.0099,
+      "step": 3304
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 9.996694857459e-05,
+      "loss": 0.0095,
+      "step": 3308
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 9.99660957660311e-05,
+      "loss": 0.0057,
+      "step": 3312
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 9.996523209861204e-05,
+      "loss": 0.0047,
+      "step": 3316
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 9.996435757252052e-05,
+      "loss": 0.0108,
+      "step": 3320
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 9.996347218794656e-05,
+      "loss": 0.0081,
+      "step": 3324
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 9.99625759450826e-05,
+      "loss": 0.0095,
+      "step": 3328
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 9.996166884412342e-05,
+      "loss": 0.0069,
+      "step": 3332
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 9.996075088526615e-05,
+      "loss": 0.0106,
+      "step": 3336
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 9.995982206871029e-05,
+      "loss": 0.009,
+      "step": 3340
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 9.995888239465768e-05,
+      "loss": 0.0081,
+      "step": 3344
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 9.995793186331253e-05,
+      "loss": 0.0075,
+      "step": 3348
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 9.995697047488142e-05,
+      "loss": 0.0068,
+      "step": 3352
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 9.99559982295733e-05,
+      "loss": 0.0093,
+      "step": 3356
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 9.995501512759946e-05,
+      "loss": 0.0068,
+      "step": 3360
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 9.995402116917352e-05,
+      "loss": 0.0098,
+      "step": 3364
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 9.995301635451155e-05,
+      "loss": 0.0096,
+      "step": 3368
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 9.995200068383186e-05,
+      "loss": 0.0102,
+      "step": 3372
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 9.995097415735521e-05,
+      "loss": 0.0081,
+      "step": 3376
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 9.994993677530468e-05,
+      "loss": 0.0099,
+      "step": 3380
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 9.994888853790574e-05,
+      "loss": 0.0122,
+      "step": 3384
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 9.994782944538618e-05,
+      "loss": 0.0105,
+      "step": 3388
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 9.994675949797616e-05,
+      "loss": 0.007,
+      "step": 3392
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 9.994567869590822e-05,
+      "loss": 0.0078,
+      "step": 3396
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 9.994458703941723e-05,
+      "loss": 0.0104,
+      "step": 3400
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 9.994348452874046e-05,
+      "loss": 0.0082,
+      "step": 3404
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 9.99423711641175e-05,
+      "loss": 0.006,
+      "step": 3408
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 9.994124694579028e-05,
+      "loss": 0.0098,
+      "step": 3412
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 9.994011187400317e-05,
+      "loss": 0.0094,
+      "step": 3416
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 9.993896594900281e-05,
+      "loss": 0.0062,
+      "step": 3420
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 9.993780917103825e-05,
+      "loss": 0.0067,
+      "step": 3424
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 9.993664154036091e-05,
+      "loss": 0.0069,
+      "step": 3428
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 9.993546305722452e-05,
+      "loss": 0.0079,
+      "step": 3432
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 9.99342737218852e-05,
+      "loss": 0.0086,
+      "step": 3436
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 9.993307353460142e-05,
+      "loss": 0.0068,
+      "step": 3440
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 9.993186249563401e-05,
+      "loss": 0.0079,
+      "step": 3444
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 9.993064060524615e-05,
+      "loss": 0.0081,
+      "step": 3448
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 9.992940786370342e-05,
+      "loss": 0.0072,
+      "step": 3452
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 9.992816427127368e-05,
+      "loss": 0.006,
+      "step": 3456
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 9.992690982822721e-05,
+      "loss": 0.0094,
+      "step": 3460
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.992564453483667e-05,
+      "loss": 0.0078,
+      "step": 3464
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 9.992436839137696e-05,
+      "loss": 0.0045,
+      "step": 3468
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 9.992308139812549e-05,
+      "loss": 0.0055,
+      "step": 3472
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 9.992178355536192e-05,
+      "loss": 0.0059,
+      "step": 3476
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 9.99204748633683e-05,
+      "loss": 0.0087,
+      "step": 3480
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 9.991915532242908e-05,
+      "loss": 0.0082,
+      "step": 3484
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 9.991782493283097e-05,
+      "loss": 0.0094,
+      "step": 3488
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 9.991648369486314e-05,
+      "loss": 0.0077,
+      "step": 3492
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 9.991513160881706e-05,
+      "loss": 0.008,
+      "step": 3496
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 9.991376867498658e-05,
+      "loss": 0.0089,
+      "step": 3500
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 9.991239489366788e-05,
+      "loss": 0.0089,
+      "step": 3504
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 9.991101026515952e-05,
+      "loss": 0.0076,
+      "step": 3508
+    },
+    {
+      "epoch": 30.02,
+      "learning_rate": 9.990961478976244e-05,
+      "loss": 0.0082,
+      "step": 3512
+    },
+    {
+      "epoch": 30.05,
+      "learning_rate": 9.990820846777988e-05,
+      "loss": 0.0099,
+      "step": 3516
+    },
+    {
+      "epoch": 30.09,
+      "learning_rate": 9.990679129951749e-05,
+      "loss": 0.0075,
+      "step": 3520
+    },
+    {
+      "epoch": 30.12,
+      "learning_rate": 9.990536328528324e-05,
+      "loss": 0.0061,
+      "step": 3524
+    },
+    {
+      "epoch": 30.15,
+      "learning_rate": 9.990392442538747e-05,
+      "loss": 0.0067,
+      "step": 3528
+    },
+    {
+      "epoch": 30.19,
+      "learning_rate": 9.990247472014291e-05,
+      "loss": 0.0074,
+      "step": 3532
+    },
+    {
+      "epoch": 30.22,
+      "learning_rate": 9.990101416986457e-05,
+      "loss": 0.0051,
+      "step": 3536
+    },
+    {
+      "epoch": 30.26,
+      "learning_rate": 9.98995427748699e-05,
+      "loss": 0.0063,
+      "step": 3540
+    },
+    {
+      "epoch": 30.29,
+      "learning_rate": 9.989806053547866e-05,
+      "loss": 0.0062,
+      "step": 3544
+    },
+    {
+      "epoch": 30.32,
+      "learning_rate": 9.989656745201298e-05,
+      "loss": 0.0071,
+      "step": 3548
+    },
+    {
+      "epoch": 30.36,
+      "learning_rate": 9.989506352479734e-05,
+      "loss": 0.0062,
+      "step": 3552
+    },
+    {
+      "epoch": 30.39,
+      "learning_rate": 9.989354875415857e-05,
+      "loss": 0.0066,
+      "step": 3556
+    },
+    {
+      "epoch": 30.43,
+      "learning_rate": 9.989202314042588e-05,
+      "loss": 0.0071,
+      "step": 3560
+    },
+    {
+      "epoch": 30.46,
+      "learning_rate": 9.989048668393082e-05,
+      "loss": 0.0087,
+      "step": 3564
+    },
+    {
+      "epoch": 30.5,
+      "learning_rate": 9.988893938500731e-05,
+      "loss": 0.0075,
+      "step": 3568
+    },
+    {
+      "epoch": 30.53,
+      "learning_rate": 9.988738124399159e-05,
+      "loss": 0.0102,
+      "step": 3572
+    },
+    {
+      "epoch": 30.56,
+      "learning_rate": 9.988581226122231e-05,
+      "loss": 0.0075,
+      "step": 3576
+    },
+    {
+      "epoch": 30.6,
+      "learning_rate": 9.988423243704042e-05,
+      "loss": 0.0095,
+      "step": 3580
+    },
+    {
+      "epoch": 30.63,
+      "learning_rate": 9.988264177178929e-05,
+      "loss": 0.0061,
+      "step": 3584
+    },
+    {
+      "epoch": 30.67,
+      "learning_rate": 9.988104026581458e-05,
+      "loss": 0.0063,
+      "step": 3588
+    },
+    {
+      "epoch": 30.7,
+      "learning_rate": 9.987942791946436e-05,
+      "loss": 0.0057,
+      "step": 3592
+    },
+    {
+      "epoch": 30.74,
+      "learning_rate": 9.987780473308901e-05,
+      "loss": 0.0054,
+      "step": 3596
+    },
+    {
+      "epoch": 30.77,
+      "learning_rate": 9.98761707070413e-05,
+      "loss": 0.0053,
+      "step": 3600
+    },
+    {
+      "epoch": 30.8,
+      "learning_rate": 9.987452584167635e-05,
+      "loss": 0.0096,
+      "step": 3604
+    },
+    {
+      "epoch": 30.84,
+      "learning_rate": 9.987287013735161e-05,
+      "loss": 0.0066,
+      "step": 3608
+    },
+    {
+      "epoch": 30.87,
+      "learning_rate": 9.987120359442693e-05,
+      "loss": 0.0088,
+      "step": 3612
+    },
+    {
+      "epoch": 30.91,
+      "learning_rate": 9.986952621326448e-05,
+      "loss": 0.0064,
+      "step": 3616
+    },
+    {
+      "epoch": 30.94,
+      "learning_rate": 9.986783799422879e-05,
+      "loss": 0.0091,
+      "step": 3620
+    },
+    {
+      "epoch": 30.97,
+      "learning_rate": 9.986613893768675e-05,
+      "loss": 0.0064,
+      "step": 3624
+    },
+    {
+      "epoch": 31.01,
+      "learning_rate": 9.986442904400763e-05,
+      "loss": 0.0065,
+      "step": 3628
+    },
+    {
+      "epoch": 31.04,
+      "learning_rate": 9.9862708313563e-05,
+      "loss": 0.0055,
+      "step": 3632
+    },
+    {
+      "epoch": 31.08,
+      "learning_rate": 9.986097674672683e-05,
+      "loss": 0.0041,
+      "step": 3636
+    },
+    {
+      "epoch": 31.11,
+      "learning_rate": 9.985923434387545e-05,
+      "loss": 0.0072,
+      "step": 3640
+    },
+    {
+      "epoch": 31.15,
+      "learning_rate": 9.98574811053875e-05,
+      "loss": 0.0084,
+      "step": 3644
+    },
+    {
+      "epoch": 31.18,
+      "learning_rate": 9.985571703164402e-05,
+      "loss": 0.0074,
+      "step": 3648
+    },
+    {
+      "epoch": 31.21,
+      "learning_rate": 9.985394212302836e-05,
+      "loss": 0.0073,
+      "step": 3652
+    },
+    {
+      "epoch": 31.25,
+      "learning_rate": 9.98521563799263e-05,
+      "loss": 0.0059,
+      "step": 3656
+    },
+    {
+      "epoch": 31.28,
+      "learning_rate": 9.985035980272588e-05,
+      "loss": 0.0065,
+      "step": 3660
+    },
+    {
+      "epoch": 31.32,
+      "learning_rate": 9.984855239181755e-05,
+      "loss": 0.0058,
+      "step": 3664
+    },
+    {
+      "epoch": 31.35,
+      "learning_rate": 9.984673414759411e-05,
+      "loss": 0.0069,
+      "step": 3668
+    },
+    {
+      "epoch": 31.38,
+      "learning_rate": 9.984490507045073e-05,
+      "loss": 0.0068,
+      "step": 3672
+    },
+    {
+      "epoch": 31.42,
+      "learning_rate": 9.984306516078488e-05,
+      "loss": 0.006,
+      "step": 3676
+    },
+    {
+      "epoch": 31.45,
+      "learning_rate": 9.984121441899644e-05,
+      "loss": 0.0071,
+      "step": 3680
+    },
+    {
+      "epoch": 31.49,
+      "learning_rate": 9.983935284548761e-05,
+      "loss": 0.0052,
+      "step": 3684
+    },
+    {
+      "epoch": 31.52,
+      "learning_rate": 9.983748044066296e-05,
+      "loss": 0.0055,
+      "step": 3688
+    },
+    {
+      "epoch": 31.56,
+      "learning_rate": 9.983559720492941e-05,
+      "loss": 0.0065,
+      "step": 3692
+    },
+    {
+      "epoch": 31.59,
+      "learning_rate": 9.983370313869622e-05,
+      "loss": 0.0063,
+      "step": 3696
+    },
+    {
+      "epoch": 31.62,
+      "learning_rate": 9.983179824237505e-05,
+      "loss": 0.0045,
+      "step": 3700
+    },
+    {
+      "epoch": 31.66,
+      "learning_rate": 9.982988251637983e-05,
+      "loss": 0.0054,
+      "step": 3704
+    },
+    {
+      "epoch": 31.69,
+      "learning_rate": 9.982795596112695e-05,
+      "loss": 0.0062,
+      "step": 3708
+    },
+    {
+      "epoch": 31.73,
+      "learning_rate": 9.982601857703507e-05,
+      "loss": 0.009,
+      "step": 3712
+    },
+    {
+      "epoch": 31.76,
+      "learning_rate": 9.982407036452523e-05,
+      "loss": 0.0089,
+      "step": 3716
+    },
+    {
+      "epoch": 31.79,
+      "learning_rate": 9.982211132402082e-05,
+      "loss": 0.0065,
+      "step": 3720
+    },
+    {
+      "epoch": 31.83,
+      "learning_rate": 9.98201414559476e-05,
+      "loss": 0.0053,
+      "step": 3724
+    },
+    {
+      "epoch": 31.86,
+      "learning_rate": 9.981816076073368e-05,
+      "loss": 0.0073,
+      "step": 3728
+    },
+    {
+      "epoch": 31.9,
+      "learning_rate": 9.981616923880947e-05,
+      "loss": 0.0063,
+      "step": 3732
+    },
+    {
+      "epoch": 31.93,
+      "learning_rate": 9.981416689060784e-05,
+      "loss": 0.0052,
+      "step": 3736
+    },
+    {
+      "epoch": 31.97,
+      "learning_rate": 9.981215371656388e-05,
+      "loss": 0.0062,
+      "step": 3740
+    },
+    {
+      "epoch": 32.0,
+      "learning_rate": 9.981012971711516e-05,
+      "loss": 0.0058,
+      "step": 3744
+    },
+    {
+      "epoch": 32.03,
+      "learning_rate": 9.980809489270152e-05,
+      "loss": 0.0065,
+      "step": 3748
+    },
+    {
+      "epoch": 32.07,
+      "learning_rate": 9.980604924376518e-05,
+      "loss": 0.0044,
+      "step": 3752
+    },
+    {
+      "epoch": 32.1,
+      "learning_rate": 9.98039927707507e-05,
+      "loss": 0.0059,
+      "step": 3756
+    },
+    {
+      "epoch": 32.14,
+      "learning_rate": 9.980192547410503e-05,
+      "loss": 0.0069,
+      "step": 3760
+    },
+    {
+      "epoch": 32.17,
+      "learning_rate": 9.979984735427741e-05,
+      "loss": 0.006,
+      "step": 3764
+    },
+    {
+      "epoch": 32.21,
+      "learning_rate": 9.979775841171949e-05,
+      "loss": 0.0053,
+      "step": 3768
+    },
+    {
+      "epoch": 32.24,
+      "learning_rate": 9.979565864688524e-05,
+      "loss": 0.005,
+      "step": 3772
+    },
+    {
+      "epoch": 32.27,
+      "learning_rate": 9.9793548060231e-05,
+      "loss": 0.0054,
+      "step": 3776
+    },
+    {
+      "epoch": 32.31,
+      "learning_rate": 9.979142665221544e-05,
+      "loss": 0.0042,
+      "step": 3780
+    },
+    {
+      "epoch": 32.34,
+      "learning_rate": 9.978929442329959e-05,
+      "loss": 0.0063,
+      "step": 3784
+    },
+    {
+      "epoch": 32.38,
+      "learning_rate": 9.978715137394685e-05,
+      "loss": 0.0053,
+      "step": 3788
+    },
+    {
+      "epoch": 32.41,
+      "learning_rate": 9.978499750462295e-05,
+      "loss": 0.0052,
+      "step": 3792
+    },
+    {
+      "epoch": 32.44,
+      "learning_rate": 9.978283281579598e-05,
+      "loss": 0.005,
+      "step": 3796
+    },
+    {
+      "epoch": 32.48,
+      "learning_rate": 9.978065730793638e-05,
+      "loss": 0.0068,
+      "step": 3800
+    },
+    {
+      "epoch": 32.51,
+      "learning_rate": 9.977847098151694e-05,
+      "loss": 0.0057,
+      "step": 3804
+    },
+    {
+      "epoch": 32.55,
+      "learning_rate": 9.97762738370128e-05,
+      "loss": 0.0074,
+      "step": 3808
+    },
+    {
+      "epoch": 32.58,
+      "learning_rate": 9.977406587490146e-05,
+      "loss": 0.005,
+      "step": 3812
+    },
+    {
+      "epoch": 32.62,
+      "learning_rate": 9.977184709566277e-05,
+      "loss": 0.0055,
+      "step": 3816
+    },
+    {
+      "epoch": 32.65,
+      "learning_rate": 9.976961749977892e-05,
+      "loss": 0.0075,
+      "step": 3820
+    },
+    {
+      "epoch": 32.68,
+      "learning_rate": 9.976737708773445e-05,
+      "loss": 0.0045,
+      "step": 3824
+    },
+    {
+      "epoch": 32.72,
+      "learning_rate": 9.976512586001625e-05,
+      "loss": 0.0066,
+      "step": 3828
+    },
+    {
+      "epoch": 32.75,
+      "learning_rate": 9.976286381711357e-05,
+      "loss": 0.0064,
+      "step": 3832
+    },
+    {
+      "epoch": 32.79,
+      "learning_rate": 9.976059095951804e-05,
+      "loss": 0.0086,
+      "step": 3836
+    },
+    {
+      "epoch": 32.82,
+      "learning_rate": 9.975830728772355e-05,
+      "loss": 0.0069,
+      "step": 3840
+    },
+    {
+      "epoch": 32.85,
+      "learning_rate": 9.975601280222647e-05,
+      "loss": 0.006,
+      "step": 3844
+    },
+    {
+      "epoch": 32.89,
+      "learning_rate": 9.975370750352538e-05,
+      "loss": 0.0046,
+      "step": 3848
+    },
+    {
+      "epoch": 32.92,
+      "learning_rate": 9.975139139212131e-05,
+      "loss": 0.0065,
+      "step": 3852
+    },
+    {
+      "epoch": 32.96,
+      "learning_rate": 9.97490644685176e-05,
+      "loss": 0.007,
+      "step": 3856
+    },
+    {
+      "epoch": 32.99,
+      "learning_rate": 9.974672673321995e-05,
+      "loss": 0.0056,
+      "step": 3860
+    },
+    {
+      "epoch": 33.03,
+      "learning_rate": 9.974437818673642e-05,
+      "loss": 0.0051,
+      "step": 3864
+    },
+    {
+      "epoch": 33.06,
+      "learning_rate": 9.974201882957739e-05,
+      "loss": 0.0041,
+      "step": 3868
+    },
+    {
+      "epoch": 33.09,
+      "learning_rate": 9.97396486622556e-05,
+      "loss": 0.0049,
+      "step": 3872
+    },
+    {
+      "epoch": 33.13,
+      "learning_rate": 9.973726768528618e-05,
+      "loss": 0.006,
+      "step": 3876
+    },
+    {
+      "epoch": 33.16,
+      "learning_rate": 9.973487589918655e-05,
+      "loss": 0.0065,
+      "step": 3880
+    },
+    {
+      "epoch": 33.2,
+      "learning_rate": 9.973247330447649e-05,
+      "loss": 0.0057,
+      "step": 3884
+    },
+    {
+      "epoch": 33.23,
+      "learning_rate": 9.973005990167816e-05,
+      "loss": 0.0059,
+      "step": 3888
+    },
+    {
+      "epoch": 33.26,
+      "learning_rate": 9.972763569131606e-05,
+      "loss": 0.0059,
+      "step": 3892
+    },
+    {
+      "epoch": 33.3,
+      "learning_rate": 9.972520067391702e-05,
+      "loss": 0.0051,
+      "step": 3896
+    },
+    {
+      "epoch": 33.33,
+      "learning_rate": 9.972275485001024e-05,
+      "loss": 0.0061,
+      "step": 3900
+    },
+    {
+      "epoch": 33.37,
+      "learning_rate": 9.972029822012722e-05,
+      "loss": 0.006,
+      "step": 3904
+    },
+    {
+      "epoch": 33.4,
+      "learning_rate": 9.971783078480191e-05,
+      "loss": 0.0049,
+      "step": 3908
+    },
+    {
+      "epoch": 33.44,
+      "learning_rate": 9.97153525445705e-05,
+      "loss": 0.004,
+      "step": 3912
+    },
+    {
+      "epoch": 33.47,
+      "learning_rate": 9.971286349997156e-05,
+      "loss": 0.0072,
+      "step": 3916
+    },
+    {
+      "epoch": 33.5,
+      "learning_rate": 9.971036365154604e-05,
+      "loss": 0.0062,
+      "step": 3920
+    },
+    {
+      "epoch": 33.54,
+      "learning_rate": 9.970785299983725e-05,
+      "loss": 0.0049,
+      "step": 3924
+    },
+    {
+      "epoch": 33.57,
+      "learning_rate": 9.970533154539077e-05,
+      "loss": 0.0048,
+      "step": 3928
+    },
+    {
+      "epoch": 33.61,
+      "learning_rate": 9.970279928875458e-05,
+      "loss": 0.0048,
+      "step": 3932
+    },
+    {
+      "epoch": 33.64,
+      "learning_rate": 9.970025623047902e-05,
+      "loss": 0.0066,
+      "step": 3936
+    },
+    {
+      "epoch": 33.68,
+      "learning_rate": 9.969770237111676e-05,
+      "loss": 0.0068,
+      "step": 3940
+    },
+    {
+      "epoch": 33.71,
+      "learning_rate": 9.969513771122278e-05,
+      "loss": 0.0052,
+      "step": 3944
+    },
+    {
+      "epoch": 33.74,
+      "learning_rate": 9.96925622513545e-05,
+      "loss": 0.0049,
+      "step": 3948
+    },
+    {
+      "epoch": 33.78,
+      "learning_rate": 9.96899759920716e-05,
+      "loss": 0.0068,
+      "step": 3952
+    },
+    {
+      "epoch": 33.81,
+      "learning_rate": 9.968737893393612e-05,
+      "loss": 0.0072,
+      "step": 3956
+    },
+    {
+      "epoch": 33.85,
+      "learning_rate": 9.96847710775125e-05,
+      "loss": 0.0118,
+      "step": 3960
+    },
+    {
+      "epoch": 33.88,
+      "learning_rate": 9.968215242336746e-05,
+      "loss": 0.0051,
+      "step": 3964
+    },
+    {
+      "epoch": 33.91,
+      "learning_rate": 9.967952297207013e-05,
+      "loss": 0.0039,
+      "step": 3968
+    },
+    {
+      "epoch": 33.95,
+      "learning_rate": 9.967688272419193e-05,
+      "loss": 0.0049,
+      "step": 3972
+    },
+    {
+      "epoch": 33.98,
+      "learning_rate": 9.967423168030665e-05,
+      "loss": 0.0058,
+      "step": 3976
+    },
+    {
+      "epoch": 34.02,
+      "learning_rate": 9.967156984099044e-05,
+      "loss": 0.0049,
+      "step": 3980
+    },
+    {
+      "epoch": 34.05,
+      "learning_rate": 9.966889720682177e-05,
+      "loss": 0.005,
+      "step": 3984
+    },
+    {
+      "epoch": 34.09,
+      "learning_rate": 9.96662137783815e-05,
+      "loss": 0.0055,
+      "step": 3988
+    },
+    {
+      "epoch": 34.12,
+      "learning_rate": 9.966351955625277e-05,
+      "loss": 0.0038,
+      "step": 3992
+    },
+    {
+      "epoch": 34.15,
+      "learning_rate": 9.96608145410211e-05,
+      "loss": 0.0038,
+      "step": 3996
+    },
+    {
+      "epoch": 34.19,
+      "learning_rate": 9.965809873327439e-05,
+      "loss": 0.0054,
+      "step": 4000
+    },
+    {
+      "epoch": 34.19,
+      "eval_exact_match": 0.4896049896049896,
+      "eval_loss": 0.668160617351532,
+      "eval_runtime": 89.1426,
+      "eval_samples_per_second": 10.792,
+      "step": 4000
+    },
+    {
+      "epoch": 34.22,
+      "learning_rate": 9.965537213360282e-05,
+      "loss": 0.0065,
+      "step": 4004
+    },
+    {
+      "epoch": 34.26,
+      "learning_rate": 9.965263474259896e-05,
+      "loss": 0.006,
+      "step": 4008
+    },
+    {
+      "epoch": 34.29,
+      "learning_rate": 9.964988656085771e-05,
+      "loss": 0.0042,
+      "step": 4012
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 9.964712758897632e-05,
+      "loss": 0.0054,
+      "step": 4016
+    },
+    {
+      "epoch": 34.36,
+      "learning_rate": 9.964435782755437e-05,
+      "loss": 0.0047,
+      "step": 4020
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 9.964157727719381e-05,
+      "loss": 0.0032,
+      "step": 4024
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 9.963878593849893e-05,
+      "loss": 0.0052,
+      "step": 4028
+    },
+    {
+      "epoch": 34.46,
+      "learning_rate": 9.963598381207632e-05,
+      "loss": 0.0039,
+      "step": 4032
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 9.963317089853502e-05,
+      "loss": 0.0038,
+      "step": 4036
+    },
+    {
+      "epoch": 34.53,
+      "learning_rate": 9.963034719848626e-05,
+      "loss": 0.0049,
+      "step": 4040
+    },
+    {
+      "epoch": 34.56,
+      "learning_rate": 9.962751271254375e-05,
+      "loss": 0.0105,
+      "step": 4044
+    },
+    {
+      "epoch": 34.6,
+      "learning_rate": 9.96246674413235e-05,
+      "loss": 0.0042,
+      "step": 4048
+    },
+    {
+      "epoch": 34.63,
+      "learning_rate": 9.962181138544383e-05,
+      "loss": 0.0053,
+      "step": 4052
+    },
+    {
+      "epoch": 34.67,
+      "learning_rate": 9.961894454552545e-05,
+      "loss": 0.0056,
+      "step": 4056
+    },
+    {
+      "epoch": 34.7,
+      "learning_rate": 9.961606692219137e-05,
+      "loss": 0.0054,
+      "step": 4060
+    },
+    {
+      "epoch": 34.74,
+      "learning_rate": 9.961317851606701e-05,
+      "loss": 0.0051,
+      "step": 4064
+    },
+    {
+      "epoch": 34.77,
+      "learning_rate": 9.961027932778005e-05,
+      "loss": 0.0037,
+      "step": 4068
+    },
+    {
+      "epoch": 34.8,
+      "learning_rate": 9.960736935796058e-05,
+      "loss": 0.004,
+      "step": 4072
+    },
+    {
+      "epoch": 34.84,
+      "learning_rate": 9.9604448607241e-05,
+      "loss": 0.0058,
+      "step": 4076
+    },
+    {
+      "epoch": 34.87,
+      "learning_rate": 9.960151707625605e-05,
+      "loss": 0.0044,
+      "step": 4080
+    },
+    {
+      "epoch": 34.91,
+      "learning_rate": 9.959857476564285e-05,
+      "loss": 0.0056,
+      "step": 4084
+    },
+    {
+      "epoch": 34.94,
+      "learning_rate": 9.959562167604082e-05,
+      "loss": 0.0066,
+      "step": 4088
+    },
+    {
+      "epoch": 34.97,
+      "learning_rate": 9.959265780809172e-05,
+      "loss": 0.0058,
+      "step": 4092
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 9.958968316243972e-05,
+      "loss": 0.0067,
+      "step": 4096
+    },
+    {
+      "epoch": 35.04,
+      "learning_rate": 9.958669773973123e-05,
+      "loss": 0.0067,
+      "step": 4100
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 9.958370154061511e-05,
+      "loss": 0.0054,
+      "step": 4104
+    },
+    {
+      "epoch": 35.11,
+      "learning_rate": 9.958069456574246e-05,
+      "loss": 0.0042,
+      "step": 4108
+    },
+    {
+      "epoch": 35.15,
+      "learning_rate": 9.957767681576679e-05,
+      "loss": 0.0037,
+      "step": 4112
+    },
+    {
+      "epoch": 35.18,
+      "learning_rate": 9.957464829134391e-05,
+      "loss": 0.0043,
+      "step": 4116
+    },
+    {
+      "epoch": 35.21,
+      "learning_rate": 9.957160899313205e-05,
+      "loss": 0.0041,
+      "step": 4120
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 9.956855892179167e-05,
+      "loss": 0.0047,
+      "step": 4124
+    },
+    {
+      "epoch": 35.28,
+      "learning_rate": 9.956549807798563e-05,
+      "loss": 0.0033,
+      "step": 4128
+    },
+    {
+      "epoch": 35.32,
+      "learning_rate": 9.956242646237914e-05,
+      "loss": 0.0049,
+      "step": 4132
+    },
+    {
+      "epoch": 35.35,
+      "learning_rate": 9.955934407563974e-05,
+      "loss": 0.0048,
+      "step": 4136
+    },
+    {
+      "epoch": 35.38,
+      "learning_rate": 9.95562509184373e-05,
+      "loss": 0.005,
+      "step": 4140
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 9.955314699144406e-05,
+      "loss": 0.0036,
+      "step": 4144
+    },
+    {
+      "epoch": 35.45,
+      "learning_rate": 9.955003229533455e-05,
+      "loss": 0.0044,
+      "step": 4148
+    },
+    {
+      "epoch": 35.49,
+      "learning_rate": 9.954690683078569e-05,
+      "loss": 0.0039,
+      "step": 4152
+    },
+    {
+      "epoch": 35.52,
+      "learning_rate": 9.954377059847669e-05,
+      "loss": 0.004,
+      "step": 4156
+    },
+    {
+      "epoch": 35.56,
+      "learning_rate": 9.954062359908918e-05,
+      "loss": 0.0055,
+      "step": 4160
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 9.953746583330703e-05,
+      "loss": 0.0043,
+      "step": 4164
+    },
+    {
+      "epoch": 35.62,
+      "learning_rate": 9.953429730181653e-05,
+      "loss": 0.0056,
+      "step": 4168
+    },
+    {
+      "epoch": 35.66,
+      "learning_rate": 9.953111800530628e-05,
+      "loss": 0.0053,
+      "step": 4172
+    },
+    {
+      "epoch": 35.69,
+      "learning_rate": 9.952792794446722e-05,
+      "loss": 0.003,
+      "step": 4176
+    },
+    {
+      "epoch": 35.73,
+      "learning_rate": 9.952472711999261e-05,
+      "loss": 0.0057,
+      "step": 4180
+    },
+    {
+      "epoch": 35.76,
+      "learning_rate": 9.952151553257809e-05,
+      "loss": 0.0052,
+      "step": 4184
+    },
+    {
+      "epoch": 35.79,
+      "learning_rate": 9.951829318292159e-05,
+      "loss": 0.0046,
+      "step": 4188
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 9.951506007172343e-05,
+      "loss": 0.0028,
+      "step": 4192
+    },
+    {
+      "epoch": 35.86,
+      "learning_rate": 9.951181619968624e-05,
+      "loss": 0.0056,
+      "step": 4196
+    },
+    {
+      "epoch": 35.9,
+      "learning_rate": 9.950856156751498e-05,
+      "loss": 0.0029,
+      "step": 4200
+    },
+    {
+      "epoch": 35.93,
+      "learning_rate": 9.950529617591698e-05,
+      "loss": 0.0046,
+      "step": 4204
+    },
+    {
+      "epoch": 35.97,
+      "learning_rate": 9.950202002560187e-05,
+      "loss": 0.0031,
+      "step": 4208
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 9.949873311728166e-05,
+      "loss": 0.0036,
+      "step": 4212
+    },
+    {
+      "epoch": 36.03,
+      "learning_rate": 9.949543545167066e-05,
+      "loss": 0.0046,
+      "step": 4216
+    },
+    {
+      "epoch": 36.07,
+      "learning_rate": 9.949212702948552e-05,
+      "loss": 0.0026,
+      "step": 4220
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 9.948880785144528e-05,
+      "loss": 0.0036,
+      "step": 4224
+    },
+    {
+      "epoch": 36.14,
+      "learning_rate": 9.948547791827126e-05,
+      "loss": 0.0046,
+      "step": 4228
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 9.948213723068712e-05,
+      "loss": 0.0025,
+      "step": 4232
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 9.94787857894189e-05,
+      "loss": 0.0043,
+      "step": 4236
+    },
+    {
+      "epoch": 36.24,
+      "learning_rate": 9.947542359519492e-05,
+      "loss": 0.0045,
+      "step": 4240
+    },
+    {
+      "epoch": 36.27,
+      "learning_rate": 9.947205064874591e-05,
+      "loss": 0.0057,
+      "step": 4244
+    },
+    {
+      "epoch": 36.31,
+      "learning_rate": 9.946866695080486e-05,
+      "loss": 0.0023,
+      "step": 4248
+    },
+    {
+      "epoch": 36.34,
+      "learning_rate": 9.946527250210715e-05,
+      "loss": 0.0033,
+      "step": 4252
+    },
+    {
+      "epoch": 36.38,
+      "learning_rate": 9.946186730339047e-05,
+      "loss": 0.0053,
+      "step": 4256
+    },
+    {
+      "epoch": 36.41,
+      "learning_rate": 9.945845135539483e-05,
+      "loss": 0.0028,
+      "step": 4260
+    },
+    {
+      "epoch": 36.44,
+      "learning_rate": 9.945502465886263e-05,
+      "loss": 0.0037,
+      "step": 4264
+    },
+    {
+      "epoch": 36.48,
+      "learning_rate": 9.945158721453857e-05,
+      "loss": 0.007,
+      "step": 4268
+    },
+    {
+      "epoch": 36.51,
+      "learning_rate": 9.94481390231697e-05,
+      "loss": 0.0027,
+      "step": 4272
+    },
+    {
+      "epoch": 36.55,
+      "learning_rate": 9.944468008550536e-05,
+      "loss": 0.0021,
+      "step": 4276
+    },
+    {
+      "epoch": 36.58,
+      "learning_rate": 9.94412104022973e-05,
+      "loss": 0.0047,
+      "step": 4280
+    },
+    {
+      "epoch": 36.62,
+      "learning_rate": 9.943772997429954e-05,
+      "loss": 0.0056,
+      "step": 4284
+    },
+    {
+      "epoch": 36.65,
+      "learning_rate": 9.943423880226848e-05,
+      "loss": 0.0046,
+      "step": 4288
+    },
+    {
+      "epoch": 36.68,
+      "learning_rate": 9.943073688696285e-05,
+      "loss": 0.0053,
+      "step": 4292
+    },
+    {
+      "epoch": 36.72,
+      "learning_rate": 9.942722422914367e-05,
+      "loss": 0.0032,
+      "step": 4296
+    },
+    {
+      "epoch": 36.75,
+      "learning_rate": 9.942370082957435e-05,
+      "loss": 0.0034,
+      "step": 4300
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 9.942016668902058e-05,
+      "loss": 0.0041,
+      "step": 4304
+    },
+    {
+      "epoch": 36.82,
+      "learning_rate": 9.941662180825048e-05,
+      "loss": 0.004,
+      "step": 4308
+    },
+    {
+      "epoch": 36.85,
+      "learning_rate": 9.941306618803436e-05,
+      "loss": 0.0058,
+      "step": 4312
+    },
+    {
+      "epoch": 36.89,
+      "learning_rate": 9.940949982914502e-05,
+      "loss": 0.0031,
+      "step": 4316
+    },
+    {
+      "epoch": 36.92,
+      "learning_rate": 9.940592273235744e-05,
+      "loss": 0.0065,
+      "step": 4320
+    },
+    {
+      "epoch": 36.96,
+      "learning_rate": 9.94023348984491e-05,
+      "loss": 0.0041,
+      "step": 4324
+    },
+    {
+      "epoch": 36.99,
+      "learning_rate": 9.939873632819964e-05,
+      "loss": 0.0054,
+      "step": 4328
+    },
+    {
+      "epoch": 37.03,
+      "learning_rate": 9.939512702239116e-05,
+      "loss": 0.0051,
+      "step": 4332
+    },
+    {
+      "epoch": 37.06,
+      "learning_rate": 9.939150698180804e-05,
+      "loss": 0.0032,
+      "step": 4336
+    },
+    {
+      "epoch": 37.09,
+      "learning_rate": 9.938787620723702e-05,
+      "loss": 0.0047,
+      "step": 4340
+    },
+    {
+      "epoch": 37.13,
+      "learning_rate": 9.938423469946713e-05,
+      "loss": 0.006,
+      "step": 4344
+    },
+    {
+      "epoch": 37.16,
+      "learning_rate": 9.938058245928978e-05,
+      "loss": 0.0044,
+      "step": 4348
+    },
+    {
+      "epoch": 37.2,
+      "learning_rate": 9.937691948749869e-05,
+      "loss": 0.0051,
+      "step": 4352
+    },
+    {
+      "epoch": 37.23,
+      "learning_rate": 9.93732457848899e-05,
+      "loss": 0.0065,
+      "step": 4356
+    },
+    {
+      "epoch": 37.26,
+      "learning_rate": 9.936956135226181e-05,
+      "loss": 0.0022,
+      "step": 4360
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 9.936586619041514e-05,
+      "loss": 0.0027,
+      "step": 4364
+    },
+    {
+      "epoch": 37.33,
+      "learning_rate": 9.936216030015291e-05,
+      "loss": 0.0039,
+      "step": 4368
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 9.935844368228054e-05,
+      "loss": 0.0032,
+      "step": 4372
+    },
+    {
+      "epoch": 37.4,
+      "learning_rate": 9.935471633760573e-05,
+      "loss": 0.0056,
+      "step": 4376
+    },
+    {
+      "epoch": 37.44,
+      "learning_rate": 9.935097826693851e-05,
+      "loss": 0.0026,
+      "step": 4380
+    },
+    {
+      "epoch": 37.47,
+      "learning_rate": 9.934722947109125e-05,
+      "loss": 0.003,
+      "step": 4384
+    },
+    {
+      "epoch": 37.5,
+      "learning_rate": 9.934346995087868e-05,
+      "loss": 0.0032,
+      "step": 4388
+    },
+    {
+      "epoch": 37.54,
+      "learning_rate": 9.933969970711783e-05,
+      "loss": 0.0028,
+      "step": 4392
+    },
+    {
+      "epoch": 37.57,
+      "learning_rate": 9.933591874062806e-05,
+      "loss": 0.004,
+      "step": 4396
+    },
+    {
+      "epoch": 37.61,
+      "learning_rate": 9.933212705223107e-05,
+      "loss": 0.0041,
+      "step": 4400
+    },
+    {
+      "epoch": 37.64,
+      "learning_rate": 9.932832464275088e-05,
+      "loss": 0.004,
+      "step": 4404
+    },
+    {
+      "epoch": 37.68,
+      "learning_rate": 9.932451151301386e-05,
+      "loss": 0.0051,
+      "step": 4408
+    },
+    {
+      "epoch": 37.71,
+      "learning_rate": 9.932068766384868e-05,
+      "loss": 0.0033,
+      "step": 4412
+    },
+    {
+      "epoch": 37.74,
+      "learning_rate": 9.931685309608636e-05,
+      "loss": 0.003,
+      "step": 4416
+    },
+    {
+      "epoch": 37.78,
+      "learning_rate": 9.931300781056027e-05,
+      "loss": 0.0048,
+      "step": 4420
+    },
+    {
+      "epoch": 37.81,
+      "learning_rate": 9.930915180810605e-05,
+      "loss": 0.0047,
+      "step": 4424
+    },
+    {
+      "epoch": 37.85,
+      "learning_rate": 9.930528508956172e-05,
+      "loss": 0.0037,
+      "step": 4428
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 9.930140765576763e-05,
+      "loss": 0.0055,
+      "step": 4432
+    },
+    {
+      "epoch": 37.91,
+      "learning_rate": 9.929751950756641e-05,
+      "loss": 0.0041,
+      "step": 4436
+    },
+    {
+      "epoch": 37.95,
+      "learning_rate": 9.929362064580307e-05,
+      "loss": 0.004,
+      "step": 4440
+    },
+    {
+      "epoch": 37.98,
+      "learning_rate": 9.928971107132493e-05,
+      "loss": 0.0047,
+      "step": 4444
+    },
+    {
+      "epoch": 38.02,
+      "learning_rate": 9.928579078498161e-05,
+      "loss": 0.0033,
+      "step": 4448
+    },
+    {
+      "epoch": 38.05,
+      "learning_rate": 9.928185978762512e-05,
+      "loss": 0.003,
+      "step": 4452
+    },
+    {
+      "epoch": 38.09,
+      "learning_rate": 9.927791808010976e-05,
+      "loss": 0.004,
+      "step": 4456
+    },
+    {
+      "epoch": 38.12,
+      "learning_rate": 9.927396566329212e-05,
+      "loss": 0.0041,
+      "step": 4460
+    },
+    {
+      "epoch": 38.15,
+      "learning_rate": 9.92700025380312e-05,
+      "loss": 0.0055,
+      "step": 4464
+    },
+    {
+      "epoch": 38.19,
+      "learning_rate": 9.926602870518826e-05,
+      "loss": 0.0038,
+      "step": 4468
+    },
+    {
+      "epoch": 38.22,
+      "learning_rate": 9.926204416562692e-05,
+      "loss": 0.005,
+      "step": 4472
+    },
+    {
+      "epoch": 38.26,
+      "learning_rate": 9.925804892021313e-05,
+      "loss": 0.0057,
+      "step": 4476
+    },
+    {
+      "epoch": 38.29,
+      "learning_rate": 9.925404296981513e-05,
+      "loss": 0.0034,
+      "step": 4480
+    },
+    {
+      "epoch": 38.32,
+      "learning_rate": 9.925002631530353e-05,
+      "loss": 0.0039,
+      "step": 4484
+    },
+    {
+      "epoch": 38.36,
+      "learning_rate": 9.924599895755126e-05,
+      "loss": 0.0037,
+      "step": 4488
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 9.924196089743352e-05,
+      "loss": 0.0038,
+      "step": 4492
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 9.923791213582791e-05,
+      "loss": 0.0027,
+      "step": 4496
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 9.923385267361434e-05,
+      "loss": 0.0042,
+      "step": 4500
+    },
+    {
+      "epoch": 38.5,
+      "learning_rate": 9.9229782511675e-05,
+      "loss": 0.0054,
+      "step": 4504
+    },
+    {
+      "epoch": 38.53,
+      "learning_rate": 9.922570165089445e-05,
+      "loss": 0.0038,
+      "step": 4508
+    },
+    {
+      "epoch": 38.56,
+      "learning_rate": 9.922161009215956e-05,
+      "loss": 0.004,
+      "step": 4512
+    },
+    {
+      "epoch": 38.6,
+      "learning_rate": 9.921750783635952e-05,
+      "loss": 0.003,
+      "step": 4516
+    },
+    {
+      "epoch": 38.63,
+      "learning_rate": 9.921339488438585e-05,
+      "loss": 0.0033,
+      "step": 4520
+    },
+    {
+      "epoch": 38.67,
+      "learning_rate": 9.920927123713242e-05,
+      "loss": 0.0029,
+      "step": 4524
+    },
+    {
+      "epoch": 38.7,
+      "learning_rate": 9.920513689549537e-05,
+      "loss": 0.0035,
+      "step": 4528
+    },
+    {
+      "epoch": 38.74,
+      "learning_rate": 9.920099186037321e-05,
+      "loss": 0.0051,
+      "step": 4532
+    },
+    {
+      "epoch": 38.77,
+      "learning_rate": 9.919683613266677e-05,
+      "loss": 0.0037,
+      "step": 4536
+    },
+    {
+      "epoch": 38.8,
+      "learning_rate": 9.919266971327916e-05,
+      "loss": 0.0032,
+      "step": 4540
+    },
+    {
+      "epoch": 38.84,
+      "learning_rate": 9.918849260311586e-05,
+      "loss": 0.0044,
+      "step": 4544
+    },
+    {
+      "epoch": 38.87,
+      "learning_rate": 9.918430480308466e-05,
+      "loss": 0.0022,
+      "step": 4548
+    },
+    {
+      "epoch": 38.91,
+      "learning_rate": 9.918010631409568e-05,
+      "loss": 0.0035,
+      "step": 4552
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 9.917589713706135e-05,
+      "loss": 0.0048,
+      "step": 4556
+    },
+    {
+      "epoch": 38.97,
+      "learning_rate": 9.91716772728964e-05,
+      "loss": 0.0029,
+      "step": 4560
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 9.916744672251795e-05,
+      "loss": 0.0055,
+      "step": 4564
+    },
+    {
+      "epoch": 39.04,
+      "learning_rate": 9.91632054868454e-05,
+      "loss": 0.0036,
+      "step": 4568
+    },
+    {
+      "epoch": 39.08,
+      "learning_rate": 9.915895356680043e-05,
+      "loss": 0.0035,
+      "step": 4572
+    },
+    {
+      "epoch": 39.11,
+      "learning_rate": 9.915469096330714e-05,
+      "loss": 0.0036,
+      "step": 4576
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 9.915041767729188e-05,
+      "loss": 0.002,
+      "step": 4580
+    },
+    {
+      "epoch": 39.18,
+      "learning_rate": 9.914613370968333e-05,
+      "loss": 0.0031,
+      "step": 4584
+    },
+    {
+      "epoch": 39.21,
+      "learning_rate": 9.914183906141252e-05,
+      "loss": 0.0043,
+      "step": 4588
+    },
+    {
+      "epoch": 39.25,
+      "learning_rate": 9.913753373341274e-05,
+      "loss": 0.0044,
+      "step": 4592
+    },
+    {
+      "epoch": 39.28,
+      "learning_rate": 9.91332177266197e-05,
+      "loss": 0.0039,
+      "step": 4596
+    },
+    {
+      "epoch": 39.32,
+      "learning_rate": 9.912889104197134e-05,
+      "loss": 0.004,
+      "step": 4600
+    },
+    {
+      "epoch": 39.35,
+      "learning_rate": 9.912455368040797e-05,
+      "loss": 0.0034,
+      "step": 4604
+    },
+    {
+      "epoch": 39.38,
+      "learning_rate": 9.912020564287218e-05,
+      "loss": 0.0032,
+      "step": 4608
+    },
+    {
+      "epoch": 39.42,
+      "learning_rate": 9.911584693030895e-05,
+      "loss": 0.004,
+      "step": 4612
+    },
+    {
+      "epoch": 39.45,
+      "learning_rate": 9.91114775436655e-05,
+      "loss": 0.0057,
+      "step": 4616
+    },
+    {
+      "epoch": 39.49,
+      "learning_rate": 9.910709748389141e-05,
+      "loss": 0.0039,
+      "step": 4620
+    },
+    {
+      "epoch": 39.52,
+      "learning_rate": 9.910270675193858e-05,
+      "loss": 0.0025,
+      "step": 4624
+    },
+    {
+      "epoch": 39.56,
+      "learning_rate": 9.909830534876123e-05,
+      "loss": 0.0036,
+      "step": 4628
+    },
+    {
+      "epoch": 39.59,
+      "learning_rate": 9.909389327531588e-05,
+      "loss": 0.0035,
+      "step": 4632
+    },
+    {
+      "epoch": 39.62,
+      "learning_rate": 9.908947053256139e-05,
+      "loss": 0.0049,
+      "step": 4636
+    },
+    {
+      "epoch": 39.66,
+      "learning_rate": 9.908503712145892e-05,
+      "loss": 0.0037,
+      "step": 4640
+    },
+    {
+      "epoch": 39.69,
+      "learning_rate": 9.908059304297198e-05,
+      "loss": 0.0029,
+      "step": 4644
+    },
+    {
+      "epoch": 39.73,
+      "learning_rate": 9.907613829806637e-05,
+      "loss": 0.0036,
+      "step": 4648
+    },
+    {
+      "epoch": 39.76,
+      "learning_rate": 9.907167288771019e-05,
+      "loss": 0.0044,
+      "step": 4652
+    },
+    {
+      "epoch": 39.79,
+      "learning_rate": 9.90671968128739e-05,
+      "loss": 0.0043,
+      "step": 4656
+    },
+    {
+      "epoch": 39.83,
+      "learning_rate": 9.90627100745303e-05,
+      "loss": 0.0073,
+      "step": 4660
+    },
+    {
+      "epoch": 39.86,
+      "learning_rate": 9.90582126736544e-05,
+      "loss": 0.003,
+      "step": 4664
+    },
+    {
+      "epoch": 39.9,
+      "learning_rate": 9.905370461122366e-05,
+      "loss": 0.0029,
+      "step": 4668
+    },
+    {
+      "epoch": 39.93,
+      "learning_rate": 9.904918588821775e-05,
+      "loss": 0.0045,
+      "step": 4672
+    },
+    {
+      "epoch": 39.97,
+      "learning_rate": 9.904465650561869e-05,
+      "loss": 0.0047,
+      "step": 4676
+    },
+    {
+      "epoch": 40.0,
+      "learning_rate": 9.904011646441087e-05,
+      "loss": 0.0046,
+      "step": 4680
+    },
+    {
+      "epoch": 40.03,
+      "learning_rate": 9.903556576558093e-05,
+      "loss": 0.0058,
+      "step": 4684
+    },
+    {
+      "epoch": 40.07,
+      "learning_rate": 9.903100441011783e-05,
+      "loss": 0.0035,
+      "step": 4688
+    },
+    {
+      "epoch": 40.1,
+      "learning_rate": 9.90264323990129e-05,
+      "loss": 0.0069,
+      "step": 4692
+    },
+    {
+      "epoch": 40.14,
+      "learning_rate": 9.902184973325975e-05,
+      "loss": 0.0032,
+      "step": 4696
+    },
+    {
+      "epoch": 40.17,
+      "learning_rate": 9.901725641385425e-05,
+      "loss": 0.0029,
+      "step": 4700
+    },
+    {
+      "epoch": 40.21,
+      "learning_rate": 9.90126524417947e-05,
+      "loss": 0.0033,
+      "step": 4704
+    },
+    {
+      "epoch": 40.24,
+      "learning_rate": 9.900803781808164e-05,
+      "loss": 0.0023,
+      "step": 4708
+    },
+    {
+      "epoch": 40.27,
+      "learning_rate": 9.900341254371794e-05,
+      "loss": 0.0029,
+      "step": 4712
+    },
+    {
+      "epoch": 40.31,
+      "learning_rate": 9.899877661970877e-05,
+      "loss": 0.0033,
+      "step": 4716
+    },
+    {
+      "epoch": 40.34,
+      "learning_rate": 9.899413004706166e-05,
+      "loss": 0.0025,
+      "step": 4720
+    },
+    {
+      "epoch": 40.38,
+      "learning_rate": 9.898947282678642e-05,
+      "loss": 0.0042,
+      "step": 4724
+    },
+    {
+      "epoch": 40.41,
+      "learning_rate": 9.898480495989514e-05,
+      "loss": 0.0042,
+      "step": 4728
+    },
+    {
+      "epoch": 40.44,
+      "learning_rate": 9.898012644740229e-05,
+      "loss": 0.0021,
+      "step": 4732
+    },
+    {
+      "epoch": 40.48,
+      "learning_rate": 9.897543729032463e-05,
+      "loss": 0.0033,
+      "step": 4736
+    },
+    {
+      "epoch": 40.51,
+      "learning_rate": 9.897073748968124e-05,
+      "loss": 0.005,
+      "step": 4740
+    },
+    {
+      "epoch": 40.55,
+      "learning_rate": 9.896602704649346e-05,
+      "loss": 0.0048,
+      "step": 4744
+    },
+    {
+      "epoch": 40.58,
+      "learning_rate": 9.896130596178503e-05,
+      "loss": 0.0023,
+      "step": 4748
+    },
+    {
+      "epoch": 40.62,
+      "learning_rate": 9.895657423658193e-05,
+      "loss": 0.0027,
+      "step": 4752
+    },
+    {
+      "epoch": 40.65,
+      "learning_rate": 9.895183187191251e-05,
+      "loss": 0.0038,
+      "step": 4756
+    },
+    {
+      "epoch": 40.68,
+      "learning_rate": 9.894707886880735e-05,
+      "loss": 0.0034,
+      "step": 4760
+    },
+    {
+      "epoch": 40.72,
+      "learning_rate": 9.894231522829944e-05,
+      "loss": 0.0034,
+      "step": 4764
+    },
+    {
+      "epoch": 40.75,
+      "learning_rate": 9.893754095142403e-05,
+      "loss": 0.0039,
+      "step": 4768
+    },
+    {
+      "epoch": 40.79,
+      "learning_rate": 9.893275603921865e-05,
+      "loss": 0.0024,
+      "step": 4772
+    },
+    {
+      "epoch": 40.82,
+      "learning_rate": 9.892796049272324e-05,
+      "loss": 0.002,
+      "step": 4776
+    },
+    {
+      "epoch": 40.85,
+      "learning_rate": 9.892315431297994e-05,
+      "loss": 0.0036,
+      "step": 4780
+    },
+    {
+      "epoch": 40.89,
+      "learning_rate": 9.891833750103328e-05,
+      "loss": 0.0037,
+      "step": 4784
+    },
+    {
+      "epoch": 40.92,
+      "learning_rate": 9.891351005793005e-05,
+      "loss": 0.004,
+      "step": 4788
+    },
+    {
+      "epoch": 40.96,
+      "learning_rate": 9.890867198471938e-05,
+      "loss": 0.0036,
+      "step": 4792
+    },
+    {
+      "epoch": 40.99,
+      "learning_rate": 9.890382328245271e-05,
+      "loss": 0.0057,
+      "step": 4796
+    },
+    {
+      "epoch": 41.03,
+      "learning_rate": 9.889896395218378e-05,
+      "loss": 0.0028,
+      "step": 4800
+    },
+    {
+      "epoch": 41.06,
+      "learning_rate": 9.889409399496865e-05,
+      "loss": 0.0018,
+      "step": 4804
+    },
+    {
+      "epoch": 41.09,
+      "learning_rate": 9.888921341186566e-05,
+      "loss": 0.0027,
+      "step": 4808
+    },
+    {
+      "epoch": 41.13,
+      "learning_rate": 9.888432220393549e-05,
+      "loss": 0.0047,
+      "step": 4812
+    },
+    {
+      "epoch": 41.16,
+      "learning_rate": 9.887942037224111e-05,
+      "loss": 0.0037,
+      "step": 4816
+    },
+    {
+      "epoch": 41.2,
+      "learning_rate": 9.887450791784782e-05,
+      "loss": 0.0034,
+      "step": 4820
+    },
+    {
+      "epoch": 41.23,
+      "learning_rate": 9.886958484182324e-05,
+      "loss": 0.0037,
+      "step": 4824
+    },
+    {
+      "epoch": 41.26,
+      "learning_rate": 9.886465114523724e-05,
+      "loss": 0.0048,
+      "step": 4828
+    },
+    {
+      "epoch": 41.3,
+      "learning_rate": 9.885970682916204e-05,
+      "loss": 0.0028,
+      "step": 4832
+    },
+    {
+      "epoch": 41.33,
+      "learning_rate": 9.885475189467217e-05,
+      "loss": 0.0036,
+      "step": 4836
+    },
+    {
+      "epoch": 41.37,
+      "learning_rate": 9.884978634284445e-05,
+      "loss": 0.0055,
+      "step": 4840
+    },
+    {
+      "epoch": 41.4,
+      "learning_rate": 9.884481017475801e-05,
+      "loss": 0.0049,
+      "step": 4844
+    },
+    {
+      "epoch": 41.44,
+      "learning_rate": 9.883982339149433e-05,
+      "loss": 0.0031,
+      "step": 4848
+    },
+    {
+      "epoch": 41.47,
+      "learning_rate": 9.883482599413712e-05,
+      "loss": 0.0038,
+      "step": 4852
+    },
+    {
+      "epoch": 41.5,
+      "learning_rate": 9.882981798377247e-05,
+      "loss": 0.0017,
+      "step": 4856
+    },
+    {
+      "epoch": 41.54,
+      "learning_rate": 9.882479936148872e-05,
+      "loss": 0.004,
+      "step": 4860
+    },
+    {
+      "epoch": 41.57,
+      "learning_rate": 9.881977012837653e-05,
+      "loss": 0.0024,
+      "step": 4864
+    },
+    {
+      "epoch": 41.61,
+      "learning_rate": 9.88147302855289e-05,
+      "loss": 0.003,
+      "step": 4868
+    },
+    {
+      "epoch": 41.64,
+      "learning_rate": 9.880967983404111e-05,
+      "loss": 0.0016,
+      "step": 4872
+    },
+    {
+      "epoch": 41.68,
+      "learning_rate": 9.880461877501075e-05,
+      "loss": 0.0047,
+      "step": 4876
+    },
+    {
+      "epoch": 41.71,
+      "learning_rate": 9.879954710953769e-05,
+      "loss": 0.0038,
+      "step": 4880
+    },
+    {
+      "epoch": 41.74,
+      "learning_rate": 9.879446483872414e-05,
+      "loss": 0.0022,
+      "step": 4884
+    },
+    {
+      "epoch": 41.78,
+      "learning_rate": 9.878937196367462e-05,
+      "loss": 0.004,
+      "step": 4888
+    },
+    {
+      "epoch": 41.81,
+      "learning_rate": 9.878426848549589e-05,
+      "loss": 0.0031,
+      "step": 4892
+    },
+    {
+      "epoch": 41.85,
+      "learning_rate": 9.877915440529711e-05,
+      "loss": 0.0027,
+      "step": 4896
+    },
+    {
+      "epoch": 41.88,
+      "learning_rate": 9.877402972418968e-05,
+      "loss": 0.003,
+      "step": 4900
+    },
+    {
+      "epoch": 41.91,
+      "learning_rate": 9.876889444328731e-05,
+      "loss": 0.0024,
+      "step": 4904
+    },
+    {
+      "epoch": 41.95,
+      "learning_rate": 9.876374856370603e-05,
+      "loss": 0.0033,
+      "step": 4908
+    },
+    {
+      "epoch": 41.98,
+      "learning_rate": 9.875859208656418e-05,
+      "loss": 0.003,
+      "step": 4912
+    },
+    {
+      "epoch": 42.02,
+      "learning_rate": 9.875342501298235e-05,
+      "loss": 0.003,
+      "step": 4916
+    },
+    {
+      "epoch": 42.05,
+      "learning_rate": 9.874824734408351e-05,
+      "loss": 0.0038,
+      "step": 4920
+    },
+    {
+      "epoch": 42.09,
+      "learning_rate": 9.874305908099288e-05,
+      "loss": 0.0034,
+      "step": 4924
+    },
+    {
+      "epoch": 42.12,
+      "learning_rate": 9.8737860224838e-05,
+      "loss": 0.0038,
+      "step": 4928
+    },
+    {
+      "epoch": 42.15,
+      "learning_rate": 9.87326507767487e-05,
+      "loss": 0.0023,
+      "step": 4932
+    },
+    {
+      "epoch": 42.19,
+      "learning_rate": 9.872743073785713e-05,
+      "loss": 0.0036,
+      "step": 4936
+    },
+    {
+      "epoch": 42.22,
+      "learning_rate": 9.872220010929772e-05,
+      "loss": 0.0052,
+      "step": 4940
+    },
+    {
+      "epoch": 42.26,
+      "learning_rate": 9.871695889220725e-05,
+      "loss": 0.0019,
+      "step": 4944
+    },
+    {
+      "epoch": 42.29,
+      "learning_rate": 9.871170708772471e-05,
+      "loss": 0.0019,
+      "step": 4948
+    },
+    {
+      "epoch": 42.32,
+      "learning_rate": 9.87064446969915e-05,
+      "loss": 0.0055,
+      "step": 4952
+    },
+    {
+      "epoch": 42.36,
+      "learning_rate": 9.870117172115123e-05,
+      "loss": 0.0031,
+      "step": 4956
+    },
+    {
+      "epoch": 42.39,
+      "learning_rate": 9.869588816134987e-05,
+      "loss": 0.0028,
+      "step": 4960
+    },
+    {
+      "epoch": 42.43,
+      "learning_rate": 9.869059401873564e-05,
+      "loss": 0.0028,
+      "step": 4964
+    },
+    {
+      "epoch": 42.46,
+      "learning_rate": 9.868528929445912e-05,
+      "loss": 0.0019,
+      "step": 4968
+    },
+    {
+      "epoch": 42.5,
+      "learning_rate": 9.867997398967313e-05,
+      "loss": 0.002,
+      "step": 4972
+    },
+    {
+      "epoch": 42.53,
+      "learning_rate": 9.867464810553283e-05,
+      "loss": 0.0035,
+      "step": 4976
+    },
+    {
+      "epoch": 42.56,
+      "learning_rate": 9.866931164319568e-05,
+      "loss": 0.0044,
+      "step": 4980
+    },
+    {
+      "epoch": 42.6,
+      "learning_rate": 9.866396460382139e-05,
+      "loss": 0.0069,
+      "step": 4984
+    },
+    {
+      "epoch": 42.63,
+      "learning_rate": 9.865860698857204e-05,
+      "loss": 0.0025,
+      "step": 4988
+    },
+    {
+      "epoch": 42.67,
+      "learning_rate": 9.865323879861194e-05,
+      "loss": 0.0032,
+      "step": 4992
+    },
+    {
+      "epoch": 42.7,
+      "learning_rate": 9.864786003510773e-05,
+      "loss": 0.0047,
+      "step": 4996
+    },
+    {
+      "epoch": 42.74,
+      "learning_rate": 9.864247069922838e-05,
+      "loss": 0.0027,
+      "step": 5000
+    },
+    {
+      "epoch": 42.74,
+      "eval_exact_match": 0.4875259875259875,
+      "eval_loss": 0.7059731483459473,
+      "eval_runtime": 86.6231,
+      "eval_samples_per_second": 11.106,
+      "step": 5000
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 8.352528793189786e-05,
+      "loss": 0.0021,
+      "step": 5004
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 8.359205474878986e-05,
+      "loss": 0.002,
+      "step": 5008
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 8.365882156568185e-05,
+      "loss": 0.0031,
+      "step": 5012
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 8.372558838257387e-05,
+      "loss": 0.0031,
+      "step": 5016
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 8.379235519946587e-05,
+      "loss": 0.0036,
+      "step": 5020
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 8.385912201635787e-05,
+      "loss": 0.0018,
+      "step": 5024
+    },
+    {
+      "epoch": 21.49,
+      "learning_rate": 8.392588883324987e-05,
+      "loss": 0.0016,
+      "step": 5028
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 8.399265565014188e-05,
+      "loss": 0.0026,
+      "step": 5032
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 8.405942246703389e-05,
+      "loss": 0.0018,
+      "step": 5036
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 8.41261892839259e-05,
+      "loss": 0.0022,
+      "step": 5040
+    },
+    {
+      "epoch": 21.56,
+      "learning_rate": 8.41929561008179e-05,
+      "loss": 0.0008,
+      "step": 5044
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 8.42597229177099e-05,
+      "loss": 0.0035,
+      "step": 5048
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 8.43264897346019e-05,
+      "loss": 0.0017,
+      "step": 5052
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 8.439325655149392e-05,
+      "loss": 0.004,
+      "step": 5056
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 8.446002336838592e-05,
+      "loss": 0.0023,
+      "step": 5060
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 8.452679018527791e-05,
+      "loss": 0.0032,
+      "step": 5064
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 8.459355700216993e-05,
+      "loss": 0.003,
+      "step": 5068
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 8.466032381906193e-05,
+      "loss": 0.0038,
+      "step": 5072
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 8.472709063595393e-05,
+      "loss": 0.0053,
+      "step": 5076
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 8.479385745284594e-05,
+      "loss": 0.0018,
+      "step": 5080
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 8.486062426973794e-05,
+      "loss": 0.0018,
+      "step": 5084
+    },
+    {
+      "epoch": 21.74,
+      "learning_rate": 8.492739108662995e-05,
+      "loss": 0.0059,
+      "step": 5088
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 8.499415790352196e-05,
+      "loss": 0.002,
+      "step": 5092
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 8.506092472041396e-05,
+      "loss": 0.0042,
+      "step": 5096
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 8.512769153730596e-05,
+      "loss": 0.0048,
+      "step": 5100
+    },
+    {
+      "epoch": 21.81,
+      "learning_rate": 8.519445835419796e-05,
+      "loss": 0.002,
+      "step": 5104
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 8.526122517108998e-05,
+      "loss": 0.0034,
+      "step": 5108
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 8.532799198798198e-05,
+      "loss": 0.0044,
+      "step": 5112
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 8.539475880487399e-05,
+      "loss": 0.0045,
+      "step": 5116
+    },
+    {
+      "epoch": 21.88,
+      "learning_rate": 8.546152562176597e-05,
+      "loss": 0.002,
+      "step": 5120
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 8.552829243865799e-05,
+      "loss": 0.0023,
+      "step": 5124
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 8.559505925555e-05,
+      "loss": 0.0027,
+      "step": 5128
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 8.5661826072442e-05,
+      "loss": 0.0055,
+      "step": 5132
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 8.5728592889334e-05,
+      "loss": 0.0035,
+      "step": 5136
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 8.579535970622602e-05,
+      "loss": 0.0038,
+      "step": 5140
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 8.586212652311802e-05,
+      "loss": 0.0031,
+      "step": 5144
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 8.592889334001002e-05,
+      "loss": 0.0043,
+      "step": 5148
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 8.599566015690202e-05,
+      "loss": 0.0034,
+      "step": 5152
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 8.606242697379403e-05,
+      "loss": 0.0042,
+      "step": 5156
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 8.612919379068604e-05,
+      "loss": 0.0022,
+      "step": 5160
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 8.619596060757804e-05,
+      "loss": 0.0039,
+      "step": 5164
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 8.626272742447005e-05,
+      "loss": 0.0026,
+      "step": 5168
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 8.632949424136205e-05,
+      "loss": 0.0022,
+      "step": 5172
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 8.639626105825405e-05,
+      "loss": 0.0055,
+      "step": 5176
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 8.646302787514605e-05,
+      "loss": 0.0034,
+      "step": 5180
+    },
+    {
+      "epoch": 22.15,
+      "learning_rate": 8.652979469203806e-05,
+      "loss": 0.0043,
+      "step": 5184
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 8.659656150893006e-05,
+      "loss": 0.0033,
+      "step": 5188
+    },
+    {
+      "epoch": 22.19,
+      "learning_rate": 8.666332832582206e-05,
+      "loss": 0.0072,
+      "step": 5192
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 8.673009514271408e-05,
+      "loss": 0.0042,
+      "step": 5196
+    },
+    {
+      "epoch": 22.22,
+      "learning_rate": 8.679686195960608e-05,
+      "loss": 0.0062,
+      "step": 5200
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 8.686362877649808e-05,
+      "loss": 0.0025,
+      "step": 5204
+    },
+    {
+      "epoch": 22.26,
+      "learning_rate": 8.693039559339009e-05,
+      "loss": 0.0044,
+      "step": 5208
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 8.699716241028209e-05,
+      "loss": 0.0037,
+      "step": 5212
+    },
+    {
+      "epoch": 22.29,
+      "learning_rate": 8.70639292271741e-05,
+      "loss": 0.002,
+      "step": 5216
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 8.713069604406611e-05,
+      "loss": 0.0024,
+      "step": 5220
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 8.719746286095811e-05,
+      "loss": 0.0018,
+      "step": 5224
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 8.726422967785011e-05,
+      "loss": 0.0025,
+      "step": 5228
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 8.733099649474212e-05,
+      "loss": 0.0051,
+      "step": 5232
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 8.739776331163412e-05,
+      "loss": 0.0027,
+      "step": 5236
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 8.746453012852612e-05,
+      "loss": 0.0043,
+      "step": 5240
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 8.753129694541812e-05,
+      "loss": 0.0048,
+      "step": 5244
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 8.759806376231014e-05,
+      "loss": 0.0014,
+      "step": 5248
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 8.766483057920214e-05,
+      "loss": 0.0019,
+      "step": 5252
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 8.773159739609415e-05,
+      "loss": 0.0038,
+      "step": 5256
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 8.779836421298615e-05,
+      "loss": 0.0026,
+      "step": 5260
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 8.786513102987815e-05,
+      "loss": 0.0027,
+      "step": 5264
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 8.793189784677017e-05,
+      "loss": 0.0042,
+      "step": 5268
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 8.799866466366217e-05,
+      "loss": 0.0026,
+      "step": 5272
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 8.806543148055417e-05,
+      "loss": 0.0037,
+      "step": 5276
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 8.813219829744617e-05,
+      "loss": 0.0058,
+      "step": 5280
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 8.819896511433818e-05,
+      "loss": 0.004,
+      "step": 5284
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 8.826573193123018e-05,
+      "loss": 0.0026,
+      "step": 5288
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 8.833249874812218e-05,
+      "loss": 0.0037,
+      "step": 5292
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 8.839926556501418e-05,
+      "loss": 0.0024,
+      "step": 5296
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 8.84660323819062e-05,
+      "loss": 0.0026,
+      "step": 5300
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 8.85327991987982e-05,
+      "loss": 0.0032,
+      "step": 5304
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 8.85995660156902e-05,
+      "loss": 0.0031,
+      "step": 5308
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 8.866633283258221e-05,
+      "loss": 0.0068,
+      "step": 5312
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 8.873309964947421e-05,
+      "loss": 0.0037,
+      "step": 5316
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 8.879986646636623e-05,
+      "loss": 0.0068,
+      "step": 5320
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 8.886663328325823e-05,
+      "loss": 0.0025,
+      "step": 5324
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 8.893340010015023e-05,
+      "loss": 0.0056,
+      "step": 5328
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 8.900016691704224e-05,
+      "loss": 0.0053,
+      "step": 5332
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 8.906693373393424e-05,
+      "loss": 0.0053,
+      "step": 5336
+    },
+    {
+      "epoch": 22.82,
+      "learning_rate": 8.913370055082625e-05,
+      "loss": 0.003,
+      "step": 5340
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 8.920046736771824e-05,
+      "loss": 0.0052,
+      "step": 5344
+    },
+    {
+      "epoch": 22.85,
+      "learning_rate": 8.926723418461025e-05,
+      "loss": 0.0072,
+      "step": 5348
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 8.933400100150225e-05,
+      "loss": 0.0034,
+      "step": 5352
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 8.940076781839426e-05,
+      "loss": 0.0027,
+      "step": 5356
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 8.946753463528627e-05,
+      "loss": 0.0032,
+      "step": 5360
+    },
+    {
+      "epoch": 22.92,
+      "learning_rate": 8.953430145217827e-05,
+      "loss": 0.0038,
+      "step": 5364
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 8.960106826907027e-05,
+      "loss": 0.0036,
+      "step": 5368
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 8.966783508596228e-05,
+      "loss": 0.003,
+      "step": 5372
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 8.973460190285429e-05,
+      "loss": 0.0026,
+      "step": 5376
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 8.98013687197463e-05,
+      "loss": 0.0054,
+      "step": 5380
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 8.98681355366383e-05,
+      "loss": 0.0027,
+      "step": 5384
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 8.99349023535303e-05,
+      "loss": 0.0022,
+      "step": 5388
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 9.000166917042232e-05,
+      "loss": 0.0048,
+      "step": 5392
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 9.006843598731432e-05,
+      "loss": 0.0034,
+      "step": 5396
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 9.013520280420631e-05,
+      "loss": 0.0033,
+      "step": 5400
+    },
+    {
+      "epoch": 23.09,
+      "learning_rate": 9.020196962109831e-05,
+      "loss": 0.0023,
+      "step": 5404
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 9.026873643799033e-05,
+      "loss": 0.0026,
+      "step": 5408
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 9.033550325488233e-05,
+      "loss": 0.002,
+      "step": 5412
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 9.040227007177433e-05,
+      "loss": 0.0022,
+      "step": 5416
+    },
+    {
+      "epoch": 23.16,
+      "learning_rate": 9.046903688866633e-05,
+      "loss": 0.0075,
+      "step": 5420
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 9.053580370555834e-05,
+      "loss": 0.0057,
+      "step": 5424
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 9.060257052245035e-05,
+      "loss": 0.0016,
+      "step": 5428
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 9.066933733934235e-05,
+      "loss": 0.0063,
+      "step": 5432
+    },
+    {
+      "epoch": 23.23,
+      "learning_rate": 9.073610415623436e-05,
+      "loss": 0.003,
+      "step": 5436
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 9.080287097312636e-05,
+      "loss": 0.004,
+      "step": 5440
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 9.086963779001836e-05,
+      "loss": 0.0035,
+      "step": 5444
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 9.093640460691038e-05,
+      "loss": 0.0033,
+      "step": 5448
+    },
+    {
+      "epoch": 23.3,
+      "learning_rate": 9.100317142380237e-05,
+      "loss": 0.0062,
+      "step": 5452
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 9.106993824069437e-05,
+      "loss": 0.0029,
+      "step": 5456
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 9.113670505758639e-05,
+      "loss": 0.0055,
+      "step": 5460
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 9.120347187447839e-05,
+      "loss": 0.0037,
+      "step": 5464
+    },
+    {
+      "epoch": 23.37,
+      "learning_rate": 9.127023869137039e-05,
+      "loss": 0.0021,
+      "step": 5468
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 9.13370055082624e-05,
+      "loss": 0.0035,
+      "step": 5472
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 9.14037723251544e-05,
+      "loss": 0.0022,
+      "step": 5476
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 9.147053914204641e-05,
+      "loss": 0.0019,
+      "step": 5480
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 9.153730595893842e-05,
+      "loss": 0.004,
+      "step": 5484
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 9.160407277583042e-05,
+      "loss": 0.0026,
+      "step": 5488
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 9.167083959272242e-05,
+      "loss": 0.0039,
+      "step": 5492
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 9.173760640961442e-05,
+      "loss": 0.0048,
+      "step": 5496
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 9.180437322650644e-05,
+      "loss": 0.0026,
+      "step": 5500
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 9.187114004339844e-05,
+      "loss": 0.005,
+      "step": 5504
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 9.193790686029043e-05,
+      "loss": 0.0035,
+      "step": 5508
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 9.200467367718243e-05,
+      "loss": 0.0015,
+      "step": 5512
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 9.207144049407445e-05,
+      "loss": 0.005,
+      "step": 5516
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 9.213820731096645e-05,
+      "loss": 0.0016,
+      "step": 5520
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 9.220497412785846e-05,
+      "loss": 0.0032,
+      "step": 5524
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 9.227174094475046e-05,
+      "loss": 0.0027,
+      "step": 5528
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 9.233850776164246e-05,
+      "loss": 0.0033,
+      "step": 5532
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 9.240527457853448e-05,
+      "loss": 0.0021,
+      "step": 5536
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 9.247204139542648e-05,
+      "loss": 0.0024,
+      "step": 5540
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 9.253880821231848e-05,
+      "loss": 0.0036,
+      "step": 5544
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 9.260557502921048e-05,
+      "loss": 0.0051,
+      "step": 5548
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 9.26723418461025e-05,
+      "loss": 0.0022,
+      "step": 5552
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 9.27391086629945e-05,
+      "loss": 0.0042,
+      "step": 5556
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 9.28058754798865e-05,
+      "loss": 0.003,
+      "step": 5560
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 9.28726422967785e-05,
+      "loss": 0.0024,
+      "step": 5564
+    },
+    {
+      "epoch": 23.79,
+      "learning_rate": 9.293940911367051e-05,
+      "loss": 0.0043,
+      "step": 5568
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 9.300617593056251e-05,
+      "loss": 0.0047,
+      "step": 5572
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 9.307294274745452e-05,
+      "loss": 0.0029,
+      "step": 5576
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 9.313970956434652e-05,
+      "loss": 0.0028,
+      "step": 5580
+    },
+    {
+      "epoch": 23.86,
+      "learning_rate": 9.320647638123852e-05,
+      "loss": 0.0057,
+      "step": 5584
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 9.327324319813054e-05,
+      "loss": 0.0042,
+      "step": 5588
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 9.334001001502254e-05,
+      "loss": 0.0024,
+      "step": 5592
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 9.340677683191454e-05,
+      "loss": 0.0037,
+      "step": 5596
+    },
+    {
+      "epoch": 23.93,
+      "learning_rate": 9.347354364880655e-05,
+      "loss": 0.005,
+      "step": 5600
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 9.354031046569855e-05,
+      "loss": 0.006,
+      "step": 5604
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 9.360707728259056e-05,
+      "loss": 0.0077,
+      "step": 5608
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 9.367384409948257e-05,
+      "loss": 0.0032,
+      "step": 5612
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 9.374061091637457e-05,
+      "loss": 0.0043,
+      "step": 5616
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 9.380737773326657e-05,
+      "loss": 0.0035,
+      "step": 5620
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 9.387414455015858e-05,
+      "loss": 0.0032,
+      "step": 5624
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 9.394091136705058e-05,
+      "loss": 0.0015,
+      "step": 5628
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 9.400767818394258e-05,
+      "loss": 0.0036,
+      "step": 5632
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 9.407444500083458e-05,
+      "loss": 0.0022,
+      "step": 5636
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 9.41412118177266e-05,
+      "loss": 0.0055,
+      "step": 5640
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 9.42079786346186e-05,
+      "loss": 0.0033,
+      "step": 5644
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 9.42747454515106e-05,
+      "loss": 0.0061,
+      "step": 5648
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 9.434151226840261e-05,
+      "loss": 0.0037,
+      "step": 5652
+    },
+    {
+      "epoch": 24.17,
+      "learning_rate": 9.440827908529461e-05,
+      "loss": 0.004,
+      "step": 5656
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 9.447504590218663e-05,
+      "loss": 0.0035,
+      "step": 5660
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 9.454181271907863e-05,
+      "loss": 0.005,
+      "step": 5664
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 9.460857953597063e-05,
+      "loss": 0.002,
+      "step": 5668
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 9.467534635286263e-05,
+      "loss": 0.0067,
+      "step": 5672
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 9.474211316975464e-05,
+      "loss": 0.005,
+      "step": 5676
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 9.480887998664664e-05,
+      "loss": 0.0032,
+      "step": 5680
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 9.487564680353864e-05,
+      "loss": 0.0039,
+      "step": 5684
+    },
+    {
+      "epoch": 24.31,
+      "learning_rate": 9.494241362043064e-05,
+      "loss": 0.0032,
+      "step": 5688
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 9.500918043732265e-05,
+      "loss": 0.0042,
+      "step": 5692
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 9.507594725421466e-05,
+      "loss": 0.0044,
+      "step": 5696
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 9.514271407110667e-05,
+      "loss": 0.0024,
+      "step": 5700
+    },
+    {
+      "epoch": 24.38,
+      "learning_rate": 9.520948088799867e-05,
+      "loss": 0.005,
+      "step": 5704
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 9.527624770489067e-05,
+      "loss": 0.0025,
+      "step": 5708
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 9.534301452178269e-05,
+      "loss": 0.0047,
+      "step": 5712
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 9.540978133867469e-05,
+      "loss": 0.003,
+      "step": 5716
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 9.547654815556669e-05,
+      "loss": 0.0025,
+      "step": 5720
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 9.55433149724587e-05,
+      "loss": 0.0034,
+      "step": 5724
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 9.56100817893507e-05,
+      "loss": 0.004,
+      "step": 5728
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 9.56768486062427e-05,
+      "loss": 0.0024,
+      "step": 5732
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 9.57436154231347e-05,
+      "loss": 0.0027,
+      "step": 5736
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 9.58103822400267e-05,
+      "loss": 0.0026,
+      "step": 5740
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 9.587714905691871e-05,
+      "loss": 0.0017,
+      "step": 5744
+    },
+    {
+      "epoch": 24.56,
+      "learning_rate": 9.594391587381072e-05,
+      "loss": 0.0067,
+      "step": 5748
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 9.601068269070273e-05,
+      "loss": 0.0032,
+      "step": 5752
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 9.607744950759473e-05,
+      "loss": 0.0078,
+      "step": 5756
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 9.614421632448673e-05,
+      "loss": 0.0082,
+      "step": 5760
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 9.621098314137873e-05,
+      "loss": 0.0049,
+      "step": 5764
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 9.627774995827075e-05,
+      "loss": 0.0036,
+      "step": 5768
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 9.634451677516275e-05,
+      "loss": 0.0045,
+      "step": 5772
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 9.641128359205476e-05,
+      "loss": 0.0068,
+      "step": 5776
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 9.647805040894676e-05,
+      "loss": 0.007,
+      "step": 5780
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 9.654481722583876e-05,
+      "loss": 0.0026,
+      "step": 5784
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 9.661158404273076e-05,
+      "loss": 0.0066,
+      "step": 5788
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 9.667835085962277e-05,
+      "loss": 0.002,
+      "step": 5792
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 9.674511767651477e-05,
+      "loss": 0.0017,
+      "step": 5796
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 9.681188449340679e-05,
+      "loss": 0.0021,
+      "step": 5800
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 9.687865131029879e-05,
+      "loss": 0.0039,
+      "step": 5804
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 9.694541812719079e-05,
+      "loss": 0.0055,
+      "step": 5808
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 9.701218494408279e-05,
+      "loss": 0.0042,
+      "step": 5812
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 9.70789517609748e-05,
+      "loss": 0.0037,
+      "step": 5816
+    },
+    {
+      "epoch": 24.87,
+      "learning_rate": 9.714571857786681e-05,
+      "loss": 0.0039,
+      "step": 5820
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 9.721248539475881e-05,
+      "loss": 0.0026,
+      "step": 5824
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 9.727925221165082e-05,
+      "loss": 0.0043,
+      "step": 5828
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 9.734601902854282e-05,
+      "loss": 0.0046,
+      "step": 5832
+    },
+    {
+      "epoch": 24.94,
+      "learning_rate": 9.741278584543482e-05,
+      "loss": 0.0033,
+      "step": 5836
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 9.747955266232682e-05,
+      "loss": 0.004,
+      "step": 5840
+    },
+    {
+      "epoch": 24.97,
+      "learning_rate": 9.754631947921883e-05,
+      "loss": 0.0051,
+      "step": 5844
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 9.761308629611083e-05,
+      "loss": 0.0041,
+      "step": 5848
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 9.767985311300283e-05,
+      "loss": 0.0039,
+      "step": 5852
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 9.774661992989485e-05,
+      "loss": 0.0039,
+      "step": 5856
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 9.781338674678685e-05,
+      "loss": 0.0026,
+      "step": 5860
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 9.788015356367885e-05,
+      "loss": 0.0023,
+      "step": 5864
+    },
+    {
+      "epoch": 25.08,
+      "learning_rate": 9.794692038057086e-05,
+      "loss": 0.0033,
+      "step": 5868
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 9.801368719746287e-05,
+      "loss": 0.0032,
+      "step": 5872
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 9.808045401435488e-05,
+      "loss": 0.0048,
+      "step": 5876
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 9.814722083124688e-05,
+      "loss": 0.0074,
+      "step": 5880
+    },
+    {
+      "epoch": 25.15,
+      "learning_rate": 9.821398764813888e-05,
+      "loss": 0.0021,
+      "step": 5884
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 9.828075446503088e-05,
+      "loss": 0.0019,
+      "step": 5888
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 9.83475212819229e-05,
+      "loss": 0.0014,
+      "step": 5892
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 9.841428809881489e-05,
+      "loss": 0.0038,
+      "step": 5896
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 9.848105491570689e-05,
+      "loss": 0.0046,
+      "step": 5900
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 9.85478217325989e-05,
+      "loss": 0.0021,
+      "step": 5904
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 9.861458854949091e-05,
+      "loss": 0.0037,
+      "step": 5908
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 9.868135536638291e-05,
+      "loss": 0.0019,
+      "step": 5912
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 9.874812218327492e-05,
+      "loss": 0.0013,
+      "step": 5916
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 9.881488900016692e-05,
+      "loss": 0.0026,
+      "step": 5920
+    },
+    {
+      "epoch": 25.32,
+      "learning_rate": 9.888165581705892e-05,
+      "loss": 0.0034,
+      "step": 5924
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 9.894842263395094e-05,
+      "loss": 0.0028,
+      "step": 5928
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 9.901518945084294e-05,
+      "loss": 0.0069,
+      "step": 5932
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 9.908195626773494e-05,
+      "loss": 0.0031,
+      "step": 5936
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 9.914872308462694e-05,
+      "loss": 0.0047,
+      "step": 5940
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 9.921548990151896e-05,
+      "loss": 0.0029,
+      "step": 5944
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 9.928225671841096e-05,
+      "loss": 0.0026,
+      "step": 5948
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 9.934902353530295e-05,
+      "loss": 0.002,
+      "step": 5952
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 9.941579035219495e-05,
+      "loss": 0.0019,
+      "step": 5956
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 9.948255716908697e-05,
+      "loss": 0.0021,
+      "step": 5960
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 9.954932398597897e-05,
+      "loss": 0.0015,
+      "step": 5964
+    },
+    {
+      "epoch": 25.5,
+      "learning_rate": 9.961609080287098e-05,
+      "loss": 0.0042,
+      "step": 5968
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 9.968285761976298e-05,
+      "loss": 0.0036,
+      "step": 5972
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 9.974962443665498e-05,
+      "loss": 0.005,
+      "step": 5976
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 9.9816391253547e-05,
+      "loss": 0.0019,
+      "step": 5980
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 9.9883158070439e-05,
+      "loss": 0.0046,
+      "step": 5984
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 9.9949924887331e-05,
+      "loss": 0.0044,
+      "step": 5988
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 9.999999991511074e-05,
+      "loss": 0.002,
+      "step": 5992
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 9.999999787776848e-05,
+      "loss": 0.004,
+      "step": 5996
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 9.999999312396997e-05,
+      "loss": 0.0056,
+      "step": 6000
+    },
+    {
+      "epoch": 25.64,
+      "eval_exact_match": 0.5,
+      "eval_loss": 0.6754465699195862,
+      "eval_runtime": 141.2031,
+      "eval_samples_per_second": 6.813,
+      "step": 6000
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 9.999998565371547e-05,
+      "loss": 0.0071,
+      "step": 6004
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 9.999997546700539e-05,
+      "loss": 0.0036,
+      "step": 6008
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 9.999996256384028e-05,
+      "loss": 0.0021,
+      "step": 6012
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 9.999994694422086e-05,
+      "loss": 0.0037,
+      "step": 6016
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 9.999992860814793e-05,
+      "loss": 0.0028,
+      "step": 6020
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 9.999990755562253e-05,
+      "loss": 0.0073,
+      "step": 6024
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 9.999988378664581e-05,
+      "loss": 0.0037,
+      "step": 6028
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 9.999985730121903e-05,
+      "loss": 0.0035,
+      "step": 6032
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 9.999982809934364e-05,
+      "loss": 0.004,
+      "step": 6036
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 9.999979618102123e-05,
+      "loss": 0.0018,
+      "step": 6040
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 9.999976154625354e-05,
+      "loss": 0.0028,
+      "step": 6044
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 9.999972419504244e-05,
+      "loss": 0.003,
+      "step": 6048
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 9.999968412738995e-05,
+      "loss": 0.0025,
+      "step": 6052
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 9.999964134329827e-05,
+      "loss": 0.0024,
+      "step": 6056
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 9.999959584276973e-05,
+      "loss": 0.0038,
+      "step": 6060
+    },
+    {
+      "epoch": 25.91,
+      "learning_rate": 9.999954762580675e-05,
+      "loss": 0.0027,
+      "step": 6064
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 9.999949669241202e-05,
+      "loss": 0.0019,
+      "step": 6068
+    },
+    {
+      "epoch": 25.95,
+      "learning_rate": 9.999944304258826e-05,
+      "loss": 0.0034,
+      "step": 6072
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 9.999938667633839e-05,
+      "loss": 0.0026,
+      "step": 6076
+    },
+    {
+      "epoch": 25.98,
+      "learning_rate": 9.999932759366551e-05,
+      "loss": 0.0041,
+      "step": 6080
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 9.999926579457277e-05,
+      "loss": 0.0026,
+      "step": 6084
+    },
+    {
+      "epoch": 26.02,
+      "learning_rate": 9.999920127906357e-05,
+      "loss": 0.0028,
+      "step": 6088
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 9.999913404714139e-05,
+      "loss": 0.0047,
+      "step": 6092
+    },
+    {
+      "epoch": 26.05,
+      "learning_rate": 9.999906409880991e-05,
+      "loss": 0.0028,
+      "step": 6096
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 9.999899143407291e-05,
+      "loss": 0.0046,
+      "step": 6100
+    },
+    {
+      "epoch": 26.09,
+      "learning_rate": 9.999891605293434e-05,
+      "loss": 0.0027,
+      "step": 6104
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 9.999883795539832e-05,
+      "loss": 0.0024,
+      "step": 6108
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 9.999875714146904e-05,
+      "loss": 0.0074,
+      "step": 6112
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 9.999867361115096e-05,
+      "loss": 0.003,
+      "step": 6116
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 9.999858736444856e-05,
+      "loss": 0.0034,
+      "step": 6120
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 9.999849840136655e-05,
+      "loss": 0.0035,
+      "step": 6124
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 9.999840672190975e-05,
+      "loss": 0.0023,
+      "step": 6128
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 9.999831232608316e-05,
+      "loss": 0.0037,
+      "step": 6132
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 9.99982152138919e-05,
+      "loss": 0.0031,
+      "step": 6136
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 9.999811538534123e-05,
+      "loss": 0.0053,
+      "step": 6140
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 9.99980128404366e-05,
+      "loss": 0.0042,
+      "step": 6144
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 9.999790757918357e-05,
+      "loss": 0.0031,
+      "step": 6148
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 9.999779960158786e-05,
+      "loss": 0.0015,
+      "step": 6152
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 9.999768890765533e-05,
+      "loss": 0.0042,
+      "step": 6156
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 9.999757549739201e-05,
+      "loss": 0.0029,
+      "step": 6160
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 9.999745937080403e-05,
+      "loss": 0.0037,
+      "step": 6164
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 9.999734052789775e-05,
+      "loss": 0.0027,
+      "step": 6168
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 9.999721896867957e-05,
+      "loss": 0.0019,
+      "step": 6172
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 9.999709469315612e-05,
+      "loss": 0.0031,
+      "step": 6176
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 9.999696770133417e-05,
+      "loss": 0.0037,
+      "step": 6180
+    },
+    {
+      "epoch": 26.43,
+      "learning_rate": 9.999683799322059e-05,
+      "loss": 0.0018,
+      "step": 6184
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 9.999670556882244e-05,
+      "loss": 0.0025,
+      "step": 6188
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 9.99965704281469e-05,
+      "loss": 0.0022,
+      "step": 6192
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 9.999643257120135e-05,
+      "loss": 0.0022,
+      "step": 6196
+    },
+    {
+      "epoch": 26.5,
+      "learning_rate": 9.999629199799323e-05,
+      "loss": 0.0032,
+      "step": 6200
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 9.999614870853023e-05,
+      "loss": 0.0041,
+      "step": 6204
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 9.999600270282007e-05,
+      "loss": 0.005,
+      "step": 6208
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 9.999585398087074e-05,
+      "loss": 0.0037,
+      "step": 6212
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 9.99957025426903e-05,
+      "loss": 0.005,
+      "step": 6216
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 9.999554838828698e-05,
+      "loss": 0.0039,
+      "step": 6220
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 9.999539151766915e-05,
+      "loss": 0.003,
+      "step": 6224
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 9.999523193084532e-05,
+      "loss": 0.002,
+      "step": 6228
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 9.999506962782418e-05,
+      "loss": 0.0018,
+      "step": 6232
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 9.999490460861454e-05,
+      "loss": 0.0031,
+      "step": 6236
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 9.999473687322538e-05,
+      "loss": 0.0029,
+      "step": 6240
+    },
+    {
+      "epoch": 26.68,
+      "learning_rate": 9.999456642166577e-05,
+      "loss": 0.0034,
+      "step": 6244
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 9.999439325394504e-05,
+      "loss": 0.002,
+      "step": 6248
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 9.999421737007254e-05,
+      "loss": 0.0023,
+      "step": 6252
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 9.999403877005783e-05,
+      "loss": 0.003,
+      "step": 6256
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 9.999385745391065e-05,
+      "loss": 0.0034,
+      "step": 6260
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 9.999367342164083e-05,
+      "loss": 0.0033,
+      "step": 6264
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 9.999348667325836e-05,
+      "loss": 0.0037,
+      "step": 6268
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 9.99932972087734e-05,
+      "loss": 0.0022,
+      "step": 6272
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 9.999310502819623e-05,
+      "loss": 0.0036,
+      "step": 6276
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 9.999291013153729e-05,
+      "loss": 0.0065,
+      "step": 6280
+    },
+    {
+      "epoch": 26.85,
+      "learning_rate": 9.999271251880718e-05,
+      "loss": 0.0047,
+      "step": 6284
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 9.999251219001662e-05,
+      "loss": 0.0015,
+      "step": 6288
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 9.999230914517653e-05,
+      "loss": 0.0022,
+      "step": 6292
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 9.99921033842979e-05,
+      "loss": 0.0016,
+      "step": 6296
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 9.999189490739191e-05,
+      "loss": 0.0035,
+      "step": 6300
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 9.999168371446993e-05,
+      "loss": 0.0022,
+      "step": 6304
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 9.99914698055434e-05,
+      "loss": 0.0029,
+      "step": 6308
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 9.999125318062394e-05,
+      "loss": 0.004,
+      "step": 6312
+    },
+    {
+      "epoch": 26.99,
+      "learning_rate": 9.999103383972333e-05,
+      "loss": 0.0027,
+      "step": 6316
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 9.999081178285347e-05,
+      "loss": 0.0031,
+      "step": 6320
+    },
+    {
+      "epoch": 27.03,
+      "learning_rate": 9.999058701002645e-05,
+      "loss": 0.0044,
+      "step": 6324
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 9.999035952125448e-05,
+      "loss": 0.0014,
+      "step": 6328
+    },
+    {
+      "epoch": 27.06,
+      "learning_rate": 9.99901293165499e-05,
+      "loss": 0.0071,
+      "step": 6332
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 9.998989639592521e-05,
+      "loss": 0.0028,
+      "step": 6336
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 9.998966075939309e-05,
+      "loss": 0.0018,
+      "step": 6340
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 9.998942240696633e-05,
+      "loss": 0.0015,
+      "step": 6344
+    },
+    {
+      "epoch": 27.13,
+      "learning_rate": 9.998918133865788e-05,
+      "loss": 0.0018,
+      "step": 6348
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 9.998893755448085e-05,
+      "loss": 0.0031,
+      "step": 6352
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 9.998869105444845e-05,
+      "loss": 0.0053,
+      "step": 6356
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 9.998844183857412e-05,
+      "loss": 0.0033,
+      "step": 6360
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 9.998818990687135e-05,
+      "loss": 0.0021,
+      "step": 6364
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 9.998793525935387e-05,
+      "loss": 0.002,
+      "step": 6368
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 9.998767789603547e-05,
+      "loss": 0.0011,
+      "step": 6372
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 9.998741781693017e-05,
+      "loss": 0.0034,
+      "step": 6376
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 9.99871550220521e-05,
+      "loss": 0.0034,
+      "step": 6380
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 9.998688951141552e-05,
+      "loss": 0.0018,
+      "step": 6384
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 9.998662128503487e-05,
+      "loss": 0.0038,
+      "step": 6388
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 9.998635034292469e-05,
+      "loss": 0.0009,
+      "step": 6392
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 9.998607668509975e-05,
+      "loss": 0.0019,
+      "step": 6396
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 9.998580031157486e-05,
+      "loss": 0.0036,
+      "step": 6400
+    },
+    {
+      "epoch": 27.37,
+      "learning_rate": 9.998552122236509e-05,
+      "loss": 0.0019,
+      "step": 6404
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 9.998523941748556e-05,
+      "loss": 0.0038,
+      "step": 6408
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 9.998495489695161e-05,
+      "loss": 0.0042,
+      "step": 6412
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 9.998466766077868e-05,
+      "loss": 0.0027,
+      "step": 6416
+    },
+    {
+      "epoch": 27.44,
+      "learning_rate": 9.998437770898239e-05,
+      "loss": 0.0026,
+      "step": 6420
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 9.998408504157849e-05,
+      "loss": 0.0027,
+      "step": 6424
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 9.998378965858286e-05,
+      "loss": 0.0043,
+      "step": 6428
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 9.998349156001156e-05,
+      "loss": 0.0031,
+      "step": 6432
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 9.998319074588081e-05,
+      "loss": 0.0018,
+      "step": 6436
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 9.998288721620692e-05,
+      "loss": 0.0033,
+      "step": 6440
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 9.998258097100639e-05,
+      "loss": 0.0009,
+      "step": 6444
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 9.998227201029586e-05,
+      "loss": 0.0035,
+      "step": 6448
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 9.998196033409212e-05,
+      "loss": 0.0034,
+      "step": 6452
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 9.998164594241208e-05,
+      "loss": 0.0023,
+      "step": 6456
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 9.998132883527287e-05,
+      "loss": 0.0042,
+      "step": 6460
+    },
+    {
+      "epoch": 27.62,
+      "learning_rate": 9.998100901269167e-05,
+      "loss": 0.0011,
+      "step": 6464
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 9.998068647468587e-05,
+      "loss": 0.0047,
+      "step": 6468
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 9.9980361221273e-05,
+      "loss": 0.0023,
+      "step": 6472
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 9.998003325247071e-05,
+      "loss": 0.0025,
+      "step": 6476
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 9.997970256829687e-05,
+      "loss": 0.003,
+      "step": 6480
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 9.997936916876939e-05,
+      "loss": 0.0016,
+      "step": 6484
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 9.997903305390641e-05,
+      "loss": 0.0027,
+      "step": 6488
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 9.997869422372618e-05,
+      "loss": 0.0038,
+      "step": 6492
+    },
+    {
+      "epoch": 27.76,
+      "learning_rate": 9.997835267824712e-05,
+      "loss": 0.0047,
+      "step": 6496
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 9.997800841748777e-05,
+      "loss": 0.002,
+      "step": 6500
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 9.997766144146685e-05,
+      "loss": 0.0065,
+      "step": 6504
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 9.997731175020319e-05,
+      "loss": 0.0062,
+      "step": 6508
+    },
+    {
+      "epoch": 27.83,
+      "learning_rate": 9.997695934371583e-05,
+      "loss": 0.0043,
+      "step": 6512
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 9.997660422202386e-05,
+      "loss": 0.006,
+      "step": 6516
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 9.997624638514661e-05,
+      "loss": 0.0015,
+      "step": 6520
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 9.997588583310352e-05,
+      "loss": 0.0029,
+      "step": 6524
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 9.997552256591416e-05,
+      "loss": 0.0052,
+      "step": 6528
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 9.997515658359828e-05,
+      "loss": 0.0014,
+      "step": 6532
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 9.997478788617576e-05,
+      "loss": 0.0018,
+      "step": 6536
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 9.997441647366663e-05,
+      "loss": 0.0025,
+      "step": 6540
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 9.997404234609107e-05,
+      "loss": 0.0022,
+      "step": 6544
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 9.997366550346943e-05,
+      "loss": 0.0025,
+      "step": 6548
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 9.997328594582213e-05,
+      "loss": 0.0033,
+      "step": 6552
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 9.997290367316982e-05,
+      "loss": 0.0018,
+      "step": 6556
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 9.997251868553328e-05,
+      "loss": 0.0025,
+      "step": 6560
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 9.997213098293341e-05,
+      "loss": 0.0038,
+      "step": 6564
+    },
+    {
+      "epoch": 28.07,
+      "learning_rate": 9.997174056539129e-05,
+      "loss": 0.0038,
+      "step": 6568
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 9.997134743292811e-05,
+      "loss": 0.0027,
+      "step": 6572
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 9.997095158556525e-05,
+      "loss": 0.0028,
+      "step": 6576
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 9.997055302332419e-05,
+      "loss": 0.003,
+      "step": 6580
+    },
+    {
+      "epoch": 28.14,
+      "learning_rate": 9.99701517462266e-05,
+      "loss": 0.0031,
+      "step": 6584
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 9.996974775429427e-05,
+      "loss": 0.0022,
+      "step": 6588
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 9.996934104754916e-05,
+      "loss": 0.0034,
+      "step": 6592
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 9.996893162601337e-05,
+      "loss": 0.0033,
+      "step": 6596
+    },
+    {
+      "epoch": 28.21,
+      "learning_rate": 9.996851948970915e-05,
+      "loss": 0.0024,
+      "step": 6600
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 9.996810463865886e-05,
+      "loss": 0.0016,
+      "step": 6604
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 9.996768707288506e-05,
+      "loss": 0.0049,
+      "step": 6608
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 9.996726679241043e-05,
+      "loss": 0.0062,
+      "step": 6612
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 9.99668437972578e-05,
+      "loss": 0.0058,
+      "step": 6616
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 9.996641808745017e-05,
+      "loss": 0.0044,
+      "step": 6620
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 9.996598966301065e-05,
+      "loss": 0.0026,
+      "step": 6624
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 9.996555852396252e-05,
+      "loss": 0.002,
+      "step": 6628
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 9.996512467032922e-05,
+      "loss": 0.0022,
+      "step": 6632
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 9.996468810213428e-05,
+      "loss": 0.0025,
+      "step": 6636
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 9.996424881940146e-05,
+      "loss": 0.0037,
+      "step": 6640
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 9.996380682215461e-05,
+      "loss": 0.0019,
+      "step": 6644
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 9.996336211041773e-05,
+      "loss": 0.0028,
+      "step": 6648
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 9.996291468421501e-05,
+      "loss": 0.0028,
+      "step": 6652
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 9.996246454357073e-05,
+      "loss": 0.0025,
+      "step": 6656
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 9.996201168850936e-05,
+      "loss": 0.0007,
+      "step": 6660
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 9.996155611905551e-05,
+      "loss": 0.002,
+      "step": 6664
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 9.99610978352339e-05,
+      "loss": 0.0035,
+      "step": 6668
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 9.996063683706946e-05,
+      "loss": 0.0032,
+      "step": 6672
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 9.996017312458724e-05,
+      "loss": 0.0017,
+      "step": 6676
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 9.995970669781241e-05,
+      "loss": 0.0016,
+      "step": 6680
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 9.995923755677033e-05,
+      "loss": 0.0049,
+      "step": 6684
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 9.995876570148646e-05,
+      "loss": 0.0024,
+      "step": 6688
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 9.995829113198645e-05,
+      "loss": 0.0041,
+      "step": 6692
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 9.995781384829609e-05,
+      "loss": 0.0042,
+      "step": 6696
+    },
+    {
+      "epoch": 28.63,
+      "learning_rate": 9.99573338504413e-05,
+      "loss": 0.0036,
+      "step": 6700
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 9.995685113844818e-05,
+      "loss": 0.0018,
+      "step": 6704
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 9.995636571234293e-05,
+      "loss": 0.0019,
+      "step": 6708
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 9.995587757215194e-05,
+      "loss": 0.0012,
+      "step": 6712
+    },
+    {
+      "epoch": 28.7,
+      "learning_rate": 9.995538671790171e-05,
+      "loss": 0.0021,
+      "step": 6716
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 9.995489314961892e-05,
+      "loss": 0.0041,
+      "step": 6720
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 9.995439686733039e-05,
+      "loss": 0.0026,
+      "step": 6724
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 9.995389787106309e-05,
+      "loss": 0.0012,
+      "step": 6728
+    },
+    {
+      "epoch": 28.77,
+      "learning_rate": 9.995339616084409e-05,
+      "loss": 0.0023,
+      "step": 6732
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 9.995289173670071e-05,
+      "loss": 0.002,
+      "step": 6736
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 9.995238459866029e-05,
+      "loss": 0.0035,
+      "step": 6740
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 9.995187474675042e-05,
+      "loss": 0.0012,
+      "step": 6744
+    },
+    {
+      "epoch": 28.84,
+      "learning_rate": 9.995136218099878e-05,
+      "loss": 0.0016,
+      "step": 6748
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 9.995084690143326e-05,
+      "loss": 0.0047,
+      "step": 6752
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 9.99503289080818e-05,
+      "loss": 0.0015,
+      "step": 6756
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 9.994980820097258e-05,
+      "loss": 0.0055,
+      "step": 6760
+    },
+    {
+      "epoch": 28.91,
+      "learning_rate": 9.994928478013387e-05,
+      "loss": 0.0013,
+      "step": 6764
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 9.994875864559413e-05,
+      "loss": 0.0028,
+      "step": 6768
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 9.99482297973819e-05,
+      "loss": 0.0032,
+      "step": 6772
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 9.994769823552594e-05,
+      "loss": 0.0039,
+      "step": 6776
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 9.994716396005515e-05,
+      "loss": 0.002,
+      "step": 6780
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 9.994662697099851e-05,
+      "loss": 0.0041,
+      "step": 6784
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 9.994608726838523e-05,
+      "loss": 0.0013,
+      "step": 6788
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 9.994554485224463e-05,
+      "loss": 0.001,
+      "step": 6792
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 9.994499972260616e-05,
+      "loss": 0.0017,
+      "step": 6796
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 9.994445187949944e-05,
+      "loss": 0.0041,
+      "step": 6800
+    },
+    {
+      "epoch": 29.08,
+      "learning_rate": 9.994390132295426e-05,
+      "loss": 0.0044,
+      "step": 6804
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 9.99433480530005e-05,
+      "loss": 0.0037,
+      "step": 6808
+    },
+    {
+      "epoch": 29.11,
+      "learning_rate": 9.994279206966824e-05,
+      "loss": 0.0013,
+      "step": 6812
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 9.994223337298764e-05,
+      "loss": 0.0019,
+      "step": 6816
+    },
+    {
+      "epoch": 29.15,
+      "learning_rate": 9.994167196298914e-05,
+      "loss": 0.0017,
+      "step": 6820
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 9.994110783970316e-05,
+      "loss": 0.0024,
+      "step": 6824
+    },
+    {
+      "epoch": 29.18,
+      "learning_rate": 9.994054100316038e-05,
+      "loss": 0.0036,
+      "step": 6828
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 9.99399714533916e-05,
+      "loss": 0.0047,
+      "step": 6832
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 9.993939919042776e-05,
+      "loss": 0.0036,
+      "step": 6836
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 9.993882421429994e-05,
+      "loss": 0.0024,
+      "step": 6840
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 9.993824652503939e-05,
+      "loss": 0.0018,
+      "step": 6844
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 9.99376661226775e-05,
+      "loss": 0.0035,
+      "step": 6848
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 9.993708300724577e-05,
+      "loss": 0.0021,
+      "step": 6852
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 9.993649717877593e-05,
+      "loss": 0.0027,
+      "step": 6856
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 9.993590863729978e-05,
+      "loss": 0.0031,
+      "step": 6860
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 9.993531738284929e-05,
+      "loss": 0.0026,
+      "step": 6864
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 9.99347234154566e-05,
+      "loss": 0.0019,
+      "step": 6868
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 9.993412673515396e-05,
+      "loss": 0.0034,
+      "step": 6872
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 9.993352734197379e-05,
+      "loss": 0.0022,
+      "step": 6876
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 9.993292523594867e-05,
+      "loss": 0.0054,
+      "step": 6880
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 9.993232041711129e-05,
+      "loss": 0.0025,
+      "step": 6884
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 9.993171288549454e-05,
+      "loss": 0.004,
+      "step": 6888
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 9.99311026411314e-05,
+      "loss": 0.003,
+      "step": 6892
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 9.993048968405502e-05,
+      "loss": 0.0017,
+      "step": 6896
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 9.992987401429873e-05,
+      "loss": 0.0017,
+      "step": 6900
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 9.992925563189595e-05,
+      "loss": 0.0028,
+      "step": 6904
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 9.992863453688031e-05,
+      "loss": 0.0049,
+      "step": 6908
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 9.992801072928551e-05,
+      "loss": 0.0042,
+      "step": 6912
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 9.992738420914546e-05,
+      "loss": 0.0019,
+      "step": 6916
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 9.992675497649424e-05,
+      "loss": 0.0044,
+      "step": 6920
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 9.992612303136597e-05,
+      "loss": 0.0049,
+      "step": 6924
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 9.992548837379501e-05,
+      "loss": 0.0028,
+      "step": 6928
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 9.992485100381584e-05,
+      "loss": 0.0036,
+      "step": 6932
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 9.992421092146309e-05,
+      "loss": 0.003,
+      "step": 6936
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 9.992356812677152e-05,
+      "loss": 0.0064,
+      "step": 6940
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 9.992292261977609e-05,
+      "loss": 0.0013,
+      "step": 6944
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 9.992227440051184e-05,
+      "loss": 0.0043,
+      "step": 6948
+    },
+    {
+      "epoch": 29.71,
+      "learning_rate": 9.992162346901399e-05,
+      "loss": 0.0018,
+      "step": 6952
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 9.99209698253179e-05,
+      "loss": 0.0048,
+      "step": 6956
+    },
+    {
+      "epoch": 29.74,
+      "learning_rate": 9.992031346945911e-05,
+      "loss": 0.0025,
+      "step": 6960
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 9.991965440147324e-05,
+      "loss": 0.0017,
+      "step": 6964
+    },
+    {
+      "epoch": 29.78,
+      "learning_rate": 9.991899262139612e-05,
+      "loss": 0.0022,
+      "step": 6968
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 9.991832812926371e-05,
+      "loss": 0.0033,
+      "step": 6972
+    },
+    {
+      "epoch": 29.81,
+      "learning_rate": 9.99176609251121e-05,
+      "loss": 0.0036,
+      "step": 6976
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 9.991699100897753e-05,
+      "loss": 0.0047,
+      "step": 6980
+    },
+    {
+      "epoch": 29.85,
+      "learning_rate": 9.99163183808964e-05,
+      "loss": 0.0021,
+      "step": 6984
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 9.991564304090525e-05,
+      "loss": 0.002,
+      "step": 6988
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 9.99149649890408e-05,
+      "loss": 0.0044,
+      "step": 6992
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 9.991428422533987e-05,
+      "loss": 0.0025,
+      "step": 6996
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 9.991360074983943e-05,
+      "loss": 0.006,
+      "step": 7000
+    },
+    {
+      "epoch": 29.91,
+      "eval_exact_match": 0.502079002079002,
+      "eval_loss": 0.7090530395507812,
+      "eval_runtime": 143.1626,
+      "eval_samples_per_second": 6.72,
+      "step": 7000
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 9.991291456257663e-05,
+      "loss": 0.0016,
+      "step": 7004
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 9.991222566358874e-05,
+      "loss": 0.0021,
+      "step": 7008
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 9.99115340529132e-05,
+      "loss": 0.0012,
+      "step": 7012
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 9.991083973058757e-05,
+      "loss": 0.0042,
+      "step": 7016
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 9.991014269664958e-05,
+      "loss": 0.0028,
+      "step": 7020
+    },
+    {
+      "epoch": 30.02,
+      "learning_rate": 9.990944295113711e-05,
+      "loss": 0.0038,
+      "step": 7024
+    },
+    {
+      "epoch": 30.03,
+      "learning_rate": 9.990874049408815e-05,
+      "loss": 0.0027,
+      "step": 7028
+    },
+    {
+      "epoch": 30.05,
+      "learning_rate": 9.990803532554087e-05,
+      "loss": 0.0031,
+      "step": 7032
+    },
+    {
+      "epoch": 30.07,
+      "learning_rate": 9.990732744553361e-05,
+      "loss": 0.0041,
+      "step": 7036
+    },
+    {
+      "epoch": 30.09,
+      "learning_rate": 9.990661685410481e-05,
+      "loss": 0.002,
+      "step": 7040
+    },
+    {
+      "epoch": 30.1,
+      "learning_rate": 9.990590355129306e-05,
+      "loss": 0.0017,
+      "step": 7044
+    },
+    {
+      "epoch": 30.12,
+      "learning_rate": 9.990518753713712e-05,
+      "loss": 0.0021,
+      "step": 7048
+    },
+    {
+      "epoch": 30.14,
+      "learning_rate": 9.990446881167592e-05,
+      "loss": 0.0021,
+      "step": 7052
+    },
+    {
+      "epoch": 30.15,
+      "learning_rate": 9.990374737494848e-05,
+      "loss": 0.0017,
+      "step": 7056
+    },
+    {
+      "epoch": 30.17,
+      "learning_rate": 9.990302322699399e-05,
+      "loss": 0.0041,
+      "step": 7060
+    },
+    {
+      "epoch": 30.19,
+      "learning_rate": 9.99022963678518e-05,
+      "loss": 0.0013,
+      "step": 7064
+    },
+    {
+      "epoch": 30.21,
+      "learning_rate": 9.99015667975614e-05,
+      "loss": 0.0035,
+      "step": 7068
+    },
+    {
+      "epoch": 30.22,
+      "learning_rate": 9.990083451616245e-05,
+      "loss": 0.0028,
+      "step": 7072
+    },
+    {
+      "epoch": 30.24,
+      "learning_rate": 9.990009952369471e-05,
+      "loss": 0.0013,
+      "step": 7076
+    },
+    {
+      "epoch": 30.26,
+      "learning_rate": 9.98993618201981e-05,
+      "loss": 0.0031,
+      "step": 7080
+    },
+    {
+      "epoch": 30.27,
+      "learning_rate": 9.98986214057127e-05,
+      "loss": 0.0015,
+      "step": 7084
+    },
+    {
+      "epoch": 30.29,
+      "learning_rate": 9.98978782802788e-05,
+      "loss": 0.004,
+      "step": 7088
+    },
+    {
+      "epoch": 30.31,
+      "learning_rate": 9.989713244393669e-05,
+      "loss": 0.002,
+      "step": 7092
+    },
+    {
+      "epoch": 30.32,
+      "learning_rate": 9.989638389672691e-05,
+      "loss": 0.0028,
+      "step": 7096
+    },
+    {
+      "epoch": 30.34,
+      "learning_rate": 9.989563263869016e-05,
+      "loss": 0.0037,
+      "step": 7100
+    },
+    {
+      "epoch": 30.36,
+      "learning_rate": 9.989487866986724e-05,
+      "loss": 0.0011,
+      "step": 7104
+    },
+    {
+      "epoch": 30.38,
+      "learning_rate": 9.98941219902991e-05,
+      "loss": 0.0018,
+      "step": 7108
+    },
+    {
+      "epoch": 30.39,
+      "learning_rate": 9.989336260002687e-05,
+      "loss": 0.0015,
+      "step": 7112
+    },
+    {
+      "epoch": 30.41,
+      "learning_rate": 9.989260049909179e-05,
+      "loss": 0.0031,
+      "step": 7116
+    },
+    {
+      "epoch": 30.43,
+      "learning_rate": 9.989183568753527e-05,
+      "loss": 0.0039,
+      "step": 7120
+    },
+    {
+      "epoch": 30.44,
+      "learning_rate": 9.989106816539885e-05,
+      "loss": 0.0032,
+      "step": 7124
+    },
+    {
+      "epoch": 30.46,
+      "learning_rate": 9.989029793272425e-05,
+      "loss": 0.0017,
+      "step": 7128
+    },
+    {
+      "epoch": 30.48,
+      "learning_rate": 9.988952498955331e-05,
+      "loss": 0.0011,
+      "step": 7132
+    },
+    {
+      "epoch": 30.5,
+      "learning_rate": 9.988874933592803e-05,
+      "loss": 0.0035,
+      "step": 7136
+    },
+    {
+      "epoch": 30.51,
+      "learning_rate": 9.988797097189052e-05,
+      "loss": 0.0041,
+      "step": 7140
+    },
+    {
+      "epoch": 30.53,
+      "learning_rate": 9.988718989748309e-05,
+      "loss": 0.0044,
+      "step": 7144
+    },
+    {
+      "epoch": 30.55,
+      "learning_rate": 9.988640611274819e-05,
+      "loss": 0.0022,
+      "step": 7148
+    },
+    {
+      "epoch": 30.56,
+      "learning_rate": 9.988561961772838e-05,
+      "loss": 0.0027,
+      "step": 7152
+    },
+    {
+      "epoch": 30.58,
+      "learning_rate": 9.988483041246639e-05,
+      "loss": 0.0018,
+      "step": 7156
+    },
+    {
+      "epoch": 30.6,
+      "learning_rate": 9.988403849700511e-05,
+      "loss": 0.0016,
+      "step": 7160
+    },
+    {
+      "epoch": 30.62,
+      "learning_rate": 9.988324387138754e-05,
+      "loss": 0.0048,
+      "step": 7164
+    },
+    {
+      "epoch": 30.63,
+      "learning_rate": 9.988244653565688e-05,
+      "loss": 0.001,
+      "step": 7168
+    },
+    {
+      "epoch": 30.65,
+      "learning_rate": 9.988164648985644e-05,
+      "loss": 0.004,
+      "step": 7172
+    },
+    {
+      "epoch": 30.67,
+      "learning_rate": 9.988084373402968e-05,
+      "loss": 0.002,
+      "step": 7176
+    },
+    {
+      "epoch": 30.68,
+      "learning_rate": 9.988003826822022e-05,
+      "loss": 0.0025,
+      "step": 7180
+    },
+    {
+      "epoch": 30.7,
+      "learning_rate": 9.987923009247179e-05,
+      "loss": 0.0029,
+      "step": 7184
+    },
+    {
+      "epoch": 30.72,
+      "learning_rate": 9.987841920682834e-05,
+      "loss": 0.0022,
+      "step": 7188
+    },
+    {
+      "epoch": 30.74,
+      "learning_rate": 9.987760561133391e-05,
+      "loss": 0.0023,
+      "step": 7192
+    },
+    {
+      "epoch": 30.75,
+      "learning_rate": 9.98767893060327e-05,
+      "loss": 0.0011,
+      "step": 7196
+    },
+    {
+      "epoch": 30.77,
+      "learning_rate": 9.987597029096907e-05,
+      "loss": 0.002,
+      "step": 7200
+    },
+    {
+      "epoch": 30.79,
+      "learning_rate": 9.987514856618748e-05,
+      "loss": 0.0014,
+      "step": 7204
+    },
+    {
+      "epoch": 30.8,
+      "learning_rate": 9.987432413173262e-05,
+      "loss": 0.0025,
+      "step": 7208
+    },
+    {
+      "epoch": 30.82,
+      "learning_rate": 9.987349698764923e-05,
+      "loss": 0.0033,
+      "step": 7212
+    },
+    {
+      "epoch": 30.84,
+      "learning_rate": 9.987266713398232e-05,
+      "loss": 0.0023,
+      "step": 7216
+    },
+    {
+      "epoch": 30.85,
+      "learning_rate": 9.987183457077689e-05,
+      "loss": 0.0026,
+      "step": 7220
+    },
+    {
+      "epoch": 30.87,
+      "learning_rate": 9.987099929807824e-05,
+      "loss": 0.0027,
+      "step": 7224
+    },
+    {
+      "epoch": 30.89,
+      "learning_rate": 9.987016131593173e-05,
+      "loss": 0.0017,
+      "step": 7228
+    },
+    {
+      "epoch": 30.91,
+      "learning_rate": 9.986932062438285e-05,
+      "loss": 0.0015,
+      "step": 7232
+    },
+    {
+      "epoch": 30.92,
+      "learning_rate": 9.986847722347734e-05,
+      "loss": 0.0025,
+      "step": 7236
+    },
+    {
+      "epoch": 30.94,
+      "learning_rate": 9.986763111326096e-05,
+      "loss": 0.0048,
+      "step": 7240
+    },
+    {
+      "epoch": 30.96,
+      "learning_rate": 9.986678229377973e-05,
+      "loss": 0.0034,
+      "step": 7244
+    },
+    {
+      "epoch": 30.97,
+      "learning_rate": 9.986593076507974e-05,
+      "loss": 0.004,
+      "step": 7248
+    },
+    {
+      "epoch": 30.99,
+      "learning_rate": 9.986507652720723e-05,
+      "loss": 0.0027,
+      "step": 7252
+    },
+    {
+      "epoch": 31.01,
+      "learning_rate": 9.986421958020863e-05,
+      "loss": 0.004,
+      "step": 7256
+    },
+    {
+      "epoch": 31.03,
+      "learning_rate": 9.986335992413054e-05,
+      "loss": 0.0015,
+      "step": 7260
+    },
+    {
+      "epoch": 31.04,
+      "learning_rate": 9.98624975590196e-05,
+      "loss": 0.0023,
+      "step": 7264
+    },
+    {
+      "epoch": 31.06,
+      "learning_rate": 9.986163248492268e-05,
+      "loss": 0.0063,
+      "step": 7268
+    },
+    {
+      "epoch": 31.08,
+      "learning_rate": 9.98607647018868e-05,
+      "loss": 0.0012,
+      "step": 7272
+    },
+    {
+      "epoch": 31.09,
+      "learning_rate": 9.98598942099591e-05,
+      "loss": 0.0052,
+      "step": 7276
+    },
+    {
+      "epoch": 31.11,
+      "learning_rate": 9.985902100918684e-05,
+      "loss": 0.0011,
+      "step": 7280
+    },
+    {
+      "epoch": 31.13,
+      "learning_rate": 9.985814509961748e-05,
+      "loss": 0.0039,
+      "step": 7284
+    },
+    {
+      "epoch": 31.15,
+      "learning_rate": 9.985726648129863e-05,
+      "loss": 0.0085,
+      "step": 7288
+    },
+    {
+      "epoch": 31.16,
+      "learning_rate": 9.9856385154278e-05,
+      "loss": 0.0036,
+      "step": 7292
+    },
+    {
+      "epoch": 31.18,
+      "learning_rate": 9.98555011186035e-05,
+      "loss": 0.0015,
+      "step": 7296
+    },
+    {
+      "epoch": 31.2,
+      "learning_rate": 9.985461437432311e-05,
+      "loss": 0.0014,
+      "step": 7300
+    },
+    {
+      "epoch": 31.21,
+      "learning_rate": 9.985372492148504e-05,
+      "loss": 0.003,
+      "step": 7304
+    },
+    {
+      "epoch": 31.23,
+      "learning_rate": 9.985283276013761e-05,
+      "loss": 0.0022,
+      "step": 7308
+    },
+    {
+      "epoch": 31.25,
+      "learning_rate": 9.98519378903293e-05,
+      "loss": 0.0007,
+      "step": 7312
+    },
+    {
+      "epoch": 31.26,
+      "learning_rate": 9.985104031210869e-05,
+      "loss": 0.0033,
+      "step": 7316
+    },
+    {
+      "epoch": 31.28,
+      "learning_rate": 9.985014002552457e-05,
+      "loss": 0.0025,
+      "step": 7320
+    },
+    {
+      "epoch": 31.3,
+      "learning_rate": 9.984923703062588e-05,
+      "loss": 0.004,
+      "step": 7324
+    },
+    {
+      "epoch": 31.32,
+      "learning_rate": 9.984833132746163e-05,
+      "loss": 0.0019,
+      "step": 7328
+    },
+    {
+      "epoch": 31.33,
+      "learning_rate": 9.984742291608104e-05,
+      "loss": 0.002,
+      "step": 7332
+    },
+    {
+      "epoch": 31.35,
+      "learning_rate": 9.984651179653348e-05,
+      "loss": 0.0011,
+      "step": 7336
+    },
+    {
+      "epoch": 31.37,
+      "learning_rate": 9.984559796886844e-05,
+      "loss": 0.0012,
+      "step": 7340
+    },
+    {
+      "epoch": 31.38,
+      "learning_rate": 9.984468143313555e-05,
+      "loss": 0.0013,
+      "step": 7344
+    },
+    {
+      "epoch": 31.4,
+      "learning_rate": 9.984376218938463e-05,
+      "loss": 0.0024,
+      "step": 7348
+    },
+    {
+      "epoch": 31.42,
+      "learning_rate": 9.984284023766562e-05,
+      "loss": 0.0057,
+      "step": 7352
+    },
+    {
+      "epoch": 31.44,
+      "learning_rate": 9.98419155780286e-05,
+      "loss": 0.0017,
+      "step": 7356
+    },
+    {
+      "epoch": 31.45,
+      "learning_rate": 9.98409882105238e-05,
+      "loss": 0.0012,
+      "step": 7360
+    },
+    {
+      "epoch": 31.47,
+      "learning_rate": 9.984005813520162e-05,
+      "loss": 0.0014,
+      "step": 7364
+    },
+    {
+      "epoch": 31.49,
+      "learning_rate": 9.983912535211258e-05,
+      "loss": 0.0015,
+      "step": 7368
+    },
+    {
+      "epoch": 31.5,
+      "learning_rate": 9.983818986130736e-05,
+      "loss": 0.0021,
+      "step": 7372
+    },
+    {
+      "epoch": 31.52,
+      "learning_rate": 9.983725166283676e-05,
+      "loss": 0.001,
+      "step": 7376
+    },
+    {
+      "epoch": 31.54,
+      "learning_rate": 9.983631075675179e-05,
+      "loss": 0.0026,
+      "step": 7380
+    },
+    {
+      "epoch": 31.56,
+      "learning_rate": 9.983536714310355e-05,
+      "loss": 0.0015,
+      "step": 7384
+    },
+    {
+      "epoch": 31.57,
+      "learning_rate": 9.983442082194333e-05,
+      "loss": 0.0016,
+      "step": 7388
+    },
+    {
+      "epoch": 31.59,
+      "learning_rate": 9.98334717933225e-05,
+      "loss": 0.0013,
+      "step": 7392
+    },
+    {
+      "epoch": 31.61,
+      "learning_rate": 9.983252005729264e-05,
+      "loss": 0.0035,
+      "step": 7396
+    },
+    {
+      "epoch": 31.62,
+      "learning_rate": 9.983156561390547e-05,
+      "loss": 0.0008,
+      "step": 7400
+    },
+    {
+      "epoch": 31.64,
+      "learning_rate": 9.983060846321282e-05,
+      "loss": 0.0021,
+      "step": 7404
+    },
+    {
+      "epoch": 31.66,
+      "learning_rate": 9.982964860526671e-05,
+      "loss": 0.0017,
+      "step": 7408
+    },
+    {
+      "epoch": 31.68,
+      "learning_rate": 9.982868604011929e-05,
+      "loss": 0.0025,
+      "step": 7412
+    },
+    {
+      "epoch": 31.69,
+      "learning_rate": 9.982772076782283e-05,
+      "loss": 0.0009,
+      "step": 7416
+    },
+    {
+      "epoch": 31.71,
+      "learning_rate": 9.982675278842981e-05,
+      "loss": 0.0026,
+      "step": 7420
+    },
+    {
+      "epoch": 31.73,
+      "learning_rate": 9.982578210199279e-05,
+      "loss": 0.0022,
+      "step": 7424
+    },
+    {
+      "epoch": 31.74,
+      "learning_rate": 9.98248087085645e-05,
+      "loss": 0.002,
+      "step": 7428
+    },
+    {
+      "epoch": 31.76,
+      "learning_rate": 9.982383260819785e-05,
+      "loss": 0.002,
+      "step": 7432
+    },
+    {
+      "epoch": 31.78,
+      "learning_rate": 9.982285380094587e-05,
+      "loss": 0.0018,
+      "step": 7436
+    },
+    {
+      "epoch": 31.79,
+      "learning_rate": 9.982187228686172e-05,
+      "loss": 0.0017,
+      "step": 7440
+    },
+    {
+      "epoch": 31.81,
+      "learning_rate": 9.982088806599874e-05,
+      "loss": 0.0035,
+      "step": 7444
+    },
+    {
+      "epoch": 31.83,
+      "learning_rate": 9.981990113841038e-05,
+      "loss": 0.0027,
+      "step": 7448
+    },
+    {
+      "epoch": 31.85,
+      "learning_rate": 9.981891150415029e-05,
+      "loss": 0.0017,
+      "step": 7452
+    },
+    {
+      "epoch": 31.86,
+      "learning_rate": 9.981791916327218e-05,
+      "loss": 0.0029,
+      "step": 7456
+    },
+    {
+      "epoch": 31.88,
+      "learning_rate": 9.981692411583005e-05,
+      "loss": 0.0012,
+      "step": 7460
+    },
+    {
+      "epoch": 31.9,
+      "learning_rate": 9.981592636187789e-05,
+      "loss": 0.0023,
+      "step": 7464
+    },
+    {
+      "epoch": 31.91,
+      "learning_rate": 9.981492590146992e-05,
+      "loss": 0.0023,
+      "step": 7468
+    },
+    {
+      "epoch": 31.93,
+      "learning_rate": 9.981392273466053e-05,
+      "loss": 0.0024,
+      "step": 7472
+    },
+    {
+      "epoch": 31.95,
+      "learning_rate": 9.981291686150418e-05,
+      "loss": 0.0035,
+      "step": 7476
+    },
+    {
+      "epoch": 31.97,
+      "learning_rate": 9.981190828205553e-05,
+      "loss": 0.0029,
+      "step": 7480
+    },
+    {
+      "epoch": 31.98,
+      "learning_rate": 9.981089699636939e-05,
+      "loss": 0.0013,
+      "step": 7484
+    },
+    {
+      "epoch": 32.0,
+      "learning_rate": 9.980988300450067e-05,
+      "loss": 0.0011,
+      "step": 7488
+    },
+    {
+      "epoch": 32.02,
+      "learning_rate": 9.98088663065045e-05,
+      "loss": 0.0022,
+      "step": 7492
+    },
+    {
+      "epoch": 32.03,
+      "learning_rate": 9.980784690243608e-05,
+      "loss": 0.0028,
+      "step": 7496
+    },
+    {
+      "epoch": 32.05,
+      "learning_rate": 9.980682479235082e-05,
+      "loss": 0.002,
+      "step": 7500
+    },
+    {
+      "epoch": 32.07,
+      "learning_rate": 9.980579997630423e-05,
+      "loss": 0.0013,
+      "step": 7504
+    },
+    {
+      "epoch": 32.09,
+      "learning_rate": 9.9804772454352e-05,
+      "loss": 0.0044,
+      "step": 7508
+    },
+    {
+      "epoch": 32.1,
+      "learning_rate": 9.980374222654996e-05,
+      "loss": 0.002,
+      "step": 7512
+    },
+    {
+      "epoch": 32.12,
+      "learning_rate": 9.980270929295406e-05,
+      "loss": 0.0019,
+      "step": 7516
+    },
+    {
+      "epoch": 32.14,
+      "learning_rate": 9.980167365362044e-05,
+      "loss": 0.0014,
+      "step": 7520
+    },
+    {
+      "epoch": 32.15,
+      "learning_rate": 9.980063530860535e-05,
+      "loss": 0.0061,
+      "step": 7524
+    },
+    {
+      "epoch": 32.17,
+      "learning_rate": 9.979959425796521e-05,
+      "loss": 0.0006,
+      "step": 7528
+    },
+    {
+      "epoch": 32.19,
+      "learning_rate": 9.979855050175659e-05,
+      "loss": 0.0022,
+      "step": 7532
+    },
+    {
+      "epoch": 32.21,
+      "learning_rate": 9.979750404003617e-05,
+      "loss": 0.004,
+      "step": 7536
+    },
+    {
+      "epoch": 32.22,
+      "learning_rate": 9.979645487286082e-05,
+      "loss": 0.0024,
+      "step": 7540
+    },
+    {
+      "epoch": 32.24,
+      "learning_rate": 9.979540300028755e-05,
+      "loss": 0.001,
+      "step": 7544
+    },
+    {
+      "epoch": 32.26,
+      "learning_rate": 9.979434842237347e-05,
+      "loss": 0.0016,
+      "step": 7548
+    },
+    {
+      "epoch": 32.27,
+      "learning_rate": 9.979329113917593e-05,
+      "loss": 0.0019,
+      "step": 7552
+    },
+    {
+      "epoch": 32.29,
+      "learning_rate": 9.979223115075233e-05,
+      "loss": 0.008,
+      "step": 7556
+    },
+    {
+      "epoch": 32.31,
+      "learning_rate": 9.979116845716026e-05,
+      "loss": 0.0041,
+      "step": 7560
+    },
+    {
+      "epoch": 32.32,
+      "learning_rate": 9.979010305845747e-05,
+      "loss": 0.0019,
+      "step": 7564
+    },
+    {
+      "epoch": 32.34,
+      "learning_rate": 9.978903495470184e-05,
+      "loss": 0.0021,
+      "step": 7568
+    },
+    {
+      "epoch": 32.36,
+      "learning_rate": 9.97879641459514e-05,
+      "loss": 0.0012,
+      "step": 7572
+    },
+    {
+      "epoch": 32.38,
+      "learning_rate": 9.978689063226432e-05,
+      "loss": 0.0005,
+      "step": 7576
+    },
+    {
+      "epoch": 32.39,
+      "learning_rate": 9.978581441369892e-05,
+      "loss": 0.0009,
+      "step": 7580
+    },
+    {
+      "epoch": 32.41,
+      "learning_rate": 9.978473549031368e-05,
+      "loss": 0.0023,
+      "step": 7584
+    },
+    {
+      "epoch": 32.43,
+      "learning_rate": 9.97836538621672e-05,
+      "loss": 0.0013,
+      "step": 7588
+    },
+    {
+      "epoch": 32.44,
+      "learning_rate": 9.978256952931827e-05,
+      "loss": 0.0018,
+      "step": 7592
+    },
+    {
+      "epoch": 32.46,
+      "learning_rate": 9.978148249182578e-05,
+      "loss": 0.0034,
+      "step": 7596
+    },
+    {
+      "epoch": 32.48,
+      "learning_rate": 9.97803927497488e-05,
+      "loss": 0.0006,
+      "step": 7600
+    },
+    {
+      "epoch": 32.5,
+      "learning_rate": 9.977930030314653e-05,
+      "loss": 0.0017,
+      "step": 7604
+    },
+    {
+      "epoch": 32.51,
+      "learning_rate": 9.977820515207831e-05,
+      "loss": 0.0005,
+      "step": 7608
+    },
+    {
+      "epoch": 32.53,
+      "learning_rate": 9.977710729660365e-05,
+      "loss": 0.0008,
+      "step": 7612
+    },
+    {
+      "epoch": 32.55,
+      "learning_rate": 9.97760067367822e-05,
+      "loss": 0.0013,
+      "step": 7616
+    },
+    {
+      "epoch": 32.56,
+      "learning_rate": 9.977490347267375e-05,
+      "loss": 0.0021,
+      "step": 7620
+    },
+    {
+      "epoch": 32.58,
+      "learning_rate": 9.977379750433824e-05,
+      "loss": 0.0037,
+      "step": 7624
+    },
+    {
+      "epoch": 32.6,
+      "learning_rate": 9.977268883183575e-05,
+      "loss": 0.0007,
+      "step": 7628
+    },
+    {
+      "epoch": 32.62,
+      "learning_rate": 9.977157745522652e-05,
+      "loss": 0.0025,
+      "step": 7632
+    },
+    {
+      "epoch": 32.63,
+      "learning_rate": 9.977046337457092e-05,
+      "loss": 0.0029,
+      "step": 7636
+    },
+    {
+      "epoch": 32.65,
+      "learning_rate": 9.976934658992949e-05,
+      "loss": 0.0022,
+      "step": 7640
+    },
+    {
+      "epoch": 32.67,
+      "learning_rate": 9.97682271013629e-05,
+      "loss": 0.002,
+      "step": 7644
+    },
+    {
+      "epoch": 32.68,
+      "learning_rate": 9.976710490893195e-05,
+      "loss": 0.0019,
+      "step": 7648
+    },
+    {
+      "epoch": 32.7,
+      "learning_rate": 9.976598001269765e-05,
+      "loss": 0.002,
+      "step": 7652
+    },
+    {
+      "epoch": 32.72,
+      "learning_rate": 9.976485241272107e-05,
+      "loss": 0.0009,
+      "step": 7656
+    },
+    {
+      "epoch": 32.74,
+      "learning_rate": 9.97637221090635e-05,
+      "loss": 0.0046,
+      "step": 7660
+    },
+    {
+      "epoch": 32.75,
+      "learning_rate": 9.976258910178635e-05,
+      "loss": 0.002,
+      "step": 7664
+    },
+    {
+      "epoch": 32.77,
+      "learning_rate": 9.976145339095115e-05,
+      "loss": 0.001,
+      "step": 7668
+    },
+    {
+      "epoch": 32.79,
+      "learning_rate": 9.976031497661964e-05,
+      "loss": 0.0008,
+      "step": 7672
+    },
+    {
+      "epoch": 32.8,
+      "learning_rate": 9.975917385885363e-05,
+      "loss": 0.0016,
+      "step": 7676
+    },
+    {
+      "epoch": 32.82,
+      "learning_rate": 9.975803003771513e-05,
+      "loss": 0.0009,
+      "step": 7680
+    },
+    {
+      "epoch": 32.84,
+      "learning_rate": 9.97568835132663e-05,
+      "loss": 0.0007,
+      "step": 7684
+    },
+    {
+      "epoch": 32.85,
+      "learning_rate": 9.975573428556941e-05,
+      "loss": 0.0008,
+      "step": 7688
+    },
+    {
+      "epoch": 32.87,
+      "learning_rate": 9.975458235468691e-05,
+      "loss": 0.0007,
+      "step": 7692
+    },
+    {
+      "epoch": 32.89,
+      "learning_rate": 9.975342772068137e-05,
+      "loss": 0.0004,
+      "step": 7696
+    },
+    {
+      "epoch": 32.91,
+      "learning_rate": 9.975227038361552e-05,
+      "loss": 0.0009,
+      "step": 7700
+    },
+    {
+      "epoch": 32.92,
+      "learning_rate": 9.975111034355225e-05,
+      "loss": 0.0006,
+      "step": 7704
+    },
+    {
+      "epoch": 32.94,
+      "learning_rate": 9.974994760055457e-05,
+      "loss": 0.0029,
+      "step": 7708
+    },
+    {
+      "epoch": 32.96,
+      "learning_rate": 9.974878215468566e-05,
+      "loss": 0.0026,
+      "step": 7712
+    },
+    {
+      "epoch": 32.97,
+      "learning_rate": 9.974761400600883e-05,
+      "loss": 0.0011,
+      "step": 7716
+    },
+    {
+      "epoch": 32.99,
+      "learning_rate": 9.974644315458756e-05,
+      "loss": 0.001,
+      "step": 7720
+    },
+    {
+      "epoch": 33.01,
+      "learning_rate": 9.974526960048545e-05,
+      "loss": 0.0004,
+      "step": 7724
+    },
+    {
+      "epoch": 33.03,
+      "learning_rate": 9.974409334376626e-05,
+      "loss": 0.0034,
+      "step": 7728
+    },
+    {
+      "epoch": 33.04,
+      "learning_rate": 9.974291438449388e-05,
+      "loss": 0.0015,
+      "step": 7732
+    },
+    {
+      "epoch": 33.06,
+      "learning_rate": 9.974173272273239e-05,
+      "loss": 0.002,
+      "step": 7736
+    },
+    {
+      "epoch": 33.08,
+      "learning_rate": 9.974054835854598e-05,
+      "loss": 0.0028,
+      "step": 7740
+    },
+    {
+      "epoch": 33.09,
+      "learning_rate": 9.973936129199898e-05,
+      "loss": 0.003,
+      "step": 7744
+    },
+    {
+      "epoch": 33.11,
+      "learning_rate": 9.97381715231559e-05,
+      "loss": 0.0017,
+      "step": 7748
+    },
+    {
+      "epoch": 33.13,
+      "learning_rate": 9.973697905208136e-05,
+      "loss": 0.0019,
+      "step": 7752
+    },
+    {
+      "epoch": 33.15,
+      "learning_rate": 9.973578387884017e-05,
+      "loss": 0.0009,
+      "step": 7756
+    },
+    {
+      "epoch": 33.16,
+      "learning_rate": 9.973458600349724e-05,
+      "loss": 0.001,
+      "step": 7760
+    },
+    {
+      "epoch": 33.18,
+      "learning_rate": 9.973338542611766e-05,
+      "loss": 0.0003,
+      "step": 7764
+    },
+    {
+      "epoch": 33.2,
+      "learning_rate": 9.973218214676668e-05,
+      "loss": 0.0007,
+      "step": 7768
+    },
+    {
+      "epoch": 33.21,
+      "learning_rate": 9.973097616550961e-05,
+      "loss": 0.0006,
+      "step": 7772
+    },
+    {
+      "epoch": 33.23,
+      "learning_rate": 9.972976748241203e-05,
+      "loss": 0.0048,
+      "step": 7776
+    },
+    {
+      "epoch": 33.25,
+      "learning_rate": 9.972855609753959e-05,
+      "loss": 0.0006,
+      "step": 7780
+    },
+    {
+      "epoch": 33.26,
+      "learning_rate": 9.97273420109581e-05,
+      "loss": 0.0005,
+      "step": 7784
+    },
+    {
+      "epoch": 33.28,
+      "learning_rate": 9.972612522273352e-05,
+      "loss": 0.0021,
+      "step": 7788
+    },
+    {
+      "epoch": 33.3,
+      "learning_rate": 9.972490573293196e-05,
+      "loss": 0.0008,
+      "step": 7792
+    },
+    {
+      "epoch": 33.32,
+      "learning_rate": 9.972368354161965e-05,
+      "loss": 0.0013,
+      "step": 7796
+    },
+    {
+      "epoch": 33.33,
+      "learning_rate": 9.972245864886302e-05,
+      "loss": 0.0017,
+      "step": 7800
+    },
+    {
+      "epoch": 33.35,
+      "learning_rate": 9.972123105472862e-05,
+      "loss": 0.0012,
+      "step": 7804
+    },
+    {
+      "epoch": 33.37,
+      "learning_rate": 9.972000075928311e-05,
+      "loss": 0.001,
+      "step": 7808
+    },
+    {
+      "epoch": 33.38,
+      "learning_rate": 9.971876776259339e-05,
+      "loss": 0.0016,
+      "step": 7812
+    },
+    {
+      "epoch": 33.4,
+      "learning_rate": 9.971753206472638e-05,
+      "loss": 0.002,
+      "step": 7816
+    },
+    {
+      "epoch": 33.42,
+      "learning_rate": 9.971629366574926e-05,
+      "loss": 0.0023,
+      "step": 7820
+    },
+    {
+      "epoch": 33.44,
+      "learning_rate": 9.971505256572928e-05,
+      "loss": 0.0011,
+      "step": 7824
+    },
+    {
+      "epoch": 33.45,
+      "learning_rate": 9.971380876473391e-05,
+      "loss": 0.0006,
+      "step": 7828
+    },
+    {
+      "epoch": 33.47,
+      "learning_rate": 9.971256226283068e-05,
+      "loss": 0.0011,
+      "step": 7832
+    },
+    {
+      "epoch": 33.49,
+      "learning_rate": 9.971131306008734e-05,
+      "loss": 0.0006,
+      "step": 7836
+    },
+    {
+      "epoch": 33.5,
+      "learning_rate": 9.971006115657174e-05,
+      "loss": 0.0006,
+      "step": 7840
+    },
+    {
+      "epoch": 33.52,
+      "learning_rate": 9.970880655235191e-05,
+      "loss": 0.001,
+      "step": 7844
+    },
+    {
+      "epoch": 33.54,
+      "learning_rate": 9.970754924749601e-05,
+      "loss": 0.0006,
+      "step": 7848
+    },
+    {
+      "epoch": 33.56,
+      "learning_rate": 9.970628924207232e-05,
+      "loss": 0.0049,
+      "step": 7852
+    },
+    {
+      "epoch": 33.57,
+      "learning_rate": 9.970502653614934e-05,
+      "loss": 0.0012,
+      "step": 7856
+    },
+    {
+      "epoch": 33.59,
+      "learning_rate": 9.970376112979564e-05,
+      "loss": 0.001,
+      "step": 7860
+    },
+    {
+      "epoch": 33.61,
+      "learning_rate": 9.970249302307999e-05,
+      "loss": 0.0016,
+      "step": 7864
+    },
+    {
+      "epoch": 33.62,
+      "learning_rate": 9.970122221607128e-05,
+      "loss": 0.0008,
+      "step": 7868
+    },
+    {
+      "epoch": 33.64,
+      "learning_rate": 9.969994870883852e-05,
+      "loss": 0.0011,
+      "step": 7872
+    },
+    {
+      "epoch": 33.66,
+      "learning_rate": 9.969867250145094e-05,
+      "loss": 0.0013,
+      "step": 7876
+    },
+    {
+      "epoch": 33.68,
+      "learning_rate": 9.969739359397785e-05,
+      "loss": 0.002,
+      "step": 7880
+    },
+    {
+      "epoch": 33.69,
+      "learning_rate": 9.969611198648876e-05,
+      "loss": 0.002,
+      "step": 7884
+    },
+    {
+      "epoch": 33.71,
+      "learning_rate": 9.969482767905326e-05,
+      "loss": 0.0014,
+      "step": 7888
+    },
+    {
+      "epoch": 33.73,
+      "learning_rate": 9.969354067174115e-05,
+      "loss": 0.0008,
+      "step": 7892
+    },
+    {
+      "epoch": 33.74,
+      "learning_rate": 9.969225096462234e-05,
+      "loss": 0.0013,
+      "step": 7896
+    },
+    {
+      "epoch": 33.76,
+      "learning_rate": 9.969095855776692e-05,
+      "loss": 0.0043,
+      "step": 7900
+    },
+    {
+      "epoch": 33.78,
+      "learning_rate": 9.968966345124509e-05,
+      "loss": 0.0015,
+      "step": 7904
+    },
+    {
+      "epoch": 33.79,
+      "learning_rate": 9.96883656451272e-05,
+      "loss": 0.0008,
+      "step": 7908
+    },
+    {
+      "epoch": 33.81,
+      "learning_rate": 9.968706513948378e-05,
+      "loss": 0.0025,
+      "step": 7912
+    },
+    {
+      "epoch": 33.83,
+      "learning_rate": 9.968576193438548e-05,
+      "loss": 0.0021,
+      "step": 7916
+    },
+    {
+      "epoch": 33.85,
+      "learning_rate": 9.968445602990309e-05,
+      "loss": 0.0016,
+      "step": 7920
+    },
+    {
+      "epoch": 33.86,
+      "learning_rate": 9.968314742610758e-05,
+      "loss": 0.0035,
+      "step": 7924
+    },
+    {
+      "epoch": 33.88,
+      "learning_rate": 9.968183612307003e-05,
+      "loss": 0.0033,
+      "step": 7928
+    },
+    {
+      "epoch": 33.9,
+      "learning_rate": 9.968052212086168e-05,
+      "loss": 0.0039,
+      "step": 7932
+    },
+    {
+      "epoch": 33.91,
+      "learning_rate": 9.967920541955392e-05,
+      "loss": 0.0021,
+      "step": 7936
+    },
+    {
+      "epoch": 33.93,
+      "learning_rate": 9.96778860192183e-05,
+      "loss": 0.0035,
+      "step": 7940
+    },
+    {
+      "epoch": 33.95,
+      "learning_rate": 9.967656391992647e-05,
+      "loss": 0.0005,
+      "step": 7944
+    },
+    {
+      "epoch": 33.97,
+      "learning_rate": 9.96752391217503e-05,
+      "loss": 0.0016,
+      "step": 7948
+    },
+    {
+      "epoch": 33.98,
+      "learning_rate": 9.967391162476174e-05,
+      "loss": 0.0014,
+      "step": 7952
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 9.967258142903291e-05,
+      "loss": 0.0031,
+      "step": 7956
+    },
+    {
+      "epoch": 34.02,
+      "learning_rate": 9.967124853463608e-05,
+      "loss": 0.0027,
+      "step": 7960
+    },
+    {
+      "epoch": 34.03,
+      "learning_rate": 9.966991294164368e-05,
+      "loss": 0.0011,
+      "step": 7964
+    },
+    {
+      "epoch": 34.05,
+      "learning_rate": 9.966857465012827e-05,
+      "loss": 0.0019,
+      "step": 7968
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 9.966723366016253e-05,
+      "loss": 0.003,
+      "step": 7972
+    },
+    {
+      "epoch": 34.09,
+      "learning_rate": 9.966588997181933e-05,
+      "loss": 0.0013,
+      "step": 7976
+    },
+    {
+      "epoch": 34.1,
+      "learning_rate": 9.96645435851717e-05,
+      "loss": 0.002,
+      "step": 7980
+    },
+    {
+      "epoch": 34.12,
+      "learning_rate": 9.966319450029274e-05,
+      "loss": 0.0021,
+      "step": 7984
+    },
+    {
+      "epoch": 34.14,
+      "learning_rate": 9.966184271725578e-05,
+      "loss": 0.0033,
+      "step": 7988
+    },
+    {
+      "epoch": 34.15,
+      "learning_rate": 9.966048823613425e-05,
+      "loss": 0.001,
+      "step": 7992
+    },
+    {
+      "epoch": 34.17,
+      "learning_rate": 9.965913105700173e-05,
+      "loss": 0.0015,
+      "step": 7996
+    },
+    {
+      "epoch": 34.19,
+      "learning_rate": 9.965777117993196e-05,
+      "loss": 0.0013,
+      "step": 8000
+    },
+    {
+      "epoch": 34.19,
+      "eval_exact_match": 0.5103950103950103,
+      "eval_loss": 0.726246178150177,
+      "eval_runtime": 130.3318,
+      "eval_samples_per_second": 7.381,
+      "step": 8000
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 9.965640860499883e-05,
+      "loss": 0.0044,
+      "step": 8004
+    },
+    {
+      "epoch": 34.22,
+      "learning_rate": 9.965504333227636e-05,
+      "loss": 0.0018,
+      "step": 8008
+    },
+    {
+      "epoch": 34.24,
+      "learning_rate": 9.965367536183872e-05,
+      "loss": 0.0017,
+      "step": 8012
+    },
+    {
+      "epoch": 34.26,
+      "learning_rate": 9.965230469376024e-05,
+      "loss": 0.002,
+      "step": 8016
+    },
+    {
+      "epoch": 34.27,
+      "learning_rate": 9.965093132811539e-05,
+      "loss": 0.0006,
+      "step": 8020
+    },
+    {
+      "epoch": 34.29,
+      "learning_rate": 9.964955526497874e-05,
+      "loss": 0.0014,
+      "step": 8024
+    },
+    {
+      "epoch": 34.31,
+      "learning_rate": 9.964817650442512e-05,
+      "loss": 0.0015,
+      "step": 8028
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 9.964679504652939e-05,
+      "loss": 0.0038,
+      "step": 8032
+    },
+    {
+      "epoch": 34.34,
+      "learning_rate": 9.964541089136661e-05,
+      "loss": 0.0006,
+      "step": 8036
+    },
+    {
+      "epoch": 34.36,
+      "learning_rate": 9.9644024039012e-05,
+      "loss": 0.0024,
+      "step": 8040
+    },
+    {
+      "epoch": 34.38,
+      "learning_rate": 9.964263448954087e-05,
+      "loss": 0.0022,
+      "step": 8044
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 9.964124224302877e-05,
+      "loss": 0.0045,
+      "step": 8048
+    },
+    {
+      "epoch": 34.41,
+      "learning_rate": 9.963984729955128e-05,
+      "loss": 0.0012,
+      "step": 8052
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 9.96384496591842e-05,
+      "loss": 0.0018,
+      "step": 8056
+    },
+    {
+      "epoch": 34.44,
+      "learning_rate": 9.963704932200351e-05,
+      "loss": 0.0014,
+      "step": 8060
+    },
+    {
+      "epoch": 34.46,
+      "learning_rate": 9.963564628808523e-05,
+      "loss": 0.0019,
+      "step": 8064
+    },
+    {
+      "epoch": 34.48,
+      "learning_rate": 9.963424055750561e-05,
+      "loss": 0.0006,
+      "step": 8068
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 9.963283213034103e-05,
+      "loss": 0.0013,
+      "step": 8072
+    },
+    {
+      "epoch": 34.51,
+      "learning_rate": 9.963142100666799e-05,
+      "loss": 0.0026,
+      "step": 8076
+    },
+    {
+      "epoch": 34.53,
+      "learning_rate": 9.963000718656316e-05,
+      "loss": 0.0021,
+      "step": 8080
+    },
+    {
+      "epoch": 34.55,
+      "learning_rate": 9.962859067010335e-05,
+      "loss": 0.0031,
+      "step": 8084
+    },
+    {
+      "epoch": 34.56,
+      "learning_rate": 9.962717145736554e-05,
+      "loss": 0.0012,
+      "step": 8088
+    },
+    {
+      "epoch": 34.58,
+      "learning_rate": 9.962574954842682e-05,
+      "loss": 0.0004,
+      "step": 8092
+    },
+    {
+      "epoch": 34.6,
+      "learning_rate": 9.962432494336442e-05,
+      "loss": 0.0014,
+      "step": 8096
+    },
+    {
+      "epoch": 34.62,
+      "learning_rate": 9.962289764225579e-05,
+      "loss": 0.0045,
+      "step": 8100
+    },
+    {
+      "epoch": 34.63,
+      "learning_rate": 9.962146764517841e-05,
+      "loss": 0.0006,
+      "step": 8104
+    },
+    {
+      "epoch": 34.65,
+      "learning_rate": 9.962003495221002e-05,
+      "loss": 0.0021,
+      "step": 8108
+    },
+    {
+      "epoch": 34.67,
+      "learning_rate": 9.961859956342843e-05,
+      "loss": 0.0015,
+      "step": 8112
+    },
+    {
+      "epoch": 34.68,
+      "learning_rate": 9.961716147891163e-05,
+      "loss": 0.0008,
+      "step": 8116
+    },
+    {
+      "epoch": 34.7,
+      "learning_rate": 9.961572069873775e-05,
+      "loss": 0.0037,
+      "step": 8120
+    },
+    {
+      "epoch": 34.72,
+      "learning_rate": 9.961427722298507e-05,
+      "loss": 0.0009,
+      "step": 8124
+    },
+    {
+      "epoch": 34.74,
+      "learning_rate": 9.961283105173202e-05,
+      "loss": 0.0023,
+      "step": 8128
+    },
+    {
+      "epoch": 34.75,
+      "learning_rate": 9.961138218505714e-05,
+      "loss": 0.0029,
+      "step": 8132
+    },
+    {
+      "epoch": 34.77,
+      "learning_rate": 9.960993062303919e-05,
+      "loss": 0.0008,
+      "step": 8136
+    },
+    {
+      "epoch": 34.79,
+      "learning_rate": 9.960847636575699e-05,
+      "loss": 0.0018,
+      "step": 8140
+    },
+    {
+      "epoch": 34.8,
+      "learning_rate": 9.960701941328958e-05,
+      "loss": 0.0014,
+      "step": 8144
+    },
+    {
+      "epoch": 34.82,
+      "learning_rate": 9.96055597657161e-05,
+      "loss": 0.0011,
+      "step": 8148
+    },
+    {
+      "epoch": 34.84,
+      "learning_rate": 9.960409742311584e-05,
+      "loss": 0.0009,
+      "step": 8152
+    },
+    {
+      "epoch": 34.85,
+      "learning_rate": 9.960263238556828e-05,
+      "loss": 0.0048,
+      "step": 8156
+    },
+    {
+      "epoch": 34.87,
+      "learning_rate": 9.9601164653153e-05,
+      "loss": 0.0022,
+      "step": 8160
+    },
+    {
+      "epoch": 34.89,
+      "learning_rate": 9.95996942259497e-05,
+      "loss": 0.0086,
+      "step": 8164
+    },
+    {
+      "epoch": 34.91,
+      "learning_rate": 9.959822110403832e-05,
+      "loss": 0.0022,
+      "step": 8168
+    },
+    {
+      "epoch": 34.92,
+      "learning_rate": 9.959674528749888e-05,
+      "loss": 0.0008,
+      "step": 8172
+    },
+    {
+      "epoch": 34.94,
+      "learning_rate": 9.959526677641156e-05,
+      "loss": 0.0015,
+      "step": 8176
+    },
+    {
+      "epoch": 34.96,
+      "learning_rate": 9.959378557085668e-05,
+      "loss": 0.0016,
+      "step": 8180
+    },
+    {
+      "epoch": 34.97,
+      "learning_rate": 9.959230167091471e-05,
+      "loss": 0.0005,
+      "step": 8184
+    },
+    {
+      "epoch": 34.99,
+      "learning_rate": 9.959081507666626e-05,
+      "loss": 0.0009,
+      "step": 8188
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 9.958932578819213e-05,
+      "loss": 0.0011,
+      "step": 8192
+    },
+    {
+      "epoch": 35.03,
+      "learning_rate": 9.95878338055732e-05,
+      "loss": 0.0018,
+      "step": 8196
+    },
+    {
+      "epoch": 35.04,
+      "learning_rate": 9.958633912889054e-05,
+      "loss": 0.0006,
+      "step": 8200
+    },
+    {
+      "epoch": 35.06,
+      "learning_rate": 9.958484175822535e-05,
+      "loss": 0.0028,
+      "step": 8204
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 9.958334169365899e-05,
+      "loss": 0.0011,
+      "step": 8208
+    },
+    {
+      "epoch": 35.09,
+      "learning_rate": 9.958183893527296e-05,
+      "loss": 0.0039,
+      "step": 8212
+    },
+    {
+      "epoch": 35.11,
+      "learning_rate": 9.958033348314888e-05,
+      "loss": 0.0019,
+      "step": 8216
+    },
+    {
+      "epoch": 35.13,
+      "learning_rate": 9.957882533736856e-05,
+      "loss": 0.0022,
+      "step": 8220
+    },
+    {
+      "epoch": 35.15,
+      "learning_rate": 9.957731449801393e-05,
+      "loss": 0.002,
+      "step": 8224
+    },
+    {
+      "epoch": 35.16,
+      "learning_rate": 9.957580096516707e-05,
+      "loss": 0.0025,
+      "step": 8228
+    },
+    {
+      "epoch": 35.18,
+      "learning_rate": 9.957428473891023e-05,
+      "loss": 0.0025,
+      "step": 8232
+    },
+    {
+      "epoch": 35.2,
+      "learning_rate": 9.957276581932575e-05,
+      "loss": 0.0018,
+      "step": 8236
+    },
+    {
+      "epoch": 35.21,
+      "learning_rate": 9.957124420649617e-05,
+      "loss": 0.0011,
+      "step": 8240
+    },
+    {
+      "epoch": 35.23,
+      "learning_rate": 9.956971990050417e-05,
+      "loss": 0.0028,
+      "step": 8244
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 9.956819290143255e-05,
+      "loss": 0.0021,
+      "step": 8248
+    },
+    {
+      "epoch": 35.26,
+      "learning_rate": 9.956666320936425e-05,
+      "loss": 0.0012,
+      "step": 8252
+    },
+    {
+      "epoch": 35.28,
+      "learning_rate": 9.956513082438243e-05,
+      "loss": 0.0011,
+      "step": 8256
+    },
+    {
+      "epoch": 35.3,
+      "learning_rate": 9.956359574657028e-05,
+      "loss": 0.0027,
+      "step": 8260
+    },
+    {
+      "epoch": 35.32,
+      "learning_rate": 9.956205797601126e-05,
+      "loss": 0.0009,
+      "step": 8264
+    },
+    {
+      "epoch": 35.33,
+      "learning_rate": 9.956051751278888e-05,
+      "loss": 0.0021,
+      "step": 8268
+    },
+    {
+      "epoch": 35.35,
+      "learning_rate": 9.955897435698683e-05,
+      "loss": 0.0034,
+      "step": 8272
+    },
+    {
+      "epoch": 35.37,
+      "learning_rate": 9.955742850868895e-05,
+      "loss": 0.0011,
+      "step": 8276
+    },
+    {
+      "epoch": 35.38,
+      "learning_rate": 9.955587996797925e-05,
+      "loss": 0.0055,
+      "step": 8280
+    },
+    {
+      "epoch": 35.4,
+      "learning_rate": 9.955432873494183e-05,
+      "loss": 0.0007,
+      "step": 8284
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 9.955277480966098e-05,
+      "loss": 0.002,
+      "step": 8288
+    },
+    {
+      "epoch": 35.44,
+      "learning_rate": 9.955121819222112e-05,
+      "loss": 0.0027,
+      "step": 8292
+    },
+    {
+      "epoch": 35.45,
+      "learning_rate": 9.954965888270684e-05,
+      "loss": 0.0018,
+      "step": 8296
+    },
+    {
+      "epoch": 35.47,
+      "learning_rate": 9.954809688120284e-05,
+      "loss": 0.0014,
+      "step": 8300
+    },
+    {
+      "epoch": 35.49,
+      "learning_rate": 9.954653218779397e-05,
+      "loss": 0.0024,
+      "step": 8304
+    },
+    {
+      "epoch": 35.5,
+      "learning_rate": 9.954496480256525e-05,
+      "loss": 0.0034,
+      "step": 8308
+    },
+    {
+      "epoch": 35.52,
+      "learning_rate": 9.954339472560183e-05,
+      "loss": 0.0037,
+      "step": 8312
+    },
+    {
+      "epoch": 35.54,
+      "learning_rate": 9.954182195698902e-05,
+      "loss": 0.0012,
+      "step": 8316
+    },
+    {
+      "epoch": 35.56,
+      "learning_rate": 9.954024649681227e-05,
+      "loss": 0.0008,
+      "step": 8320
+    },
+    {
+      "epoch": 35.57,
+      "learning_rate": 9.953866834515717e-05,
+      "loss": 0.0009,
+      "step": 8324
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 9.953708750210943e-05,
+      "loss": 0.0021,
+      "step": 8328
+    },
+    {
+      "epoch": 35.61,
+      "learning_rate": 9.953550396775498e-05,
+      "loss": 0.0036,
+      "step": 8332
+    },
+    {
+      "epoch": 35.62,
+      "learning_rate": 9.953391774217984e-05,
+      "loss": 0.0013,
+      "step": 8336
+    },
+    {
+      "epoch": 35.64,
+      "learning_rate": 9.953232882547017e-05,
+      "loss": 0.0008,
+      "step": 8340
+    },
+    {
+      "epoch": 35.66,
+      "learning_rate": 9.95307372177123e-05,
+      "loss": 0.0011,
+      "step": 8344
+    },
+    {
+      "epoch": 35.68,
+      "learning_rate": 9.952914291899271e-05,
+      "loss": 0.0033,
+      "step": 8348
+    },
+    {
+      "epoch": 35.69,
+      "learning_rate": 9.952754592939801e-05,
+      "loss": 0.0012,
+      "step": 8352
+    },
+    {
+      "epoch": 35.71,
+      "learning_rate": 9.952594624901498e-05,
+      "loss": 0.0017,
+      "step": 8356
+    },
+    {
+      "epoch": 35.73,
+      "learning_rate": 9.95243438779305e-05,
+      "loss": 0.002,
+      "step": 8360
+    },
+    {
+      "epoch": 35.74,
+      "learning_rate": 9.952273881623166e-05,
+      "loss": 0.0019,
+      "step": 8364
+    },
+    {
+      "epoch": 35.76,
+      "learning_rate": 9.952113106400562e-05,
+      "loss": 0.0042,
+      "step": 8368
+    },
+    {
+      "epoch": 35.78,
+      "learning_rate": 9.951952062133977e-05,
+      "loss": 0.0017,
+      "step": 8372
+    },
+    {
+      "epoch": 35.79,
+      "learning_rate": 9.951790748832158e-05,
+      "loss": 0.0017,
+      "step": 8376
+    },
+    {
+      "epoch": 35.81,
+      "learning_rate": 9.951629166503868e-05,
+      "loss": 0.0014,
+      "step": 8380
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 9.95146731515789e-05,
+      "loss": 0.0016,
+      "step": 8384
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 9.951305194803013e-05,
+      "loss": 0.0005,
+      "step": 8388
+    },
+    {
+      "epoch": 35.86,
+      "learning_rate": 9.951142805448046e-05,
+      "loss": 0.0008,
+      "step": 8392
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 9.95098014710181e-05,
+      "loss": 0.0029,
+      "step": 8396
+    },
+    {
+      "epoch": 35.9,
+      "learning_rate": 9.950817219773145e-05,
+      "loss": 0.0026,
+      "step": 8400
+    },
+    {
+      "epoch": 35.91,
+      "learning_rate": 9.950654023470902e-05,
+      "loss": 0.0044,
+      "step": 8404
+    },
+    {
+      "epoch": 35.93,
+      "learning_rate": 9.950490558203947e-05,
+      "loss": 0.0011,
+      "step": 8408
+    },
+    {
+      "epoch": 35.95,
+      "learning_rate": 9.95032682398116e-05,
+      "loss": 0.0013,
+      "step": 8412
+    },
+    {
+      "epoch": 35.97,
+      "learning_rate": 9.950162820811438e-05,
+      "loss": 0.0014,
+      "step": 8416
+    },
+    {
+      "epoch": 35.98,
+      "learning_rate": 9.949998548703688e-05,
+      "loss": 0.002,
+      "step": 8420
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 9.949834007666838e-05,
+      "loss": 0.0011,
+      "step": 8424
+    },
+    {
+      "epoch": 36.02,
+      "learning_rate": 9.949669197709827e-05,
+      "loss": 0.0014,
+      "step": 8428
+    },
+    {
+      "epoch": 36.03,
+      "learning_rate": 9.94950411884161e-05,
+      "loss": 0.0015,
+      "step": 8432
+    },
+    {
+      "epoch": 36.05,
+      "learning_rate": 9.949338771071152e-05,
+      "loss": 0.0008,
+      "step": 8436
+    },
+    {
+      "epoch": 36.07,
+      "learning_rate": 9.94917315440744e-05,
+      "loss": 0.0014,
+      "step": 8440
+    },
+    {
+      "epoch": 36.09,
+      "learning_rate": 9.94900726885947e-05,
+      "loss": 0.001,
+      "step": 8444
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 9.948841114436253e-05,
+      "loss": 0.0024,
+      "step": 8448
+    },
+    {
+      "epoch": 36.12,
+      "learning_rate": 9.948674691146817e-05,
+      "loss": 0.0012,
+      "step": 8452
+    },
+    {
+      "epoch": 36.14,
+      "learning_rate": 9.948507999000205e-05,
+      "loss": 0.0011,
+      "step": 8456
+    },
+    {
+      "epoch": 36.15,
+      "learning_rate": 9.948341038005474e-05,
+      "loss": 0.0014,
+      "step": 8460
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 9.948173808171691e-05,
+      "loss": 0.0017,
+      "step": 8464
+    },
+    {
+      "epoch": 36.19,
+      "learning_rate": 9.948006309507944e-05,
+      "loss": 0.0021,
+      "step": 8468
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 9.947838542023334e-05,
+      "loss": 0.0021,
+      "step": 8472
+    },
+    {
+      "epoch": 36.22,
+      "learning_rate": 9.947670505726974e-05,
+      "loss": 0.0008,
+      "step": 8476
+    },
+    {
+      "epoch": 36.24,
+      "learning_rate": 9.947502200627994e-05,
+      "loss": 0.0036,
+      "step": 8480
+    },
+    {
+      "epoch": 36.26,
+      "learning_rate": 9.947333626735535e-05,
+      "loss": 0.0006,
+      "step": 8484
+    },
+    {
+      "epoch": 36.27,
+      "learning_rate": 9.947164784058761e-05,
+      "loss": 0.0019,
+      "step": 8488
+    },
+    {
+      "epoch": 36.29,
+      "learning_rate": 9.94699567260684e-05,
+      "loss": 0.0004,
+      "step": 8492
+    },
+    {
+      "epoch": 36.31,
+      "learning_rate": 9.946826292388964e-05,
+      "loss": 0.0028,
+      "step": 8496
+    },
+    {
+      "epoch": 36.32,
+      "learning_rate": 9.946656643414331e-05,
+      "loss": 0.0039,
+      "step": 8500
+    },
+    {
+      "epoch": 36.34,
+      "learning_rate": 9.946486725692161e-05,
+      "loss": 0.0011,
+      "step": 8504
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 9.946316539231683e-05,
+      "loss": 0.0007,
+      "step": 8508
+    },
+    {
+      "epoch": 36.38,
+      "learning_rate": 9.946146084042148e-05,
+      "loss": 0.0022,
+      "step": 8512
+    },
+    {
+      "epoch": 36.39,
+      "learning_rate": 9.94597536013281e-05,
+      "loss": 0.0009,
+      "step": 8516
+    },
+    {
+      "epoch": 36.41,
+      "learning_rate": 9.945804367512948e-05,
+      "loss": 0.0015,
+      "step": 8520
+    },
+    {
+      "epoch": 36.43,
+      "learning_rate": 9.945633106191851e-05,
+      "loss": 0.001,
+      "step": 8524
+    },
+    {
+      "epoch": 36.44,
+      "learning_rate": 9.945461576178825e-05,
+      "loss": 0.0004,
+      "step": 8528
+    },
+    {
+      "epoch": 36.46,
+      "learning_rate": 9.945289777483186e-05,
+      "loss": 0.0011,
+      "step": 8532
+    },
+    {
+      "epoch": 36.48,
+      "learning_rate": 9.945117710114272e-05,
+      "loss": 0.002,
+      "step": 8536
+    },
+    {
+      "epoch": 36.5,
+      "learning_rate": 9.944945374081427e-05,
+      "loss": 0.0013,
+      "step": 8540
+    },
+    {
+      "epoch": 36.51,
+      "learning_rate": 9.944772769394018e-05,
+      "loss": 0.0008,
+      "step": 8544
+    },
+    {
+      "epoch": 36.53,
+      "learning_rate": 9.944599896061417e-05,
+      "loss": 0.0004,
+      "step": 8548
+    },
+    {
+      "epoch": 36.55,
+      "learning_rate": 9.944426754093021e-05,
+      "loss": 0.0007,
+      "step": 8552
+    },
+    {
+      "epoch": 36.56,
+      "learning_rate": 9.944253343498234e-05,
+      "loss": 0.0017,
+      "step": 8556
+    },
+    {
+      "epoch": 36.58,
+      "learning_rate": 9.944079664286479e-05,
+      "loss": 0.0013,
+      "step": 8560
+    },
+    {
+      "epoch": 36.6,
+      "learning_rate": 9.94390571646719e-05,
+      "loss": 0.0013,
+      "step": 8564
+    },
+    {
+      "epoch": 36.62,
+      "learning_rate": 9.94373150004982e-05,
+      "loss": 0.0007,
+      "step": 8568
+    },
+    {
+      "epoch": 36.63,
+      "learning_rate": 9.943557015043832e-05,
+      "loss": 0.0038,
+      "step": 8572
+    },
+    {
+      "epoch": 36.65,
+      "learning_rate": 9.943382261458705e-05,
+      "loss": 0.0022,
+      "step": 8576
+    },
+    {
+      "epoch": 36.67,
+      "learning_rate": 9.943207239303935e-05,
+      "loss": 0.0011,
+      "step": 8580
+    },
+    {
+      "epoch": 36.68,
+      "learning_rate": 9.943031948589029e-05,
+      "loss": 0.0009,
+      "step": 8584
+    },
+    {
+      "epoch": 36.7,
+      "learning_rate": 9.942856389323512e-05,
+      "loss": 0.0012,
+      "step": 8588
+    },
+    {
+      "epoch": 36.72,
+      "learning_rate": 9.942680561516921e-05,
+      "loss": 0.0004,
+      "step": 8592
+    },
+    {
+      "epoch": 36.74,
+      "learning_rate": 9.94250446517881e-05,
+      "loss": 0.001,
+      "step": 8596
+    },
+    {
+      "epoch": 36.75,
+      "learning_rate": 9.942328100318745e-05,
+      "loss": 0.0008,
+      "step": 8600
+    },
+    {
+      "epoch": 36.77,
+      "learning_rate": 9.942151466946308e-05,
+      "loss": 0.0009,
+      "step": 8604
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 9.941974565071094e-05,
+      "loss": 0.0004,
+      "step": 8608
+    },
+    {
+      "epoch": 36.8,
+      "learning_rate": 9.941797394702717e-05,
+      "loss": 0.0011,
+      "step": 8612
+    },
+    {
+      "epoch": 36.82,
+      "learning_rate": 9.941619955850798e-05,
+      "loss": 0.0014,
+      "step": 8616
+    },
+    {
+      "epoch": 36.84,
+      "learning_rate": 9.941442248524981e-05,
+      "loss": 0.0029,
+      "step": 8620
+    },
+    {
+      "epoch": 36.85,
+      "learning_rate": 9.94126427273492e-05,
+      "loss": 0.0015,
+      "step": 8624
+    },
+    {
+      "epoch": 36.87,
+      "learning_rate": 9.941086028490284e-05,
+      "loss": 0.0011,
+      "step": 8628
+    },
+    {
+      "epoch": 36.89,
+      "learning_rate": 9.940907515800755e-05,
+      "loss": 0.0017,
+      "step": 8632
+    },
+    {
+      "epoch": 36.91,
+      "learning_rate": 9.940728734676035e-05,
+      "loss": 0.0012,
+      "step": 8636
+    },
+    {
+      "epoch": 36.92,
+      "learning_rate": 9.940549685125834e-05,
+      "loss": 0.0006,
+      "step": 8640
+    },
+    {
+      "epoch": 36.94,
+      "learning_rate": 9.940370367159881e-05,
+      "loss": 0.0022,
+      "step": 8644
+    },
+    {
+      "epoch": 36.96,
+      "learning_rate": 9.940190780787917e-05,
+      "loss": 0.0028,
+      "step": 8648
+    },
+    {
+      "epoch": 36.97,
+      "learning_rate": 9.9400109260197e-05,
+      "loss": 0.0019,
+      "step": 8652
+    },
+    {
+      "epoch": 36.99,
+      "learning_rate": 9.939830802865002e-05,
+      "loss": 0.002,
+      "step": 8656
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 9.939650411333606e-05,
+      "loss": 0.0009,
+      "step": 8660
+    },
+    {
+      "epoch": 37.03,
+      "learning_rate": 9.939469751435315e-05,
+      "loss": 0.0087,
+      "step": 8664
+    },
+    {
+      "epoch": 37.04,
+      "learning_rate": 9.939288823179945e-05,
+      "loss": 0.0025,
+      "step": 8668
+    },
+    {
+      "epoch": 37.06,
+      "learning_rate": 9.939107626577324e-05,
+      "loss": 0.001,
+      "step": 8672
+    },
+    {
+      "epoch": 37.08,
+      "learning_rate": 9.938926161637294e-05,
+      "loss": 0.0009,
+      "step": 8676
+    },
+    {
+      "epoch": 37.09,
+      "learning_rate": 9.938744428369717e-05,
+      "loss": 0.0068,
+      "step": 8680
+    },
+    {
+      "epoch": 37.11,
+      "learning_rate": 9.938562426784468e-05,
+      "loss": 0.0018,
+      "step": 8684
+    },
+    {
+      "epoch": 37.13,
+      "learning_rate": 9.93838015689143e-05,
+      "loss": 0.0011,
+      "step": 8688
+    },
+    {
+      "epoch": 37.15,
+      "learning_rate": 9.93819761870051e-05,
+      "loss": 0.0017,
+      "step": 8692
+    },
+    {
+      "epoch": 37.16,
+      "learning_rate": 9.938014812221622e-05,
+      "loss": 0.0017,
+      "step": 8696
+    },
+    {
+      "epoch": 37.18,
+      "learning_rate": 9.9378317374647e-05,
+      "loss": 0.0018,
+      "step": 8700
+    },
+    {
+      "epoch": 37.2,
+      "learning_rate": 9.937648394439689e-05,
+      "loss": 0.0014,
+      "step": 8704
+    },
+    {
+      "epoch": 37.21,
+      "learning_rate": 9.937464783156549e-05,
+      "loss": 0.0024,
+      "step": 8708
+    },
+    {
+      "epoch": 37.23,
+      "learning_rate": 9.937280903625259e-05,
+      "loss": 0.0018,
+      "step": 8712
+    },
+    {
+      "epoch": 37.25,
+      "learning_rate": 9.937096755855806e-05,
+      "loss": 0.0004,
+      "step": 8716
+    },
+    {
+      "epoch": 37.26,
+      "learning_rate": 9.936912339858194e-05,
+      "loss": 0.0026,
+      "step": 8720
+    },
+    {
+      "epoch": 37.28,
+      "learning_rate": 9.936727655642443e-05,
+      "loss": 0.0006,
+      "step": 8724
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 9.936542703218589e-05,
+      "loss": 0.0003,
+      "step": 8728
+    },
+    {
+      "epoch": 37.32,
+      "learning_rate": 9.936357482596679e-05,
+      "loss": 0.0018,
+      "step": 8732
+    },
+    {
+      "epoch": 37.33,
+      "learning_rate": 9.936171993786773e-05,
+      "loss": 0.001,
+      "step": 8736
+    },
+    {
+      "epoch": 37.35,
+      "learning_rate": 9.935986236798953e-05,
+      "loss": 0.0009,
+      "step": 8740
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 9.935800211643307e-05,
+      "loss": 0.0017,
+      "step": 8744
+    },
+    {
+      "epoch": 37.38,
+      "learning_rate": 9.935613918329945e-05,
+      "loss": 0.0028,
+      "step": 8748
+    },
+    {
+      "epoch": 37.4,
+      "learning_rate": 9.935427356868985e-05,
+      "loss": 0.0015,
+      "step": 8752
+    },
+    {
+      "epoch": 37.42,
+      "learning_rate": 9.935240527270566e-05,
+      "loss": 0.0004,
+      "step": 8756
+    },
+    {
+      "epoch": 37.44,
+      "learning_rate": 9.935053429544835e-05,
+      "loss": 0.0015,
+      "step": 8760
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 9.93486606370196e-05,
+      "loss": 0.001,
+      "step": 8764
+    },
+    {
+      "epoch": 37.47,
+      "learning_rate": 9.934678429752117e-05,
+      "loss": 0.0014,
+      "step": 8768
+    },
+    {
+      "epoch": 37.49,
+      "learning_rate": 9.934490527705502e-05,
+      "loss": 0.0007,
+      "step": 8772
+    },
+    {
+      "epoch": 37.5,
+      "learning_rate": 9.934302357572324e-05,
+      "loss": 0.0042,
+      "step": 8776
+    },
+    {
+      "epoch": 37.52,
+      "learning_rate": 9.934113919362807e-05,
+      "loss": 0.0016,
+      "step": 8780
+    },
+    {
+      "epoch": 37.54,
+      "learning_rate": 9.933925213087184e-05,
+      "loss": 0.002,
+      "step": 8784
+    },
+    {
+      "epoch": 37.56,
+      "learning_rate": 9.933736238755714e-05,
+      "loss": 0.0013,
+      "step": 8788
+    },
+    {
+      "epoch": 37.57,
+      "learning_rate": 9.933546996378657e-05,
+      "loss": 0.0036,
+      "step": 8792
+    },
+    {
+      "epoch": 37.59,
+      "learning_rate": 9.9333574859663e-05,
+      "loss": 0.002,
+      "step": 8796
+    },
+    {
+      "epoch": 37.61,
+      "learning_rate": 9.933167707528934e-05,
+      "loss": 0.0006,
+      "step": 8800
+    },
+    {
+      "epoch": 37.62,
+      "learning_rate": 9.932977661076873e-05,
+      "loss": 0.0012,
+      "step": 8804
+    },
+    {
+      "epoch": 37.64,
+      "learning_rate": 9.932787346620443e-05,
+      "loss": 0.0007,
+      "step": 8808
+    },
+    {
+      "epoch": 37.66,
+      "learning_rate": 9.932596764169979e-05,
+      "loss": 0.001,
+      "step": 8812
+    },
+    {
+      "epoch": 37.68,
+      "learning_rate": 9.93240591373584e-05,
+      "loss": 0.0026,
+      "step": 8816
+    },
+    {
+      "epoch": 37.69,
+      "learning_rate": 9.932214795328391e-05,
+      "loss": 0.0009,
+      "step": 8820
+    },
+    {
+      "epoch": 37.71,
+      "learning_rate": 9.932023408958018e-05,
+      "loss": 0.0014,
+      "step": 8824
+    },
+    {
+      "epoch": 37.73,
+      "learning_rate": 9.931831754635117e-05,
+      "loss": 0.0003,
+      "step": 8828
+    },
+    {
+      "epoch": 37.74,
+      "learning_rate": 9.9316398323701e-05,
+      "loss": 0.0014,
+      "step": 8832
+    },
+    {
+      "epoch": 37.76,
+      "learning_rate": 9.931447642173398e-05,
+      "loss": 0.0021,
+      "step": 8836
+    },
+    {
+      "epoch": 37.78,
+      "learning_rate": 9.93125518405545e-05,
+      "loss": 0.0007,
+      "step": 8840
+    },
+    {
+      "epoch": 37.79,
+      "learning_rate": 9.93106245802671e-05,
+      "loss": 0.001,
+      "step": 8844
+    },
+    {
+      "epoch": 37.81,
+      "learning_rate": 9.93086946409765e-05,
+      "loss": 0.0007,
+      "step": 8848
+    },
+    {
+      "epoch": 37.83,
+      "learning_rate": 9.930676202278756e-05,
+      "loss": 0.0011,
+      "step": 8852
+    },
+    {
+      "epoch": 37.85,
+      "learning_rate": 9.930482672580528e-05,
+      "loss": 0.0006,
+      "step": 8856
+    },
+    {
+      "epoch": 37.86,
+      "learning_rate": 9.930288875013479e-05,
+      "loss": 0.0017,
+      "step": 8860
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 9.930094809588139e-05,
+      "loss": 0.0027,
+      "step": 8864
+    },
+    {
+      "epoch": 37.9,
+      "learning_rate": 9.92990047631505e-05,
+      "loss": 0.0022,
+      "step": 8868
+    },
+    {
+      "epoch": 37.91,
+      "learning_rate": 9.929705875204771e-05,
+      "loss": 0.0028,
+      "step": 8872
+    },
+    {
+      "epoch": 37.93,
+      "learning_rate": 9.929511006267876e-05,
+      "loss": 0.0015,
+      "step": 8876
+    },
+    {
+      "epoch": 37.95,
+      "learning_rate": 9.92931586951495e-05,
+      "loss": 0.0009,
+      "step": 8880
+    },
+    {
+      "epoch": 37.97,
+      "learning_rate": 9.929120464956594e-05,
+      "loss": 0.0012,
+      "step": 8884
+    },
+    {
+      "epoch": 37.98,
+      "learning_rate": 9.928924792603424e-05,
+      "loss": 0.0008,
+      "step": 8888
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 9.928728852466073e-05,
+      "loss": 0.0062,
+      "step": 8892
+    },
+    {
+      "epoch": 38.02,
+      "learning_rate": 9.928532644555186e-05,
+      "loss": 0.0005,
+      "step": 8896
+    },
+    {
+      "epoch": 38.03,
+      "learning_rate": 9.928336168881423e-05,
+      "loss": 0.0003,
+      "step": 8900
+    },
+    {
+      "epoch": 38.05,
+      "learning_rate": 9.928139425455455e-05,
+      "loss": 0.0015,
+      "step": 8904
+    },
+    {
+      "epoch": 38.07,
+      "learning_rate": 9.927942414287974e-05,
+      "loss": 0.0005,
+      "step": 8908
+    },
+    {
+      "epoch": 38.09,
+      "learning_rate": 9.927745135389684e-05,
+      "loss": 0.0007,
+      "step": 8912
+    },
+    {
+      "epoch": 38.1,
+      "learning_rate": 9.927547588771299e-05,
+      "loss": 0.0008,
+      "step": 8916
+    },
+    {
+      "epoch": 38.12,
+      "learning_rate": 9.927349774443555e-05,
+      "loss": 0.0008,
+      "step": 8920
+    },
+    {
+      "epoch": 38.14,
+      "learning_rate": 9.9271516924172e-05,
+      "loss": 0.0013,
+      "step": 8924
+    },
+    {
+      "epoch": 38.15,
+      "learning_rate": 9.926953342702991e-05,
+      "loss": 0.0017,
+      "step": 8928
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 9.926754725311709e-05,
+      "loss": 0.0021,
+      "step": 8932
+    },
+    {
+      "epoch": 38.19,
+      "learning_rate": 9.926555840254144e-05,
+      "loss": 0.0008,
+      "step": 8936
+    },
+    {
+      "epoch": 38.21,
+      "learning_rate": 9.926356687541098e-05,
+      "loss": 0.0002,
+      "step": 8940
+    },
+    {
+      "epoch": 38.22,
+      "learning_rate": 9.926157267183394e-05,
+      "loss": 0.0037,
+      "step": 8944
+    },
+    {
+      "epoch": 38.24,
+      "learning_rate": 9.925957579191863e-05,
+      "loss": 0.0024,
+      "step": 8948
+    },
+    {
+      "epoch": 38.26,
+      "learning_rate": 9.925757623577358e-05,
+      "loss": 0.001,
+      "step": 8952
+    },
+    {
+      "epoch": 38.27,
+      "learning_rate": 9.92555740035074e-05,
+      "loss": 0.0007,
+      "step": 8956
+    },
+    {
+      "epoch": 38.29,
+      "learning_rate": 9.925356909522887e-05,
+      "loss": 0.0007,
+      "step": 8960
+    },
+    {
+      "epoch": 38.31,
+      "learning_rate": 9.925156151104692e-05,
+      "loss": 0.0011,
+      "step": 8964
+    },
+    {
+      "epoch": 38.32,
+      "learning_rate": 9.924955125107063e-05,
+      "loss": 0.0011,
+      "step": 8968
+    },
+    {
+      "epoch": 38.34,
+      "learning_rate": 9.924753831540918e-05,
+      "loss": 0.0019,
+      "step": 8972
+    },
+    {
+      "epoch": 38.36,
+      "learning_rate": 9.924552270417198e-05,
+      "loss": 0.0008,
+      "step": 8976
+    },
+    {
+      "epoch": 38.38,
+      "learning_rate": 9.92435044174685e-05,
+      "loss": 0.0014,
+      "step": 8980
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 9.92414834554084e-05,
+      "loss": 0.0024,
+      "step": 8984
+    },
+    {
+      "epoch": 38.41,
+      "learning_rate": 9.92394598181015e-05,
+      "loss": 0.0007,
+      "step": 8988
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 9.923743350565771e-05,
+      "loss": 0.0006,
+      "step": 8992
+    },
+    {
+      "epoch": 38.44,
+      "learning_rate": 9.923540451818713e-05,
+      "loss": 0.0015,
+      "step": 8996
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 9.923337285579999e-05,
+      "loss": 0.0006,
+      "step": 9000
+    },
+    {
+      "epoch": 38.46,
+      "eval_exact_match": 0.501039501039501,
+      "eval_loss": 0.7777053117752075,
+      "eval_runtime": 134.1347,
+      "eval_samples_per_second": 7.172,
+      "step": 9000
+    },
+    {
+      "epoch": 38.48,
+      "learning_rate": 9.923133851860669e-05,
+      "loss": 0.0006,
+      "step": 9004
+    },
+    {
+      "epoch": 38.5,
+      "learning_rate": 9.922930150671773e-05,
+      "loss": 0.0008,
+      "step": 9008
+    },
+    {
+      "epoch": 38.51,
+      "learning_rate": 9.922726182024378e-05,
+      "loss": 0.0009,
+      "step": 9012
+    },
+    {
+      "epoch": 38.53,
+      "learning_rate": 9.922521945929567e-05,
+      "loss": 0.0014,
+      "step": 9016
+    },
+    {
+      "epoch": 38.55,
+      "learning_rate": 9.922317442398433e-05,
+      "loss": 0.0031,
+      "step": 9020
+    },
+    {
+      "epoch": 38.56,
+      "learning_rate": 9.92211267144209e-05,
+      "loss": 0.0017,
+      "step": 9024
+    },
+    {
+      "epoch": 38.58,
+      "learning_rate": 9.921907633071661e-05,
+      "loss": 0.0018,
+      "step": 9028
+    },
+    {
+      "epoch": 38.6,
+      "learning_rate": 9.921702327298286e-05,
+      "loss": 0.001,
+      "step": 9032
+    },
+    {
+      "epoch": 38.62,
+      "learning_rate": 9.92149675413312e-05,
+      "loss": 0.0011,
+      "step": 9036
+    },
+    {
+      "epoch": 38.63,
+      "learning_rate": 9.921290913587328e-05,
+      "loss": 0.0012,
+      "step": 9040
+    },
+    {
+      "epoch": 38.65,
+      "learning_rate": 9.921084805672098e-05,
+      "loss": 0.0023,
+      "step": 9044
+    },
+    {
+      "epoch": 38.67,
+      "learning_rate": 9.920878430398626e-05,
+      "loss": 0.0031,
+      "step": 9048
+    },
+    {
+      "epoch": 38.68,
+      "learning_rate": 9.920671787778124e-05,
+      "loss": 0.0007,
+      "step": 9052
+    },
+    {
+      "epoch": 38.7,
+      "learning_rate": 9.920464877821818e-05,
+      "loss": 0.0012,
+      "step": 9056
+    },
+    {
+      "epoch": 38.72,
+      "learning_rate": 9.920257700540949e-05,
+      "loss": 0.0025,
+      "step": 9060
+    },
+    {
+      "epoch": 38.74,
+      "learning_rate": 9.920050255946774e-05,
+      "loss": 0.0034,
+      "step": 9064
+    },
+    {
+      "epoch": 38.75,
+      "learning_rate": 9.919842544050561e-05,
+      "loss": 0.0022,
+      "step": 9068
+    },
+    {
+      "epoch": 38.77,
+      "learning_rate": 9.919634564863599e-05,
+      "loss": 0.0008,
+      "step": 9072
+    },
+    {
+      "epoch": 38.79,
+      "learning_rate": 9.919426318397185e-05,
+      "loss": 0.0012,
+      "step": 9076
+    },
+    {
+      "epoch": 38.8,
+      "learning_rate": 9.91921780466263e-05,
+      "loss": 0.0018,
+      "step": 9080
+    },
+    {
+      "epoch": 38.82,
+      "learning_rate": 9.919009023671268e-05,
+      "loss": 0.0012,
+      "step": 9084
+    },
+    {
+      "epoch": 38.84,
+      "learning_rate": 9.918799975434436e-05,
+      "loss": 0.0006,
+      "step": 9088
+    },
+    {
+      "epoch": 38.85,
+      "learning_rate": 9.918590659963497e-05,
+      "loss": 0.0011,
+      "step": 9092
+    },
+    {
+      "epoch": 38.87,
+      "learning_rate": 9.91838107726982e-05,
+      "loss": 0.0006,
+      "step": 9096
+    },
+    {
+      "epoch": 38.89,
+      "learning_rate": 9.918171227364791e-05,
+      "loss": 0.0004,
+      "step": 9100
+    },
+    {
+      "epoch": 38.91,
+      "learning_rate": 9.917961110259811e-05,
+      "loss": 0.0009,
+      "step": 9104
+    },
+    {
+      "epoch": 38.92,
+      "learning_rate": 9.917750725966297e-05,
+      "loss": 0.0012,
+      "step": 9108
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 9.917540074495679e-05,
+      "loss": 0.0006,
+      "step": 9112
+    },
+    {
+      "epoch": 38.96,
+      "learning_rate": 9.917329155859401e-05,
+      "loss": 0.0015,
+      "step": 9116
+    },
+    {
+      "epoch": 38.97,
+      "learning_rate": 9.917117970068919e-05,
+      "loss": 0.0003,
+      "step": 9120
+    },
+    {
+      "epoch": 38.99,
+      "learning_rate": 9.91690651713571e-05,
+      "loss": 0.002,
+      "step": 9124
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 9.916694797071265e-05,
+      "loss": 0.0006,
+      "step": 9128
+    },
+    {
+      "epoch": 39.03,
+      "learning_rate": 9.91648280988708e-05,
+      "loss": 0.0005,
+      "step": 9132
+    },
+    {
+      "epoch": 39.04,
+      "learning_rate": 9.916270555594674e-05,
+      "loss": 0.0003,
+      "step": 9136
+    },
+    {
+      "epoch": 39.06,
+      "learning_rate": 9.91605803420558e-05,
+      "loss": 0.0011,
+      "step": 9140
+    },
+    {
+      "epoch": 39.08,
+      "learning_rate": 9.915845245731345e-05,
+      "loss": 0.0024,
+      "step": 9144
+    },
+    {
+      "epoch": 39.09,
+      "learning_rate": 9.915632190183528e-05,
+      "loss": 0.0017,
+      "step": 9148
+    },
+    {
+      "epoch": 39.11,
+      "learning_rate": 9.915418867573703e-05,
+      "loss": 0.001,
+      "step": 9152
+    },
+    {
+      "epoch": 39.13,
+      "learning_rate": 9.915205277913462e-05,
+      "loss": 0.0003,
+      "step": 9156
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 9.914991421214408e-05,
+      "loss": 0.0006,
+      "step": 9160
+    },
+    {
+      "epoch": 39.16,
+      "learning_rate": 9.91477729748816e-05,
+      "loss": 0.0002,
+      "step": 9164
+    },
+    {
+      "epoch": 39.18,
+      "learning_rate": 9.914562906746353e-05,
+      "loss": 0.0021,
+      "step": 9168
+    },
+    {
+      "epoch": 39.2,
+      "learning_rate": 9.914348249000629e-05,
+      "loss": 0.0006,
+      "step": 9172
+    },
+    {
+      "epoch": 39.21,
+      "learning_rate": 9.914133324262656e-05,
+      "loss": 0.001,
+      "step": 9176
+    },
+    {
+      "epoch": 39.23,
+      "learning_rate": 9.913918132544107e-05,
+      "loss": 0.002,
+      "step": 9180
+    },
+    {
+      "epoch": 39.25,
+      "learning_rate": 9.913702673856675e-05,
+      "loss": 0.0009,
+      "step": 9184
+    },
+    {
+      "epoch": 39.26,
+      "learning_rate": 9.913486948212066e-05,
+      "loss": 0.0021,
+      "step": 9188
+    },
+    {
+      "epoch": 39.28,
+      "learning_rate": 9.913270955622e-05,
+      "loss": 0.0021,
+      "step": 9192
+    },
+    {
+      "epoch": 39.3,
+      "learning_rate": 9.91305469609821e-05,
+      "loss": 0.0012,
+      "step": 9196
+    },
+    {
+      "epoch": 39.32,
+      "learning_rate": 9.912838169652446e-05,
+      "loss": 0.0025,
+      "step": 9200
+    },
+    {
+      "epoch": 39.33,
+      "learning_rate": 9.912621376296474e-05,
+      "loss": 0.001,
+      "step": 9204
+    },
+    {
+      "epoch": 39.35,
+      "learning_rate": 9.91240431604207e-05,
+      "loss": 0.0045,
+      "step": 9208
+    },
+    {
+      "epoch": 39.37,
+      "learning_rate": 9.912186988901025e-05,
+      "loss": 0.0017,
+      "step": 9212
+    },
+    {
+      "epoch": 39.38,
+      "learning_rate": 9.91196939488515e-05,
+      "loss": 0.0008,
+      "step": 9216
+    },
+    {
+      "epoch": 39.4,
+      "learning_rate": 9.911751534006264e-05,
+      "loss": 0.0014,
+      "step": 9220
+    },
+    {
+      "epoch": 39.42,
+      "learning_rate": 9.911533406276203e-05,
+      "loss": 0.001,
+      "step": 9224
+    },
+    {
+      "epoch": 39.44,
+      "learning_rate": 9.911315011706822e-05,
+      "loss": 0.0025,
+      "step": 9228
+    },
+    {
+      "epoch": 39.45,
+      "learning_rate": 9.911096350309979e-05,
+      "loss": 0.003,
+      "step": 9232
+    },
+    {
+      "epoch": 39.47,
+      "learning_rate": 9.910877422097559e-05,
+      "loss": 0.0009,
+      "step": 9236
+    },
+    {
+      "epoch": 39.49,
+      "learning_rate": 9.910658227081457e-05,
+      "loss": 0.0009,
+      "step": 9240
+    },
+    {
+      "epoch": 39.5,
+      "learning_rate": 9.910438765273578e-05,
+      "loss": 0.0023,
+      "step": 9244
+    },
+    {
+      "epoch": 39.52,
+      "learning_rate": 9.910219036685845e-05,
+      "loss": 0.0013,
+      "step": 9248
+    },
+    {
+      "epoch": 39.54,
+      "learning_rate": 9.9099990413302e-05,
+      "loss": 0.002,
+      "step": 9252
+    },
+    {
+      "epoch": 39.56,
+      "learning_rate": 9.90977877921859e-05,
+      "loss": 0.0004,
+      "step": 9256
+    },
+    {
+      "epoch": 39.57,
+      "learning_rate": 9.909558250362986e-05,
+      "loss": 0.0003,
+      "step": 9260
+    },
+    {
+      "epoch": 39.59,
+      "learning_rate": 9.909337454775367e-05,
+      "loss": 0.0011,
+      "step": 9264
+    },
+    {
+      "epoch": 39.61,
+      "learning_rate": 9.909116392467728e-05,
+      "loss": 0.0019,
+      "step": 9268
+    },
+    {
+      "epoch": 39.62,
+      "learning_rate": 9.908895063452079e-05,
+      "loss": 0.0007,
+      "step": 9272
+    },
+    {
+      "epoch": 39.64,
+      "learning_rate": 9.908673467740448e-05,
+      "loss": 0.001,
+      "step": 9276
+    },
+    {
+      "epoch": 39.66,
+      "learning_rate": 9.90845160534487e-05,
+      "loss": 0.0012,
+      "step": 9280
+    },
+    {
+      "epoch": 39.68,
+      "learning_rate": 9.908229476277401e-05,
+      "loss": 0.0009,
+      "step": 9284
+    },
+    {
+      "epoch": 39.69,
+      "learning_rate": 9.908007080550109e-05,
+      "loss": 0.0016,
+      "step": 9288
+    },
+    {
+      "epoch": 39.71,
+      "learning_rate": 9.907784418175075e-05,
+      "loss": 0.0013,
+      "step": 9292
+    },
+    {
+      "epoch": 39.73,
+      "learning_rate": 9.907561489164397e-05,
+      "loss": 0.002,
+      "step": 9296
+    },
+    {
+      "epoch": 39.74,
+      "learning_rate": 9.907338293530187e-05,
+      "loss": 0.0015,
+      "step": 9300
+    },
+    {
+      "epoch": 39.76,
+      "learning_rate": 9.90711483128457e-05,
+      "loss": 0.0024,
+      "step": 9304
+    },
+    {
+      "epoch": 39.78,
+      "learning_rate": 9.906891102439688e-05,
+      "loss": 0.0007,
+      "step": 9308
+    },
+    {
+      "epoch": 39.79,
+      "learning_rate": 9.906667107007694e-05,
+      "loss": 0.0021,
+      "step": 9312
+    },
+    {
+      "epoch": 39.81,
+      "learning_rate": 9.90644284500076e-05,
+      "loss": 0.0008,
+      "step": 9316
+    },
+    {
+      "epoch": 39.83,
+      "learning_rate": 9.906218316431066e-05,
+      "loss": 0.0034,
+      "step": 9320
+    },
+    {
+      "epoch": 39.85,
+      "learning_rate": 9.905993521310816e-05,
+      "loss": 0.0005,
+      "step": 9324
+    },
+    {
+      "epoch": 39.86,
+      "learning_rate": 9.905768459652218e-05,
+      "loss": 0.0036,
+      "step": 9328
+    },
+    {
+      "epoch": 39.88,
+      "learning_rate": 9.905543131467501e-05,
+      "loss": 0.0008,
+      "step": 9332
+    },
+    {
+      "epoch": 39.9,
+      "learning_rate": 9.905317536768908e-05,
+      "loss": 0.0011,
+      "step": 9336
+    },
+    {
+      "epoch": 39.91,
+      "learning_rate": 9.905091675568694e-05,
+      "loss": 0.0012,
+      "step": 9340
+    },
+    {
+      "epoch": 39.93,
+      "learning_rate": 9.904865547879131e-05,
+      "loss": 0.0009,
+      "step": 9344
+    },
+    {
+      "epoch": 39.95,
+      "learning_rate": 9.904639153712503e-05,
+      "loss": 0.0006,
+      "step": 9348
+    },
+    {
+      "epoch": 39.97,
+      "learning_rate": 9.90441249308111e-05,
+      "loss": 0.002,
+      "step": 9352
+    },
+    {
+      "epoch": 39.98,
+      "learning_rate": 9.904185565997269e-05,
+      "loss": 0.0036,
+      "step": 9356
+    },
+    {
+      "epoch": 40.0,
+      "learning_rate": 9.903958372473303e-05,
+      "loss": 0.0006,
+      "step": 9360
+    },
+    {
+      "epoch": 40.02,
+      "learning_rate": 9.903730912521561e-05,
+      "loss": 0.001,
+      "step": 9364
+    },
+    {
+      "epoch": 40.03,
+      "learning_rate": 9.903503186154399e-05,
+      "loss": 0.0019,
+      "step": 9368
+    },
+    {
+      "epoch": 40.05,
+      "learning_rate": 9.903275193384188e-05,
+      "loss": 0.0008,
+      "step": 9372
+    },
+    {
+      "epoch": 40.07,
+      "learning_rate": 9.903046934223315e-05,
+      "loss": 0.002,
+      "step": 9376
+    },
+    {
+      "epoch": 40.09,
+      "learning_rate": 9.90281840868418e-05,
+      "loss": 0.0023,
+      "step": 9380
+    },
+    {
+      "epoch": 40.1,
+      "learning_rate": 9.902589616779201e-05,
+      "loss": 0.0011,
+      "step": 9384
+    },
+    {
+      "epoch": 40.12,
+      "learning_rate": 9.902360558520807e-05,
+      "loss": 0.0016,
+      "step": 9388
+    },
+    {
+      "epoch": 40.14,
+      "learning_rate": 9.902131233921443e-05,
+      "loss": 0.0041,
+      "step": 9392
+    },
+    {
+      "epoch": 40.15,
+      "learning_rate": 9.901901642993567e-05,
+      "loss": 0.0029,
+      "step": 9396
+    },
+    {
+      "epoch": 40.17,
+      "learning_rate": 9.901671785749654e-05,
+      "loss": 0.0011,
+      "step": 9400
+    },
+    {
+      "epoch": 40.19,
+      "learning_rate": 9.90144166220219e-05,
+      "loss": 0.001,
+      "step": 9404
+    },
+    {
+      "epoch": 40.21,
+      "learning_rate": 9.901211272363679e-05,
+      "loss": 0.0007,
+      "step": 9408
+    },
+    {
+      "epoch": 40.22,
+      "learning_rate": 9.900980616246637e-05,
+      "loss": 0.0021,
+      "step": 9412
+    },
+    {
+      "epoch": 40.24,
+      "learning_rate": 9.900749693863596e-05,
+      "loss": 0.0009,
+      "step": 9416
+    },
+    {
+      "epoch": 40.26,
+      "learning_rate": 9.900518505227101e-05,
+      "loss": 0.0004,
+      "step": 9420
+    },
+    {
+      "epoch": 40.27,
+      "learning_rate": 9.900287050349712e-05,
+      "loss": 0.001,
+      "step": 9424
+    },
+    {
+      "epoch": 40.29,
+      "learning_rate": 9.900055329244004e-05,
+      "loss": 0.0022,
+      "step": 9428
+    },
+    {
+      "epoch": 40.31,
+      "learning_rate": 9.89982334192257e-05,
+      "loss": 0.0014,
+      "step": 9432
+    },
+    {
+      "epoch": 40.32,
+      "learning_rate": 9.899591088398007e-05,
+      "loss": 0.0029,
+      "step": 9436
+    },
+    {
+      "epoch": 40.34,
+      "learning_rate": 9.899358568682938e-05,
+      "loss": 0.0004,
+      "step": 9440
+    },
+    {
+      "epoch": 40.36,
+      "learning_rate": 9.899125782789993e-05,
+      "loss": 0.0006,
+      "step": 9444
+    },
+    {
+      "epoch": 40.38,
+      "learning_rate": 9.898892730731821e-05,
+      "loss": 0.0019,
+      "step": 9448
+    },
+    {
+      "epoch": 40.39,
+      "learning_rate": 9.898659412521082e-05,
+      "loss": 0.0016,
+      "step": 9452
+    },
+    {
+      "epoch": 40.41,
+      "learning_rate": 9.898425828170452e-05,
+      "loss": 0.0029,
+      "step": 9456
+    },
+    {
+      "epoch": 40.43,
+      "learning_rate": 9.898191977692623e-05,
+      "loss": 0.0018,
+      "step": 9460
+    },
+    {
+      "epoch": 40.44,
+      "learning_rate": 9.8979578611003e-05,
+      "loss": 0.0005,
+      "step": 9464
+    },
+    {
+      "epoch": 40.46,
+      "learning_rate": 9.8977234784062e-05,
+      "loss": 0.0007,
+      "step": 9468
+    },
+    {
+      "epoch": 40.48,
+      "learning_rate": 9.897488829623059e-05,
+      "loss": 0.0022,
+      "step": 9472
+    },
+    {
+      "epoch": 40.5,
+      "learning_rate": 9.897253914763622e-05,
+      "loss": 0.0034,
+      "step": 9476
+    },
+    {
+      "epoch": 40.51,
+      "learning_rate": 9.897018733840656e-05,
+      "loss": 0.0019,
+      "step": 9480
+    },
+    {
+      "epoch": 40.53,
+      "learning_rate": 9.896783286866935e-05,
+      "loss": 0.0009,
+      "step": 9484
+    },
+    {
+      "epoch": 40.55,
+      "learning_rate": 9.896547573855253e-05,
+      "loss": 0.0006,
+      "step": 9488
+    },
+    {
+      "epoch": 40.56,
+      "learning_rate": 9.896311594818415e-05,
+      "loss": 0.0004,
+      "step": 9492
+    },
+    {
+      "epoch": 40.58,
+      "learning_rate": 9.896075349769241e-05,
+      "loss": 0.0038,
+      "step": 9496
+    },
+    {
+      "epoch": 40.6,
+      "learning_rate": 9.895838838720565e-05,
+      "loss": 0.0023,
+      "step": 9500
+    },
+    {
+      "epoch": 40.62,
+      "learning_rate": 9.89560206168524e-05,
+      "loss": 0.0005,
+      "step": 9504
+    },
+    {
+      "epoch": 40.63,
+      "learning_rate": 9.895365018676128e-05,
+      "loss": 0.0008,
+      "step": 9508
+    },
+    {
+      "epoch": 40.65,
+      "learning_rate": 9.895127709706106e-05,
+      "loss": 0.0016,
+      "step": 9512
+    },
+    {
+      "epoch": 40.67,
+      "learning_rate": 9.894890134788068e-05,
+      "loss": 0.0013,
+      "step": 9516
+    },
+    {
+      "epoch": 40.68,
+      "learning_rate": 9.894652293934921e-05,
+      "loss": 0.0011,
+      "step": 9520
+    },
+    {
+      "epoch": 40.7,
+      "learning_rate": 9.894414187159587e-05,
+      "loss": 0.0006,
+      "step": 9524
+    },
+    {
+      "epoch": 40.72,
+      "learning_rate": 9.894175814475002e-05,
+      "loss": 0.0008,
+      "step": 9528
+    },
+    {
+      "epoch": 40.74,
+      "learning_rate": 9.893937175894117e-05,
+      "loss": 0.0008,
+      "step": 9532
+    },
+    {
+      "epoch": 40.75,
+      "learning_rate": 9.893698271429896e-05,
+      "loss": 0.001,
+      "step": 9536
+    },
+    {
+      "epoch": 40.77,
+      "learning_rate": 9.893459101095319e-05,
+      "loss": 0.0006,
+      "step": 9540
+    },
+    {
+      "epoch": 40.79,
+      "learning_rate": 9.893219664903382e-05,
+      "loss": 0.0009,
+      "step": 9544
+    },
+    {
+      "epoch": 40.8,
+      "learning_rate": 9.892979962867088e-05,
+      "loss": 0.0005,
+      "step": 9548
+    },
+    {
+      "epoch": 40.82,
+      "learning_rate": 9.892739994999466e-05,
+      "loss": 0.0013,
+      "step": 9552
+    },
+    {
+      "epoch": 40.84,
+      "learning_rate": 9.89249976131355e-05,
+      "loss": 0.0012,
+      "step": 9556
+    },
+    {
+      "epoch": 40.85,
+      "learning_rate": 9.89225926182239e-05,
+      "loss": 0.0007,
+      "step": 9560
+    },
+    {
+      "epoch": 40.87,
+      "learning_rate": 9.892018496539055e-05,
+      "loss": 0.0019,
+      "step": 9564
+    },
+    {
+      "epoch": 40.89,
+      "learning_rate": 9.891777465476627e-05,
+      "loss": 0.0026,
+      "step": 9568
+    },
+    {
+      "epoch": 40.91,
+      "learning_rate": 9.891536168648195e-05,
+      "loss": 0.0009,
+      "step": 9572
+    },
+    {
+      "epoch": 40.92,
+      "learning_rate": 9.891294606066875e-05,
+      "loss": 0.0008,
+      "step": 9576
+    },
+    {
+      "epoch": 40.94,
+      "learning_rate": 9.891052777745786e-05,
+      "loss": 0.0007,
+      "step": 9580
+    },
+    {
+      "epoch": 40.96,
+      "learning_rate": 9.89081068369807e-05,
+      "loss": 0.0015,
+      "step": 9584
+    },
+    {
+      "epoch": 40.97,
+      "learning_rate": 9.890568323936877e-05,
+      "loss": 0.0007,
+      "step": 9588
+    },
+    {
+      "epoch": 40.99,
+      "learning_rate": 9.890325698475377e-05,
+      "loss": 0.0007,
+      "step": 9592
+    },
+    {
+      "epoch": 41.01,
+      "learning_rate": 9.89008280732675e-05,
+      "loss": 0.0029,
+      "step": 9596
+    },
+    {
+      "epoch": 41.03,
+      "learning_rate": 9.889839650504191e-05,
+      "loss": 0.0007,
+      "step": 9600
+    },
+    {
+      "epoch": 41.04,
+      "learning_rate": 9.889596228020912e-05,
+      "loss": 0.0013,
+      "step": 9604
+    },
+    {
+      "epoch": 41.06,
+      "learning_rate": 9.889352539890136e-05,
+      "loss": 0.0062,
+      "step": 9608
+    },
+    {
+      "epoch": 41.08,
+      "learning_rate": 9.889108586125105e-05,
+      "loss": 0.0003,
+      "step": 9612
+    },
+    {
+      "epoch": 41.09,
+      "learning_rate": 9.888864366739073e-05,
+      "loss": 0.001,
+      "step": 9616
+    },
+    {
+      "epoch": 41.11,
+      "learning_rate": 9.888619881745305e-05,
+      "loss": 0.0005,
+      "step": 9620
+    },
+    {
+      "epoch": 41.13,
+      "learning_rate": 9.888375131157085e-05,
+      "loss": 0.0016,
+      "step": 9624
+    },
+    {
+      "epoch": 41.15,
+      "learning_rate": 9.888130114987712e-05,
+      "loss": 0.0013,
+      "step": 9628
+    },
+    {
+      "epoch": 41.16,
+      "learning_rate": 9.887884833250496e-05,
+      "loss": 0.0008,
+      "step": 9632
+    },
+    {
+      "epoch": 41.18,
+      "learning_rate": 9.887639285958763e-05,
+      "loss": 0.0025,
+      "step": 9636
+    },
+    {
+      "epoch": 41.2,
+      "learning_rate": 9.887393473125854e-05,
+      "loss": 0.001,
+      "step": 9640
+    },
+    {
+      "epoch": 41.21,
+      "learning_rate": 9.887147394765122e-05,
+      "loss": 0.0007,
+      "step": 9644
+    },
+    {
+      "epoch": 41.23,
+      "learning_rate": 9.886901050889938e-05,
+      "loss": 0.0018,
+      "step": 9648
+    },
+    {
+      "epoch": 41.25,
+      "learning_rate": 9.886654441513686e-05,
+      "loss": 0.0007,
+      "step": 9652
+    },
+    {
+      "epoch": 41.26,
+      "learning_rate": 9.886407566649762e-05,
+      "loss": 0.0013,
+      "step": 9656
+    },
+    {
+      "epoch": 41.28,
+      "learning_rate": 9.886160426311581e-05,
+      "loss": 0.0006,
+      "step": 9660
+    },
+    {
+      "epoch": 41.3,
+      "learning_rate": 9.885913020512568e-05,
+      "loss": 0.0021,
+      "step": 9664
+    },
+    {
+      "epoch": 41.32,
+      "learning_rate": 9.885665349266163e-05,
+      "loss": 0.0017,
+      "step": 9668
+    },
+    {
+      "epoch": 41.33,
+      "learning_rate": 9.885417412585826e-05,
+      "loss": 0.0004,
+      "step": 9672
+    },
+    {
+      "epoch": 41.35,
+      "learning_rate": 9.885169210485023e-05,
+      "loss": 0.0027,
+      "step": 9676
+    },
+    {
+      "epoch": 41.37,
+      "learning_rate": 9.884920742977243e-05,
+      "loss": 0.0028,
+      "step": 9680
+    },
+    {
+      "epoch": 41.38,
+      "learning_rate": 9.884672010075982e-05,
+      "loss": 0.0042,
+      "step": 9684
+    },
+    {
+      "epoch": 41.4,
+      "learning_rate": 9.884423011794753e-05,
+      "loss": 0.0006,
+      "step": 9688
+    },
+    {
+      "epoch": 41.42,
+      "learning_rate": 9.884173748147084e-05,
+      "loss": 0.0021,
+      "step": 9692
+    },
+    {
+      "epoch": 41.44,
+      "learning_rate": 9.883924219146519e-05,
+      "loss": 0.0016,
+      "step": 9696
+    },
+    {
+      "epoch": 41.45,
+      "learning_rate": 9.883674424806614e-05,
+      "loss": 0.0003,
+      "step": 9700
+    },
+    {
+      "epoch": 41.47,
+      "learning_rate": 9.88342436514094e-05,
+      "loss": 0.0007,
+      "step": 9704
+    },
+    {
+      "epoch": 41.49,
+      "learning_rate": 9.883174040163081e-05,
+      "loss": 0.0015,
+      "step": 9708
+    },
+    {
+      "epoch": 41.5,
+      "learning_rate": 9.882923449886641e-05,
+      "loss": 0.0033,
+      "step": 9712
+    },
+    {
+      "epoch": 41.52,
+      "learning_rate": 9.882672594325228e-05,
+      "loss": 0.0012,
+      "step": 9716
+    },
+    {
+      "epoch": 41.54,
+      "learning_rate": 9.882421473492478e-05,
+      "loss": 0.0012,
+      "step": 9720
+    },
+    {
+      "epoch": 41.56,
+      "learning_rate": 9.88217008740203e-05,
+      "loss": 0.0007,
+      "step": 9724
+    },
+    {
+      "epoch": 41.57,
+      "learning_rate": 9.881918436067541e-05,
+      "loss": 0.0033,
+      "step": 9728
+    },
+    {
+      "epoch": 41.59,
+      "learning_rate": 9.881666519502684e-05,
+      "loss": 0.001,
+      "step": 9732
+    },
+    {
+      "epoch": 41.61,
+      "learning_rate": 9.881414337721146e-05,
+      "loss": 0.0011,
+      "step": 9736
+    },
+    {
+      "epoch": 41.62,
+      "learning_rate": 9.881161890736627e-05,
+      "loss": 0.0009,
+      "step": 9740
+    },
+    {
+      "epoch": 41.64,
+      "learning_rate": 9.880909178562845e-05,
+      "loss": 0.0018,
+      "step": 9744
+    },
+    {
+      "epoch": 41.66,
+      "learning_rate": 9.880656201213525e-05,
+      "loss": 0.0008,
+      "step": 9748
+    },
+    {
+      "epoch": 41.68,
+      "learning_rate": 9.880402958702415e-05,
+      "loss": 0.0016,
+      "step": 9752
+    },
+    {
+      "epoch": 41.69,
+      "learning_rate": 9.88014945104327e-05,
+      "loss": 0.0017,
+      "step": 9756
+    },
+    {
+      "epoch": 41.71,
+      "learning_rate": 9.879895678249866e-05,
+      "loss": 0.0011,
+      "step": 9760
+    },
+    {
+      "epoch": 41.73,
+      "learning_rate": 9.879641640335988e-05,
+      "loss": 0.0003,
+      "step": 9764
+    },
+    {
+      "epoch": 41.74,
+      "learning_rate": 9.87938733731544e-05,
+      "loss": 0.0009,
+      "step": 9768
+    },
+    {
+      "epoch": 41.76,
+      "learning_rate": 9.879132769202035e-05,
+      "loss": 0.0007,
+      "step": 9772
+    },
+    {
+      "epoch": 41.78,
+      "learning_rate": 9.878877936009605e-05,
+      "loss": 0.0007,
+      "step": 9776
+    },
+    {
+      "epoch": 41.79,
+      "learning_rate": 9.878622837751997e-05,
+      "loss": 0.0005,
+      "step": 9780
+    },
+    {
+      "epoch": 41.81,
+      "learning_rate": 9.878367474443066e-05,
+      "loss": 0.0005,
+      "step": 9784
+    },
+    {
+      "epoch": 41.83,
+      "learning_rate": 9.87811184609669e-05,
+      "loss": 0.0012,
+      "step": 9788
+    },
+    {
+      "epoch": 41.85,
+      "learning_rate": 9.877855952726752e-05,
+      "loss": 0.0019,
+      "step": 9792
+    },
+    {
+      "epoch": 41.86,
+      "learning_rate": 9.877599794347158e-05,
+      "loss": 0.0015,
+      "step": 9796
+    },
+    {
+      "epoch": 41.88,
+      "learning_rate": 9.877343370971827e-05,
+      "loss": 0.0008,
+      "step": 9800
+    },
+    {
+      "epoch": 41.9,
+      "learning_rate": 9.877086682614686e-05,
+      "loss": 0.0003,
+      "step": 9804
+    },
+    {
+      "epoch": 41.91,
+      "learning_rate": 9.876829729289682e-05,
+      "loss": 0.0005,
+      "step": 9808
+    },
+    {
+      "epoch": 41.93,
+      "learning_rate": 9.876572511010775e-05,
+      "loss": 0.0009,
+      "step": 9812
+    },
+    {
+      "epoch": 41.95,
+      "learning_rate": 9.87631502779194e-05,
+      "loss": 0.002,
+      "step": 9816
+    },
+    {
+      "epoch": 41.97,
+      "learning_rate": 9.876057279647165e-05,
+      "loss": 0.001,
+      "step": 9820
+    },
+    {
+      "epoch": 41.98,
+      "learning_rate": 9.875799266590453e-05,
+      "loss": 0.0019,
+      "step": 9824
+    },
+    {
+      "epoch": 42.0,
+      "learning_rate": 9.875540988635824e-05,
+      "loss": 0.0003,
+      "step": 9828
+    },
+    {
+      "epoch": 42.02,
+      "learning_rate": 9.875282445797308e-05,
+      "loss": 0.0003,
+      "step": 9832
+    },
+    {
+      "epoch": 42.03,
+      "learning_rate": 9.875023638088951e-05,
+      "loss": 0.0013,
+      "step": 9836
+    },
+    {
+      "epoch": 42.05,
+      "learning_rate": 9.874764565524816e-05,
+      "loss": 0.0004,
+      "step": 9840
+    },
+    {
+      "epoch": 42.07,
+      "learning_rate": 9.874505228118976e-05,
+      "loss": 0.0006,
+      "step": 9844
+    },
+    {
+      "epoch": 42.09,
+      "learning_rate": 9.874245625885521e-05,
+      "loss": 0.0004,
+      "step": 9848
+    },
+    {
+      "epoch": 42.1,
+      "learning_rate": 9.873985758838555e-05,
+      "loss": 0.0003,
+      "step": 9852
+    },
+    {
+      "epoch": 42.12,
+      "learning_rate": 9.873725626992199e-05,
+      "loss": 0.0023,
+      "step": 9856
+    },
+    {
+      "epoch": 42.14,
+      "learning_rate": 9.873465230360582e-05,
+      "loss": 0.0017,
+      "step": 9860
+    },
+    {
+      "epoch": 42.15,
+      "learning_rate": 9.87320456895785e-05,
+      "loss": 0.0009,
+      "step": 9864
+    },
+    {
+      "epoch": 42.17,
+      "learning_rate": 9.87294364279817e-05,
+      "loss": 0.001,
+      "step": 9868
+    },
+    {
+      "epoch": 42.19,
+      "learning_rate": 9.872682451895714e-05,
+      "loss": 0.0005,
+      "step": 9872
+    },
+    {
+      "epoch": 42.21,
+      "learning_rate": 9.872420996264673e-05,
+      "loss": 0.0007,
+      "step": 9876
+    },
+    {
+      "epoch": 42.22,
+      "learning_rate": 9.872159275919253e-05,
+      "loss": 0.001,
+      "step": 9880
+    },
+    {
+      "epoch": 42.24,
+      "learning_rate": 9.87189729087367e-05,
+      "loss": 0.0009,
+      "step": 9884
+    },
+    {
+      "epoch": 42.26,
+      "learning_rate": 9.87163504114216e-05,
+      "loss": 0.0012,
+      "step": 9888
+    },
+    {
+      "epoch": 42.27,
+      "learning_rate": 9.871372526738971e-05,
+      "loss": 0.0007,
+      "step": 9892
+    },
+    {
+      "epoch": 42.29,
+      "learning_rate": 9.871109747678364e-05,
+      "loss": 0.0005,
+      "step": 9896
+    },
+    {
+      "epoch": 42.31,
+      "learning_rate": 9.870846703974614e-05,
+      "loss": 0.0009,
+      "step": 9900
+    },
+    {
+      "epoch": 42.32,
+      "learning_rate": 9.870583395642017e-05,
+      "loss": 0.0005,
+      "step": 9904
+    },
+    {
+      "epoch": 42.34,
+      "learning_rate": 9.870319822694874e-05,
+      "loss": 0.0007,
+      "step": 9908
+    },
+    {
+      "epoch": 42.36,
+      "learning_rate": 9.870055985147503e-05,
+      "loss": 0.0015,
+      "step": 9912
+    },
+    {
+      "epoch": 42.38,
+      "learning_rate": 9.869791883014243e-05,
+      "loss": 0.0015,
+      "step": 9916
+    },
+    {
+      "epoch": 42.39,
+      "learning_rate": 9.869527516309441e-05,
+      "loss": 0.0014,
+      "step": 9920
+    },
+    {
+      "epoch": 42.41,
+      "learning_rate": 9.869262885047459e-05,
+      "loss": 0.0015,
+      "step": 9924
+    },
+    {
+      "epoch": 42.43,
+      "learning_rate": 9.868997989242672e-05,
+      "loss": 0.0018,
+      "step": 9928
+    },
+    {
+      "epoch": 42.44,
+      "learning_rate": 9.868732828909476e-05,
+      "loss": 0.0018,
+      "step": 9932
+    },
+    {
+      "epoch": 42.46,
+      "learning_rate": 9.868467404062273e-05,
+      "loss": 0.0009,
+      "step": 9936
+    },
+    {
+      "epoch": 42.48,
+      "learning_rate": 9.868201714715489e-05,
+      "loss": 0.001,
+      "step": 9940
+    },
+    {
+      "epoch": 42.5,
+      "learning_rate": 9.867935760883551e-05,
+      "loss": 0.0009,
+      "step": 9944
+    },
+    {
+      "epoch": 42.51,
+      "learning_rate": 9.867669542580912e-05,
+      "loss": 0.0009,
+      "step": 9948
+    },
+    {
+      "epoch": 42.53,
+      "learning_rate": 9.867403059822035e-05,
+      "loss": 0.0008,
+      "step": 9952
+    },
+    {
+      "epoch": 42.55,
+      "learning_rate": 9.867136312621398e-05,
+      "loss": 0.0011,
+      "step": 9956
+    },
+    {
+      "epoch": 42.56,
+      "learning_rate": 9.866869300993496e-05,
+      "loss": 0.002,
+      "step": 9960
+    },
+    {
+      "epoch": 42.58,
+      "learning_rate": 9.866602024952828e-05,
+      "loss": 0.001,
+      "step": 9964
+    },
+    {
+      "epoch": 42.6,
+      "learning_rate": 9.866334484513923e-05,
+      "loss": 0.0008,
+      "step": 9968
+    },
+    {
+      "epoch": 42.62,
+      "learning_rate": 9.866066679691313e-05,
+      "loss": 0.0006,
+      "step": 9972
+    },
+    {
+      "epoch": 42.63,
+      "learning_rate": 9.865798610499546e-05,
+      "loss": 0.0012,
+      "step": 9976
+    },
+    {
+      "epoch": 42.65,
+      "learning_rate": 9.865530276953187e-05,
+      "loss": 0.0004,
+      "step": 9980
+    },
+    {
+      "epoch": 42.67,
+      "learning_rate": 9.865261679066814e-05,
+      "loss": 0.0033,
+      "step": 9984
+    },
+    {
+      "epoch": 42.68,
+      "learning_rate": 9.864992816855021e-05,
+      "loss": 0.0021,
+      "step": 9988
+    },
+    {
+      "epoch": 42.7,
+      "learning_rate": 9.864723690332415e-05,
+      "loss": 0.0006,
+      "step": 9992
+    },
+    {
+      "epoch": 42.72,
+      "learning_rate": 9.864454299513616e-05,
+      "loss": 0.002,
+      "step": 9996
+    },
+    {
+      "epoch": 42.74,
+      "learning_rate": 9.864184644413262e-05,
+      "loss": 0.0016,
+      "step": 10000
+    },
+    {
+      "epoch": 42.74,
+      "eval_exact_match": 0.5031185031185031,
+      "eval_loss": 0.7922365665435791,
+      "eval_runtime": 139.7876,
+      "eval_samples_per_second": 6.882,
+      "step": 10000
+    },
+    {
+      "epoch": 42.75,
+      "learning_rate": 9.863914725046e-05,
+      "loss": 0.0025,
+      "step": 10004
+    },
+    {
+      "epoch": 42.77,
+      "learning_rate": 9.863644541426496e-05,
+      "loss": 0.0007,
+      "step": 10008
+    },
+    {
+      "epoch": 42.79,
+      "learning_rate": 9.863374093569428e-05,
+      "loss": 0.0017,
+      "step": 10012
+    },
+    {
+      "epoch": 42.8,
+      "learning_rate": 9.863103381489493e-05,
+      "loss": 0.0003,
+      "step": 10016
+    },
+    {
+      "epoch": 42.82,
+      "learning_rate": 9.862832405201395e-05,
+      "loss": 0.0007,
+      "step": 10020
+    },
+    {
+      "epoch": 42.84,
+      "learning_rate": 9.862561164719855e-05,
+      "loss": 0.0023,
+      "step": 10024
+    },
+    {
+      "epoch": 42.85,
+      "learning_rate": 9.862289660059611e-05,
+      "loss": 0.0014,
+      "step": 10028
+    },
+    {
+      "epoch": 42.87,
+      "learning_rate": 9.862017891235416e-05,
+      "loss": 0.0014,
+      "step": 10032
+    },
+    {
+      "epoch": 42.89,
+      "learning_rate": 9.86174585826203e-05,
+      "loss": 0.0044,
+      "step": 10036
+    },
+    {
+      "epoch": 42.91,
+      "learning_rate": 9.861473561154236e-05,
+      "loss": 0.0004,
+      "step": 10040
+    },
+    {
+      "epoch": 42.92,
+      "learning_rate": 9.861200999926826e-05,
+      "loss": 0.0011,
+      "step": 10044
+    },
+    {
+      "epoch": 42.94,
+      "learning_rate": 9.860928174594608e-05,
+      "loss": 0.0016,
+      "step": 10048
+    },
+    {
+      "epoch": 42.96,
+      "learning_rate": 9.860655085172405e-05,
+      "loss": 0.0008,
+      "step": 10052
+    },
+    {
+      "epoch": 42.97,
+      "learning_rate": 9.860381731675055e-05,
+      "loss": 0.0011,
+      "step": 10056
+    },
+    {
+      "epoch": 42.99,
+      "learning_rate": 9.860108114117406e-05,
+      "loss": 0.0011,
+      "step": 10060
+    },
+    {
+      "epoch": 43.01,
+      "learning_rate": 9.859834232514324e-05,
+      "loss": 0.0004,
+      "step": 10064
+    },
+    {
+      "epoch": 43.03,
+      "learning_rate": 9.859560086880693e-05,
+      "loss": 0.0004,
+      "step": 10068
+    },
+    {
+      "epoch": 43.04,
+      "learning_rate": 9.859285677231402e-05,
+      "loss": 0.0013,
+      "step": 10072
+    },
+    {
+      "epoch": 43.06,
+      "learning_rate": 9.85901100358136e-05,
+      "loss": 0.0013,
+      "step": 10076
+    },
+    {
+      "epoch": 43.08,
+      "learning_rate": 9.858736065945493e-05,
+      "loss": 0.0012,
+      "step": 10080
+    },
+    {
+      "epoch": 43.09,
+      "learning_rate": 9.858460864338735e-05,
+      "loss": 0.0007,
+      "step": 10084
+    },
+    {
+      "epoch": 43.11,
+      "learning_rate": 9.858185398776039e-05,
+      "loss": 0.0031,
+      "step": 10088
+    },
+    {
+      "epoch": 43.13,
+      "learning_rate": 9.85790966927237e-05,
+      "loss": 0.0007,
+      "step": 10092
+    },
+    {
+      "epoch": 43.15,
+      "learning_rate": 9.85763367584271e-05,
+      "loss": 0.0012,
+      "step": 10096
+    },
+    {
+      "epoch": 43.16,
+      "learning_rate": 9.857357418502051e-05,
+      "loss": 0.0033,
+      "step": 10100
+    },
+    {
+      "epoch": 43.18,
+      "learning_rate": 9.857080897265403e-05,
+      "loss": 0.0031,
+      "step": 10104
+    },
+    {
+      "epoch": 43.2,
+      "learning_rate": 9.856804112147788e-05,
+      "loss": 0.001,
+      "step": 10108
+    },
+    {
+      "epoch": 43.21,
+      "learning_rate": 9.856527063164246e-05,
+      "loss": 0.0013,
+      "step": 10112
+    },
+    {
+      "epoch": 43.23,
+      "learning_rate": 9.856249750329827e-05,
+      "loss": 0.0018,
+      "step": 10116
+    },
+    {
+      "epoch": 43.25,
+      "learning_rate": 9.855972173659597e-05,
+      "loss": 0.0009,
+      "step": 10120
+    },
+    {
+      "epoch": 43.26,
+      "learning_rate": 9.855694333168638e-05,
+      "loss": 0.0004,
+      "step": 10124
+    },
+    {
+      "epoch": 43.28,
+      "learning_rate": 9.855416228872043e-05,
+      "loss": 0.0004,
+      "step": 10128
+    },
+    {
+      "epoch": 43.3,
+      "learning_rate": 9.855137860784923e-05,
+      "loss": 0.0002,
+      "step": 10132
+    },
+    {
+      "epoch": 43.32,
+      "learning_rate": 9.854859228922398e-05,
+      "loss": 0.0004,
+      "step": 10136
+    },
+    {
+      "epoch": 43.33,
+      "learning_rate": 9.85458033329961e-05,
+      "loss": 0.0013,
+      "step": 10140
+    },
+    {
+      "epoch": 43.35,
+      "learning_rate": 9.85430117393171e-05,
+      "loss": 0.0029,
+      "step": 10144
+    },
+    {
+      "epoch": 43.37,
+      "learning_rate": 9.854021750833864e-05,
+      "loss": 0.0011,
+      "step": 10148
+    },
+    {
+      "epoch": 43.38,
+      "learning_rate": 9.853742064021252e-05,
+      "loss": 0.0003,
+      "step": 10152
+    },
+    {
+      "epoch": 43.4,
+      "learning_rate": 9.85346211350907e-05,
+      "loss": 0.0012,
+      "step": 10156
+    },
+    {
+      "epoch": 43.42,
+      "learning_rate": 9.853181899312527e-05,
+      "loss": 0.0012,
+      "step": 10160
+    },
+    {
+      "epoch": 43.44,
+      "learning_rate": 9.852901421446849e-05,
+      "loss": 0.0008,
+      "step": 10164
+    },
+    {
+      "epoch": 43.45,
+      "learning_rate": 9.85262067992727e-05,
+      "loss": 0.0026,
+      "step": 10168
+    },
+    {
+      "epoch": 43.47,
+      "learning_rate": 9.852339674769047e-05,
+      "loss": 0.0006,
+      "step": 10172
+    },
+    {
+      "epoch": 43.49,
+      "learning_rate": 9.852058405987444e-05,
+      "loss": 0.0016,
+      "step": 10176
+    },
+    {
+      "epoch": 43.5,
+      "learning_rate": 9.85177687359774e-05,
+      "loss": 0.0003,
+      "step": 10180
+    },
+    {
+      "epoch": 43.52,
+      "learning_rate": 9.851495077615236e-05,
+      "loss": 0.0004,
+      "step": 10184
+    },
+    {
+      "epoch": 43.54,
+      "learning_rate": 9.851213018055237e-05,
+      "loss": 0.0014,
+      "step": 10188
+    },
+    {
+      "epoch": 43.56,
+      "learning_rate": 9.850930694933069e-05,
+      "loss": 0.0008,
+      "step": 10192
+    },
+    {
+      "epoch": 43.57,
+      "learning_rate": 9.85064810826407e-05,
+      "loss": 0.0003,
+      "step": 10196
+    },
+    {
+      "epoch": 43.59,
+      "learning_rate": 9.850365258063595e-05,
+      "loss": 0.0015,
+      "step": 10200
+    },
+    {
+      "epoch": 43.61,
+      "learning_rate": 9.850082144347007e-05,
+      "loss": 0.0016,
+      "step": 10204
+    },
+    {
+      "epoch": 43.62,
+      "learning_rate": 9.84979876712969e-05,
+      "loss": 0.0032,
+      "step": 10208
+    },
+    {
+      "epoch": 43.64,
+      "learning_rate": 9.849515126427038e-05,
+      "loss": 0.0004,
+      "step": 10212
+    },
+    {
+      "epoch": 43.66,
+      "learning_rate": 9.849231222254463e-05,
+      "loss": 0.0004,
+      "step": 10216
+    },
+    {
+      "epoch": 43.68,
+      "learning_rate": 9.848947054627386e-05,
+      "loss": 0.0007,
+      "step": 10220
+    },
+    {
+      "epoch": 43.69,
+      "learning_rate": 9.848662623561248e-05,
+      "loss": 0.0019,
+      "step": 10224
+    },
+    {
+      "epoch": 43.71,
+      "learning_rate": 9.848377929071503e-05,
+      "loss": 0.0004,
+      "step": 10228
+    },
+    {
+      "epoch": 43.73,
+      "learning_rate": 9.848092971173615e-05,
+      "loss": 0.0017,
+      "step": 10232
+    },
+    {
+      "epoch": 43.74,
+      "learning_rate": 9.847807749883069e-05,
+      "loss": 0.0007,
+      "step": 10236
+    },
+    {
+      "epoch": 43.76,
+      "learning_rate": 9.847522265215357e-05,
+      "loss": 0.0009,
+      "step": 10240
+    },
+    {
+      "epoch": 43.78,
+      "learning_rate": 9.847236517185994e-05,
+      "loss": 0.002,
+      "step": 10244
+    },
+    {
+      "epoch": 43.79,
+      "learning_rate": 9.846950505810499e-05,
+      "loss": 0.0019,
+      "step": 10248
+    },
+    {
+      "epoch": 43.81,
+      "learning_rate": 9.846664231104414e-05,
+      "loss": 0.0013,
+      "step": 10252
+    },
+    {
+      "epoch": 43.83,
+      "learning_rate": 9.846377693083292e-05,
+      "loss": 0.0004,
+      "step": 10256
+    },
+    {
+      "epoch": 43.85,
+      "learning_rate": 9.8460908917627e-05,
+      "loss": 0.0003,
+      "step": 10260
+    },
+    {
+      "epoch": 43.86,
+      "learning_rate": 9.845803827158218e-05,
+      "loss": 0.0014,
+      "step": 10264
+    },
+    {
+      "epoch": 43.88,
+      "learning_rate": 9.845516499285444e-05,
+      "loss": 0.0009,
+      "step": 10268
+    },
+    {
+      "epoch": 43.9,
+      "learning_rate": 9.845228908159988e-05,
+      "loss": 0.0004,
+      "step": 10272
+    },
+    {
+      "epoch": 43.91,
+      "learning_rate": 9.844941053797474e-05,
+      "loss": 0.0007,
+      "step": 10276
+    },
+    {
+      "epoch": 43.93,
+      "learning_rate": 9.844652936213541e-05,
+      "loss": 0.0017,
+      "step": 10280
+    },
+    {
+      "epoch": 43.95,
+      "learning_rate": 9.844364555423844e-05,
+      "loss": 0.0019,
+      "step": 10284
+    },
+    {
+      "epoch": 43.97,
+      "learning_rate": 9.844075911444046e-05,
+      "loss": 0.0004,
+      "step": 10288
+    },
+    {
+      "epoch": 43.98,
+      "learning_rate": 9.843787004289833e-05,
+      "loss": 0.0006,
+      "step": 10292
+    },
+    {
+      "epoch": 44.0,
+      "learning_rate": 9.8434978339769e-05,
+      "loss": 0.0007,
+      "step": 10296
+    },
+    {
+      "epoch": 44.02,
+      "learning_rate": 9.843208400520956e-05,
+      "loss": 0.0002,
+      "step": 10300
+    },
+    {
+      "epoch": 44.03,
+      "learning_rate": 9.842918703937727e-05,
+      "loss": 0.0005,
+      "step": 10304
+    },
+    {
+      "epoch": 44.05,
+      "learning_rate": 9.842628744242953e-05,
+      "loss": 0.0021,
+      "step": 10308
+    },
+    {
+      "epoch": 44.07,
+      "learning_rate": 9.842338521452384e-05,
+      "loss": 0.0002,
+      "step": 10312
+    },
+    {
+      "epoch": 44.09,
+      "learning_rate": 9.84204803558179e-05,
+      "loss": 0.0006,
+      "step": 10316
+    },
+    {
+      "epoch": 44.1,
+      "learning_rate": 9.841757286646953e-05,
+      "loss": 0.0013,
+      "step": 10320
+    },
+    {
+      "epoch": 44.12,
+      "learning_rate": 9.841466274663669e-05,
+      "loss": 0.0006,
+      "step": 10324
+    },
+    {
+      "epoch": 44.14,
+      "learning_rate": 9.841174999647747e-05,
+      "loss": 0.0006,
+      "step": 10328
+    },
+    {
+      "epoch": 44.15,
+      "learning_rate": 9.840883461615011e-05,
+      "loss": 0.0013,
+      "step": 10332
+    },
+    {
+      "epoch": 44.17,
+      "learning_rate": 9.840591660581304e-05,
+      "loss": 0.0004,
+      "step": 10336
+    },
+    {
+      "epoch": 44.19,
+      "learning_rate": 9.840299596562475e-05,
+      "loss": 0.0012,
+      "step": 10340
+    },
+    {
+      "epoch": 44.21,
+      "learning_rate": 9.840007269574394e-05,
+      "loss": 0.0007,
+      "step": 10344
+    },
+    {
+      "epoch": 44.22,
+      "learning_rate": 9.839714679632943e-05,
+      "loss": 0.0012,
+      "step": 10348
+    },
+    {
+      "epoch": 44.24,
+      "learning_rate": 9.839421826754016e-05,
+      "loss": 0.0021,
+      "step": 10352
+    },
+    {
+      "epoch": 44.26,
+      "learning_rate": 9.839128710953527e-05,
+      "loss": 0.0015,
+      "step": 10356
+    },
+    {
+      "epoch": 44.27,
+      "learning_rate": 9.838835332247397e-05,
+      "loss": 0.0038,
+      "step": 10360
+    },
+    {
+      "epoch": 44.29,
+      "learning_rate": 9.838541690651569e-05,
+      "loss": 0.0006,
+      "step": 10364
+    },
+    {
+      "epoch": 44.31,
+      "learning_rate": 9.83824778618199e-05,
+      "loss": 0.0007,
+      "step": 10368
+    },
+    {
+      "epoch": 44.32,
+      "learning_rate": 9.837953618854634e-05,
+      "loss": 0.0015,
+      "step": 10372
+    },
+    {
+      "epoch": 44.34,
+      "learning_rate": 9.83765918868548e-05,
+      "loss": 0.0002,
+      "step": 10376
+    },
+    {
+      "epoch": 44.36,
+      "learning_rate": 9.837364495690525e-05,
+      "loss": 0.0015,
+      "step": 10380
+    },
+    {
+      "epoch": 44.38,
+      "learning_rate": 9.83706953988578e-05,
+      "loss": 0.0015,
+      "step": 10384
+    },
+    {
+      "epoch": 44.39,
+      "learning_rate": 9.836774321287265e-05,
+      "loss": 0.0022,
+      "step": 10388
+    },
+    {
+      "epoch": 44.41,
+      "learning_rate": 9.836478839911025e-05,
+      "loss": 0.0007,
+      "step": 10392
+    },
+    {
+      "epoch": 44.43,
+      "learning_rate": 9.83618309577311e-05,
+      "loss": 0.0006,
+      "step": 10396
+    },
+    {
+      "epoch": 44.44,
+      "learning_rate": 9.83588708888959e-05,
+      "loss": 0.0009,
+      "step": 10400
+    },
+    {
+      "epoch": 44.46,
+      "learning_rate": 9.835590819276543e-05,
+      "loss": 0.0023,
+      "step": 10404
+    },
+    {
+      "epoch": 44.48,
+      "learning_rate": 9.835294286950069e-05,
+      "loss": 0.0028,
+      "step": 10408
+    },
+    {
+      "epoch": 44.5,
+      "learning_rate": 9.834997491926275e-05,
+      "loss": 0.0035,
+      "step": 10412
+    },
+    {
+      "epoch": 44.51,
+      "learning_rate": 9.83470043422129e-05,
+      "loss": 0.0011,
+      "step": 10416
+    },
+    {
+      "epoch": 44.53,
+      "learning_rate": 9.834403113851247e-05,
+      "loss": 0.0004,
+      "step": 10420
+    },
+    {
+      "epoch": 44.55,
+      "learning_rate": 9.834105530832302e-05,
+      "loss": 0.0017,
+      "step": 10424
+    },
+    {
+      "epoch": 44.56,
+      "learning_rate": 9.833807685180624e-05,
+      "loss": 0.0011,
+      "step": 10428
+    },
+    {
+      "epoch": 44.58,
+      "learning_rate": 9.833509576912394e-05,
+      "loss": 0.0005,
+      "step": 10432
+    },
+    {
+      "epoch": 44.6,
+      "learning_rate": 9.833211206043807e-05,
+      "loss": 0.0012,
+      "step": 10436
+    },
+    {
+      "epoch": 44.62,
+      "learning_rate": 9.832912572591072e-05,
+      "loss": 0.003,
+      "step": 10440
+    },
+    {
+      "epoch": 44.63,
+      "learning_rate": 9.832613676570417e-05,
+      "loss": 0.001,
+      "step": 10444
+    },
+    {
+      "epoch": 44.65,
+      "learning_rate": 9.832314517998076e-05,
+      "loss": 0.0005,
+      "step": 10448
+    },
+    {
+      "epoch": 44.67,
+      "learning_rate": 9.832015096890308e-05,
+      "loss": 0.0006,
+      "step": 10452
+    },
+    {
+      "epoch": 44.68,
+      "learning_rate": 9.831715413263375e-05,
+      "loss": 0.0025,
+      "step": 10456
+    },
+    {
+      "epoch": 44.7,
+      "learning_rate": 9.831415467133562e-05,
+      "loss": 0.0005,
+      "step": 10460
+    },
+    {
+      "epoch": 44.72,
+      "learning_rate": 9.831115258517163e-05,
+      "loss": 0.0008,
+      "step": 10464
+    },
+    {
+      "epoch": 44.74,
+      "learning_rate": 9.830814787430486e-05,
+      "loss": 0.0003,
+      "step": 10468
+    },
+    {
+      "epoch": 44.75,
+      "learning_rate": 9.830514053889861e-05,
+      "loss": 0.0008,
+      "step": 10472
+    },
+    {
+      "epoch": 44.77,
+      "learning_rate": 9.830213057911624e-05,
+      "loss": 0.0035,
+      "step": 10476
+    },
+    {
+      "epoch": 44.79,
+      "learning_rate": 9.829911799512124e-05,
+      "loss": 0.0015,
+      "step": 10480
+    },
+    {
+      "epoch": 44.8,
+      "learning_rate": 9.829610278707735e-05,
+      "loss": 0.0002,
+      "step": 10484
+    },
+    {
+      "epoch": 44.82,
+      "learning_rate": 9.829308495514832e-05,
+      "loss": 0.0005,
+      "step": 10488
+    },
+    {
+      "epoch": 44.84,
+      "learning_rate": 9.829006449949813e-05,
+      "loss": 0.0015,
+      "step": 10492
+    },
+    {
+      "epoch": 44.85,
+      "learning_rate": 9.82870414202909e-05,
+      "loss": 0.0007,
+      "step": 10496
+    },
+    {
+      "epoch": 44.87,
+      "learning_rate": 9.828401571769086e-05,
+      "loss": 0.0004,
+      "step": 10500
+    },
+    {
+      "epoch": 44.89,
+      "learning_rate": 9.828098739186238e-05,
+      "loss": 0.0006,
+      "step": 10504
+    },
+    {
+      "epoch": 44.91,
+      "learning_rate": 9.827795644296998e-05,
+      "loss": 0.001,
+      "step": 10508
+    },
+    {
+      "epoch": 44.92,
+      "learning_rate": 9.827492287117836e-05,
+      "loss": 0.0011,
+      "step": 10512
+    },
+    {
+      "epoch": 44.94,
+      "learning_rate": 9.82718866766523e-05,
+      "loss": 0.0008,
+      "step": 10516
+    },
+    {
+      "epoch": 44.96,
+      "learning_rate": 9.826884785955679e-05,
+      "loss": 0.003,
+      "step": 10520
+    },
+    {
+      "epoch": 44.97,
+      "learning_rate": 9.826580642005688e-05,
+      "loss": 0.0006,
+      "step": 10524
+    },
+    {
+      "epoch": 44.99,
+      "learning_rate": 9.826276235831785e-05,
+      "loss": 0.0011,
+      "step": 10528
+    },
+    {
+      "epoch": 45.01,
+      "learning_rate": 9.825971567450505e-05,
+      "loss": 0.0002,
+      "step": 10532
+    },
+    {
+      "epoch": 45.03,
+      "learning_rate": 9.825666636878403e-05,
+      "loss": 0.0008,
+      "step": 10536
+    },
+    {
+      "epoch": 45.04,
+      "learning_rate": 9.825361444132044e-05,
+      "loss": 0.0004,
+      "step": 10540
+    },
+    {
+      "epoch": 45.06,
+      "learning_rate": 9.82505598922801e-05,
+      "loss": 0.0006,
+      "step": 10544
+    },
+    {
+      "epoch": 45.08,
+      "learning_rate": 9.824750272182894e-05,
+      "loss": 0.0005,
+      "step": 10548
+    },
+    {
+      "epoch": 45.09,
+      "learning_rate": 9.824444293013306e-05,
+      "loss": 0.0002,
+      "step": 10552
+    },
+    {
+      "epoch": 45.11,
+      "learning_rate": 9.824138051735872e-05,
+      "loss": 0.0014,
+      "step": 10556
+    },
+    {
+      "epoch": 45.13,
+      "learning_rate": 9.823831548367226e-05,
+      "loss": 0.0008,
+      "step": 10560
+    },
+    {
+      "epoch": 45.15,
+      "learning_rate": 9.823524782924024e-05,
+      "loss": 0.0008,
+      "step": 10564
+    },
+    {
+      "epoch": 45.16,
+      "learning_rate": 9.82321775542293e-05,
+      "loss": 0.0003,
+      "step": 10568
+    },
+    {
+      "epoch": 45.18,
+      "learning_rate": 9.822910465880625e-05,
+      "loss": 0.0006,
+      "step": 10572
+    },
+    {
+      "epoch": 45.2,
+      "learning_rate": 9.822602914313803e-05,
+      "loss": 0.0015,
+      "step": 10576
+    },
+    {
+      "epoch": 45.21,
+      "learning_rate": 9.822295100739176e-05,
+      "loss": 0.0048,
+      "step": 10580
+    },
+    {
+      "epoch": 45.23,
+      "learning_rate": 9.821987025173462e-05,
+      "loss": 0.0041,
+      "step": 10584
+    },
+    {
+      "epoch": 45.25,
+      "learning_rate": 9.821678687633404e-05,
+      "loss": 0.002,
+      "step": 10588
+    },
+    {
+      "epoch": 45.26,
+      "learning_rate": 9.821370088135748e-05,
+      "loss": 0.0005,
+      "step": 10592
+    },
+    {
+      "epoch": 45.28,
+      "learning_rate": 9.821061226697266e-05,
+      "loss": 0.0007,
+      "step": 10596
+    },
+    {
+      "epoch": 45.3,
+      "learning_rate": 9.820752103334734e-05,
+      "loss": 0.0008,
+      "step": 10600
+    },
+    {
+      "epoch": 45.32,
+      "learning_rate": 9.820442718064948e-05,
+      "loss": 0.0018,
+      "step": 10604
+    },
+    {
+      "epoch": 45.33,
+      "learning_rate": 9.820133070904717e-05,
+      "loss": 0.001,
+      "step": 10608
+    },
+    {
+      "epoch": 45.35,
+      "learning_rate": 9.819823161870863e-05,
+      "loss": 0.0002,
+      "step": 10612
+    },
+    {
+      "epoch": 45.37,
+      "learning_rate": 9.819512990980223e-05,
+      "loss": 0.0004,
+      "step": 10616
+    },
+    {
+      "epoch": 45.38,
+      "learning_rate": 9.819202558249649e-05,
+      "loss": 0.0015,
+      "step": 10620
+    },
+    {
+      "epoch": 45.4,
+      "learning_rate": 9.818891863696004e-05,
+      "loss": 0.0005,
+      "step": 10624
+    },
+    {
+      "epoch": 45.42,
+      "learning_rate": 9.818580907336173e-05,
+      "loss": 0.0009,
+      "step": 10628
+    },
+    {
+      "epoch": 45.44,
+      "learning_rate": 9.818269689187045e-05,
+      "loss": 0.0006,
+      "step": 10632
+    },
+    {
+      "epoch": 45.45,
+      "learning_rate": 9.817958209265531e-05,
+      "loss": 0.0014,
+      "step": 10636
+    },
+    {
+      "epoch": 45.47,
+      "learning_rate": 9.817646467588551e-05,
+      "loss": 0.0006,
+      "step": 10640
+    },
+    {
+      "epoch": 45.49,
+      "learning_rate": 9.817334464173045e-05,
+      "loss": 0.0011,
+      "step": 10644
+    },
+    {
+      "epoch": 45.5,
+      "learning_rate": 9.817022199035961e-05,
+      "loss": 0.0003,
+      "step": 10648
+    },
+    {
+      "epoch": 45.52,
+      "learning_rate": 9.816709672194265e-05,
+      "loss": 0.0007,
+      "step": 10652
+    },
+    {
+      "epoch": 45.54,
+      "learning_rate": 9.816396883664938e-05,
+      "loss": 0.0002,
+      "step": 10656
+    },
+    {
+      "epoch": 45.56,
+      "learning_rate": 9.81608383346497e-05,
+      "loss": 0.0004,
+      "step": 10660
+    },
+    {
+      "epoch": 45.57,
+      "learning_rate": 9.815770521611371e-05,
+      "loss": 0.0011,
+      "step": 10664
+    },
+    {
+      "epoch": 45.59,
+      "learning_rate": 9.815456948121164e-05,
+      "loss": 0.0026,
+      "step": 10668
+    },
+    {
+      "epoch": 45.61,
+      "learning_rate": 9.815143113011383e-05,
+      "loss": 0.0032,
+      "step": 10672
+    },
+    {
+      "epoch": 45.62,
+      "learning_rate": 9.81482901629908e-05,
+      "loss": 0.0008,
+      "step": 10676
+    },
+    {
+      "epoch": 45.64,
+      "learning_rate": 9.814514658001317e-05,
+      "loss": 0.0014,
+      "step": 10680
+    },
+    {
+      "epoch": 45.66,
+      "learning_rate": 9.814200038135177e-05,
+      "loss": 0.0004,
+      "step": 10684
+    },
+    {
+      "epoch": 45.68,
+      "learning_rate": 9.813885156717749e-05,
+      "loss": 0.0008,
+      "step": 10688
+    },
+    {
+      "epoch": 45.69,
+      "learning_rate": 9.813570013766142e-05,
+      "loss": 0.0032,
+      "step": 10692
+    },
+    {
+      "epoch": 45.71,
+      "learning_rate": 9.813254609297477e-05,
+      "loss": 0.0004,
+      "step": 10696
+    },
+    {
+      "epoch": 45.73,
+      "learning_rate": 9.812938943328892e-05,
+      "loss": 0.0017,
+      "step": 10700
+    },
+    {
+      "epoch": 45.74,
+      "learning_rate": 9.812623015877533e-05,
+      "loss": 0.0024,
+      "step": 10704
+    },
+    {
+      "epoch": 45.76,
+      "learning_rate": 9.812306826960566e-05,
+      "loss": 0.0007,
+      "step": 10708
+    },
+    {
+      "epoch": 45.78,
+      "learning_rate": 9.81199037659517e-05,
+      "loss": 0.0013,
+      "step": 10712
+    },
+    {
+      "epoch": 45.79,
+      "learning_rate": 9.811673664798538e-05,
+      "loss": 0.0008,
+      "step": 10716
+    },
+    {
+      "epoch": 45.81,
+      "learning_rate": 9.811356691587874e-05,
+      "loss": 0.0003,
+      "step": 10720
+    },
+    {
+      "epoch": 45.83,
+      "learning_rate": 9.8110394569804e-05,
+      "loss": 0.0026,
+      "step": 10724
+    },
+    {
+      "epoch": 45.85,
+      "learning_rate": 9.81072196099335e-05,
+      "loss": 0.0004,
+      "step": 10728
+    },
+    {
+      "epoch": 45.86,
+      "learning_rate": 9.810404203643977e-05,
+      "loss": 0.0007,
+      "step": 10732
+    },
+    {
+      "epoch": 45.88,
+      "learning_rate": 9.81008618494954e-05,
+      "loss": 0.0005,
+      "step": 10736
+    },
+    {
+      "epoch": 45.9,
+      "learning_rate": 9.809767904927321e-05,
+      "loss": 0.0029,
+      "step": 10740
+    },
+    {
+      "epoch": 45.91,
+      "learning_rate": 9.809449363594607e-05,
+      "loss": 0.0022,
+      "step": 10744
+    },
+    {
+      "epoch": 45.93,
+      "learning_rate": 9.809130560968709e-05,
+      "loss": 0.002,
+      "step": 10748
+    },
+    {
+      "epoch": 45.95,
+      "learning_rate": 9.808811497066943e-05,
+      "loss": 0.0028,
+      "step": 10752
+    },
+    {
+      "epoch": 45.97,
+      "learning_rate": 9.808492171906647e-05,
+      "loss": 0.0023,
+      "step": 10756
+    },
+    {
+      "epoch": 45.98,
+      "learning_rate": 9.808172585505167e-05,
+      "loss": 0.0015,
+      "step": 10760
+    },
+    {
+      "epoch": 46.0,
+      "learning_rate": 9.807852737879868e-05,
+      "loss": 0.0014,
+      "step": 10764
+    },
+    {
+      "epoch": 46.02,
+      "learning_rate": 9.807532629048125e-05,
+      "loss": 0.003,
+      "step": 10768
+    },
+    {
+      "epoch": 46.03,
+      "learning_rate": 9.807212259027329e-05,
+      "loss": 0.0003,
+      "step": 10772
+    },
+    {
+      "epoch": 46.05,
+      "learning_rate": 9.806891627834889e-05,
+      "loss": 0.002,
+      "step": 10776
+    },
+    {
+      "epoch": 46.07,
+      "learning_rate": 9.80657073548822e-05,
+      "loss": 0.001,
+      "step": 10780
+    },
+    {
+      "epoch": 46.09,
+      "learning_rate": 9.80624958200476e-05,
+      "loss": 0.0014,
+      "step": 10784
+    },
+    {
+      "epoch": 46.1,
+      "learning_rate": 9.805928167401955e-05,
+      "loss": 0.001,
+      "step": 10788
+    },
+    {
+      "epoch": 46.12,
+      "learning_rate": 9.805606491697267e-05,
+      "loss": 0.003,
+      "step": 10792
+    },
+    {
+      "epoch": 46.14,
+      "learning_rate": 9.805284554908172e-05,
+      "loss": 0.0012,
+      "step": 10796
+    },
+    {
+      "epoch": 46.15,
+      "learning_rate": 9.80496235705216e-05,
+      "loss": 0.0004,
+      "step": 10800
+    },
+    {
+      "epoch": 46.17,
+      "learning_rate": 9.804639898146739e-05,
+      "loss": 0.0006,
+      "step": 10804
+    },
+    {
+      "epoch": 46.19,
+      "learning_rate": 9.804317178209426e-05,
+      "loss": 0.0007,
+      "step": 10808
+    },
+    {
+      "epoch": 46.21,
+      "learning_rate": 9.803994197257751e-05,
+      "loss": 0.0003,
+      "step": 10812
+    },
+    {
+      "epoch": 46.22,
+      "learning_rate": 9.803670955309265e-05,
+      "loss": 0.0002,
+      "step": 10816
+    },
+    {
+      "epoch": 46.24,
+      "learning_rate": 9.80334745238153e-05,
+      "loss": 0.0038,
+      "step": 10820
+    },
+    {
+      "epoch": 46.26,
+      "learning_rate": 9.80302368849212e-05,
+      "loss": 0.0003,
+      "step": 10824
+    },
+    {
+      "epoch": 46.27,
+      "learning_rate": 9.802699663658625e-05,
+      "loss": 0.0006,
+      "step": 10828
+    },
+    {
+      "epoch": 46.29,
+      "learning_rate": 9.802375377898649e-05,
+      "loss": 0.0004,
+      "step": 10832
+    },
+    {
+      "epoch": 46.31,
+      "learning_rate": 9.802050831229808e-05,
+      "loss": 0.0003,
+      "step": 10836
+    },
+    {
+      "epoch": 46.32,
+      "learning_rate": 9.801726023669739e-05,
+      "loss": 0.0002,
+      "step": 10840
+    },
+    {
+      "epoch": 46.34,
+      "learning_rate": 9.801400955236084e-05,
+      "loss": 0.0009,
+      "step": 10844
+    },
+    {
+      "epoch": 46.36,
+      "learning_rate": 9.801075625946508e-05,
+      "loss": 0.0012,
+      "step": 10848
+    },
+    {
+      "epoch": 46.38,
+      "learning_rate": 9.800750035818682e-05,
+      "loss": 0.0005,
+      "step": 10852
+    },
+    {
+      "epoch": 46.39,
+      "learning_rate": 9.800424184870298e-05,
+      "loss": 0.0019,
+      "step": 10856
+    },
+    {
+      "epoch": 46.41,
+      "learning_rate": 9.800098073119055e-05,
+      "loss": 0.0015,
+      "step": 10860
+    },
+    {
+      "epoch": 46.43,
+      "learning_rate": 9.799771700582676e-05,
+      "loss": 0.0003,
+      "step": 10864
+    },
+    {
+      "epoch": 46.44,
+      "learning_rate": 9.799445067278888e-05,
+      "loss": 0.0005,
+      "step": 10868
+    },
+    {
+      "epoch": 46.46,
+      "learning_rate": 9.799118173225438e-05,
+      "loss": 0.0009,
+      "step": 10872
+    },
+    {
+      "epoch": 46.48,
+      "learning_rate": 9.798791018440088e-05,
+      "loss": 0.0005,
+      "step": 10876
+    },
+    {
+      "epoch": 46.5,
+      "learning_rate": 9.798463602940608e-05,
+      "loss": 0.0007,
+      "step": 10880
+    },
+    {
+      "epoch": 46.51,
+      "learning_rate": 9.798135926744791e-05,
+      "loss": 0.0009,
+      "step": 10884
+    },
+    {
+      "epoch": 46.53,
+      "learning_rate": 9.797807989870435e-05,
+      "loss": 0.0043,
+      "step": 10888
+    },
+    {
+      "epoch": 46.55,
+      "learning_rate": 9.79747979233536e-05,
+      "loss": 0.0027,
+      "step": 10892
+    },
+    {
+      "epoch": 46.56,
+      "learning_rate": 9.797151334157393e-05,
+      "loss": 0.0012,
+      "step": 10896
+    },
+    {
+      "epoch": 46.58,
+      "learning_rate": 9.796822615354383e-05,
+      "loss": 0.0004,
+      "step": 10900
+    },
+    {
+      "epoch": 46.6,
+      "learning_rate": 9.796493635944186e-05,
+      "loss": 0.0007,
+      "step": 10904
+    },
+    {
+      "epoch": 46.62,
+      "learning_rate": 9.796164395944675e-05,
+      "loss": 0.0009,
+      "step": 10908
+    },
+    {
+      "epoch": 46.63,
+      "learning_rate": 9.795834895373739e-05,
+      "loss": 0.0008,
+      "step": 10912
+    },
+    {
+      "epoch": 46.65,
+      "learning_rate": 9.79550513424928e-05,
+      "loss": 0.0002,
+      "step": 10916
+    },
+    {
+      "epoch": 46.67,
+      "learning_rate": 9.795175112589213e-05,
+      "loss": 0.0009,
+      "step": 10920
+    },
+    {
+      "epoch": 46.68,
+      "learning_rate": 9.794844830411465e-05,
+      "loss": 0.0002,
+      "step": 10924
+    },
+    {
+      "epoch": 46.7,
+      "learning_rate": 9.794514287733983e-05,
+      "loss": 0.0005,
+      "step": 10928
+    },
+    {
+      "epoch": 46.72,
+      "learning_rate": 9.794183484574724e-05,
+      "loss": 0.001,
+      "step": 10932
+    },
+    {
+      "epoch": 46.74,
+      "learning_rate": 9.793852420951661e-05,
+      "loss": 0.001,
+      "step": 10936
+    },
+    {
+      "epoch": 46.75,
+      "learning_rate": 9.79352109688278e-05,
+      "loss": 0.0018,
+      "step": 10940
+    },
+    {
+      "epoch": 46.77,
+      "learning_rate": 9.793189512386081e-05,
+      "loss": 0.0014,
+      "step": 10944
+    },
+    {
+      "epoch": 46.79,
+      "learning_rate": 9.792857667479581e-05,
+      "loss": 0.0002,
+      "step": 10948
+    },
+    {
+      "epoch": 46.8,
+      "learning_rate": 9.792525562181305e-05,
+      "loss": 0.0008,
+      "step": 10952
+    },
+    {
+      "epoch": 46.82,
+      "learning_rate": 9.792193196509299e-05,
+      "loss": 0.0017,
+      "step": 10956
+    },
+    {
+      "epoch": 46.84,
+      "learning_rate": 9.791860570481618e-05,
+      "loss": 0.0006,
+      "step": 10960
+    },
+    {
+      "epoch": 46.85,
+      "learning_rate": 9.791527684116336e-05,
+      "loss": 0.0017,
+      "step": 10964
+    },
+    {
+      "epoch": 46.87,
+      "learning_rate": 9.791194537431536e-05,
+      "loss": 0.0005,
+      "step": 10968
+    },
+    {
+      "epoch": 46.89,
+      "learning_rate": 9.79086113044532e-05,
+      "loss": 0.0002,
+      "step": 10972
+    },
+    {
+      "epoch": 46.91,
+      "learning_rate": 9.790527463175799e-05,
+      "loss": 0.0005,
+      "step": 10976
+    },
+    {
+      "epoch": 46.92,
+      "learning_rate": 9.790193535641101e-05,
+      "loss": 0.0008,
+      "step": 10980
+    },
+    {
+      "epoch": 46.94,
+      "learning_rate": 9.789859347859369e-05,
+      "loss": 0.0004,
+      "step": 10984
+    },
+    {
+      "epoch": 46.96,
+      "learning_rate": 9.789524899848761e-05,
+      "loss": 0.0016,
+      "step": 10988
+    },
+    {
+      "epoch": 46.97,
+      "learning_rate": 9.789190191627444e-05,
+      "loss": 0.0003,
+      "step": 10992
+    },
+    {
+      "epoch": 46.99,
+      "learning_rate": 9.788855223213605e-05,
+      "loss": 0.002,
+      "step": 10996
+    },
+    {
+      "epoch": 47.01,
+      "learning_rate": 9.788519994625441e-05,
+      "loss": 0.0004,
+      "step": 11000
+    },
+    {
+      "epoch": 47.01,
+      "eval_exact_match": 0.5207900207900208,
+      "eval_loss": 0.7863634824752808,
+      "eval_runtime": 130.5406,
+      "eval_samples_per_second": 7.369,
+      "step": 11000
+    },
+    {
+      "epoch": 47.03,
+      "learning_rate": 9.788184505881164e-05,
+      "loss": 0.0013,
+      "step": 11004
+    },
+    {
+      "epoch": 47.04,
+      "learning_rate": 9.787848756999002e-05,
+      "loss": 0.0003,
+      "step": 11008
+    },
+    {
+      "epoch": 47.06,
+      "learning_rate": 9.787512747997197e-05,
+      "loss": 0.0006,
+      "step": 11012
+    },
+    {
+      "epoch": 47.08,
+      "learning_rate": 9.787176478894002e-05,
+      "loss": 0.0008,
+      "step": 11016
+    },
+    {
+      "epoch": 47.09,
+      "learning_rate": 9.786839949707686e-05,
+      "loss": 0.0003,
+      "step": 11020
+    },
+    {
+      "epoch": 47.11,
+      "learning_rate": 9.786503160456536e-05,
+      "loss": 0.0004,
+      "step": 11024
+    },
+    {
+      "epoch": 47.13,
+      "learning_rate": 9.786166111158845e-05,
+      "loss": 0.0004,
+      "step": 11028
+    },
+    {
+      "epoch": 47.15,
+      "learning_rate": 9.785828801832928e-05,
+      "loss": 0.0006,
+      "step": 11032
+    },
+    {
+      "epoch": 47.16,
+      "learning_rate": 9.785491232497108e-05,
+      "loss": 0.0005,
+      "step": 11036
+    },
+    {
+      "epoch": 47.18,
+      "learning_rate": 9.785153403169726e-05,
+      "loss": 0.0003,
+      "step": 11040
+    },
+    {
+      "epoch": 47.2,
+      "learning_rate": 9.784815313869136e-05,
+      "loss": 0.0014,
+      "step": 11044
+    },
+    {
+      "epoch": 47.21,
+      "learning_rate": 9.784476964613707e-05,
+      "loss": 0.0006,
+      "step": 11048
+    },
+    {
+      "epoch": 47.23,
+      "learning_rate": 9.784138355421819e-05,
+      "loss": 0.0005,
+      "step": 11052
+    },
+    {
+      "epoch": 47.25,
+      "learning_rate": 9.78379948631187e-05,
+      "loss": 0.001,
+      "step": 11056
+    },
+    {
+      "epoch": 47.26,
+      "learning_rate": 9.783460357302271e-05,
+      "loss": 0.0006,
+      "step": 11060
+    },
+    {
+      "epoch": 47.28,
+      "learning_rate": 9.783120968411445e-05,
+      "loss": 0.0011,
+      "step": 11064
+    },
+    {
+      "epoch": 47.3,
+      "learning_rate": 9.782781319657832e-05,
+      "loss": 0.0002,
+      "step": 11068
+    },
+    {
+      "epoch": 47.32,
+      "learning_rate": 9.782441411059884e-05,
+      "loss": 0.0006,
+      "step": 11072
+    },
+    {
+      "epoch": 47.33,
+      "learning_rate": 9.782101242636068e-05,
+      "loss": 0.001,
+      "step": 11076
+    },
+    {
+      "epoch": 47.35,
+      "learning_rate": 9.781760814404864e-05,
+      "loss": 0.0009,
+      "step": 11080
+    },
+    {
+      "epoch": 47.37,
+      "learning_rate": 9.78142012638477e-05,
+      "loss": 0.0017,
+      "step": 11084
+    },
+    {
+      "epoch": 47.38,
+      "learning_rate": 9.781079178594294e-05,
+      "loss": 0.0029,
+      "step": 11088
+    },
+    {
+      "epoch": 47.4,
+      "learning_rate": 9.780737971051957e-05,
+      "loss": 0.0012,
+      "step": 11092
+    },
+    {
+      "epoch": 47.42,
+      "learning_rate": 9.780396503776301e-05,
+      "loss": 0.0015,
+      "step": 11096
+    },
+    {
+      "epoch": 47.44,
+      "learning_rate": 9.780054776785875e-05,
+      "loss": 0.0013,
+      "step": 11100
+    },
+    {
+      "epoch": 47.45,
+      "learning_rate": 9.779712790099242e-05,
+      "loss": 0.0045,
+      "step": 11104
+    },
+    {
+      "epoch": 47.47,
+      "learning_rate": 9.779370543734987e-05,
+      "loss": 0.0019,
+      "step": 11108
+    },
+    {
+      "epoch": 47.49,
+      "learning_rate": 9.779028037711701e-05,
+      "loss": 0.0006,
+      "step": 11112
+    },
+    {
+      "epoch": 47.5,
+      "learning_rate": 9.778685272047994e-05,
+      "loss": 0.0004,
+      "step": 11116
+    },
+    {
+      "epoch": 47.52,
+      "learning_rate": 9.778342246762486e-05,
+      "loss": 0.0006,
+      "step": 11120
+    },
+    {
+      "epoch": 47.54,
+      "learning_rate": 9.777998961873815e-05,
+      "loss": 0.0004,
+      "step": 11124
+    },
+    {
+      "epoch": 47.56,
+      "learning_rate": 9.77765541740063e-05,
+      "loss": 0.0003,
+      "step": 11128
+    },
+    {
+      "epoch": 47.57,
+      "learning_rate": 9.777311613361596e-05,
+      "loss": 0.0017,
+      "step": 11132
+    },
+    {
+      "epoch": 47.59,
+      "learning_rate": 9.776967549775392e-05,
+      "loss": 0.0008,
+      "step": 11136
+    },
+    {
+      "epoch": 47.61,
+      "learning_rate": 9.77662322666071e-05,
+      "loss": 0.0004,
+      "step": 11140
+    },
+    {
+      "epoch": 47.62,
+      "learning_rate": 9.776278644036257e-05,
+      "loss": 0.0006,
+      "step": 11144
+    },
+    {
+      "epoch": 47.64,
+      "learning_rate": 9.775933801920754e-05,
+      "loss": 0.0032,
+      "step": 11148
+    },
+    {
+      "epoch": 47.66,
+      "learning_rate": 9.775588700332935e-05,
+      "loss": 0.0021,
+      "step": 11152
+    },
+    {
+      "epoch": 47.68,
+      "learning_rate": 9.77524333929155e-05,
+      "loss": 0.0006,
+      "step": 11156
+    },
+    {
+      "epoch": 47.69,
+      "learning_rate": 9.774897718815364e-05,
+      "loss": 0.0006,
+      "step": 11160
+    },
+    {
+      "epoch": 47.71,
+      "learning_rate": 9.77455183892315e-05,
+      "loss": 0.0006,
+      "step": 11164
+    },
+    {
+      "epoch": 47.73,
+      "learning_rate": 9.774205699633704e-05,
+      "loss": 0.0004,
+      "step": 11168
+    },
+    {
+      "epoch": 47.74,
+      "learning_rate": 9.773859300965828e-05,
+      "loss": 0.0005,
+      "step": 11172
+    },
+    {
+      "epoch": 47.76,
+      "learning_rate": 9.773512642938343e-05,
+      "loss": 0.0003,
+      "step": 11176
+    },
+    {
+      "epoch": 47.78,
+      "learning_rate": 9.773165725570081e-05,
+      "loss": 0.0007,
+      "step": 11180
+    },
+    {
+      "epoch": 47.79,
+      "learning_rate": 9.772818548879893e-05,
+      "loss": 0.0021,
+      "step": 11184
+    },
+    {
+      "epoch": 47.81,
+      "learning_rate": 9.772471112886639e-05,
+      "loss": 0.0003,
+      "step": 11188
+    },
+    {
+      "epoch": 47.83,
+      "learning_rate": 9.772123417609194e-05,
+      "loss": 0.0006,
+      "step": 11192
+    },
+    {
+      "epoch": 47.85,
+      "learning_rate": 9.771775463066449e-05,
+      "loss": 0.0006,
+      "step": 11196
+    },
+    {
+      "epoch": 47.86,
+      "learning_rate": 9.771427249277306e-05,
+      "loss": 0.0028,
+      "step": 11200
+    },
+    {
+      "epoch": 47.88,
+      "learning_rate": 9.771078776260687e-05,
+      "loss": 0.0002,
+      "step": 11204
+    },
+    {
+      "epoch": 47.9,
+      "learning_rate": 9.77073004403552e-05,
+      "loss": 0.0003,
+      "step": 11208
+    },
+    {
+      "epoch": 47.91,
+      "learning_rate": 9.770381052620756e-05,
+      "loss": 0.0033,
+      "step": 11212
+    },
+    {
+      "epoch": 47.93,
+      "learning_rate": 9.770031802035353e-05,
+      "loss": 0.0002,
+      "step": 11216
+    },
+    {
+      "epoch": 47.95,
+      "learning_rate": 9.769682292298284e-05,
+      "loss": 0.0012,
+      "step": 11220
+    },
+    {
+      "epoch": 47.97,
+      "learning_rate": 9.76933252342854e-05,
+      "loss": 0.0012,
+      "step": 11224
+    },
+    {
+      "epoch": 47.98,
+      "learning_rate": 9.768982495445122e-05,
+      "loss": 0.0008,
+      "step": 11228
+    },
+    {
+      "epoch": 48.0,
+      "learning_rate": 9.768632208367048e-05,
+      "loss": 0.0015,
+      "step": 11232
+    },
+    {
+      "epoch": 48.02,
+      "learning_rate": 9.768281662213347e-05,
+      "loss": 0.001,
+      "step": 11236
+    },
+    {
+      "epoch": 48.03,
+      "learning_rate": 9.767930857003065e-05,
+      "loss": 0.0006,
+      "step": 11240
+    },
+    {
+      "epoch": 48.05,
+      "learning_rate": 9.767579792755262e-05,
+      "loss": 0.0005,
+      "step": 11244
+    },
+    {
+      "epoch": 48.07,
+      "learning_rate": 9.76722846948901e-05,
+      "loss": 0.0004,
+      "step": 11248
+    },
+    {
+      "epoch": 48.09,
+      "learning_rate": 9.766876887223393e-05,
+      "loss": 0.001,
+      "step": 11252
+    },
+    {
+      "epoch": 48.1,
+      "learning_rate": 9.766525045977519e-05,
+      "loss": 0.0016,
+      "step": 11256
+    },
+    {
+      "epoch": 48.12,
+      "learning_rate": 9.766172945770497e-05,
+      "loss": 0.0009,
+      "step": 11260
+    },
+    {
+      "epoch": 48.14,
+      "learning_rate": 9.76582058662146e-05,
+      "loss": 0.0044,
+      "step": 11264
+    },
+    {
+      "epoch": 48.15,
+      "learning_rate": 9.765467968549549e-05,
+      "loss": 0.0007,
+      "step": 11268
+    },
+    {
+      "epoch": 48.17,
+      "learning_rate": 9.765115091573923e-05,
+      "loss": 0.0008,
+      "step": 11272
+    },
+    {
+      "epoch": 48.19,
+      "learning_rate": 9.764761955713752e-05,
+      "loss": 0.0004,
+      "step": 11276
+    },
+    {
+      "epoch": 48.21,
+      "learning_rate": 9.764408560988225e-05,
+      "loss": 0.001,
+      "step": 11280
+    },
+    {
+      "epoch": 48.22,
+      "learning_rate": 9.764054907416537e-05,
+      "loss": 0.0012,
+      "step": 11284
+    },
+    {
+      "epoch": 48.24,
+      "learning_rate": 9.763700995017905e-05,
+      "loss": 0.0007,
+      "step": 11288
+    },
+    {
+      "epoch": 48.26,
+      "learning_rate": 9.763346823811555e-05,
+      "loss": 0.0012,
+      "step": 11292
+    },
+    {
+      "epoch": 48.27,
+      "learning_rate": 9.76299239381673e-05,
+      "loss": 0.0009,
+      "step": 11296
+    },
+    {
+      "epoch": 48.29,
+      "learning_rate": 9.762637705052684e-05,
+      "loss": 0.0007,
+      "step": 11300
+    },
+    {
+      "epoch": 48.31,
+      "learning_rate": 9.76228275753869e-05,
+      "loss": 0.0009,
+      "step": 11304
+    },
+    {
+      "epoch": 48.32,
+      "learning_rate": 9.76192755129403e-05,
+      "loss": 0.0014,
+      "step": 11308
+    },
+    {
+      "epoch": 48.34,
+      "learning_rate": 9.761572086338002e-05,
+      "loss": 0.0006,
+      "step": 11312
+    },
+    {
+      "epoch": 48.36,
+      "learning_rate": 9.761216362689918e-05,
+      "loss": 0.0006,
+      "step": 11316
+    },
+    {
+      "epoch": 48.38,
+      "learning_rate": 9.760860380369105e-05,
+      "loss": 0.001,
+      "step": 11320
+    },
+    {
+      "epoch": 48.39,
+      "learning_rate": 9.760504139394902e-05,
+      "loss": 0.0011,
+      "step": 11324
+    },
+    {
+      "epoch": 48.41,
+      "learning_rate": 9.760147639786665e-05,
+      "loss": 0.0001,
+      "step": 11328
+    },
+    {
+      "epoch": 48.43,
+      "learning_rate": 9.75979088156376e-05,
+      "loss": 0.0004,
+      "step": 11332
+    },
+    {
+      "epoch": 48.44,
+      "learning_rate": 9.759433864745571e-05,
+      "loss": 0.0002,
+      "step": 11336
+    },
+    {
+      "epoch": 48.46,
+      "learning_rate": 9.759076589351494e-05,
+      "loss": 0.0007,
+      "step": 11340
+    },
+    {
+      "epoch": 48.48,
+      "learning_rate": 9.75871905540094e-05,
+      "loss": 0.0013,
+      "step": 11344
+    },
+    {
+      "epoch": 48.5,
+      "learning_rate": 9.758361262913334e-05,
+      "loss": 0.0004,
+      "step": 11348
+    },
+    {
+      "epoch": 48.51,
+      "learning_rate": 9.758003211908111e-05,
+      "loss": 0.0007,
+      "step": 11352
+    },
+    {
+      "epoch": 48.53,
+      "learning_rate": 9.757644902404728e-05,
+      "loss": 0.0005,
+      "step": 11356
+    },
+    {
+      "epoch": 48.55,
+      "learning_rate": 9.75728633442265e-05,
+      "loss": 0.0002,
+      "step": 11360
+    },
+    {
+      "epoch": 48.56,
+      "learning_rate": 9.756927507981357e-05,
+      "loss": 0.0011,
+      "step": 11364
+    },
+    {
+      "epoch": 48.58,
+      "learning_rate": 9.756568423100344e-05,
+      "loss": 0.0003,
+      "step": 11368
+    },
+    {
+      "epoch": 48.6,
+      "learning_rate": 9.756209079799121e-05,
+      "loss": 0.001,
+      "step": 11372
+    },
+    {
+      "epoch": 48.62,
+      "learning_rate": 9.75584947809721e-05,
+      "loss": 0.0003,
+      "step": 11376
+    },
+    {
+      "epoch": 48.63,
+      "learning_rate": 9.755489618014147e-05,
+      "loss": 0.0002,
+      "step": 11380
+    },
+    {
+      "epoch": 48.65,
+      "learning_rate": 9.755129499569482e-05,
+      "loss": 0.0019,
+      "step": 11384
+    },
+    {
+      "epoch": 48.67,
+      "learning_rate": 9.754769122782784e-05,
+      "loss": 0.0008,
+      "step": 11388
+    },
+    {
+      "epoch": 48.68,
+      "learning_rate": 9.754408487673628e-05,
+      "loss": 0.0002,
+      "step": 11392
+    },
+    {
+      "epoch": 48.7,
+      "learning_rate": 9.754047594261609e-05,
+      "loss": 0.0004,
+      "step": 11396
+    },
+    {
+      "epoch": 48.72,
+      "learning_rate": 9.753686442566334e-05,
+      "loss": 0.0005,
+      "step": 11400
+    },
+    {
+      "epoch": 48.74,
+      "learning_rate": 9.753325032607423e-05,
+      "loss": 0.0001,
+      "step": 11404
+    },
+    {
+      "epoch": 48.75,
+      "learning_rate": 9.75296336440451e-05,
+      "loss": 0.0004,
+      "step": 11408
+    },
+    {
+      "epoch": 48.77,
+      "learning_rate": 9.752601437977248e-05,
+      "loss": 0.0018,
+      "step": 11412
+    },
+    {
+      "epoch": 48.79,
+      "learning_rate": 9.752239253345297e-05,
+      "loss": 0.0003,
+      "step": 11416
+    },
+    {
+      "epoch": 48.8,
+      "learning_rate": 9.751876810528335e-05,
+      "loss": 0.001,
+      "step": 11420
+    },
+    {
+      "epoch": 48.82,
+      "learning_rate": 9.751514109546054e-05,
+      "loss": 0.0031,
+      "step": 11424
+    },
+    {
+      "epoch": 48.84,
+      "learning_rate": 9.751151150418158e-05,
+      "loss": 0.0004,
+      "step": 11428
+    },
+    {
+      "epoch": 48.85,
+      "learning_rate": 9.750787933164366e-05,
+      "loss": 0.0027,
+      "step": 11432
+    },
+    {
+      "epoch": 48.87,
+      "learning_rate": 9.750424457804412e-05,
+      "loss": 0.0003,
+      "step": 11436
+    },
+    {
+      "epoch": 48.89,
+      "learning_rate": 9.750060724358045e-05,
+      "loss": 0.0021,
+      "step": 11440
+    },
+    {
+      "epoch": 48.91,
+      "learning_rate": 9.749696732845021e-05,
+      "loss": 0.0033,
+      "step": 11444
+    },
+    {
+      "epoch": 48.92,
+      "learning_rate": 9.749332483285123e-05,
+      "loss": 0.0011,
+      "step": 11448
+    },
+    {
+      "epoch": 48.94,
+      "learning_rate": 9.748967975698135e-05,
+      "loss": 0.0002,
+      "step": 11452
+    },
+    {
+      "epoch": 48.96,
+      "learning_rate": 9.748603210103862e-05,
+      "loss": 0.0012,
+      "step": 11456
+    },
+    {
+      "epoch": 48.97,
+      "learning_rate": 9.748238186522119e-05,
+      "loss": 0.0025,
+      "step": 11460
+    },
+    {
+      "epoch": 48.99,
+      "learning_rate": 9.747872904972741e-05,
+      "loss": 0.006,
+      "step": 11464
+    },
+    {
+      "epoch": 49.01,
+      "learning_rate": 9.747507365475573e-05,
+      "loss": 0.0002,
+      "step": 11468
+    },
+    {
+      "epoch": 49.03,
+      "learning_rate": 9.747141568050472e-05,
+      "loss": 0.0004,
+      "step": 11472
+    },
+    {
+      "epoch": 49.04,
+      "learning_rate": 9.746775512717312e-05,
+      "loss": 0.0005,
+      "step": 11476
+    },
+    {
+      "epoch": 49.06,
+      "learning_rate": 9.746409199495983e-05,
+      "loss": 0.0004,
+      "step": 11480
+    },
+    {
+      "epoch": 49.08,
+      "learning_rate": 9.746042628406384e-05,
+      "loss": 0.0016,
+      "step": 11484
+    },
+    {
+      "epoch": 49.09,
+      "learning_rate": 9.745675799468432e-05,
+      "loss": 0.0002,
+      "step": 11488
+    },
+    {
+      "epoch": 49.11,
+      "learning_rate": 9.745308712702055e-05,
+      "loss": 0.0008,
+      "step": 11492
+    },
+    {
+      "epoch": 49.13,
+      "learning_rate": 9.744941368127199e-05,
+      "loss": 0.0002,
+      "step": 11496
+    },
+    {
+      "epoch": 49.15,
+      "learning_rate": 9.744573765763816e-05,
+      "loss": 0.0006,
+      "step": 11500
+    },
+    {
+      "epoch": 49.16,
+      "learning_rate": 9.744205905631884e-05,
+      "loss": 0.0013,
+      "step": 11504
+    },
+    {
+      "epoch": 49.18,
+      "learning_rate": 9.743837787751386e-05,
+      "loss": 0.0009,
+      "step": 11508
+    },
+    {
+      "epoch": 49.2,
+      "learning_rate": 9.743469412142321e-05,
+      "loss": 0.0004,
+      "step": 11512
+    },
+    {
+      "epoch": 49.21,
+      "learning_rate": 9.743100778824703e-05,
+      "loss": 0.0003,
+      "step": 11516
+    },
+    {
+      "epoch": 49.23,
+      "learning_rate": 9.742731887818558e-05,
+      "loss": 0.0003,
+      "step": 11520
+    },
+    {
+      "epoch": 49.25,
+      "learning_rate": 9.74236273914393e-05,
+      "loss": 0.0002,
+      "step": 11524
+    },
+    {
+      "epoch": 49.26,
+      "learning_rate": 9.741993332820873e-05,
+      "loss": 0.0004,
+      "step": 11528
+    },
+    {
+      "epoch": 49.28,
+      "learning_rate": 9.741623668869457e-05,
+      "loss": 0.0002,
+      "step": 11532
+    },
+    {
+      "epoch": 49.3,
+      "learning_rate": 9.741253747309767e-05,
+      "loss": 0.0002,
+      "step": 11536
+    },
+    {
+      "epoch": 49.32,
+      "learning_rate": 9.740883568161897e-05,
+      "loss": 0.0003,
+      "step": 11540
+    },
+    {
+      "epoch": 49.33,
+      "learning_rate": 9.740513131445961e-05,
+      "loss": 0.0007,
+      "step": 11544
+    },
+    {
+      "epoch": 49.35,
+      "learning_rate": 9.740142437182084e-05,
+      "loss": 0.0024,
+      "step": 11548
+    },
+    {
+      "epoch": 49.37,
+      "learning_rate": 9.739771485390406e-05,
+      "loss": 0.0004,
+      "step": 11552
+    },
+    {
+      "epoch": 49.38,
+      "learning_rate": 9.73940027609108e-05,
+      "loss": 0.0014,
+      "step": 11556
+    },
+    {
+      "epoch": 49.4,
+      "learning_rate": 9.739028809304274e-05,
+      "loss": 0.001,
+      "step": 11560
+    },
+    {
+      "epoch": 49.42,
+      "learning_rate": 9.738657085050167e-05,
+      "loss": 0.0013,
+      "step": 11564
+    },
+    {
+      "epoch": 49.44,
+      "learning_rate": 9.738285103348958e-05,
+      "loss": 0.0003,
+      "step": 11568
+    },
+    {
+      "epoch": 49.45,
+      "learning_rate": 9.737912864220856e-05,
+      "loss": 0.0017,
+      "step": 11572
+    },
+    {
+      "epoch": 49.47,
+      "learning_rate": 9.73754036768608e-05,
+      "loss": 0.0009,
+      "step": 11576
+    },
+    {
+      "epoch": 49.49,
+      "learning_rate": 9.737167613764873e-05,
+      "loss": 0.0004,
+      "step": 11580
+    },
+    {
+      "epoch": 49.5,
+      "learning_rate": 9.736794602477483e-05,
+      "loss": 0.0008,
+      "step": 11584
+    },
+    {
+      "epoch": 49.52,
+      "learning_rate": 9.736421333844177e-05,
+      "loss": 0.0032,
+      "step": 11588
+    },
+    {
+      "epoch": 49.54,
+      "learning_rate": 9.736047807885233e-05,
+      "loss": 0.0004,
+      "step": 11592
+    },
+    {
+      "epoch": 49.56,
+      "learning_rate": 9.735674024620947e-05,
+      "loss": 0.0011,
+      "step": 11596
+    },
+    {
+      "epoch": 49.57,
+      "learning_rate": 9.735299984071621e-05,
+      "loss": 0.0023,
+      "step": 11600
+    },
+    {
+      "epoch": 49.59,
+      "learning_rate": 9.734925686257584e-05,
+      "loss": 0.001,
+      "step": 11604
+    },
+    {
+      "epoch": 49.61,
+      "learning_rate": 9.734551131199163e-05,
+      "loss": 0.0008,
+      "step": 11608
+    },
+    {
+      "epoch": 49.62,
+      "learning_rate": 9.734176318916715e-05,
+      "loss": 0.0002,
+      "step": 11612
+    },
+    {
+      "epoch": 49.64,
+      "learning_rate": 9.733801249430596e-05,
+      "loss": 0.0006,
+      "step": 11616
+    },
+    {
+      "epoch": 49.66,
+      "learning_rate": 9.73342592276119e-05,
+      "loss": 0.0016,
+      "step": 11620
+    },
+    {
+      "epoch": 49.68,
+      "learning_rate": 9.733050338928882e-05,
+      "loss": 0.0007,
+      "step": 11624
+    },
+    {
+      "epoch": 49.69,
+      "learning_rate": 9.732674497954082e-05,
+      "loss": 0.0009,
+      "step": 11628
+    },
+    {
+      "epoch": 49.71,
+      "learning_rate": 9.732298399857206e-05,
+      "loss": 0.0012,
+      "step": 11632
+    },
+    {
+      "epoch": 49.73,
+      "learning_rate": 9.73192204465869e-05,
+      "loss": 0.0019,
+      "step": 11636
+    },
+    {
+      "epoch": 49.74,
+      "learning_rate": 9.731545432378977e-05,
+      "loss": 0.0004,
+      "step": 11640
+    },
+    {
+      "epoch": 49.76,
+      "learning_rate": 9.73116856303853e-05,
+      "loss": 0.0004,
+      "step": 11644
+    },
+    {
+      "epoch": 49.78,
+      "learning_rate": 9.730791436657827e-05,
+      "loss": 0.001,
+      "step": 11648
+    },
+    {
+      "epoch": 49.79,
+      "learning_rate": 9.730414053257352e-05,
+      "loss": 0.0002,
+      "step": 11652
+    },
+    {
+      "epoch": 49.81,
+      "learning_rate": 9.730036412857611e-05,
+      "loss": 0.0007,
+      "step": 11656
+    },
+    {
+      "epoch": 49.83,
+      "learning_rate": 9.72965851547912e-05,
+      "loss": 0.0002,
+      "step": 11660
+    },
+    {
+      "epoch": 49.85,
+      "learning_rate": 9.72928036114241e-05,
+      "loss": 0.0021,
+      "step": 11664
+    },
+    {
+      "epoch": 49.86,
+      "learning_rate": 9.728901949868024e-05,
+      "loss": 0.0029,
+      "step": 11668
+    },
+    {
+      "epoch": 49.88,
+      "learning_rate": 9.728523281676524e-05,
+      "loss": 0.0008,
+      "step": 11672
+    },
+    {
+      "epoch": 49.9,
+      "learning_rate": 9.728144356588481e-05,
+      "loss": 0.0002,
+      "step": 11676
+    },
+    {
+      "epoch": 49.91,
+      "learning_rate": 9.727765174624482e-05,
+      "loss": 0.0004,
+      "step": 11680
+    },
+    {
+      "epoch": 49.93,
+      "learning_rate": 9.727385735805126e-05,
+      "loss": 0.0002,
+      "step": 11684
+    },
+    {
+      "epoch": 49.95,
+      "learning_rate": 9.72700604015103e-05,
+      "loss": 0.0006,
+      "step": 11688
+    },
+    {
+      "epoch": 49.97,
+      "learning_rate": 9.726626087682821e-05,
+      "loss": 0.0015,
+      "step": 11692
+    },
+    {
+      "epoch": 49.98,
+      "learning_rate": 9.726245878421142e-05,
+      "loss": 0.0017,
+      "step": 11696
+    },
+    {
+      "epoch": 50.0,
+      "learning_rate": 9.72586541238665e-05,
+      "loss": 0.0007,
+      "step": 11700
+    },
+    {
+      "epoch": 50.02,
+      "learning_rate": 9.725484689600013e-05,
+      "loss": 0.0005,
+      "step": 11704
+    },
+    {
+      "epoch": 50.03,
+      "learning_rate": 9.72510371008192e-05,
+      "loss": 0.0006,
+      "step": 11708
+    },
+    {
+      "epoch": 50.05,
+      "learning_rate": 9.724722473853065e-05,
+      "loss": 0.0004,
+      "step": 11712
+    },
+    {
+      "epoch": 50.07,
+      "learning_rate": 9.724340980934162e-05,
+      "loss": 0.0003,
+      "step": 11716
+    },
+    {
+      "epoch": 50.09,
+      "learning_rate": 9.723959231345936e-05,
+      "loss": 0.0008,
+      "step": 11720
+    },
+    {
+      "epoch": 50.1,
+      "learning_rate": 9.723577225109128e-05,
+      "loss": 0.0009,
+      "step": 11724
+    },
+    {
+      "epoch": 50.12,
+      "learning_rate": 9.723194962244493e-05,
+      "loss": 0.0015,
+      "step": 11728
+    },
+    {
+      "epoch": 50.14,
+      "learning_rate": 9.722812442772798e-05,
+      "loss": 0.0001,
+      "step": 11732
+    },
+    {
+      "epoch": 50.15,
+      "learning_rate": 9.722429666714823e-05,
+      "loss": 0.0045,
+      "step": 11736
+    },
+    {
+      "epoch": 50.17,
+      "learning_rate": 9.722046634091367e-05,
+      "loss": 0.0069,
+      "step": 11740
+    },
+    {
+      "epoch": 50.19,
+      "learning_rate": 9.72166334492324e-05,
+      "loss": 0.0003,
+      "step": 11744
+    },
+    {
+      "epoch": 50.21,
+      "learning_rate": 9.721279799231263e-05,
+      "loss": 0.0005,
+      "step": 11748
+    },
+    {
+      "epoch": 50.22,
+      "learning_rate": 9.720895997036275e-05,
+      "loss": 0.0008,
+      "step": 11752
+    },
+    {
+      "epoch": 50.24,
+      "learning_rate": 9.72051193835913e-05,
+      "loss": 0.0003,
+      "step": 11756
+    },
+    {
+      "epoch": 50.26,
+      "learning_rate": 9.720127623220689e-05,
+      "loss": 0.0002,
+      "step": 11760
+    },
+    {
+      "epoch": 50.27,
+      "learning_rate": 9.719743051641834e-05,
+      "loss": 0.0008,
+      "step": 11764
+    },
+    {
+      "epoch": 50.29,
+      "learning_rate": 9.719358223643459e-05,
+      "loss": 0.0003,
+      "step": 11768
+    },
+    {
+      "epoch": 50.31,
+      "learning_rate": 9.718973139246471e-05,
+      "loss": 0.0004,
+      "step": 11772
+    },
+    {
+      "epoch": 50.32,
+      "learning_rate": 9.71858779847179e-05,
+      "loss": 0.0002,
+      "step": 11776
+    },
+    {
+      "epoch": 50.34,
+      "learning_rate": 9.718202201340353e-05,
+      "loss": 0.0012,
+      "step": 11780
+    },
+    {
+      "epoch": 50.36,
+      "learning_rate": 9.717816347873108e-05,
+      "loss": 0.001,
+      "step": 11784
+    },
+    {
+      "epoch": 50.38,
+      "learning_rate": 9.717430238091019e-05,
+      "loss": 0.0002,
+      "step": 11788
+    },
+    {
+      "epoch": 50.39,
+      "learning_rate": 9.717043872015061e-05,
+      "loss": 0.0011,
+      "step": 11792
+    },
+    {
+      "epoch": 50.41,
+      "learning_rate": 9.716657249666228e-05,
+      "loss": 0.001,
+      "step": 11796
+    },
+    {
+      "epoch": 50.43,
+      "learning_rate": 9.716270371065523e-05,
+      "loss": 0.001,
+      "step": 11800
+    },
+    {
+      "epoch": 50.44,
+      "learning_rate": 9.715883236233965e-05,
+      "loss": 0.0002,
+      "step": 11804
+    },
+    {
+      "epoch": 50.46,
+      "learning_rate": 9.715495845192586e-05,
+      "loss": 0.0004,
+      "step": 11808
+    },
+    {
+      "epoch": 50.48,
+      "learning_rate": 9.715108197962435e-05,
+      "loss": 0.0017,
+      "step": 11812
+    },
+    {
+      "epoch": 50.5,
+      "learning_rate": 9.714720294564569e-05,
+      "loss": 0.0007,
+      "step": 11816
+    },
+    {
+      "epoch": 50.51,
+      "learning_rate": 9.714332135020066e-05,
+      "loss": 0.0006,
+      "step": 11820
+    },
+    {
+      "epoch": 50.53,
+      "learning_rate": 9.71394371935001e-05,
+      "loss": 0.0011,
+      "step": 11824
+    },
+    {
+      "epoch": 50.55,
+      "learning_rate": 9.713555047575509e-05,
+      "loss": 0.0003,
+      "step": 11828
+    },
+    {
+      "epoch": 50.56,
+      "learning_rate": 9.713166119717674e-05,
+      "loss": 0.0006,
+      "step": 11832
+    },
+    {
+      "epoch": 50.58,
+      "learning_rate": 9.712776935797638e-05,
+      "loss": 0.0009,
+      "step": 11836
+    },
+    {
+      "epoch": 50.6,
+      "learning_rate": 9.712387495836544e-05,
+      "loss": 0.0004,
+      "step": 11840
+    },
+    {
+      "epoch": 50.62,
+      "learning_rate": 9.711997799855552e-05,
+      "loss": 0.0002,
+      "step": 11844
+    },
+    {
+      "epoch": 50.63,
+      "learning_rate": 9.71160784787583e-05,
+      "loss": 0.0003,
+      "step": 11848
+    },
+    {
+      "epoch": 50.65,
+      "learning_rate": 9.711217639918567e-05,
+      "loss": 0.0008,
+      "step": 11852
+    },
+    {
+      "epoch": 50.67,
+      "learning_rate": 9.710827176004961e-05,
+      "loss": 0.0006,
+      "step": 11856
+    },
+    {
+      "epoch": 50.68,
+      "learning_rate": 9.710436456156226e-05,
+      "loss": 0.0004,
+      "step": 11860
+    },
+    {
+      "epoch": 50.7,
+      "learning_rate": 9.71004548039359e-05,
+      "loss": 0.0003,
+      "step": 11864
+    },
+    {
+      "epoch": 50.72,
+      "learning_rate": 9.709654248738293e-05,
+      "loss": 0.0018,
+      "step": 11868
+    },
+    {
+      "epoch": 50.74,
+      "learning_rate": 9.709262761211592e-05,
+      "loss": 0.0012,
+      "step": 11872
+    },
+    {
+      "epoch": 50.75,
+      "learning_rate": 9.708871017834756e-05,
+      "loss": 0.0006,
+      "step": 11876
+    },
+    {
+      "epoch": 50.77,
+      "learning_rate": 9.708479018629066e-05,
+      "loss": 0.0007,
+      "step": 11880
+    },
+    {
+      "epoch": 50.79,
+      "learning_rate": 9.70808676361582e-05,
+      "loss": 0.0006,
+      "step": 11884
+    },
+    {
+      "epoch": 50.8,
+      "learning_rate": 9.707694252816331e-05,
+      "loss": 0.0004,
+      "step": 11888
+    },
+    {
+      "epoch": 50.82,
+      "learning_rate": 9.707301486251921e-05,
+      "loss": 0.0002,
+      "step": 11892
+    },
+    {
+      "epoch": 50.84,
+      "learning_rate": 9.70690846394393e-05,
+      "loss": 0.0018,
+      "step": 11896
+    },
+    {
+      "epoch": 50.85,
+      "learning_rate": 9.70651518591371e-05,
+      "loss": 0.0038,
+      "step": 11900
+    },
+    {
+      "epoch": 50.87,
+      "learning_rate": 9.706121652182628e-05,
+      "loss": 0.0006,
+      "step": 11904
+    },
+    {
+      "epoch": 50.89,
+      "learning_rate": 9.705727862772064e-05,
+      "loss": 0.0008,
+      "step": 11908
+    },
+    {
+      "epoch": 50.91,
+      "learning_rate": 9.705333817703413e-05,
+      "loss": 0.0001,
+      "step": 11912
+    },
+    {
+      "epoch": 50.92,
+      "learning_rate": 9.704939516998082e-05,
+      "loss": 0.0004,
+      "step": 11916
+    },
+    {
+      "epoch": 50.94,
+      "learning_rate": 9.704544960677492e-05,
+      "loss": 0.0007,
+      "step": 11920
+    },
+    {
+      "epoch": 50.96,
+      "learning_rate": 9.704150148763082e-05,
+      "loss": 0.0006,
+      "step": 11924
+    },
+    {
+      "epoch": 50.97,
+      "learning_rate": 9.7037550812763e-05,
+      "loss": 0.0005,
+      "step": 11928
+    },
+    {
+      "epoch": 50.99,
+      "learning_rate": 9.703359758238609e-05,
+      "loss": 0.0013,
+      "step": 11932
+    },
+    {
+      "epoch": 51.01,
+      "learning_rate": 9.702964179671488e-05,
+      "loss": 0.0037,
+      "step": 11936
+    },
+    {
+      "epoch": 51.03,
+      "learning_rate": 9.702568345596428e-05,
+      "loss": 0.0006,
+      "step": 11940
+    },
+    {
+      "epoch": 51.04,
+      "learning_rate": 9.702172256034933e-05,
+      "loss": 0.004,
+      "step": 11944
+    },
+    {
+      "epoch": 51.06,
+      "learning_rate": 9.701775911008524e-05,
+      "loss": 0.0008,
+      "step": 11948
+    },
+    {
+      "epoch": 51.08,
+      "learning_rate": 9.701379310538733e-05,
+      "loss": 0.0007,
+      "step": 11952
+    },
+    {
+      "epoch": 51.09,
+      "learning_rate": 9.700982454647108e-05,
+      "loss": 0.0003,
+      "step": 11956
+    },
+    {
+      "epoch": 51.11,
+      "learning_rate": 9.700585343355208e-05,
+      "loss": 0.0029,
+      "step": 11960
+    },
+    {
+      "epoch": 51.13,
+      "learning_rate": 9.70018797668461e-05,
+      "loss": 0.0006,
+      "step": 11964
+    },
+    {
+      "epoch": 51.15,
+      "learning_rate": 9.699790354656902e-05,
+      "loss": 0.0013,
+      "step": 11968
+    },
+    {
+      "epoch": 51.16,
+      "learning_rate": 9.699392477293686e-05,
+      "loss": 0.0005,
+      "step": 11972
+    },
+    {
+      "epoch": 51.18,
+      "learning_rate": 9.698994344616577e-05,
+      "loss": 0.0003,
+      "step": 11976
+    },
+    {
+      "epoch": 51.2,
+      "learning_rate": 9.698595956647205e-05,
+      "loss": 0.0008,
+      "step": 11980
+    },
+    {
+      "epoch": 51.21,
+      "learning_rate": 9.698197313407218e-05,
+      "loss": 0.0002,
+      "step": 11984
+    },
+    {
+      "epoch": 51.23,
+      "learning_rate": 9.697798414918271e-05,
+      "loss": 0.0024,
+      "step": 11988
+    },
+    {
+      "epoch": 51.25,
+      "learning_rate": 9.697399261202036e-05,
+      "loss": 0.0003,
+      "step": 11992
+    },
+    {
+      "epoch": 51.26,
+      "learning_rate": 9.6969998522802e-05,
+      "loss": 0.0009,
+      "step": 11996
+    },
+    {
+      "epoch": 51.28,
+      "learning_rate": 9.696600188174459e-05,
+      "loss": 0.0019,
+      "step": 12000
+    },
+    {
+      "epoch": 51.28,
+      "eval_exact_match": 0.5166320166320166,
+      "eval_loss": 0.8132425546646118,
+      "eval_runtime": 140.4988,
+      "eval_samples_per_second": 6.847,
+      "step": 12000
+    },
+    {
+      "epoch": 51.3,
+      "learning_rate": 9.696200268906532e-05,
+      "loss": 0.0009,
+      "step": 12004
+    },
+    {
+      "epoch": 51.32,
+      "learning_rate": 9.69580009449814e-05,
+      "loss": 0.0016,
+      "step": 12008
+    },
+    {
+      "epoch": 51.33,
+      "learning_rate": 9.69539966497103e-05,
+      "loss": 0.0015,
+      "step": 12012
+    },
+    {
+      "epoch": 51.35,
+      "learning_rate": 9.694998980346952e-05,
+      "loss": 0.001,
+      "step": 12016
+    },
+    {
+      "epoch": 51.37,
+      "learning_rate": 9.694598040647679e-05,
+      "loss": 0.0005,
+      "step": 12020
+    },
+    {
+      "epoch": 51.38,
+      "learning_rate": 9.694196845894992e-05,
+      "loss": 0.0003,
+      "step": 12024
+    },
+    {
+      "epoch": 51.4,
+      "learning_rate": 9.693795396110686e-05,
+      "loss": 0.0008,
+      "step": 12028
+    },
+    {
+      "epoch": 51.42,
+      "learning_rate": 9.693393691316572e-05,
+      "loss": 0.0011,
+      "step": 12032
+    },
+    {
+      "epoch": 51.44,
+      "learning_rate": 9.692991731534477e-05,
+      "loss": 0.0025,
+      "step": 12036
+    },
+    {
+      "epoch": 51.45,
+      "learning_rate": 9.692589516786237e-05,
+      "loss": 0.0011,
+      "step": 12040
+    },
+    {
+      "epoch": 51.47,
+      "learning_rate": 9.692187047093703e-05,
+      "loss": 0.0015,
+      "step": 12044
+    },
+    {
+      "epoch": 51.49,
+      "learning_rate": 9.691784322478743e-05,
+      "loss": 0.0006,
+      "step": 12048
+    },
+    {
+      "epoch": 51.5,
+      "learning_rate": 9.691381342963235e-05,
+      "loss": 0.0016,
+      "step": 12052
+    },
+    {
+      "epoch": 51.52,
+      "learning_rate": 9.690978108569073e-05,
+      "loss": 0.0007,
+      "step": 12056
+    },
+    {
+      "epoch": 51.54,
+      "learning_rate": 9.690574619318166e-05,
+      "loss": 0.0009,
+      "step": 12060
+    },
+    {
+      "epoch": 51.56,
+      "learning_rate": 9.690170875232431e-05,
+      "loss": 0.0016,
+      "step": 12064
+    },
+    {
+      "epoch": 51.57,
+      "learning_rate": 9.689766876333809e-05,
+      "loss": 0.0001,
+      "step": 12068
+    },
+    {
+      "epoch": 51.59,
+      "learning_rate": 9.689362622644244e-05,
+      "loss": 0.0005,
+      "step": 12072
+    },
+    {
+      "epoch": 51.61,
+      "learning_rate": 9.688958114185702e-05,
+      "loss": 0.0002,
+      "step": 12076
+    },
+    {
+      "epoch": 51.62,
+      "learning_rate": 9.688553350980157e-05,
+      "loss": 0.002,
+      "step": 12080
+    },
+    {
+      "epoch": 51.64,
+      "learning_rate": 9.6881483330496e-05,
+      "loss": 0.0004,
+      "step": 12084
+    },
+    {
+      "epoch": 51.66,
+      "learning_rate": 9.687743060416036e-05,
+      "loss": 0.0014,
+      "step": 12088
+    },
+    {
+      "epoch": 51.68,
+      "learning_rate": 9.687337533101484e-05,
+      "loss": 0.0008,
+      "step": 12092
+    },
+    {
+      "epoch": 51.69,
+      "learning_rate": 9.686931751127974e-05,
+      "loss": 0.0001,
+      "step": 12096
+    },
+    {
+      "epoch": 51.71,
+      "learning_rate": 9.686525714517554e-05,
+      "loss": 0.0004,
+      "step": 12100
+    },
+    {
+      "epoch": 51.73,
+      "learning_rate": 9.68611942329228e-05,
+      "loss": 0.0002,
+      "step": 12104
+    },
+    {
+      "epoch": 51.74,
+      "learning_rate": 9.685712877474229e-05,
+      "loss": 0.0003,
+      "step": 12108
+    },
+    {
+      "epoch": 51.76,
+      "learning_rate": 9.685306077085488e-05,
+      "loss": 0.0014,
+      "step": 12112
+    },
+    {
+      "epoch": 51.78,
+      "learning_rate": 9.684899022148154e-05,
+      "loss": 0.0002,
+      "step": 12116
+    },
+    {
+      "epoch": 51.79,
+      "learning_rate": 9.684491712684347e-05,
+      "loss": 0.0005,
+      "step": 12120
+    },
+    {
+      "epoch": 51.81,
+      "learning_rate": 9.684084148716195e-05,
+      "loss": 0.0002,
+      "step": 12124
+    },
+    {
+      "epoch": 51.83,
+      "learning_rate": 9.683676330265838e-05,
+      "loss": 0.0002,
+      "step": 12128
+    },
+    {
+      "epoch": 51.85,
+      "learning_rate": 9.683268257355433e-05,
+      "loss": 0.0036,
+      "step": 12132
+    },
+    {
+      "epoch": 51.86,
+      "learning_rate": 9.682859930007153e-05,
+      "loss": 0.0003,
+      "step": 12136
+    },
+    {
+      "epoch": 51.88,
+      "learning_rate": 9.682451348243178e-05,
+      "loss": 0.0016,
+      "step": 12140
+    },
+    {
+      "epoch": 51.9,
+      "learning_rate": 9.68204251208571e-05,
+      "loss": 0.0014,
+      "step": 12144
+    },
+    {
+      "epoch": 51.91,
+      "learning_rate": 9.681633421556957e-05,
+      "loss": 0.0015,
+      "step": 12148
+    },
+    {
+      "epoch": 51.93,
+      "learning_rate": 9.681224076679147e-05,
+      "loss": 0.0002,
+      "step": 12152
+    },
+    {
+      "epoch": 51.95,
+      "learning_rate": 9.680814477474518e-05,
+      "loss": 0.0008,
+      "step": 12156
+    },
+    {
+      "epoch": 51.97,
+      "learning_rate": 9.680404623965325e-05,
+      "loss": 0.003,
+      "step": 12160
+    },
+    {
+      "epoch": 51.98,
+      "learning_rate": 9.679994516173834e-05,
+      "loss": 0.0016,
+      "step": 12164
+    },
+    {
+      "epoch": 52.0,
+      "learning_rate": 9.679584154122323e-05,
+      "loss": 0.0004,
+      "step": 12168
+    },
+    {
+      "epoch": 52.02,
+      "learning_rate": 9.679173537833092e-05,
+      "loss": 0.0037,
+      "step": 12172
+    },
+    {
+      "epoch": 52.03,
+      "learning_rate": 9.678762667328445e-05,
+      "loss": 0.0006,
+      "step": 12176
+    },
+    {
+      "epoch": 52.05,
+      "learning_rate": 9.678351542630706e-05,
+      "loss": 0.0008,
+      "step": 12180
+    },
+    {
+      "epoch": 52.07,
+      "learning_rate": 9.677940163762212e-05,
+      "loss": 0.0005,
+      "step": 12184
+    },
+    {
+      "epoch": 52.09,
+      "learning_rate": 9.67752853074531e-05,
+      "loss": 0.0003,
+      "step": 12188
+    },
+    {
+      "epoch": 52.1,
+      "learning_rate": 9.677116643602367e-05,
+      "loss": 0.0002,
+      "step": 12192
+    },
+    {
+      "epoch": 52.12,
+      "learning_rate": 9.676704502355757e-05,
+      "loss": 0.0004,
+      "step": 12196
+    },
+    {
+      "epoch": 52.14,
+      "learning_rate": 9.676292107027876e-05,
+      "loss": 0.0009,
+      "step": 12200
+    },
+    {
+      "epoch": 52.15,
+      "learning_rate": 9.675879457641124e-05,
+      "loss": 0.0003,
+      "step": 12204
+    },
+    {
+      "epoch": 52.17,
+      "learning_rate": 9.675466554217922e-05,
+      "loss": 0.0007,
+      "step": 12208
+    },
+    {
+      "epoch": 52.19,
+      "learning_rate": 9.675053396780704e-05,
+      "loss": 0.0003,
+      "step": 12212
+    },
+    {
+      "epoch": 52.21,
+      "learning_rate": 9.674639985351915e-05,
+      "loss": 0.0002,
+      "step": 12216
+    },
+    {
+      "epoch": 52.22,
+      "learning_rate": 9.674226319954014e-05,
+      "loss": 0.0001,
+      "step": 12220
+    },
+    {
+      "epoch": 52.24,
+      "learning_rate": 9.673812400609479e-05,
+      "loss": 0.0002,
+      "step": 12224
+    },
+    {
+      "epoch": 52.26,
+      "learning_rate": 9.673398227340794e-05,
+      "loss": 0.0002,
+      "step": 12228
+    },
+    {
+      "epoch": 52.27,
+      "learning_rate": 9.672983800170464e-05,
+      "loss": 0.0002,
+      "step": 12232
+    },
+    {
+      "epoch": 52.29,
+      "learning_rate": 9.672569119121e-05,
+      "loss": 0.0005,
+      "step": 12236
+    },
+    {
+      "epoch": 52.31,
+      "learning_rate": 9.672154184214934e-05,
+      "loss": 0.0017,
+      "step": 12240
+    },
+    {
+      "epoch": 52.32,
+      "learning_rate": 9.67173899547481e-05,
+      "loss": 0.0002,
+      "step": 12244
+    },
+    {
+      "epoch": 52.34,
+      "learning_rate": 9.671323552923184e-05,
+      "loss": 0.0009,
+      "step": 12248
+    },
+    {
+      "epoch": 52.36,
+      "learning_rate": 9.670907856582627e-05,
+      "loss": 0.0002,
+      "step": 12252
+    },
+    {
+      "epoch": 52.38,
+      "learning_rate": 9.670491906475722e-05,
+      "loss": 0.0002,
+      "step": 12256
+    },
+    {
+      "epoch": 52.39,
+      "learning_rate": 9.670075702625068e-05,
+      "loss": 0.0004,
+      "step": 12260
+    },
+    {
+      "epoch": 52.41,
+      "learning_rate": 9.669659245053278e-05,
+      "loss": 0.0017,
+      "step": 12264
+    },
+    {
+      "epoch": 52.43,
+      "learning_rate": 9.669242533782974e-05,
+      "loss": 0.0018,
+      "step": 12268
+    },
+    {
+      "epoch": 52.44,
+      "learning_rate": 9.668825568836801e-05,
+      "loss": 0.0011,
+      "step": 12272
+    },
+    {
+      "epoch": 52.46,
+      "learning_rate": 9.66840835023741e-05,
+      "loss": 0.0002,
+      "step": 12276
+    },
+    {
+      "epoch": 52.48,
+      "learning_rate": 9.667990878007466e-05,
+      "loss": 0.0014,
+      "step": 12280
+    },
+    {
+      "epoch": 52.5,
+      "learning_rate": 9.667573152169652e-05,
+      "loss": 0.0002,
+      "step": 12284
+    },
+    {
+      "epoch": 52.51,
+      "learning_rate": 9.667155172746663e-05,
+      "loss": 0.001,
+      "step": 12288
+    },
+    {
+      "epoch": 52.53,
+      "learning_rate": 9.666736939761207e-05,
+      "loss": 0.0007,
+      "step": 12292
+    },
+    {
+      "epoch": 52.55,
+      "learning_rate": 9.666318453236006e-05,
+      "loss": 0.0021,
+      "step": 12296
+    },
+    {
+      "epoch": 52.56,
+      "learning_rate": 9.665899713193797e-05,
+      "loss": 0.0002,
+      "step": 12300
+    },
+    {
+      "epoch": 52.58,
+      "learning_rate": 9.665480719657327e-05,
+      "loss": 0.0011,
+      "step": 12304
+    },
+    {
+      "epoch": 52.6,
+      "learning_rate": 9.665061472649364e-05,
+      "loss": 0.0013,
+      "step": 12308
+    },
+    {
+      "epoch": 52.62,
+      "learning_rate": 9.66464197219268e-05,
+      "loss": 0.0003,
+      "step": 12312
+    },
+    {
+      "epoch": 52.63,
+      "learning_rate": 9.664222218310071e-05,
+      "loss": 0.0006,
+      "step": 12316
+    },
+    {
+      "epoch": 52.65,
+      "learning_rate": 9.663802211024337e-05,
+      "loss": 0.0003,
+      "step": 12320
+    },
+    {
+      "epoch": 52.67,
+      "learning_rate": 9.663381950358302e-05,
+      "loss": 0.0003,
+      "step": 12324
+    },
+    {
+      "epoch": 52.68,
+      "learning_rate": 9.662961436334795e-05,
+      "loss": 0.0006,
+      "step": 12328
+    },
+    {
+      "epoch": 52.7,
+      "learning_rate": 9.662540668976662e-05,
+      "loss": 0.0002,
+      "step": 12332
+    },
+    {
+      "epoch": 52.72,
+      "learning_rate": 9.662119648306767e-05,
+      "loss": 0.0003,
+      "step": 12336
+    },
+    {
+      "epoch": 52.74,
+      "learning_rate": 9.661698374347978e-05,
+      "loss": 0.0009,
+      "step": 12340
+    },
+    {
+      "epoch": 52.75,
+      "learning_rate": 9.661276847123184e-05,
+      "loss": 0.0005,
+      "step": 12344
+    },
+    {
+      "epoch": 52.77,
+      "learning_rate": 9.66085506665529e-05,
+      "loss": 0.002,
+      "step": 12348
+    },
+    {
+      "epoch": 52.79,
+      "learning_rate": 9.660433032967205e-05,
+      "loss": 0.0039,
+      "step": 12352
+    },
+    {
+      "epoch": 52.8,
+      "learning_rate": 9.660010746081864e-05,
+      "loss": 0.0002,
+      "step": 12356
+    },
+    {
+      "epoch": 52.82,
+      "learning_rate": 9.659588206022203e-05,
+      "loss": 0.0014,
+      "step": 12360
+    },
+    {
+      "epoch": 52.84,
+      "learning_rate": 9.659165412811184e-05,
+      "loss": 0.0005,
+      "step": 12364
+    },
+    {
+      "epoch": 52.85,
+      "learning_rate": 9.658742366471773e-05,
+      "loss": 0.0002,
+      "step": 12368
+    },
+    {
+      "epoch": 52.87,
+      "learning_rate": 9.658319067026958e-05,
+      "loss": 0.0003,
+      "step": 12372
+    },
+    {
+      "epoch": 52.89,
+      "learning_rate": 9.657895514499731e-05,
+      "loss": 0.002,
+      "step": 12376
+    },
+    {
+      "epoch": 52.91,
+      "learning_rate": 9.657471708913108e-05,
+      "loss": 0.0008,
+      "step": 12380
+    },
+    {
+      "epoch": 52.92,
+      "learning_rate": 9.65704765029011e-05,
+      "loss": 0.0003,
+      "step": 12384
+    },
+    {
+      "epoch": 52.94,
+      "learning_rate": 9.656623338653778e-05,
+      "loss": 0.0002,
+      "step": 12388
+    },
+    {
+      "epoch": 52.96,
+      "learning_rate": 9.656198774027167e-05,
+      "loss": 0.0003,
+      "step": 12392
+    },
+    {
+      "epoch": 52.97,
+      "learning_rate": 9.655773956433339e-05,
+      "loss": 0.001,
+      "step": 12396
+    },
+    {
+      "epoch": 52.99,
+      "learning_rate": 9.655348885895375e-05,
+      "loss": 0.0004,
+      "step": 12400
+    },
+    {
+      "epoch": 53.01,
+      "learning_rate": 9.654923562436368e-05,
+      "loss": 0.0008,
+      "step": 12404
+    },
+    {
+      "epoch": 53.03,
+      "learning_rate": 9.654497986079427e-05,
+      "loss": 0.0006,
+      "step": 12408
+    },
+    {
+      "epoch": 53.04,
+      "learning_rate": 9.654072156847673e-05,
+      "loss": 0.0003,
+      "step": 12412
+    },
+    {
+      "epoch": 53.06,
+      "learning_rate": 9.653646074764242e-05,
+      "loss": 0.0026,
+      "step": 12416
+    },
+    {
+      "epoch": 53.08,
+      "learning_rate": 9.653219739852281e-05,
+      "loss": 0.0017,
+      "step": 12420
+    },
+    {
+      "epoch": 53.09,
+      "learning_rate": 9.652793152134951e-05,
+      "loss": 0.0005,
+      "step": 12424
+    },
+    {
+      "epoch": 53.11,
+      "learning_rate": 9.652366311635432e-05,
+      "loss": 0.0005,
+      "step": 12428
+    },
+    {
+      "epoch": 53.13,
+      "learning_rate": 9.65193921837691e-05,
+      "loss": 0.0003,
+      "step": 12432
+    },
+    {
+      "epoch": 53.15,
+      "learning_rate": 9.651511872382593e-05,
+      "loss": 0.0009,
+      "step": 12436
+    },
+    {
+      "epoch": 53.16,
+      "learning_rate": 9.651084273675695e-05,
+      "loss": 0.0003,
+      "step": 12440
+    },
+    {
+      "epoch": 53.18,
+      "learning_rate": 9.650656422279446e-05,
+      "loss": 0.0009,
+      "step": 12444
+    },
+    {
+      "epoch": 53.2,
+      "learning_rate": 9.650228318217094e-05,
+      "loss": 0.001,
+      "step": 12448
+    },
+    {
+      "epoch": 53.21,
+      "learning_rate": 9.649799961511895e-05,
+      "loss": 0.0011,
+      "step": 12452
+    },
+    {
+      "epoch": 53.23,
+      "learning_rate": 9.649371352187123e-05,
+      "loss": 0.0002,
+      "step": 12456
+    },
+    {
+      "epoch": 53.25,
+      "learning_rate": 9.648942490266064e-05,
+      "loss": 0.0004,
+      "step": 12460
+    },
+    {
+      "epoch": 53.26,
+      "learning_rate": 9.648513375772016e-05,
+      "loss": 0.0004,
+      "step": 12464
+    },
+    {
+      "epoch": 53.28,
+      "learning_rate": 9.648084008728293e-05,
+      "loss": 0.0008,
+      "step": 12468
+    },
+    {
+      "epoch": 53.3,
+      "learning_rate": 9.647654389158223e-05,
+      "loss": 0.0003,
+      "step": 12472
+    },
+    {
+      "epoch": 53.32,
+      "learning_rate": 9.647224517085147e-05,
+      "loss": 0.0012,
+      "step": 12476
+    },
+    {
+      "epoch": 53.33,
+      "learning_rate": 9.64679439253242e-05,
+      "loss": 0.0001,
+      "step": 12480
+    },
+    {
+      "epoch": 53.35,
+      "learning_rate": 9.646364015523408e-05,
+      "loss": 0.001,
+      "step": 12484
+    },
+    {
+      "epoch": 53.37,
+      "learning_rate": 9.645933386081494e-05,
+      "loss": 0.0009,
+      "step": 12488
+    },
+    {
+      "epoch": 53.38,
+      "learning_rate": 9.645502504230075e-05,
+      "loss": 0.0007,
+      "step": 12492
+    },
+    {
+      "epoch": 53.4,
+      "learning_rate": 9.645071369992559e-05,
+      "loss": 0.0005,
+      "step": 12496
+    },
+    {
+      "epoch": 53.42,
+      "learning_rate": 9.644639983392369e-05,
+      "loss": 0.0004,
+      "step": 12500
+    },
+    {
+      "epoch": 53.44,
+      "learning_rate": 9.644208344452943e-05,
+      "loss": 0.0002,
+      "step": 12504
+    },
+    {
+      "epoch": 53.45,
+      "learning_rate": 9.64377645319773e-05,
+      "loss": 0.0002,
+      "step": 12508
+    },
+    {
+      "epoch": 53.47,
+      "learning_rate": 9.643344309650197e-05,
+      "loss": 0.0003,
+      "step": 12512
+    },
+    {
+      "epoch": 53.49,
+      "learning_rate": 9.642911913833821e-05,
+      "loss": 0.0004,
+      "step": 12516
+    },
+    {
+      "epoch": 53.5,
+      "learning_rate": 9.642479265772091e-05,
+      "loss": 0.0001,
+      "step": 12520
+    },
+    {
+      "epoch": 53.52,
+      "learning_rate": 9.642046365488516e-05,
+      "loss": 0.0009,
+      "step": 12524
+    },
+    {
+      "epoch": 53.54,
+      "learning_rate": 9.64161321300661e-05,
+      "loss": 0.0019,
+      "step": 12528
+    },
+    {
+      "epoch": 53.56,
+      "learning_rate": 9.641179808349913e-05,
+      "loss": 0.0002,
+      "step": 12532
+    },
+    {
+      "epoch": 53.57,
+      "learning_rate": 9.640746151541966e-05,
+      "loss": 0.0009,
+      "step": 12536
+    },
+    {
+      "epoch": 53.59,
+      "learning_rate": 9.640312242606332e-05,
+      "loss": 0.0007,
+      "step": 12540
+    },
+    {
+      "epoch": 53.61,
+      "learning_rate": 9.639878081566582e-05,
+      "loss": 0.0006,
+      "step": 12544
+    },
+    {
+      "epoch": 53.62,
+      "learning_rate": 9.639443668446307e-05,
+      "loss": 0.0004,
+      "step": 12548
+    },
+    {
+      "epoch": 53.64,
+      "learning_rate": 9.639009003269108e-05,
+      "loss": 0.0008,
+      "step": 12552
+    },
+    {
+      "epoch": 53.66,
+      "learning_rate": 9.638574086058597e-05,
+      "loss": 0.0005,
+      "step": 12556
+    },
+    {
+      "epoch": 53.68,
+      "learning_rate": 9.638138916838404e-05,
+      "loss": 0.0005,
+      "step": 12560
+    },
+    {
+      "epoch": 53.69,
+      "learning_rate": 9.637703495632171e-05,
+      "loss": 0.0003,
+      "step": 12564
+    },
+    {
+      "epoch": 53.71,
+      "learning_rate": 9.637267822463556e-05,
+      "loss": 0.0019,
+      "step": 12568
+    },
+    {
+      "epoch": 53.73,
+      "learning_rate": 9.63683189735623e-05,
+      "loss": 0.002,
+      "step": 12572
+    },
+    {
+      "epoch": 53.74,
+      "learning_rate": 9.63639572033387e-05,
+      "loss": 0.0005,
+      "step": 12576
+    },
+    {
+      "epoch": 53.76,
+      "learning_rate": 9.635959291420178e-05,
+      "loss": 0.0002,
+      "step": 12580
+    },
+    {
+      "epoch": 53.78,
+      "learning_rate": 9.635522610638865e-05,
+      "loss": 0.0006,
+      "step": 12584
+    },
+    {
+      "epoch": 53.79,
+      "learning_rate": 9.635085678013653e-05,
+      "loss": 0.0015,
+      "step": 12588
+    },
+    {
+      "epoch": 53.81,
+      "learning_rate": 9.634648493568284e-05,
+      "loss": 0.001,
+      "step": 12592
+    },
+    {
+      "epoch": 53.83,
+      "learning_rate": 9.634211057326506e-05,
+      "loss": 0.0006,
+      "step": 12596
+    },
+    {
+      "epoch": 53.85,
+      "learning_rate": 9.633773369312085e-05,
+      "loss": 0.0018,
+      "step": 12600
+    },
+    {
+      "epoch": 53.86,
+      "learning_rate": 9.633335429548801e-05,
+      "loss": 0.0001,
+      "step": 12604
+    },
+    {
+      "epoch": 53.88,
+      "learning_rate": 9.632897238060448e-05,
+      "loss": 0.0004,
+      "step": 12608
+    },
+    {
+      "epoch": 53.9,
+      "learning_rate": 9.632458794870831e-05,
+      "loss": 0.0008,
+      "step": 12612
+    },
+    {
+      "epoch": 53.91,
+      "learning_rate": 9.632020100003771e-05,
+      "loss": 0.0008,
+      "step": 12616
+    },
+    {
+      "epoch": 53.93,
+      "learning_rate": 9.631581153483102e-05,
+      "loss": 0.0005,
+      "step": 12620
+    },
+    {
+      "epoch": 53.95,
+      "learning_rate": 9.631141955332673e-05,
+      "loss": 0.0011,
+      "step": 12624
+    },
+    {
+      "epoch": 53.97,
+      "learning_rate": 9.630702505576341e-05,
+      "loss": 0.0003,
+      "step": 12628
+    },
+    {
+      "epoch": 53.98,
+      "learning_rate": 9.630262804237984e-05,
+      "loss": 0.0006,
+      "step": 12632
+    },
+    {
+      "epoch": 54.0,
+      "learning_rate": 9.62982285134149e-05,
+      "loss": 0.0003,
+      "step": 12636
+    },
+    {
+      "epoch": 54.02,
+      "learning_rate": 9.629382646910762e-05,
+      "loss": 0.0003,
+      "step": 12640
+    },
+    {
+      "epoch": 54.03,
+      "learning_rate": 9.628942190969715e-05,
+      "loss": 0.0023,
+      "step": 12644
+    },
+    {
+      "epoch": 54.05,
+      "learning_rate": 9.628501483542278e-05,
+      "loss": 0.0008,
+      "step": 12648
+    },
+    {
+      "epoch": 54.07,
+      "learning_rate": 9.628060524652395e-05,
+      "loss": 0.0004,
+      "step": 12652
+    },
+    {
+      "epoch": 54.09,
+      "learning_rate": 9.627619314324025e-05,
+      "loss": 0.0001,
+      "step": 12656
+    },
+    {
+      "epoch": 54.1,
+      "learning_rate": 9.627177852581135e-05,
+      "loss": 0.0003,
+      "step": 12660
+    },
+    {
+      "epoch": 54.12,
+      "learning_rate": 9.626736139447711e-05,
+      "loss": 0.0003,
+      "step": 12664
+    },
+    {
+      "epoch": 54.14,
+      "learning_rate": 9.62629417494775e-05,
+      "loss": 0.0002,
+      "step": 12668
+    },
+    {
+      "epoch": 54.15,
+      "learning_rate": 9.625851959105265e-05,
+      "loss": 0.0006,
+      "step": 12672
+    },
+    {
+      "epoch": 54.17,
+      "learning_rate": 9.625409491944282e-05,
+      "loss": 0.0021,
+      "step": 12676
+    },
+    {
+      "epoch": 54.19,
+      "learning_rate": 9.624966773488834e-05,
+      "loss": 0.0002,
+      "step": 12680
+    },
+    {
+      "epoch": 54.21,
+      "learning_rate": 9.624523803762982e-05,
+      "loss": 0.0002,
+      "step": 12684
+    },
+    {
+      "epoch": 54.22,
+      "learning_rate": 9.624080582790786e-05,
+      "loss": 0.0003,
+      "step": 12688
+    },
+    {
+      "epoch": 54.24,
+      "learning_rate": 9.623637110596328e-05,
+      "loss": 0.0006,
+      "step": 12692
+    },
+    {
+      "epoch": 54.26,
+      "learning_rate": 9.623193387203701e-05,
+      "loss": 0.0001,
+      "step": 12696
+    },
+    {
+      "epoch": 54.27,
+      "learning_rate": 9.622749412637014e-05,
+      "loss": 0.0001,
+      "step": 12700
+    },
+    {
+      "epoch": 54.29,
+      "learning_rate": 9.622305186920384e-05,
+      "loss": 0.0003,
+      "step": 12704
+    },
+    {
+      "epoch": 54.31,
+      "learning_rate": 9.621860710077949e-05,
+      "loss": 0.0013,
+      "step": 12708
+    },
+    {
+      "epoch": 54.32,
+      "learning_rate": 9.621415982133855e-05,
+      "loss": 0.0003,
+      "step": 12712
+    },
+    {
+      "epoch": 54.34,
+      "learning_rate": 9.620971003112263e-05,
+      "loss": 0.0003,
+      "step": 12716
+    },
+    {
+      "epoch": 54.36,
+      "learning_rate": 9.62052577303735e-05,
+      "loss": 0.0001,
+      "step": 12720
+    },
+    {
+      "epoch": 54.38,
+      "learning_rate": 9.620080291933307e-05,
+      "loss": 0.0009,
+      "step": 12724
+    },
+    {
+      "epoch": 54.39,
+      "learning_rate": 9.619634559824331e-05,
+      "loss": 0.0008,
+      "step": 12728
+    },
+    {
+      "epoch": 54.41,
+      "learning_rate": 9.619188576734642e-05,
+      "loss": 0.0001,
+      "step": 12732
+    },
+    {
+      "epoch": 54.43,
+      "learning_rate": 9.618742342688469e-05,
+      "loss": 0.0011,
+      "step": 12736
+    },
+    {
+      "epoch": 54.44,
+      "learning_rate": 9.618295857710057e-05,
+      "loss": 0.002,
+      "step": 12740
+    },
+    {
+      "epoch": 54.46,
+      "learning_rate": 9.61784912182366e-05,
+      "loss": 0.0002,
+      "step": 12744
+    },
+    {
+      "epoch": 54.48,
+      "learning_rate": 9.617402135053552e-05,
+      "loss": 0.0001,
+      "step": 12748
+    },
+    {
+      "epoch": 54.5,
+      "learning_rate": 9.616954897424015e-05,
+      "loss": 0.0009,
+      "step": 12752
+    },
+    {
+      "epoch": 54.51,
+      "learning_rate": 9.61650740895935e-05,
+      "loss": 0.0002,
+      "step": 12756
+    },
+    {
+      "epoch": 54.53,
+      "learning_rate": 9.616059669683864e-05,
+      "loss": 0.0002,
+      "step": 12760
+    },
+    {
+      "epoch": 54.55,
+      "learning_rate": 9.615611679621887e-05,
+      "loss": 0.0007,
+      "step": 12764
+    },
+    {
+      "epoch": 54.56,
+      "learning_rate": 9.615163438797756e-05,
+      "loss": 0.0003,
+      "step": 12768
+    },
+    {
+      "epoch": 54.58,
+      "learning_rate": 9.614714947235821e-05,
+      "loss": 0.0008,
+      "step": 12772
+    },
+    {
+      "epoch": 54.6,
+      "learning_rate": 9.614266204960453e-05,
+      "loss": 0.0005,
+      "step": 12776
+    },
+    {
+      "epoch": 54.62,
+      "learning_rate": 9.613817211996027e-05,
+      "loss": 0.0013,
+      "step": 12780
+    },
+    {
+      "epoch": 54.63,
+      "learning_rate": 9.613367968366941e-05,
+      "loss": 0.0002,
+      "step": 12784
+    },
+    {
+      "epoch": 54.65,
+      "learning_rate": 9.612918474097599e-05,
+      "loss": 0.0007,
+      "step": 12788
+    },
+    {
+      "epoch": 54.67,
+      "learning_rate": 9.612468729212423e-05,
+      "loss": 0.0006,
+      "step": 12792
+    },
+    {
+      "epoch": 54.68,
+      "learning_rate": 9.612018733735844e-05,
+      "loss": 0.0001,
+      "step": 12796
+    },
+    {
+      "epoch": 54.7,
+      "learning_rate": 9.611568487692316e-05,
+      "loss": 0.0004,
+      "step": 12800
+    },
+    {
+      "epoch": 54.72,
+      "learning_rate": 9.611117991106295e-05,
+      "loss": 0.0002,
+      "step": 12804
+    },
+    {
+      "epoch": 54.74,
+      "learning_rate": 9.610667244002259e-05,
+      "loss": 0.0006,
+      "step": 12808
+    },
+    {
+      "epoch": 54.75,
+      "learning_rate": 9.610216246404694e-05,
+      "loss": 0.0022,
+      "step": 12812
+    },
+    {
+      "epoch": 54.77,
+      "learning_rate": 9.609764998338105e-05,
+      "loss": 0.0012,
+      "step": 12816
+    },
+    {
+      "epoch": 54.79,
+      "learning_rate": 9.609313499827008e-05,
+      "loss": 0.0018,
+      "step": 12820
+    },
+    {
+      "epoch": 54.8,
+      "learning_rate": 9.60886175089593e-05,
+      "loss": 0.0006,
+      "step": 12824
+    },
+    {
+      "epoch": 54.82,
+      "learning_rate": 9.608409751569417e-05,
+      "loss": 0.0012,
+      "step": 12828
+    },
+    {
+      "epoch": 54.84,
+      "learning_rate": 9.607957501872022e-05,
+      "loss": 0.0001,
+      "step": 12832
+    },
+    {
+      "epoch": 54.85,
+      "learning_rate": 9.607505001828321e-05,
+      "loss": 0.0002,
+      "step": 12836
+    },
+    {
+      "epoch": 54.87,
+      "learning_rate": 9.607052251462892e-05,
+      "loss": 0.0004,
+      "step": 12840
+    },
+    {
+      "epoch": 54.89,
+      "learning_rate": 9.606599250800336e-05,
+      "loss": 0.0002,
+      "step": 12844
+    },
+    {
+      "epoch": 54.91,
+      "learning_rate": 9.606145999865264e-05,
+      "loss": 0.0028,
+      "step": 12848
+    },
+    {
+      "epoch": 54.92,
+      "learning_rate": 9.6056924986823e-05,
+      "loss": 0.0002,
+      "step": 12852
+    },
+    {
+      "epoch": 54.94,
+      "learning_rate": 9.60523874727608e-05,
+      "loss": 0.0002,
+      "step": 12856
+    },
+    {
+      "epoch": 54.96,
+      "learning_rate": 9.60478474567126e-05,
+      "loss": 0.0014,
+      "step": 12860
+    },
+    {
+      "epoch": 54.97,
+      "learning_rate": 9.604330493892504e-05,
+      "loss": 0.0003,
+      "step": 12864
+    },
+    {
+      "epoch": 54.99,
+      "learning_rate": 9.60387599196449e-05,
+      "loss": 0.0014,
+      "step": 12868
+    },
+    {
+      "epoch": 55.01,
+      "learning_rate": 9.603421239911912e-05,
+      "loss": 0.0002,
+      "step": 12872
+    },
+    {
+      "epoch": 55.03,
+      "learning_rate": 9.602966237759475e-05,
+      "loss": 0.0001,
+      "step": 12876
+    },
+    {
+      "epoch": 55.04,
+      "learning_rate": 9.6025109855319e-05,
+      "loss": 0.0002,
+      "step": 12880
+    },
+    {
+      "epoch": 55.06,
+      "learning_rate": 9.602055483253923e-05,
+      "loss": 0.0002,
+      "step": 12884
+    },
+    {
+      "epoch": 55.08,
+      "learning_rate": 9.601599730950284e-05,
+      "loss": 0.0005,
+      "step": 12888
+    },
+    {
+      "epoch": 55.09,
+      "learning_rate": 9.601143728645749e-05,
+      "loss": 0.0013,
+      "step": 12892
+    },
+    {
+      "epoch": 55.11,
+      "learning_rate": 9.600687476365092e-05,
+      "loss": 0.0005,
+      "step": 12896
+    },
+    {
+      "epoch": 55.13,
+      "learning_rate": 9.600230974133099e-05,
+      "loss": 0.0002,
+      "step": 12900
+    },
+    {
+      "epoch": 55.15,
+      "learning_rate": 9.599774221974573e-05,
+      "loss": 0.0003,
+      "step": 12904
+    },
+    {
+      "epoch": 55.16,
+      "learning_rate": 9.599317219914328e-05,
+      "loss": 0.0005,
+      "step": 12908
+    },
+    {
+      "epoch": 55.18,
+      "learning_rate": 9.598859967977193e-05,
+      "loss": 0.002,
+      "step": 12912
+    },
+    {
+      "epoch": 55.2,
+      "learning_rate": 9.598402466188009e-05,
+      "loss": 0.0008,
+      "step": 12916
+    },
+    {
+      "epoch": 55.21,
+      "learning_rate": 9.597944714571631e-05,
+      "loss": 0.0002,
+      "step": 12920
+    },
+    {
+      "epoch": 55.23,
+      "learning_rate": 9.597486713152933e-05,
+      "loss": 0.0002,
+      "step": 12924
+    },
+    {
+      "epoch": 55.25,
+      "learning_rate": 9.597028461956792e-05,
+      "loss": 0.0011,
+      "step": 12928
+    },
+    {
+      "epoch": 55.26,
+      "learning_rate": 9.596569961008108e-05,
+      "loss": 0.0002,
+      "step": 12932
+    },
+    {
+      "epoch": 55.28,
+      "learning_rate": 9.596111210331788e-05,
+      "loss": 0.0002,
+      "step": 12936
+    },
+    {
+      "epoch": 55.3,
+      "learning_rate": 9.59565220995276e-05,
+      "loss": 0.0016,
+      "step": 12940
+    },
+    {
+      "epoch": 55.32,
+      "learning_rate": 9.595192959895958e-05,
+      "loss": 0.0009,
+      "step": 12944
+    },
+    {
+      "epoch": 55.33,
+      "learning_rate": 9.594733460186331e-05,
+      "loss": 0.0012,
+      "step": 12948
+    },
+    {
+      "epoch": 55.35,
+      "learning_rate": 9.594273710848846e-05,
+      "loss": 0.0018,
+      "step": 12952
+    },
+    {
+      "epoch": 55.37,
+      "learning_rate": 9.593813711908482e-05,
+      "loss": 0.001,
+      "step": 12956
+    },
+    {
+      "epoch": 55.38,
+      "learning_rate": 9.593353463390227e-05,
+      "loss": 0.0003,
+      "step": 12960
+    },
+    {
+      "epoch": 55.4,
+      "learning_rate": 9.592892965319088e-05,
+      "loss": 0.0011,
+      "step": 12964
+    },
+    {
+      "epoch": 55.42,
+      "learning_rate": 9.592432217720082e-05,
+      "loss": 0.0001,
+      "step": 12968
+    },
+    {
+      "epoch": 55.44,
+      "learning_rate": 9.59197122061824e-05,
+      "loss": 0.0009,
+      "step": 12972
+    },
+    {
+      "epoch": 55.45,
+      "learning_rate": 9.591509974038612e-05,
+      "loss": 0.0019,
+      "step": 12976
+    },
+    {
+      "epoch": 55.47,
+      "learning_rate": 9.591048478006255e-05,
+      "loss": 0.001,
+      "step": 12980
+    },
+    {
+      "epoch": 55.49,
+      "learning_rate": 9.590586732546241e-05,
+      "loss": 0.0003,
+      "step": 12984
+    },
+    {
+      "epoch": 55.5,
+      "learning_rate": 9.590124737683655e-05,
+      "loss": 0.0017,
+      "step": 12988
+    },
+    {
+      "epoch": 55.52,
+      "learning_rate": 9.589662493443599e-05,
+      "loss": 0.0005,
+      "step": 12992
+    },
+    {
+      "epoch": 55.54,
+      "learning_rate": 9.589199999851183e-05,
+      "loss": 0.0007,
+      "step": 12996
+    },
+    {
+      "epoch": 55.56,
+      "learning_rate": 9.58873725693154e-05,
+      "loss": 0.0002,
+      "step": 13000
+    },
+    {
+      "epoch": 55.56,
+      "eval_exact_match": 0.5145530145530145,
+      "eval_loss": 0.8539115190505981,
+      "eval_runtime": 141.8733,
+      "eval_samples_per_second": 6.781,
+      "step": 13000
+    },
+    {
+      "epoch": 55.57,
+      "learning_rate": 9.588274264709806e-05,
+      "loss": 0.0008,
+      "step": 13004
+    },
+    {
+      "epoch": 55.59,
+      "learning_rate": 9.587811023211135e-05,
+      "loss": 0.0002,
+      "step": 13008
+    },
+    {
+      "epoch": 55.61,
+      "learning_rate": 9.587347532460695e-05,
+      "loss": 0.0002,
+      "step": 13012
+    },
+    {
+      "epoch": 55.62,
+      "learning_rate": 9.586883792483668e-05,
+      "loss": 0.0003,
+      "step": 13016
+    },
+    {
+      "epoch": 55.64,
+      "learning_rate": 9.586419803305248e-05,
+      "loss": 0.0012,
+      "step": 13020
+    },
+    {
+      "epoch": 55.66,
+      "learning_rate": 9.585955564950643e-05,
+      "loss": 0.0015,
+      "step": 13024
+    },
+    {
+      "epoch": 55.68,
+      "learning_rate": 9.585491077445075e-05,
+      "loss": 0.0005,
+      "step": 13028
+    },
+    {
+      "epoch": 55.69,
+      "learning_rate": 9.585026340813776e-05,
+      "loss": 0.0013,
+      "step": 13032
+    },
+    {
+      "epoch": 55.71,
+      "learning_rate": 9.584561355082001e-05,
+      "loss": 0.0002,
+      "step": 13036
+    },
+    {
+      "epoch": 55.73,
+      "learning_rate": 9.584096120275007e-05,
+      "loss": 0.001,
+      "step": 13040
+    },
+    {
+      "epoch": 55.74,
+      "learning_rate": 9.583630636418074e-05,
+      "loss": 0.0005,
+      "step": 13044
+    },
+    {
+      "epoch": 55.76,
+      "learning_rate": 9.583164903536488e-05,
+      "loss": 0.0005,
+      "step": 13048
+    },
+    {
+      "epoch": 55.78,
+      "learning_rate": 9.582698921655551e-05,
+      "loss": 0.0014,
+      "step": 13052
+    },
+    {
+      "epoch": 55.79,
+      "learning_rate": 9.582232690800584e-05,
+      "loss": 0.0001,
+      "step": 13056
+    },
+    {
+      "epoch": 55.81,
+      "learning_rate": 9.581766210996911e-05,
+      "loss": 0.0007,
+      "step": 13060
+    },
+    {
+      "epoch": 55.83,
+      "learning_rate": 9.581299482269882e-05,
+      "loss": 0.0001,
+      "step": 13064
+    },
+    {
+      "epoch": 55.85,
+      "learning_rate": 9.58083250464485e-05,
+      "loss": 0.0005,
+      "step": 13068
+    },
+    {
+      "epoch": 55.86,
+      "learning_rate": 9.580365278147185e-05,
+      "loss": 0.0008,
+      "step": 13072
+    },
+    {
+      "epoch": 55.88,
+      "learning_rate": 9.579897802802273e-05,
+      "loss": 0.0006,
+      "step": 13076
+    },
+    {
+      "epoch": 55.9,
+      "learning_rate": 9.579430078635509e-05,
+      "loss": 0.0004,
+      "step": 13080
+    },
+    {
+      "epoch": 55.91,
+      "learning_rate": 9.578962105672307e-05,
+      "loss": 0.0019,
+      "step": 13084
+    },
+    {
+      "epoch": 55.93,
+      "learning_rate": 9.57849388393809e-05,
+      "loss": 0.0005,
+      "step": 13088
+    },
+    {
+      "epoch": 55.95,
+      "learning_rate": 9.578025413458296e-05,
+      "loss": 0.0002,
+      "step": 13092
+    },
+    {
+      "epoch": 55.97,
+      "learning_rate": 9.577556694258378e-05,
+      "loss": 0.0002,
+      "step": 13096
+    },
+    {
+      "epoch": 55.98,
+      "learning_rate": 9.577087726363798e-05,
+      "loss": 0.0028,
+      "step": 13100
+    },
+    {
+      "epoch": 56.0,
+      "learning_rate": 9.57661850980004e-05,
+      "loss": 0.0002,
+      "step": 13104
+    },
+    {
+      "epoch": 56.02,
+      "learning_rate": 9.576149044592589e-05,
+      "loss": 0.0006,
+      "step": 13108
+    },
+    {
+      "epoch": 56.03,
+      "learning_rate": 9.575679330766955e-05,
+      "loss": 0.0017,
+      "step": 13112
+    },
+    {
+      "epoch": 56.05,
+      "learning_rate": 9.575209368348659e-05,
+      "loss": 0.0006,
+      "step": 13116
+    },
+    {
+      "epoch": 56.07,
+      "learning_rate": 9.57473915736323e-05,
+      "loss": 0.0005,
+      "step": 13120
+    },
+    {
+      "epoch": 56.09,
+      "learning_rate": 9.574268697836216e-05,
+      "loss": 0.0003,
+      "step": 13124
+    },
+    {
+      "epoch": 56.1,
+      "learning_rate": 9.573797989793174e-05,
+      "loss": 0.0005,
+      "step": 13128
+    },
+    {
+      "epoch": 56.12,
+      "learning_rate": 9.57332703325968e-05,
+      "loss": 0.0001,
+      "step": 13132
+    },
+    {
+      "epoch": 56.14,
+      "learning_rate": 9.57285582826132e-05,
+      "loss": 0.0011,
+      "step": 13136
+    },
+    {
+      "epoch": 56.15,
+      "learning_rate": 9.572384374823695e-05,
+      "loss": 0.0011,
+      "step": 13140
+    },
+    {
+      "epoch": 56.17,
+      "learning_rate": 9.571912672972416e-05,
+      "loss": 0.0019,
+      "step": 13144
+    },
+    {
+      "epoch": 56.19,
+      "learning_rate": 9.571440722733112e-05,
+      "loss": 0.0002,
+      "step": 13148
+    },
+    {
+      "epoch": 56.21,
+      "learning_rate": 9.570968524131423e-05,
+      "loss": 0.002,
+      "step": 13152
+    },
+    {
+      "epoch": 56.22,
+      "learning_rate": 9.570496077193005e-05,
+      "loss": 0.0003,
+      "step": 13156
+    },
+    {
+      "epoch": 56.24,
+      "learning_rate": 9.570023381943523e-05,
+      "loss": 0.001,
+      "step": 13160
+    },
+    {
+      "epoch": 56.26,
+      "learning_rate": 9.56955043840866e-05,
+      "loss": 0.0004,
+      "step": 13164
+    },
+    {
+      "epoch": 56.27,
+      "learning_rate": 9.56907724661411e-05,
+      "loss": 0.0009,
+      "step": 13168
+    },
+    {
+      "epoch": 56.29,
+      "learning_rate": 9.568603806585581e-05,
+      "loss": 0.0002,
+      "step": 13172
+    },
+    {
+      "epoch": 56.31,
+      "learning_rate": 9.568130118348793e-05,
+      "loss": 0.0009,
+      "step": 13176
+    },
+    {
+      "epoch": 56.32,
+      "learning_rate": 9.567656181929485e-05,
+      "loss": 0.0005,
+      "step": 13180
+    },
+    {
+      "epoch": 56.34,
+      "learning_rate": 9.567181997353403e-05,
+      "loss": 0.0015,
+      "step": 13184
+    },
+    {
+      "epoch": 56.36,
+      "learning_rate": 9.566707564646307e-05,
+      "loss": 0.0003,
+      "step": 13188
+    },
+    {
+      "epoch": 56.38,
+      "learning_rate": 9.566232883833977e-05,
+      "loss": 0.0001,
+      "step": 13192
+    },
+    {
+      "epoch": 56.39,
+      "learning_rate": 9.565757954942198e-05,
+      "loss": 0.0012,
+      "step": 13196
+    },
+    {
+      "epoch": 56.41,
+      "learning_rate": 9.565282777996775e-05,
+      "loss": 0.0071,
+      "step": 13200
+    },
+    {
+      "epoch": 56.43,
+      "learning_rate": 9.564807353023523e-05,
+      "loss": 0.0007,
+      "step": 13204
+    },
+    {
+      "epoch": 56.44,
+      "learning_rate": 9.564331680048273e-05,
+      "loss": 0.0008,
+      "step": 13208
+    },
+    {
+      "epoch": 56.46,
+      "learning_rate": 9.563855759096864e-05,
+      "loss": 0.0009,
+      "step": 13212
+    },
+    {
+      "epoch": 56.48,
+      "learning_rate": 9.563379590195157e-05,
+      "loss": 0.0002,
+      "step": 13216
+    },
+    {
+      "epoch": 56.5,
+      "learning_rate": 9.56290317336902e-05,
+      "loss": 0.001,
+      "step": 13220
+    },
+    {
+      "epoch": 56.51,
+      "learning_rate": 9.562426508644336e-05,
+      "loss": 0.0009,
+      "step": 13224
+    },
+    {
+      "epoch": 56.53,
+      "learning_rate": 9.561949596046999e-05,
+      "loss": 0.0003,
+      "step": 13228
+    },
+    {
+      "epoch": 56.55,
+      "learning_rate": 9.561472435602924e-05,
+      "loss": 0.0023,
+      "step": 13232
+    },
+    {
+      "epoch": 56.56,
+      "learning_rate": 9.560995027338033e-05,
+      "loss": 0.0003,
+      "step": 13236
+    },
+    {
+      "epoch": 56.58,
+      "learning_rate": 9.560517371278262e-05,
+      "loss": 0.0014,
+      "step": 13240
+    },
+    {
+      "epoch": 56.6,
+      "learning_rate": 9.560039467449563e-05,
+      "loss": 0.0001,
+      "step": 13244
+    },
+    {
+      "epoch": 56.62,
+      "learning_rate": 9.5595613158779e-05,
+      "loss": 0.0003,
+      "step": 13248
+    },
+    {
+      "epoch": 56.63,
+      "learning_rate": 9.55908291658925e-05,
+      "loss": 0.0009,
+      "step": 13252
+    },
+    {
+      "epoch": 56.65,
+      "learning_rate": 9.558604269609604e-05,
+      "loss": 0.0009,
+      "step": 13256
+    },
+    {
+      "epoch": 56.67,
+      "learning_rate": 9.558125374964967e-05,
+      "loss": 0.001,
+      "step": 13260
+    },
+    {
+      "epoch": 56.68,
+      "learning_rate": 9.557646232681356e-05,
+      "loss": 0.0001,
+      "step": 13264
+    },
+    {
+      "epoch": 56.7,
+      "learning_rate": 9.557166842784804e-05,
+      "loss": 0.0002,
+      "step": 13268
+    },
+    {
+      "epoch": 56.72,
+      "learning_rate": 9.556687205301354e-05,
+      "loss": 0.0011,
+      "step": 13272
+    },
+    {
+      "epoch": 56.74,
+      "learning_rate": 9.556207320257065e-05,
+      "loss": 0.0011,
+      "step": 13276
+    },
+    {
+      "epoch": 56.75,
+      "learning_rate": 9.55572718767801e-05,
+      "loss": 0.0009,
+      "step": 13280
+    },
+    {
+      "epoch": 56.77,
+      "learning_rate": 9.555246807590273e-05,
+      "loss": 0.0006,
+      "step": 13284
+    },
+    {
+      "epoch": 56.79,
+      "learning_rate": 9.554766180019952e-05,
+      "loss": 0.0018,
+      "step": 13288
+    },
+    {
+      "epoch": 56.8,
+      "learning_rate": 9.55428530499316e-05,
+      "loss": 0.0004,
+      "step": 13292
+    },
+    {
+      "epoch": 56.82,
+      "learning_rate": 9.553804182536021e-05,
+      "loss": 0.0015,
+      "step": 13296
+    },
+    {
+      "epoch": 56.84,
+      "learning_rate": 9.553322812674676e-05,
+      "loss": 0.0004,
+      "step": 13300
+    },
+    {
+      "epoch": 56.85,
+      "learning_rate": 9.552841195435278e-05,
+      "loss": 0.0004,
+      "step": 13304
+    },
+    {
+      "epoch": 56.87,
+      "learning_rate": 9.552359330843991e-05,
+      "loss": 0.0004,
+      "step": 13308
+    },
+    {
+      "epoch": 56.89,
+      "learning_rate": 9.551877218926994e-05,
+      "loss": 0.0014,
+      "step": 13312
+    },
+    {
+      "epoch": 56.91,
+      "learning_rate": 9.55139485971048e-05,
+      "loss": 0.0002,
+      "step": 13316
+    },
+    {
+      "epoch": 56.92,
+      "learning_rate": 9.550912253220656e-05,
+      "loss": 0.0017,
+      "step": 13320
+    },
+    {
+      "epoch": 56.94,
+      "learning_rate": 9.550429399483743e-05,
+      "loss": 0.0005,
+      "step": 13324
+    },
+    {
+      "epoch": 56.96,
+      "learning_rate": 9.549946298525971e-05,
+      "loss": 0.0004,
+      "step": 13328
+    },
+    {
+      "epoch": 56.97,
+      "learning_rate": 9.549462950373587e-05,
+      "loss": 0.001,
+      "step": 13332
+    },
+    {
+      "epoch": 56.99,
+      "learning_rate": 9.548979355052853e-05,
+      "loss": 0.0013,
+      "step": 13336
+    },
+    {
+      "epoch": 57.01,
+      "learning_rate": 9.54849551259004e-05,
+      "loss": 0.0002,
+      "step": 13340
+    },
+    {
+      "epoch": 57.03,
+      "learning_rate": 9.548011423011435e-05,
+      "loss": 0.0002,
+      "step": 13344
+    },
+    {
+      "epoch": 57.04,
+      "learning_rate": 9.54752708634334e-05,
+      "loss": 0.0002,
+      "step": 13348
+    },
+    {
+      "epoch": 57.06,
+      "learning_rate": 9.547042502612067e-05,
+      "loss": 0.0004,
+      "step": 13352
+    },
+    {
+      "epoch": 57.08,
+      "learning_rate": 9.546557671843944e-05,
+      "loss": 0.0008,
+      "step": 13356
+    },
+    {
+      "epoch": 57.09,
+      "learning_rate": 9.54607259406531e-05,
+      "loss": 0.001,
+      "step": 13360
+    },
+    {
+      "epoch": 57.11,
+      "learning_rate": 9.545587269302521e-05,
+      "loss": 0.0007,
+      "step": 13364
+    },
+    {
+      "epoch": 57.13,
+      "learning_rate": 9.545101697581942e-05,
+      "loss": 0.0003,
+      "step": 13368
+    },
+    {
+      "epoch": 57.15,
+      "learning_rate": 9.544615878929955e-05,
+      "loss": 0.0014,
+      "step": 13372
+    },
+    {
+      "epoch": 57.16,
+      "learning_rate": 9.544129813372953e-05,
+      "loss": 0.0009,
+      "step": 13376
+    },
+    {
+      "epoch": 57.18,
+      "learning_rate": 9.543643500937345e-05,
+      "loss": 0.0004,
+      "step": 13380
+    },
+    {
+      "epoch": 57.2,
+      "learning_rate": 9.54315694164955e-05,
+      "loss": 0.0001,
+      "step": 13384
+    },
+    {
+      "epoch": 57.21,
+      "learning_rate": 9.542670135536005e-05,
+      "loss": 0.0054,
+      "step": 13388
+    },
+    {
+      "epoch": 57.23,
+      "learning_rate": 9.542183082623155e-05,
+      "loss": 0.0005,
+      "step": 13392
+    },
+    {
+      "epoch": 57.25,
+      "learning_rate": 9.541695782937463e-05,
+      "loss": 0.0027,
+      "step": 13396
+    },
+    {
+      "epoch": 57.26,
+      "learning_rate": 9.541208236505403e-05,
+      "loss": 0.0008,
+      "step": 13400
+    },
+    {
+      "epoch": 57.28,
+      "learning_rate": 9.540720443353464e-05,
+      "loss": 0.001,
+      "step": 13404
+    },
+    {
+      "epoch": 57.3,
+      "learning_rate": 9.540232403508146e-05,
+      "loss": 0.0014,
+      "step": 13408
+    },
+    {
+      "epoch": 57.32,
+      "learning_rate": 9.539744116995964e-05,
+      "loss": 0.0016,
+      "step": 13412
+    },
+    {
+      "epoch": 57.33,
+      "learning_rate": 9.539255583843445e-05,
+      "loss": 0.0004,
+      "step": 13416
+    },
+    {
+      "epoch": 57.35,
+      "learning_rate": 9.538766804077134e-05,
+      "loss": 0.0004,
+      "step": 13420
+    },
+    {
+      "epoch": 57.37,
+      "learning_rate": 9.538277777723583e-05,
+      "loss": 0.0001,
+      "step": 13424
+    },
+    {
+      "epoch": 57.38,
+      "learning_rate": 9.53778850480936e-05,
+      "loss": 0.0002,
+      "step": 13428
+    },
+    {
+      "epoch": 57.4,
+      "learning_rate": 9.537298985361049e-05,
+      "loss": 0.0002,
+      "step": 13432
+    },
+    {
+      "epoch": 57.42,
+      "learning_rate": 9.536809219405245e-05,
+      "loss": 0.0015,
+      "step": 13436
+    },
+    {
+      "epoch": 57.44,
+      "learning_rate": 9.536319206968556e-05,
+      "loss": 0.0014,
+      "step": 13440
+    },
+    {
+      "epoch": 57.45,
+      "learning_rate": 9.535828948077602e-05,
+      "loss": 0.0006,
+      "step": 13444
+    },
+    {
+      "epoch": 57.47,
+      "learning_rate": 9.53533844275902e-05,
+      "loss": 0.0005,
+      "step": 13448
+    },
+    {
+      "epoch": 57.49,
+      "learning_rate": 9.534847691039461e-05,
+      "loss": 0.0007,
+      "step": 13452
+    },
+    {
+      "epoch": 57.5,
+      "learning_rate": 9.534356692945583e-05,
+      "loss": 0.0002,
+      "step": 13456
+    },
+    {
+      "epoch": 57.52,
+      "learning_rate": 9.533865448504064e-05,
+      "loss": 0.0002,
+      "step": 13460
+    },
+    {
+      "epoch": 57.54,
+      "learning_rate": 9.533373957741594e-05,
+      "loss": 0.0012,
+      "step": 13464
+    },
+    {
+      "epoch": 57.56,
+      "learning_rate": 9.532882220684871e-05,
+      "loss": 0.0016,
+      "step": 13468
+    },
+    {
+      "epoch": 57.57,
+      "learning_rate": 9.532390237360613e-05,
+      "loss": 0.0004,
+      "step": 13472
+    },
+    {
+      "epoch": 57.59,
+      "learning_rate": 9.531898007795552e-05,
+      "loss": 0.0001,
+      "step": 13476
+    },
+    {
+      "epoch": 57.61,
+      "learning_rate": 9.531405532016424e-05,
+      "loss": 0.0002,
+      "step": 13480
+    },
+    {
+      "epoch": 57.62,
+      "learning_rate": 9.53091281004999e-05,
+      "loss": 0.0002,
+      "step": 13484
+    },
+    {
+      "epoch": 57.64,
+      "learning_rate": 9.530419841923018e-05,
+      "loss": 0.0002,
+      "step": 13488
+    },
+    {
+      "epoch": 57.66,
+      "learning_rate": 9.52992662766229e-05,
+      "loss": 0.001,
+      "step": 13492
+    },
+    {
+      "epoch": 57.68,
+      "learning_rate": 9.529433167294601e-05,
+      "loss": 0.0024,
+      "step": 13496
+    },
+    {
+      "epoch": 57.69,
+      "learning_rate": 9.528939460846764e-05,
+      "loss": 0.0004,
+      "step": 13500
+    },
+    {
+      "epoch": 57.71,
+      "learning_rate": 9.528445508345596e-05,
+      "loss": 0.0003,
+      "step": 13504
+    },
+    {
+      "epoch": 57.73,
+      "learning_rate": 9.527951309817936e-05,
+      "loss": 0.0004,
+      "step": 13508
+    },
+    {
+      "epoch": 57.74,
+      "learning_rate": 9.527456865290633e-05,
+      "loss": 0.0018,
+      "step": 13512
+    },
+    {
+      "epoch": 57.76,
+      "learning_rate": 9.526962174790552e-05,
+      "loss": 0.0003,
+      "step": 13516
+    },
+    {
+      "epoch": 57.78,
+      "learning_rate": 9.526467238344566e-05,
+      "loss": 0.0001,
+      "step": 13520
+    },
+    {
+      "epoch": 57.79,
+      "learning_rate": 9.525972055979566e-05,
+      "loss": 0.0003,
+      "step": 13524
+    },
+    {
+      "epoch": 57.81,
+      "learning_rate": 9.525476627722454e-05,
+      "loss": 0.0001,
+      "step": 13528
+    },
+    {
+      "epoch": 57.83,
+      "learning_rate": 9.524980953600146e-05,
+      "loss": 0.0003,
+      "step": 13532
+    },
+    {
+      "epoch": 57.85,
+      "learning_rate": 9.524485033639574e-05,
+      "loss": 0.0043,
+      "step": 13536
+    },
+    {
+      "epoch": 57.86,
+      "learning_rate": 9.523988867867676e-05,
+      "loss": 0.0001,
+      "step": 13540
+    },
+    {
+      "epoch": 57.88,
+      "learning_rate": 9.523492456311414e-05,
+      "loss": 0.0002,
+      "step": 13544
+    },
+    {
+      "epoch": 57.9,
+      "learning_rate": 9.522995798997754e-05,
+      "loss": 0.0003,
+      "step": 13548
+    },
+    {
+      "epoch": 57.91,
+      "learning_rate": 9.52249889595368e-05,
+      "loss": 0.0023,
+      "step": 13552
+    },
+    {
+      "epoch": 57.93,
+      "learning_rate": 9.522001747206186e-05,
+      "loss": 0.0001,
+      "step": 13556
+    },
+    {
+      "epoch": 57.95,
+      "learning_rate": 9.521504352782287e-05,
+      "loss": 0.0009,
+      "step": 13560
+    },
+    {
+      "epoch": 57.97,
+      "learning_rate": 9.521006712709001e-05,
+      "loss": 0.0022,
+      "step": 13564
+    },
+    {
+      "epoch": 57.98,
+      "learning_rate": 9.520508827013366e-05,
+      "loss": 0.0015,
+      "step": 13568
+    },
+    {
+      "epoch": 58.0,
+      "learning_rate": 9.520010695722432e-05,
+      "loss": 0.0005,
+      "step": 13572
+    },
+    {
+      "epoch": 58.02,
+      "learning_rate": 9.519512318863263e-05,
+      "loss": 0.0001,
+      "step": 13576
+    },
+    {
+      "epoch": 58.03,
+      "learning_rate": 9.519013696462934e-05,
+      "loss": 0.0006,
+      "step": 13580
+    },
+    {
+      "epoch": 58.05,
+      "learning_rate": 9.518514828548535e-05,
+      "loss": 0.0002,
+      "step": 13584
+    },
+    {
+      "epoch": 58.07,
+      "learning_rate": 9.518015715147168e-05,
+      "loss": 0.0044,
+      "step": 13588
+    },
+    {
+      "epoch": 58.09,
+      "learning_rate": 9.517516356285951e-05,
+      "loss": 0.0003,
+      "step": 13592
+    },
+    {
+      "epoch": 58.1,
+      "learning_rate": 9.517016751992013e-05,
+      "loss": 0.0006,
+      "step": 13596
+    },
+    {
+      "epoch": 58.12,
+      "learning_rate": 9.516516902292498e-05,
+      "loss": 0.0052,
+      "step": 13600
+    },
+    {
+      "epoch": 58.14,
+      "learning_rate": 9.51601680721456e-05,
+      "loss": 0.0003,
+      "step": 13604
+    },
+    {
+      "epoch": 58.15,
+      "learning_rate": 9.515516466785373e-05,
+      "loss": 0.0003,
+      "step": 13608
+    },
+    {
+      "epoch": 58.17,
+      "learning_rate": 9.515015881032116e-05,
+      "loss": 0.0005,
+      "step": 13612
+    },
+    {
+      "epoch": 58.19,
+      "learning_rate": 9.514515049981987e-05,
+      "loss": 0.0002,
+      "step": 13616
+    },
+    {
+      "epoch": 58.21,
+      "learning_rate": 9.514013973662195e-05,
+      "loss": 0.0002,
+      "step": 13620
+    },
+    {
+      "epoch": 58.22,
+      "learning_rate": 9.513512652099963e-05,
+      "loss": 0.0006,
+      "step": 13624
+    },
+    {
+      "epoch": 58.24,
+      "learning_rate": 9.51301108532253e-05,
+      "loss": 0.0015,
+      "step": 13628
+    },
+    {
+      "epoch": 58.26,
+      "learning_rate": 9.512509273357141e-05,
+      "loss": 0.0006,
+      "step": 13632
+    },
+    {
+      "epoch": 58.27,
+      "learning_rate": 9.512007216231063e-05,
+      "loss": 0.0003,
+      "step": 13636
+    },
+    {
+      "epoch": 58.29,
+      "learning_rate": 9.511504913971572e-05,
+      "loss": 0.0006,
+      "step": 13640
+    },
+    {
+      "epoch": 58.31,
+      "learning_rate": 9.511002366605954e-05,
+      "loss": 0.0003,
+      "step": 13644
+    },
+    {
+      "epoch": 58.32,
+      "learning_rate": 9.510499574161517e-05,
+      "loss": 0.0002,
+      "step": 13648
+    },
+    {
+      "epoch": 58.34,
+      "learning_rate": 9.509996536665573e-05,
+      "loss": 0.0019,
+      "step": 13652
+    },
+    {
+      "epoch": 58.36,
+      "learning_rate": 9.509493254145455e-05,
+      "loss": 0.0013,
+      "step": 13656
+    },
+    {
+      "epoch": 58.38,
+      "learning_rate": 9.508989726628503e-05,
+      "loss": 0.0002,
+      "step": 13660
+    },
+    {
+      "epoch": 58.39,
+      "learning_rate": 9.508485954142076e-05,
+      "loss": 0.0001,
+      "step": 13664
+    },
+    {
+      "epoch": 58.41,
+      "learning_rate": 9.507981936713541e-05,
+      "loss": 0.0002,
+      "step": 13668
+    },
+    {
+      "epoch": 58.43,
+      "learning_rate": 9.507477674370282e-05,
+      "loss": 0.0004,
+      "step": 13672
+    },
+    {
+      "epoch": 58.44,
+      "learning_rate": 9.506973167139694e-05,
+      "loss": 0.0002,
+      "step": 13676
+    },
+    {
+      "epoch": 58.46,
+      "learning_rate": 9.50646841504919e-05,
+      "loss": 0.0006,
+      "step": 13680
+    },
+    {
+      "epoch": 58.48,
+      "learning_rate": 9.505963418126188e-05,
+      "loss": 0.0001,
+      "step": 13684
+    },
+    {
+      "epoch": 58.5,
+      "learning_rate": 9.505458176398128e-05,
+      "loss": 0.0004,
+      "step": 13688
+    },
+    {
+      "epoch": 58.51,
+      "learning_rate": 9.504952689892456e-05,
+      "loss": 0.0009,
+      "step": 13692
+    },
+    {
+      "epoch": 58.53,
+      "learning_rate": 9.504446958636635e-05,
+      "loss": 0.0003,
+      "step": 13696
+    },
+    {
+      "epoch": 58.55,
+      "learning_rate": 9.503940982658145e-05,
+      "loss": 0.0006,
+      "step": 13700
+    },
+    {
+      "epoch": 58.56,
+      "learning_rate": 9.50343476198447e-05,
+      "loss": 0.0001,
+      "step": 13704
+    },
+    {
+      "epoch": 58.58,
+      "learning_rate": 9.502928296643117e-05,
+      "loss": 0.0003,
+      "step": 13708
+    },
+    {
+      "epoch": 58.6,
+      "learning_rate": 9.502421586661598e-05,
+      "loss": 0.0004,
+      "step": 13712
+    },
+    {
+      "epoch": 58.62,
+      "learning_rate": 9.501914632067446e-05,
+      "loss": 0.0009,
+      "step": 13716
+    },
+    {
+      "epoch": 58.63,
+      "learning_rate": 9.501407432888199e-05,
+      "loss": 0.0002,
+      "step": 13720
+    },
+    {
+      "epoch": 58.65,
+      "learning_rate": 9.500899989151416e-05,
+      "loss": 0.0003,
+      "step": 13724
+    },
+    {
+      "epoch": 58.67,
+      "learning_rate": 9.500392300884665e-05,
+      "loss": 0.0002,
+      "step": 13728
+    },
+    {
+      "epoch": 58.68,
+      "learning_rate": 9.499884368115528e-05,
+      "loss": 0.0002,
+      "step": 13732
+    },
+    {
+      "epoch": 58.7,
+      "learning_rate": 9.4993761908716e-05,
+      "loss": 0.0002,
+      "step": 13736
+    },
+    {
+      "epoch": 58.72,
+      "learning_rate": 9.498867769180491e-05,
+      "loss": 0.0001,
+      "step": 13740
+    },
+    {
+      "epoch": 58.74,
+      "learning_rate": 9.498359103069823e-05,
+      "loss": 0.0001,
+      "step": 13744
+    },
+    {
+      "epoch": 58.75,
+      "learning_rate": 9.49785019256723e-05,
+      "loss": 0.0005,
+      "step": 13748
+    },
+    {
+      "epoch": 58.77,
+      "learning_rate": 9.497341037700363e-05,
+      "loss": 0.0006,
+      "step": 13752
+    },
+    {
+      "epoch": 58.79,
+      "learning_rate": 9.496831638496883e-05,
+      "loss": 0.0017,
+      "step": 13756
+    },
+    {
+      "epoch": 58.8,
+      "learning_rate": 9.496321994984463e-05,
+      "loss": 0.0002,
+      "step": 13760
+    },
+    {
+      "epoch": 58.82,
+      "learning_rate": 9.495812107190795e-05,
+      "loss": 0.0007,
+      "step": 13764
+    },
+    {
+      "epoch": 58.84,
+      "learning_rate": 9.495301975143578e-05,
+      "loss": 0.0004,
+      "step": 13768
+    },
+    {
+      "epoch": 58.85,
+      "learning_rate": 9.494791598870528e-05,
+      "loss": 0.0016,
+      "step": 13772
+    },
+    {
+      "epoch": 58.87,
+      "learning_rate": 9.494280978399374e-05,
+      "loss": 0.0001,
+      "step": 13776
+    },
+    {
+      "epoch": 58.89,
+      "learning_rate": 9.493770113757855e-05,
+      "loss": 0.0004,
+      "step": 13780
+    },
+    {
+      "epoch": 58.91,
+      "learning_rate": 9.493259004973732e-05,
+      "loss": 0.0002,
+      "step": 13784
+    },
+    {
+      "epoch": 58.92,
+      "learning_rate": 9.492747652074766e-05,
+      "loss": 0.0003,
+      "step": 13788
+    },
+    {
+      "epoch": 58.94,
+      "learning_rate": 9.492236055088741e-05,
+      "loss": 0.0019,
+      "step": 13792
+    },
+    {
+      "epoch": 58.96,
+      "learning_rate": 9.491724214043453e-05,
+      "loss": 0.0003,
+      "step": 13796
+    },
+    {
+      "epoch": 58.97,
+      "learning_rate": 9.491212128966709e-05,
+      "loss": 0.0005,
+      "step": 13800
+    },
+    {
+      "epoch": 58.99,
+      "learning_rate": 9.49069979988633e-05,
+      "loss": 0.0002,
+      "step": 13804
+    },
+    {
+      "epoch": 59.01,
+      "learning_rate": 9.490187226830148e-05,
+      "loss": 0.0001,
+      "step": 13808
+    },
+    {
+      "epoch": 59.03,
+      "learning_rate": 9.489674409826015e-05,
+      "loss": 0.0018,
+      "step": 13812
+    },
+    {
+      "epoch": 59.04,
+      "learning_rate": 9.48916134890179e-05,
+      "loss": 0.0022,
+      "step": 13816
+    },
+    {
+      "epoch": 59.06,
+      "learning_rate": 9.488648044085345e-05,
+      "loss": 0.0004,
+      "step": 13820
+    },
+    {
+      "epoch": 59.08,
+      "learning_rate": 9.488134495404572e-05,
+      "loss": 0.0002,
+      "step": 13824
+    },
+    {
+      "epoch": 59.09,
+      "learning_rate": 9.487620702887368e-05,
+      "loss": 0.0005,
+      "step": 13828
+    },
+    {
+      "epoch": 59.11,
+      "learning_rate": 9.487106666561647e-05,
+      "loss": 0.0006,
+      "step": 13832
+    },
+    {
+      "epoch": 59.13,
+      "learning_rate": 9.486592386455338e-05,
+      "loss": 0.0002,
+      "step": 13836
+    },
+    {
+      "epoch": 59.15,
+      "learning_rate": 9.486077862596379e-05,
+      "loss": 0.0003,
+      "step": 13840
+    },
+    {
+      "epoch": 59.16,
+      "learning_rate": 9.485563095012727e-05,
+      "loss": 0.0006,
+      "step": 13844
+    },
+    {
+      "epoch": 59.18,
+      "learning_rate": 9.485048083732346e-05,
+      "loss": 0.0017,
+      "step": 13848
+    },
+    {
+      "epoch": 59.2,
+      "learning_rate": 9.484532828783217e-05,
+      "loss": 0.0001,
+      "step": 13852
+    },
+    {
+      "epoch": 59.21,
+      "learning_rate": 9.484017330193333e-05,
+      "loss": 0.0005,
+      "step": 13856
+    },
+    {
+      "epoch": 59.23,
+      "learning_rate": 9.483501587990703e-05,
+      "loss": 0.0004,
+      "step": 13860
+    },
+    {
+      "epoch": 59.25,
+      "learning_rate": 9.482985602203343e-05,
+      "loss": 0.0009,
+      "step": 13864
+    },
+    {
+      "epoch": 59.26,
+      "learning_rate": 9.482469372859288e-05,
+      "loss": 0.0009,
+      "step": 13868
+    },
+    {
+      "epoch": 59.28,
+      "learning_rate": 9.481952899986586e-05,
+      "loss": 0.0015,
+      "step": 13872
+    },
+    {
+      "epoch": 59.3,
+      "learning_rate": 9.481436183613293e-05,
+      "loss": 0.0009,
+      "step": 13876
+    },
+    {
+      "epoch": 59.32,
+      "learning_rate": 9.480919223767484e-05,
+      "loss": 0.0002,
+      "step": 13880
+    },
+    {
+      "epoch": 59.33,
+      "learning_rate": 9.480402020477243e-05,
+      "loss": 0.0015,
+      "step": 13884
+    },
+    {
+      "epoch": 59.35,
+      "learning_rate": 9.479884573770674e-05,
+      "loss": 0.0029,
+      "step": 13888
+    },
+    {
+      "epoch": 59.37,
+      "learning_rate": 9.479366883675884e-05,
+      "loss": 0.0001,
+      "step": 13892
+    },
+    {
+      "epoch": 59.38,
+      "learning_rate": 9.478848950221e-05,
+      "loss": 0.0019,
+      "step": 13896
+    },
+    {
+      "epoch": 59.4,
+      "learning_rate": 9.478330773434163e-05,
+      "loss": 0.0006,
+      "step": 13900
+    },
+    {
+      "epoch": 59.42,
+      "learning_rate": 9.477812353343522e-05,
+      "loss": 0.0002,
+      "step": 13904
+    },
+    {
+      "epoch": 59.44,
+      "learning_rate": 9.477293689977245e-05,
+      "loss": 0.0003,
+      "step": 13908
+    },
+    {
+      "epoch": 59.45,
+      "learning_rate": 9.476774783363511e-05,
+      "loss": 0.0022,
+      "step": 13912
+    },
+    {
+      "epoch": 59.47,
+      "learning_rate": 9.476255633530508e-05,
+      "loss": 0.0003,
+      "step": 13916
+    },
+    {
+      "epoch": 59.49,
+      "learning_rate": 9.475736240506444e-05,
+      "loss": 0.0001,
+      "step": 13920
+    },
+    {
+      "epoch": 59.5,
+      "learning_rate": 9.475216604319536e-05,
+      "loss": 0.0004,
+      "step": 13924
+    },
+    {
+      "epoch": 59.52,
+      "learning_rate": 9.474696724998017e-05,
+      "loss": 0.0004,
+      "step": 13928
+    },
+    {
+      "epoch": 59.54,
+      "learning_rate": 9.47417660257013e-05,
+      "loss": 0.0003,
+      "step": 13932
+    },
+    {
+      "epoch": 59.56,
+      "learning_rate": 9.473656237064133e-05,
+      "loss": 0.001,
+      "step": 13936
+    },
+    {
+      "epoch": 59.57,
+      "learning_rate": 9.473135628508296e-05,
+      "loss": 0.0003,
+      "step": 13940
+    },
+    {
+      "epoch": 59.59,
+      "learning_rate": 9.472614776930906e-05,
+      "loss": 0.0003,
+      "step": 13944
+    },
+    {
+      "epoch": 59.61,
+      "learning_rate": 9.47209368236026e-05,
+      "loss": 0.0006,
+      "step": 13948
+    },
+    {
+      "epoch": 59.62,
+      "learning_rate": 9.471572344824664e-05,
+      "loss": 0.0011,
+      "step": 13952
+    },
+    {
+      "epoch": 59.64,
+      "learning_rate": 9.471050764352447e-05,
+      "loss": 0.0007,
+      "step": 13956
+    },
+    {
+      "epoch": 59.66,
+      "learning_rate": 9.470528940971944e-05,
+      "loss": 0.0001,
+      "step": 13960
+    },
+    {
+      "epoch": 59.68,
+      "learning_rate": 9.470006874711505e-05,
+      "loss": 0.0019,
+      "step": 13964
+    },
+    {
+      "epoch": 59.69,
+      "learning_rate": 9.469484565599495e-05,
+      "loss": 0.0002,
+      "step": 13968
+    },
+    {
+      "epoch": 59.71,
+      "learning_rate": 9.468962013664288e-05,
+      "loss": 0.0001,
+      "step": 13972
+    },
+    {
+      "epoch": 59.73,
+      "learning_rate": 9.468439218934276e-05,
+      "loss": 0.0005,
+      "step": 13976
+    },
+    {
+      "epoch": 59.74,
+      "learning_rate": 9.467916181437862e-05,
+      "loss": 0.0017,
+      "step": 13980
+    },
+    {
+      "epoch": 59.76,
+      "learning_rate": 9.46739290120346e-05,
+      "loss": 0.0001,
+      "step": 13984
+    },
+    {
+      "epoch": 59.78,
+      "learning_rate": 9.466869378259502e-05,
+      "loss": 0.0001,
+      "step": 13988
+    },
+    {
+      "epoch": 59.79,
+      "learning_rate": 9.466345612634428e-05,
+      "loss": 0.0012,
+      "step": 13992
+    },
+    {
+      "epoch": 59.81,
+      "learning_rate": 9.465821604356696e-05,
+      "loss": 0.0004,
+      "step": 13996
+    },
+    {
+      "epoch": 59.83,
+      "learning_rate": 9.465297353454774e-05,
+      "loss": 0.0006,
+      "step": 14000
+    },
+    {
+      "epoch": 59.83,
+      "eval_exact_match": 0.524948024948025,
+      "eval_loss": 0.8439415097236633,
+      "eval_runtime": 156.3849,
+      "eval_samples_per_second": 6.151,
+      "step": 14000
+    },
+    {
+      "epoch": 59.85,
+      "learning_rate": 9.464772859957143e-05,
+      "loss": 0.0005,
+      "step": 14004
+    },
+    {
+      "epoch": 59.86,
+      "learning_rate": 9.4642481238923e-05,
+      "loss": 0.0002,
+      "step": 14008
+    },
+    {
+      "epoch": 59.88,
+      "learning_rate": 9.463723145288752e-05,
+      "loss": 0.0002,
+      "step": 14012
+    },
+    {
+      "epoch": 59.9,
+      "learning_rate": 9.463197924175022e-05,
+      "loss": 0.0005,
+      "step": 14016
+    },
+    {
+      "epoch": 59.91,
+      "learning_rate": 9.462672460579645e-05,
+      "loss": 0.0009,
+      "step": 14020
+    },
+    {
+      "epoch": 59.93,
+      "learning_rate": 9.462146754531167e-05,
+      "loss": 0.0002,
+      "step": 14024
+    },
+    {
+      "epoch": 59.95,
+      "learning_rate": 9.461620806058152e-05,
+      "loss": 0.001,
+      "step": 14028
+    },
+    {
+      "epoch": 59.97,
+      "learning_rate": 9.46109461518917e-05,
+      "loss": 0.0003,
+      "step": 14032
+    },
+    {
+      "epoch": 59.98,
+      "learning_rate": 9.460568181952813e-05,
+      "loss": 0.0004,
+      "step": 14036
+    },
+    {
+      "epoch": 60.0,
+      "learning_rate": 9.460041506377679e-05,
+      "loss": 0.0001,
+      "step": 14040
+    },
+    {
+      "epoch": 60.02,
+      "learning_rate": 9.459514588492382e-05,
+      "loss": 0.0002,
+      "step": 14044
+    },
+    {
+      "epoch": 60.03,
+      "learning_rate": 9.45898742832555e-05,
+      "loss": 0.0001,
+      "step": 14048
+    },
+    {
+      "epoch": 60.05,
+      "learning_rate": 9.458460025905824e-05,
+      "loss": 0.0016,
+      "step": 14052
+    },
+    {
+      "epoch": 60.07,
+      "learning_rate": 9.457932381261855e-05,
+      "loss": 0.0012,
+      "step": 14056
+    },
+    {
+      "epoch": 60.09,
+      "learning_rate": 9.45740449442231e-05,
+      "loss": 0.0002,
+      "step": 14060
+    },
+    {
+      "epoch": 60.1,
+      "learning_rate": 9.456876365415871e-05,
+      "loss": 0.0013,
+      "step": 14064
+    },
+    {
+      "epoch": 60.12,
+      "learning_rate": 9.456347994271229e-05,
+      "loss": 0.0001,
+      "step": 14068
+    },
+    {
+      "epoch": 60.14,
+      "learning_rate": 9.455819381017089e-05,
+      "loss": 0.0002,
+      "step": 14072
+    },
+    {
+      "epoch": 60.15,
+      "learning_rate": 9.45529052568217e-05,
+      "loss": 0.001,
+      "step": 14076
+    },
+    {
+      "epoch": 60.17,
+      "learning_rate": 9.454761428295206e-05,
+      "loss": 0.0001,
+      "step": 14080
+    },
+    {
+      "epoch": 60.19,
+      "learning_rate": 9.454232088884942e-05,
+      "loss": 0.0002,
+      "step": 14084
+    },
+    {
+      "epoch": 60.21,
+      "learning_rate": 9.453702507480136e-05,
+      "loss": 0.0003,
+      "step": 14088
+    },
+    {
+      "epoch": 60.22,
+      "learning_rate": 9.453172684109559e-05,
+      "loss": 0.0002,
+      "step": 14092
+    },
+    {
+      "epoch": 60.24,
+      "learning_rate": 9.452642618801997e-05,
+      "loss": 0.0007,
+      "step": 14096
+    },
+    {
+      "epoch": 60.26,
+      "learning_rate": 9.452112311586247e-05,
+      "loss": 0.001,
+      "step": 14100
+    },
+    {
+      "epoch": 60.27,
+      "learning_rate": 9.451581762491124e-05,
+      "loss": 0.0013,
+      "step": 14104
+    },
+    {
+      "epoch": 60.29,
+      "learning_rate": 9.451050971545447e-05,
+      "loss": 0.0014,
+      "step": 14108
+    },
+    {
+      "epoch": 60.31,
+      "learning_rate": 9.450519938778056e-05,
+      "loss": 0.0011,
+      "step": 14112
+    },
+    {
+      "epoch": 60.32,
+      "learning_rate": 9.449988664217799e-05,
+      "loss": 0.0006,
+      "step": 14116
+    },
+    {
+      "epoch": 60.34,
+      "learning_rate": 9.449457147893543e-05,
+      "loss": 0.001,
+      "step": 14120
+    },
+    {
+      "epoch": 60.36,
+      "learning_rate": 9.448925389834165e-05,
+      "loss": 0.0007,
+      "step": 14124
+    },
+    {
+      "epoch": 60.38,
+      "learning_rate": 9.448393390068553e-05,
+      "loss": 0.001,
+      "step": 14128
+    },
+    {
+      "epoch": 60.39,
+      "learning_rate": 9.447861148625609e-05,
+      "loss": 0.0005,
+      "step": 14132
+    },
+    {
+      "epoch": 60.41,
+      "learning_rate": 9.447328665534253e-05,
+      "loss": 0.0002,
+      "step": 14136
+    },
+    {
+      "epoch": 60.43,
+      "learning_rate": 9.446795940823411e-05,
+      "loss": 0.002,
+      "step": 14140
+    },
+    {
+      "epoch": 60.44,
+      "learning_rate": 9.446262974522026e-05,
+      "loss": 0.0011,
+      "step": 14144
+    },
+    {
+      "epoch": 60.46,
+      "learning_rate": 9.445729766659055e-05,
+      "loss": 0.0004,
+      "step": 14148
+    },
+    {
+      "epoch": 60.48,
+      "learning_rate": 9.445196317263467e-05,
+      "loss": 0.0003,
+      "step": 14152
+    },
+    {
+      "epoch": 60.5,
+      "learning_rate": 9.44466262636424e-05,
+      "loss": 0.0003,
+      "step": 14156
+    },
+    {
+      "epoch": 60.51,
+      "learning_rate": 9.444128693990375e-05,
+      "loss": 0.0001,
+      "step": 14160
+    },
+    {
+      "epoch": 60.53,
+      "learning_rate": 9.443594520170876e-05,
+      "loss": 0.0003,
+      "step": 14164
+    },
+    {
+      "epoch": 60.55,
+      "learning_rate": 9.443060104934762e-05,
+      "loss": 0.0013,
+      "step": 14168
+    },
+    {
+      "epoch": 60.56,
+      "learning_rate": 9.442525448311072e-05,
+      "loss": 0.0007,
+      "step": 14172
+    },
+    {
+      "epoch": 60.58,
+      "learning_rate": 9.441990550328854e-05,
+      "loss": 0.0001,
+      "step": 14176
+    },
+    {
+      "epoch": 60.6,
+      "learning_rate": 9.441455411017164e-05,
+      "loss": 0.0003,
+      "step": 14180
+    },
+    {
+      "epoch": 60.62,
+      "learning_rate": 9.440920030405078e-05,
+      "loss": 0.0002,
+      "step": 14184
+    },
+    {
+      "epoch": 60.63,
+      "learning_rate": 9.440384408521684e-05,
+      "loss": 0.0001,
+      "step": 14188
+    },
+    {
+      "epoch": 60.65,
+      "learning_rate": 9.439848545396079e-05,
+      "loss": 0.0005,
+      "step": 14192
+    },
+    {
+      "epoch": 60.67,
+      "learning_rate": 9.439312441057379e-05,
+      "loss": 0.0001,
+      "step": 14196
+    },
+    {
+      "epoch": 60.68,
+      "learning_rate": 9.438776095534708e-05,
+      "loss": 0.0004,
+      "step": 14200
+    },
+    {
+      "epoch": 60.7,
+      "learning_rate": 9.438239508857207e-05,
+      "loss": 0.0016,
+      "step": 14204
+    },
+    {
+      "epoch": 60.72,
+      "learning_rate": 9.437702681054025e-05,
+      "loss": 0.0005,
+      "step": 14208
+    },
+    {
+      "epoch": 60.74,
+      "learning_rate": 9.437165612154331e-05,
+      "loss": 0.0004,
+      "step": 14212
+    },
+    {
+      "epoch": 60.75,
+      "learning_rate": 9.4366283021873e-05,
+      "loss": 0.001,
+      "step": 14216
+    },
+    {
+      "epoch": 60.77,
+      "learning_rate": 9.436090751182128e-05,
+      "loss": 0.0016,
+      "step": 14220
+    },
+    {
+      "epoch": 60.79,
+      "learning_rate": 9.435552959168016e-05,
+      "loss": 0.0014,
+      "step": 14224
+    },
+    {
+      "epoch": 60.8,
+      "learning_rate": 9.435014926174183e-05,
+      "loss": 0.0003,
+      "step": 14228
+    },
+    {
+      "epoch": 60.82,
+      "learning_rate": 9.43447665222986e-05,
+      "loss": 0.0003,
+      "step": 14232
+    },
+    {
+      "epoch": 60.84,
+      "learning_rate": 9.43393813736429e-05,
+      "loss": 0.0004,
+      "step": 14236
+    },
+    {
+      "epoch": 60.85,
+      "learning_rate": 9.433399381606733e-05,
+      "loss": 0.0003,
+      "step": 14240
+    },
+    {
+      "epoch": 60.87,
+      "learning_rate": 9.432860384986456e-05,
+      "loss": 0.0008,
+      "step": 14244
+    },
+    {
+      "epoch": 60.89,
+      "learning_rate": 9.432321147532743e-05,
+      "loss": 0.0015,
+      "step": 14248
+    },
+    {
+      "epoch": 60.91,
+      "learning_rate": 9.431781669274891e-05,
+      "loss": 0.0003,
+      "step": 14252
+    },
+    {
+      "epoch": 60.92,
+      "learning_rate": 9.431241950242207e-05,
+      "loss": 0.0002,
+      "step": 14256
+    },
+    {
+      "epoch": 60.94,
+      "learning_rate": 9.430701990464017e-05,
+      "loss": 0.0003,
+      "step": 14260
+    },
+    {
+      "epoch": 60.96,
+      "learning_rate": 9.430161789969654e-05,
+      "loss": 0.0011,
+      "step": 14264
+    },
+    {
+      "epoch": 60.97,
+      "learning_rate": 9.429621348788468e-05,
+      "loss": 0.0001,
+      "step": 14268
+    },
+    {
+      "epoch": 60.99,
+      "learning_rate": 9.429080666949821e-05,
+      "loss": 0.0009,
+      "step": 14272
+    },
+    {
+      "epoch": 61.01,
+      "learning_rate": 9.428539744483085e-05,
+      "loss": 0.0005,
+      "step": 14276
+    },
+    {
+      "epoch": 61.03,
+      "learning_rate": 9.427998581417652e-05,
+      "loss": 0.001,
+      "step": 14280
+    },
+    {
+      "epoch": 61.04,
+      "learning_rate": 9.42745717778292e-05,
+      "loss": 0.0003,
+      "step": 14284
+    },
+    {
+      "epoch": 61.06,
+      "learning_rate": 9.426915533608304e-05,
+      "loss": 0.0011,
+      "step": 14288
+    },
+    {
+      "epoch": 61.08,
+      "learning_rate": 9.42637364892323e-05,
+      "loss": 0.0006,
+      "step": 14292
+    },
+    {
+      "epoch": 61.09,
+      "learning_rate": 9.425831523757139e-05,
+      "loss": 0.0002,
+      "step": 14296
+    },
+    {
+      "epoch": 61.11,
+      "learning_rate": 9.425289158139486e-05,
+      "loss": 0.0001,
+      "step": 14300
+    },
+    {
+      "epoch": 61.13,
+      "learning_rate": 9.424746552099733e-05,
+      "loss": 0.0002,
+      "step": 14304
+    },
+    {
+      "epoch": 61.15,
+      "learning_rate": 9.424203705667363e-05,
+      "loss": 0.0003,
+      "step": 14308
+    },
+    {
+      "epoch": 61.16,
+      "learning_rate": 9.423660618871866e-05,
+      "loss": 0.0001,
+      "step": 14312
+    },
+    {
+      "epoch": 61.18,
+      "learning_rate": 9.423117291742749e-05,
+      "loss": 0.0013,
+      "step": 14316
+    },
+    {
+      "epoch": 61.2,
+      "learning_rate": 9.42257372430953e-05,
+      "loss": 0.0006,
+      "step": 14320
+    },
+    {
+      "epoch": 61.21,
+      "learning_rate": 9.42202991660174e-05,
+      "loss": 0.0003,
+      "step": 14324
+    },
+    {
+      "epoch": 61.23,
+      "learning_rate": 9.421485868648925e-05,
+      "loss": 0.0002,
+      "step": 14328
+    },
+    {
+      "epoch": 61.25,
+      "learning_rate": 9.420941580480642e-05,
+      "loss": 0.0009,
+      "step": 14332
+    },
+    {
+      "epoch": 61.26,
+      "learning_rate": 9.42039705212646e-05,
+      "loss": 0.0004,
+      "step": 14336
+    },
+    {
+      "epoch": 61.28,
+      "learning_rate": 9.419852283615964e-05,
+      "loss": 0.0001,
+      "step": 14340
+    },
+    {
+      "epoch": 61.3,
+      "learning_rate": 9.419307274978753e-05,
+      "loss": 0.0004,
+      "step": 14344
+    },
+    {
+      "epoch": 61.32,
+      "learning_rate": 9.418762026244434e-05,
+      "loss": 0.0002,
+      "step": 14348
+    },
+    {
+      "epoch": 61.33,
+      "learning_rate": 9.41821653744263e-05,
+      "loss": 0.0008,
+      "step": 14352
+    },
+    {
+      "epoch": 61.35,
+      "learning_rate": 9.417670808602978e-05,
+      "loss": 0.0034,
+      "step": 14356
+    },
+    {
+      "epoch": 61.37,
+      "learning_rate": 9.417124839755127e-05,
+      "loss": 0.0001,
+      "step": 14360
+    },
+    {
+      "epoch": 61.38,
+      "learning_rate": 9.416578630928738e-05,
+      "loss": 0.0018,
+      "step": 14364
+    },
+    {
+      "epoch": 61.4,
+      "learning_rate": 9.416032182153488e-05,
+      "loss": 0.0002,
+      "step": 14368
+    },
+    {
+      "epoch": 61.42,
+      "learning_rate": 9.415485493459063e-05,
+      "loss": 0.0002,
+      "step": 14372
+    },
+    {
+      "epoch": 61.44,
+      "learning_rate": 9.414938564875164e-05,
+      "loss": 0.001,
+      "step": 14376
+    },
+    {
+      "epoch": 61.45,
+      "learning_rate": 9.414391396431507e-05,
+      "loss": 0.0015,
+      "step": 14380
+    },
+    {
+      "epoch": 61.47,
+      "learning_rate": 9.41384398815782e-05,
+      "loss": 0.0003,
+      "step": 14384
+    },
+    {
+      "epoch": 61.49,
+      "learning_rate": 9.41329634008384e-05,
+      "loss": 0.0021,
+      "step": 14388
+    },
+    {
+      "epoch": 61.5,
+      "learning_rate": 9.412748452239321e-05,
+      "loss": 0.0008,
+      "step": 14392
+    },
+    {
+      "epoch": 61.52,
+      "learning_rate": 9.41220032465403e-05,
+      "loss": 0.0006,
+      "step": 14396
+    },
+    {
+      "epoch": 61.54,
+      "learning_rate": 9.411651957357746e-05,
+      "loss": 0.0003,
+      "step": 14400
+    },
+    {
+      "epoch": 61.56,
+      "learning_rate": 9.411103350380263e-05,
+      "loss": 0.0031,
+      "step": 14404
+    },
+    {
+      "epoch": 61.57,
+      "learning_rate": 9.410554503751383e-05,
+      "loss": 0.0003,
+      "step": 14408
+    },
+    {
+      "epoch": 61.59,
+      "learning_rate": 9.410005417500928e-05,
+      "loss": 0.0004,
+      "step": 14412
+    },
+    {
+      "epoch": 61.61,
+      "learning_rate": 9.409456091658726e-05,
+      "loss": 0.0019,
+      "step": 14416
+    },
+    {
+      "epoch": 61.62,
+      "learning_rate": 9.408906526254624e-05,
+      "loss": 0.0001,
+      "step": 14420
+    },
+    {
+      "epoch": 61.64,
+      "learning_rate": 9.408356721318479e-05,
+      "loss": 0.0002,
+      "step": 14424
+    },
+    {
+      "epoch": 61.66,
+      "learning_rate": 9.40780667688016e-05,
+      "loss": 0.0003,
+      "step": 14428
+    },
+    {
+      "epoch": 61.68,
+      "learning_rate": 9.407256392969552e-05,
+      "loss": 0.0002,
+      "step": 14432
+    },
+    {
+      "epoch": 61.69,
+      "learning_rate": 9.40670586961655e-05,
+      "loss": 0.0013,
+      "step": 14436
+    },
+    {
+      "epoch": 61.71,
+      "learning_rate": 9.406155106851064e-05,
+      "loss": 0.0003,
+      "step": 14440
+    },
+    {
+      "epoch": 61.73,
+      "learning_rate": 9.405604104703019e-05,
+      "loss": 0.0003,
+      "step": 14444
+    },
+    {
+      "epoch": 61.74,
+      "learning_rate": 9.405052863202346e-05,
+      "loss": 0.0003,
+      "step": 14448
+    },
+    {
+      "epoch": 61.76,
+      "learning_rate": 9.404501382378996e-05,
+      "loss": 0.0004,
+      "step": 14452
+    },
+    {
+      "epoch": 61.78,
+      "learning_rate": 9.40394966226293e-05,
+      "loss": 0.0019,
+      "step": 14456
+    },
+    {
+      "epoch": 61.79,
+      "learning_rate": 9.403397702884123e-05,
+      "loss": 0.0004,
+      "step": 14460
+    },
+    {
+      "epoch": 61.81,
+      "learning_rate": 9.40284550427256e-05,
+      "loss": 0.0008,
+      "step": 14464
+    },
+    {
+      "epoch": 61.83,
+      "learning_rate": 9.402293066458246e-05,
+      "loss": 0.0001,
+      "step": 14468
+    },
+    {
+      "epoch": 61.85,
+      "learning_rate": 9.401740389471191e-05,
+      "loss": 0.0005,
+      "step": 14472
+    },
+    {
+      "epoch": 61.86,
+      "learning_rate": 9.401187473341422e-05,
+      "loss": 0.0002,
+      "step": 14476
+    },
+    {
+      "epoch": 61.88,
+      "learning_rate": 9.400634318098979e-05,
+      "loss": 0.0002,
+      "step": 14480
+    },
+    {
+      "epoch": 61.9,
+      "learning_rate": 9.400080923773915e-05,
+      "loss": 0.0005,
+      "step": 14484
+    },
+    {
+      "epoch": 61.91,
+      "learning_rate": 9.399527290396293e-05,
+      "loss": 0.0002,
+      "step": 14488
+    },
+    {
+      "epoch": 61.93,
+      "learning_rate": 9.398973417996195e-05,
+      "loss": 0.0007,
+      "step": 14492
+    },
+    {
+      "epoch": 61.95,
+      "learning_rate": 9.39841930660371e-05,
+      "loss": 0.0002,
+      "step": 14496
+    },
+    {
+      "epoch": 61.97,
+      "learning_rate": 9.397864956248942e-05,
+      "loss": 0.001,
+      "step": 14500
+    },
+    {
+      "epoch": 61.98,
+      "learning_rate": 9.39731036696201e-05,
+      "loss": 0.0002,
+      "step": 14504
+    },
+    {
+      "epoch": 62.0,
+      "learning_rate": 9.396755538773045e-05,
+      "loss": 0.0014,
+      "step": 14508
+    },
+    {
+      "epoch": 62.02,
+      "learning_rate": 9.396200471712188e-05,
+      "loss": 0.0001,
+      "step": 14512
+    },
+    {
+      "epoch": 62.03,
+      "learning_rate": 9.395645165809597e-05,
+      "loss": 0.0004,
+      "step": 14516
+    },
+    {
+      "epoch": 62.05,
+      "learning_rate": 9.395089621095439e-05,
+      "loss": 0.0003,
+      "step": 14520
+    },
+    {
+      "epoch": 62.07,
+      "learning_rate": 9.3945338375999e-05,
+      "loss": 0.0008,
+      "step": 14524
+    },
+    {
+      "epoch": 62.09,
+      "learning_rate": 9.393977815353169e-05,
+      "loss": 0.0001,
+      "step": 14528
+    },
+    {
+      "epoch": 62.1,
+      "learning_rate": 9.393421554385461e-05,
+      "loss": 0.0003,
+      "step": 14532
+    },
+    {
+      "epoch": 62.12,
+      "learning_rate": 9.392865054726996e-05,
+      "loss": 0.0004,
+      "step": 14536
+    },
+    {
+      "epoch": 62.14,
+      "learning_rate": 9.392308316408005e-05,
+      "loss": 0.0001,
+      "step": 14540
+    },
+    {
+      "epoch": 62.15,
+      "learning_rate": 9.391751339458737e-05,
+      "loss": 0.0014,
+      "step": 14544
+    },
+    {
+      "epoch": 62.17,
+      "learning_rate": 9.391194123909451e-05,
+      "loss": 0.0002,
+      "step": 14548
+    },
+    {
+      "epoch": 62.19,
+      "learning_rate": 9.390636669790422e-05,
+      "loss": 0.0001,
+      "step": 14552
+    },
+    {
+      "epoch": 62.21,
+      "learning_rate": 9.390078977131933e-05,
+      "loss": 0.0005,
+      "step": 14556
+    },
+    {
+      "epoch": 62.22,
+      "learning_rate": 9.389521045964287e-05,
+      "loss": 0.0005,
+      "step": 14560
+    },
+    {
+      "epoch": 62.24,
+      "learning_rate": 9.388962876317793e-05,
+      "loss": 0.0003,
+      "step": 14564
+    },
+    {
+      "epoch": 62.26,
+      "learning_rate": 9.388404468222775e-05,
+      "loss": 0.0002,
+      "step": 14568
+    },
+    {
+      "epoch": 62.27,
+      "learning_rate": 9.387845821709573e-05,
+      "loss": 0.0003,
+      "step": 14572
+    },
+    {
+      "epoch": 62.29,
+      "learning_rate": 9.387286936808539e-05,
+      "loss": 0.0001,
+      "step": 14576
+    },
+    {
+      "epoch": 62.31,
+      "learning_rate": 9.386727813550034e-05,
+      "loss": 0.001,
+      "step": 14580
+    },
+    {
+      "epoch": 62.32,
+      "learning_rate": 9.386168451964434e-05,
+      "loss": 0.0002,
+      "step": 14584
+    },
+    {
+      "epoch": 62.34,
+      "learning_rate": 9.38560885208213e-05,
+      "loss": 0.0004,
+      "step": 14588
+    },
+    {
+      "epoch": 62.36,
+      "learning_rate": 9.385049013933526e-05,
+      "loss": 0.0005,
+      "step": 14592
+    },
+    {
+      "epoch": 62.38,
+      "learning_rate": 9.384488937549036e-05,
+      "loss": 0.0004,
+      "step": 14596
+    },
+    {
+      "epoch": 62.39,
+      "learning_rate": 9.383928622959091e-05,
+      "loss": 0.0003,
+      "step": 14600
+    },
+    {
+      "epoch": 62.41,
+      "learning_rate": 9.383368070194127e-05,
+      "loss": 0.0001,
+      "step": 14604
+    },
+    {
+      "epoch": 62.43,
+      "learning_rate": 9.382807279284602e-05,
+      "loss": 0.0005,
+      "step": 14608
+    },
+    {
+      "epoch": 62.44,
+      "learning_rate": 9.382246250260984e-05,
+      "loss": 0.0026,
+      "step": 14612
+    },
+    {
+      "epoch": 62.46,
+      "learning_rate": 9.381684983153751e-05,
+      "loss": 0.0001,
+      "step": 14616
+    },
+    {
+      "epoch": 62.48,
+      "learning_rate": 9.381123477993398e-05,
+      "loss": 0.0002,
+      "step": 14620
+    },
+    {
+      "epoch": 62.5,
+      "learning_rate": 9.380561734810429e-05,
+      "loss": 0.0002,
+      "step": 14624
+    },
+    {
+      "epoch": 62.51,
+      "learning_rate": 9.379999753635365e-05,
+      "loss": 0.0001,
+      "step": 14628
+    },
+    {
+      "epoch": 62.53,
+      "learning_rate": 9.379437534498737e-05,
+      "loss": 0.0002,
+      "step": 14632
+    },
+    {
+      "epoch": 62.55,
+      "learning_rate": 9.37887507743109e-05,
+      "loss": 0.0002,
+      "step": 14636
+    },
+    {
+      "epoch": 62.56,
+      "learning_rate": 9.378312382462981e-05,
+      "loss": 0.0022,
+      "step": 14640
+    },
+    {
+      "epoch": 62.58,
+      "learning_rate": 9.377749449624984e-05,
+      "loss": 0.0001,
+      "step": 14644
+    },
+    {
+      "epoch": 62.6,
+      "learning_rate": 9.377186278947678e-05,
+      "loss": 0.002,
+      "step": 14648
+    },
+    {
+      "epoch": 62.62,
+      "learning_rate": 9.376622870461663e-05,
+      "loss": 0.0005,
+      "step": 14652
+    },
+    {
+      "epoch": 62.63,
+      "learning_rate": 9.376059224197547e-05,
+      "loss": 0.0003,
+      "step": 14656
+    },
+    {
+      "epoch": 62.65,
+      "learning_rate": 9.375495340185952e-05,
+      "loss": 0.0012,
+      "step": 14660
+    },
+    {
+      "epoch": 62.67,
+      "learning_rate": 9.374931218457515e-05,
+      "loss": 0.0001,
+      "step": 14664
+    },
+    {
+      "epoch": 62.68,
+      "learning_rate": 9.374366859042883e-05,
+      "loss": 0.0001,
+      "step": 14668
+    },
+    {
+      "epoch": 62.7,
+      "learning_rate": 9.373802261972717e-05,
+      "loss": 0.0002,
+      "step": 14672
+    },
+    {
+      "epoch": 62.72,
+      "learning_rate": 9.373237427277692e-05,
+      "loss": 0.0001,
+      "step": 14676
+    },
+    {
+      "epoch": 62.74,
+      "learning_rate": 9.372672354988494e-05,
+      "loss": 0.0003,
+      "step": 14680
+    },
+    {
+      "epoch": 62.75,
+      "learning_rate": 9.372107045135823e-05,
+      "loss": 0.0001,
+      "step": 14684
+    },
+    {
+      "epoch": 62.77,
+      "learning_rate": 9.371541497750393e-05,
+      "loss": 0.0016,
+      "step": 14688
+    },
+    {
+      "epoch": 62.79,
+      "learning_rate": 9.370975712862929e-05,
+      "loss": 0.0001,
+      "step": 14692
+    },
+    {
+      "epoch": 62.8,
+      "learning_rate": 9.370409690504169e-05,
+      "loss": 0.0005,
+      "step": 14696
+    },
+    {
+      "epoch": 62.82,
+      "learning_rate": 9.369843430704864e-05,
+      "loss": 0.0001,
+      "step": 14700
+    },
+    {
+      "epoch": 62.84,
+      "learning_rate": 9.369276933495781e-05,
+      "loss": 0.0009,
+      "step": 14704
+    },
+    {
+      "epoch": 62.85,
+      "learning_rate": 9.368710198907695e-05,
+      "loss": 0.0001,
+      "step": 14708
+    },
+    {
+      "epoch": 62.87,
+      "learning_rate": 9.368143226971395e-05,
+      "loss": 0.0001,
+      "step": 14712
+    },
+    {
+      "epoch": 62.89,
+      "learning_rate": 9.367576017717689e-05,
+      "loss": 0.0002,
+      "step": 14716
+    },
+    {
+      "epoch": 62.91,
+      "learning_rate": 9.367008571177388e-05,
+      "loss": 0.0023,
+      "step": 14720
+    },
+    {
+      "epoch": 62.92,
+      "learning_rate": 9.366440887381324e-05,
+      "loss": 0.0011,
+      "step": 14724
+    },
+    {
+      "epoch": 62.94,
+      "learning_rate": 9.365872966360336e-05,
+      "loss": 0.0002,
+      "step": 14728
+    },
+    {
+      "epoch": 62.96,
+      "learning_rate": 9.36530480814528e-05,
+      "loss": 0.0006,
+      "step": 14732
+    },
+    {
+      "epoch": 62.97,
+      "learning_rate": 9.364736412767026e-05,
+      "loss": 0.0002,
+      "step": 14736
+    },
+    {
+      "epoch": 62.99,
+      "learning_rate": 9.364167780256452e-05,
+      "loss": 0.0006,
+      "step": 14740
+    },
+    {
+      "epoch": 63.01,
+      "learning_rate": 9.363598910644449e-05,
+      "loss": 0.0007,
+      "step": 14744
+    },
+    {
+      "epoch": 63.03,
+      "learning_rate": 9.363029803961927e-05,
+      "loss": 0.0003,
+      "step": 14748
+    },
+    {
+      "epoch": 63.04,
+      "learning_rate": 9.362460460239804e-05,
+      "loss": 0.0001,
+      "step": 14752
+    },
+    {
+      "epoch": 63.06,
+      "learning_rate": 9.361890879509012e-05,
+      "loss": 0.0001,
+      "step": 14756
+    },
+    {
+      "epoch": 63.08,
+      "learning_rate": 9.361321061800494e-05,
+      "loss": 0.0003,
+      "step": 14760
+    },
+    {
+      "epoch": 63.09,
+      "learning_rate": 9.36075100714521e-05,
+      "loss": 0.0003,
+      "step": 14764
+    },
+    {
+      "epoch": 63.11,
+      "learning_rate": 9.36018071557413e-05,
+      "loss": 0.0007,
+      "step": 14768
+    },
+    {
+      "epoch": 63.13,
+      "learning_rate": 9.359610187118238e-05,
+      "loss": 0.0001,
+      "step": 14772
+    },
+    {
+      "epoch": 63.15,
+      "learning_rate": 9.359039421808529e-05,
+      "loss": 0.0002,
+      "step": 14776
+    },
+    {
+      "epoch": 63.16,
+      "learning_rate": 9.358468419676012e-05,
+      "loss": 0.0013,
+      "step": 14780
+    },
+    {
+      "epoch": 63.18,
+      "learning_rate": 9.357897180751709e-05,
+      "loss": 0.0001,
+      "step": 14784
+    },
+    {
+      "epoch": 63.2,
+      "learning_rate": 9.357325705066658e-05,
+      "loss": 0.0001,
+      "step": 14788
+    },
+    {
+      "epoch": 63.21,
+      "learning_rate": 9.356753992651903e-05,
+      "loss": 0.0008,
+      "step": 14792
+    },
+    {
+      "epoch": 63.23,
+      "learning_rate": 9.356182043538506e-05,
+      "loss": 0.0001,
+      "step": 14796
+    },
+    {
+      "epoch": 63.25,
+      "learning_rate": 9.355609857757541e-05,
+      "loss": 0.0001,
+      "step": 14800
+    },
+    {
+      "epoch": 63.26,
+      "learning_rate": 9.355037435340095e-05,
+      "loss": 0.0002,
+      "step": 14804
+    },
+    {
+      "epoch": 63.28,
+      "learning_rate": 9.354464776317265e-05,
+      "loss": 0.0002,
+      "step": 14808
+    },
+    {
+      "epoch": 63.3,
+      "learning_rate": 9.353891880720163e-05,
+      "loss": 0.0002,
+      "step": 14812
+    },
+    {
+      "epoch": 63.32,
+      "learning_rate": 9.353318748579916e-05,
+      "loss": 0.0008,
+      "step": 14816
+    },
+    {
+      "epoch": 63.33,
+      "learning_rate": 9.352745379927661e-05,
+      "loss": 0.0008,
+      "step": 14820
+    },
+    {
+      "epoch": 63.35,
+      "learning_rate": 9.35217177479455e-05,
+      "loss": 0.0002,
+      "step": 14824
+    },
+    {
+      "epoch": 63.37,
+      "learning_rate": 9.351597933211742e-05,
+      "loss": 0.0002,
+      "step": 14828
+    },
+    {
+      "epoch": 63.38,
+      "learning_rate": 9.35102385521042e-05,
+      "loss": 0.0001,
+      "step": 14832
+    },
+    {
+      "epoch": 63.4,
+      "learning_rate": 9.350449540821767e-05,
+      "loss": 0.0002,
+      "step": 14836
+    },
+    {
+      "epoch": 63.42,
+      "learning_rate": 9.349874990076988e-05,
+      "loss": 0.0002,
+      "step": 14840
+    },
+    {
+      "epoch": 63.44,
+      "learning_rate": 9.349300203007296e-05,
+      "loss": 0.0021,
+      "step": 14844
+    },
+    {
+      "epoch": 63.45,
+      "learning_rate": 9.348725179643922e-05,
+      "loss": 0.0005,
+      "step": 14848
+    },
+    {
+      "epoch": 63.47,
+      "learning_rate": 9.348149920018104e-05,
+      "loss": 0.0001,
+      "step": 14852
+    },
+    {
+      "epoch": 63.49,
+      "learning_rate": 9.347574424161093e-05,
+      "loss": 0.0002,
+      "step": 14856
+    },
+    {
+      "epoch": 63.5,
+      "learning_rate": 9.346998692104162e-05,
+      "loss": 0.0009,
+      "step": 14860
+    },
+    {
+      "epoch": 63.52,
+      "learning_rate": 9.346422723878585e-05,
+      "loss": 0.0012,
+      "step": 14864
+    },
+    {
+      "epoch": 63.54,
+      "learning_rate": 9.345846519515654e-05,
+      "loss": 0.0001,
+      "step": 14868
+    },
+    {
+      "epoch": 63.56,
+      "learning_rate": 9.345270079046675e-05,
+      "loss": 0.001,
+      "step": 14872
+    },
+    {
+      "epoch": 63.57,
+      "learning_rate": 9.344693402502964e-05,
+      "loss": 0.0013,
+      "step": 14876
+    },
+    {
+      "epoch": 63.59,
+      "learning_rate": 9.344116489915855e-05,
+      "loss": 0.0005,
+      "step": 14880
+    },
+    {
+      "epoch": 63.61,
+      "learning_rate": 9.343539341316687e-05,
+      "loss": 0.0007,
+      "step": 14884
+    },
+    {
+      "epoch": 63.62,
+      "learning_rate": 9.342961956736817e-05,
+      "loss": 0.0009,
+      "step": 14888
+    },
+    {
+      "epoch": 63.64,
+      "learning_rate": 9.342384336207614e-05,
+      "loss": 0.0002,
+      "step": 14892
+    },
+    {
+      "epoch": 63.66,
+      "learning_rate": 9.341806479760463e-05,
+      "loss": 0.0014,
+      "step": 14896
+    },
+    {
+      "epoch": 63.68,
+      "learning_rate": 9.341228387426752e-05,
+      "loss": 0.0001,
+      "step": 14900
+    },
+    {
+      "epoch": 63.69,
+      "learning_rate": 9.340650059237893e-05,
+      "loss": 0.0002,
+      "step": 14904
+    },
+    {
+      "epoch": 63.71,
+      "learning_rate": 9.340071495225305e-05,
+      "loss": 0.0008,
+      "step": 14908
+    },
+    {
+      "epoch": 63.73,
+      "learning_rate": 9.33949269542042e-05,
+      "loss": 0.001,
+      "step": 14912
+    },
+    {
+      "epoch": 63.74,
+      "learning_rate": 9.338913659854684e-05,
+      "loss": 0.0001,
+      "step": 14916
+    },
+    {
+      "epoch": 63.76,
+      "learning_rate": 9.338334388559557e-05,
+      "loss": 0.0004,
+      "step": 14920
+    },
+    {
+      "epoch": 63.78,
+      "learning_rate": 9.33775488156651e-05,
+      "loss": 0.0003,
+      "step": 14924
+    },
+    {
+      "epoch": 63.79,
+      "learning_rate": 9.337175138907024e-05,
+      "loss": 0.0007,
+      "step": 14928
+    },
+    {
+      "epoch": 63.81,
+      "learning_rate": 9.336595160612598e-05,
+      "loss": 0.0004,
+      "step": 14932
+    },
+    {
+      "epoch": 63.83,
+      "learning_rate": 9.336014946714743e-05,
+      "loss": 0.0002,
+      "step": 14936
+    },
+    {
+      "epoch": 63.85,
+      "learning_rate": 9.335434497244979e-05,
+      "loss": 0.0003,
+      "step": 14940
+    },
+    {
+      "epoch": 63.86,
+      "learning_rate": 9.334853812234845e-05,
+      "loss": 0.0003,
+      "step": 14944
+    },
+    {
+      "epoch": 63.88,
+      "learning_rate": 9.334272891715885e-05,
+      "loss": 0.0005,
+      "step": 14948
+    },
+    {
+      "epoch": 63.9,
+      "learning_rate": 9.333691735719662e-05,
+      "loss": 0.0001,
+      "step": 14952
+    },
+    {
+      "epoch": 63.91,
+      "learning_rate": 9.333110344277749e-05,
+      "loss": 0.0002,
+      "step": 14956
+    },
+    {
+      "epoch": 63.93,
+      "learning_rate": 9.332528717421734e-05,
+      "loss": 0.001,
+      "step": 14960
+    },
+    {
+      "epoch": 63.95,
+      "learning_rate": 9.331946855183215e-05,
+      "loss": 0.0001,
+      "step": 14964
+    },
+    {
+      "epoch": 63.97,
+      "learning_rate": 9.331364757593803e-05,
+      "loss": 0.0008,
+      "step": 14968
+    },
+    {
+      "epoch": 63.98,
+      "learning_rate": 9.330782424685125e-05,
+      "loss": 0.0002,
+      "step": 14972
+    },
+    {
+      "epoch": 64.0,
+      "learning_rate": 9.330199856488817e-05,
+      "loss": 0.0005,
+      "step": 14976
+    },
+    {
+      "epoch": 64.02,
+      "learning_rate": 9.329617053036531e-05,
+      "loss": 0.0001,
+      "step": 14980
+    },
+    {
+      "epoch": 64.03,
+      "learning_rate": 9.329034014359929e-05,
+      "loss": 0.0002,
+      "step": 14984
+    },
+    {
+      "epoch": 64.05,
+      "learning_rate": 9.328450740490687e-05,
+      "loss": 0.0001,
+      "step": 14988
+    },
+    {
+      "epoch": 64.07,
+      "learning_rate": 9.327867231460495e-05,
+      "loss": 0.0004,
+      "step": 14992
+    },
+    {
+      "epoch": 64.09,
+      "learning_rate": 9.327283487301052e-05,
+      "loss": 0.0003,
+      "step": 14996
+    },
+    {
+      "epoch": 64.1,
+      "learning_rate": 9.326699508044075e-05,
+      "loss": 0.0028,
+      "step": 15000
+    },
+    {
+      "epoch": 64.1,
+      "eval_exact_match": 0.501039501039501,
+      "eval_loss": 0.8530089259147644,
+      "eval_runtime": 139.8575,
+      "eval_samples_per_second": 6.878,
+      "step": 15000
+    },
+    {
+      "epoch": 64.12,
+      "learning_rate": 9.326115293721289e-05,
+      "loss": 0.0002,
+      "step": 15004
+    },
+    {
+      "epoch": 64.14,
+      "learning_rate": 9.325530844364437e-05,
+      "loss": 0.0003,
+      "step": 15008
+    },
+    {
+      "epoch": 64.15,
+      "learning_rate": 9.324946160005268e-05,
+      "loss": 0.0002,
+      "step": 15012
+    },
+    {
+      "epoch": 64.17,
+      "learning_rate": 9.324361240675548e-05,
+      "loss": 0.0002,
+      "step": 15016
+    },
+    {
+      "epoch": 64.19,
+      "learning_rate": 9.323776086407058e-05,
+      "loss": 0.0002,
+      "step": 15020
+    },
+    {
+      "epoch": 64.21,
+      "learning_rate": 9.323190697231586e-05,
+      "loss": 0.0001,
+      "step": 15024
+    },
+    {
+      "epoch": 64.22,
+      "learning_rate": 9.322605073180937e-05,
+      "loss": 0.0003,
+      "step": 15028
+    },
+    {
+      "epoch": 64.24,
+      "learning_rate": 9.322019214286926e-05,
+      "loss": 0.0001,
+      "step": 15032
+    },
+    {
+      "epoch": 64.26,
+      "learning_rate": 9.321433120581384e-05,
+      "loss": 0.0007,
+      "step": 15036
+    },
+    {
+      "epoch": 64.27,
+      "learning_rate": 9.320846792096152e-05,
+      "loss": 0.0002,
+      "step": 15040
+    },
+    {
+      "epoch": 64.29,
+      "learning_rate": 9.320260228863084e-05,
+      "loss": 0.0002,
+      "step": 15044
+    },
+    {
+      "epoch": 64.31,
+      "learning_rate": 9.31967343091405e-05,
+      "loss": 0.0007,
+      "step": 15048
+    },
+    {
+      "epoch": 64.32,
+      "learning_rate": 9.319086398280928e-05,
+      "loss": 0.0001,
+      "step": 15052
+    },
+    {
+      "epoch": 64.34,
+      "learning_rate": 9.318499130995613e-05,
+      "loss": 0.0007,
+      "step": 15056
+    },
+    {
+      "epoch": 64.36,
+      "learning_rate": 9.317911629090007e-05,
+      "loss": 0.0001,
+      "step": 15060
+    },
+    {
+      "epoch": 64.38,
+      "learning_rate": 9.317323892596033e-05,
+      "loss": 0.0003,
+      "step": 15064
+    },
+    {
+      "epoch": 64.39,
+      "learning_rate": 9.316735921545618e-05,
+      "loss": 0.0001,
+      "step": 15068
+    },
+    {
+      "epoch": 64.41,
+      "learning_rate": 9.316147715970709e-05,
+      "loss": 0.0001,
+      "step": 15072
+    },
+    {
+      "epoch": 64.43,
+      "learning_rate": 9.315559275903261e-05,
+      "loss": 0.0005,
+      "step": 15076
+    },
+    {
+      "epoch": 64.44,
+      "learning_rate": 9.314970601375245e-05,
+      "loss": 0.0005,
+      "step": 15080
+    },
+    {
+      "epoch": 64.46,
+      "learning_rate": 9.314381692418642e-05,
+      "loss": 0.0001,
+      "step": 15084
+    },
+    {
+      "epoch": 64.48,
+      "learning_rate": 9.313792549065446e-05,
+      "loss": 0.0003,
+      "step": 15088
+    },
+    {
+      "epoch": 64.5,
+      "learning_rate": 9.313203171347667e-05,
+      "loss": 0.0002,
+      "step": 15092
+    },
+    {
+      "epoch": 64.51,
+      "learning_rate": 9.312613559297325e-05,
+      "loss": 0.002,
+      "step": 15096
+    },
+    {
+      "epoch": 64.53,
+      "learning_rate": 9.312023712946452e-05,
+      "loss": 0.0001,
+      "step": 15100
+    },
+    {
+      "epoch": 64.55,
+      "learning_rate": 9.311433632327093e-05,
+      "loss": 0.0001,
+      "step": 15104
+    },
+    {
+      "epoch": 64.56,
+      "learning_rate": 9.310843317471308e-05,
+      "loss": 0.0005,
+      "step": 15108
+    },
+    {
+      "epoch": 64.58,
+      "learning_rate": 9.310252768411167e-05,
+      "loss": 0.0017,
+      "step": 15112
+    },
+    {
+      "epoch": 64.6,
+      "learning_rate": 9.309661985178756e-05,
+      "loss": 0.0002,
+      "step": 15116
+    },
+    {
+      "epoch": 64.62,
+      "learning_rate": 9.309070967806171e-05,
+      "loss": 0.0001,
+      "step": 15120
+    },
+    {
+      "epoch": 64.63,
+      "learning_rate": 9.308479716325522e-05,
+      "loss": 0.0001,
+      "step": 15124
+    },
+    {
+      "epoch": 64.65,
+      "learning_rate": 9.307888230768929e-05,
+      "loss": 0.0001,
+      "step": 15128
+    },
+    {
+      "epoch": 64.67,
+      "learning_rate": 9.307296511168528e-05,
+      "loss": 0.0001,
+      "step": 15132
+    },
+    {
+      "epoch": 64.68,
+      "learning_rate": 9.306704557556468e-05,
+      "loss": 0.0014,
+      "step": 15136
+    },
+    {
+      "epoch": 64.7,
+      "learning_rate": 9.306112369964908e-05,
+      "loss": 0.0001,
+      "step": 15140
+    },
+    {
+      "epoch": 64.72,
+      "learning_rate": 9.305519948426022e-05,
+      "loss": 0.0001,
+      "step": 15144
+    },
+    {
+      "epoch": 64.74,
+      "learning_rate": 9.304927292971994e-05,
+      "loss": 0.0002,
+      "step": 15148
+    },
+    {
+      "epoch": 64.75,
+      "learning_rate": 9.304334403635023e-05,
+      "loss": 0.0003,
+      "step": 15152
+    },
+    {
+      "epoch": 64.77,
+      "learning_rate": 9.303741280447322e-05,
+      "loss": 0.0007,
+      "step": 15156
+    },
+    {
+      "epoch": 64.79,
+      "learning_rate": 9.303147923441112e-05,
+      "loss": 0.0001,
+      "step": 15160
+    },
+    {
+      "epoch": 64.8,
+      "learning_rate": 9.302554332648633e-05,
+      "loss": 0.0001,
+      "step": 15164
+    },
+    {
+      "epoch": 64.82,
+      "learning_rate": 9.30196050810213e-05,
+      "loss": 0.0003,
+      "step": 15168
+    },
+    {
+      "epoch": 64.84,
+      "learning_rate": 9.301366449833868e-05,
+      "loss": 0.0007,
+      "step": 15172
+    },
+    {
+      "epoch": 64.85,
+      "learning_rate": 9.30077215787612e-05,
+      "loss": 0.0002,
+      "step": 15176
+    },
+    {
+      "epoch": 64.87,
+      "learning_rate": 9.300177632261176e-05,
+      "loss": 0.0001,
+      "step": 15180
+    },
+    {
+      "epoch": 64.89,
+      "learning_rate": 9.299582873021333e-05,
+      "loss": 0.0021,
+      "step": 15184
+    },
+    {
+      "epoch": 64.91,
+      "learning_rate": 9.298987880188907e-05,
+      "loss": 0.0002,
+      "step": 15188
+    },
+    {
+      "epoch": 64.92,
+      "learning_rate": 9.29839265379622e-05,
+      "loss": 0.0003,
+      "step": 15192
+    },
+    {
+      "epoch": 64.94,
+      "learning_rate": 9.297797193875612e-05,
+      "loss": 0.0003,
+      "step": 15196
+    },
+    {
+      "epoch": 64.96,
+      "learning_rate": 9.297201500459433e-05,
+      "loss": 0.0001,
+      "step": 15200
+    },
+    {
+      "epoch": 64.97,
+      "learning_rate": 9.296605573580049e-05,
+      "loss": 0.0009,
+      "step": 15204
+    },
+    {
+      "epoch": 64.99,
+      "learning_rate": 9.296009413269833e-05,
+      "loss": 0.0003,
+      "step": 15208
+    },
+    {
+      "epoch": 65.01,
+      "learning_rate": 9.295413019561174e-05,
+      "loss": 0.0018,
+      "step": 15212
+    },
+    {
+      "epoch": 65.03,
+      "learning_rate": 9.294816392486476e-05,
+      "loss": 0.0001,
+      "step": 15216
+    },
+    {
+      "epoch": 65.04,
+      "learning_rate": 9.294219532078151e-05,
+      "loss": 0.0002,
+      "step": 15220
+    },
+    {
+      "epoch": 65.06,
+      "learning_rate": 9.293622438368628e-05,
+      "loss": 0.0002,
+      "step": 15224
+    },
+    {
+      "epoch": 65.08,
+      "learning_rate": 9.293025111390343e-05,
+      "loss": 0.0001,
+      "step": 15228
+    },
+    {
+      "epoch": 65.09,
+      "learning_rate": 9.292427551175753e-05,
+      "loss": 0.0001,
+      "step": 15232
+    },
+    {
+      "epoch": 65.11,
+      "learning_rate": 9.291829757757319e-05,
+      "loss": 0.0003,
+      "step": 15236
+    },
+    {
+      "epoch": 65.13,
+      "learning_rate": 9.291231731167521e-05,
+      "loss": 0.0008,
+      "step": 15240
+    },
+    {
+      "epoch": 65.15,
+      "learning_rate": 9.290633471438847e-05,
+      "loss": 0.0004,
+      "step": 15244
+    },
+    {
+      "epoch": 65.16,
+      "learning_rate": 9.290034978603803e-05,
+      "loss": 0.0008,
+      "step": 15248
+    },
+    {
+      "epoch": 65.18,
+      "learning_rate": 9.289436252694902e-05,
+      "loss": 0.0009,
+      "step": 15252
+    },
+    {
+      "epoch": 65.2,
+      "learning_rate": 9.288837293744673e-05,
+      "loss": 0.0004,
+      "step": 15256
+    },
+    {
+      "epoch": 65.21,
+      "learning_rate": 9.288238101785658e-05,
+      "loss": 0.0007,
+      "step": 15260
+    },
+    {
+      "epoch": 65.23,
+      "learning_rate": 9.287638676850409e-05,
+      "loss": 0.0002,
+      "step": 15264
+    },
+    {
+      "epoch": 65.25,
+      "learning_rate": 9.287039018971493e-05,
+      "loss": 0.0006,
+      "step": 15268
+    },
+    {
+      "epoch": 65.26,
+      "learning_rate": 9.286439128181488e-05,
+      "loss": 0.0001,
+      "step": 15272
+    },
+    {
+      "epoch": 65.28,
+      "learning_rate": 9.285839004512986e-05,
+      "loss": 0.0006,
+      "step": 15276
+    },
+    {
+      "epoch": 65.3,
+      "learning_rate": 9.285238647998592e-05,
+      "loss": 0.0001,
+      "step": 15280
+    },
+    {
+      "epoch": 65.32,
+      "learning_rate": 9.284638058670924e-05,
+      "loss": 0.0003,
+      "step": 15284
+    },
+    {
+      "epoch": 65.33,
+      "learning_rate": 9.284037236562608e-05,
+      "loss": 0.0002,
+      "step": 15288
+    },
+    {
+      "epoch": 65.35,
+      "learning_rate": 9.283436181706288e-05,
+      "loss": 0.0003,
+      "step": 15292
+    },
+    {
+      "epoch": 65.37,
+      "learning_rate": 9.28283489413462e-05,
+      "loss": 0.0007,
+      "step": 15296
+    },
+    {
+      "epoch": 65.38,
+      "learning_rate": 9.28223337388027e-05,
+      "loss": 0.0004,
+      "step": 15300
+    },
+    {
+      "epoch": 65.4,
+      "learning_rate": 9.281631620975917e-05,
+      "loss": 0.0001,
+      "step": 15304
+    },
+    {
+      "epoch": 65.42,
+      "learning_rate": 9.281029635454256e-05,
+      "loss": 0.0004,
+      "step": 15308
+    },
+    {
+      "epoch": 65.44,
+      "learning_rate": 9.280427417347991e-05,
+      "loss": 0.0004,
+      "step": 15312
+    },
+    {
+      "epoch": 65.45,
+      "learning_rate": 9.27982496668984e-05,
+      "loss": 0.0006,
+      "step": 15316
+    },
+    {
+      "epoch": 65.47,
+      "learning_rate": 9.279222283512534e-05,
+      "loss": 0.0002,
+      "step": 15320
+    },
+    {
+      "epoch": 65.49,
+      "learning_rate": 9.278619367848818e-05,
+      "loss": 0.0008,
+      "step": 15324
+    },
+    {
+      "epoch": 65.5,
+      "learning_rate": 9.278016219731442e-05,
+      "loss": 0.0005,
+      "step": 15328
+    },
+    {
+      "epoch": 65.52,
+      "learning_rate": 9.277412839193183e-05,
+      "loss": 0.0011,
+      "step": 15332
+    },
+    {
+      "epoch": 65.54,
+      "learning_rate": 9.276809226266814e-05,
+      "loss": 0.0004,
+      "step": 15336
+    },
+    {
+      "epoch": 65.56,
+      "learning_rate": 9.276205380985134e-05,
+      "loss": 0.0008,
+      "step": 15340
+    },
+    {
+      "epoch": 65.57,
+      "learning_rate": 9.27560130338095e-05,
+      "loss": 0.0004,
+      "step": 15344
+    },
+    {
+      "epoch": 65.59,
+      "learning_rate": 9.274996993487074e-05,
+      "loss": 0.0003,
+      "step": 15348
+    },
+    {
+      "epoch": 65.61,
+      "learning_rate": 9.274392451336346e-05,
+      "loss": 0.0013,
+      "step": 15352
+    },
+    {
+      "epoch": 65.62,
+      "learning_rate": 9.273787676961607e-05,
+      "loss": 0.0002,
+      "step": 15356
+    },
+    {
+      "epoch": 65.64,
+      "learning_rate": 9.273182670395713e-05,
+      "loss": 0.0005,
+      "step": 15360
+    },
+    {
+      "epoch": 65.66,
+      "learning_rate": 9.272577431671534e-05,
+      "loss": 0.0003,
+      "step": 15364
+    },
+    {
+      "epoch": 65.68,
+      "learning_rate": 9.271971960821952e-05,
+      "loss": 0.0006,
+      "step": 15368
+    },
+    {
+      "epoch": 65.69,
+      "learning_rate": 9.271366257879861e-05,
+      "loss": 0.0004,
+      "step": 15372
+    },
+    {
+      "epoch": 65.71,
+      "learning_rate": 9.27076032287817e-05,
+      "loss": 0.0003,
+      "step": 15376
+    },
+    {
+      "epoch": 65.73,
+      "learning_rate": 9.270154155849799e-05,
+      "loss": 0.0013,
+      "step": 15380
+    },
+    {
+      "epoch": 65.74,
+      "learning_rate": 9.269547756827679e-05,
+      "loss": 0.001,
+      "step": 15384
+    },
+    {
+      "epoch": 65.76,
+      "learning_rate": 9.268941125844755e-05,
+      "loss": 0.0001,
+      "step": 15388
+    },
+    {
+      "epoch": 65.78,
+      "learning_rate": 9.268334262933986e-05,
+      "loss": 0.0002,
+      "step": 15392
+    },
+    {
+      "epoch": 65.79,
+      "learning_rate": 9.267727168128342e-05,
+      "loss": 0.0002,
+      "step": 15396
+    },
+    {
+      "epoch": 65.81,
+      "learning_rate": 9.267119841460805e-05,
+      "loss": 0.0002,
+      "step": 15400
+    },
+    {
+      "epoch": 65.83,
+      "learning_rate": 9.266512282964372e-05,
+      "loss": 0.0006,
+      "step": 15404
+    },
+    {
+      "epoch": 65.85,
+      "learning_rate": 9.26590449267205e-05,
+      "loss": 0.0001,
+      "step": 15408
+    },
+    {
+      "epoch": 65.86,
+      "learning_rate": 9.265296470616861e-05,
+      "loss": 0.0001,
+      "step": 15412
+    },
+    {
+      "epoch": 65.88,
+      "learning_rate": 9.264688216831836e-05,
+      "loss": 0.0002,
+      "step": 15416
+    },
+    {
+      "epoch": 65.9,
+      "learning_rate": 9.264079731350024e-05,
+      "loss": 0.0001,
+      "step": 15420
+    },
+    {
+      "epoch": 65.91,
+      "learning_rate": 9.263471014204481e-05,
+      "loss": 0.0001,
+      "step": 15424
+    },
+    {
+      "epoch": 65.93,
+      "learning_rate": 9.262862065428278e-05,
+      "loss": 0.0001,
+      "step": 15428
+    },
+    {
+      "epoch": 65.95,
+      "learning_rate": 9.2622528850545e-05,
+      "loss": 0.0001,
+      "step": 15432
+    },
+    {
+      "epoch": 65.97,
+      "learning_rate": 9.261643473116244e-05,
+      "loss": 0.0003,
+      "step": 15436
+    },
+    {
+      "epoch": 65.98,
+      "learning_rate": 9.261033829646617e-05,
+      "loss": 0.0002,
+      "step": 15440
+    },
+    {
+      "epoch": 66.0,
+      "learning_rate": 9.26042395467874e-05,
+      "loss": 0.0001,
+      "step": 15444
+    },
+    {
+      "epoch": 66.02,
+      "learning_rate": 9.259813848245749e-05,
+      "loss": 0.0004,
+      "step": 15448
+    },
+    {
+      "epoch": 66.03,
+      "learning_rate": 9.259203510380789e-05,
+      "loss": 0.0001,
+      "step": 15452
+    },
+    {
+      "epoch": 66.05,
+      "learning_rate": 9.25859294111702e-05,
+      "loss": 0.0008,
+      "step": 15456
+    },
+    {
+      "epoch": 66.07,
+      "learning_rate": 9.257982140487613e-05,
+      "loss": 0.0002,
+      "step": 15460
+    },
+    {
+      "epoch": 66.09,
+      "learning_rate": 9.257371108525753e-05,
+      "loss": 0.0001,
+      "step": 15464
+    },
+    {
+      "epoch": 66.1,
+      "learning_rate": 9.256759845264636e-05,
+      "loss": 0.0004,
+      "step": 15468
+    },
+    {
+      "epoch": 66.12,
+      "learning_rate": 9.256148350737472e-05,
+      "loss": 0.0002,
+      "step": 15472
+    },
+    {
+      "epoch": 66.14,
+      "learning_rate": 9.255536624977484e-05,
+      "loss": 0.0001,
+      "step": 15476
+    },
+    {
+      "epoch": 66.15,
+      "learning_rate": 9.254924668017903e-05,
+      "loss": 0.0004,
+      "step": 15480
+    },
+    {
+      "epoch": 66.17,
+      "learning_rate": 9.25431247989198e-05,
+      "loss": 0.0001,
+      "step": 15484
+    },
+    {
+      "epoch": 66.19,
+      "learning_rate": 9.253700060632972e-05,
+      "loss": 0.0005,
+      "step": 15488
+    },
+    {
+      "epoch": 66.21,
+      "learning_rate": 9.253087410274153e-05,
+      "loss": 0.0001,
+      "step": 15492
+    },
+    {
+      "epoch": 66.22,
+      "learning_rate": 9.252474528848807e-05,
+      "loss": 0.0001,
+      "step": 15496
+    },
+    {
+      "epoch": 66.24,
+      "learning_rate": 9.251861416390232e-05,
+      "loss": 0.0002,
+      "step": 15500
+    },
+    {
+      "epoch": 66.26,
+      "learning_rate": 9.251248072931737e-05,
+      "loss": 0.0005,
+      "step": 15504
+    },
+    {
+      "epoch": 66.27,
+      "learning_rate": 9.250634498506643e-05,
+      "loss": 0.0001,
+      "step": 15508
+    },
+    {
+      "epoch": 66.29,
+      "learning_rate": 9.25002069314829e-05,
+      "loss": 0.0011,
+      "step": 15512
+    },
+    {
+      "epoch": 66.31,
+      "learning_rate": 9.249406656890019e-05,
+      "loss": 0.0002,
+      "step": 15516
+    },
+    {
+      "epoch": 66.32,
+      "learning_rate": 9.248792389765193e-05,
+      "loss": 0.0,
+      "step": 15520
+    },
+    {
+      "epoch": 66.34,
+      "learning_rate": 9.248177891807186e-05,
+      "loss": 0.0002,
+      "step": 15524
+    },
+    {
+      "epoch": 66.36,
+      "learning_rate": 9.247563163049381e-05,
+      "loss": 0.0012,
+      "step": 15528
+    },
+    {
+      "epoch": 66.38,
+      "learning_rate": 9.246948203525176e-05,
+      "loss": 0.0001,
+      "step": 15532
+    },
+    {
+      "epoch": 66.39,
+      "learning_rate": 9.246333013267983e-05,
+      "loss": 0.0001,
+      "step": 15536
+    },
+    {
+      "epoch": 66.41,
+      "learning_rate": 9.245717592311225e-05,
+      "loss": 0.0003,
+      "step": 15540
+    },
+    {
+      "epoch": 66.43,
+      "learning_rate": 9.245101940688333e-05,
+      "loss": 0.0001,
+      "step": 15544
+    },
+    {
+      "epoch": 66.44,
+      "learning_rate": 9.24448605843276e-05,
+      "loss": 0.0014,
+      "step": 15548
+    },
+    {
+      "epoch": 66.46,
+      "learning_rate": 9.243869945577962e-05,
+      "loss": 0.0013,
+      "step": 15552
+    },
+    {
+      "epoch": 66.48,
+      "learning_rate": 9.243253602157413e-05,
+      "loss": 0.0001,
+      "step": 15556
+    },
+    {
+      "epoch": 66.5,
+      "learning_rate": 9.242637028204602e-05,
+      "loss": 0.0004,
+      "step": 15560
+    },
+    {
+      "epoch": 66.51,
+      "learning_rate": 9.242020223753023e-05,
+      "loss": 0.0001,
+      "step": 15564
+    },
+    {
+      "epoch": 66.53,
+      "learning_rate": 9.241403188836188e-05,
+      "loss": 0.0011,
+      "step": 15568
+    },
+    {
+      "epoch": 66.55,
+      "learning_rate": 9.240785923487619e-05,
+      "loss": 0.0002,
+      "step": 15572
+    },
+    {
+      "epoch": 66.56,
+      "learning_rate": 9.240168427740852e-05,
+      "loss": 0.0004,
+      "step": 15576
+    },
+    {
+      "epoch": 66.58,
+      "learning_rate": 9.239550701629436e-05,
+      "loss": 0.0001,
+      "step": 15580
+    },
+    {
+      "epoch": 66.6,
+      "learning_rate": 9.23893274518693e-05,
+      "loss": 0.0001,
+      "step": 15584
+    },
+    {
+      "epoch": 66.62,
+      "learning_rate": 9.238314558446907e-05,
+      "loss": 0.0012,
+      "step": 15588
+    },
+    {
+      "epoch": 66.63,
+      "learning_rate": 9.237696141442952e-05,
+      "loss": 0.0015,
+      "step": 15592
+    },
+    {
+      "epoch": 66.65,
+      "learning_rate": 9.237077494208668e-05,
+      "loss": 0.0014,
+      "step": 15596
+    },
+    {
+      "epoch": 66.67,
+      "learning_rate": 9.236458616777659e-05,
+      "loss": 0.0001,
+      "step": 15600
+    },
+    {
+      "epoch": 66.68,
+      "learning_rate": 9.235839509183551e-05,
+      "loss": 0.0001,
+      "step": 15604
+    },
+    {
+      "epoch": 66.7,
+      "learning_rate": 9.23522017145998e-05,
+      "loss": 0.0002,
+      "step": 15608
+    },
+    {
+      "epoch": 66.72,
+      "learning_rate": 9.234600603640594e-05,
+      "loss": 0.0013,
+      "step": 15612
+    },
+    {
+      "epoch": 66.74,
+      "learning_rate": 9.233980805759052e-05,
+      "loss": 0.0002,
+      "step": 15616
+    },
+    {
+      "epoch": 66.75,
+      "learning_rate": 9.23336077784903e-05,
+      "loss": 0.0004,
+      "step": 15620
+    },
+    {
+      "epoch": 66.77,
+      "learning_rate": 9.23274051994421e-05,
+      "loss": 0.0002,
+      "step": 15624
+    },
+    {
+      "epoch": 66.79,
+      "learning_rate": 9.232120032078292e-05,
+      "loss": 0.0003,
+      "step": 15628
+    },
+    {
+      "epoch": 66.8,
+      "learning_rate": 9.231499314284987e-05,
+      "loss": 0.0001,
+      "step": 15632
+    },
+    {
+      "epoch": 66.82,
+      "learning_rate": 9.230878366598017e-05,
+      "loss": 0.0001,
+      "step": 15636
+    },
+    {
+      "epoch": 66.84,
+      "learning_rate": 9.230257189051118e-05,
+      "loss": 0.0003,
+      "step": 15640
+    },
+    {
+      "epoch": 66.85,
+      "learning_rate": 9.229635781678038e-05,
+      "loss": 0.0003,
+      "step": 15644
+    },
+    {
+      "epoch": 66.87,
+      "learning_rate": 9.229014144512537e-05,
+      "loss": 0.0002,
+      "step": 15648
+    },
+    {
+      "epoch": 66.89,
+      "learning_rate": 9.22839227758839e-05,
+      "loss": 0.0004,
+      "step": 15652
+    },
+    {
+      "epoch": 66.91,
+      "learning_rate": 9.22777018093938e-05,
+      "loss": 0.0004,
+      "step": 15656
+    },
+    {
+      "epoch": 66.92,
+      "learning_rate": 9.227147854599306e-05,
+      "loss": 0.0017,
+      "step": 15660
+    },
+    {
+      "epoch": 66.94,
+      "learning_rate": 9.226525298601979e-05,
+      "loss": 0.0001,
+      "step": 15664
+    },
+    {
+      "epoch": 66.96,
+      "learning_rate": 9.225902512981221e-05,
+      "loss": 0.0003,
+      "step": 15668
+    },
+    {
+      "epoch": 66.97,
+      "learning_rate": 9.225279497770867e-05,
+      "loss": 0.0001,
+      "step": 15672
+    },
+    {
+      "epoch": 66.99,
+      "learning_rate": 9.224656253004766e-05,
+      "loss": 0.0015,
+      "step": 15676
+    },
+    {
+      "epoch": 67.01,
+      "learning_rate": 9.224032778716781e-05,
+      "loss": 0.0005,
+      "step": 15680
+    },
+    {
+      "epoch": 67.03,
+      "learning_rate": 9.223409074940777e-05,
+      "loss": 0.0001,
+      "step": 15684
+    },
+    {
+      "epoch": 67.04,
+      "learning_rate": 9.222785141710646e-05,
+      "loss": 0.0002,
+      "step": 15688
+    },
+    {
+      "epoch": 67.06,
+      "learning_rate": 9.222160979060286e-05,
+      "loss": 0.0003,
+      "step": 15692
+    },
+    {
+      "epoch": 67.08,
+      "learning_rate": 9.221536587023603e-05,
+      "loss": 0.0005,
+      "step": 15696
+    },
+    {
+      "epoch": 67.09,
+      "learning_rate": 9.220911965634521e-05,
+      "loss": 0.0008,
+      "step": 15700
+    },
+    {
+      "epoch": 67.11,
+      "learning_rate": 9.220287114926978e-05,
+      "loss": 0.0001,
+      "step": 15704
+    },
+    {
+      "epoch": 67.13,
+      "learning_rate": 9.219662034934917e-05,
+      "loss": 0.0,
+      "step": 15708
+    },
+    {
+      "epoch": 67.15,
+      "learning_rate": 9.219036725692302e-05,
+      "loss": 0.0002,
+      "step": 15712
+    },
+    {
+      "epoch": 67.16,
+      "learning_rate": 9.218411187233103e-05,
+      "loss": 0.0,
+      "step": 15716
+    },
+    {
+      "epoch": 67.18,
+      "learning_rate": 9.217785419591307e-05,
+      "loss": 0.0012,
+      "step": 15720
+    },
+    {
+      "epoch": 67.2,
+      "learning_rate": 9.217159422800909e-05,
+      "loss": 0.0002,
+      "step": 15724
+    },
+    {
+      "epoch": 67.21,
+      "learning_rate": 9.216533196895922e-05,
+      "loss": 0.0004,
+      "step": 15728
+    },
+    {
+      "epoch": 67.23,
+      "learning_rate": 9.215906741910365e-05,
+      "loss": 0.0005,
+      "step": 15732
+    },
+    {
+      "epoch": 67.25,
+      "learning_rate": 9.215280057878276e-05,
+      "loss": 0.0001,
+      "step": 15736
+    },
+    {
+      "epoch": 67.26,
+      "learning_rate": 9.214653144833699e-05,
+      "loss": 0.0001,
+      "step": 15740
+    },
+    {
+      "epoch": 67.28,
+      "learning_rate": 9.214026002810697e-05,
+      "loss": 0.0001,
+      "step": 15744
+    },
+    {
+      "epoch": 67.3,
+      "learning_rate": 9.213398631843337e-05,
+      "loss": 0.0005,
+      "step": 15748
+    },
+    {
+      "epoch": 67.32,
+      "learning_rate": 9.212771031965709e-05,
+      "loss": 0.0008,
+      "step": 15752
+    },
+    {
+      "epoch": 67.33,
+      "learning_rate": 9.212143203211907e-05,
+      "loss": 0.0003,
+      "step": 15756
+    },
+    {
+      "epoch": 67.35,
+      "learning_rate": 9.21151514561604e-05,
+      "loss": 0.0001,
+      "step": 15760
+    },
+    {
+      "epoch": 67.37,
+      "learning_rate": 9.210886859212233e-05,
+      "loss": 0.0001,
+      "step": 15764
+    },
+    {
+      "epoch": 67.38,
+      "learning_rate": 9.210258344034617e-05,
+      "loss": 0.0001,
+      "step": 15768
+    },
+    {
+      "epoch": 67.4,
+      "learning_rate": 9.209629600117341e-05,
+      "loss": 0.0001,
+      "step": 15772
+    },
+    {
+      "epoch": 67.42,
+      "learning_rate": 9.209000627494562e-05,
+      "loss": 0.0001,
+      "step": 15776
+    },
+    {
+      "epoch": 67.44,
+      "learning_rate": 9.208371426200453e-05,
+      "loss": 0.0002,
+      "step": 15780
+    },
+    {
+      "epoch": 67.45,
+      "learning_rate": 9.207741996269197e-05,
+      "loss": 0.0001,
+      "step": 15784
+    },
+    {
+      "epoch": 67.47,
+      "learning_rate": 9.207112337734989e-05,
+      "loss": 0.0014,
+      "step": 15788
+    },
+    {
+      "epoch": 67.49,
+      "learning_rate": 9.206482450632041e-05,
+      "loss": 0.0001,
+      "step": 15792
+    },
+    {
+      "epoch": 67.5,
+      "learning_rate": 9.205852334994574e-05,
+      "loss": 0.0001,
+      "step": 15796
+    },
+    {
+      "epoch": 67.52,
+      "learning_rate": 9.20522199085682e-05,
+      "loss": 0.0005,
+      "step": 15800
+    },
+    {
+      "epoch": 67.54,
+      "learning_rate": 9.204591418253023e-05,
+      "loss": 0.0004,
+      "step": 15804
+    },
+    {
+      "epoch": 67.56,
+      "learning_rate": 9.203960617217446e-05,
+      "loss": 0.0001,
+      "step": 15808
+    },
+    {
+      "epoch": 67.57,
+      "learning_rate": 9.203329587784359e-05,
+      "loss": 0.0003,
+      "step": 15812
+    },
+    {
+      "epoch": 67.59,
+      "learning_rate": 9.202698329988042e-05,
+      "loss": 0.0002,
+      "step": 15816
+    },
+    {
+      "epoch": 67.61,
+      "learning_rate": 9.202066843862794e-05,
+      "loss": 0.0003,
+      "step": 15820
+    },
+    {
+      "epoch": 67.62,
+      "learning_rate": 9.201435129442922e-05,
+      "loss": 0.0006,
+      "step": 15824
+    },
+    {
+      "epoch": 67.64,
+      "learning_rate": 9.200803186762745e-05,
+      "loss": 0.0001,
+      "step": 15828
+    },
+    {
+      "epoch": 67.66,
+      "learning_rate": 9.200171015856597e-05,
+      "loss": 0.0001,
+      "step": 15832
+    },
+    {
+      "epoch": 67.68,
+      "learning_rate": 9.199538616758827e-05,
+      "loss": 0.0003,
+      "step": 15836
+    },
+    {
+      "epoch": 67.69,
+      "learning_rate": 9.198905989503785e-05,
+      "loss": 0.0001,
+      "step": 15840
+    },
+    {
+      "epoch": 67.71,
+      "learning_rate": 9.198273134125848e-05,
+      "loss": 0.0005,
+      "step": 15844
+    },
+    {
+      "epoch": 67.73,
+      "learning_rate": 9.197640050659396e-05,
+      "loss": 0.0006,
+      "step": 15848
+    },
+    {
+      "epoch": 67.74,
+      "learning_rate": 9.197006739138822e-05,
+      "loss": 0.0003,
+      "step": 15852
+    },
+    {
+      "epoch": 67.76,
+      "learning_rate": 9.196373199598536e-05,
+      "loss": 0.0006,
+      "step": 15856
+    },
+    {
+      "epoch": 67.78,
+      "learning_rate": 9.195739432072955e-05,
+      "loss": 0.0002,
+      "step": 15860
+    },
+    {
+      "epoch": 67.79,
+      "learning_rate": 9.195105436596514e-05,
+      "loss": 0.0003,
+      "step": 15864
+    },
+    {
+      "epoch": 67.81,
+      "learning_rate": 9.194471213203655e-05,
+      "loss": 0.0001,
+      "step": 15868
+    },
+    {
+      "epoch": 67.83,
+      "learning_rate": 9.193836761928838e-05,
+      "loss": 0.0001,
+      "step": 15872
+    },
+    {
+      "epoch": 67.85,
+      "learning_rate": 9.193202082806528e-05,
+      "loss": 0.0003,
+      "step": 15876
+    },
+    {
+      "epoch": 67.86,
+      "learning_rate": 9.192567175871209e-05,
+      "loss": 0.0001,
+      "step": 15880
+    },
+    {
+      "epoch": 67.88,
+      "learning_rate": 9.191932041157375e-05,
+      "loss": 0.0001,
+      "step": 15884
+    },
+    {
+      "epoch": 67.9,
+      "learning_rate": 9.191296678699532e-05,
+      "loss": 0.0005,
+      "step": 15888
+    },
+    {
+      "epoch": 67.91,
+      "learning_rate": 9.190661088532198e-05,
+      "loss": 0.0006,
+      "step": 15892
+    },
+    {
+      "epoch": 67.93,
+      "learning_rate": 9.190025270689905e-05,
+      "loss": 0.0001,
+      "step": 15896
+    },
+    {
+      "epoch": 67.95,
+      "learning_rate": 9.189389225207196e-05,
+      "loss": 0.0001,
+      "step": 15900
+    },
+    {
+      "epoch": 67.97,
+      "learning_rate": 9.188752952118627e-05,
+      "loss": 0.0002,
+      "step": 15904
+    },
+    {
+      "epoch": 67.98,
+      "learning_rate": 9.188116451458765e-05,
+      "loss": 0.0024,
+      "step": 15908
+    },
+    {
+      "epoch": 68.0,
+      "learning_rate": 9.187479723262193e-05,
+      "loss": 0.0001,
+      "step": 15912
+    },
+    {
+      "epoch": 68.02,
+      "learning_rate": 9.186842767563501e-05,
+      "loss": 0.0005,
+      "step": 15916
+    },
+    {
+      "epoch": 68.03,
+      "learning_rate": 9.186205584397296e-05,
+      "loss": 0.0006,
+      "step": 15920
+    },
+    {
+      "epoch": 68.05,
+      "learning_rate": 9.185568173798195e-05,
+      "loss": 0.0001,
+      "step": 15924
+    },
+    {
+      "epoch": 68.07,
+      "learning_rate": 9.184930535800828e-05,
+      "loss": 0.0001,
+      "step": 15928
+    },
+    {
+      "epoch": 68.09,
+      "learning_rate": 9.184292670439839e-05,
+      "loss": 0.0004,
+      "step": 15932
+    },
+    {
+      "epoch": 68.1,
+      "learning_rate": 9.18365457774988e-05,
+      "loss": 0.0004,
+      "step": 15936
+    },
+    {
+      "epoch": 68.12,
+      "learning_rate": 9.183016257765619e-05,
+      "loss": 0.0001,
+      "step": 15940
+    },
+    {
+      "epoch": 68.14,
+      "learning_rate": 9.182377710521737e-05,
+      "loss": 0.0001,
+      "step": 15944
+    },
+    {
+      "epoch": 68.15,
+      "learning_rate": 9.181738936052924e-05,
+      "loss": 0.0001,
+      "step": 15948
+    },
+    {
+      "epoch": 68.17,
+      "learning_rate": 9.181099934393884e-05,
+      "loss": 0.0005,
+      "step": 15952
+    },
+    {
+      "epoch": 68.19,
+      "learning_rate": 9.180460705579333e-05,
+      "loss": 0.0003,
+      "step": 15956
+    },
+    {
+      "epoch": 68.21,
+      "learning_rate": 9.179821249644e-05,
+      "loss": 0.0001,
+      "step": 15960
+    },
+    {
+      "epoch": 68.22,
+      "learning_rate": 9.179181566622628e-05,
+      "loss": 0.0001,
+      "step": 15964
+    },
+    {
+      "epoch": 68.24,
+      "learning_rate": 9.178541656549971e-05,
+      "loss": 0.0001,
+      "step": 15968
+    },
+    {
+      "epoch": 68.26,
+      "learning_rate": 9.17790151946079e-05,
+      "loss": 0.0004,
+      "step": 15972
+    },
+    {
+      "epoch": 68.27,
+      "learning_rate": 9.177261155389864e-05,
+      "loss": 0.0006,
+      "step": 15976
+    },
+    {
+      "epoch": 68.29,
+      "learning_rate": 9.17662056437199e-05,
+      "loss": 0.001,
+      "step": 15980
+    },
+    {
+      "epoch": 68.31,
+      "learning_rate": 9.175979746441963e-05,
+      "loss": 0.0002,
+      "step": 15984
+    },
+    {
+      "epoch": 68.32,
+      "learning_rate": 9.175338701634601e-05,
+      "loss": 0.0002,
+      "step": 15988
+    },
+    {
+      "epoch": 68.34,
+      "learning_rate": 9.174697429984732e-05,
+      "loss": 0.0001,
+      "step": 15992
+    },
+    {
+      "epoch": 68.36,
+      "learning_rate": 9.174055931527197e-05,
+      "loss": 0.0008,
+      "step": 15996
+    },
+    {
+      "epoch": 68.38,
+      "learning_rate": 9.173414206296844e-05,
+      "loss": 0.0014,
+      "step": 16000
+    },
+    {
+      "epoch": 68.38,
+      "eval_exact_match": 0.5135135135135135,
+      "eval_loss": 0.9037817120552063,
+      "eval_runtime": 142.7035,
+      "eval_samples_per_second": 6.741,
+      "step": 16000
+    },
+    {
+      "epoch": 68.39,
+      "learning_rate": 9.17277225432854e-05,
+      "loss": 0.0002,
+      "step": 16004
+    },
+    {
+      "epoch": 68.41,
+      "learning_rate": 9.172130075657161e-05,
+      "loss": 0.0001,
+      "step": 16008
+    },
+    {
+      "epoch": 68.43,
+      "learning_rate": 9.171487670317598e-05,
+      "loss": 0.0015,
+      "step": 16012
+    },
+    {
+      "epoch": 68.44,
+      "learning_rate": 9.170845038344749e-05,
+      "loss": 0.0004,
+      "step": 16016
+    },
+    {
+      "epoch": 68.46,
+      "learning_rate": 9.17020217977353e-05,
+      "loss": 0.0005,
+      "step": 16020
+    },
+    {
+      "epoch": 68.48,
+      "learning_rate": 9.169559094638866e-05,
+      "loss": 0.0002,
+      "step": 16024
+    },
+    {
+      "epoch": 68.5,
+      "learning_rate": 9.168915782975696e-05,
+      "loss": 0.0001,
+      "step": 16028
+    },
+    {
+      "epoch": 68.51,
+      "learning_rate": 9.168272244818971e-05,
+      "loss": 0.0002,
+      "step": 16032
+    },
+    {
+      "epoch": 68.53,
+      "learning_rate": 9.167628480203652e-05,
+      "loss": 0.0009,
+      "step": 16036
+    },
+    {
+      "epoch": 68.55,
+      "learning_rate": 9.166984489164714e-05,
+      "loss": 0.0,
+      "step": 16040
+    },
+    {
+      "epoch": 68.56,
+      "learning_rate": 9.166340271737146e-05,
+      "loss": 0.0012,
+      "step": 16044
+    },
+    {
+      "epoch": 68.58,
+      "learning_rate": 9.16569582795595e-05,
+      "loss": 0.0001,
+      "step": 16048
+    },
+    {
+      "epoch": 68.6,
+      "learning_rate": 9.165051157856132e-05,
+      "loss": 0.0002,
+      "step": 16052
+    },
+    {
+      "epoch": 68.62,
+      "learning_rate": 9.164406261472721e-05,
+      "loss": 0.0001,
+      "step": 16056
+    },
+    {
+      "epoch": 68.63,
+      "learning_rate": 9.163761138840752e-05,
+      "loss": 0.0002,
+      "step": 16060
+    },
+    {
+      "epoch": 68.65,
+      "learning_rate": 9.163115789995276e-05,
+      "loss": 0.0002,
+      "step": 16064
+    },
+    {
+      "epoch": 68.67,
+      "learning_rate": 9.162470214971351e-05,
+      "loss": 0.002,
+      "step": 16068
+    },
+    {
+      "epoch": 68.68,
+      "learning_rate": 9.161824413804052e-05,
+      "loss": 0.0009,
+      "step": 16072
+    },
+    {
+      "epoch": 68.7,
+      "learning_rate": 9.161178386528467e-05,
+      "loss": 0.0001,
+      "step": 16076
+    },
+    {
+      "epoch": 68.72,
+      "learning_rate": 9.160532133179689e-05,
+      "loss": 0.0012,
+      "step": 16080
+    },
+    {
+      "epoch": 68.74,
+      "learning_rate": 9.159885653792834e-05,
+      "loss": 0.0001,
+      "step": 16084
+    },
+    {
+      "epoch": 68.75,
+      "learning_rate": 9.159238948403022e-05,
+      "loss": 0.0001,
+      "step": 16088
+    },
+    {
+      "epoch": 68.77,
+      "learning_rate": 9.158592017045387e-05,
+      "loss": 0.0002,
+      "step": 16092
+    },
+    {
+      "epoch": 68.79,
+      "learning_rate": 9.157944859755076e-05,
+      "loss": 0.0003,
+      "step": 16096
+    },
+    {
+      "epoch": 68.8,
+      "learning_rate": 9.157297476567253e-05,
+      "loss": 0.0002,
+      "step": 16100
+    },
+    {
+      "epoch": 68.82,
+      "learning_rate": 9.156649867517084e-05,
+      "loss": 0.0002,
+      "step": 16104
+    },
+    {
+      "epoch": 68.84,
+      "learning_rate": 9.156002032639757e-05,
+      "loss": 0.0007,
+      "step": 16108
+    },
+    {
+      "epoch": 68.85,
+      "learning_rate": 9.155353971970466e-05,
+      "loss": 0.0001,
+      "step": 16112
+    },
+    {
+      "epoch": 68.87,
+      "learning_rate": 9.15470568554442e-05,
+      "loss": 0.0042,
+      "step": 16116
+    },
+    {
+      "epoch": 68.89,
+      "learning_rate": 9.154057173396842e-05,
+      "loss": 0.0009,
+      "step": 16120
+    },
+    {
+      "epoch": 68.91,
+      "learning_rate": 9.153408435562961e-05,
+      "loss": 0.0009,
+      "step": 16124
+    },
+    {
+      "epoch": 68.92,
+      "learning_rate": 9.152759472078027e-05,
+      "loss": 0.0022,
+      "step": 16128
+    },
+    {
+      "epoch": 68.94,
+      "learning_rate": 9.152110282977293e-05,
+      "loss": 0.0005,
+      "step": 16132
+    },
+    {
+      "epoch": 68.96,
+      "learning_rate": 9.151460868296034e-05,
+      "loss": 0.0013,
+      "step": 16136
+    },
+    {
+      "epoch": 68.97,
+      "learning_rate": 9.150811228069527e-05,
+      "loss": 0.0002,
+      "step": 16140
+    },
+    {
+      "epoch": 68.99,
+      "learning_rate": 9.15016136233307e-05,
+      "loss": 0.0009,
+      "step": 16144
+    },
+    {
+      "epoch": 69.01,
+      "learning_rate": 9.149511271121967e-05,
+      "loss": 0.0008,
+      "step": 16148
+    },
+    {
+      "epoch": 69.03,
+      "learning_rate": 9.14886095447154e-05,
+      "loss": 0.0001,
+      "step": 16152
+    },
+    {
+      "epoch": 69.04,
+      "learning_rate": 9.148210412417117e-05,
+      "loss": 0.0003,
+      "step": 16156
+    },
+    {
+      "epoch": 69.06,
+      "learning_rate": 9.147559644994044e-05,
+      "loss": 0.0031,
+      "step": 16160
+    },
+    {
+      "epoch": 69.08,
+      "learning_rate": 9.146908652237676e-05,
+      "loss": 0.0001,
+      "step": 16164
+    },
+    {
+      "epoch": 69.09,
+      "learning_rate": 9.146257434183379e-05,
+      "loss": 0.0004,
+      "step": 16168
+    },
+    {
+      "epoch": 69.11,
+      "learning_rate": 9.145605990866534e-05,
+      "loss": 0.0001,
+      "step": 16172
+    },
+    {
+      "epoch": 69.13,
+      "learning_rate": 9.144954322322534e-05,
+      "loss": 0.0001,
+      "step": 16176
+    },
+    {
+      "epoch": 69.15,
+      "learning_rate": 9.144302428586783e-05,
+      "loss": 0.0004,
+      "step": 16180
+    },
+    {
+      "epoch": 69.16,
+      "learning_rate": 9.143650309694699e-05,
+      "loss": 0.0001,
+      "step": 16184
+    },
+    {
+      "epoch": 69.18,
+      "learning_rate": 9.14299796568171e-05,
+      "loss": 0.0002,
+      "step": 16188
+    },
+    {
+      "epoch": 69.2,
+      "learning_rate": 9.142345396583257e-05,
+      "loss": 0.0007,
+      "step": 16192
+    },
+    {
+      "epoch": 69.21,
+      "learning_rate": 9.141692602434795e-05,
+      "loss": 0.0001,
+      "step": 16196
+    },
+    {
+      "epoch": 69.23,
+      "learning_rate": 9.141039583271787e-05,
+      "loss": 0.0001,
+      "step": 16200
+    },
+    {
+      "epoch": 69.25,
+      "learning_rate": 9.140386339129714e-05,
+      "loss": 0.0018,
+      "step": 16204
+    },
+    {
+      "epoch": 69.26,
+      "learning_rate": 9.139732870044064e-05,
+      "loss": 0.0001,
+      "step": 16208
+    },
+    {
+      "epoch": 69.28,
+      "learning_rate": 9.139079176050341e-05,
+      "loss": 0.0004,
+      "step": 16212
+    },
+    {
+      "epoch": 69.3,
+      "learning_rate": 9.138425257184059e-05,
+      "loss": 0.0019,
+      "step": 16216
+    },
+    {
+      "epoch": 69.32,
+      "learning_rate": 9.137771113480744e-05,
+      "loss": 0.0013,
+      "step": 16220
+    },
+    {
+      "epoch": 69.33,
+      "learning_rate": 9.137116744975937e-05,
+      "loss": 0.0011,
+      "step": 16224
+    },
+    {
+      "epoch": 69.35,
+      "learning_rate": 9.136462151705187e-05,
+      "loss": 0.0002,
+      "step": 16228
+    },
+    {
+      "epoch": 69.37,
+      "learning_rate": 9.135807333704058e-05,
+      "loss": 0.0001,
+      "step": 16232
+    },
+    {
+      "epoch": 69.38,
+      "learning_rate": 9.135152291008128e-05,
+      "loss": 0.0001,
+      "step": 16236
+    },
+    {
+      "epoch": 69.4,
+      "learning_rate": 9.134497023652982e-05,
+      "loss": 0.0002,
+      "step": 16240
+    },
+    {
+      "epoch": 69.42,
+      "learning_rate": 9.13384153167422e-05,
+      "loss": 0.0001,
+      "step": 16244
+    },
+    {
+      "epoch": 69.44,
+      "learning_rate": 9.133185815107458e-05,
+      "loss": 0.0001,
+      "step": 16248
+    },
+    {
+      "epoch": 69.45,
+      "learning_rate": 9.132529873988317e-05,
+      "loss": 0.0,
+      "step": 16252
+    },
+    {
+      "epoch": 69.47,
+      "learning_rate": 9.131873708352434e-05,
+      "loss": 0.0017,
+      "step": 16256
+    },
+    {
+      "epoch": 69.49,
+      "learning_rate": 9.13121731823546e-05,
+      "loss": 0.0003,
+      "step": 16260
+    },
+    {
+      "epoch": 69.5,
+      "learning_rate": 9.130560703673054e-05,
+      "loss": 0.0002,
+      "step": 16264
+    },
+    {
+      "epoch": 69.52,
+      "learning_rate": 9.12990386470089e-05,
+      "loss": 0.0001,
+      "step": 16268
+    },
+    {
+      "epoch": 69.54,
+      "learning_rate": 9.129246801354654e-05,
+      "loss": 0.0002,
+      "step": 16272
+    },
+    {
+      "epoch": 69.56,
+      "learning_rate": 9.128589513670043e-05,
+      "loss": 0.0001,
+      "step": 16276
+    },
+    {
+      "epoch": 69.57,
+      "learning_rate": 9.127932001682766e-05,
+      "loss": 0.0003,
+      "step": 16280
+    },
+    {
+      "epoch": 69.59,
+      "learning_rate": 9.127274265428548e-05,
+      "loss": 0.0001,
+      "step": 16284
+    },
+    {
+      "epoch": 69.61,
+      "learning_rate": 9.12661630494312e-05,
+      "loss": 0.0004,
+      "step": 16288
+    },
+    {
+      "epoch": 69.62,
+      "learning_rate": 9.12595812026223e-05,
+      "loss": 0.0006,
+      "step": 16292
+    },
+    {
+      "epoch": 69.64,
+      "learning_rate": 9.125299711421638e-05,
+      "loss": 0.0001,
+      "step": 16296
+    },
+    {
+      "epoch": 69.66,
+      "learning_rate": 9.12464107845711e-05,
+      "loss": 0.0002,
+      "step": 16300
+    },
+    {
+      "epoch": 69.68,
+      "learning_rate": 9.123982221404435e-05,
+      "loss": 0.0001,
+      "step": 16304
+    },
+    {
+      "epoch": 69.69,
+      "learning_rate": 9.123323140299404e-05,
+      "loss": 0.0002,
+      "step": 16308
+    },
+    {
+      "epoch": 69.71,
+      "learning_rate": 9.122663835177825e-05,
+      "loss": 0.001,
+      "step": 16312
+    },
+    {
+      "epoch": 69.73,
+      "learning_rate": 9.12200430607552e-05,
+      "loss": 0.0004,
+      "step": 16316
+    },
+    {
+      "epoch": 69.74,
+      "learning_rate": 9.121344553028317e-05,
+      "loss": 0.0001,
+      "step": 16320
+    },
+    {
+      "epoch": 69.76,
+      "learning_rate": 9.120684576072062e-05,
+      "loss": 0.0017,
+      "step": 16324
+    },
+    {
+      "epoch": 69.78,
+      "learning_rate": 9.12002437524261e-05,
+      "loss": 0.0008,
+      "step": 16328
+    },
+    {
+      "epoch": 69.79,
+      "learning_rate": 9.11936395057583e-05,
+      "loss": 0.0001,
+      "step": 16332
+    },
+    {
+      "epoch": 69.81,
+      "learning_rate": 9.118703302107603e-05,
+      "loss": 0.0006,
+      "step": 16336
+    },
+    {
+      "epoch": 69.83,
+      "learning_rate": 9.118042429873818e-05,
+      "loss": 0.0005,
+      "step": 16340
+    },
+    {
+      "epoch": 69.85,
+      "learning_rate": 9.117381333910385e-05,
+      "loss": 0.0001,
+      "step": 16344
+    },
+    {
+      "epoch": 69.86,
+      "learning_rate": 9.116720014253215e-05,
+      "loss": 0.0001,
+      "step": 16348
+    },
+    {
+      "epoch": 69.88,
+      "learning_rate": 9.116058470938241e-05,
+      "loss": 0.0021,
+      "step": 16352
+    },
+    {
+      "epoch": 69.9,
+      "learning_rate": 9.115396704001401e-05,
+      "loss": 0.0004,
+      "step": 16356
+    },
+    {
+      "epoch": 69.91,
+      "learning_rate": 9.11473471347865e-05,
+      "loss": 0.0001,
+      "step": 16360
+    },
+    {
+      "epoch": 69.93,
+      "learning_rate": 9.114072499405955e-05,
+      "loss": 0.0002,
+      "step": 16364
+    },
+    {
+      "epoch": 69.95,
+      "learning_rate": 9.113410061819292e-05,
+      "loss": 0.0001,
+      "step": 16368
+    },
+    {
+      "epoch": 69.97,
+      "learning_rate": 9.112747400754648e-05,
+      "loss": 0.0001,
+      "step": 16372
+    },
+    {
+      "epoch": 69.98,
+      "learning_rate": 9.112084516248029e-05,
+      "loss": 0.0002,
+      "step": 16376
+    },
+    {
+      "epoch": 70.0,
+      "learning_rate": 9.111421408335447e-05,
+      "loss": 0.0002,
+      "step": 16380
+    },
+    {
+      "epoch": 70.02,
+      "learning_rate": 9.11075807705293e-05,
+      "loss": 0.0007,
+      "step": 16384
+    },
+    {
+      "epoch": 70.03,
+      "learning_rate": 9.110094522436512e-05,
+      "loss": 0.0002,
+      "step": 16388
+    },
+    {
+      "epoch": 70.05,
+      "learning_rate": 9.109430744522247e-05,
+      "loss": 0.0017,
+      "step": 16392
+    },
+    {
+      "epoch": 70.07,
+      "learning_rate": 9.108766743346195e-05,
+      "loss": 0.0003,
+      "step": 16396
+    },
+    {
+      "epoch": 70.09,
+      "learning_rate": 9.108102518944433e-05,
+      "loss": 0.0001,
+      "step": 16400
+    },
+    {
+      "epoch": 70.1,
+      "learning_rate": 9.107438071353047e-05,
+      "loss": 0.0004,
+      "step": 16404
+    },
+    {
+      "epoch": 70.12,
+      "learning_rate": 9.106773400608135e-05,
+      "loss": 0.0003,
+      "step": 16408
+    },
+    {
+      "epoch": 70.14,
+      "learning_rate": 9.106108506745807e-05,
+      "loss": 0.0002,
+      "step": 16412
+    },
+    {
+      "epoch": 70.15,
+      "learning_rate": 9.105443389802191e-05,
+      "loss": 0.0003,
+      "step": 16416
+    },
+    {
+      "epoch": 70.17,
+      "learning_rate": 9.104778049813417e-05,
+      "loss": 0.0001,
+      "step": 16420
+    },
+    {
+      "epoch": 70.19,
+      "learning_rate": 9.104112486815633e-05,
+      "loss": 0.0019,
+      "step": 16424
+    },
+    {
+      "epoch": 70.21,
+      "learning_rate": 9.103446700845e-05,
+      "loss": 0.0001,
+      "step": 16428
+    },
+    {
+      "epoch": 70.22,
+      "learning_rate": 9.102780691937689e-05,
+      "loss": 0.0015,
+      "step": 16432
+    },
+    {
+      "epoch": 70.24,
+      "learning_rate": 9.102114460129885e-05,
+      "loss": 0.0001,
+      "step": 16436
+    },
+    {
+      "epoch": 70.26,
+      "learning_rate": 9.101448005457783e-05,
+      "loss": 0.0002,
+      "step": 16440
+    },
+    {
+      "epoch": 70.27,
+      "learning_rate": 9.100781327957588e-05,
+      "loss": 0.0001,
+      "step": 16444
+    },
+    {
+      "epoch": 70.29,
+      "learning_rate": 9.100114427665524e-05,
+      "loss": 0.0001,
+      "step": 16448
+    },
+    {
+      "epoch": 70.31,
+      "learning_rate": 9.099447304617823e-05,
+      "loss": 0.0001,
+      "step": 16452
+    },
+    {
+      "epoch": 70.32,
+      "learning_rate": 9.098779958850728e-05,
+      "loss": 0.0001,
+      "step": 16456
+    },
+    {
+      "epoch": 70.34,
+      "learning_rate": 9.098112390400492e-05,
+      "loss": 0.0012,
+      "step": 16460
+    },
+    {
+      "epoch": 70.36,
+      "learning_rate": 9.097444599303389e-05,
+      "loss": 0.0002,
+      "step": 16464
+    },
+    {
+      "epoch": 70.38,
+      "learning_rate": 9.096776585595697e-05,
+      "loss": 0.0001,
+      "step": 16468
+    },
+    {
+      "epoch": 70.39,
+      "learning_rate": 9.096108349313707e-05,
+      "loss": 0.0006,
+      "step": 16472
+    },
+    {
+      "epoch": 70.41,
+      "learning_rate": 9.095439890493728e-05,
+      "loss": 0.0002,
+      "step": 16476
+    },
+    {
+      "epoch": 70.43,
+      "learning_rate": 9.094771209172072e-05,
+      "loss": 0.0003,
+      "step": 16480
+    },
+    {
+      "epoch": 70.44,
+      "learning_rate": 9.094102305385072e-05,
+      "loss": 0.0003,
+      "step": 16484
+    },
+    {
+      "epoch": 70.46,
+      "learning_rate": 9.093433179169065e-05,
+      "loss": 0.0002,
+      "step": 16488
+    },
+    {
+      "epoch": 70.48,
+      "learning_rate": 9.092763830560407e-05,
+      "loss": 0.0002,
+      "step": 16492
+    },
+    {
+      "epoch": 70.5,
+      "learning_rate": 9.092094259595462e-05,
+      "loss": 0.0001,
+      "step": 16496
+    },
+    {
+      "epoch": 70.51,
+      "learning_rate": 9.091424466310608e-05,
+      "loss": 0.0002,
+      "step": 16500
+    },
+    {
+      "epoch": 70.53,
+      "learning_rate": 9.090754450742233e-05,
+      "loss": 0.0001,
+      "step": 16504
+    },
+    {
+      "epoch": 70.55,
+      "learning_rate": 9.090084212926739e-05,
+      "loss": 0.0001,
+      "step": 16508
+    },
+    {
+      "epoch": 70.56,
+      "learning_rate": 9.08941375290054e-05,
+      "loss": 0.0002,
+      "step": 16512
+    },
+    {
+      "epoch": 70.58,
+      "learning_rate": 9.088743070700061e-05,
+      "loss": 0.0001,
+      "step": 16516
+    },
+    {
+      "epoch": 70.6,
+      "learning_rate": 9.088072166361738e-05,
+      "loss": 0.0001,
+      "step": 16520
+    },
+    {
+      "epoch": 70.62,
+      "learning_rate": 9.087401039922025e-05,
+      "loss": 0.0002,
+      "step": 16524
+    },
+    {
+      "epoch": 70.63,
+      "learning_rate": 9.08672969141738e-05,
+      "loss": 0.0001,
+      "step": 16528
+    },
+    {
+      "epoch": 70.65,
+      "learning_rate": 9.086058120884275e-05,
+      "loss": 0.0005,
+      "step": 16532
+    },
+    {
+      "epoch": 70.67,
+      "learning_rate": 9.085386328359202e-05,
+      "loss": 0.0001,
+      "step": 16536
+    },
+    {
+      "epoch": 70.68,
+      "learning_rate": 9.084714313878653e-05,
+      "loss": 0.0007,
+      "step": 16540
+    },
+    {
+      "epoch": 70.7,
+      "learning_rate": 9.084042077479142e-05,
+      "loss": 0.0003,
+      "step": 16544
+    },
+    {
+      "epoch": 70.72,
+      "learning_rate": 9.083369619197189e-05,
+      "loss": 0.0,
+      "step": 16548
+    },
+    {
+      "epoch": 70.74,
+      "learning_rate": 9.082696939069329e-05,
+      "loss": 0.0001,
+      "step": 16552
+    },
+    {
+      "epoch": 70.75,
+      "learning_rate": 9.082024037132106e-05,
+      "loss": 0.0005,
+      "step": 16556
+    },
+    {
+      "epoch": 70.77,
+      "learning_rate": 9.08135091342208e-05,
+      "loss": 0.0002,
+      "step": 16560
+    },
+    {
+      "epoch": 70.79,
+      "learning_rate": 9.080677567975823e-05,
+      "loss": 0.0002,
+      "step": 16564
+    },
+    {
+      "epoch": 70.8,
+      "learning_rate": 9.080004000829913e-05,
+      "loss": 0.0002,
+      "step": 16568
+    },
+    {
+      "epoch": 70.82,
+      "learning_rate": 9.079330212020949e-05,
+      "loss": 0.0001,
+      "step": 16572
+    },
+    {
+      "epoch": 70.84,
+      "learning_rate": 9.078656201585533e-05,
+      "loss": 0.0001,
+      "step": 16576
+    },
+    {
+      "epoch": 70.85,
+      "learning_rate": 9.077981969560285e-05,
+      "loss": 0.0012,
+      "step": 16580
+    },
+    {
+      "epoch": 70.87,
+      "learning_rate": 9.077307515981837e-05,
+      "loss": 0.0005,
+      "step": 16584
+    },
+    {
+      "epoch": 70.89,
+      "learning_rate": 9.07663284088683e-05,
+      "loss": 0.0001,
+      "step": 16588
+    },
+    {
+      "epoch": 70.91,
+      "learning_rate": 9.07595794431192e-05,
+      "loss": 0.0007,
+      "step": 16592
+    },
+    {
+      "epoch": 70.92,
+      "learning_rate": 9.075282826293769e-05,
+      "loss": 0.0006,
+      "step": 16596
+    },
+    {
+      "epoch": 70.94,
+      "learning_rate": 9.074607486869062e-05,
+      "loss": 0.0001,
+      "step": 16600
+    },
+    {
+      "epoch": 70.96,
+      "learning_rate": 9.073931926074486e-05,
+      "loss": 0.0003,
+      "step": 16604
+    },
+    {
+      "epoch": 70.97,
+      "learning_rate": 9.073256143946742e-05,
+      "loss": 0.0008,
+      "step": 16608
+    },
+    {
+      "epoch": 70.99,
+      "learning_rate": 9.072580140522549e-05,
+      "loss": 0.0003,
+      "step": 16612
+    },
+    {
+      "epoch": 71.01,
+      "learning_rate": 9.071903915838631e-05,
+      "loss": 0.0003,
+      "step": 16616
+    },
+    {
+      "epoch": 71.03,
+      "learning_rate": 9.071227469931727e-05,
+      "loss": 0.0003,
+      "step": 16620
+    },
+    {
+      "epoch": 71.04,
+      "learning_rate": 9.070550802838587e-05,
+      "loss": 0.0001,
+      "step": 16624
+    },
+    {
+      "epoch": 71.06,
+      "learning_rate": 9.069873914595975e-05,
+      "loss": 0.0001,
+      "step": 16628
+    },
+    {
+      "epoch": 71.08,
+      "learning_rate": 9.069196805240666e-05,
+      "loss": 0.0016,
+      "step": 16632
+    },
+    {
+      "epoch": 71.09,
+      "learning_rate": 9.068519474809446e-05,
+      "loss": 0.0001,
+      "step": 16636
+    },
+    {
+      "epoch": 71.11,
+      "learning_rate": 9.067841923339113e-05,
+      "loss": 0.0001,
+      "step": 16640
+    },
+    {
+      "epoch": 71.13,
+      "learning_rate": 9.06716415086648e-05,
+      "loss": 0.0001,
+      "step": 16644
+    },
+    {
+      "epoch": 71.15,
+      "learning_rate": 9.066486157428368e-05,
+      "loss": 0.0003,
+      "step": 16648
+    },
+    {
+      "epoch": 71.16,
+      "learning_rate": 9.065807943061612e-05,
+      "loss": 0.0004,
+      "step": 16652
+    },
+    {
+      "epoch": 71.18,
+      "learning_rate": 9.065129507803058e-05,
+      "loss": 0.0003,
+      "step": 16656
+    },
+    {
+      "epoch": 71.2,
+      "learning_rate": 9.064450851689567e-05,
+      "loss": 0.0007,
+      "step": 16660
+    },
+    {
+      "epoch": 71.21,
+      "learning_rate": 9.063771974758009e-05,
+      "loss": 0.0005,
+      "step": 16664
+    },
+    {
+      "epoch": 71.23,
+      "learning_rate": 9.063092877045265e-05,
+      "loss": 0.0034,
+      "step": 16668
+    },
+    {
+      "epoch": 71.25,
+      "learning_rate": 9.062413558588232e-05,
+      "loss": 0.0002,
+      "step": 16672
+    },
+    {
+      "epoch": 71.26,
+      "learning_rate": 9.061734019423817e-05,
+      "loss": 0.0001,
+      "step": 16676
+    },
+    {
+      "epoch": 71.28,
+      "learning_rate": 9.061054259588936e-05,
+      "loss": 0.0001,
+      "step": 16680
+    },
+    {
+      "epoch": 71.3,
+      "learning_rate": 9.060374279120523e-05,
+      "loss": 0.0001,
+      "step": 16684
+    },
+    {
+      "epoch": 71.32,
+      "learning_rate": 9.059694078055517e-05,
+      "loss": 0.0001,
+      "step": 16688
+    },
+    {
+      "epoch": 71.33,
+      "learning_rate": 9.059013656430877e-05,
+      "loss": 0.0001,
+      "step": 16692
+    },
+    {
+      "epoch": 71.35,
+      "learning_rate": 9.058333014283566e-05,
+      "loss": 0.0001,
+      "step": 16696
+    },
+    {
+      "epoch": 71.37,
+      "learning_rate": 9.057652151650566e-05,
+      "loss": 0.0013,
+      "step": 16700
+    },
+    {
+      "epoch": 71.38,
+      "learning_rate": 9.056971068568866e-05,
+      "loss": 0.0001,
+      "step": 16704
+    },
+    {
+      "epoch": 71.4,
+      "learning_rate": 9.056289765075468e-05,
+      "loss": 0.0003,
+      "step": 16708
+    },
+    {
+      "epoch": 71.42,
+      "learning_rate": 9.055608241207387e-05,
+      "loss": 0.0003,
+      "step": 16712
+    },
+    {
+      "epoch": 71.44,
+      "learning_rate": 9.054926497001651e-05,
+      "loss": 0.0001,
+      "step": 16716
+    },
+    {
+      "epoch": 71.45,
+      "learning_rate": 9.054244532495297e-05,
+      "loss": 0.0001,
+      "step": 16720
+    },
+    {
+      "epoch": 71.47,
+      "learning_rate": 9.053562347725377e-05,
+      "loss": 0.0003,
+      "step": 16724
+    },
+    {
+      "epoch": 71.49,
+      "learning_rate": 9.05287994272895e-05,
+      "loss": 0.0001,
+      "step": 16728
+    },
+    {
+      "epoch": 71.5,
+      "learning_rate": 9.052197317543096e-05,
+      "loss": 0.0008,
+      "step": 16732
+    },
+    {
+      "epoch": 71.52,
+      "learning_rate": 9.051514472204896e-05,
+      "loss": 0.002,
+      "step": 16736
+    },
+    {
+      "epoch": 71.54,
+      "learning_rate": 9.050831406751452e-05,
+      "loss": 0.0001,
+      "step": 16740
+    },
+    {
+      "epoch": 71.56,
+      "learning_rate": 9.050148121219873e-05,
+      "loss": 0.0001,
+      "step": 16744
+    },
+    {
+      "epoch": 71.57,
+      "learning_rate": 9.04946461564728e-05,
+      "loss": 0.0002,
+      "step": 16748
+    },
+    {
+      "epoch": 71.59,
+      "learning_rate": 9.048780890070811e-05,
+      "loss": 0.0007,
+      "step": 16752
+    },
+    {
+      "epoch": 71.61,
+      "learning_rate": 9.048096944527609e-05,
+      "loss": 0.0001,
+      "step": 16756
+    },
+    {
+      "epoch": 71.62,
+      "learning_rate": 9.047412779054834e-05,
+      "loss": 0.0001,
+      "step": 16760
+    },
+    {
+      "epoch": 71.64,
+      "learning_rate": 9.046728393689654e-05,
+      "loss": 0.0007,
+      "step": 16764
+    },
+    {
+      "epoch": 71.66,
+      "learning_rate": 9.046043788469252e-05,
+      "loss": 0.0005,
+      "step": 16768
+    },
+    {
+      "epoch": 71.68,
+      "learning_rate": 9.045358963430824e-05,
+      "loss": 0.0006,
+      "step": 16772
+    },
+    {
+      "epoch": 71.69,
+      "learning_rate": 9.044673918611573e-05,
+      "loss": 0.0002,
+      "step": 16776
+    },
+    {
+      "epoch": 71.71,
+      "learning_rate": 9.043988654048719e-05,
+      "loss": 0.0004,
+      "step": 16780
+    },
+    {
+      "epoch": 71.73,
+      "learning_rate": 9.04330316977949e-05,
+      "loss": 0.0006,
+      "step": 16784
+    },
+    {
+      "epoch": 71.74,
+      "learning_rate": 9.042617465841127e-05,
+      "loss": 0.0009,
+      "step": 16788
+    },
+    {
+      "epoch": 71.76,
+      "learning_rate": 9.041931542270888e-05,
+      "loss": 0.0001,
+      "step": 16792
+    },
+    {
+      "epoch": 71.78,
+      "learning_rate": 9.041245399106036e-05,
+      "loss": 0.0003,
+      "step": 16796
+    },
+    {
+      "epoch": 71.79,
+      "learning_rate": 9.040559036383848e-05,
+      "loss": 0.0001,
+      "step": 16800
+    },
+    {
+      "epoch": 71.81,
+      "learning_rate": 9.039872454141613e-05,
+      "loss": 0.0001,
+      "step": 16804
+    },
+    {
+      "epoch": 71.83,
+      "learning_rate": 9.039185652416635e-05,
+      "loss": 0.0002,
+      "step": 16808
+    },
+    {
+      "epoch": 71.85,
+      "learning_rate": 9.038498631246227e-05,
+      "loss": 0.0003,
+      "step": 16812
+    },
+    {
+      "epoch": 71.86,
+      "learning_rate": 9.037811390667711e-05,
+      "loss": 0.0002,
+      "step": 16816
+    },
+    {
+      "epoch": 71.88,
+      "learning_rate": 9.037123930718426e-05,
+      "loss": 0.0001,
+      "step": 16820
+    },
+    {
+      "epoch": 71.9,
+      "learning_rate": 9.036436251435723e-05,
+      "loss": 0.0001,
+      "step": 16824
+    },
+    {
+      "epoch": 71.91,
+      "learning_rate": 9.03574835285696e-05,
+      "loss": 0.0002,
+      "step": 16828
+    },
+    {
+      "epoch": 71.93,
+      "learning_rate": 9.035060235019513e-05,
+      "loss": 0.0001,
+      "step": 16832
+    },
+    {
+      "epoch": 71.95,
+      "learning_rate": 9.034371897960763e-05,
+      "loss": 0.0011,
+      "step": 16836
+    },
+    {
+      "epoch": 71.97,
+      "learning_rate": 9.033683341718112e-05,
+      "loss": 0.0001,
+      "step": 16840
+    },
+    {
+      "epoch": 71.98,
+      "learning_rate": 9.032994566328963e-05,
+      "loss": 0.0005,
+      "step": 16844
+    },
+    {
+      "epoch": 72.0,
+      "learning_rate": 9.032305571830739e-05,
+      "loss": 0.0001,
+      "step": 16848
+    },
+    {
+      "epoch": 72.02,
+      "learning_rate": 9.031616358260873e-05,
+      "loss": 0.0006,
+      "step": 16852
+    },
+    {
+      "epoch": 72.03,
+      "learning_rate": 9.03092692565681e-05,
+      "loss": 0.0006,
+      "step": 16856
+    },
+    {
+      "epoch": 72.05,
+      "learning_rate": 9.030237274056003e-05,
+      "loss": 0.0001,
+      "step": 16860
+    },
+    {
+      "epoch": 72.07,
+      "learning_rate": 9.029547403495925e-05,
+      "loss": 0.0001,
+      "step": 16864
+    },
+    {
+      "epoch": 72.09,
+      "learning_rate": 9.028857314014052e-05,
+      "loss": 0.0001,
+      "step": 16868
+    },
+    {
+      "epoch": 72.1,
+      "learning_rate": 9.028167005647878e-05,
+      "loss": 0.0001,
+      "step": 16872
+    },
+    {
+      "epoch": 72.12,
+      "learning_rate": 9.027476478434906e-05,
+      "loss": 0.001,
+      "step": 16876
+    },
+    {
+      "epoch": 72.14,
+      "learning_rate": 9.026785732412652e-05,
+      "loss": 0.0002,
+      "step": 16880
+    },
+    {
+      "epoch": 72.15,
+      "learning_rate": 9.026094767618645e-05,
+      "loss": 0.0001,
+      "step": 16884
+    },
+    {
+      "epoch": 72.17,
+      "learning_rate": 9.025403584090421e-05,
+      "loss": 0.0007,
+      "step": 16888
+    },
+    {
+      "epoch": 72.19,
+      "learning_rate": 9.024712181865536e-05,
+      "loss": 0.0001,
+      "step": 16892
+    },
+    {
+      "epoch": 72.21,
+      "learning_rate": 9.02402056098155e-05,
+      "loss": 0.0001,
+      "step": 16896
+    },
+    {
+      "epoch": 72.22,
+      "learning_rate": 9.023328721476037e-05,
+      "loss": 0.0006,
+      "step": 16900
+    },
+    {
+      "epoch": 72.24,
+      "learning_rate": 9.022636663386587e-05,
+      "loss": 0.004,
+      "step": 16904
+    },
+    {
+      "epoch": 72.26,
+      "learning_rate": 9.0219443867508e-05,
+      "loss": 0.0001,
+      "step": 16908
+    },
+    {
+      "epoch": 72.27,
+      "learning_rate": 9.021251891606284e-05,
+      "loss": 0.0001,
+      "step": 16912
+    },
+    {
+      "epoch": 72.29,
+      "learning_rate": 9.020559177990662e-05,
+      "loss": 0.0001,
+      "step": 16916
+    },
+    {
+      "epoch": 72.31,
+      "learning_rate": 9.019866245941569e-05,
+      "loss": 0.0001,
+      "step": 16920
+    },
+    {
+      "epoch": 72.32,
+      "learning_rate": 9.01917309549665e-05,
+      "loss": 0.0001,
+      "step": 16924
+    },
+    {
+      "epoch": 72.34,
+      "learning_rate": 9.018479726693565e-05,
+      "loss": 0.0001,
+      "step": 16928
+    },
+    {
+      "epoch": 72.36,
+      "learning_rate": 9.017786139569985e-05,
+      "loss": 0.0001,
+      "step": 16932
+    },
+    {
+      "epoch": 72.38,
+      "learning_rate": 9.017092334163591e-05,
+      "loss": 0.0001,
+      "step": 16936
+    },
+    {
+      "epoch": 72.39,
+      "learning_rate": 9.016398310512075e-05,
+      "loss": 0.0001,
+      "step": 16940
+    },
+    {
+      "epoch": 72.41,
+      "learning_rate": 9.015704068653144e-05,
+      "loss": 0.0002,
+      "step": 16944
+    },
+    {
+      "epoch": 72.43,
+      "learning_rate": 9.015009608624516e-05,
+      "loss": 0.0001,
+      "step": 16948
+    },
+    {
+      "epoch": 72.44,
+      "learning_rate": 9.01431493046392e-05,
+      "loss": 0.0001,
+      "step": 16952
+    },
+    {
+      "epoch": 72.46,
+      "learning_rate": 9.0136200342091e-05,
+      "loss": 0.0006,
+      "step": 16956
+    },
+    {
+      "epoch": 72.48,
+      "learning_rate": 9.012924919897805e-05,
+      "loss": 0.0003,
+      "step": 16960
+    },
+    {
+      "epoch": 72.5,
+      "learning_rate": 9.0122295875678e-05,
+      "loss": 0.0002,
+      "step": 16964
+    },
+    {
+      "epoch": 72.51,
+      "learning_rate": 9.011534037256866e-05,
+      "loss": 0.0002,
+      "step": 16968
+    },
+    {
+      "epoch": 72.53,
+      "learning_rate": 9.010838269002787e-05,
+      "loss": 0.0004,
+      "step": 16972
+    },
+    {
+      "epoch": 72.55,
+      "learning_rate": 9.010142282843365e-05,
+      "loss": 0.0009,
+      "step": 16976
+    },
+    {
+      "epoch": 72.56,
+      "learning_rate": 9.009446078816414e-05,
+      "loss": 0.0002,
+      "step": 16980
+    },
+    {
+      "epoch": 72.58,
+      "learning_rate": 9.008749656959758e-05,
+      "loss": 0.0004,
+      "step": 16984
+    },
+    {
+      "epoch": 72.6,
+      "learning_rate": 9.00805301731123e-05,
+      "loss": 0.0,
+      "step": 16988
+    },
+    {
+      "epoch": 72.62,
+      "learning_rate": 9.00735615990868e-05,
+      "loss": 0.0005,
+      "step": 16992
+    },
+    {
+      "epoch": 72.63,
+      "learning_rate": 9.006659084789967e-05,
+      "loss": 0.0002,
+      "step": 16996
+    },
+    {
+      "epoch": 72.65,
+      "learning_rate": 9.005961791992965e-05,
+      "loss": 0.0018,
+      "step": 17000
+    },
+    {
+      "epoch": 72.65,
+      "eval_exact_match": 0.5197505197505198,
+      "eval_loss": 0.90228271484375,
+      "eval_runtime": 140.6077,
+      "eval_samples_per_second": 6.842,
+      "step": 17000
+    },
+    {
+      "epoch": 72.67,
+      "learning_rate": 9.005264281555554e-05,
+      "loss": 0.0001,
+      "step": 17004
+    },
+    {
+      "epoch": 72.68,
+      "learning_rate": 9.004566553515629e-05,
+      "loss": 0.0001,
+      "step": 17008
+    },
+    {
+      "epoch": 72.7,
+      "learning_rate": 9.0038686079111e-05,
+      "loss": 0.0002,
+      "step": 17012
+    },
+    {
+      "epoch": 72.72,
+      "learning_rate": 9.003170444779882e-05,
+      "loss": 0.0006,
+      "step": 17016
+    },
+    {
+      "epoch": 72.74,
+      "learning_rate": 9.00247206415991e-05,
+      "loss": 0.0011,
+      "step": 17020
+    },
+    {
+      "epoch": 72.75,
+      "learning_rate": 9.001773466089123e-05,
+      "loss": 0.0013,
+      "step": 17024
+    },
+    {
+      "epoch": 72.77,
+      "learning_rate": 9.001074650605477e-05,
+      "loss": 0.0001,
+      "step": 17028
+    },
+    {
+      "epoch": 72.79,
+      "learning_rate": 9.000375617746937e-05,
+      "loss": 0.0012,
+      "step": 17032
+    },
+    {
+      "epoch": 72.8,
+      "learning_rate": 8.999676367551479e-05,
+      "loss": 0.0004,
+      "step": 17036
+    },
+    {
+      "epoch": 72.82,
+      "learning_rate": 8.998976900057097e-05,
+      "loss": 0.0001,
+      "step": 17040
+    },
+    {
+      "epoch": 72.84,
+      "learning_rate": 8.99827721530179e-05,
+      "loss": 0.0001,
+      "step": 17044
+    },
+    {
+      "epoch": 72.85,
+      "learning_rate": 8.997577313323571e-05,
+      "loss": 0.0001,
+      "step": 17048
+    },
+    {
+      "epoch": 72.87,
+      "learning_rate": 8.996877194160466e-05,
+      "loss": 0.0002,
+      "step": 17052
+    },
+    {
+      "epoch": 72.89,
+      "learning_rate": 8.996176857850511e-05,
+      "loss": 0.0001,
+      "step": 17056
+    },
+    {
+      "epoch": 72.91,
+      "learning_rate": 8.995476304431756e-05,
+      "loss": 0.0002,
+      "step": 17060
+    },
+    {
+      "epoch": 72.92,
+      "learning_rate": 8.99477553394226e-05,
+      "loss": 0.0008,
+      "step": 17064
+    },
+    {
+      "epoch": 72.94,
+      "learning_rate": 8.994074546420096e-05,
+      "loss": 0.0001,
+      "step": 17068
+    },
+    {
+      "epoch": 72.96,
+      "learning_rate": 8.993373341903348e-05,
+      "loss": 0.0,
+      "step": 17072
+    },
+    {
+      "epoch": 72.97,
+      "learning_rate": 8.992671920430111e-05,
+      "loss": 0.0011,
+      "step": 17076
+    },
+    {
+      "epoch": 72.99,
+      "learning_rate": 8.991970282038493e-05,
+      "loss": 0.0002,
+      "step": 17080
+    },
+    {
+      "epoch": 73.01,
+      "learning_rate": 8.991268426766616e-05,
+      "loss": 0.0002,
+      "step": 17084
+    },
+    {
+      "epoch": 73.03,
+      "learning_rate": 8.990566354652606e-05,
+      "loss": 0.0002,
+      "step": 17088
+    },
+    {
+      "epoch": 73.04,
+      "learning_rate": 8.98986406573461e-05,
+      "loss": 0.0001,
+      "step": 17092
+    },
+    {
+      "epoch": 73.06,
+      "learning_rate": 8.989161560050782e-05,
+      "loss": 0.0015,
+      "step": 17096
+    },
+    {
+      "epoch": 73.08,
+      "learning_rate": 8.988458837639289e-05,
+      "loss": 0.0009,
+      "step": 17100
+    },
+    {
+      "epoch": 73.09,
+      "learning_rate": 8.987755898538307e-05,
+      "loss": 0.0001,
+      "step": 17104
+    },
+    {
+      "epoch": 73.11,
+      "learning_rate": 8.987052742786028e-05,
+      "loss": 0.0001,
+      "step": 17108
+    },
+    {
+      "epoch": 73.13,
+      "learning_rate": 8.986349370420652e-05,
+      "loss": 0.0029,
+      "step": 17112
+    },
+    {
+      "epoch": 73.15,
+      "learning_rate": 8.985645781480396e-05,
+      "loss": 0.0005,
+      "step": 17116
+    },
+    {
+      "epoch": 73.16,
+      "learning_rate": 8.984941976003481e-05,
+      "loss": 0.0,
+      "step": 17120
+    },
+    {
+      "epoch": 73.18,
+      "learning_rate": 8.984237954028148e-05,
+      "loss": 0.0001,
+      "step": 17124
+    },
+    {
+      "epoch": 73.2,
+      "learning_rate": 8.983533715592645e-05,
+      "loss": 0.0002,
+      "step": 17128
+    },
+    {
+      "epoch": 73.21,
+      "learning_rate": 8.982829260735231e-05,
+      "loss": 0.0008,
+      "step": 17132
+    },
+    {
+      "epoch": 73.23,
+      "learning_rate": 8.982124589494178e-05,
+      "loss": 0.0001,
+      "step": 17136
+    },
+    {
+      "epoch": 73.25,
+      "learning_rate": 8.981419701907773e-05,
+      "loss": 0.0015,
+      "step": 17140
+    },
+    {
+      "epoch": 73.26,
+      "learning_rate": 8.980714598014311e-05,
+      "loss": 0.0002,
+      "step": 17144
+    },
+    {
+      "epoch": 73.28,
+      "learning_rate": 8.980009277852099e-05,
+      "loss": 0.0001,
+      "step": 17148
+    },
+    {
+      "epoch": 73.3,
+      "learning_rate": 8.979303741459457e-05,
+      "loss": 0.0001,
+      "step": 17152
+    },
+    {
+      "epoch": 73.32,
+      "learning_rate": 8.978597988874715e-05,
+      "loss": 0.0001,
+      "step": 17156
+    },
+    {
+      "epoch": 73.33,
+      "learning_rate": 8.977892020136216e-05,
+      "loss": 0.0,
+      "step": 17160
+    },
+    {
+      "epoch": 73.35,
+      "learning_rate": 8.977185835282316e-05,
+      "loss": 0.0001,
+      "step": 17164
+    },
+    {
+      "epoch": 73.37,
+      "learning_rate": 8.976479434351382e-05,
+      "loss": 0.0003,
+      "step": 17168
+    },
+    {
+      "epoch": 73.38,
+      "learning_rate": 8.975772817381789e-05,
+      "loss": 0.0001,
+      "step": 17172
+    },
+    {
+      "epoch": 73.4,
+      "learning_rate": 8.97506598441193e-05,
+      "loss": 0.0002,
+      "step": 17176
+    },
+    {
+      "epoch": 73.42,
+      "learning_rate": 8.974358935480205e-05,
+      "loss": 0.0002,
+      "step": 17180
+    },
+    {
+      "epoch": 73.44,
+      "learning_rate": 8.973651670625028e-05,
+      "loss": 0.0001,
+      "step": 17184
+    },
+    {
+      "epoch": 73.45,
+      "learning_rate": 8.972944189884824e-05,
+      "loss": 0.0001,
+      "step": 17188
+    },
+    {
+      "epoch": 73.47,
+      "learning_rate": 8.97223649329803e-05,
+      "loss": 0.0001,
+      "step": 17192
+    },
+    {
+      "epoch": 73.49,
+      "learning_rate": 8.971528580903093e-05,
+      "loss": 0.0001,
+      "step": 17196
+    },
+    {
+      "epoch": 73.5,
+      "learning_rate": 8.970820452738475e-05,
+      "loss": 0.0003,
+      "step": 17200
+    },
+    {
+      "epoch": 73.52,
+      "learning_rate": 8.970112108842649e-05,
+      "loss": 0.0001,
+      "step": 17204
+    },
+    {
+      "epoch": 73.54,
+      "learning_rate": 8.969403549254097e-05,
+      "loss": 0.0001,
+      "step": 17208
+    },
+    {
+      "epoch": 73.56,
+      "learning_rate": 8.968694774011312e-05,
+      "loss": 0.0001,
+      "step": 17212
+    },
+    {
+      "epoch": 73.57,
+      "learning_rate": 8.967985783152805e-05,
+      "loss": 0.0,
+      "step": 17216
+    },
+    {
+      "epoch": 73.59,
+      "learning_rate": 8.967276576717094e-05,
+      "loss": 0.0001,
+      "step": 17220
+    },
+    {
+      "epoch": 73.61,
+      "learning_rate": 8.966567154742709e-05,
+      "loss": 0.0001,
+      "step": 17224
+    },
+    {
+      "epoch": 73.62,
+      "learning_rate": 8.965857517268193e-05,
+      "loss": 0.0,
+      "step": 17228
+    },
+    {
+      "epoch": 73.64,
+      "learning_rate": 8.965147664332099e-05,
+      "loss": 0.0005,
+      "step": 17232
+    },
+    {
+      "epoch": 73.66,
+      "learning_rate": 8.964437595972993e-05,
+      "loss": 0.0001,
+      "step": 17236
+    },
+    {
+      "epoch": 73.68,
+      "learning_rate": 8.963727312229452e-05,
+      "loss": 0.0005,
+      "step": 17240
+    },
+    {
+      "epoch": 73.69,
+      "learning_rate": 8.963016813140066e-05,
+      "loss": 0.0002,
+      "step": 17244
+    },
+    {
+      "epoch": 73.71,
+      "learning_rate": 8.962306098743435e-05,
+      "loss": 0.0003,
+      "step": 17248
+    },
+    {
+      "epoch": 73.73,
+      "learning_rate": 8.961595169078172e-05,
+      "loss": 0.0001,
+      "step": 17252
+    },
+    {
+      "epoch": 73.74,
+      "learning_rate": 8.960884024182902e-05,
+      "loss": 0.0001,
+      "step": 17256
+    },
+    {
+      "epoch": 73.76,
+      "learning_rate": 8.96017266409626e-05,
+      "loss": 0.0003,
+      "step": 17260
+    },
+    {
+      "epoch": 73.78,
+      "learning_rate": 8.959461088856893e-05,
+      "loss": 0.0,
+      "step": 17264
+    },
+    {
+      "epoch": 73.79,
+      "learning_rate": 8.95874929850346e-05,
+      "loss": 0.0,
+      "step": 17268
+    },
+    {
+      "epoch": 73.81,
+      "learning_rate": 8.958037293074634e-05,
+      "loss": 0.0004,
+      "step": 17272
+    },
+    {
+      "epoch": 73.83,
+      "learning_rate": 8.957325072609095e-05,
+      "loss": 0.0001,
+      "step": 17276
+    },
+    {
+      "epoch": 73.85,
+      "learning_rate": 8.95661263714554e-05,
+      "loss": 0.0001,
+      "step": 17280
+    },
+    {
+      "epoch": 73.86,
+      "learning_rate": 8.955899986722673e-05,
+      "loss": 0.001,
+      "step": 17284
+    },
+    {
+      "epoch": 73.88,
+      "learning_rate": 8.955187121379213e-05,
+      "loss": 0.0003,
+      "step": 17288
+    },
+    {
+      "epoch": 73.9,
+      "learning_rate": 8.954474041153889e-05,
+      "loss": 0.0001,
+      "step": 17292
+    },
+    {
+      "epoch": 73.91,
+      "learning_rate": 8.95376074608544e-05,
+      "loss": 0.0007,
+      "step": 17296
+    },
+    {
+      "epoch": 73.93,
+      "learning_rate": 8.953047236212621e-05,
+      "loss": 0.0001,
+      "step": 17300
+    },
+    {
+      "epoch": 73.95,
+      "learning_rate": 8.952333511574196e-05,
+      "loss": 0.0001,
+      "step": 17304
+    },
+    {
+      "epoch": 73.97,
+      "learning_rate": 8.951619572208942e-05,
+      "loss": 0.0001,
+      "step": 17308
+    },
+    {
+      "epoch": 73.98,
+      "learning_rate": 8.950905418155644e-05,
+      "loss": 0.0024,
+      "step": 17312
+    },
+    {
+      "epoch": 74.0,
+      "learning_rate": 8.950191049453104e-05,
+      "loss": 0.0003,
+      "step": 17316
+    },
+    {
+      "epoch": 74.02,
+      "learning_rate": 8.949476466140132e-05,
+      "loss": 0.0022,
+      "step": 17320
+    },
+    {
+      "epoch": 74.03,
+      "learning_rate": 8.94876166825555e-05,
+      "loss": 0.0001,
+      "step": 17324
+    },
+    {
+      "epoch": 74.05,
+      "learning_rate": 8.948046655838193e-05,
+      "loss": 0.0001,
+      "step": 17328
+    },
+    {
+      "epoch": 74.07,
+      "learning_rate": 8.947331428926906e-05,
+      "loss": 0.0008,
+      "step": 17332
+    },
+    {
+      "epoch": 74.09,
+      "learning_rate": 8.94661598756055e-05,
+      "loss": 0.0004,
+      "step": 17336
+    },
+    {
+      "epoch": 74.1,
+      "learning_rate": 8.94590033177799e-05,
+      "loss": 0.0002,
+      "step": 17340
+    },
+    {
+      "epoch": 74.12,
+      "learning_rate": 8.945184461618111e-05,
+      "loss": 0.0008,
+      "step": 17344
+    },
+    {
+      "epoch": 74.14,
+      "learning_rate": 8.944468377119801e-05,
+      "loss": 0.0002,
+      "step": 17348
+    },
+    {
+      "epoch": 74.15,
+      "learning_rate": 8.943752078321968e-05,
+      "loss": 0.0005,
+      "step": 17352
+    },
+    {
+      "epoch": 74.17,
+      "learning_rate": 8.943035565263526e-05,
+      "loss": 0.0011,
+      "step": 17356
+    },
+    {
+      "epoch": 74.19,
+      "learning_rate": 8.942318837983403e-05,
+      "loss": 0.0002,
+      "step": 17360
+    },
+    {
+      "epoch": 74.21,
+      "learning_rate": 8.94160189652054e-05,
+      "loss": 0.0031,
+      "step": 17364
+    },
+    {
+      "epoch": 74.22,
+      "learning_rate": 8.940884740913884e-05,
+      "loss": 0.0001,
+      "step": 17368
+    },
+    {
+      "epoch": 74.24,
+      "learning_rate": 8.940167371202401e-05,
+      "loss": 0.0001,
+      "step": 17372
+    },
+    {
+      "epoch": 74.26,
+      "learning_rate": 8.939449787425062e-05,
+      "loss": 0.0028,
+      "step": 17376
+    },
+    {
+      "epoch": 74.27,
+      "learning_rate": 8.938731989620857e-05,
+      "loss": 0.0,
+      "step": 17380
+    },
+    {
+      "epoch": 74.29,
+      "learning_rate": 8.938013977828778e-05,
+      "loss": 0.0,
+      "step": 17384
+    },
+    {
+      "epoch": 74.31,
+      "learning_rate": 8.937295752087839e-05,
+      "loss": 0.0001,
+      "step": 17388
+    },
+    {
+      "epoch": 74.32,
+      "learning_rate": 8.936577312437056e-05,
+      "loss": 0.0016,
+      "step": 17392
+    },
+    {
+      "epoch": 74.34,
+      "learning_rate": 8.935858658915466e-05,
+      "loss": 0.0001,
+      "step": 17396
+    },
+    {
+      "epoch": 74.36,
+      "learning_rate": 8.935139791562109e-05,
+      "loss": 0.0,
+      "step": 17400
+    },
+    {
+      "epoch": 74.38,
+      "learning_rate": 8.93442071041604e-05,
+      "loss": 0.0002,
+      "step": 17404
+    },
+    {
+      "epoch": 74.39,
+      "learning_rate": 8.933701415516329e-05,
+      "loss": 0.0018,
+      "step": 17408
+    },
+    {
+      "epoch": 74.41,
+      "learning_rate": 8.932981906902053e-05,
+      "loss": 0.0,
+      "step": 17412
+    },
+    {
+      "epoch": 74.43,
+      "learning_rate": 8.932262184612304e-05,
+      "loss": 0.0001,
+      "step": 17416
+    },
+    {
+      "epoch": 74.44,
+      "learning_rate": 8.93154224868618e-05,
+      "loss": 0.0001,
+      "step": 17420
+    },
+    {
+      "epoch": 74.46,
+      "learning_rate": 8.930822099162798e-05,
+      "loss": 0.0002,
+      "step": 17424
+    },
+    {
+      "epoch": 74.48,
+      "learning_rate": 8.930101736081284e-05,
+      "loss": 0.0001,
+      "step": 17428
+    },
+    {
+      "epoch": 74.5,
+      "learning_rate": 8.929381159480772e-05,
+      "loss": 0.0001,
+      "step": 17432
+    },
+    {
+      "epoch": 74.51,
+      "learning_rate": 8.92866036940041e-05,
+      "loss": 0.0001,
+      "step": 17436
+    },
+    {
+      "epoch": 74.53,
+      "learning_rate": 8.927939365879359e-05,
+      "loss": 0.0003,
+      "step": 17440
+    },
+    {
+      "epoch": 74.55,
+      "learning_rate": 8.92721814895679e-05,
+      "loss": 0.0001,
+      "step": 17444
+    },
+    {
+      "epoch": 74.56,
+      "learning_rate": 8.926496718671889e-05,
+      "loss": 0.0005,
+      "step": 17448
+    },
+    {
+      "epoch": 74.58,
+      "learning_rate": 8.925775075063847e-05,
+      "loss": 0.0005,
+      "step": 17452
+    },
+    {
+      "epoch": 74.6,
+      "learning_rate": 8.925053218171871e-05,
+      "loss": 0.0001,
+      "step": 17456
+    },
+    {
+      "epoch": 74.62,
+      "learning_rate": 8.924331148035179e-05,
+      "loss": 0.0001,
+      "step": 17460
+    },
+    {
+      "epoch": 74.63,
+      "learning_rate": 8.923608864693003e-05,
+      "loss": 0.0001,
+      "step": 17464
+    },
+    {
+      "epoch": 74.65,
+      "learning_rate": 8.92288636818458e-05,
+      "loss": 0.0001,
+      "step": 17468
+    },
+    {
+      "epoch": 74.67,
+      "learning_rate": 8.922163658549164e-05,
+      "loss": 0.0001,
+      "step": 17472
+    },
+    {
+      "epoch": 74.68,
+      "learning_rate": 8.92144073582602e-05,
+      "loss": 0.0001,
+      "step": 17476
+    },
+    {
+      "epoch": 74.7,
+      "learning_rate": 8.920717600054425e-05,
+      "loss": 0.0005,
+      "step": 17480
+    },
+    {
+      "epoch": 74.72,
+      "learning_rate": 8.919994251273664e-05,
+      "loss": 0.0001,
+      "step": 17484
+    },
+    {
+      "epoch": 74.74,
+      "learning_rate": 8.919270689523036e-05,
+      "loss": 0.0001,
+      "step": 17488
+    },
+    {
+      "epoch": 74.75,
+      "learning_rate": 8.918546914841853e-05,
+      "loss": 0.0001,
+      "step": 17492
+    },
+    {
+      "epoch": 74.77,
+      "learning_rate": 8.917822927269434e-05,
+      "loss": 0.0009,
+      "step": 17496
+    },
+    {
+      "epoch": 74.79,
+      "learning_rate": 8.917098726845117e-05,
+      "loss": 0.0,
+      "step": 17500
+    },
+    {
+      "epoch": 74.8,
+      "learning_rate": 8.916374313608244e-05,
+      "loss": 0.0009,
+      "step": 17504
+    },
+    {
+      "epoch": 74.82,
+      "learning_rate": 8.915649687598174e-05,
+      "loss": 0.0001,
+      "step": 17508
+    },
+    {
+      "epoch": 74.84,
+      "learning_rate": 8.914924848854271e-05,
+      "loss": 0.0001,
+      "step": 17512
+    },
+    {
+      "epoch": 74.85,
+      "learning_rate": 8.914199797415921e-05,
+      "loss": 0.0004,
+      "step": 17516
+    },
+    {
+      "epoch": 74.87,
+      "learning_rate": 8.91347453332251e-05,
+      "loss": 0.0001,
+      "step": 17520
+    },
+    {
+      "epoch": 74.89,
+      "learning_rate": 8.912749056613444e-05,
+      "loss": 0.0023,
+      "step": 17524
+    },
+    {
+      "epoch": 74.91,
+      "learning_rate": 8.912023367328137e-05,
+      "loss": 0.0002,
+      "step": 17528
+    },
+    {
+      "epoch": 74.92,
+      "learning_rate": 8.911297465506015e-05,
+      "loss": 0.0001,
+      "step": 17532
+    },
+    {
+      "epoch": 74.94,
+      "learning_rate": 8.910571351186515e-05,
+      "loss": 0.0003,
+      "step": 17536
+    },
+    {
+      "epoch": 74.96,
+      "learning_rate": 8.909845024409086e-05,
+      "loss": 0.0011,
+      "step": 17540
+    },
+    {
+      "epoch": 74.97,
+      "learning_rate": 8.90911848521319e-05,
+      "loss": 0.0001,
+      "step": 17544
+    },
+    {
+      "epoch": 74.99,
+      "learning_rate": 8.908391733638299e-05,
+      "loss": 0.0001,
+      "step": 17548
+    },
+    {
+      "epoch": 75.01,
+      "learning_rate": 8.907664769723895e-05,
+      "loss": 0.0001,
+      "step": 17552
+    },
+    {
+      "epoch": 75.03,
+      "learning_rate": 8.906937593509476e-05,
+      "loss": 0.0001,
+      "step": 17556
+    },
+    {
+      "epoch": 75.04,
+      "learning_rate": 8.906210205034546e-05,
+      "loss": 0.0003,
+      "step": 17560
+    },
+    {
+      "epoch": 75.06,
+      "learning_rate": 8.905482604338626e-05,
+      "loss": 0.0012,
+      "step": 17564
+    },
+    {
+      "epoch": 75.08,
+      "learning_rate": 8.904754791461245e-05,
+      "loss": 0.0001,
+      "step": 17568
+    },
+    {
+      "epoch": 75.09,
+      "learning_rate": 8.904026766441944e-05,
+      "loss": 0.0002,
+      "step": 17572
+    },
+    {
+      "epoch": 75.11,
+      "learning_rate": 8.903298529320275e-05,
+      "loss": 0.0004,
+      "step": 17576
+    },
+    {
+      "epoch": 75.13,
+      "learning_rate": 8.902570080135805e-05,
+      "loss": 0.0003,
+      "step": 17580
+    },
+    {
+      "epoch": 75.15,
+      "learning_rate": 8.901841418928108e-05,
+      "loss": 0.0,
+      "step": 17584
+    },
+    {
+      "epoch": 75.16,
+      "learning_rate": 8.901112545736771e-05,
+      "loss": 0.0002,
+      "step": 17588
+    },
+    {
+      "epoch": 75.18,
+      "learning_rate": 8.900383460601395e-05,
+      "loss": 0.0002,
+      "step": 17592
+    },
+    {
+      "epoch": 75.2,
+      "learning_rate": 8.899654163561592e-05,
+      "loss": 0.0013,
+      "step": 17596
+    },
+    {
+      "epoch": 75.21,
+      "learning_rate": 8.898924654656979e-05,
+      "loss": 0.0003,
+      "step": 17600
+    },
+    {
+      "epoch": 75.23,
+      "learning_rate": 8.898194933927194e-05,
+      "loss": 0.0001,
+      "step": 17604
+    },
+    {
+      "epoch": 75.25,
+      "learning_rate": 8.89746500141188e-05,
+      "loss": 0.0002,
+      "step": 17608
+    },
+    {
+      "epoch": 75.26,
+      "learning_rate": 8.896734857150695e-05,
+      "loss": 0.0004,
+      "step": 17612
+    },
+    {
+      "epoch": 75.28,
+      "learning_rate": 8.896004501183305e-05,
+      "loss": 0.0003,
+      "step": 17616
+    },
+    {
+      "epoch": 75.3,
+      "learning_rate": 8.895273933549391e-05,
+      "loss": 0.0006,
+      "step": 17620
+    },
+    {
+      "epoch": 75.32,
+      "learning_rate": 8.894543154288644e-05,
+      "loss": 0.0005,
+      "step": 17624
+    },
+    {
+      "epoch": 75.33,
+      "learning_rate": 8.893812163440767e-05,
+      "loss": 0.001,
+      "step": 17628
+    },
+    {
+      "epoch": 75.35,
+      "learning_rate": 8.893080961045472e-05,
+      "loss": 0.0002,
+      "step": 17632
+    },
+    {
+      "epoch": 75.37,
+      "learning_rate": 8.892349547142488e-05,
+      "loss": 0.0001,
+      "step": 17636
+    },
+    {
+      "epoch": 75.38,
+      "learning_rate": 8.891617921771548e-05,
+      "loss": 0.0001,
+      "step": 17640
+    },
+    {
+      "epoch": 75.4,
+      "learning_rate": 8.890886084972406e-05,
+      "loss": 0.001,
+      "step": 17644
+    },
+    {
+      "epoch": 75.42,
+      "learning_rate": 8.890154036784817e-05,
+      "loss": 0.0003,
+      "step": 17648
+    },
+    {
+      "epoch": 75.44,
+      "learning_rate": 8.889421777248556e-05,
+      "loss": 0.0002,
+      "step": 17652
+    },
+    {
+      "epoch": 75.45,
+      "learning_rate": 8.888689306403402e-05,
+      "loss": 0.001,
+      "step": 17656
+    },
+    {
+      "epoch": 75.47,
+      "learning_rate": 8.887956624289154e-05,
+      "loss": 0.0001,
+      "step": 17660
+    },
+    {
+      "epoch": 75.49,
+      "learning_rate": 8.887223730945616e-05,
+      "loss": 0.0001,
+      "step": 17664
+    },
+    {
+      "epoch": 75.5,
+      "learning_rate": 8.886490626412604e-05,
+      "loss": 0.0004,
+      "step": 17668
+    },
+    {
+      "epoch": 75.52,
+      "learning_rate": 8.885757310729948e-05,
+      "loss": 0.0002,
+      "step": 17672
+    },
+    {
+      "epoch": 75.54,
+      "learning_rate": 8.885023783937491e-05,
+      "loss": 0.0007,
+      "step": 17676
+    },
+    {
+      "epoch": 75.56,
+      "learning_rate": 8.88429004607508e-05,
+      "loss": 0.0001,
+      "step": 17680
+    },
+    {
+      "epoch": 75.57,
+      "learning_rate": 8.883556097182582e-05,
+      "loss": 0.0003,
+      "step": 17684
+    },
+    {
+      "epoch": 75.59,
+      "learning_rate": 8.882821937299873e-05,
+      "loss": 0.0004,
+      "step": 17688
+    },
+    {
+      "epoch": 75.61,
+      "learning_rate": 8.882087566466833e-05,
+      "loss": 0.0008,
+      "step": 17692
+    },
+    {
+      "epoch": 75.62,
+      "learning_rate": 8.881352984723365e-05,
+      "loss": 0.0001,
+      "step": 17696
+    },
+    {
+      "epoch": 75.64,
+      "learning_rate": 8.880618192109379e-05,
+      "loss": 0.0001,
+      "step": 17700
+    },
+    {
+      "epoch": 75.66,
+      "learning_rate": 8.879883188664793e-05,
+      "loss": 0.0003,
+      "step": 17704
+    },
+    {
+      "epoch": 75.68,
+      "learning_rate": 8.879147974429538e-05,
+      "loss": 0.0001,
+      "step": 17708
+    },
+    {
+      "epoch": 75.69,
+      "learning_rate": 8.87841254944356e-05,
+      "loss": 0.0013,
+      "step": 17712
+    },
+    {
+      "epoch": 75.71,
+      "learning_rate": 8.877676913746813e-05,
+      "loss": 0.0001,
+      "step": 17716
+    },
+    {
+      "epoch": 75.73,
+      "learning_rate": 8.876941067379264e-05,
+      "loss": 0.0001,
+      "step": 17720
+    },
+    {
+      "epoch": 75.74,
+      "learning_rate": 8.876205010380891e-05,
+      "loss": 0.0002,
+      "step": 17724
+    },
+    {
+      "epoch": 75.76,
+      "learning_rate": 8.875468742791682e-05,
+      "loss": 0.0002,
+      "step": 17728
+    },
+    {
+      "epoch": 75.78,
+      "learning_rate": 8.874732264651639e-05,
+      "loss": 0.0001,
+      "step": 17732
+    },
+    {
+      "epoch": 75.79,
+      "learning_rate": 8.873995576000774e-05,
+      "loss": 0.0004,
+      "step": 17736
+    },
+    {
+      "epoch": 75.81,
+      "learning_rate": 8.87325867687911e-05,
+      "loss": 0.0001,
+      "step": 17740
+    },
+    {
+      "epoch": 75.83,
+      "learning_rate": 8.872521567326683e-05,
+      "loss": 0.0001,
+      "step": 17744
+    },
+    {
+      "epoch": 75.85,
+      "learning_rate": 8.871784247383539e-05,
+      "loss": 0.0002,
+      "step": 17748
+    },
+    {
+      "epoch": 75.86,
+      "learning_rate": 8.871046717089735e-05,
+      "loss": 0.0001,
+      "step": 17752
+    },
+    {
+      "epoch": 75.88,
+      "learning_rate": 8.870308976485344e-05,
+      "loss": 0.0,
+      "step": 17756
+    },
+    {
+      "epoch": 75.9,
+      "learning_rate": 8.869571025610442e-05,
+      "loss": 0.0002,
+      "step": 17760
+    },
+    {
+      "epoch": 75.91,
+      "learning_rate": 8.868832864505125e-05,
+      "loss": 0.0,
+      "step": 17764
+    },
+    {
+      "epoch": 75.93,
+      "learning_rate": 8.868094493209493e-05,
+      "loss": 0.0001,
+      "step": 17768
+    },
+    {
+      "epoch": 75.95,
+      "learning_rate": 8.867355911763667e-05,
+      "loss": 0.0005,
+      "step": 17772
+    },
+    {
+      "epoch": 75.97,
+      "learning_rate": 8.866617120207767e-05,
+      "loss": 0.0002,
+      "step": 17776
+    },
+    {
+      "epoch": 75.98,
+      "learning_rate": 8.865878118581932e-05,
+      "loss": 0.0005,
+      "step": 17780
+    },
+    {
+      "epoch": 76.0,
+      "learning_rate": 8.865138906926316e-05,
+      "loss": 0.0015,
+      "step": 17784
+    },
+    {
+      "epoch": 76.02,
+      "learning_rate": 8.864399485281074e-05,
+      "loss": 0.0001,
+      "step": 17788
+    },
+    {
+      "epoch": 76.03,
+      "learning_rate": 8.863659853686384e-05,
+      "loss": 0.0001,
+      "step": 17792
+    },
+    {
+      "epoch": 76.05,
+      "learning_rate": 8.862920012182423e-05,
+      "loss": 0.0004,
+      "step": 17796
+    },
+    {
+      "epoch": 76.07,
+      "learning_rate": 8.862179960809391e-05,
+      "loss": 0.0004,
+      "step": 17800
+    },
+    {
+      "epoch": 76.09,
+      "learning_rate": 8.861439699607492e-05,
+      "loss": 0.0001,
+      "step": 17804
+    },
+    {
+      "epoch": 76.1,
+      "learning_rate": 8.860699228616945e-05,
+      "loss": 0.0001,
+      "step": 17808
+    },
+    {
+      "epoch": 76.12,
+      "learning_rate": 8.859958547877978e-05,
+      "loss": 0.0006,
+      "step": 17812
+    },
+    {
+      "epoch": 76.14,
+      "learning_rate": 8.859217657430831e-05,
+      "loss": 0.0012,
+      "step": 17816
+    },
+    {
+      "epoch": 76.15,
+      "learning_rate": 8.858476557315758e-05,
+      "loss": 0.0001,
+      "step": 17820
+    },
+    {
+      "epoch": 76.17,
+      "learning_rate": 8.857735247573022e-05,
+      "loss": 0.0005,
+      "step": 17824
+    },
+    {
+      "epoch": 76.19,
+      "learning_rate": 8.856993728242896e-05,
+      "loss": 0.0001,
+      "step": 17828
+    },
+    {
+      "epoch": 76.21,
+      "learning_rate": 8.856251999365666e-05,
+      "loss": 0.0002,
+      "step": 17832
+    },
+    {
+      "epoch": 76.22,
+      "learning_rate": 8.855510060981633e-05,
+      "loss": 0.0001,
+      "step": 17836
+    },
+    {
+      "epoch": 76.24,
+      "learning_rate": 8.8547679131311e-05,
+      "loss": 0.0004,
+      "step": 17840
+    },
+    {
+      "epoch": 76.26,
+      "learning_rate": 8.854025555854395e-05,
+      "loss": 0.0001,
+      "step": 17844
+    },
+    {
+      "epoch": 76.27,
+      "learning_rate": 8.853282989191842e-05,
+      "loss": 0.0001,
+      "step": 17848
+    },
+    {
+      "epoch": 76.29,
+      "learning_rate": 8.852540213183789e-05,
+      "loss": 0.0001,
+      "step": 17852
+    },
+    {
+      "epoch": 76.31,
+      "learning_rate": 8.851797227870589e-05,
+      "loss": 0.001,
+      "step": 17856
+    },
+    {
+      "epoch": 76.32,
+      "learning_rate": 8.851054033292604e-05,
+      "loss": 0.0002,
+      "step": 17860
+    },
+    {
+      "epoch": 76.34,
+      "learning_rate": 8.850310629490218e-05,
+      "loss": 0.0001,
+      "step": 17864
+    },
+    {
+      "epoch": 76.36,
+      "learning_rate": 8.849567016503814e-05,
+      "loss": 0.0005,
+      "step": 17868
+    },
+    {
+      "epoch": 76.38,
+      "learning_rate": 8.848823194373796e-05,
+      "loss": 0.0002,
+      "step": 17872
+    },
+    {
+      "epoch": 76.39,
+      "learning_rate": 8.848079163140573e-05,
+      "loss": 0.0002,
+      "step": 17876
+    },
+    {
+      "epoch": 76.41,
+      "learning_rate": 8.847334922844566e-05,
+      "loss": 0.0001,
+      "step": 17880
+    },
+    {
+      "epoch": 76.43,
+      "learning_rate": 8.846590473526209e-05,
+      "loss": 0.0004,
+      "step": 17884
+    },
+    {
+      "epoch": 76.44,
+      "learning_rate": 8.845845815225951e-05,
+      "loss": 0.0012,
+      "step": 17888
+    },
+    {
+      "epoch": 76.46,
+      "learning_rate": 8.845100947984246e-05,
+      "loss": 0.0001,
+      "step": 17892
+    },
+    {
+      "epoch": 76.48,
+      "learning_rate": 8.844355871841563e-05,
+      "loss": 0.0002,
+      "step": 17896
+    },
+    {
+      "epoch": 76.5,
+      "learning_rate": 8.84361058683838e-05,
+      "loss": 0.0005,
+      "step": 17900
+    },
+    {
+      "epoch": 76.51,
+      "learning_rate": 8.842865093015189e-05,
+      "loss": 0.0001,
+      "step": 17904
+    },
+    {
+      "epoch": 76.53,
+      "learning_rate": 8.84211939041249e-05,
+      "loss": 0.0005,
+      "step": 17908
+    },
+    {
+      "epoch": 76.55,
+      "learning_rate": 8.8413734790708e-05,
+      "loss": 0.0001,
+      "step": 17912
+    },
+    {
+      "epoch": 76.56,
+      "learning_rate": 8.840627359030642e-05,
+      "loss": 0.0012,
+      "step": 17916
+    },
+    {
+      "epoch": 76.58,
+      "learning_rate": 8.83988103033255e-05,
+      "loss": 0.0001,
+      "step": 17920
+    },
+    {
+      "epoch": 76.6,
+      "learning_rate": 8.839134493017074e-05,
+      "loss": 0.0003,
+      "step": 17924
+    },
+    {
+      "epoch": 76.62,
+      "learning_rate": 8.83838774712477e-05,
+      "loss": 0.0002,
+      "step": 17928
+    },
+    {
+      "epoch": 76.63,
+      "learning_rate": 8.837640792696212e-05,
+      "loss": 0.0002,
+      "step": 17932
+    },
+    {
+      "epoch": 76.65,
+      "learning_rate": 8.836893629771977e-05,
+      "loss": 0.0006,
+      "step": 17936
+    },
+    {
+      "epoch": 76.67,
+      "learning_rate": 8.836146258392661e-05,
+      "loss": 0.0004,
+      "step": 17940
+    },
+    {
+      "epoch": 76.68,
+      "learning_rate": 8.835398678598867e-05,
+      "loss": 0.0008,
+      "step": 17944
+    },
+    {
+      "epoch": 76.7,
+      "learning_rate": 8.83465089043121e-05,
+      "loss": 0.0001,
+      "step": 17948
+    },
+    {
+      "epoch": 76.72,
+      "learning_rate": 8.833902893930317e-05,
+      "loss": 0.0006,
+      "step": 17952
+    },
+    {
+      "epoch": 76.74,
+      "learning_rate": 8.833154689136826e-05,
+      "loss": 0.0016,
+      "step": 17956
+    },
+    {
+      "epoch": 76.75,
+      "learning_rate": 8.832406276091386e-05,
+      "loss": 0.0003,
+      "step": 17960
+    },
+    {
+      "epoch": 76.77,
+      "learning_rate": 8.831657654834658e-05,
+      "loss": 0.0006,
+      "step": 17964
+    },
+    {
+      "epoch": 76.79,
+      "learning_rate": 8.830908825407314e-05,
+      "loss": 0.0005,
+      "step": 17968
+    },
+    {
+      "epoch": 76.8,
+      "learning_rate": 8.830159787850036e-05,
+      "loss": 0.0008,
+      "step": 17972
+    },
+    {
+      "epoch": 76.82,
+      "learning_rate": 8.829410542203522e-05,
+      "loss": 0.0011,
+      "step": 17976
+    },
+    {
+      "epoch": 76.84,
+      "learning_rate": 8.828661088508473e-05,
+      "loss": 0.0003,
+      "step": 17980
+    },
+    {
+      "epoch": 76.85,
+      "learning_rate": 8.82791142680561e-05,
+      "loss": 0.0014,
+      "step": 17984
+    },
+    {
+      "epoch": 76.87,
+      "learning_rate": 8.827161557135659e-05,
+      "loss": 0.0001,
+      "step": 17988
+    },
+    {
+      "epoch": 76.89,
+      "learning_rate": 8.826411479539363e-05,
+      "loss": 0.0001,
+      "step": 17992
+    },
+    {
+      "epoch": 76.91,
+      "learning_rate": 8.825661194057469e-05,
+      "loss": 0.001,
+      "step": 17996
+    },
+    {
+      "epoch": 76.92,
+      "learning_rate": 8.824910700730742e-05,
+      "loss": 0.0001,
+      "step": 18000
+    },
+    {
+      "epoch": 76.92,
+      "eval_exact_match": 0.5135135135135135,
+      "eval_loss": 0.9052860736846924,
+      "eval_runtime": 171.5376,
+      "eval_samples_per_second": 5.608,
+      "step": 18000
+    },
+    {
+      "epoch": 76.94,
+      "learning_rate": 8.824159999599955e-05,
+      "loss": 0.0002,
+      "step": 18004
+    },
+    {
+      "epoch": 76.96,
+      "learning_rate": 8.823409090705892e-05,
+      "loss": 0.0014,
+      "step": 18008
+    },
+    {
+      "epoch": 76.97,
+      "learning_rate": 8.822657974089352e-05,
+      "loss": 0.0001,
+      "step": 18012
+    },
+    {
+      "epoch": 76.99,
+      "learning_rate": 8.82190664979114e-05,
+      "loss": 0.0014,
+      "step": 18016
+    },
+    {
+      "epoch": 77.01,
+      "learning_rate": 8.821155117852074e-05,
+      "loss": 0.0003,
+      "step": 18020
+    },
+    {
+      "epoch": 77.03,
+      "learning_rate": 8.820403378312987e-05,
+      "loss": 0.0002,
+      "step": 18024
+    },
+    {
+      "epoch": 77.04,
+      "learning_rate": 8.819651431214717e-05,
+      "loss": 0.0,
+      "step": 18028
+    },
+    {
+      "epoch": 77.06,
+      "learning_rate": 8.81889927659812e-05,
+      "loss": 0.0001,
+      "step": 18032
+    },
+    {
+      "epoch": 77.08,
+      "learning_rate": 8.818146914504058e-05,
+      "loss": 0.0001,
+      "step": 18036
+    },
+    {
+      "epoch": 77.09,
+      "learning_rate": 8.817394344973406e-05,
+      "loss": 0.0001,
+      "step": 18040
+    },
+    {
+      "epoch": 77.11,
+      "learning_rate": 8.816641568047052e-05,
+      "loss": 0.0004,
+      "step": 18044
+    },
+    {
+      "epoch": 77.13,
+      "learning_rate": 8.815888583765893e-05,
+      "loss": 0.0001,
+      "step": 18048
+    },
+    {
+      "epoch": 77.15,
+      "learning_rate": 8.815135392170836e-05,
+      "loss": 0.0001,
+      "step": 18052
+    },
+    {
+      "epoch": 77.16,
+      "learning_rate": 8.814381993302805e-05,
+      "loss": 0.0001,
+      "step": 18056
+    },
+    {
+      "epoch": 77.18,
+      "learning_rate": 8.813628387202728e-05,
+      "loss": 0.0002,
+      "step": 18060
+    },
+    {
+      "epoch": 77.2,
+      "learning_rate": 8.81287457391155e-05,
+      "loss": 0.0003,
+      "step": 18064
+    },
+    {
+      "epoch": 77.21,
+      "learning_rate": 8.812120553470223e-05,
+      "loss": 0.0001,
+      "step": 18068
+    },
+    {
+      "epoch": 77.23,
+      "learning_rate": 8.811366325919715e-05,
+      "loss": 0.0011,
+      "step": 18072
+    },
+    {
+      "epoch": 77.25,
+      "learning_rate": 8.810611891301e-05,
+      "loss": 0.0001,
+      "step": 18076
+    },
+    {
+      "epoch": 77.26,
+      "learning_rate": 8.809857249655066e-05,
+      "loss": 0.0003,
+      "step": 18080
+    },
+    {
+      "epoch": 77.28,
+      "learning_rate": 8.809102401022916e-05,
+      "loss": 0.0001,
+      "step": 18084
+    },
+    {
+      "epoch": 77.3,
+      "learning_rate": 8.808347345445555e-05,
+      "loss": 0.0001,
+      "step": 18088
+    },
+    {
+      "epoch": 77.32,
+      "learning_rate": 8.807592082964006e-05,
+      "loss": 0.0001,
+      "step": 18092
+    },
+    {
+      "epoch": 77.33,
+      "learning_rate": 8.806836613619303e-05,
+      "loss": 0.0001,
+      "step": 18096
+    },
+    {
+      "epoch": 77.35,
+      "learning_rate": 8.806080937452489e-05,
+      "loss": 0.0,
+      "step": 18100
+    },
+    {
+      "epoch": 77.37,
+      "learning_rate": 8.80532505450462e-05,
+      "loss": 0.0005,
+      "step": 18104
+    },
+    {
+      "epoch": 77.38,
+      "learning_rate": 8.804568964816762e-05,
+      "loss": 0.0009,
+      "step": 18108
+    },
+    {
+      "epoch": 77.4,
+      "learning_rate": 8.803812668429992e-05,
+      "loss": 0.0001,
+      "step": 18112
+    },
+    {
+      "epoch": 77.42,
+      "learning_rate": 8.8030561653854e-05,
+      "loss": 0.0025,
+      "step": 18116
+    },
+    {
+      "epoch": 77.44,
+      "learning_rate": 8.802299455724086e-05,
+      "loss": 0.0012,
+      "step": 18120
+    },
+    {
+      "epoch": 77.45,
+      "learning_rate": 8.80154253948716e-05,
+      "loss": 0.0001,
+      "step": 18124
+    },
+    {
+      "epoch": 77.47,
+      "learning_rate": 8.800785416715747e-05,
+      "loss": 0.0003,
+      "step": 18128
+    },
+    {
+      "epoch": 77.49,
+      "learning_rate": 8.80002808745098e-05,
+      "loss": 0.0005,
+      "step": 18132
+    },
+    {
+      "epoch": 77.5,
+      "learning_rate": 8.799270551734002e-05,
+      "loss": 0.0017,
+      "step": 18136
+    },
+    {
+      "epoch": 77.52,
+      "learning_rate": 8.798512809605973e-05,
+      "loss": 0.0009,
+      "step": 18140
+    },
+    {
+      "epoch": 77.54,
+      "learning_rate": 8.797754861108056e-05,
+      "loss": 0.0006,
+      "step": 18144
+    },
+    {
+      "epoch": 77.56,
+      "learning_rate": 8.796996706281433e-05,
+      "loss": 0.0003,
+      "step": 18148
+    },
+    {
+      "epoch": 77.57,
+      "learning_rate": 8.796238345167293e-05,
+      "loss": 0.0001,
+      "step": 18152
+    },
+    {
+      "epoch": 77.59,
+      "learning_rate": 8.795479777806838e-05,
+      "loss": 0.0001,
+      "step": 18156
+    },
+    {
+      "epoch": 77.61,
+      "learning_rate": 8.794721004241277e-05,
+      "loss": 0.0001,
+      "step": 18160
+    },
+    {
+      "epoch": 77.62,
+      "learning_rate": 8.79396202451184e-05,
+      "loss": 0.0007,
+      "step": 18164
+    },
+    {
+      "epoch": 77.64,
+      "learning_rate": 8.793202838659753e-05,
+      "loss": 0.0001,
+      "step": 18168
+    },
+    {
+      "epoch": 77.66,
+      "learning_rate": 8.792443446726268e-05,
+      "loss": 0.0017,
+      "step": 18172
+    },
+    {
+      "epoch": 77.68,
+      "learning_rate": 8.79168384875264e-05,
+      "loss": 0.0001,
+      "step": 18176
+    },
+    {
+      "epoch": 77.69,
+      "learning_rate": 8.790924044780139e-05,
+      "loss": 0.0001,
+      "step": 18180
+    },
+    {
+      "epoch": 77.71,
+      "learning_rate": 8.790164034850043e-05,
+      "loss": 0.0002,
+      "step": 18184
+    },
+    {
+      "epoch": 77.73,
+      "learning_rate": 8.789403819003642e-05,
+      "loss": 0.0003,
+      "step": 18188
+    },
+    {
+      "epoch": 77.74,
+      "learning_rate": 8.78864339728224e-05,
+      "loss": 0.0001,
+      "step": 18192
+    },
+    {
+      "epoch": 77.76,
+      "learning_rate": 8.78788276972715e-05,
+      "loss": 0.0002,
+      "step": 18196
+    },
+    {
+      "epoch": 77.78,
+      "learning_rate": 8.787121936379692e-05,
+      "loss": 0.0001,
+      "step": 18200
+    },
+    {
+      "epoch": 77.79,
+      "learning_rate": 8.786360897281206e-05,
+      "loss": 0.0013,
+      "step": 18204
+    },
+    {
+      "epoch": 77.81,
+      "learning_rate": 8.785599652473037e-05,
+      "loss": 0.0012,
+      "step": 18208
+    },
+    {
+      "epoch": 77.83,
+      "learning_rate": 8.784838201996544e-05,
+      "loss": 0.0003,
+      "step": 18212
+    },
+    {
+      "epoch": 77.85,
+      "learning_rate": 8.784076545893094e-05,
+      "loss": 0.0001,
+      "step": 18216
+    },
+    {
+      "epoch": 77.86,
+      "learning_rate": 8.783314684204067e-05,
+      "loss": 0.0001,
+      "step": 18220
+    },
+    {
+      "epoch": 77.88,
+      "learning_rate": 8.782552616970856e-05,
+      "loss": 0.0007,
+      "step": 18224
+    },
+    {
+      "epoch": 77.9,
+      "learning_rate": 8.781790344234863e-05,
+      "loss": 0.0005,
+      "step": 18228
+    },
+    {
+      "epoch": 77.91,
+      "learning_rate": 8.781027866037501e-05,
+      "loss": 0.0012,
+      "step": 18232
+    },
+    {
+      "epoch": 77.93,
+      "learning_rate": 8.780265182420195e-05,
+      "loss": 0.0,
+      "step": 18236
+    },
+    {
+      "epoch": 77.95,
+      "learning_rate": 8.779502293424381e-05,
+      "loss": 0.0002,
+      "step": 18240
+    },
+    {
+      "epoch": 77.97,
+      "learning_rate": 8.778739199091506e-05,
+      "loss": 0.0003,
+      "step": 18244
+    },
+    {
+      "epoch": 77.98,
+      "learning_rate": 8.777975899463029e-05,
+      "loss": 0.0002,
+      "step": 18248
+    },
+    {
+      "epoch": 78.0,
+      "learning_rate": 8.777212394580418e-05,
+      "loss": 0.0012,
+      "step": 18252
+    },
+    {
+      "epoch": 78.02,
+      "learning_rate": 8.776448684485155e-05,
+      "loss": 0.0001,
+      "step": 18256
+    },
+    {
+      "epoch": 78.03,
+      "learning_rate": 8.775684769218731e-05,
+      "loss": 0.0,
+      "step": 18260
+    },
+    {
+      "epoch": 78.05,
+      "learning_rate": 8.774920648822646e-05,
+      "loss": 0.0002,
+      "step": 18264
+    },
+    {
+      "epoch": 78.07,
+      "learning_rate": 8.77415632333842e-05,
+      "loss": 0.0002,
+      "step": 18268
+    },
+    {
+      "epoch": 78.09,
+      "learning_rate": 8.773391792807575e-05,
+      "loss": 0.0001,
+      "step": 18272
+    },
+    {
+      "epoch": 78.1,
+      "learning_rate": 8.772627057271646e-05,
+      "loss": 0.0002,
+      "step": 18276
+    },
+    {
+      "epoch": 78.12,
+      "learning_rate": 8.771862116772182e-05,
+      "loss": 0.0001,
+      "step": 18280
+    },
+    {
+      "epoch": 78.14,
+      "learning_rate": 8.771096971350741e-05,
+      "loss": 0.0,
+      "step": 18284
+    },
+    {
+      "epoch": 78.15,
+      "learning_rate": 8.770331621048893e-05,
+      "loss": 0.0001,
+      "step": 18288
+    },
+    {
+      "epoch": 78.17,
+      "learning_rate": 8.769566065908219e-05,
+      "loss": 0.0001,
+      "step": 18292
+    },
+    {
+      "epoch": 78.19,
+      "learning_rate": 8.76880030597031e-05,
+      "loss": 0.0001,
+      "step": 18296
+    },
+    {
+      "epoch": 78.21,
+      "learning_rate": 8.768034341276772e-05,
+      "loss": 0.0009,
+      "step": 18300
+    },
+    {
+      "epoch": 78.22,
+      "learning_rate": 8.767268171869214e-05,
+      "loss": 0.0002,
+      "step": 18304
+    },
+    {
+      "epoch": 78.24,
+      "learning_rate": 8.766501797789266e-05,
+      "loss": 0.0001,
+      "step": 18308
+    },
+    {
+      "epoch": 78.26,
+      "learning_rate": 8.765735219078561e-05,
+      "loss": 0.0,
+      "step": 18312
+    },
+    {
+      "epoch": 78.27,
+      "learning_rate": 8.764968435778751e-05,
+      "loss": 0.0001,
+      "step": 18316
+    },
+    {
+      "epoch": 78.29,
+      "learning_rate": 8.76420144793149e-05,
+      "loss": 0.0002,
+      "step": 18320
+    },
+    {
+      "epoch": 78.31,
+      "learning_rate": 8.763434255578449e-05,
+      "loss": 0.0004,
+      "step": 18324
+    },
+    {
+      "epoch": 78.32,
+      "learning_rate": 8.76266685876131e-05,
+      "loss": 0.0006,
+      "step": 18328
+    },
+    {
+      "epoch": 78.34,
+      "learning_rate": 8.761899257521766e-05,
+      "loss": 0.0001,
+      "step": 18332
+    },
+    {
+      "epoch": 78.36,
+      "learning_rate": 8.761131451901517e-05,
+      "loss": 0.0001,
+      "step": 18336
+    },
+    {
+      "epoch": 78.38,
+      "learning_rate": 8.760363441942279e-05,
+      "loss": 0.0,
+      "step": 18340
+    },
+    {
+      "epoch": 78.39,
+      "learning_rate": 8.759595227685778e-05,
+      "loss": 0.0001,
+      "step": 18344
+    },
+    {
+      "epoch": 78.41,
+      "learning_rate": 8.75882680917375e-05,
+      "loss": 0.0001,
+      "step": 18348
+    },
+    {
+      "epoch": 78.43,
+      "learning_rate": 8.758058186447942e-05,
+      "loss": 0.0014,
+      "step": 18352
+    },
+    {
+      "epoch": 78.44,
+      "learning_rate": 8.757289359550111e-05,
+      "loss": 0.0011,
+      "step": 18356
+    },
+    {
+      "epoch": 78.46,
+      "learning_rate": 8.75652032852203e-05,
+      "loss": 0.0004,
+      "step": 18360
+    },
+    {
+      "epoch": 78.48,
+      "learning_rate": 8.755751093405478e-05,
+      "loss": 0.0001,
+      "step": 18364
+    },
+    {
+      "epoch": 78.5,
+      "learning_rate": 8.754981654242246e-05,
+      "loss": 0.0001,
+      "step": 18368
+    },
+    {
+      "epoch": 78.51,
+      "learning_rate": 8.754212011074139e-05,
+      "loss": 0.0,
+      "step": 18372
+    },
+    {
+      "epoch": 78.53,
+      "learning_rate": 8.753442163942969e-05,
+      "loss": 0.0017,
+      "step": 18376
+    },
+    {
+      "epoch": 78.55,
+      "learning_rate": 8.752672112890563e-05,
+      "loss": 0.0001,
+      "step": 18380
+    },
+    {
+      "epoch": 78.56,
+      "learning_rate": 8.751901857958756e-05,
+      "loss": 0.0002,
+      "step": 18384
+    },
+    {
+      "epoch": 78.58,
+      "learning_rate": 8.751131399189396e-05,
+      "loss": 0.0001,
+      "step": 18388
+    },
+    {
+      "epoch": 78.6,
+      "learning_rate": 8.750360736624342e-05,
+      "loss": 0.0008,
+      "step": 18392
+    },
+    {
+      "epoch": 78.62,
+      "learning_rate": 8.749589870305462e-05,
+      "loss": 0.0021,
+      "step": 18396
+    },
+    {
+      "epoch": 78.63,
+      "learning_rate": 8.748818800274635e-05,
+      "loss": 0.0002,
+      "step": 18400
+    },
+    {
+      "epoch": 78.65,
+      "learning_rate": 8.748047526573755e-05,
+      "loss": 0.0001,
+      "step": 18404
+    },
+    {
+      "epoch": 78.67,
+      "learning_rate": 8.747276049244726e-05,
+      "loss": 0.0001,
+      "step": 18408
+    },
+    {
+      "epoch": 78.68,
+      "learning_rate": 8.746504368329457e-05,
+      "loss": 0.0001,
+      "step": 18412
+    },
+    {
+      "epoch": 78.7,
+      "learning_rate": 8.74573248386988e-05,
+      "loss": 0.0001,
+      "step": 18416
+    },
+    {
+      "epoch": 78.72,
+      "learning_rate": 8.744960395907922e-05,
+      "loss": 0.0003,
+      "step": 18420
+    },
+    {
+      "epoch": 78.74,
+      "learning_rate": 8.744188104485535e-05,
+      "loss": 0.0001,
+      "step": 18424
+    },
+    {
+      "epoch": 78.75,
+      "learning_rate": 8.743415609644678e-05,
+      "loss": 0.0001,
+      "step": 18428
+    },
+    {
+      "epoch": 78.77,
+      "learning_rate": 8.742642911427317e-05,
+      "loss": 0.0001,
+      "step": 18432
+    },
+    {
+      "epoch": 78.79,
+      "learning_rate": 8.741870009875434e-05,
+      "loss": 0.0,
+      "step": 18436
+    },
+    {
+      "epoch": 78.8,
+      "learning_rate": 8.741096905031017e-05,
+      "loss": 0.0,
+      "step": 18440
+    },
+    {
+      "epoch": 78.82,
+      "learning_rate": 8.740323596936073e-05,
+      "loss": 0.0001,
+      "step": 18444
+    },
+    {
+      "epoch": 78.84,
+      "learning_rate": 8.739550085632611e-05,
+      "loss": 0.0002,
+      "step": 18448
+    },
+    {
+      "epoch": 78.85,
+      "learning_rate": 8.738776371162657e-05,
+      "loss": 0.0004,
+      "step": 18452
+    },
+    {
+      "epoch": 78.87,
+      "learning_rate": 8.738002453568245e-05,
+      "loss": 0.0001,
+      "step": 18456
+    },
+    {
+      "epoch": 78.89,
+      "learning_rate": 8.737228332891423e-05,
+      "loss": 0.0007,
+      "step": 18460
+    },
+    {
+      "epoch": 78.91,
+      "learning_rate": 8.736454009174249e-05,
+      "loss": 0.0018,
+      "step": 18464
+    },
+    {
+      "epoch": 78.92,
+      "learning_rate": 8.735679482458787e-05,
+      "loss": 0.0001,
+      "step": 18468
+    },
+    {
+      "epoch": 78.94,
+      "learning_rate": 8.734904752787121e-05,
+      "loss": 0.0001,
+      "step": 18472
+    },
+    {
+      "epoch": 78.96,
+      "learning_rate": 8.734129820201339e-05,
+      "loss": 0.0005,
+      "step": 18476
+    },
+    {
+      "epoch": 78.97,
+      "learning_rate": 8.733354684743542e-05,
+      "loss": 0.0001,
+      "step": 18480
+    },
+    {
+      "epoch": 78.99,
+      "learning_rate": 8.732579346455844e-05,
+      "loss": 0.0006,
+      "step": 18484
+    },
+    {
+      "epoch": 79.01,
+      "learning_rate": 8.731803805380369e-05,
+      "loss": 0.0001,
+      "step": 18488
+    },
+    {
+      "epoch": 79.03,
+      "learning_rate": 8.731028061559249e-05,
+      "loss": 0.0001,
+      "step": 18492
+    },
+    {
+      "epoch": 79.04,
+      "learning_rate": 8.730252115034632e-05,
+      "loss": 0.0001,
+      "step": 18496
+    },
+    {
+      "epoch": 79.06,
+      "learning_rate": 8.729475965848673e-05,
+      "loss": 0.0001,
+      "step": 18500
+    },
+    {
+      "epoch": 79.08,
+      "learning_rate": 8.728699614043539e-05,
+      "loss": 0.0001,
+      "step": 18504
+    },
+    {
+      "epoch": 79.09,
+      "learning_rate": 8.72792305966141e-05,
+      "loss": 0.0002,
+      "step": 18508
+    },
+    {
+      "epoch": 79.11,
+      "learning_rate": 8.727146302744473e-05,
+      "loss": 0.0001,
+      "step": 18512
+    },
+    {
+      "epoch": 79.13,
+      "learning_rate": 8.726369343334932e-05,
+      "loss": 0.0001,
+      "step": 18516
+    },
+    {
+      "epoch": 79.15,
+      "learning_rate": 8.725592181474997e-05,
+      "loss": 0.0002,
+      "step": 18520
+    },
+    {
+      "epoch": 79.16,
+      "learning_rate": 8.72481481720689e-05,
+      "loss": 0.0004,
+      "step": 18524
+    },
+    {
+      "epoch": 79.18,
+      "learning_rate": 8.724037250572845e-05,
+      "loss": 0.0001,
+      "step": 18528
+    },
+    {
+      "epoch": 79.2,
+      "learning_rate": 8.723259481615107e-05,
+      "loss": 0.0007,
+      "step": 18532
+    },
+    {
+      "epoch": 79.21,
+      "learning_rate": 8.72248151037593e-05,
+      "loss": 0.0008,
+      "step": 18536
+    },
+    {
+      "epoch": 79.23,
+      "learning_rate": 8.721703336897582e-05,
+      "loss": 0.0001,
+      "step": 18540
+    },
+    {
+      "epoch": 79.25,
+      "learning_rate": 8.72092496122234e-05,
+      "loss": 0.0,
+      "step": 18544
+    },
+    {
+      "epoch": 79.26,
+      "learning_rate": 8.720146383392492e-05,
+      "loss": 0.0001,
+      "step": 18548
+    },
+    {
+      "epoch": 79.28,
+      "learning_rate": 8.719367603450338e-05,
+      "loss": 0.0001,
+      "step": 18552
+    },
+    {
+      "epoch": 79.3,
+      "learning_rate": 8.718588621438188e-05,
+      "loss": 0.0001,
+      "step": 18556
+    },
+    {
+      "epoch": 79.32,
+      "learning_rate": 8.717809437398366e-05,
+      "loss": 0.0003,
+      "step": 18560
+    },
+    {
+      "epoch": 79.33,
+      "learning_rate": 8.717030051373199e-05,
+      "loss": 0.0001,
+      "step": 18564
+    },
+    {
+      "epoch": 79.35,
+      "learning_rate": 8.716250463405034e-05,
+      "loss": 0.0002,
+      "step": 18568
+    },
+    {
+      "epoch": 79.37,
+      "learning_rate": 8.715470673536226e-05,
+      "loss": 0.0002,
+      "step": 18572
+    },
+    {
+      "epoch": 79.38,
+      "learning_rate": 8.714690681809138e-05,
+      "loss": 0.0001,
+      "step": 18576
+    },
+    {
+      "epoch": 79.4,
+      "learning_rate": 8.713910488266148e-05,
+      "loss": 0.0,
+      "step": 18580
+    },
+    {
+      "epoch": 79.42,
+      "learning_rate": 8.713130092949644e-05,
+      "loss": 0.0001,
+      "step": 18584
+    },
+    {
+      "epoch": 79.44,
+      "learning_rate": 8.712349495902021e-05,
+      "loss": 0.0002,
+      "step": 18588
+    },
+    {
+      "epoch": 79.45,
+      "learning_rate": 8.711568697165691e-05,
+      "loss": 0.0001,
+      "step": 18592
+    },
+    {
+      "epoch": 79.47,
+      "learning_rate": 8.710787696783072e-05,
+      "loss": 0.0001,
+      "step": 18596
+    },
+    {
+      "epoch": 79.49,
+      "learning_rate": 8.710006494796597e-05,
+      "loss": 0.0006,
+      "step": 18600
+    },
+    {
+      "epoch": 79.5,
+      "learning_rate": 8.709225091248708e-05,
+      "loss": 0.0005,
+      "step": 18604
+    },
+    {
+      "epoch": 79.52,
+      "learning_rate": 8.708443486181855e-05,
+      "loss": 0.0004,
+      "step": 18608
+    },
+    {
+      "epoch": 79.54,
+      "learning_rate": 8.707661679638505e-05,
+      "loss": 0.0002,
+      "step": 18612
+    },
+    {
+      "epoch": 79.56,
+      "learning_rate": 8.706879671661132e-05,
+      "loss": 0.0001,
+      "step": 18616
+    },
+    {
+      "epoch": 79.57,
+      "learning_rate": 8.706097462292223e-05,
+      "loss": 0.0001,
+      "step": 18620
+    },
+    {
+      "epoch": 79.59,
+      "learning_rate": 8.70531505157427e-05,
+      "loss": 0.0001,
+      "step": 18624
+    },
+    {
+      "epoch": 79.61,
+      "learning_rate": 8.704532439549787e-05,
+      "loss": 0.0003,
+      "step": 18628
+    },
+    {
+      "epoch": 79.62,
+      "learning_rate": 8.703749626261289e-05,
+      "loss": 0.0002,
+      "step": 18632
+    },
+    {
+      "epoch": 79.64,
+      "learning_rate": 8.702966611751306e-05,
+      "loss": 0.0001,
+      "step": 18636
+    },
+    {
+      "epoch": 79.66,
+      "learning_rate": 8.70218339606238e-05,
+      "loss": 0.0004,
+      "step": 18640
+    },
+    {
+      "epoch": 79.68,
+      "learning_rate": 8.70139997923706e-05,
+      "loss": 0.0001,
+      "step": 18644
+    },
+    {
+      "epoch": 79.69,
+      "learning_rate": 8.70061636131791e-05,
+      "loss": 0.0006,
+      "step": 18648
+    },
+    {
+      "epoch": 79.71,
+      "learning_rate": 8.699832542347504e-05,
+      "loss": 0.0001,
+      "step": 18652
+    },
+    {
+      "epoch": 79.73,
+      "learning_rate": 8.699048522368425e-05,
+      "loss": 0.0007,
+      "step": 18656
+    },
+    {
+      "epoch": 79.74,
+      "learning_rate": 8.698264301423267e-05,
+      "loss": 0.0013,
+      "step": 18660
+    },
+    {
+      "epoch": 79.76,
+      "learning_rate": 8.697479879554638e-05,
+      "loss": 0.0,
+      "step": 18664
+    },
+    {
+      "epoch": 79.78,
+      "learning_rate": 8.696695256805154e-05,
+      "loss": 0.0002,
+      "step": 18668
+    },
+    {
+      "epoch": 79.79,
+      "learning_rate": 8.695910433217443e-05,
+      "loss": 0.0004,
+      "step": 18672
+    },
+    {
+      "epoch": 79.81,
+      "learning_rate": 8.695125408834145e-05,
+      "loss": 0.0,
+      "step": 18676
+    },
+    {
+      "epoch": 79.83,
+      "learning_rate": 8.694340183697908e-05,
+      "loss": 0.0001,
+      "step": 18680
+    },
+    {
+      "epoch": 79.85,
+      "learning_rate": 8.693554757851392e-05,
+      "loss": 0.0005,
+      "step": 18684
+    },
+    {
+      "epoch": 79.86,
+      "learning_rate": 8.692769131337271e-05,
+      "loss": 0.0001,
+      "step": 18688
+    },
+    {
+      "epoch": 79.88,
+      "learning_rate": 8.691983304198225e-05,
+      "loss": 0.0003,
+      "step": 18692
+    },
+    {
+      "epoch": 79.9,
+      "learning_rate": 8.69119727647695e-05,
+      "loss": 0.0001,
+      "step": 18696
+    },
+    {
+      "epoch": 79.91,
+      "learning_rate": 8.690411048216147e-05,
+      "loss": 0.0001,
+      "step": 18700
+    },
+    {
+      "epoch": 79.93,
+      "learning_rate": 8.689624619458534e-05,
+      "loss": 0.0003,
+      "step": 18704
+    },
+    {
+      "epoch": 79.95,
+      "learning_rate": 8.688837990246834e-05,
+      "loss": 0.0001,
+      "step": 18708
+    },
+    {
+      "epoch": 79.97,
+      "learning_rate": 8.688051160623786e-05,
+      "loss": 0.0003,
+      "step": 18712
+    },
+    {
+      "epoch": 79.98,
+      "learning_rate": 8.68726413063214e-05,
+      "loss": 0.0008,
+      "step": 18716
+    },
+    {
+      "epoch": 80.0,
+      "learning_rate": 8.686476900314648e-05,
+      "loss": 0.0009,
+      "step": 18720
+    },
+    {
+      "epoch": 80.02,
+      "learning_rate": 8.685689469714086e-05,
+      "loss": 0.0,
+      "step": 18724
+    },
+    {
+      "epoch": 80.03,
+      "learning_rate": 8.684901838873232e-05,
+      "loss": 0.0003,
+      "step": 18728
+    },
+    {
+      "epoch": 80.05,
+      "learning_rate": 8.684114007834876e-05,
+      "loss": 0.0001,
+      "step": 18732
+    },
+    {
+      "epoch": 80.07,
+      "learning_rate": 8.683325976641823e-05,
+      "loss": 0.0,
+      "step": 18736
+    },
+    {
+      "epoch": 80.09,
+      "learning_rate": 8.682537745336882e-05,
+      "loss": 0.0007,
+      "step": 18740
+    },
+    {
+      "epoch": 80.1,
+      "learning_rate": 8.681749313962882e-05,
+      "loss": 0.0,
+      "step": 18744
+    },
+    {
+      "epoch": 80.12,
+      "learning_rate": 8.680960682562653e-05,
+      "loss": 0.0001,
+      "step": 18748
+    },
+    {
+      "epoch": 80.14,
+      "learning_rate": 8.680171851179044e-05,
+      "loss": 0.0001,
+      "step": 18752
+    },
+    {
+      "epoch": 80.15,
+      "learning_rate": 8.679382819854908e-05,
+      "loss": 0.0001,
+      "step": 18756
+    },
+    {
+      "epoch": 80.17,
+      "learning_rate": 8.678593588633118e-05,
+      "loss": 0.0002,
+      "step": 18760
+    },
+    {
+      "epoch": 80.19,
+      "learning_rate": 8.677804157556549e-05,
+      "loss": 0.0003,
+      "step": 18764
+    },
+    {
+      "epoch": 80.21,
+      "learning_rate": 8.677014526668087e-05,
+      "loss": 0.0006,
+      "step": 18768
+    },
+    {
+      "epoch": 80.22,
+      "learning_rate": 8.676224696010637e-05,
+      "loss": 0.0015,
+      "step": 18772
+    },
+    {
+      "epoch": 80.24,
+      "learning_rate": 8.675434665627107e-05,
+      "loss": 0.0001,
+      "step": 18776
+    },
+    {
+      "epoch": 80.26,
+      "learning_rate": 8.674644435560419e-05,
+      "loss": 0.0001,
+      "step": 18780
+    },
+    {
+      "epoch": 80.27,
+      "learning_rate": 8.673854005853508e-05,
+      "loss": 0.0001,
+      "step": 18784
+    },
+    {
+      "epoch": 80.29,
+      "learning_rate": 8.673063376549315e-05,
+      "loss": 0.0,
+      "step": 18788
+    },
+    {
+      "epoch": 80.31,
+      "learning_rate": 8.672272547690793e-05,
+      "loss": 0.0001,
+      "step": 18792
+    },
+    {
+      "epoch": 80.32,
+      "learning_rate": 8.67148151932091e-05,
+      "loss": 0.0001,
+      "step": 18796
+    },
+    {
+      "epoch": 80.34,
+      "learning_rate": 8.67069029148264e-05,
+      "loss": 0.0,
+      "step": 18800
+    },
+    {
+      "epoch": 80.36,
+      "learning_rate": 8.669898864218968e-05,
+      "loss": 0.0,
+      "step": 18804
+    },
+    {
+      "epoch": 80.38,
+      "learning_rate": 8.669107237572896e-05,
+      "loss": 0.0,
+      "step": 18808
+    },
+    {
+      "epoch": 80.39,
+      "learning_rate": 8.668315411587432e-05,
+      "loss": 0.0001,
+      "step": 18812
+    },
+    {
+      "epoch": 80.41,
+      "learning_rate": 8.66752338630559e-05,
+      "loss": 0.0007,
+      "step": 18816
+    },
+    {
+      "epoch": 80.43,
+      "learning_rate": 8.666731161770404e-05,
+      "loss": 0.0001,
+      "step": 18820
+    },
+    {
+      "epoch": 80.44,
+      "learning_rate": 8.665938738024915e-05,
+      "loss": 0.0003,
+      "step": 18824
+    },
+    {
+      "epoch": 80.46,
+      "learning_rate": 8.665146115112173e-05,
+      "loss": 0.0,
+      "step": 18828
+    },
+    {
+      "epoch": 80.48,
+      "learning_rate": 8.664353293075243e-05,
+      "loss": 0.0001,
+      "step": 18832
+    },
+    {
+      "epoch": 80.5,
+      "learning_rate": 8.663560271957196e-05,
+      "loss": 0.0001,
+      "step": 18836
+    },
+    {
+      "epoch": 80.51,
+      "learning_rate": 8.662767051801116e-05,
+      "loss": 0.0001,
+      "step": 18840
+    },
+    {
+      "epoch": 80.53,
+      "learning_rate": 8.6619736326501e-05,
+      "loss": 0.0001,
+      "step": 18844
+    },
+    {
+      "epoch": 80.55,
+      "learning_rate": 8.661180014547252e-05,
+      "loss": 0.0001,
+      "step": 18848
+    },
+    {
+      "epoch": 80.56,
+      "learning_rate": 8.66038619753569e-05,
+      "loss": 0.0,
+      "step": 18852
+    },
+    {
+      "epoch": 80.58,
+      "learning_rate": 8.65959218165854e-05,
+      "loss": 0.0007,
+      "step": 18856
+    },
+    {
+      "epoch": 80.6,
+      "learning_rate": 8.65879796695894e-05,
+      "loss": 0.0002,
+      "step": 18860
+    },
+    {
+      "epoch": 80.62,
+      "learning_rate": 8.658003553480042e-05,
+      "loss": 0.0007,
+      "step": 18864
+    },
+    {
+      "epoch": 80.63,
+      "learning_rate": 8.657208941265002e-05,
+      "loss": 0.0002,
+      "step": 18868
+    },
+    {
+      "epoch": 80.65,
+      "learning_rate": 8.656414130356994e-05,
+      "loss": 0.0,
+      "step": 18872
+    },
+    {
+      "epoch": 80.67,
+      "learning_rate": 8.655619120799195e-05,
+      "loss": 0.0001,
+      "step": 18876
+    },
+    {
+      "epoch": 80.68,
+      "learning_rate": 8.654823912634801e-05,
+      "loss": 0.0001,
+      "step": 18880
+    },
+    {
+      "epoch": 80.7,
+      "learning_rate": 8.654028505907012e-05,
+      "loss": 0.0002,
+      "step": 18884
+    },
+    {
+      "epoch": 80.72,
+      "learning_rate": 8.653232900659045e-05,
+      "loss": 0.0001,
+      "step": 18888
+    },
+    {
+      "epoch": 80.74,
+      "learning_rate": 8.652437096934124e-05,
+      "loss": 0.0002,
+      "step": 18892
+    },
+    {
+      "epoch": 80.75,
+      "learning_rate": 8.651641094775483e-05,
+      "loss": 0.0,
+      "step": 18896
+    },
+    {
+      "epoch": 80.77,
+      "learning_rate": 8.650844894226366e-05,
+      "loss": 0.0003,
+      "step": 18900
+    },
+    {
+      "epoch": 80.79,
+      "learning_rate": 8.650048495330034e-05,
+      "loss": 0.0002,
+      "step": 18904
+    },
+    {
+      "epoch": 80.8,
+      "learning_rate": 8.649251898129754e-05,
+      "loss": 0.0001,
+      "step": 18908
+    },
+    {
+      "epoch": 80.82,
+      "learning_rate": 8.648455102668802e-05,
+      "loss": 0.0001,
+      "step": 18912
+    },
+    {
+      "epoch": 80.84,
+      "learning_rate": 8.647658108990469e-05,
+      "loss": 0.0001,
+      "step": 18916
+    },
+    {
+      "epoch": 80.85,
+      "learning_rate": 8.646860917138055e-05,
+      "loss": 0.0021,
+      "step": 18920
+    },
+    {
+      "epoch": 80.87,
+      "learning_rate": 8.646063527154869e-05,
+      "loss": 0.0003,
+      "step": 18924
+    },
+    {
+      "epoch": 80.89,
+      "learning_rate": 8.645265939084235e-05,
+      "loss": 0.0,
+      "step": 18928
+    },
+    {
+      "epoch": 80.91,
+      "learning_rate": 8.644468152969482e-05,
+      "loss": 0.0003,
+      "step": 18932
+    },
+    {
+      "epoch": 80.92,
+      "learning_rate": 8.643670168853957e-05,
+      "loss": 0.0002,
+      "step": 18936
+    },
+    {
+      "epoch": 80.94,
+      "learning_rate": 8.642871986781012e-05,
+      "loss": 0.0001,
+      "step": 18940
+    },
+    {
+      "epoch": 80.96,
+      "learning_rate": 8.642073606794011e-05,
+      "loss": 0.0001,
+      "step": 18944
+    },
+    {
+      "epoch": 80.97,
+      "learning_rate": 8.64127502893633e-05,
+      "loss": 0.0001,
+      "step": 18948
+    },
+    {
+      "epoch": 80.99,
+      "learning_rate": 8.640476253251354e-05,
+      "loss": 0.0009,
+      "step": 18952
+    },
+    {
+      "epoch": 81.01,
+      "learning_rate": 8.639677279782481e-05,
+      "loss": 0.0001,
+      "step": 18956
+    },
+    {
+      "epoch": 81.03,
+      "learning_rate": 8.638878108573117e-05,
+      "loss": 0.0001,
+      "step": 18960
+    },
+    {
+      "epoch": 81.04,
+      "learning_rate": 8.638078739666683e-05,
+      "loss": 0.0,
+      "step": 18964
+    },
+    {
+      "epoch": 81.06,
+      "learning_rate": 8.637279173106605e-05,
+      "loss": 0.0001,
+      "step": 18968
+    },
+    {
+      "epoch": 81.08,
+      "learning_rate": 8.636479408936324e-05,
+      "loss": 0.0002,
+      "step": 18972
+    },
+    {
+      "epoch": 81.09,
+      "learning_rate": 8.635679447199292e-05,
+      "loss": 0.0001,
+      "step": 18976
+    },
+    {
+      "epoch": 81.11,
+      "learning_rate": 8.634879287938969e-05,
+      "loss": 0.0004,
+      "step": 18980
+    },
+    {
+      "epoch": 81.13,
+      "learning_rate": 8.634078931198825e-05,
+      "loss": 0.0001,
+      "step": 18984
+    },
+    {
+      "epoch": 81.15,
+      "learning_rate": 8.633278377022345e-05,
+      "loss": 0.0002,
+      "step": 18988
+    },
+    {
+      "epoch": 81.16,
+      "learning_rate": 8.632477625453021e-05,
+      "loss": 0.0002,
+      "step": 18992
+    },
+    {
+      "epoch": 81.18,
+      "learning_rate": 8.631676676534359e-05,
+      "loss": 0.0001,
+      "step": 18996
+    },
+    {
+      "epoch": 81.2,
+      "learning_rate": 8.630875530309873e-05,
+      "loss": 0.0001,
+      "step": 19000
+    },
+    {
+      "epoch": 81.2,
+      "eval_exact_match": 0.5166320166320166,
+      "eval_loss": 0.9038861393928528,
+      "eval_runtime": 138.7634,
+      "eval_samples_per_second": 6.933,
+      "step": 19000
+    },
+    {
+      "epoch": 81.21,
+      "learning_rate": 8.630074186823088e-05,
+      "loss": 0.0009,
+      "step": 19004
+    },
+    {
+      "epoch": 81.23,
+      "learning_rate": 8.629272646117542e-05,
+      "loss": 0.0001,
+      "step": 19008
+    },
+    {
+      "epoch": 81.25,
+      "learning_rate": 8.628470908236779e-05,
+      "loss": 0.0015,
+      "step": 19012
+    },
+    {
+      "epoch": 81.26,
+      "learning_rate": 8.627668973224358e-05,
+      "loss": 0.0001,
+      "step": 19016
+    },
+    {
+      "epoch": 81.28,
+      "learning_rate": 8.626866841123849e-05,
+      "loss": 0.0,
+      "step": 19020
+    },
+    {
+      "epoch": 81.3,
+      "learning_rate": 8.626064511978831e-05,
+      "loss": 0.0025,
+      "step": 19024
+    },
+    {
+      "epoch": 81.32,
+      "learning_rate": 8.62526198583289e-05,
+      "loss": 0.0014,
+      "step": 19028
+    },
+    {
+      "epoch": 81.33,
+      "learning_rate": 8.624459262729631e-05,
+      "loss": 0.0003,
+      "step": 19032
+    },
+    {
+      "epoch": 81.35,
+      "learning_rate": 8.623656342712664e-05,
+      "loss": 0.0005,
+      "step": 19036
+    },
+    {
+      "epoch": 81.37,
+      "learning_rate": 8.622853225825611e-05,
+      "loss": 0.0001,
+      "step": 19040
+    },
+    {
+      "epoch": 81.38,
+      "learning_rate": 8.622049912112103e-05,
+      "loss": 0.0004,
+      "step": 19044
+    },
+    {
+      "epoch": 81.4,
+      "learning_rate": 8.621246401615786e-05,
+      "loss": 0.0003,
+      "step": 19048
+    },
+    {
+      "epoch": 81.42,
+      "learning_rate": 8.620442694380311e-05,
+      "loss": 0.0,
+      "step": 19052
+    },
+    {
+      "epoch": 81.44,
+      "learning_rate": 8.619638790449345e-05,
+      "loss": 0.0001,
+      "step": 19056
+    },
+    {
+      "epoch": 81.45,
+      "learning_rate": 8.618834689866562e-05,
+      "loss": 0.0001,
+      "step": 19060
+    },
+    {
+      "epoch": 81.47,
+      "learning_rate": 8.618030392675649e-05,
+      "loss": 0.0002,
+      "step": 19064
+    },
+    {
+      "epoch": 81.49,
+      "learning_rate": 8.617225898920302e-05,
+      "loss": 0.0003,
+      "step": 19068
+    },
+    {
+      "epoch": 81.5,
+      "learning_rate": 8.61642120864423e-05,
+      "loss": 0.0001,
+      "step": 19072
+    },
+    {
+      "epoch": 81.52,
+      "learning_rate": 8.615616321891151e-05,
+      "loss": 0.0001,
+      "step": 19076
+    },
+    {
+      "epoch": 81.54,
+      "learning_rate": 8.614811238704791e-05,
+      "loss": 0.0005,
+      "step": 19080
+    },
+    {
+      "epoch": 81.56,
+      "learning_rate": 8.614005959128892e-05,
+      "loss": 0.0006,
+      "step": 19084
+    },
+    {
+      "epoch": 81.57,
+      "learning_rate": 8.613200483207205e-05,
+      "loss": 0.0001,
+      "step": 19088
+    },
+    {
+      "epoch": 81.59,
+      "learning_rate": 8.612394810983487e-05,
+      "loss": 0.0001,
+      "step": 19092
+    },
+    {
+      "epoch": 81.61,
+      "learning_rate": 8.611588942501512e-05,
+      "loss": 0.0006,
+      "step": 19096
+    },
+    {
+      "epoch": 81.62,
+      "learning_rate": 8.610782877805063e-05,
+      "loss": 0.0001,
+      "step": 19100
+    },
+    {
+      "epoch": 81.64,
+      "learning_rate": 8.609976616937931e-05,
+      "loss": 0.0001,
+      "step": 19104
+    },
+    {
+      "epoch": 81.66,
+      "learning_rate": 8.60917015994392e-05,
+      "loss": 0.0005,
+      "step": 19108
+    },
+    {
+      "epoch": 81.68,
+      "learning_rate": 8.608363506866844e-05,
+      "loss": 0.0002,
+      "step": 19112
+    },
+    {
+      "epoch": 81.69,
+      "learning_rate": 8.607556657750528e-05,
+      "loss": 0.0002,
+      "step": 19116
+    },
+    {
+      "epoch": 81.71,
+      "learning_rate": 8.606749612638809e-05,
+      "loss": 0.0001,
+      "step": 19120
+    },
+    {
+      "epoch": 81.73,
+      "learning_rate": 8.605942371575531e-05,
+      "loss": 0.0001,
+      "step": 19124
+    },
+    {
+      "epoch": 81.74,
+      "learning_rate": 8.605134934604548e-05,
+      "loss": 0.0002,
+      "step": 19128
+    },
+    {
+      "epoch": 81.76,
+      "learning_rate": 8.604327301769735e-05,
+      "loss": 0.0003,
+      "step": 19132
+    },
+    {
+      "epoch": 81.78,
+      "learning_rate": 8.603519473114962e-05,
+      "loss": 0.0001,
+      "step": 19136
+    },
+    {
+      "epoch": 81.79,
+      "learning_rate": 8.602711448684123e-05,
+      "loss": 0.0002,
+      "step": 19140
+    },
+    {
+      "epoch": 81.81,
+      "learning_rate": 8.601903228521116e-05,
+      "loss": 0.0002,
+      "step": 19144
+    },
+    {
+      "epoch": 81.83,
+      "learning_rate": 8.601094812669849e-05,
+      "loss": 0.0006,
+      "step": 19148
+    },
+    {
+      "epoch": 81.85,
+      "learning_rate": 8.600286201174243e-05,
+      "loss": 0.0001,
+      "step": 19152
+    },
+    {
+      "epoch": 81.86,
+      "learning_rate": 8.599477394078233e-05,
+      "loss": 0.0001,
+      "step": 19156
+    },
+    {
+      "epoch": 81.88,
+      "learning_rate": 8.598668391425754e-05,
+      "loss": 0.0002,
+      "step": 19160
+    },
+    {
+      "epoch": 81.9,
+      "learning_rate": 8.597859193260765e-05,
+      "loss": 0.0004,
+      "step": 19164
+    },
+    {
+      "epoch": 81.91,
+      "learning_rate": 8.597049799627225e-05,
+      "loss": 0.0001,
+      "step": 19168
+    },
+    {
+      "epoch": 81.93,
+      "learning_rate": 8.596240210569108e-05,
+      "loss": 0.0003,
+      "step": 19172
+    },
+    {
+      "epoch": 81.95,
+      "learning_rate": 8.5954304261304e-05,
+      "loss": 0.0001,
+      "step": 19176
+    },
+    {
+      "epoch": 81.97,
+      "learning_rate": 8.594620446355096e-05,
+      "loss": 0.0001,
+      "step": 19180
+    },
+    {
+      "epoch": 81.98,
+      "learning_rate": 8.5938102712872e-05,
+      "loss": 0.0001,
+      "step": 19184
+    },
+    {
+      "epoch": 82.0,
+      "learning_rate": 8.592999900970726e-05,
+      "loss": 0.0004,
+      "step": 19188
+    },
+    {
+      "epoch": 82.02,
+      "learning_rate": 8.592189335449706e-05,
+      "loss": 0.0001,
+      "step": 19192
+    },
+    {
+      "epoch": 82.03,
+      "learning_rate": 8.591378574768174e-05,
+      "loss": 0.0,
+      "step": 19196
+    },
+    {
+      "epoch": 82.05,
+      "learning_rate": 8.590567618970178e-05,
+      "loss": 0.0001,
+      "step": 19200
+    },
+    {
+      "epoch": 82.07,
+      "learning_rate": 8.589756468099778e-05,
+      "loss": 0.0,
+      "step": 19204
+    },
+    {
+      "epoch": 82.09,
+      "learning_rate": 8.588945122201042e-05,
+      "loss": 0.0004,
+      "step": 19208
+    },
+    {
+      "epoch": 82.1,
+      "learning_rate": 8.58813358131805e-05,
+      "loss": 0.0001,
+      "step": 19212
+    },
+    {
+      "epoch": 82.12,
+      "learning_rate": 8.587321845494891e-05,
+      "loss": 0.0002,
+      "step": 19216
+    },
+    {
+      "epoch": 82.14,
+      "learning_rate": 8.586509914775667e-05,
+      "loss": 0.0,
+      "step": 19220
+    },
+    {
+      "epoch": 82.15,
+      "learning_rate": 8.585697789204493e-05,
+      "loss": 0.0009,
+      "step": 19224
+    },
+    {
+      "epoch": 82.17,
+      "learning_rate": 8.584885468825483e-05,
+      "loss": 0.0002,
+      "step": 19228
+    },
+    {
+      "epoch": 82.19,
+      "learning_rate": 8.584072953682776e-05,
+      "loss": 0.0015,
+      "step": 19232
+    },
+    {
+      "epoch": 82.21,
+      "learning_rate": 8.583260243820514e-05,
+      "loss": 0.0001,
+      "step": 19236
+    },
+    {
+      "epoch": 82.22,
+      "learning_rate": 8.582447339282851e-05,
+      "loss": 0.0001,
+      "step": 19240
+    },
+    {
+      "epoch": 82.24,
+      "learning_rate": 8.58163424011395e-05,
+      "loss": 0.0001,
+      "step": 19244
+    },
+    {
+      "epoch": 82.26,
+      "learning_rate": 8.580820946357985e-05,
+      "loss": 0.0002,
+      "step": 19248
+    },
+    {
+      "epoch": 82.27,
+      "learning_rate": 8.580007458059147e-05,
+      "loss": 0.0001,
+      "step": 19252
+    },
+    {
+      "epoch": 82.29,
+      "learning_rate": 8.579193775261627e-05,
+      "loss": 0.0006,
+      "step": 19256
+    },
+    {
+      "epoch": 82.31,
+      "learning_rate": 8.578379898009632e-05,
+      "loss": 0.0,
+      "step": 19260
+    },
+    {
+      "epoch": 82.32,
+      "learning_rate": 8.57756582634738e-05,
+      "loss": 0.0005,
+      "step": 19264
+    },
+    {
+      "epoch": 82.34,
+      "learning_rate": 8.576751560319101e-05,
+      "loss": 0.0,
+      "step": 19268
+    },
+    {
+      "epoch": 82.36,
+      "learning_rate": 8.57593709996903e-05,
+      "loss": 0.0014,
+      "step": 19272
+    },
+    {
+      "epoch": 82.38,
+      "learning_rate": 8.575122445341417e-05,
+      "loss": 0.0001,
+      "step": 19276
+    },
+    {
+      "epoch": 82.39,
+      "learning_rate": 8.574307596480522e-05,
+      "loss": 0.0006,
+      "step": 19280
+    },
+    {
+      "epoch": 82.41,
+      "learning_rate": 8.573492553430615e-05,
+      "loss": 0.0006,
+      "step": 19284
+    },
+    {
+      "epoch": 82.43,
+      "learning_rate": 8.572677316235978e-05,
+      "loss": 0.0,
+      "step": 19288
+    },
+    {
+      "epoch": 82.44,
+      "learning_rate": 8.5718618849409e-05,
+      "loss": 0.0003,
+      "step": 19292
+    },
+    {
+      "epoch": 82.46,
+      "learning_rate": 8.571046259589681e-05,
+      "loss": 0.0001,
+      "step": 19296
+    },
+    {
+      "epoch": 82.48,
+      "learning_rate": 8.570230440226639e-05,
+      "loss": 0.0001,
+      "step": 19300
+    },
+    {
+      "epoch": 82.5,
+      "learning_rate": 8.56941442689609e-05,
+      "loss": 0.0003,
+      "step": 19304
+    },
+    {
+      "epoch": 82.51,
+      "learning_rate": 8.568598219642373e-05,
+      "loss": 0.0,
+      "step": 19308
+    },
+    {
+      "epoch": 82.53,
+      "learning_rate": 8.567781818509829e-05,
+      "loss": 0.0001,
+      "step": 19312
+    },
+    {
+      "epoch": 82.55,
+      "learning_rate": 8.566965223542811e-05,
+      "loss": 0.0002,
+      "step": 19316
+    },
+    {
+      "epoch": 82.56,
+      "learning_rate": 8.566148434785686e-05,
+      "loss": 0.0001,
+      "step": 19320
+    },
+    {
+      "epoch": 82.58,
+      "learning_rate": 8.56533145228283e-05,
+      "loss": 0.0,
+      "step": 19324
+    },
+    {
+      "epoch": 82.6,
+      "learning_rate": 8.564514276078626e-05,
+      "loss": 0.0008,
+      "step": 19328
+    },
+    {
+      "epoch": 82.62,
+      "learning_rate": 8.563696906217475e-05,
+      "loss": 0.0001,
+      "step": 19332
+    },
+    {
+      "epoch": 82.63,
+      "learning_rate": 8.562879342743779e-05,
+      "loss": 0.0001,
+      "step": 19336
+    },
+    {
+      "epoch": 82.65,
+      "learning_rate": 8.562061585701961e-05,
+      "loss": 0.0001,
+      "step": 19340
+    },
+    {
+      "epoch": 82.67,
+      "learning_rate": 8.561243635136443e-05,
+      "loss": 0.0001,
+      "step": 19344
+    },
+    {
+      "epoch": 82.68,
+      "learning_rate": 8.560425491091668e-05,
+      "loss": 0.0005,
+      "step": 19348
+    },
+    {
+      "epoch": 82.7,
+      "learning_rate": 8.559607153612084e-05,
+      "loss": 0.0001,
+      "step": 19352
+    },
+    {
+      "epoch": 82.72,
+      "learning_rate": 8.558788622742149e-05,
+      "loss": 0.0001,
+      "step": 19356
+    },
+    {
+      "epoch": 82.74,
+      "learning_rate": 8.557969898526335e-05,
+      "loss": 0.0,
+      "step": 19360
+    },
+    {
+      "epoch": 82.75,
+      "learning_rate": 8.557150981009121e-05,
+      "loss": 0.0006,
+      "step": 19364
+    },
+    {
+      "epoch": 82.77,
+      "learning_rate": 8.556331870234999e-05,
+      "loss": 0.0001,
+      "step": 19368
+    },
+    {
+      "epoch": 82.79,
+      "learning_rate": 8.555512566248471e-05,
+      "loss": 0.0003,
+      "step": 19372
+    },
+    {
+      "epoch": 82.8,
+      "learning_rate": 8.554693069094048e-05,
+      "loss": 0.0002,
+      "step": 19376
+    },
+    {
+      "epoch": 82.82,
+      "learning_rate": 8.553873378816252e-05,
+      "loss": 0.0001,
+      "step": 19380
+    },
+    {
+      "epoch": 82.84,
+      "learning_rate": 8.553053495459618e-05,
+      "loss": 0.0001,
+      "step": 19384
+    },
+    {
+      "epoch": 82.85,
+      "learning_rate": 8.55223341906869e-05,
+      "loss": 0.0001,
+      "step": 19388
+    },
+    {
+      "epoch": 82.87,
+      "learning_rate": 8.551413149688019e-05,
+      "loss": 0.0007,
+      "step": 19392
+    },
+    {
+      "epoch": 82.89,
+      "learning_rate": 8.55059268736217e-05,
+      "loss": 0.0001,
+      "step": 19396
+    },
+    {
+      "epoch": 82.91,
+      "learning_rate": 8.54977203213572e-05,
+      "loss": 0.0001,
+      "step": 19400
+    },
+    {
+      "epoch": 82.92,
+      "learning_rate": 8.548951184053257e-05,
+      "loss": 0.0001,
+      "step": 19404
+    },
+    {
+      "epoch": 82.94,
+      "learning_rate": 8.548130143159369e-05,
+      "loss": 0.0004,
+      "step": 19408
+    },
+    {
+      "epoch": 82.96,
+      "learning_rate": 8.54730890949867e-05,
+      "loss": 0.0002,
+      "step": 19412
+    },
+    {
+      "epoch": 82.97,
+      "learning_rate": 8.546487483115773e-05,
+      "loss": 0.0,
+      "step": 19416
+    },
+    {
+      "epoch": 82.99,
+      "learning_rate": 8.545665864055308e-05,
+      "loss": 0.0001,
+      "step": 19420
+    },
+    {
+      "epoch": 83.01,
+      "learning_rate": 8.544844052361909e-05,
+      "loss": 0.0001,
+      "step": 19424
+    },
+    {
+      "epoch": 83.03,
+      "learning_rate": 8.544022048080227e-05,
+      "loss": 0.003,
+      "step": 19428
+    },
+    {
+      "epoch": 83.04,
+      "learning_rate": 8.54319985125492e-05,
+      "loss": 0.0001,
+      "step": 19432
+    },
+    {
+      "epoch": 83.06,
+      "learning_rate": 8.542377461930659e-05,
+      "loss": 0.0001,
+      "step": 19436
+    },
+    {
+      "epoch": 83.08,
+      "learning_rate": 8.54155488015212e-05,
+      "loss": 0.0002,
+      "step": 19440
+    },
+    {
+      "epoch": 83.09,
+      "learning_rate": 8.540732105963998e-05,
+      "loss": 0.0001,
+      "step": 19444
+    },
+    {
+      "epoch": 83.11,
+      "learning_rate": 8.539909139410989e-05,
+      "loss": 0.0002,
+      "step": 19448
+    },
+    {
+      "epoch": 83.13,
+      "learning_rate": 8.539085980537807e-05,
+      "loss": 0.0001,
+      "step": 19452
+    },
+    {
+      "epoch": 83.15,
+      "learning_rate": 8.538262629389171e-05,
+      "loss": 0.0001,
+      "step": 19456
+    },
+    {
+      "epoch": 83.16,
+      "learning_rate": 8.537439086009817e-05,
+      "loss": 0.0,
+      "step": 19460
+    },
+    {
+      "epoch": 83.18,
+      "learning_rate": 8.536615350444483e-05,
+      "loss": 0.0001,
+      "step": 19464
+    },
+    {
+      "epoch": 83.2,
+      "learning_rate": 8.535791422737924e-05,
+      "loss": 0.0005,
+      "step": 19468
+    },
+    {
+      "epoch": 83.21,
+      "learning_rate": 8.534967302934905e-05,
+      "loss": 0.0001,
+      "step": 19472
+    },
+    {
+      "epoch": 83.23,
+      "learning_rate": 8.534142991080194e-05,
+      "loss": 0.0001,
+      "step": 19476
+    },
+    {
+      "epoch": 83.25,
+      "learning_rate": 8.53331848721858e-05,
+      "loss": 0.0016,
+      "step": 19480
+    },
+    {
+      "epoch": 83.26,
+      "learning_rate": 8.532493791394857e-05,
+      "loss": 0.0009,
+      "step": 19484
+    },
+    {
+      "epoch": 83.28,
+      "learning_rate": 8.531668903653831e-05,
+      "loss": 0.0,
+      "step": 19488
+    },
+    {
+      "epoch": 83.3,
+      "learning_rate": 8.530843824040314e-05,
+      "loss": 0.0005,
+      "step": 19492
+    },
+    {
+      "epoch": 83.32,
+      "learning_rate": 8.530018552599134e-05,
+      "loss": 0.0,
+      "step": 19496
+    },
+    {
+      "epoch": 83.33,
+      "learning_rate": 8.529193089375126e-05,
+      "loss": 0.0001,
+      "step": 19500
+    },
+    {
+      "epoch": 83.35,
+      "learning_rate": 8.52836743441314e-05,
+      "loss": 0.0005,
+      "step": 19504
+    },
+    {
+      "epoch": 83.37,
+      "learning_rate": 8.52754158775803e-05,
+      "loss": 0.0,
+      "step": 19508
+    },
+    {
+      "epoch": 83.38,
+      "learning_rate": 8.526715549454664e-05,
+      "loss": 0.0004,
+      "step": 19512
+    },
+    {
+      "epoch": 83.4,
+      "learning_rate": 8.52588931954792e-05,
+      "loss": 0.0,
+      "step": 19516
+    },
+    {
+      "epoch": 83.42,
+      "learning_rate": 8.525062898082685e-05,
+      "loss": 0.0001,
+      "step": 19520
+    },
+    {
+      "epoch": 83.44,
+      "learning_rate": 8.524236285103861e-05,
+      "loss": 0.0001,
+      "step": 19524
+    },
+    {
+      "epoch": 83.45,
+      "learning_rate": 8.523409480656356e-05,
+      "loss": 0.0001,
+      "step": 19528
+    },
+    {
+      "epoch": 83.47,
+      "learning_rate": 8.522582484785088e-05,
+      "loss": 0.0011,
+      "step": 19532
+    },
+    {
+      "epoch": 83.49,
+      "learning_rate": 8.52175529753499e-05,
+      "loss": 0.0001,
+      "step": 19536
+    },
+    {
+      "epoch": 83.5,
+      "learning_rate": 8.520927918950999e-05,
+      "loss": 0.0,
+      "step": 19540
+    },
+    {
+      "epoch": 83.52,
+      "learning_rate": 8.520100349078069e-05,
+      "loss": 0.0016,
+      "step": 19544
+    },
+    {
+      "epoch": 83.54,
+      "learning_rate": 8.519272587961155e-05,
+      "loss": 0.0006,
+      "step": 19548
+    },
+    {
+      "epoch": 83.56,
+      "learning_rate": 8.518444635645237e-05,
+      "loss": 0.0019,
+      "step": 19552
+    },
+    {
+      "epoch": 83.57,
+      "learning_rate": 8.517616492175292e-05,
+      "loss": 0.0001,
+      "step": 19556
+    },
+    {
+      "epoch": 83.59,
+      "learning_rate": 8.516788157596312e-05,
+      "loss": 0.0001,
+      "step": 19560
+    },
+    {
+      "epoch": 83.61,
+      "learning_rate": 8.515959631953301e-05,
+      "loss": 0.0002,
+      "step": 19564
+    },
+    {
+      "epoch": 83.62,
+      "learning_rate": 8.515130915291271e-05,
+      "loss": 0.0008,
+      "step": 19568
+    },
+    {
+      "epoch": 83.64,
+      "learning_rate": 8.514302007655248e-05,
+      "loss": 0.0001,
+      "step": 19572
+    },
+    {
+      "epoch": 83.66,
+      "learning_rate": 8.513472909090263e-05,
+      "loss": 0.0,
+      "step": 19576
+    },
+    {
+      "epoch": 83.68,
+      "learning_rate": 8.512643619641362e-05,
+      "loss": 0.0,
+      "step": 19580
+    },
+    {
+      "epoch": 83.69,
+      "learning_rate": 8.511814139353599e-05,
+      "loss": 0.0001,
+      "step": 19584
+    },
+    {
+      "epoch": 83.71,
+      "learning_rate": 8.510984468272039e-05,
+      "loss": 0.0014,
+      "step": 19588
+    },
+    {
+      "epoch": 83.73,
+      "learning_rate": 8.510154606441756e-05,
+      "loss": 0.0014,
+      "step": 19592
+    },
+    {
+      "epoch": 83.74,
+      "learning_rate": 8.509324553907837e-05,
+      "loss": 0.0001,
+      "step": 19596
+    },
+    {
+      "epoch": 83.76,
+      "learning_rate": 8.508494310715379e-05,
+      "loss": 0.0001,
+      "step": 19600
+    },
+    {
+      "epoch": 83.78,
+      "learning_rate": 8.507663876909487e-05,
+      "loss": 0.0001,
+      "step": 19604
+    },
+    {
+      "epoch": 83.79,
+      "learning_rate": 8.506833252535277e-05,
+      "loss": 0.0,
+      "step": 19608
+    },
+    {
+      "epoch": 83.81,
+      "learning_rate": 8.506002437637879e-05,
+      "loss": 0.0001,
+      "step": 19612
+    },
+    {
+      "epoch": 83.83,
+      "learning_rate": 8.505171432262427e-05,
+      "loss": 0.0,
+      "step": 19616
+    },
+    {
+      "epoch": 83.85,
+      "learning_rate": 8.504340236454071e-05,
+      "loss": 0.0013,
+      "step": 19620
+    },
+    {
+      "epoch": 83.86,
+      "learning_rate": 8.503508850257968e-05,
+      "loss": 0.0002,
+      "step": 19624
+    },
+    {
+      "epoch": 83.88,
+      "learning_rate": 8.502677273719287e-05,
+      "loss": 0.0,
+      "step": 19628
+    },
+    {
+      "epoch": 83.9,
+      "learning_rate": 8.501845506883208e-05,
+      "loss": 0.0002,
+      "step": 19632
+    },
+    {
+      "epoch": 83.91,
+      "learning_rate": 8.501013549794917e-05,
+      "loss": 0.0012,
+      "step": 19636
+    },
+    {
+      "epoch": 83.93,
+      "learning_rate": 8.500181402499617e-05,
+      "loss": 0.0005,
+      "step": 19640
+    },
+    {
+      "epoch": 83.95,
+      "learning_rate": 8.499349065042516e-05,
+      "loss": 0.001,
+      "step": 19644
+    },
+    {
+      "epoch": 83.97,
+      "learning_rate": 8.498516537468834e-05,
+      "loss": 0.0001,
+      "step": 19648
+    },
+    {
+      "epoch": 83.98,
+      "learning_rate": 8.497683819823801e-05,
+      "loss": 0.0009,
+      "step": 19652
+    },
+    {
+      "epoch": 84.0,
+      "learning_rate": 8.496850912152661e-05,
+      "loss": 0.0001,
+      "step": 19656
+    },
+    {
+      "epoch": 84.02,
+      "learning_rate": 8.496017814500661e-05,
+      "loss": 0.0001,
+      "step": 19660
+    },
+    {
+      "epoch": 84.03,
+      "learning_rate": 8.495184526913066e-05,
+      "loss": 0.0002,
+      "step": 19664
+    },
+    {
+      "epoch": 84.05,
+      "learning_rate": 8.494351049435145e-05,
+      "loss": 0.0001,
+      "step": 19668
+    },
+    {
+      "epoch": 84.07,
+      "learning_rate": 8.493517382112182e-05,
+      "loss": 0.0001,
+      "step": 19672
+    },
+    {
+      "epoch": 84.09,
+      "learning_rate": 8.492683524989467e-05,
+      "loss": 0.0001,
+      "step": 19676
+    },
+    {
+      "epoch": 84.1,
+      "learning_rate": 8.491849478112307e-05,
+      "loss": 0.0,
+      "step": 19680
+    },
+    {
+      "epoch": 84.12,
+      "learning_rate": 8.491015241526011e-05,
+      "loss": 0.0,
+      "step": 19684
+    },
+    {
+      "epoch": 84.14,
+      "learning_rate": 8.490180815275906e-05,
+      "loss": 0.0002,
+      "step": 19688
+    },
+    {
+      "epoch": 84.15,
+      "learning_rate": 8.489346199407321e-05,
+      "loss": 0.0017,
+      "step": 19692
+    },
+    {
+      "epoch": 84.17,
+      "learning_rate": 8.488511393965601e-05,
+      "loss": 0.0002,
+      "step": 19696
+    },
+    {
+      "epoch": 84.19,
+      "learning_rate": 8.487676398996105e-05,
+      "loss": 0.0004,
+      "step": 19700
+    },
+    {
+      "epoch": 84.21,
+      "learning_rate": 8.486841214544194e-05,
+      "loss": 0.0002,
+      "step": 19704
+    },
+    {
+      "epoch": 84.22,
+      "learning_rate": 8.48600584065524e-05,
+      "loss": 0.0001,
+      "step": 19708
+    },
+    {
+      "epoch": 84.24,
+      "learning_rate": 8.485170277374635e-05,
+      "loss": 0.0,
+      "step": 19712
+    },
+    {
+      "epoch": 84.26,
+      "learning_rate": 8.484334524747767e-05,
+      "loss": 0.0001,
+      "step": 19716
+    },
+    {
+      "epoch": 84.27,
+      "learning_rate": 8.483498582820048e-05,
+      "loss": 0.0,
+      "step": 19720
+    },
+    {
+      "epoch": 84.29,
+      "learning_rate": 8.482662451636891e-05,
+      "loss": 0.0001,
+      "step": 19724
+    },
+    {
+      "epoch": 84.31,
+      "learning_rate": 8.481826131243722e-05,
+      "loss": 0.0,
+      "step": 19728
+    },
+    {
+      "epoch": 84.32,
+      "learning_rate": 8.480989621685979e-05,
+      "loss": 0.0002,
+      "step": 19732
+    },
+    {
+      "epoch": 84.34,
+      "learning_rate": 8.480152923009107e-05,
+      "loss": 0.0001,
+      "step": 19736
+    },
+    {
+      "epoch": 84.36,
+      "learning_rate": 8.479316035258565e-05,
+      "loss": 0.0016,
+      "step": 19740
+    },
+    {
+      "epoch": 84.38,
+      "learning_rate": 8.478478958479819e-05,
+      "loss": 0.0001,
+      "step": 19744
+    },
+    {
+      "epoch": 84.39,
+      "learning_rate": 8.477641692718348e-05,
+      "loss": 0.0,
+      "step": 19748
+    },
+    {
+      "epoch": 84.41,
+      "learning_rate": 8.476804238019638e-05,
+      "loss": 0.0004,
+      "step": 19752
+    },
+    {
+      "epoch": 84.43,
+      "learning_rate": 8.475966594429188e-05,
+      "loss": 0.0001,
+      "step": 19756
+    },
+    {
+      "epoch": 84.44,
+      "learning_rate": 8.475128761992506e-05,
+      "loss": 0.0001,
+      "step": 19760
+    },
+    {
+      "epoch": 84.46,
+      "learning_rate": 8.474290740755113e-05,
+      "loss": 0.0001,
+      "step": 19764
+    },
+    {
+      "epoch": 84.48,
+      "learning_rate": 8.473452530762535e-05,
+      "loss": 0.0001,
+      "step": 19768
+    },
+    {
+      "epoch": 84.5,
+      "learning_rate": 8.472614132060314e-05,
+      "loss": 0.0016,
+      "step": 19772
+    },
+    {
+      "epoch": 84.51,
+      "learning_rate": 8.471775544693998e-05,
+      "loss": 0.0002,
+      "step": 19776
+    },
+    {
+      "epoch": 84.53,
+      "learning_rate": 8.470936768709146e-05,
+      "loss": 0.0,
+      "step": 19780
+    },
+    {
+      "epoch": 84.55,
+      "learning_rate": 8.47009780415133e-05,
+      "loss": 0.0004,
+      "step": 19784
+    },
+    {
+      "epoch": 84.56,
+      "learning_rate": 8.469258651066128e-05,
+      "loss": 0.0001,
+      "step": 19788
+    },
+    {
+      "epoch": 84.58,
+      "learning_rate": 8.468419309499131e-05,
+      "loss": 0.0002,
+      "step": 19792
+    },
+    {
+      "epoch": 84.6,
+      "learning_rate": 8.467579779495939e-05,
+      "loss": 0.0015,
+      "step": 19796
+    },
+    {
+      "epoch": 84.62,
+      "learning_rate": 8.466740061102166e-05,
+      "loss": 0.0,
+      "step": 19800
+    },
+    {
+      "epoch": 84.63,
+      "learning_rate": 8.465900154363431e-05,
+      "loss": 0.0003,
+      "step": 19804
+    },
+    {
+      "epoch": 84.65,
+      "learning_rate": 8.465060059325365e-05,
+      "loss": 0.0002,
+      "step": 19808
+    },
+    {
+      "epoch": 84.67,
+      "learning_rate": 8.464219776033611e-05,
+      "loss": 0.0001,
+      "step": 19812
+    },
+    {
+      "epoch": 84.68,
+      "learning_rate": 8.463379304533818e-05,
+      "loss": 0.0001,
+      "step": 19816
+    },
+    {
+      "epoch": 84.7,
+      "learning_rate": 8.462538644871653e-05,
+      "loss": 0.0008,
+      "step": 19820
+    },
+    {
+      "epoch": 84.72,
+      "learning_rate": 8.461697797092784e-05,
+      "loss": 0.0007,
+      "step": 19824
+    },
+    {
+      "epoch": 84.74,
+      "learning_rate": 8.460856761242894e-05,
+      "loss": 0.0,
+      "step": 19828
+    },
+    {
+      "epoch": 84.75,
+      "learning_rate": 8.46001553736768e-05,
+      "loss": 0.0,
+      "step": 19832
+    },
+    {
+      "epoch": 84.77,
+      "learning_rate": 8.459174125512838e-05,
+      "loss": 0.0003,
+      "step": 19836
+    },
+    {
+      "epoch": 84.79,
+      "learning_rate": 8.458332525724086e-05,
+      "loss": 0.001,
+      "step": 19840
+    },
+    {
+      "epoch": 84.8,
+      "learning_rate": 8.457490738047147e-05,
+      "loss": 0.0003,
+      "step": 19844
+    },
+    {
+      "epoch": 84.82,
+      "learning_rate": 8.456648762527755e-05,
+      "loss": 0.0003,
+      "step": 19848
+    },
+    {
+      "epoch": 84.84,
+      "learning_rate": 8.455806599211651e-05,
+      "loss": 0.0001,
+      "step": 19852
+    },
+    {
+      "epoch": 84.85,
+      "learning_rate": 8.45496424814459e-05,
+      "loss": 0.0001,
+      "step": 19856
+    },
+    {
+      "epoch": 84.87,
+      "learning_rate": 8.454121709372339e-05,
+      "loss": 0.0001,
+      "step": 19860
+    },
+    {
+      "epoch": 84.89,
+      "learning_rate": 8.453278982940667e-05,
+      "loss": 0.0013,
+      "step": 19864
+    },
+    {
+      "epoch": 84.91,
+      "learning_rate": 8.452436068895365e-05,
+      "loss": 0.0004,
+      "step": 19868
+    },
+    {
+      "epoch": 84.92,
+      "learning_rate": 8.451592967282222e-05,
+      "loss": 0.0017,
+      "step": 19872
+    },
+    {
+      "epoch": 84.94,
+      "learning_rate": 8.450749678147048e-05,
+      "loss": 0.0001,
+      "step": 19876
+    },
+    {
+      "epoch": 84.96,
+      "learning_rate": 8.449906201535653e-05,
+      "loss": 0.0004,
+      "step": 19880
+    },
+    {
+      "epoch": 84.97,
+      "learning_rate": 8.449062537493868e-05,
+      "loss": 0.0001,
+      "step": 19884
+    },
+    {
+      "epoch": 84.99,
+      "learning_rate": 8.448218686067524e-05,
+      "loss": 0.0003,
+      "step": 19888
+    },
+    {
+      "epoch": 85.01,
+      "learning_rate": 8.447374647302469e-05,
+      "loss": 0.0001,
+      "step": 19892
+    },
+    {
+      "epoch": 85.03,
+      "learning_rate": 8.446530421244557e-05,
+      "loss": 0.0001,
+      "step": 19896
+    },
+    {
+      "epoch": 85.04,
+      "learning_rate": 8.445686007939657e-05,
+      "loss": 0.0002,
+      "step": 19900
+    },
+    {
+      "epoch": 85.06,
+      "learning_rate": 8.444841407433644e-05,
+      "loss": 0.0008,
+      "step": 19904
+    },
+    {
+      "epoch": 85.08,
+      "learning_rate": 8.443996619772401e-05,
+      "loss": 0.0006,
+      "step": 19908
+    },
+    {
+      "epoch": 85.09,
+      "learning_rate": 8.44315164500183e-05,
+      "loss": 0.0001,
+      "step": 19912
+    },
+    {
+      "epoch": 85.11,
+      "learning_rate": 8.442306483167833e-05,
+      "loss": 0.0001,
+      "step": 19916
+    },
+    {
+      "epoch": 85.13,
+      "learning_rate": 8.44146113431633e-05,
+      "loss": 0.0001,
+      "step": 19920
+    },
+    {
+      "epoch": 85.15,
+      "learning_rate": 8.44061559849325e-05,
+      "loss": 0.0002,
+      "step": 19924
+    },
+    {
+      "epoch": 85.16,
+      "learning_rate": 8.439769875744524e-05,
+      "loss": 0.0003,
+      "step": 19928
+    },
+    {
+      "epoch": 85.18,
+      "learning_rate": 8.438923966116104e-05,
+      "loss": 0.0004,
+      "step": 19932
+    },
+    {
+      "epoch": 85.2,
+      "learning_rate": 8.438077869653946e-05,
+      "loss": 0.0001,
+      "step": 19936
+    },
+    {
+      "epoch": 85.21,
+      "learning_rate": 8.437231586404019e-05,
+      "loss": 0.0008,
+      "step": 19940
+    },
+    {
+      "epoch": 85.23,
+      "learning_rate": 8.4363851164123e-05,
+      "loss": 0.0004,
+      "step": 19944
+    },
+    {
+      "epoch": 85.25,
+      "learning_rate": 8.435538459724775e-05,
+      "loss": 0.0002,
+      "step": 19948
+    },
+    {
+      "epoch": 85.26,
+      "learning_rate": 8.434691616387446e-05,
+      "loss": 0.0022,
+      "step": 19952
+    },
+    {
+      "epoch": 85.28,
+      "learning_rate": 8.433844586446318e-05,
+      "loss": 0.0001,
+      "step": 19956
+    },
+    {
+      "epoch": 85.3,
+      "learning_rate": 8.43299736994741e-05,
+      "loss": 0.0002,
+      "step": 19960
+    },
+    {
+      "epoch": 85.32,
+      "learning_rate": 8.432149966936754e-05,
+      "loss": 0.0001,
+      "step": 19964
+    },
+    {
+      "epoch": 85.33,
+      "learning_rate": 8.431302377460383e-05,
+      "loss": 0.0001,
+      "step": 19968
+    },
+    {
+      "epoch": 85.35,
+      "learning_rate": 8.43045460156435e-05,
+      "loss": 0.0001,
+      "step": 19972
+    },
+    {
+      "epoch": 85.37,
+      "learning_rate": 8.429606639294711e-05,
+      "loss": 0.0,
+      "step": 19976
+    },
+    {
+      "epoch": 85.38,
+      "learning_rate": 8.428758490697538e-05,
+      "loss": 0.0007,
+      "step": 19980
+    },
+    {
+      "epoch": 85.4,
+      "learning_rate": 8.427910155818909e-05,
+      "loss": 0.0015,
+      "step": 19984
+    },
+    {
+      "epoch": 85.42,
+      "learning_rate": 8.427061634704911e-05,
+      "loss": 0.0001,
+      "step": 19988
+    },
+    {
+      "epoch": 85.44,
+      "learning_rate": 8.426212927401649e-05,
+      "loss": 0.0,
+      "step": 19992
+    },
+    {
+      "epoch": 85.45,
+      "learning_rate": 8.425364033955225e-05,
+      "loss": 0.0001,
+      "step": 19996
+    },
+    {
+      "epoch": 85.47,
+      "learning_rate": 8.424514954411767e-05,
+      "loss": 0.0024,
+      "step": 20000
+    },
+    {
+      "epoch": 85.47,
+      "eval_exact_match": 0.5031185031185031,
+      "eval_loss": 0.9105172753334045,
+      "eval_runtime": 140.8685,
+      "eval_samples_per_second": 6.829,
+      "step": 20000
+    },
+    {
+      "epoch": 85.49,
+      "learning_rate": 8.423665688817397e-05,
+      "loss": 0.0006,
+      "step": 20004
+    },
+    {
+      "epoch": 85.5,
+      "learning_rate": 8.422816237218259e-05,
+      "loss": 0.0013,
+      "step": 20008
+    },
+    {
+      "epoch": 85.52,
+      "learning_rate": 8.421966599660502e-05,
+      "loss": 0.0004,
+      "step": 20012
+    },
+    {
+      "epoch": 85.54,
+      "learning_rate": 8.421116776190288e-05,
+      "loss": 0.0003,
+      "step": 20016
+    },
+    {
+      "epoch": 85.56,
+      "learning_rate": 8.420266766853784e-05,
+      "loss": 0.0001,
+      "step": 20020
+    },
+    {
+      "epoch": 85.57,
+      "learning_rate": 8.419416571697171e-05,
+      "loss": 0.0001,
+      "step": 20024
+    },
+    {
+      "epoch": 85.59,
+      "learning_rate": 8.418566190766641e-05,
+      "loss": 0.0,
+      "step": 20028
+    },
+    {
+      "epoch": 85.61,
+      "learning_rate": 8.417715624108393e-05,
+      "loss": 0.0001,
+      "step": 20032
+    },
+    {
+      "epoch": 85.62,
+      "learning_rate": 8.416864871768639e-05,
+      "loss": 0.0001,
+      "step": 20036
+    },
+    {
+      "epoch": 85.64,
+      "learning_rate": 8.416013933793599e-05,
+      "loss": 0.0003,
+      "step": 20040
+    },
+    {
+      "epoch": 85.66,
+      "learning_rate": 8.415162810229502e-05,
+      "loss": 0.0001,
+      "step": 20044
+    },
+    {
+      "epoch": 85.68,
+      "learning_rate": 8.414311501122591e-05,
+      "loss": 0.0005,
+      "step": 20048
+    },
+    {
+      "epoch": 85.69,
+      "learning_rate": 8.413460006519116e-05,
+      "loss": 0.0002,
+      "step": 20052
+    },
+    {
+      "epoch": 85.71,
+      "learning_rate": 8.412608326465337e-05,
+      "loss": 0.0001,
+      "step": 20056
+    },
+    {
+      "epoch": 85.73,
+      "learning_rate": 8.411756461007527e-05,
+      "loss": 0.0001,
+      "step": 20060
+    },
+    {
+      "epoch": 85.74,
+      "learning_rate": 8.410904410191967e-05,
+      "loss": 0.0004,
+      "step": 20064
+    },
+    {
+      "epoch": 85.76,
+      "learning_rate": 8.410052174064946e-05,
+      "loss": 0.0011,
+      "step": 20068
+    },
+    {
+      "epoch": 85.78,
+      "learning_rate": 8.409199752672767e-05,
+      "loss": 0.0002,
+      "step": 20072
+    },
+    {
+      "epoch": 85.79,
+      "learning_rate": 8.408347146061741e-05,
+      "loss": 0.0,
+      "step": 20076
+    },
+    {
+      "epoch": 85.81,
+      "learning_rate": 8.40749435427819e-05,
+      "loss": 0.0001,
+      "step": 20080
+    },
+    {
+      "epoch": 85.83,
+      "learning_rate": 8.406641377368446e-05,
+      "loss": 0.0,
+      "step": 20084
+    },
+    {
+      "epoch": 85.85,
+      "learning_rate": 8.405788215378847e-05,
+      "loss": 0.0007,
+      "step": 20088
+    },
+    {
+      "epoch": 85.86,
+      "learning_rate": 8.404934868355747e-05,
+      "loss": 0.0002,
+      "step": 20092
+    },
+    {
+      "epoch": 85.88,
+      "learning_rate": 8.404081336345507e-05,
+      "loss": 0.0003,
+      "step": 20096
+    },
+    {
+      "epoch": 85.9,
+      "learning_rate": 8.4032276193945e-05,
+      "loss": 0.0005,
+      "step": 20100
+    },
+    {
+      "epoch": 85.91,
+      "learning_rate": 8.402373717549105e-05,
+      "loss": 0.0008,
+      "step": 20104
+    },
+    {
+      "epoch": 85.93,
+      "learning_rate": 8.401519630855718e-05,
+      "loss": 0.0008,
+      "step": 20108
+    },
+    {
+      "epoch": 85.95,
+      "learning_rate": 8.400665359360737e-05,
+      "loss": 0.0006,
+      "step": 20112
+    },
+    {
+      "epoch": 85.97,
+      "learning_rate": 8.399810903110575e-05,
+      "loss": 0.0001,
+      "step": 20116
+    },
+    {
+      "epoch": 85.98,
+      "learning_rate": 8.398956262151654e-05,
+      "loss": 0.0001,
+      "step": 20120
+    },
+    {
+      "epoch": 86.0,
+      "learning_rate": 8.398101436530405e-05,
+      "loss": 0.0001,
+      "step": 20124
+    },
+    {
+      "epoch": 86.02,
+      "learning_rate": 8.397246426293273e-05,
+      "loss": 0.0001,
+      "step": 20128
+    },
+    {
+      "epoch": 86.03,
+      "learning_rate": 8.396391231486707e-05,
+      "loss": 0.0001,
+      "step": 20132
+    },
+    {
+      "epoch": 86.05,
+      "learning_rate": 8.395535852157168e-05,
+      "loss": 0.0001,
+      "step": 20136
+    },
+    {
+      "epoch": 86.07,
+      "learning_rate": 8.394680288351132e-05,
+      "loss": 0.0,
+      "step": 20140
+    },
+    {
+      "epoch": 86.09,
+      "learning_rate": 8.393824540115076e-05,
+      "loss": 0.0002,
+      "step": 20144
+    },
+    {
+      "epoch": 86.1,
+      "learning_rate": 8.392968607495497e-05,
+      "loss": 0.0002,
+      "step": 20148
+    },
+    {
+      "epoch": 86.12,
+      "learning_rate": 8.392112490538894e-05,
+      "loss": 0.0001,
+      "step": 20152
+    },
+    {
+      "epoch": 86.14,
+      "learning_rate": 8.391256189291779e-05,
+      "loss": 0.0008,
+      "step": 20156
+    },
+    {
+      "epoch": 86.15,
+      "learning_rate": 8.390399703800678e-05,
+      "loss": 0.0002,
+      "step": 20160
+    },
+    {
+      "epoch": 86.17,
+      "learning_rate": 8.389543034112117e-05,
+      "loss": 0.0001,
+      "step": 20164
+    },
+    {
+      "epoch": 86.19,
+      "learning_rate": 8.388686180272643e-05,
+      "loss": 0.0003,
+      "step": 20168
+    },
+    {
+      "epoch": 86.21,
+      "learning_rate": 8.387829142328807e-05,
+      "loss": 0.0001,
+      "step": 20172
+    },
+    {
+      "epoch": 86.22,
+      "learning_rate": 8.386971920327169e-05,
+      "loss": 0.0002,
+      "step": 20176
+    },
+    {
+      "epoch": 86.24,
+      "learning_rate": 8.386114514314303e-05,
+      "loss": 0.0001,
+      "step": 20180
+    },
+    {
+      "epoch": 86.26,
+      "learning_rate": 8.385256924336792e-05,
+      "loss": 0.0002,
+      "step": 20184
+    },
+    {
+      "epoch": 86.27,
+      "learning_rate": 8.384399150441225e-05,
+      "loss": 0.0003,
+      "step": 20188
+    },
+    {
+      "epoch": 86.29,
+      "learning_rate": 8.383541192674207e-05,
+      "loss": 0.0001,
+      "step": 20192
+    },
+    {
+      "epoch": 86.31,
+      "learning_rate": 8.382683051082349e-05,
+      "loss": 0.0001,
+      "step": 20196
+    },
+    {
+      "epoch": 86.32,
+      "learning_rate": 8.381824725712272e-05,
+      "loss": 0.0,
+      "step": 20200
+    },
+    {
+      "epoch": 86.34,
+      "learning_rate": 8.380966216610612e-05,
+      "loss": 0.0003,
+      "step": 20204
+    },
+    {
+      "epoch": 86.36,
+      "learning_rate": 8.380107523824008e-05,
+      "loss": 0.0001,
+      "step": 20208
+    },
+    {
+      "epoch": 86.38,
+      "learning_rate": 8.379248647399111e-05,
+      "loss": 0.0,
+      "step": 20212
+    },
+    {
+      "epoch": 86.39,
+      "learning_rate": 8.378389587382584e-05,
+      "loss": 0.0,
+      "step": 20216
+    },
+    {
+      "epoch": 86.41,
+      "learning_rate": 8.3775303438211e-05,
+      "loss": 0.0006,
+      "step": 20220
+    },
+    {
+      "epoch": 86.43,
+      "learning_rate": 8.37667091676134e-05,
+      "loss": 0.0001,
+      "step": 20224
+    },
+    {
+      "epoch": 86.44,
+      "learning_rate": 8.375811306249996e-05,
+      "loss": 0.0001,
+      "step": 20228
+    },
+    {
+      "epoch": 86.46,
+      "learning_rate": 8.37495151233377e-05,
+      "loss": 0.0004,
+      "step": 20232
+    },
+    {
+      "epoch": 86.48,
+      "learning_rate": 8.374091535059374e-05,
+      "loss": 0.0001,
+      "step": 20236
+    },
+    {
+      "epoch": 86.5,
+      "learning_rate": 8.373231374473531e-05,
+      "loss": 0.0001,
+      "step": 20240
+    },
+    {
+      "epoch": 86.51,
+      "learning_rate": 8.37237103062297e-05,
+      "loss": 0.0006,
+      "step": 20244
+    },
+    {
+      "epoch": 86.53,
+      "learning_rate": 8.371510503554436e-05,
+      "loss": 0.0001,
+      "step": 20248
+    },
+    {
+      "epoch": 86.55,
+      "learning_rate": 8.370649793314678e-05,
+      "loss": 0.0001,
+      "step": 20252
+    },
+    {
+      "epoch": 86.56,
+      "learning_rate": 8.369788899950457e-05,
+      "loss": 0.0019,
+      "step": 20256
+    },
+    {
+      "epoch": 86.58,
+      "learning_rate": 8.368927823508548e-05,
+      "loss": 0.0001,
+      "step": 20260
+    },
+    {
+      "epoch": 86.6,
+      "learning_rate": 8.36806656403573e-05,
+      "loss": 0.0001,
+      "step": 20264
+    },
+    {
+      "epoch": 86.62,
+      "learning_rate": 8.367205121578797e-05,
+      "loss": 0.0,
+      "step": 20268
+    },
+    {
+      "epoch": 86.63,
+      "learning_rate": 8.366343496184546e-05,
+      "loss": 0.0001,
+      "step": 20272
+    },
+    {
+      "epoch": 86.65,
+      "learning_rate": 8.365481687899793e-05,
+      "loss": 0.0016,
+      "step": 20276
+    },
+    {
+      "epoch": 86.67,
+      "learning_rate": 8.364619696771355e-05,
+      "loss": 0.0,
+      "step": 20280
+    },
+    {
+      "epoch": 86.68,
+      "learning_rate": 8.363757522846066e-05,
+      "loss": 0.0002,
+      "step": 20284
+    },
+    {
+      "epoch": 86.7,
+      "learning_rate": 8.362895166170768e-05,
+      "loss": 0.0003,
+      "step": 20288
+    },
+    {
+      "epoch": 86.72,
+      "learning_rate": 8.362032626792308e-05,
+      "loss": 0.0005,
+      "step": 20292
+    },
+    {
+      "epoch": 86.74,
+      "learning_rate": 8.361169904757553e-05,
+      "loss": 0.0002,
+      "step": 20296
+    },
+    {
+      "epoch": 86.75,
+      "learning_rate": 8.360307000113369e-05,
+      "loss": 0.0002,
+      "step": 20300
+    },
+    {
+      "epoch": 86.77,
+      "learning_rate": 8.359443912906639e-05,
+      "loss": 0.0001,
+      "step": 20304
+    },
+    {
+      "epoch": 86.79,
+      "learning_rate": 8.358580643184253e-05,
+      "loss": 0.0005,
+      "step": 20308
+    },
+    {
+      "epoch": 86.8,
+      "learning_rate": 8.357717190993113e-05,
+      "loss": 0.0003,
+      "step": 20312
+    },
+    {
+      "epoch": 86.82,
+      "learning_rate": 8.356853556380127e-05,
+      "loss": 0.0,
+      "step": 20316
+    },
+    {
+      "epoch": 86.84,
+      "learning_rate": 8.355989739392217e-05,
+      "loss": 0.0012,
+      "step": 20320
+    },
+    {
+      "epoch": 86.85,
+      "learning_rate": 8.355125740076315e-05,
+      "loss": 0.0008,
+      "step": 20324
+    },
+    {
+      "epoch": 86.87,
+      "learning_rate": 8.35426155847936e-05,
+      "loss": 0.0008,
+      "step": 20328
+    },
+    {
+      "epoch": 86.89,
+      "learning_rate": 8.3533971946483e-05,
+      "loss": 0.0018,
+      "step": 20332
+    },
+    {
+      "epoch": 86.91,
+      "learning_rate": 8.352532648630098e-05,
+      "loss": 0.0009,
+      "step": 20336
+    },
+    {
+      "epoch": 86.92,
+      "learning_rate": 8.351667920471723e-05,
+      "loss": 0.0002,
+      "step": 20340
+    },
+    {
+      "epoch": 86.94,
+      "learning_rate": 8.350803010220156e-05,
+      "loss": 0.0003,
+      "step": 20344
+    },
+    {
+      "epoch": 86.96,
+      "learning_rate": 8.349937917922386e-05,
+      "loss": 0.0001,
+      "step": 20348
+    },
+    {
+      "epoch": 86.97,
+      "learning_rate": 8.349072643625412e-05,
+      "loss": 0.0001,
+      "step": 20352
+    },
+    {
+      "epoch": 86.99,
+      "learning_rate": 8.348207187376245e-05,
+      "loss": 0.0001,
+      "step": 20356
+    },
+    {
+      "epoch": 87.01,
+      "learning_rate": 8.347341549221903e-05,
+      "loss": 0.0001,
+      "step": 20360
+    },
+    {
+      "epoch": 87.03,
+      "learning_rate": 8.346475729209416e-05,
+      "loss": 0.0001,
+      "step": 20364
+    },
+    {
+      "epoch": 87.04,
+      "learning_rate": 8.345609727385825e-05,
+      "loss": 0.0002,
+      "step": 20368
+    },
+    {
+      "epoch": 87.06,
+      "learning_rate": 8.344743543798176e-05,
+      "loss": 0.0001,
+      "step": 20372
+    },
+    {
+      "epoch": 87.08,
+      "learning_rate": 8.343877178493529e-05,
+      "loss": 0.0001,
+      "step": 20376
+    },
+    {
+      "epoch": 87.09,
+      "learning_rate": 8.343010631518955e-05,
+      "loss": 0.0003,
+      "step": 20380
+    },
+    {
+      "epoch": 87.11,
+      "learning_rate": 8.342143902921531e-05,
+      "loss": 0.0,
+      "step": 20384
+    },
+    {
+      "epoch": 87.13,
+      "learning_rate": 8.341276992748344e-05,
+      "loss": 0.0001,
+      "step": 20388
+    },
+    {
+      "epoch": 87.15,
+      "learning_rate": 8.340409901046496e-05,
+      "loss": 0.0001,
+      "step": 20392
+    },
+    {
+      "epoch": 87.16,
+      "learning_rate": 8.339542627863093e-05,
+      "loss": 0.0009,
+      "step": 20396
+    },
+    {
+      "epoch": 87.18,
+      "learning_rate": 8.338675173245254e-05,
+      "loss": 0.0001,
+      "step": 20400
+    },
+    {
+      "epoch": 87.2,
+      "learning_rate": 8.337807537240106e-05,
+      "loss": 0.0002,
+      "step": 20404
+    },
+    {
+      "epoch": 87.21,
+      "learning_rate": 8.336939719894788e-05,
+      "loss": 0.0002,
+      "step": 20408
+    },
+    {
+      "epoch": 87.23,
+      "learning_rate": 8.336071721256447e-05,
+      "loss": 0.0003,
+      "step": 20412
+    },
+    {
+      "epoch": 87.25,
+      "learning_rate": 8.335203541372242e-05,
+      "loss": 0.0,
+      "step": 20416
+    },
+    {
+      "epoch": 87.26,
+      "learning_rate": 8.33433518028934e-05,
+      "loss": 0.0005,
+      "step": 20420
+    },
+    {
+      "epoch": 87.28,
+      "learning_rate": 8.333466638054916e-05,
+      "loss": 0.0003,
+      "step": 20424
+    },
+    {
+      "epoch": 87.3,
+      "learning_rate": 8.33259791471616e-05,
+      "loss": 0.0003,
+      "step": 20428
+    },
+    {
+      "epoch": 87.32,
+      "learning_rate": 8.331729010320267e-05,
+      "loss": 0.0004,
+      "step": 20432
+    },
+    {
+      "epoch": 87.33,
+      "learning_rate": 8.330859924914444e-05,
+      "loss": 0.0005,
+      "step": 20436
+    },
+    {
+      "epoch": 87.35,
+      "learning_rate": 8.329990658545912e-05,
+      "loss": 0.0011,
+      "step": 20440
+    },
+    {
+      "epoch": 87.37,
+      "learning_rate": 8.329121211261892e-05,
+      "loss": 0.0005,
+      "step": 20444
+    },
+    {
+      "epoch": 87.38,
+      "learning_rate": 8.328251583109621e-05,
+      "loss": 0.0001,
+      "step": 20448
+    },
+    {
+      "epoch": 87.4,
+      "learning_rate": 8.327381774136347e-05,
+      "loss": 0.0,
+      "step": 20452
+    },
+    {
+      "epoch": 87.42,
+      "learning_rate": 8.326511784389326e-05,
+      "loss": 0.0001,
+      "step": 20456
+    },
+    {
+      "epoch": 87.44,
+      "learning_rate": 8.325641613915822e-05,
+      "loss": 0.0002,
+      "step": 20460
+    },
+    {
+      "epoch": 87.45,
+      "learning_rate": 8.324771262763114e-05,
+      "loss": 0.0003,
+      "step": 20464
+    },
+    {
+      "epoch": 87.47,
+      "learning_rate": 8.323900730978482e-05,
+      "loss": 0.0016,
+      "step": 20468
+    },
+    {
+      "epoch": 87.49,
+      "learning_rate": 8.323030018609225e-05,
+      "loss": 0.0001,
+      "step": 20472
+    },
+    {
+      "epoch": 87.5,
+      "learning_rate": 8.322159125702649e-05,
+      "loss": 0.0,
+      "step": 20476
+    },
+    {
+      "epoch": 87.52,
+      "learning_rate": 8.321288052306066e-05,
+      "loss": 0.0002,
+      "step": 20480
+    },
+    {
+      "epoch": 87.54,
+      "learning_rate": 8.320416798466803e-05,
+      "loss": 0.0,
+      "step": 20484
+    },
+    {
+      "epoch": 87.56,
+      "learning_rate": 8.319545364232193e-05,
+      "loss": 0.0001,
+      "step": 20488
+    },
+    {
+      "epoch": 87.57,
+      "learning_rate": 8.31867374964958e-05,
+      "loss": 0.0001,
+      "step": 20492
+    },
+    {
+      "epoch": 87.59,
+      "learning_rate": 8.317801954766318e-05,
+      "loss": 0.0007,
+      "step": 20496
+    },
+    {
+      "epoch": 87.61,
+      "learning_rate": 8.316929979629773e-05,
+      "loss": 0.0001,
+      "step": 20500
+    },
+    {
+      "epoch": 87.62,
+      "learning_rate": 8.316057824287315e-05,
+      "loss": 0.0004,
+      "step": 20504
+    },
+    {
+      "epoch": 87.64,
+      "learning_rate": 8.315185488786332e-05,
+      "loss": 0.0,
+      "step": 20508
+    },
+    {
+      "epoch": 87.66,
+      "learning_rate": 8.314312973174214e-05,
+      "loss": 0.0013,
+      "step": 20512
+    },
+    {
+      "epoch": 87.68,
+      "learning_rate": 8.313440277498366e-05,
+      "loss": 0.0004,
+      "step": 20516
+    },
+    {
+      "epoch": 87.69,
+      "learning_rate": 8.312567401806197e-05,
+      "loss": 0.0005,
+      "step": 20520
+    },
+    {
+      "epoch": 87.71,
+      "learning_rate": 8.311694346145133e-05,
+      "loss": 0.0,
+      "step": 20524
+    },
+    {
+      "epoch": 87.73,
+      "learning_rate": 8.310821110562608e-05,
+      "loss": 0.0004,
+      "step": 20528
+    },
+    {
+      "epoch": 87.74,
+      "learning_rate": 8.30994769510606e-05,
+      "loss": 0.0,
+      "step": 20532
+    },
+    {
+      "epoch": 87.76,
+      "learning_rate": 8.309074099822942e-05,
+      "loss": 0.0002,
+      "step": 20536
+    },
+    {
+      "epoch": 87.78,
+      "learning_rate": 8.308200324760717e-05,
+      "loss": 0.0003,
+      "step": 20540
+    },
+    {
+      "epoch": 87.79,
+      "learning_rate": 8.307326369966854e-05,
+      "loss": 0.0001,
+      "step": 20544
+    },
+    {
+      "epoch": 87.81,
+      "learning_rate": 8.306452235488839e-05,
+      "loss": 0.0004,
+      "step": 20548
+    },
+    {
+      "epoch": 87.83,
+      "learning_rate": 8.305577921374158e-05,
+      "loss": 0.0001,
+      "step": 20552
+    },
+    {
+      "epoch": 87.85,
+      "learning_rate": 8.304703427670313e-05,
+      "loss": 0.0001,
+      "step": 20556
+    },
+    {
+      "epoch": 87.86,
+      "learning_rate": 8.303828754424816e-05,
+      "loss": 0.0001,
+      "step": 20560
+    },
+    {
+      "epoch": 87.88,
+      "learning_rate": 8.302953901685187e-05,
+      "loss": 0.0,
+      "step": 20564
+    },
+    {
+      "epoch": 87.9,
+      "learning_rate": 8.302078869498955e-05,
+      "loss": 0.0015,
+      "step": 20568
+    },
+    {
+      "epoch": 87.91,
+      "learning_rate": 8.301203657913658e-05,
+      "loss": 0.0001,
+      "step": 20572
+    },
+    {
+      "epoch": 87.93,
+      "learning_rate": 8.300328266976848e-05,
+      "loss": 0.0,
+      "step": 20576
+    },
+    {
+      "epoch": 87.95,
+      "learning_rate": 8.299452696736086e-05,
+      "loss": 0.0002,
+      "step": 20580
+    },
+    {
+      "epoch": 87.97,
+      "learning_rate": 8.298576947238938e-05,
+      "loss": 0.0001,
+      "step": 20584
+    },
+    {
+      "epoch": 87.98,
+      "learning_rate": 8.297701018532982e-05,
+      "loss": 0.0008,
+      "step": 20588
+    },
+    {
+      "epoch": 88.0,
+      "learning_rate": 8.29682491066581e-05,
+      "loss": 0.0002,
+      "step": 20592
+    },
+    {
+      "epoch": 88.02,
+      "learning_rate": 8.295948623685014e-05,
+      "loss": 0.0004,
+      "step": 20596
+    },
+    {
+      "epoch": 88.03,
+      "learning_rate": 8.29507215763821e-05,
+      "loss": 0.0005,
+      "step": 20600
+    },
+    {
+      "epoch": 88.05,
+      "learning_rate": 8.294195512573011e-05,
+      "loss": 0.0001,
+      "step": 20604
+    },
+    {
+      "epoch": 88.07,
+      "learning_rate": 8.293318688537043e-05,
+      "loss": 0.0001,
+      "step": 20608
+    },
+    {
+      "epoch": 88.09,
+      "learning_rate": 8.292441685577946e-05,
+      "loss": 0.0,
+      "step": 20612
+    },
+    {
+      "epoch": 88.1,
+      "learning_rate": 8.291564503743365e-05,
+      "loss": 0.0001,
+      "step": 20616
+    },
+    {
+      "epoch": 88.12,
+      "learning_rate": 8.29068714308096e-05,
+      "loss": 0.0001,
+      "step": 20620
+    },
+    {
+      "epoch": 88.14,
+      "learning_rate": 8.289809603638391e-05,
+      "loss": 0.0,
+      "step": 20624
+    },
+    {
+      "epoch": 88.15,
+      "learning_rate": 8.28893188546334e-05,
+      "loss": 0.0001,
+      "step": 20628
+    },
+    {
+      "epoch": 88.17,
+      "learning_rate": 8.28805398860349e-05,
+      "loss": 0.0,
+      "step": 20632
+    },
+    {
+      "epoch": 88.19,
+      "learning_rate": 8.287175913106535e-05,
+      "loss": 0.0,
+      "step": 20636
+    },
+    {
+      "epoch": 88.21,
+      "learning_rate": 8.286297659020183e-05,
+      "loss": 0.0001,
+      "step": 20640
+    },
+    {
+      "epoch": 88.22,
+      "learning_rate": 8.285419226392148e-05,
+      "loss": 0.0,
+      "step": 20644
+    },
+    {
+      "epoch": 88.24,
+      "learning_rate": 8.284540615270152e-05,
+      "loss": 0.0001,
+      "step": 20648
+    },
+    {
+      "epoch": 88.26,
+      "learning_rate": 8.283661825701933e-05,
+      "loss": 0.0001,
+      "step": 20652
+    },
+    {
+      "epoch": 88.27,
+      "learning_rate": 8.28278285773523e-05,
+      "loss": 0.0001,
+      "step": 20656
+    },
+    {
+      "epoch": 88.29,
+      "learning_rate": 8.281903711417802e-05,
+      "loss": 0.0001,
+      "step": 20660
+    },
+    {
+      "epoch": 88.31,
+      "learning_rate": 8.28102438679741e-05,
+      "loss": 0.0003,
+      "step": 20664
+    },
+    {
+      "epoch": 88.32,
+      "learning_rate": 8.280144883921827e-05,
+      "loss": 0.0001,
+      "step": 20668
+    },
+    {
+      "epoch": 88.34,
+      "learning_rate": 8.279265202838833e-05,
+      "loss": 0.0001,
+      "step": 20672
+    },
+    {
+      "epoch": 88.36,
+      "learning_rate": 8.278385343596225e-05,
+      "loss": 0.0001,
+      "step": 20676
+    },
+    {
+      "epoch": 88.38,
+      "learning_rate": 8.277505306241799e-05,
+      "loss": 0.0002,
+      "step": 20680
+    },
+    {
+      "epoch": 88.39,
+      "learning_rate": 8.276625090823374e-05,
+      "loss": 0.0,
+      "step": 20684
+    },
+    {
+      "epoch": 88.41,
+      "learning_rate": 8.275744697388766e-05,
+      "loss": 0.0,
+      "step": 20688
+    },
+    {
+      "epoch": 88.43,
+      "learning_rate": 8.274864125985808e-05,
+      "loss": 0.0,
+      "step": 20692
+    },
+    {
+      "epoch": 88.44,
+      "learning_rate": 8.273983376662339e-05,
+      "loss": 0.0005,
+      "step": 20696
+    },
+    {
+      "epoch": 88.46,
+      "learning_rate": 8.273102449466212e-05,
+      "loss": 0.0,
+      "step": 20700
+    },
+    {
+      "epoch": 88.48,
+      "learning_rate": 8.272221344445285e-05,
+      "loss": 0.0,
+      "step": 20704
+    },
+    {
+      "epoch": 88.5,
+      "learning_rate": 8.27134006164743e-05,
+      "loss": 0.0,
+      "step": 20708
+    },
+    {
+      "epoch": 88.51,
+      "learning_rate": 8.270458601120523e-05,
+      "loss": 0.0001,
+      "step": 20712
+    },
+    {
+      "epoch": 88.53,
+      "learning_rate": 8.269576962912456e-05,
+      "loss": 0.0,
+      "step": 20716
+    },
+    {
+      "epoch": 88.55,
+      "learning_rate": 8.268695147071123e-05,
+      "loss": 0.0001,
+      "step": 20720
+    },
+    {
+      "epoch": 88.56,
+      "learning_rate": 8.267813153644439e-05,
+      "loss": 0.0003,
+      "step": 20724
+    },
+    {
+      "epoch": 88.58,
+      "learning_rate": 8.266930982680317e-05,
+      "loss": 0.0017,
+      "step": 20728
+    },
+    {
+      "epoch": 88.6,
+      "learning_rate": 8.266048634226686e-05,
+      "loss": 0.0,
+      "step": 20732
+    },
+    {
+      "epoch": 88.62,
+      "learning_rate": 8.265166108331483e-05,
+      "loss": 0.0001,
+      "step": 20736
+    },
+    {
+      "epoch": 88.63,
+      "learning_rate": 8.264283405042657e-05,
+      "loss": 0.0,
+      "step": 20740
+    },
+    {
+      "epoch": 88.65,
+      "learning_rate": 8.26340052440816e-05,
+      "loss": 0.0,
+      "step": 20744
+    },
+    {
+      "epoch": 88.67,
+      "learning_rate": 8.262517466475963e-05,
+      "loss": 0.0,
+      "step": 20748
+    },
+    {
+      "epoch": 88.68,
+      "learning_rate": 8.261634231294037e-05,
+      "loss": 0.0,
+      "step": 20752
+    },
+    {
+      "epoch": 88.7,
+      "learning_rate": 8.260750818910372e-05,
+      "loss": 0.0,
+      "step": 20756
+    },
+    {
+      "epoch": 88.72,
+      "learning_rate": 8.259867229372961e-05,
+      "loss": 0.0001,
+      "step": 20760
+    },
+    {
+      "epoch": 88.74,
+      "learning_rate": 8.258983462729808e-05,
+      "loss": 0.0003,
+      "step": 20764
+    },
+    {
+      "epoch": 88.75,
+      "learning_rate": 8.258099519028928e-05,
+      "loss": 0.0,
+      "step": 20768
+    },
+    {
+      "epoch": 88.77,
+      "learning_rate": 8.257215398318344e-05,
+      "loss": 0.0002,
+      "step": 20772
+    },
+    {
+      "epoch": 88.79,
+      "learning_rate": 8.25633110064609e-05,
+      "loss": 0.0,
+      "step": 20776
+    },
+    {
+      "epoch": 88.8,
+      "learning_rate": 8.25544662606021e-05,
+      "loss": 0.0001,
+      "step": 20780
+    },
+    {
+      "epoch": 88.82,
+      "learning_rate": 8.254561974608757e-05,
+      "loss": 0.0002,
+      "step": 20784
+    },
+    {
+      "epoch": 88.84,
+      "learning_rate": 8.253677146339791e-05,
+      "loss": 0.0001,
+      "step": 20788
+    },
+    {
+      "epoch": 88.85,
+      "learning_rate": 8.252792141301386e-05,
+      "loss": 0.0012,
+      "step": 20792
+    },
+    {
+      "epoch": 88.87,
+      "learning_rate": 8.251906959541625e-05,
+      "loss": 0.0,
+      "step": 20796
+    },
+    {
+      "epoch": 88.89,
+      "learning_rate": 8.251021601108593e-05,
+      "loss": 0.0,
+      "step": 20800
+    },
+    {
+      "epoch": 88.91,
+      "learning_rate": 8.250136066050397e-05,
+      "loss": 0.0,
+      "step": 20804
+    },
+    {
+      "epoch": 88.92,
+      "learning_rate": 8.249250354415145e-05,
+      "loss": 0.001,
+      "step": 20808
+    },
+    {
+      "epoch": 88.94,
+      "learning_rate": 8.248364466250959e-05,
+      "loss": 0.0001,
+      "step": 20812
+    },
+    {
+      "epoch": 88.96,
+      "learning_rate": 8.247478401605963e-05,
+      "loss": 0.0001,
+      "step": 20816
+    },
+    {
+      "epoch": 88.97,
+      "learning_rate": 8.246592160528302e-05,
+      "loss": 0.0001,
+      "step": 20820
+    },
+    {
+      "epoch": 88.99,
+      "learning_rate": 8.245705743066123e-05,
+      "loss": 0.0002,
+      "step": 20824
+    },
+    {
+      "epoch": 89.01,
+      "learning_rate": 8.244819149267584e-05,
+      "loss": 0.0013,
+      "step": 20828
+    },
+    {
+      "epoch": 89.03,
+      "learning_rate": 8.24393237918085e-05,
+      "loss": 0.0003,
+      "step": 20832
+    },
+    {
+      "epoch": 89.04,
+      "learning_rate": 8.243045432854102e-05,
+      "loss": 0.0001,
+      "step": 20836
+    },
+    {
+      "epoch": 89.06,
+      "learning_rate": 8.242158310335528e-05,
+      "loss": 0.0,
+      "step": 20840
+    },
+    {
+      "epoch": 89.08,
+      "learning_rate": 8.241271011673322e-05,
+      "loss": 0.0001,
+      "step": 20844
+    },
+    {
+      "epoch": 89.09,
+      "learning_rate": 8.24038353691569e-05,
+      "loss": 0.0001,
+      "step": 20848
+    },
+    {
+      "epoch": 89.11,
+      "learning_rate": 8.239495886110848e-05,
+      "loss": 0.0002,
+      "step": 20852
+    },
+    {
+      "epoch": 89.13,
+      "learning_rate": 8.238608059307023e-05,
+      "loss": 0.0001,
+      "step": 20856
+    },
+    {
+      "epoch": 89.15,
+      "learning_rate": 8.237720056552449e-05,
+      "loss": 0.0001,
+      "step": 20860
+    },
+    {
+      "epoch": 89.16,
+      "learning_rate": 8.23683187789537e-05,
+      "loss": 0.0,
+      "step": 20864
+    },
+    {
+      "epoch": 89.18,
+      "learning_rate": 8.23594352338404e-05,
+      "loss": 0.0,
+      "step": 20868
+    },
+    {
+      "epoch": 89.2,
+      "learning_rate": 8.235054993066722e-05,
+      "loss": 0.0,
+      "step": 20872
+    },
+    {
+      "epoch": 89.21,
+      "learning_rate": 8.23416628699169e-05,
+      "loss": 0.0,
+      "step": 20876
+    },
+    {
+      "epoch": 89.23,
+      "learning_rate": 8.233277405207227e-05,
+      "loss": 0.0001,
+      "step": 20880
+    },
+    {
+      "epoch": 89.25,
+      "learning_rate": 8.232388347761624e-05,
+      "loss": 0.0021,
+      "step": 20884
+    },
+    {
+      "epoch": 89.26,
+      "learning_rate": 8.231499114703184e-05,
+      "loss": 0.0004,
+      "step": 20888
+    },
+    {
+      "epoch": 89.28,
+      "learning_rate": 8.230609706080217e-05,
+      "loss": 0.0004,
+      "step": 20892
+    },
+    {
+      "epoch": 89.3,
+      "learning_rate": 8.229720121941043e-05,
+      "loss": 0.0001,
+      "step": 20896
+    },
+    {
+      "epoch": 89.32,
+      "learning_rate": 8.228830362333995e-05,
+      "loss": 0.0,
+      "step": 20900
+    },
+    {
+      "epoch": 89.33,
+      "learning_rate": 8.227940427307413e-05,
+      "loss": 0.0001,
+      "step": 20904
+    },
+    {
+      "epoch": 89.35,
+      "learning_rate": 8.227050316909642e-05,
+      "loss": 0.0001,
+      "step": 20908
+    },
+    {
+      "epoch": 89.37,
+      "learning_rate": 8.226160031189045e-05,
+      "loss": 0.0,
+      "step": 20912
+    },
+    {
+      "epoch": 89.38,
+      "learning_rate": 8.225269570193989e-05,
+      "loss": 0.0001,
+      "step": 20916
+    },
+    {
+      "epoch": 89.4,
+      "learning_rate": 8.224378933972851e-05,
+      "loss": 0.0,
+      "step": 20920
+    },
+    {
+      "epoch": 89.42,
+      "learning_rate": 8.223488122574022e-05,
+      "loss": 0.0,
+      "step": 20924
+    },
+    {
+      "epoch": 89.44,
+      "learning_rate": 8.222597136045895e-05,
+      "loss": 0.0,
+      "step": 20928
+    },
+    {
+      "epoch": 89.45,
+      "learning_rate": 8.221705974436881e-05,
+      "loss": 0.0001,
+      "step": 20932
+    },
+    {
+      "epoch": 89.47,
+      "learning_rate": 8.22081463779539e-05,
+      "loss": 0.0,
+      "step": 20936
+    },
+    {
+      "epoch": 89.49,
+      "learning_rate": 8.219923126169853e-05,
+      "loss": 0.0002,
+      "step": 20940
+    },
+    {
+      "epoch": 89.5,
+      "learning_rate": 8.219031439608702e-05,
+      "loss": 0.0003,
+      "step": 20944
+    },
+    {
+      "epoch": 89.52,
+      "learning_rate": 8.218139578160382e-05,
+      "loss": 0.0009,
+      "step": 20948
+    },
+    {
+      "epoch": 89.54,
+      "learning_rate": 8.217247541873347e-05,
+      "loss": 0.0001,
+      "step": 20952
+    },
+    {
+      "epoch": 89.56,
+      "learning_rate": 8.216355330796061e-05,
+      "loss": 0.0006,
+      "step": 20956
+    },
+    {
+      "epoch": 89.57,
+      "learning_rate": 8.215462944976999e-05,
+      "loss": 0.0,
+      "step": 20960
+    },
+    {
+      "epoch": 89.59,
+      "learning_rate": 8.21457038446464e-05,
+      "loss": 0.0002,
+      "step": 20964
+    },
+    {
+      "epoch": 89.61,
+      "learning_rate": 8.213677649307478e-05,
+      "loss": 0.0002,
+      "step": 20968
+    },
+    {
+      "epoch": 89.62,
+      "learning_rate": 8.212784739554015e-05,
+      "loss": 0.0,
+      "step": 20972
+    },
+    {
+      "epoch": 89.64,
+      "learning_rate": 8.21189165525276e-05,
+      "loss": 0.0006,
+      "step": 20976
+    },
+    {
+      "epoch": 89.66,
+      "learning_rate": 8.210998396452235e-05,
+      "loss": 0.0,
+      "step": 20980
+    },
+    {
+      "epoch": 89.68,
+      "learning_rate": 8.21010496320097e-05,
+      "loss": 0.0,
+      "step": 20984
+    },
+    {
+      "epoch": 89.69,
+      "learning_rate": 8.209211355547504e-05,
+      "loss": 0.0,
+      "step": 20988
+    },
+    {
+      "epoch": 89.71,
+      "learning_rate": 8.208317573540386e-05,
+      "loss": 0.0,
+      "step": 20992
+    },
+    {
+      "epoch": 89.73,
+      "learning_rate": 8.207423617228174e-05,
+      "loss": 0.0,
+      "step": 20996
+    },
+    {
+      "epoch": 89.74,
+      "learning_rate": 8.206529486659435e-05,
+      "loss": 0.0001,
+      "step": 21000
+    },
+    {
+      "epoch": 89.74,
+      "eval_exact_match": 0.5083160083160083,
+      "eval_loss": 0.9255384206771851,
+      "eval_runtime": 135.5229,
+      "eval_samples_per_second": 7.098,
+      "step": 21000
+    },
+    {
+      "epoch": 89.76,
+      "learning_rate": 8.20563518188275e-05,
+      "loss": 0.0,
+      "step": 21004
+    },
+    {
+      "epoch": 89.78,
+      "learning_rate": 8.204740702946702e-05,
+      "loss": 0.0001,
+      "step": 21008
+    },
+    {
+      "epoch": 89.79,
+      "learning_rate": 8.203846049899891e-05,
+      "loss": 0.0005,
+      "step": 21012
+    },
+    {
+      "epoch": 89.81,
+      "learning_rate": 8.202951222790916e-05,
+      "loss": 0.0001,
+      "step": 21016
+    },
+    {
+      "epoch": 89.83,
+      "learning_rate": 8.2020562216684e-05,
+      "loss": 0.0,
+      "step": 21020
+    },
+    {
+      "epoch": 89.85,
+      "learning_rate": 8.201161046580963e-05,
+      "loss": 0.0,
+      "step": 21024
+    },
+    {
+      "epoch": 89.86,
+      "learning_rate": 8.200265697577241e-05,
+      "loss": 0.0004,
+      "step": 21028
+    },
+    {
+      "epoch": 89.88,
+      "learning_rate": 8.199370174705876e-05,
+      "loss": 0.001,
+      "step": 21032
+    },
+    {
+      "epoch": 89.9,
+      "learning_rate": 8.198474478015521e-05,
+      "loss": 0.0001,
+      "step": 21036
+    },
+    {
+      "epoch": 89.91,
+      "learning_rate": 8.197578607554842e-05,
+      "loss": 0.0002,
+      "step": 21040
+    },
+    {
+      "epoch": 89.93,
+      "learning_rate": 8.196682563372505e-05,
+      "loss": 0.0002,
+      "step": 21044
+    },
+    {
+      "epoch": 89.95,
+      "learning_rate": 8.195786345517196e-05,
+      "loss": 0.0011,
+      "step": 21048
+    },
+    {
+      "epoch": 89.97,
+      "learning_rate": 8.194889954037603e-05,
+      "loss": 0.0001,
+      "step": 21052
+    },
+    {
+      "epoch": 89.98,
+      "learning_rate": 8.193993388982428e-05,
+      "loss": 0.0001,
+      "step": 21056
+    },
+    {
+      "epoch": 90.0,
+      "learning_rate": 8.19309665040038e-05,
+      "loss": 0.0,
+      "step": 21060
+    },
+    {
+      "epoch": 90.02,
+      "learning_rate": 8.192199738340177e-05,
+      "loss": 0.0001,
+      "step": 21064
+    },
+    {
+      "epoch": 90.03,
+      "learning_rate": 8.19130265285055e-05,
+      "loss": 0.0001,
+      "step": 21068
+    },
+    {
+      "epoch": 90.05,
+      "learning_rate": 8.190405393980234e-05,
+      "loss": 0.0001,
+      "step": 21072
+    },
+    {
+      "epoch": 90.07,
+      "learning_rate": 8.189507961777976e-05,
+      "loss": 0.0011,
+      "step": 21076
+    },
+    {
+      "epoch": 90.09,
+      "learning_rate": 8.188610356292536e-05,
+      "loss": 0.0,
+      "step": 21080
+    },
+    {
+      "epoch": 90.1,
+      "learning_rate": 8.187712577572679e-05,
+      "loss": 0.0001,
+      "step": 21084
+    },
+    {
+      "epoch": 90.12,
+      "learning_rate": 8.186814625667178e-05,
+      "loss": 0.0001,
+      "step": 21088
+    },
+    {
+      "epoch": 90.14,
+      "learning_rate": 8.185916500624821e-05,
+      "loss": 0.0001,
+      "step": 21092
+    },
+    {
+      "epoch": 90.15,
+      "learning_rate": 8.185018202494401e-05,
+      "loss": 0.0005,
+      "step": 21096
+    },
+    {
+      "epoch": 90.17,
+      "learning_rate": 8.184119731324722e-05,
+      "loss": 0.0,
+      "step": 21100
+    },
+    {
+      "epoch": 90.19,
+      "learning_rate": 8.183221087164596e-05,
+      "loss": 0.0001,
+      "step": 21104
+    },
+    {
+      "epoch": 90.21,
+      "learning_rate": 8.182322270062848e-05,
+      "loss": 0.0003,
+      "step": 21108
+    },
+    {
+      "epoch": 90.22,
+      "learning_rate": 8.181423280068308e-05,
+      "loss": 0.0,
+      "step": 21112
+    },
+    {
+      "epoch": 90.24,
+      "learning_rate": 8.180524117229818e-05,
+      "loss": 0.0,
+      "step": 21116
+    },
+    {
+      "epoch": 90.26,
+      "learning_rate": 8.179624781596226e-05,
+      "loss": 0.0001,
+      "step": 21120
+    },
+    {
+      "epoch": 90.27,
+      "learning_rate": 8.1787252732164e-05,
+      "loss": 0.0002,
+      "step": 21124
+    },
+    {
+      "epoch": 90.29,
+      "learning_rate": 8.1778255921392e-05,
+      "loss": 0.0001,
+      "step": 21128
+    },
+    {
+      "epoch": 90.31,
+      "learning_rate": 8.17692573841351e-05,
+      "loss": 0.0,
+      "step": 21132
+    },
+    {
+      "epoch": 90.32,
+      "learning_rate": 8.176025712088218e-05,
+      "loss": 0.0001,
+      "step": 21136
+    },
+    {
+      "epoch": 90.34,
+      "learning_rate": 8.175125513212221e-05,
+      "loss": 0.0001,
+      "step": 21140
+    },
+    {
+      "epoch": 90.36,
+      "learning_rate": 8.174225141834426e-05,
+      "loss": 0.0002,
+      "step": 21144
+    },
+    {
+      "epoch": 90.38,
+      "learning_rate": 8.17332459800375e-05,
+      "loss": 0.0001,
+      "step": 21148
+    },
+    {
+      "epoch": 90.39,
+      "learning_rate": 8.172423881769117e-05,
+      "loss": 0.0006,
+      "step": 21152
+    },
+    {
+      "epoch": 90.41,
+      "learning_rate": 8.171522993179463e-05,
+      "loss": 0.0001,
+      "step": 21156
+    },
+    {
+      "epoch": 90.43,
+      "learning_rate": 8.170621932283735e-05,
+      "loss": 0.0003,
+      "step": 21160
+    },
+    {
+      "epoch": 90.44,
+      "learning_rate": 8.169720699130883e-05,
+      "loss": 0.0,
+      "step": 21164
+    },
+    {
+      "epoch": 90.46,
+      "learning_rate": 8.168819293769874e-05,
+      "loss": 0.0001,
+      "step": 21168
+    },
+    {
+      "epoch": 90.48,
+      "learning_rate": 8.167917716249677e-05,
+      "loss": 0.0001,
+      "step": 21172
+    },
+    {
+      "epoch": 90.5,
+      "learning_rate": 8.167015966619274e-05,
+      "loss": 0.0,
+      "step": 21176
+    },
+    {
+      "epoch": 90.51,
+      "learning_rate": 8.16611404492766e-05,
+      "loss": 0.0,
+      "step": 21180
+    },
+    {
+      "epoch": 90.53,
+      "learning_rate": 8.165211951223832e-05,
+      "loss": 0.0001,
+      "step": 21184
+    },
+    {
+      "epoch": 90.55,
+      "learning_rate": 8.164309685556802e-05,
+      "loss": 0.0,
+      "step": 21188
+    },
+    {
+      "epoch": 90.56,
+      "learning_rate": 8.163407247975588e-05,
+      "loss": 0.0001,
+      "step": 21192
+    },
+    {
+      "epoch": 90.58,
+      "learning_rate": 8.162504638529218e-05,
+      "loss": 0.0002,
+      "step": 21196
+    },
+    {
+      "epoch": 90.6,
+      "learning_rate": 8.161601857266732e-05,
+      "loss": 0.0,
+      "step": 21200
+    },
+    {
+      "epoch": 90.62,
+      "learning_rate": 8.160698904237176e-05,
+      "loss": 0.0003,
+      "step": 21204
+    },
+    {
+      "epoch": 90.63,
+      "learning_rate": 8.15979577948961e-05,
+      "loss": 0.0001,
+      "step": 21208
+    },
+    {
+      "epoch": 90.65,
+      "learning_rate": 8.158892483073094e-05,
+      "loss": 0.0001,
+      "step": 21212
+    },
+    {
+      "epoch": 90.67,
+      "learning_rate": 8.157989015036707e-05,
+      "loss": 0.0002,
+      "step": 21216
+    },
+    {
+      "epoch": 90.68,
+      "learning_rate": 8.157085375429533e-05,
+      "loss": 0.0,
+      "step": 21220
+    },
+    {
+      "epoch": 90.7,
+      "learning_rate": 8.156181564300667e-05,
+      "loss": 0.0002,
+      "step": 21224
+    },
+    {
+      "epoch": 90.72,
+      "learning_rate": 8.155277581699212e-05,
+      "loss": 0.0001,
+      "step": 21228
+    },
+    {
+      "epoch": 90.74,
+      "learning_rate": 8.154373427674278e-05,
+      "loss": 0.0,
+      "step": 21232
+    },
+    {
+      "epoch": 90.75,
+      "learning_rate": 8.153469102274988e-05,
+      "loss": 0.0002,
+      "step": 21236
+    },
+    {
+      "epoch": 90.77,
+      "learning_rate": 8.152564605550476e-05,
+      "loss": 0.0,
+      "step": 21240
+    },
+    {
+      "epoch": 90.79,
+      "learning_rate": 8.15165993754988e-05,
+      "loss": 0.0003,
+      "step": 21244
+    },
+    {
+      "epoch": 90.8,
+      "learning_rate": 8.150755098322351e-05,
+      "loss": 0.0002,
+      "step": 21248
+    },
+    {
+      "epoch": 90.82,
+      "learning_rate": 8.149850087917045e-05,
+      "loss": 0.0002,
+      "step": 21252
+    },
+    {
+      "epoch": 90.84,
+      "learning_rate": 8.148944906383136e-05,
+      "loss": 0.0002,
+      "step": 21256
+    },
+    {
+      "epoch": 90.85,
+      "learning_rate": 8.148039553769796e-05,
+      "loss": 0.0001,
+      "step": 21260
+    },
+    {
+      "epoch": 90.87,
+      "learning_rate": 8.147134030126217e-05,
+      "loss": 0.0001,
+      "step": 21264
+    },
+    {
+      "epoch": 90.89,
+      "learning_rate": 8.146228335501591e-05,
+      "loss": 0.0001,
+      "step": 21268
+    },
+    {
+      "epoch": 90.91,
+      "learning_rate": 8.145322469945126e-05,
+      "loss": 0.0001,
+      "step": 21272
+    },
+    {
+      "epoch": 90.92,
+      "learning_rate": 8.144416433506036e-05,
+      "loss": 0.0002,
+      "step": 21276
+    },
+    {
+      "epoch": 90.94,
+      "learning_rate": 8.143510226233546e-05,
+      "loss": 0.0001,
+      "step": 21280
+    },
+    {
+      "epoch": 90.96,
+      "learning_rate": 8.14260384817689e-05,
+      "loss": 0.0002,
+      "step": 21284
+    },
+    {
+      "epoch": 90.97,
+      "learning_rate": 8.141697299385308e-05,
+      "loss": 0.0001,
+      "step": 21288
+    },
+    {
+      "epoch": 90.99,
+      "learning_rate": 8.140790579908056e-05,
+      "loss": 0.0,
+      "step": 21292
+    },
+    {
+      "epoch": 91.01,
+      "learning_rate": 8.13988368979439e-05,
+      "loss": 0.0001,
+      "step": 21296
+    },
+    {
+      "epoch": 91.03,
+      "learning_rate": 8.138976629093586e-05,
+      "loss": 0.0,
+      "step": 21300
+    },
+    {
+      "epoch": 91.04,
+      "learning_rate": 8.13806939785492e-05,
+      "loss": 0.0021,
+      "step": 21304
+    },
+    {
+      "epoch": 91.06,
+      "learning_rate": 8.137161996127682e-05,
+      "loss": 0.0011,
+      "step": 21308
+    },
+    {
+      "epoch": 91.08,
+      "learning_rate": 8.136254423961172e-05,
+      "loss": 0.0,
+      "step": 21312
+    },
+    {
+      "epoch": 91.09,
+      "learning_rate": 8.135346681404696e-05,
+      "loss": 0.0,
+      "step": 21316
+    },
+    {
+      "epoch": 91.11,
+      "learning_rate": 8.134438768507572e-05,
+      "loss": 0.0,
+      "step": 21320
+    },
+    {
+      "epoch": 91.13,
+      "learning_rate": 8.133530685319125e-05,
+      "loss": 0.0003,
+      "step": 21324
+    },
+    {
+      "epoch": 91.15,
+      "learning_rate": 8.13262243188869e-05,
+      "loss": 0.0001,
+      "step": 21328
+    },
+    {
+      "epoch": 91.16,
+      "learning_rate": 8.131714008265614e-05,
+      "loss": 0.0001,
+      "step": 21332
+    },
+    {
+      "epoch": 91.18,
+      "learning_rate": 8.130805414499247e-05,
+      "loss": 0.0,
+      "step": 21336
+    },
+    {
+      "epoch": 91.2,
+      "learning_rate": 8.129896650638956e-05,
+      "loss": 0.0001,
+      "step": 21340
+    },
+    {
+      "epoch": 91.21,
+      "learning_rate": 8.128987716734111e-05,
+      "loss": 0.0,
+      "step": 21344
+    },
+    {
+      "epoch": 91.23,
+      "learning_rate": 8.128078612834095e-05,
+      "loss": 0.0001,
+      "step": 21348
+    },
+    {
+      "epoch": 91.25,
+      "learning_rate": 8.127169338988299e-05,
+      "loss": 0.0,
+      "step": 21352
+    },
+    {
+      "epoch": 91.26,
+      "learning_rate": 8.12625989524612e-05,
+      "loss": 0.0,
+      "step": 21356
+    },
+    {
+      "epoch": 91.28,
+      "learning_rate": 8.125350281656969e-05,
+      "loss": 0.0001,
+      "step": 21360
+    },
+    {
+      "epoch": 91.3,
+      "learning_rate": 8.124440498270267e-05,
+      "loss": 0.0,
+      "step": 21364
+    },
+    {
+      "epoch": 91.32,
+      "learning_rate": 8.123530545135439e-05,
+      "loss": 0.0001,
+      "step": 21368
+    },
+    {
+      "epoch": 91.33,
+      "learning_rate": 8.122620422301923e-05,
+      "loss": 0.0003,
+      "step": 21372
+    },
+    {
+      "epoch": 91.35,
+      "learning_rate": 8.121710129819163e-05,
+      "loss": 0.0001,
+      "step": 21376
+    },
+    {
+      "epoch": 91.37,
+      "learning_rate": 8.120799667736618e-05,
+      "loss": 0.0006,
+      "step": 21380
+    },
+    {
+      "epoch": 91.38,
+      "learning_rate": 8.11988903610375e-05,
+      "loss": 0.0001,
+      "step": 21384
+    },
+    {
+      "epoch": 91.4,
+      "learning_rate": 8.118978234970035e-05,
+      "loss": 0.0,
+      "step": 21388
+    },
+    {
+      "epoch": 91.42,
+      "learning_rate": 8.118067264384955e-05,
+      "loss": 0.0003,
+      "step": 21392
+    },
+    {
+      "epoch": 91.44,
+      "learning_rate": 8.117156124397999e-05,
+      "loss": 0.0001,
+      "step": 21396
+    },
+    {
+      "epoch": 91.45,
+      "learning_rate": 8.116244815058673e-05,
+      "loss": 0.0,
+      "step": 21400
+    },
+    {
+      "epoch": 91.47,
+      "learning_rate": 8.115333336416486e-05,
+      "loss": 0.0001,
+      "step": 21404
+    },
+    {
+      "epoch": 91.49,
+      "learning_rate": 8.114421688520957e-05,
+      "loss": 0.0001,
+      "step": 21408
+    },
+    {
+      "epoch": 91.5,
+      "learning_rate": 8.113509871421618e-05,
+      "loss": 0.0001,
+      "step": 21412
+    },
+    {
+      "epoch": 91.52,
+      "learning_rate": 8.112597885168003e-05,
+      "loss": 0.0,
+      "step": 21416
+    },
+    {
+      "epoch": 91.54,
+      "learning_rate": 8.111685729809662e-05,
+      "loss": 0.0002,
+      "step": 21420
+    },
+    {
+      "epoch": 91.56,
+      "learning_rate": 8.110773405396153e-05,
+      "loss": 0.0001,
+      "step": 21424
+    },
+    {
+      "epoch": 91.57,
+      "learning_rate": 8.109860911977039e-05,
+      "loss": 0.0002,
+      "step": 21428
+    },
+    {
+      "epoch": 91.59,
+      "learning_rate": 8.108948249601894e-05,
+      "loss": 0.0002,
+      "step": 21432
+    },
+    {
+      "epoch": 91.61,
+      "learning_rate": 8.108035418320306e-05,
+      "loss": 0.0001,
+      "step": 21436
+    },
+    {
+      "epoch": 91.62,
+      "learning_rate": 8.107122418181865e-05,
+      "loss": 0.0028,
+      "step": 21440
+    },
+    {
+      "epoch": 91.64,
+      "learning_rate": 8.106209249236177e-05,
+      "loss": 0.0,
+      "step": 21444
+    },
+    {
+      "epoch": 91.66,
+      "learning_rate": 8.105295911532848e-05,
+      "loss": 0.0004,
+      "step": 21448
+    },
+    {
+      "epoch": 91.68,
+      "learning_rate": 8.104382405121506e-05,
+      "loss": 0.0002,
+      "step": 21452
+    },
+    {
+      "epoch": 91.69,
+      "learning_rate": 8.103468730051774e-05,
+      "loss": 0.0,
+      "step": 21456
+    },
+    {
+      "epoch": 91.71,
+      "learning_rate": 8.102554886373296e-05,
+      "loss": 0.0,
+      "step": 21460
+    },
+    {
+      "epoch": 91.73,
+      "learning_rate": 8.10164087413572e-05,
+      "loss": 0.0,
+      "step": 21464
+    },
+    {
+      "epoch": 91.74,
+      "learning_rate": 8.100726693388703e-05,
+      "loss": 0.0001,
+      "step": 21468
+    },
+    {
+      "epoch": 91.76,
+      "learning_rate": 8.099812344181909e-05,
+      "loss": 0.0001,
+      "step": 21472
+    },
+    {
+      "epoch": 91.78,
+      "learning_rate": 8.098897826565016e-05,
+      "loss": 0.0001,
+      "step": 21476
+    },
+    {
+      "epoch": 91.79,
+      "learning_rate": 8.09798314058771e-05,
+      "loss": 0.0001,
+      "step": 21480
+    },
+    {
+      "epoch": 91.81,
+      "learning_rate": 8.097068286299683e-05,
+      "loss": 0.0021,
+      "step": 21484
+    },
+    {
+      "epoch": 91.83,
+      "learning_rate": 8.096153263750639e-05,
+      "loss": 0.0003,
+      "step": 21488
+    },
+    {
+      "epoch": 91.85,
+      "learning_rate": 8.095238072990289e-05,
+      "loss": 0.0001,
+      "step": 21492
+    },
+    {
+      "epoch": 91.86,
+      "learning_rate": 8.094322714068358e-05,
+      "loss": 0.0,
+      "step": 21496
+    },
+    {
+      "epoch": 91.88,
+      "learning_rate": 8.093407187034576e-05,
+      "loss": 0.0018,
+      "step": 21500
+    },
+    {
+      "epoch": 91.9,
+      "learning_rate": 8.092491491938678e-05,
+      "loss": 0.0007,
+      "step": 21504
+    },
+    {
+      "epoch": 91.91,
+      "learning_rate": 8.091575628830419e-05,
+      "loss": 0.0001,
+      "step": 21508
+    },
+    {
+      "epoch": 91.93,
+      "learning_rate": 8.090659597759554e-05,
+      "loss": 0.0001,
+      "step": 21512
+    },
+    {
+      "epoch": 91.95,
+      "learning_rate": 8.089743398775849e-05,
+      "loss": 0.0,
+      "step": 21516
+    },
+    {
+      "epoch": 91.97,
+      "learning_rate": 8.088827031929081e-05,
+      "loss": 0.0,
+      "step": 21520
+    },
+    {
+      "epoch": 91.98,
+      "learning_rate": 8.08791049726904e-05,
+      "loss": 0.0001,
+      "step": 21524
+    },
+    {
+      "epoch": 92.0,
+      "learning_rate": 8.086993794845514e-05,
+      "loss": 0.0001,
+      "step": 21528
+    },
+    {
+      "epoch": 92.02,
+      "learning_rate": 8.08607692470831e-05,
+      "loss": 0.0001,
+      "step": 21532
+    },
+    {
+      "epoch": 92.03,
+      "learning_rate": 8.085159886907239e-05,
+      "loss": 0.0,
+      "step": 21536
+    },
+    {
+      "epoch": 92.05,
+      "learning_rate": 8.084242681492125e-05,
+      "loss": 0.0001,
+      "step": 21540
+    },
+    {
+      "epoch": 92.07,
+      "learning_rate": 8.083325308512799e-05,
+      "loss": 0.0,
+      "step": 21544
+    },
+    {
+      "epoch": 92.09,
+      "learning_rate": 8.082407768019099e-05,
+      "loss": 0.0,
+      "step": 21548
+    },
+    {
+      "epoch": 92.1,
+      "learning_rate": 8.081490060060875e-05,
+      "loss": 0.0005,
+      "step": 21552
+    },
+    {
+      "epoch": 92.12,
+      "learning_rate": 8.080572184687987e-05,
+      "loss": 0.0,
+      "step": 21556
+    },
+    {
+      "epoch": 92.14,
+      "learning_rate": 8.0796541419503e-05,
+      "loss": 0.0001,
+      "step": 21560
+    },
+    {
+      "epoch": 92.15,
+      "learning_rate": 8.078735931897691e-05,
+      "loss": 0.0004,
+      "step": 21564
+    },
+    {
+      "epoch": 92.17,
+      "learning_rate": 8.077817554580045e-05,
+      "loss": 0.0003,
+      "step": 21568
+    },
+    {
+      "epoch": 92.19,
+      "learning_rate": 8.076899010047259e-05,
+      "loss": 0.0,
+      "step": 21572
+    },
+    {
+      "epoch": 92.21,
+      "learning_rate": 8.075980298349235e-05,
+      "loss": 0.0002,
+      "step": 21576
+    },
+    {
+      "epoch": 92.22,
+      "learning_rate": 8.075061419535885e-05,
+      "loss": 0.0001,
+      "step": 21580
+    },
+    {
+      "epoch": 92.24,
+      "learning_rate": 8.074142373657135e-05,
+      "loss": 0.0,
+      "step": 21584
+    },
+    {
+      "epoch": 92.26,
+      "learning_rate": 8.07322316076291e-05,
+      "loss": 0.0002,
+      "step": 21588
+    },
+    {
+      "epoch": 92.27,
+      "learning_rate": 8.072303780903153e-05,
+      "loss": 0.0004,
+      "step": 21592
+    },
+    {
+      "epoch": 92.29,
+      "learning_rate": 8.071384234127813e-05,
+      "loss": 0.0,
+      "step": 21596
+    },
+    {
+      "epoch": 92.31,
+      "learning_rate": 8.070464520486849e-05,
+      "loss": 0.0018,
+      "step": 21600
+    },
+    {
+      "epoch": 92.32,
+      "learning_rate": 8.069544640030227e-05,
+      "loss": 0.0,
+      "step": 21604
+    },
+    {
+      "epoch": 92.34,
+      "learning_rate": 8.068624592807924e-05,
+      "loss": 0.0001,
+      "step": 21608
+    },
+    {
+      "epoch": 92.36,
+      "learning_rate": 8.067704378869927e-05,
+      "loss": 0.0,
+      "step": 21612
+    },
+    {
+      "epoch": 92.38,
+      "learning_rate": 8.066783998266225e-05,
+      "loss": 0.0,
+      "step": 21616
+    },
+    {
+      "epoch": 92.39,
+      "learning_rate": 8.065863451046828e-05,
+      "loss": 0.0003,
+      "step": 21620
+    },
+    {
+      "epoch": 92.41,
+      "learning_rate": 8.064942737261745e-05,
+      "loss": 0.0004,
+      "step": 21624
+    },
+    {
+      "epoch": 92.43,
+      "learning_rate": 8.064021856960998e-05,
+      "loss": 0.0001,
+      "step": 21628
+    },
+    {
+      "epoch": 92.44,
+      "learning_rate": 8.063100810194616e-05,
+      "loss": 0.0,
+      "step": 21632
+    },
+    {
+      "epoch": 92.46,
+      "learning_rate": 8.062179597012641e-05,
+      "loss": 0.0001,
+      "step": 21636
+    },
+    {
+      "epoch": 92.48,
+      "learning_rate": 8.061258217465121e-05,
+      "loss": 0.0003,
+      "step": 21640
+    },
+    {
+      "epoch": 92.5,
+      "learning_rate": 8.060336671602116e-05,
+      "loss": 0.0001,
+      "step": 21644
+    },
+    {
+      "epoch": 92.51,
+      "learning_rate": 8.059414959473689e-05,
+      "loss": 0.0001,
+      "step": 21648
+    },
+    {
+      "epoch": 92.53,
+      "learning_rate": 8.058493081129917e-05,
+      "loss": 0.0012,
+      "step": 21652
+    },
+    {
+      "epoch": 92.55,
+      "learning_rate": 8.057571036620886e-05,
+      "loss": 0.0,
+      "step": 21656
+    },
+    {
+      "epoch": 92.56,
+      "learning_rate": 8.056648825996688e-05,
+      "loss": 0.0001,
+      "step": 21660
+    },
+    {
+      "epoch": 92.58,
+      "learning_rate": 8.05572644930743e-05,
+      "loss": 0.0,
+      "step": 21664
+    },
+    {
+      "epoch": 92.6,
+      "learning_rate": 8.054803906603219e-05,
+      "loss": 0.0,
+      "step": 21668
+    },
+    {
+      "epoch": 92.62,
+      "learning_rate": 8.053881197934178e-05,
+      "loss": 0.0,
+      "step": 21672
+    },
+    {
+      "epoch": 92.63,
+      "learning_rate": 8.052958323350437e-05,
+      "loss": 0.0,
+      "step": 21676
+    },
+    {
+      "epoch": 92.65,
+      "learning_rate": 8.052035282902135e-05,
+      "loss": 0.0001,
+      "step": 21680
+    },
+    {
+      "epoch": 92.67,
+      "learning_rate": 8.05111207663942e-05,
+      "loss": 0.0,
+      "step": 21684
+    },
+    {
+      "epoch": 92.68,
+      "learning_rate": 8.050188704612448e-05,
+      "loss": 0.0002,
+      "step": 21688
+    },
+    {
+      "epoch": 92.7,
+      "learning_rate": 8.049265166871387e-05,
+      "loss": 0.0,
+      "step": 21692
+    },
+    {
+      "epoch": 92.72,
+      "learning_rate": 8.04834146346641e-05,
+      "loss": 0.0001,
+      "step": 21696
+    },
+    {
+      "epoch": 92.74,
+      "learning_rate": 8.047417594447702e-05,
+      "loss": 0.0001,
+      "step": 21700
+    },
+    {
+      "epoch": 92.75,
+      "learning_rate": 8.046493559865456e-05,
+      "loss": 0.0,
+      "step": 21704
+    },
+    {
+      "epoch": 92.77,
+      "learning_rate": 8.045569359769874e-05,
+      "loss": 0.0001,
+      "step": 21708
+    },
+    {
+      "epoch": 92.79,
+      "learning_rate": 8.044644994211165e-05,
+      "loss": 0.0006,
+      "step": 21712
+    },
+    {
+      "epoch": 92.8,
+      "learning_rate": 8.043720463239553e-05,
+      "loss": 0.0001,
+      "step": 21716
+    },
+    {
+      "epoch": 92.82,
+      "learning_rate": 8.042795766905264e-05,
+      "loss": 0.0004,
+      "step": 21720
+    },
+    {
+      "epoch": 92.84,
+      "learning_rate": 8.041870905258538e-05,
+      "loss": 0.0001,
+      "step": 21724
+    },
+    {
+      "epoch": 92.85,
+      "learning_rate": 8.040945878349617e-05,
+      "loss": 0.0001,
+      "step": 21728
+    },
+    {
+      "epoch": 92.87,
+      "learning_rate": 8.040020686228764e-05,
+      "loss": 0.0001,
+      "step": 21732
+    },
+    {
+      "epoch": 92.89,
+      "learning_rate": 8.039095328946239e-05,
+      "loss": 0.0018,
+      "step": 21736
+    },
+    {
+      "epoch": 92.91,
+      "learning_rate": 8.038169806552318e-05,
+      "loss": 0.0,
+      "step": 21740
+    },
+    {
+      "epoch": 92.92,
+      "learning_rate": 8.037244119097283e-05,
+      "loss": 0.0,
+      "step": 21744
+    },
+    {
+      "epoch": 92.94,
+      "learning_rate": 8.036318266631425e-05,
+      "loss": 0.0,
+      "step": 21748
+    },
+    {
+      "epoch": 92.96,
+      "learning_rate": 8.035392249205046e-05,
+      "loss": 0.0004,
+      "step": 21752
+    },
+    {
+      "epoch": 92.97,
+      "learning_rate": 8.034466066868454e-05,
+      "loss": 0.0026,
+      "step": 21756
+    },
+    {
+      "epoch": 92.99,
+      "learning_rate": 8.033539719671972e-05,
+      "loss": 0.0001,
+      "step": 21760
+    },
+    {
+      "epoch": 93.01,
+      "learning_rate": 8.032613207665922e-05,
+      "loss": 0.0001,
+      "step": 21764
+    },
+    {
+      "epoch": 93.03,
+      "learning_rate": 8.031686530900645e-05,
+      "loss": 0.0001,
+      "step": 21768
+    },
+    {
+      "epoch": 93.04,
+      "learning_rate": 8.030759689426484e-05,
+      "loss": 0.0,
+      "step": 21772
+    },
+    {
+      "epoch": 93.06,
+      "learning_rate": 8.029832683293795e-05,
+      "loss": 0.0001,
+      "step": 21776
+    },
+    {
+      "epoch": 93.08,
+      "learning_rate": 8.028905512552939e-05,
+      "loss": 0.0001,
+      "step": 21780
+    },
+    {
+      "epoch": 93.09,
+      "learning_rate": 8.027978177254292e-05,
+      "loss": 0.0001,
+      "step": 21784
+    },
+    {
+      "epoch": 93.11,
+      "learning_rate": 8.027050677448231e-05,
+      "loss": 0.0,
+      "step": 21788
+    },
+    {
+      "epoch": 93.13,
+      "learning_rate": 8.026123013185152e-05,
+      "loss": 0.0009,
+      "step": 21792
+    },
+    {
+      "epoch": 93.15,
+      "learning_rate": 8.025195184515448e-05,
+      "loss": 0.0008,
+      "step": 21796
+    },
+    {
+      "epoch": 93.16,
+      "learning_rate": 8.02426719148953e-05,
+      "loss": 0.0003,
+      "step": 21800
+    },
+    {
+      "epoch": 93.18,
+      "learning_rate": 8.023339034157817e-05,
+      "loss": 0.0,
+      "step": 21804
+    },
+    {
+      "epoch": 93.2,
+      "learning_rate": 8.022410712570732e-05,
+      "loss": 0.0016,
+      "step": 21808
+    },
+    {
+      "epoch": 93.21,
+      "learning_rate": 8.021482226778712e-05,
+      "loss": 0.0001,
+      "step": 21812
+    },
+    {
+      "epoch": 93.23,
+      "learning_rate": 8.020553576832197e-05,
+      "loss": 0.0001,
+      "step": 21816
+    },
+    {
+      "epoch": 93.25,
+      "learning_rate": 8.019624762781643e-05,
+      "loss": 0.0,
+      "step": 21820
+    },
+    {
+      "epoch": 93.26,
+      "learning_rate": 8.018695784677513e-05,
+      "loss": 0.002,
+      "step": 21824
+    },
+    {
+      "epoch": 93.28,
+      "learning_rate": 8.017766642570276e-05,
+      "loss": 0.0,
+      "step": 21828
+    },
+    {
+      "epoch": 93.3,
+      "learning_rate": 8.01683733651041e-05,
+      "loss": 0.0002,
+      "step": 21832
+    },
+    {
+      "epoch": 93.32,
+      "learning_rate": 8.015907866548404e-05,
+      "loss": 0.0001,
+      "step": 21836
+    },
+    {
+      "epoch": 93.33,
+      "learning_rate": 8.014978232734757e-05,
+      "loss": 0.0,
+      "step": 21840
+    },
+    {
+      "epoch": 93.35,
+      "learning_rate": 8.014048435119973e-05,
+      "loss": 0.0002,
+      "step": 21844
+    },
+    {
+      "epoch": 93.37,
+      "learning_rate": 8.013118473754568e-05,
+      "loss": 0.0,
+      "step": 21848
+    },
+    {
+      "epoch": 93.38,
+      "learning_rate": 8.012188348689068e-05,
+      "loss": 0.0,
+      "step": 21852
+    },
+    {
+      "epoch": 93.4,
+      "learning_rate": 8.011258059974002e-05,
+      "loss": 0.0003,
+      "step": 21856
+    },
+    {
+      "epoch": 93.42,
+      "learning_rate": 8.010327607659914e-05,
+      "loss": 0.0,
+      "step": 21860
+    },
+    {
+      "epoch": 93.44,
+      "learning_rate": 8.009396991797355e-05,
+      "loss": 0.0,
+      "step": 21864
+    },
+    {
+      "epoch": 93.45,
+      "learning_rate": 8.008466212436884e-05,
+      "loss": 0.0001,
+      "step": 21868
+    },
+    {
+      "epoch": 93.47,
+      "learning_rate": 8.007535269629069e-05,
+      "loss": 0.0002,
+      "step": 21872
+    },
+    {
+      "epoch": 93.49,
+      "learning_rate": 8.00660416342449e-05,
+      "loss": 0.0001,
+      "step": 21876
+    },
+    {
+      "epoch": 93.5,
+      "learning_rate": 8.005672893873729e-05,
+      "loss": 0.0,
+      "step": 21880
+    },
+    {
+      "epoch": 93.52,
+      "learning_rate": 8.004741461027381e-05,
+      "loss": 0.0001,
+      "step": 21884
+    },
+    {
+      "epoch": 93.54,
+      "learning_rate": 8.003809864936055e-05,
+      "loss": 0.0001,
+      "step": 21888
+    },
+    {
+      "epoch": 93.56,
+      "learning_rate": 8.00287810565036e-05,
+      "loss": 0.0,
+      "step": 21892
+    },
+    {
+      "epoch": 93.57,
+      "learning_rate": 8.001946183220916e-05,
+      "loss": 0.0,
+      "step": 21896
+    },
+    {
+      "epoch": 93.59,
+      "learning_rate": 8.001014097698359e-05,
+      "loss": 0.0003,
+      "step": 21900
+    },
+    {
+      "epoch": 93.61,
+      "learning_rate": 8.000081849133324e-05,
+      "loss": 0.0001,
+      "step": 21904
+    },
+    {
+      "epoch": 93.62,
+      "learning_rate": 7.999149437576461e-05,
+      "loss": 0.0,
+      "step": 21908
+    },
+    {
+      "epoch": 93.64,
+      "learning_rate": 7.998216863078427e-05,
+      "loss": 0.0,
+      "step": 21912
+    },
+    {
+      "epoch": 93.66,
+      "learning_rate": 7.997284125689886e-05,
+      "loss": 0.0,
+      "step": 21916
+    },
+    {
+      "epoch": 93.68,
+      "learning_rate": 7.996351225461516e-05,
+      "loss": 0.0001,
+      "step": 21920
+    },
+    {
+      "epoch": 93.69,
+      "learning_rate": 7.995418162443999e-05,
+      "loss": 0.0002,
+      "step": 21924
+    },
+    {
+      "epoch": 93.71,
+      "learning_rate": 7.994484936688027e-05,
+      "loss": 0.0004,
+      "step": 21928
+    },
+    {
+      "epoch": 93.73,
+      "learning_rate": 7.993551548244301e-05,
+      "loss": 0.0002,
+      "step": 21932
+    },
+    {
+      "epoch": 93.74,
+      "learning_rate": 7.992617997163533e-05,
+      "loss": 0.0,
+      "step": 21936
+    },
+    {
+      "epoch": 93.76,
+      "learning_rate": 7.991684283496442e-05,
+      "loss": 0.0001,
+      "step": 21940
+    },
+    {
+      "epoch": 93.78,
+      "learning_rate": 7.990750407293754e-05,
+      "loss": 0.0005,
+      "step": 21944
+    },
+    {
+      "epoch": 93.79,
+      "learning_rate": 7.989816368606207e-05,
+      "loss": 0.0001,
+      "step": 21948
+    },
+    {
+      "epoch": 93.81,
+      "learning_rate": 7.988882167484546e-05,
+      "loss": 0.0001,
+      "step": 21952
+    },
+    {
+      "epoch": 93.83,
+      "learning_rate": 7.987947803979525e-05,
+      "loss": 0.0001,
+      "step": 21956
+    },
+    {
+      "epoch": 93.85,
+      "learning_rate": 7.987013278141909e-05,
+      "loss": 0.0006,
+      "step": 21960
+    },
+    {
+      "epoch": 93.86,
+      "learning_rate": 7.986078590022468e-05,
+      "loss": 0.0,
+      "step": 21964
+    },
+    {
+      "epoch": 93.88,
+      "learning_rate": 7.985143739671983e-05,
+      "loss": 0.0,
+      "step": 21968
+    },
+    {
+      "epoch": 93.9,
+      "learning_rate": 7.984208727141245e-05,
+      "loss": 0.0,
+      "step": 21972
+    },
+    {
+      "epoch": 93.91,
+      "learning_rate": 7.983273552481048e-05,
+      "loss": 0.0,
+      "step": 21976
+    },
+    {
+      "epoch": 93.93,
+      "learning_rate": 7.982338215742207e-05,
+      "loss": 0.0001,
+      "step": 21980
+    },
+    {
+      "epoch": 93.95,
+      "learning_rate": 7.981402716975532e-05,
+      "loss": 0.0,
+      "step": 21984
+    },
+    {
+      "epoch": 93.97,
+      "learning_rate": 7.980467056231848e-05,
+      "loss": 0.0002,
+      "step": 21988
+    },
+    {
+      "epoch": 93.98,
+      "learning_rate": 7.979531233561992e-05,
+      "loss": 0.0002,
+      "step": 21992
+    },
+    {
+      "epoch": 94.0,
+      "learning_rate": 7.978595249016803e-05,
+      "loss": 0.0003,
+      "step": 21996
+    },
+    {
+      "epoch": 94.02,
+      "learning_rate": 7.977659102647137e-05,
+      "loss": 0.0001,
+      "step": 22000
+    },
+    {
+      "epoch": 94.02,
+      "eval_exact_match": 0.5135135135135135,
+      "eval_loss": 0.9283977746963501,
+      "eval_runtime": 174.6852,
+      "eval_samples_per_second": 5.507,
+      "step": 22000
+    },
+    {
+      "epoch": 94.03,
+      "learning_rate": 7.976722794503848e-05,
+      "loss": 0.0,
+      "step": 22004
+    },
+    {
+      "epoch": 94.05,
+      "learning_rate": 7.975786324637808e-05,
+      "loss": 0.0,
+      "step": 22008
+    },
+    {
+      "epoch": 94.07,
+      "learning_rate": 7.974849693099895e-05,
+      "loss": 0.0001,
+      "step": 22012
+    },
+    {
+      "epoch": 94.09,
+      "learning_rate": 7.973912899940994e-05,
+      "loss": 0.0008,
+      "step": 22016
+    },
+    {
+      "epoch": 94.1,
+      "learning_rate": 7.972975945212e-05,
+      "loss": 0.0001,
+      "step": 22020
+    },
+    {
+      "epoch": 94.12,
+      "learning_rate": 7.97203882896382e-05,
+      "loss": 0.0018,
+      "step": 22024
+    },
+    {
+      "epoch": 94.14,
+      "learning_rate": 7.971101551247361e-05,
+      "loss": 0.0001,
+      "step": 22028
+    },
+    {
+      "epoch": 94.15,
+      "learning_rate": 7.970164112113551e-05,
+      "loss": 0.0,
+      "step": 22032
+    },
+    {
+      "epoch": 94.17,
+      "learning_rate": 7.969226511613314e-05,
+      "loss": 0.0,
+      "step": 22036
+    },
+    {
+      "epoch": 94.19,
+      "learning_rate": 7.968288749797596e-05,
+      "loss": 0.0001,
+      "step": 22040
+    },
+    {
+      "epoch": 94.21,
+      "learning_rate": 7.967350826717338e-05,
+      "loss": 0.0004,
+      "step": 22044
+    },
+    {
+      "epoch": 94.22,
+      "learning_rate": 7.966412742423501e-05,
+      "loss": 0.0,
+      "step": 22048
+    },
+    {
+      "epoch": 94.24,
+      "learning_rate": 7.965474496967047e-05,
+      "loss": 0.0001,
+      "step": 22052
+    },
+    {
+      "epoch": 94.26,
+      "learning_rate": 7.964536090398953e-05,
+      "loss": 0.0001,
+      "step": 22056
+    },
+    {
+      "epoch": 94.27,
+      "learning_rate": 7.963597522770201e-05,
+      "loss": 0.0001,
+      "step": 22060
+    },
+    {
+      "epoch": 94.29,
+      "learning_rate": 7.962658794131781e-05,
+      "loss": 0.0001,
+      "step": 22064
+    },
+    {
+      "epoch": 94.31,
+      "learning_rate": 7.961719904534694e-05,
+      "loss": 0.0001,
+      "step": 22068
+    },
+    {
+      "epoch": 94.32,
+      "learning_rate": 7.960780854029952e-05,
+      "loss": 0.0001,
+      "step": 22072
+    },
+    {
+      "epoch": 94.34,
+      "learning_rate": 7.959841642668569e-05,
+      "loss": 0.0,
+      "step": 22076
+    },
+    {
+      "epoch": 94.36,
+      "learning_rate": 7.958902270501571e-05,
+      "loss": 0.0002,
+      "step": 22080
+    },
+    {
+      "epoch": 94.38,
+      "learning_rate": 7.957962737579998e-05,
+      "loss": 0.0,
+      "step": 22084
+    },
+    {
+      "epoch": 94.39,
+      "learning_rate": 7.95702304395489e-05,
+      "loss": 0.0001,
+      "step": 22088
+    },
+    {
+      "epoch": 94.41,
+      "learning_rate": 7.956083189677302e-05,
+      "loss": 0.0002,
+      "step": 22092
+    },
+    {
+      "epoch": 94.43,
+      "learning_rate": 7.955143174798292e-05,
+      "loss": 0.0,
+      "step": 22096
+    },
+    {
+      "epoch": 94.44,
+      "learning_rate": 7.954202999368934e-05,
+      "loss": 0.0006,
+      "step": 22100
+    },
+    {
+      "epoch": 94.46,
+      "learning_rate": 7.953262663440306e-05,
+      "loss": 0.0,
+      "step": 22104
+    },
+    {
+      "epoch": 94.48,
+      "learning_rate": 7.952322167063492e-05,
+      "loss": 0.0,
+      "step": 22108
+    },
+    {
+      "epoch": 94.5,
+      "learning_rate": 7.951381510289596e-05,
+      "loss": 0.0,
+      "step": 22112
+    },
+    {
+      "epoch": 94.51,
+      "learning_rate": 7.950440693169714e-05,
+      "loss": 0.0015,
+      "step": 22116
+    },
+    {
+      "epoch": 94.53,
+      "learning_rate": 7.949499715754967e-05,
+      "loss": 0.0001,
+      "step": 22120
+    },
+    {
+      "epoch": 94.55,
+      "learning_rate": 7.948558578096474e-05,
+      "loss": 0.0,
+      "step": 22124
+    },
+    {
+      "epoch": 94.56,
+      "learning_rate": 7.947617280245366e-05,
+      "loss": 0.0008,
+      "step": 22128
+    },
+    {
+      "epoch": 94.58,
+      "learning_rate": 7.946675822252786e-05,
+      "loss": 0.0003,
+      "step": 22132
+    },
+    {
+      "epoch": 94.6,
+      "learning_rate": 7.945734204169879e-05,
+      "loss": 0.0,
+      "step": 22136
+    },
+    {
+      "epoch": 94.62,
+      "learning_rate": 7.944792426047802e-05,
+      "loss": 0.0001,
+      "step": 22140
+    },
+    {
+      "epoch": 94.63,
+      "learning_rate": 7.943850487937725e-05,
+      "loss": 0.0,
+      "step": 22144
+    },
+    {
+      "epoch": 94.65,
+      "learning_rate": 7.94290838989082e-05,
+      "loss": 0.0,
+      "step": 22148
+    },
+    {
+      "epoch": 94.67,
+      "learning_rate": 7.941966131958272e-05,
+      "loss": 0.0012,
+      "step": 22152
+    },
+    {
+      "epoch": 94.68,
+      "learning_rate": 7.941023714191269e-05,
+      "loss": 0.0003,
+      "step": 22156
+    },
+    {
+      "epoch": 94.7,
+      "learning_rate": 7.940081136641015e-05,
+      "loss": 0.0,
+      "step": 22160
+    },
+    {
+      "epoch": 94.72,
+      "learning_rate": 7.93913839935872e-05,
+      "loss": 0.0002,
+      "step": 22164
+    },
+    {
+      "epoch": 94.74,
+      "learning_rate": 7.9381955023956e-05,
+      "loss": 0.0001,
+      "step": 22168
+    },
+    {
+      "epoch": 94.75,
+      "learning_rate": 7.937252445802884e-05,
+      "loss": 0.0007,
+      "step": 22172
+    },
+    {
+      "epoch": 94.77,
+      "learning_rate": 7.936309229631804e-05,
+      "loss": 0.0002,
+      "step": 22176
+    },
+    {
+      "epoch": 94.79,
+      "learning_rate": 7.93536585393361e-05,
+      "loss": 0.0006,
+      "step": 22180
+    },
+    {
+      "epoch": 94.8,
+      "learning_rate": 7.934422318759547e-05,
+      "loss": 0.0,
+      "step": 22184
+    },
+    {
+      "epoch": 94.82,
+      "learning_rate": 7.933478624160884e-05,
+      "loss": 0.0003,
+      "step": 22188
+    },
+    {
+      "epoch": 94.84,
+      "learning_rate": 7.932534770188886e-05,
+      "loss": 0.0001,
+      "step": 22192
+    },
+    {
+      "epoch": 94.85,
+      "learning_rate": 7.931590756894833e-05,
+      "loss": 0.0,
+      "step": 22196
+    },
+    {
+      "epoch": 94.87,
+      "learning_rate": 7.930646584330012e-05,
+      "loss": 0.0003,
+      "step": 22200
+    },
+    {
+      "epoch": 94.89,
+      "learning_rate": 7.929702252545723e-05,
+      "loss": 0.0013,
+      "step": 22204
+    },
+    {
+      "epoch": 94.91,
+      "learning_rate": 7.928757761593265e-05,
+      "loss": 0.0002,
+      "step": 22208
+    },
+    {
+      "epoch": 94.92,
+      "learning_rate": 7.927813111523955e-05,
+      "loss": 0.0008,
+      "step": 22212
+    },
+    {
+      "epoch": 94.94,
+      "learning_rate": 7.926868302389114e-05,
+      "loss": 0.0,
+      "step": 22216
+    },
+    {
+      "epoch": 94.96,
+      "learning_rate": 7.925923334240072e-05,
+      "loss": 0.0001,
+      "step": 22220
+    },
+    {
+      "epoch": 94.97,
+      "learning_rate": 7.92497820712817e-05,
+      "loss": 0.0,
+      "step": 22224
+    },
+    {
+      "epoch": 94.99,
+      "learning_rate": 7.924032921104754e-05,
+      "loss": 0.0002,
+      "step": 22228
+    },
+    {
+      "epoch": 95.01,
+      "learning_rate": 7.923087476221182e-05,
+      "loss": 0.0001,
+      "step": 22232
+    },
+    {
+      "epoch": 95.03,
+      "learning_rate": 7.922141872528817e-05,
+      "loss": 0.0,
+      "step": 22236
+    },
+    {
+      "epoch": 95.04,
+      "learning_rate": 7.921196110079037e-05,
+      "loss": 0.0001,
+      "step": 22240
+    },
+    {
+      "epoch": 95.06,
+      "learning_rate": 7.92025018892322e-05,
+      "loss": 0.0001,
+      "step": 22244
+    },
+    {
+      "epoch": 95.08,
+      "learning_rate": 7.91930410911276e-05,
+      "loss": 0.0,
+      "step": 22248
+    },
+    {
+      "epoch": 95.09,
+      "learning_rate": 7.918357870699056e-05,
+      "loss": 0.0028,
+      "step": 22252
+    },
+    {
+      "epoch": 95.11,
+      "learning_rate": 7.917411473733514e-05,
+      "loss": 0.0,
+      "step": 22256
+    },
+    {
+      "epoch": 95.13,
+      "learning_rate": 7.916464918267554e-05,
+      "loss": 0.0,
+      "step": 22260
+    },
+    {
+      "epoch": 95.15,
+      "learning_rate": 7.915518204352602e-05,
+      "loss": 0.0004,
+      "step": 22264
+    },
+    {
+      "epoch": 95.16,
+      "learning_rate": 7.914571332040088e-05,
+      "loss": 0.0006,
+      "step": 22268
+    },
+    {
+      "epoch": 95.18,
+      "learning_rate": 7.913624301381459e-05,
+      "loss": 0.0,
+      "step": 22272
+    },
+    {
+      "epoch": 95.2,
+      "learning_rate": 7.912677112428164e-05,
+      "loss": 0.0002,
+      "step": 22276
+    },
+    {
+      "epoch": 95.21,
+      "learning_rate": 7.911729765231665e-05,
+      "loss": 0.0,
+      "step": 22280
+    },
+    {
+      "epoch": 95.23,
+      "learning_rate": 7.910782259843427e-05,
+      "loss": 0.0016,
+      "step": 22284
+    },
+    {
+      "epoch": 95.25,
+      "learning_rate": 7.909834596314932e-05,
+      "loss": 0.0005,
+      "step": 22288
+    },
+    {
+      "epoch": 95.26,
+      "learning_rate": 7.908886774697661e-05,
+      "loss": 0.0003,
+      "step": 22292
+    },
+    {
+      "epoch": 95.28,
+      "learning_rate": 7.90793879504311e-05,
+      "loss": 0.0002,
+      "step": 22296
+    },
+    {
+      "epoch": 95.3,
+      "learning_rate": 7.906990657402783e-05,
+      "loss": 0.0,
+      "step": 22300
+    },
+    {
+      "epoch": 95.32,
+      "learning_rate": 7.90604236182819e-05,
+      "loss": 0.0001,
+      "step": 22304
+    },
+    {
+      "epoch": 95.33,
+      "learning_rate": 7.905093908370852e-05,
+      "loss": 0.0012,
+      "step": 22308
+    },
+    {
+      "epoch": 95.35,
+      "learning_rate": 7.904145297082299e-05,
+      "loss": 0.0,
+      "step": 22312
+    },
+    {
+      "epoch": 95.37,
+      "learning_rate": 7.903196528014065e-05,
+      "loss": 0.0,
+      "step": 22316
+    },
+    {
+      "epoch": 95.38,
+      "learning_rate": 7.902247601217699e-05,
+      "loss": 0.0,
+      "step": 22320
+    },
+    {
+      "epoch": 95.4,
+      "learning_rate": 7.901298516744753e-05,
+      "loss": 0.0001,
+      "step": 22324
+    },
+    {
+      "epoch": 95.42,
+      "learning_rate": 7.900349274646791e-05,
+      "loss": 0.0,
+      "step": 22328
+    },
+    {
+      "epoch": 95.44,
+      "learning_rate": 7.899399874975383e-05,
+      "loss": 0.0002,
+      "step": 22332
+    },
+    {
+      "epoch": 95.45,
+      "learning_rate": 7.898450317782112e-05,
+      "loss": 0.0,
+      "step": 22336
+    },
+    {
+      "epoch": 95.47,
+      "learning_rate": 7.897500603118565e-05,
+      "loss": 0.0,
+      "step": 22340
+    },
+    {
+      "epoch": 95.49,
+      "learning_rate": 7.896550731036338e-05,
+      "loss": 0.0,
+      "step": 22344
+    },
+    {
+      "epoch": 95.5,
+      "learning_rate": 7.89560070158704e-05,
+      "loss": 0.0,
+      "step": 22348
+    },
+    {
+      "epoch": 95.52,
+      "learning_rate": 7.894650514822281e-05,
+      "loss": 0.0,
+      "step": 22352
+    },
+    {
+      "epoch": 95.54,
+      "learning_rate": 7.893700170793686e-05,
+      "loss": 0.0,
+      "step": 22356
+    },
+    {
+      "epoch": 95.56,
+      "learning_rate": 7.892749669552888e-05,
+      "loss": 0.0,
+      "step": 22360
+    },
+    {
+      "epoch": 95.57,
+      "learning_rate": 7.891799011151523e-05,
+      "loss": 0.0011,
+      "step": 22364
+    },
+    {
+      "epoch": 95.59,
+      "learning_rate": 7.890848195641243e-05,
+      "loss": 0.0,
+      "step": 22368
+    },
+    {
+      "epoch": 95.61,
+      "learning_rate": 7.889897223073703e-05,
+      "loss": 0.0004,
+      "step": 22372
+    },
+    {
+      "epoch": 95.62,
+      "learning_rate": 7.888946093500568e-05,
+      "loss": 0.0002,
+      "step": 22376
+    },
+    {
+      "epoch": 95.64,
+      "learning_rate": 7.887994806973516e-05,
+      "loss": 0.0,
+      "step": 22380
+    },
+    {
+      "epoch": 95.66,
+      "learning_rate": 7.887043363544225e-05,
+      "loss": 0.0,
+      "step": 22384
+    },
+    {
+      "epoch": 95.68,
+      "learning_rate": 7.886091763264388e-05,
+      "loss": 0.0,
+      "step": 22388
+    },
+    {
+      "epoch": 95.69,
+      "learning_rate": 7.885140006185705e-05,
+      "loss": 0.0,
+      "step": 22392
+    },
+    {
+      "epoch": 95.71,
+      "learning_rate": 7.884188092359881e-05,
+      "loss": 0.0001,
+      "step": 22396
+    },
+    {
+      "epoch": 95.73,
+      "learning_rate": 7.883236021838638e-05,
+      "loss": 0.0,
+      "step": 22400
+    },
+    {
+      "epoch": 95.74,
+      "learning_rate": 7.882283794673698e-05,
+      "loss": 0.0004,
+      "step": 22404
+    },
+    {
+      "epoch": 95.76,
+      "learning_rate": 7.881331410916795e-05,
+      "loss": 0.0,
+      "step": 22408
+    },
+    {
+      "epoch": 95.78,
+      "learning_rate": 7.880378870619672e-05,
+      "loss": 0.0001,
+      "step": 22412
+    },
+    {
+      "epoch": 95.79,
+      "learning_rate": 7.879426173834076e-05,
+      "loss": 0.0006,
+      "step": 22416
+    },
+    {
+      "epoch": 95.81,
+      "learning_rate": 7.878473320611771e-05,
+      "loss": 0.0,
+      "step": 22420
+    },
+    {
+      "epoch": 95.83,
+      "learning_rate": 7.877520311004523e-05,
+      "loss": 0.0001,
+      "step": 22424
+    },
+    {
+      "epoch": 95.85,
+      "learning_rate": 7.876567145064108e-05,
+      "loss": 0.0005,
+      "step": 22428
+    },
+    {
+      "epoch": 95.86,
+      "learning_rate": 7.875613822842311e-05,
+      "loss": 0.0,
+      "step": 22432
+    },
+    {
+      "epoch": 95.88,
+      "learning_rate": 7.874660344390923e-05,
+      "loss": 0.0001,
+      "step": 22436
+    },
+    {
+      "epoch": 95.9,
+      "learning_rate": 7.873706709761748e-05,
+      "loss": 0.0,
+      "step": 22440
+    },
+    {
+      "epoch": 95.91,
+      "learning_rate": 7.872752919006596e-05,
+      "loss": 0.0001,
+      "step": 22444
+    },
+    {
+      "epoch": 95.93,
+      "learning_rate": 7.871798972177287e-05,
+      "loss": 0.0,
+      "step": 22448
+    },
+    {
+      "epoch": 95.95,
+      "learning_rate": 7.870844869325644e-05,
+      "loss": 0.0001,
+      "step": 22452
+    },
+    {
+      "epoch": 95.97,
+      "learning_rate": 7.869890610503504e-05,
+      "loss": 0.0,
+      "step": 22456
+    },
+    {
+      "epoch": 95.98,
+      "learning_rate": 7.868936195762714e-05,
+      "loss": 0.0007,
+      "step": 22460
+    },
+    {
+      "epoch": 96.0,
+      "learning_rate": 7.867981625155124e-05,
+      "loss": 0.0,
+      "step": 22464
+    },
+    {
+      "epoch": 96.02,
+      "learning_rate": 7.867026898732595e-05,
+      "loss": 0.001,
+      "step": 22468
+    },
+    {
+      "epoch": 96.03,
+      "learning_rate": 7.866072016546997e-05,
+      "loss": 0.0001,
+      "step": 22472
+    },
+    {
+      "epoch": 96.05,
+      "learning_rate": 7.865116978650208e-05,
+      "loss": 0.0,
+      "step": 22476
+    },
+    {
+      "epoch": 96.07,
+      "learning_rate": 7.864161785094114e-05,
+      "loss": 0.0001,
+      "step": 22480
+    },
+    {
+      "epoch": 96.09,
+      "learning_rate": 7.86320643593061e-05,
+      "loss": 0.0004,
+      "step": 22484
+    },
+    {
+      "epoch": 96.1,
+      "learning_rate": 7.8622509312116e-05,
+      "loss": 0.0006,
+      "step": 22488
+    },
+    {
+      "epoch": 96.12,
+      "learning_rate": 7.861295270988994e-05,
+      "loss": 0.0,
+      "step": 22492
+    },
+    {
+      "epoch": 96.14,
+      "learning_rate": 7.860339455314713e-05,
+      "loss": 0.0001,
+      "step": 22496
+    },
+    {
+      "epoch": 96.15,
+      "learning_rate": 7.859383484240687e-05,
+      "loss": 0.0007,
+      "step": 22500
+    },
+    {
+      "epoch": 96.17,
+      "learning_rate": 7.858427357818851e-05,
+      "loss": 0.0,
+      "step": 22504
+    },
+    {
+      "epoch": 96.19,
+      "learning_rate": 7.857471076101153e-05,
+      "loss": 0.0,
+      "step": 22508
+    },
+    {
+      "epoch": 96.21,
+      "learning_rate": 7.856514639139546e-05,
+      "loss": 0.0,
+      "step": 22512
+    },
+    {
+      "epoch": 96.22,
+      "learning_rate": 7.855558046985986e-05,
+      "loss": 0.0002,
+      "step": 22516
+    },
+    {
+      "epoch": 96.24,
+      "learning_rate": 7.854601299692454e-05,
+      "loss": 0.0008,
+      "step": 22520
+    },
+    {
+      "epoch": 96.26,
+      "learning_rate": 7.853644397310926e-05,
+      "loss": 0.0,
+      "step": 22524
+    },
+    {
+      "epoch": 96.27,
+      "learning_rate": 7.852687339893386e-05,
+      "loss": 0.0001,
+      "step": 22528
+    },
+    {
+      "epoch": 96.29,
+      "learning_rate": 7.851730127491834e-05,
+      "loss": 0.0,
+      "step": 22532
+    },
+    {
+      "epoch": 96.31,
+      "learning_rate": 7.850772760158273e-05,
+      "loss": 0.0,
+      "step": 22536
+    },
+    {
+      "epoch": 96.32,
+      "learning_rate": 7.849815237944716e-05,
+      "loss": 0.0001,
+      "step": 22540
+    },
+    {
+      "epoch": 96.34,
+      "learning_rate": 7.848857560903183e-05,
+      "loss": 0.0001,
+      "step": 22544
+    },
+    {
+      "epoch": 96.36,
+      "learning_rate": 7.847899729085706e-05,
+      "loss": 0.0,
+      "step": 22548
+    },
+    {
+      "epoch": 96.38,
+      "learning_rate": 7.846941742544323e-05,
+      "loss": 0.0002,
+      "step": 22552
+    },
+    {
+      "epoch": 96.39,
+      "learning_rate": 7.84598360133108e-05,
+      "loss": 0.0006,
+      "step": 22556
+    },
+    {
+      "epoch": 96.41,
+      "learning_rate": 7.84502530549803e-05,
+      "loss": 0.0006,
+      "step": 22560
+    },
+    {
+      "epoch": 96.43,
+      "learning_rate": 7.844066855097241e-05,
+      "loss": 0.0014,
+      "step": 22564
+    },
+    {
+      "epoch": 96.44,
+      "learning_rate": 7.84310825018078e-05,
+      "loss": 0.0001,
+      "step": 22568
+    },
+    {
+      "epoch": 96.46,
+      "learning_rate": 7.84214949080073e-05,
+      "loss": 0.0,
+      "step": 22572
+    },
+    {
+      "epoch": 96.48,
+      "learning_rate": 7.841190577009179e-05,
+      "loss": 0.0001,
+      "step": 22576
+    },
+    {
+      "epoch": 96.5,
+      "learning_rate": 7.840231508858222e-05,
+      "loss": 0.0001,
+      "step": 22580
+    },
+    {
+      "epoch": 96.51,
+      "learning_rate": 7.839272286399967e-05,
+      "loss": 0.0002,
+      "step": 22584
+    },
+    {
+      "epoch": 96.53,
+      "learning_rate": 7.838312909686528e-05,
+      "loss": 0.0,
+      "step": 22588
+    },
+    {
+      "epoch": 96.55,
+      "learning_rate": 7.837353378770024e-05,
+      "loss": 0.0001,
+      "step": 22592
+    },
+    {
+      "epoch": 96.56,
+      "learning_rate": 7.836393693702588e-05,
+      "loss": 0.0001,
+      "step": 22596
+    },
+    {
+      "epoch": 96.58,
+      "learning_rate": 7.835433854536359e-05,
+      "loss": 0.0001,
+      "step": 22600
+    },
+    {
+      "epoch": 96.6,
+      "learning_rate": 7.834473861323483e-05,
+      "loss": 0.0,
+      "step": 22604
+    },
+    {
+      "epoch": 96.62,
+      "learning_rate": 7.833513714116117e-05,
+      "loss": 0.0001,
+      "step": 22608
+    },
+    {
+      "epoch": 96.63,
+      "learning_rate": 7.832553412966421e-05,
+      "loss": 0.0005,
+      "step": 22612
+    },
+    {
+      "epoch": 96.65,
+      "learning_rate": 7.831592957926572e-05,
+      "loss": 0.0026,
+      "step": 22616
+    },
+    {
+      "epoch": 96.67,
+      "learning_rate": 7.830632349048748e-05,
+      "loss": 0.0,
+      "step": 22620
+    },
+    {
+      "epoch": 96.68,
+      "learning_rate": 7.829671586385142e-05,
+      "loss": 0.0,
+      "step": 22624
+    },
+    {
+      "epoch": 96.7,
+      "learning_rate": 7.828710669987946e-05,
+      "loss": 0.0002,
+      "step": 22628
+    },
+    {
+      "epoch": 96.72,
+      "learning_rate": 7.827749599909368e-05,
+      "loss": 0.0002,
+      "step": 22632
+    },
+    {
+      "epoch": 96.74,
+      "learning_rate": 7.826788376201622e-05,
+      "loss": 0.0007,
+      "step": 22636
+    },
+    {
+      "epoch": 96.75,
+      "learning_rate": 7.825826998916932e-05,
+      "loss": 0.0001,
+      "step": 22640
+    },
+    {
+      "epoch": 96.77,
+      "learning_rate": 7.824865468107528e-05,
+      "loss": 0.0002,
+      "step": 22644
+    },
+    {
+      "epoch": 96.79,
+      "learning_rate": 7.823903783825646e-05,
+      "loss": 0.0002,
+      "step": 22648
+    },
+    {
+      "epoch": 96.8,
+      "learning_rate": 7.822941946123538e-05,
+      "loss": 0.0001,
+      "step": 22652
+    },
+    {
+      "epoch": 96.82,
+      "learning_rate": 7.821979955053458e-05,
+      "loss": 0.0,
+      "step": 22656
+    },
+    {
+      "epoch": 96.84,
+      "learning_rate": 7.821017810667669e-05,
+      "loss": 0.0002,
+      "step": 22660
+    },
+    {
+      "epoch": 96.85,
+      "learning_rate": 7.820055513018445e-05,
+      "loss": 0.0,
+      "step": 22664
+    },
+    {
+      "epoch": 96.87,
+      "learning_rate": 7.819093062158067e-05,
+      "loss": 0.0,
+      "step": 22668
+    },
+    {
+      "epoch": 96.89,
+      "learning_rate": 7.818130458138823e-05,
+      "loss": 0.0001,
+      "step": 22672
+    },
+    {
+      "epoch": 96.91,
+      "learning_rate": 7.81716770101301e-05,
+      "loss": 0.0001,
+      "step": 22676
+    },
+    {
+      "epoch": 96.92,
+      "learning_rate": 7.816204790832935e-05,
+      "loss": 0.0001,
+      "step": 22680
+    },
+    {
+      "epoch": 96.94,
+      "learning_rate": 7.81524172765091e-05,
+      "loss": 0.0004,
+      "step": 22684
+    },
+    {
+      "epoch": 96.96,
+      "learning_rate": 7.814278511519262e-05,
+      "loss": 0.0001,
+      "step": 22688
+    },
+    {
+      "epoch": 96.97,
+      "learning_rate": 7.813315142490318e-05,
+      "loss": 0.0002,
+      "step": 22692
+    },
+    {
+      "epoch": 96.99,
+      "learning_rate": 7.812351620616416e-05,
+      "loss": 0.0,
+      "step": 22696
+    },
+    {
+      "epoch": 97.01,
+      "learning_rate": 7.811387945949906e-05,
+      "loss": 0.0002,
+      "step": 22700
+    },
+    {
+      "epoch": 97.03,
+      "learning_rate": 7.810424118543143e-05,
+      "loss": 0.0001,
+      "step": 22704
+    },
+    {
+      "epoch": 97.04,
+      "learning_rate": 7.809460138448491e-05,
+      "loss": 0.0,
+      "step": 22708
+    },
+    {
+      "epoch": 97.06,
+      "learning_rate": 7.808496005718323e-05,
+      "loss": 0.0,
+      "step": 22712
+    },
+    {
+      "epoch": 97.08,
+      "learning_rate": 7.807531720405015e-05,
+      "loss": 0.0002,
+      "step": 22716
+    },
+    {
+      "epoch": 97.09,
+      "learning_rate": 7.806567282560959e-05,
+      "loss": 0.0001,
+      "step": 22720
+    },
+    {
+      "epoch": 97.11,
+      "learning_rate": 7.805602692238554e-05,
+      "loss": 0.0001,
+      "step": 22724
+    },
+    {
+      "epoch": 97.13,
+      "learning_rate": 7.804637949490203e-05,
+      "loss": 0.0002,
+      "step": 22728
+    },
+    {
+      "epoch": 97.15,
+      "learning_rate": 7.80367305436832e-05,
+      "loss": 0.0001,
+      "step": 22732
+    },
+    {
+      "epoch": 97.16,
+      "learning_rate": 7.802708006925326e-05,
+      "loss": 0.0,
+      "step": 22736
+    },
+    {
+      "epoch": 97.18,
+      "learning_rate": 7.801742807213652e-05,
+      "loss": 0.0005,
+      "step": 22740
+    },
+    {
+      "epoch": 97.2,
+      "learning_rate": 7.80077745528574e-05,
+      "loss": 0.0,
+      "step": 22744
+    },
+    {
+      "epoch": 97.21,
+      "learning_rate": 7.79981195119403e-05,
+      "loss": 0.0,
+      "step": 22748
+    },
+    {
+      "epoch": 97.23,
+      "learning_rate": 7.79884629499098e-05,
+      "loss": 0.0002,
+      "step": 22752
+    },
+    {
+      "epoch": 97.25,
+      "learning_rate": 7.797880486729055e-05,
+      "loss": 0.0,
+      "step": 22756
+    },
+    {
+      "epoch": 97.26,
+      "learning_rate": 7.796914526460725e-05,
+      "loss": 0.0002,
+      "step": 22760
+    },
+    {
+      "epoch": 97.28,
+      "learning_rate": 7.79594841423847e-05,
+      "loss": 0.0,
+      "step": 22764
+    },
+    {
+      "epoch": 97.3,
+      "learning_rate": 7.794982150114778e-05,
+      "loss": 0.0006,
+      "step": 22768
+    },
+    {
+      "epoch": 97.32,
+      "learning_rate": 7.794015734142144e-05,
+      "loss": 0.0,
+      "step": 22772
+    },
+    {
+      "epoch": 97.33,
+      "learning_rate": 7.793049166373075e-05,
+      "loss": 0.0008,
+      "step": 22776
+    },
+    {
+      "epoch": 97.35,
+      "learning_rate": 7.792082446860083e-05,
+      "loss": 0.0005,
+      "step": 22780
+    },
+    {
+      "epoch": 97.37,
+      "learning_rate": 7.791115575655687e-05,
+      "loss": 0.0001,
+      "step": 22784
+    },
+    {
+      "epoch": 97.38,
+      "learning_rate": 7.790148552812417e-05,
+      "loss": 0.0,
+      "step": 22788
+    },
+    {
+      "epoch": 97.4,
+      "learning_rate": 7.789181378382816e-05,
+      "loss": 0.0001,
+      "step": 22792
+    },
+    {
+      "epoch": 97.42,
+      "learning_rate": 7.78821405241942e-05,
+      "loss": 0.0001,
+      "step": 22796
+    },
+    {
+      "epoch": 97.44,
+      "learning_rate": 7.78724657497479e-05,
+      "loss": 0.0007,
+      "step": 22800
+    },
+    {
+      "epoch": 97.45,
+      "learning_rate": 7.786278946101487e-05,
+      "loss": 0.0,
+      "step": 22804
+    },
+    {
+      "epoch": 97.47,
+      "learning_rate": 7.785311165852078e-05,
+      "loss": 0.0,
+      "step": 22808
+    },
+    {
+      "epoch": 97.49,
+      "learning_rate": 7.784343234279147e-05,
+      "loss": 0.0,
+      "step": 22812
+    },
+    {
+      "epoch": 97.5,
+      "learning_rate": 7.783375151435277e-05,
+      "loss": 0.0005,
+      "step": 22816
+    },
+    {
+      "epoch": 97.52,
+      "learning_rate": 7.782406917373065e-05,
+      "loss": 0.0,
+      "step": 22820
+    },
+    {
+      "epoch": 97.54,
+      "learning_rate": 7.781438532145114e-05,
+      "loss": 0.0001,
+      "step": 22824
+    },
+    {
+      "epoch": 97.56,
+      "learning_rate": 7.780469995804034e-05,
+      "loss": 0.0002,
+      "step": 22828
+    },
+    {
+      "epoch": 97.57,
+      "learning_rate": 7.779501308402446e-05,
+      "loss": 0.0015,
+      "step": 22832
+    },
+    {
+      "epoch": 97.59,
+      "learning_rate": 7.778532469992977e-05,
+      "loss": 0.0,
+      "step": 22836
+    },
+    {
+      "epoch": 97.61,
+      "learning_rate": 7.777563480628265e-05,
+      "loss": 0.0004,
+      "step": 22840
+    },
+    {
+      "epoch": 97.62,
+      "learning_rate": 7.776594340360954e-05,
+      "loss": 0.0001,
+      "step": 22844
+    },
+    {
+      "epoch": 97.64,
+      "learning_rate": 7.775625049243695e-05,
+      "loss": 0.0002,
+      "step": 22848
+    },
+    {
+      "epoch": 97.66,
+      "learning_rate": 7.77465560732915e-05,
+      "loss": 0.0,
+      "step": 22852
+    },
+    {
+      "epoch": 97.68,
+      "learning_rate": 7.773686014669988e-05,
+      "loss": 0.0,
+      "step": 22856
+    },
+    {
+      "epoch": 97.69,
+      "learning_rate": 7.772716271318884e-05,
+      "loss": 0.0,
+      "step": 22860
+    },
+    {
+      "epoch": 97.71,
+      "learning_rate": 7.771746377328527e-05,
+      "loss": 0.0001,
+      "step": 22864
+    },
+    {
+      "epoch": 97.73,
+      "learning_rate": 7.770776332751606e-05,
+      "loss": 0.0002,
+      "step": 22868
+    },
+    {
+      "epoch": 97.74,
+      "learning_rate": 7.769806137640827e-05,
+      "loss": 0.0001,
+      "step": 22872
+    },
+    {
+      "epoch": 97.76,
+      "learning_rate": 7.768835792048896e-05,
+      "loss": 0.0,
+      "step": 22876
+    },
+    {
+      "epoch": 97.78,
+      "learning_rate": 7.767865296028535e-05,
+      "loss": 0.0,
+      "step": 22880
+    },
+    {
+      "epoch": 97.79,
+      "learning_rate": 7.766894649632468e-05,
+      "loss": 0.0001,
+      "step": 22884
+    },
+    {
+      "epoch": 97.81,
+      "learning_rate": 7.765923852913431e-05,
+      "loss": 0.0005,
+      "step": 22888
+    },
+    {
+      "epoch": 97.83,
+      "learning_rate": 7.764952905924162e-05,
+      "loss": 0.0001,
+      "step": 22892
+    },
+    {
+      "epoch": 97.85,
+      "learning_rate": 7.763981808717416e-05,
+      "loss": 0.0001,
+      "step": 22896
+    },
+    {
+      "epoch": 97.86,
+      "learning_rate": 7.763010561345952e-05,
+      "loss": 0.0,
+      "step": 22900
+    },
+    {
+      "epoch": 97.88,
+      "learning_rate": 7.762039163862533e-05,
+      "loss": 0.0003,
+      "step": 22904
+    },
+    {
+      "epoch": 97.9,
+      "learning_rate": 7.761067616319939e-05,
+      "loss": 0.002,
+      "step": 22908
+    },
+    {
+      "epoch": 97.91,
+      "learning_rate": 7.760095918770951e-05,
+      "loss": 0.0007,
+      "step": 22912
+    },
+    {
+      "epoch": 97.93,
+      "learning_rate": 7.75912407126836e-05,
+      "loss": 0.0,
+      "step": 22916
+    },
+    {
+      "epoch": 97.95,
+      "learning_rate": 7.758152073864967e-05,
+      "loss": 0.0,
+      "step": 22920
+    },
+    {
+      "epoch": 97.97,
+      "learning_rate": 7.757179926613579e-05,
+      "loss": 0.0001,
+      "step": 22924
+    },
+    {
+      "epoch": 97.98,
+      "learning_rate": 7.756207629567011e-05,
+      "loss": 0.0001,
+      "step": 22928
+    },
+    {
+      "epoch": 98.0,
+      "learning_rate": 7.75523518277809e-05,
+      "loss": 0.0,
+      "step": 22932
+    },
+    {
+      "epoch": 98.02,
+      "learning_rate": 7.754262586299645e-05,
+      "loss": 0.0001,
+      "step": 22936
+    },
+    {
+      "epoch": 98.03,
+      "learning_rate": 7.753289840184518e-05,
+      "loss": 0.0,
+      "step": 22940
+    },
+    {
+      "epoch": 98.05,
+      "learning_rate": 7.752316944485556e-05,
+      "loss": 0.0002,
+      "step": 22944
+    },
+    {
+      "epoch": 98.07,
+      "learning_rate": 7.751343899255618e-05,
+      "loss": 0.0,
+      "step": 22948
+    },
+    {
+      "epoch": 98.09,
+      "learning_rate": 7.750370704547567e-05,
+      "loss": 0.0001,
+      "step": 22952
+    },
+    {
+      "epoch": 98.1,
+      "learning_rate": 7.749397360414274e-05,
+      "loss": 0.0001,
+      "step": 22956
+    },
+    {
+      "epoch": 98.12,
+      "learning_rate": 7.748423866908625e-05,
+      "loss": 0.0002,
+      "step": 22960
+    },
+    {
+      "epoch": 98.14,
+      "learning_rate": 7.747450224083505e-05,
+      "loss": 0.0001,
+      "step": 22964
+    },
+    {
+      "epoch": 98.15,
+      "learning_rate": 7.746476431991811e-05,
+      "loss": 0.0,
+      "step": 22968
+    },
+    {
+      "epoch": 98.17,
+      "learning_rate": 7.745502490686452e-05,
+      "loss": 0.0,
+      "step": 22972
+    },
+    {
+      "epoch": 98.19,
+      "learning_rate": 7.744528400220336e-05,
+      "loss": 0.0,
+      "step": 22976
+    },
+    {
+      "epoch": 98.21,
+      "learning_rate": 7.74355416064639e-05,
+      "loss": 0.0,
+      "step": 22980
+    },
+    {
+      "epoch": 98.22,
+      "learning_rate": 7.74257977201754e-05,
+      "loss": 0.0,
+      "step": 22984
+    },
+    {
+      "epoch": 98.24,
+      "learning_rate": 7.741605234386724e-05,
+      "loss": 0.0,
+      "step": 22988
+    },
+    {
+      "epoch": 98.26,
+      "learning_rate": 7.740630547806889e-05,
+      "loss": 0.0001,
+      "step": 22992
+    },
+    {
+      "epoch": 98.27,
+      "learning_rate": 7.739655712330989e-05,
+      "loss": 0.0001,
+      "step": 22996
+    },
+    {
+      "epoch": 98.29,
+      "learning_rate": 7.738680728011983e-05,
+      "loss": 0.0001,
+      "step": 23000
+    },
+    {
+      "epoch": 98.29,
+      "eval_exact_match": 0.5093555093555093,
+      "eval_loss": 0.9610804915428162,
+      "eval_runtime": 137.9706,
+      "eval_samples_per_second": 6.973,
+      "step": 23000
+    },
+    {
+      "epoch": 98.31,
+      "learning_rate": 7.737705594902847e-05,
+      "loss": 0.0007,
+      "step": 23004
+    },
+    {
+      "epoch": 98.32,
+      "learning_rate": 7.736730313056552e-05,
+      "loss": 0.0,
+      "step": 23008
+    },
+    {
+      "epoch": 98.34,
+      "learning_rate": 7.73575488252609e-05,
+      "loss": 0.0001,
+      "step": 23012
+    },
+    {
+      "epoch": 98.36,
+      "learning_rate": 7.73477930336445e-05,
+      "loss": 0.0001,
+      "step": 23016
+    },
+    {
+      "epoch": 98.38,
+      "learning_rate": 7.73380357562464e-05,
+      "loss": 0.0001,
+      "step": 23020
+    },
+    {
+      "epoch": 98.39,
+      "learning_rate": 7.732827699359668e-05,
+      "loss": 0.0,
+      "step": 23024
+    },
+    {
+      "epoch": 98.41,
+      "learning_rate": 7.73185167462255e-05,
+      "loss": 0.0,
+      "step": 23028
+    },
+    {
+      "epoch": 98.43,
+      "learning_rate": 7.730875501466317e-05,
+      "loss": 0.0001,
+      "step": 23032
+    },
+    {
+      "epoch": 98.44,
+      "learning_rate": 7.729899179944e-05,
+      "loss": 0.0,
+      "step": 23036
+    },
+    {
+      "epoch": 98.46,
+      "learning_rate": 7.728922710108644e-05,
+      "loss": 0.0001,
+      "step": 23040
+    },
+    {
+      "epoch": 98.48,
+      "learning_rate": 7.727946092013298e-05,
+      "loss": 0.0,
+      "step": 23044
+    },
+    {
+      "epoch": 98.5,
+      "learning_rate": 7.726969325711023e-05,
+      "loss": 0.0,
+      "step": 23048
+    },
+    {
+      "epoch": 98.51,
+      "learning_rate": 7.725992411254885e-05,
+      "loss": 0.0,
+      "step": 23052
+    },
+    {
+      "epoch": 98.53,
+      "learning_rate": 7.725015348697956e-05,
+      "loss": 0.0001,
+      "step": 23056
+    },
+    {
+      "epoch": 98.55,
+      "learning_rate": 7.724038138093324e-05,
+      "loss": 0.0001,
+      "step": 23060
+    },
+    {
+      "epoch": 98.56,
+      "learning_rate": 7.723060779494075e-05,
+      "loss": 0.0002,
+      "step": 23064
+    },
+    {
+      "epoch": 98.58,
+      "learning_rate": 7.722083272953314e-05,
+      "loss": 0.0,
+      "step": 23068
+    },
+    {
+      "epoch": 98.6,
+      "learning_rate": 7.721105618524141e-05,
+      "loss": 0.0,
+      "step": 23072
+    },
+    {
+      "epoch": 98.62,
+      "learning_rate": 7.720127816259677e-05,
+      "loss": 0.0002,
+      "step": 23076
+    },
+    {
+      "epoch": 98.63,
+      "learning_rate": 7.719149866213041e-05,
+      "loss": 0.0,
+      "step": 23080
+    },
+    {
+      "epoch": 98.65,
+      "learning_rate": 7.718171768437367e-05,
+      "loss": 0.0001,
+      "step": 23084
+    },
+    {
+      "epoch": 98.67,
+      "learning_rate": 7.717193522985794e-05,
+      "loss": 0.0003,
+      "step": 23088
+    },
+    {
+      "epoch": 98.68,
+      "learning_rate": 7.716215129911467e-05,
+      "loss": 0.0001,
+      "step": 23092
+    },
+    {
+      "epoch": 98.7,
+      "learning_rate": 7.715236589267542e-05,
+      "loss": 0.0001,
+      "step": 23096
+    },
+    {
+      "epoch": 98.72,
+      "learning_rate": 7.714257901107185e-05,
+      "loss": 0.0003,
+      "step": 23100
+    },
+    {
+      "epoch": 98.74,
+      "learning_rate": 7.713279065483565e-05,
+      "loss": 0.0001,
+      "step": 23104
+    },
+    {
+      "epoch": 98.75,
+      "learning_rate": 7.712300082449862e-05,
+      "loss": 0.0,
+      "step": 23108
+    },
+    {
+      "epoch": 98.77,
+      "learning_rate": 7.71132095205926e-05,
+      "loss": 0.0028,
+      "step": 23112
+    },
+    {
+      "epoch": 98.79,
+      "learning_rate": 7.710341674364958e-05,
+      "loss": 0.0004,
+      "step": 23116
+    },
+    {
+      "epoch": 98.8,
+      "learning_rate": 7.70936224942016e-05,
+      "loss": 0.0001,
+      "step": 23120
+    },
+    {
+      "epoch": 98.82,
+      "learning_rate": 7.708382677278074e-05,
+      "loss": 0.0003,
+      "step": 23124
+    },
+    {
+      "epoch": 98.84,
+      "learning_rate": 7.707402957991923e-05,
+      "loss": 0.0,
+      "step": 23128
+    },
+    {
+      "epoch": 98.85,
+      "learning_rate": 7.70642309161493e-05,
+      "loss": 0.0,
+      "step": 23132
+    },
+    {
+      "epoch": 98.87,
+      "learning_rate": 7.705443078200333e-05,
+      "loss": 0.0001,
+      "step": 23136
+    },
+    {
+      "epoch": 98.89,
+      "learning_rate": 7.704462917801376e-05,
+      "loss": 0.0001,
+      "step": 23140
+    },
+    {
+      "epoch": 98.91,
+      "learning_rate": 7.703482610471309e-05,
+      "loss": 0.0,
+      "step": 23144
+    },
+    {
+      "epoch": 98.92,
+      "learning_rate": 7.70250215626339e-05,
+      "loss": 0.0,
+      "step": 23148
+    },
+    {
+      "epoch": 98.94,
+      "learning_rate": 7.701521555230888e-05,
+      "loss": 0.0,
+      "step": 23152
+    },
+    {
+      "epoch": 98.96,
+      "learning_rate": 7.700540807427078e-05,
+      "loss": 0.0,
+      "step": 23156
+    },
+    {
+      "epoch": 98.97,
+      "learning_rate": 7.699559912905243e-05,
+      "loss": 0.0003,
+      "step": 23160
+    },
+    {
+      "epoch": 98.99,
+      "learning_rate": 7.698578871718673e-05,
+      "loss": 0.0,
+      "step": 23164
+    },
+    {
+      "epoch": 99.01,
+      "learning_rate": 7.697597683920669e-05,
+      "loss": 0.0006,
+      "step": 23168
+    },
+    {
+      "epoch": 99.03,
+      "learning_rate": 7.696616349564539e-05,
+      "loss": 0.0,
+      "step": 23172
+    },
+    {
+      "epoch": 99.04,
+      "learning_rate": 7.695634868703594e-05,
+      "loss": 0.0,
+      "step": 23176
+    },
+    {
+      "epoch": 99.06,
+      "learning_rate": 7.694653241391161e-05,
+      "loss": 0.0002,
+      "step": 23180
+    },
+    {
+      "epoch": 99.08,
+      "learning_rate": 7.693671467680567e-05,
+      "loss": 0.0002,
+      "step": 23184
+    },
+    {
+      "epoch": 99.09,
+      "learning_rate": 7.692689547625154e-05,
+      "loss": 0.0,
+      "step": 23188
+    },
+    {
+      "epoch": 99.11,
+      "learning_rate": 7.69170748127827e-05,
+      "loss": 0.0001,
+      "step": 23192
+    },
+    {
+      "epoch": 99.13,
+      "learning_rate": 7.690725268693266e-05,
+      "loss": 0.0,
+      "step": 23196
+    },
+    {
+      "epoch": 99.15,
+      "learning_rate": 7.689742909923505e-05,
+      "loss": 0.0,
+      "step": 23200
+    },
+    {
+      "epoch": 99.16,
+      "learning_rate": 7.688760405022362e-05,
+      "loss": 0.0,
+      "step": 23204
+    },
+    {
+      "epoch": 99.18,
+      "learning_rate": 7.687777754043212e-05,
+      "loss": 0.0015,
+      "step": 23208
+    },
+    {
+      "epoch": 99.2,
+      "learning_rate": 7.686794957039442e-05,
+      "loss": 0.0001,
+      "step": 23212
+    },
+    {
+      "epoch": 99.21,
+      "learning_rate": 7.685812014064448e-05,
+      "loss": 0.0,
+      "step": 23216
+    },
+    {
+      "epoch": 99.23,
+      "learning_rate": 7.68482892517163e-05,
+      "loss": 0.0,
+      "step": 23220
+    },
+    {
+      "epoch": 99.25,
+      "learning_rate": 7.683845690414402e-05,
+      "loss": 0.0002,
+      "step": 23224
+    },
+    {
+      "epoch": 99.26,
+      "learning_rate": 7.682862309846177e-05,
+      "loss": 0.0008,
+      "step": 23228
+    },
+    {
+      "epoch": 99.28,
+      "learning_rate": 7.681878783520386e-05,
+      "loss": 0.0,
+      "step": 23232
+    },
+    {
+      "epoch": 99.3,
+      "learning_rate": 7.680895111490462e-05,
+      "loss": 0.0,
+      "step": 23236
+    },
+    {
+      "epoch": 99.32,
+      "learning_rate": 7.679911293809846e-05,
+      "loss": 0.0002,
+      "step": 23240
+    },
+    {
+      "epoch": 99.33,
+      "learning_rate": 7.678927330531988e-05,
+      "loss": 0.0001,
+      "step": 23244
+    },
+    {
+      "epoch": 99.35,
+      "learning_rate": 7.677943221710346e-05,
+      "loss": 0.0001,
+      "step": 23248
+    },
+    {
+      "epoch": 99.37,
+      "learning_rate": 7.676958967398386e-05,
+      "loss": 0.0001,
+      "step": 23252
+    },
+    {
+      "epoch": 99.38,
+      "learning_rate": 7.675974567649583e-05,
+      "loss": 0.0001,
+      "step": 23256
+    },
+    {
+      "epoch": 99.4,
+      "learning_rate": 7.674990022517417e-05,
+      "loss": 0.0005,
+      "step": 23260
+    },
+    {
+      "epoch": 99.42,
+      "learning_rate": 7.674005332055376e-05,
+      "loss": 0.0,
+      "step": 23264
+    },
+    {
+      "epoch": 99.44,
+      "learning_rate": 7.673020496316961e-05,
+      "loss": 0.0,
+      "step": 23268
+    },
+    {
+      "epoch": 99.45,
+      "learning_rate": 7.672035515355675e-05,
+      "loss": 0.0,
+      "step": 23272
+    },
+    {
+      "epoch": 99.47,
+      "learning_rate": 7.67105038922503e-05,
+      "loss": 0.0,
+      "step": 23276
+    },
+    {
+      "epoch": 99.49,
+      "learning_rate": 7.670065117978552e-05,
+      "loss": 0.0,
+      "step": 23280
+    },
+    {
+      "epoch": 99.5,
+      "learning_rate": 7.669079701669762e-05,
+      "loss": 0.0,
+      "step": 23284
+    },
+    {
+      "epoch": 99.52,
+      "learning_rate": 7.668094140352205e-05,
+      "loss": 0.0,
+      "step": 23288
+    },
+    {
+      "epoch": 99.54,
+      "learning_rate": 7.667108434079421e-05,
+      "loss": 0.0,
+      "step": 23292
+    },
+    {
+      "epoch": 99.56,
+      "learning_rate": 7.666122582904962e-05,
+      "loss": 0.0,
+      "step": 23296
+    },
+    {
+      "epoch": 99.57,
+      "learning_rate": 7.665136586882391e-05,
+      "loss": 0.0,
+      "step": 23300
+    },
+    {
+      "epoch": 99.59,
+      "learning_rate": 7.664150446065278e-05,
+      "loss": 0.0001,
+      "step": 23304
+    },
+    {
+      "epoch": 99.61,
+      "learning_rate": 7.663164160507192e-05,
+      "loss": 0.0,
+      "step": 23308
+    },
+    {
+      "epoch": 99.62,
+      "learning_rate": 7.662177730261723e-05,
+      "loss": 0.0,
+      "step": 23312
+    },
+    {
+      "epoch": 99.64,
+      "learning_rate": 7.661191155382464e-05,
+      "loss": 0.0001,
+      "step": 23316
+    },
+    {
+      "epoch": 99.66,
+      "learning_rate": 7.66020443592301e-05,
+      "loss": 0.0,
+      "step": 23320
+    },
+    {
+      "epoch": 99.68,
+      "learning_rate": 7.659217571936969e-05,
+      "loss": 0.0,
+      "step": 23324
+    },
+    {
+      "epoch": 99.69,
+      "learning_rate": 7.658230563477962e-05,
+      "loss": 0.0001,
+      "step": 23328
+    },
+    {
+      "epoch": 99.71,
+      "learning_rate": 7.657243410599608e-05,
+      "loss": 0.0,
+      "step": 23332
+    },
+    {
+      "epoch": 99.73,
+      "learning_rate": 7.656256113355536e-05,
+      "loss": 0.0006,
+      "step": 23336
+    },
+    {
+      "epoch": 99.74,
+      "learning_rate": 7.655268671799389e-05,
+      "loss": 0.0,
+      "step": 23340
+    },
+    {
+      "epoch": 99.76,
+      "learning_rate": 7.654281085984814e-05,
+      "loss": 0.0,
+      "step": 23344
+    },
+    {
+      "epoch": 99.78,
+      "learning_rate": 7.653293355965464e-05,
+      "loss": 0.0,
+      "step": 23348
+    },
+    {
+      "epoch": 99.79,
+      "learning_rate": 7.652305481795e-05,
+      "loss": 0.0002,
+      "step": 23352
+    },
+    {
+      "epoch": 99.81,
+      "learning_rate": 7.651317463527093e-05,
+      "loss": 0.0001,
+      "step": 23356
+    },
+    {
+      "epoch": 99.83,
+      "learning_rate": 7.650329301215424e-05,
+      "loss": 0.0003,
+      "step": 23360
+    },
+    {
+      "epoch": 99.85,
+      "learning_rate": 7.649340994913677e-05,
+      "loss": 0.0001,
+      "step": 23364
+    },
+    {
+      "epoch": 99.86,
+      "learning_rate": 7.648352544675546e-05,
+      "loss": 0.0,
+      "step": 23368
+    },
+    {
+      "epoch": 99.88,
+      "learning_rate": 7.647363950554734e-05,
+      "loss": 0.0001,
+      "step": 23372
+    },
+    {
+      "epoch": 99.9,
+      "learning_rate": 7.646375212604949e-05,
+      "loss": 0.0,
+      "step": 23376
+    },
+    {
+      "epoch": 99.91,
+      "learning_rate": 7.645386330879906e-05,
+      "loss": 0.0,
+      "step": 23380
+    },
+    {
+      "epoch": 99.93,
+      "learning_rate": 7.644397305433334e-05,
+      "loss": 0.0,
+      "step": 23384
+    },
+    {
+      "epoch": 99.95,
+      "learning_rate": 7.643408136318964e-05,
+      "loss": 0.0,
+      "step": 23388
+    },
+    {
+      "epoch": 99.97,
+      "learning_rate": 7.642418823590538e-05,
+      "loss": 0.0,
+      "step": 23392
+    },
+    {
+      "epoch": 99.98,
+      "learning_rate": 7.641429367301802e-05,
+      "loss": 0.0001,
+      "step": 23396
+    },
+    {
+      "epoch": 100.0,
+      "learning_rate": 7.640439767506516e-05,
+      "loss": 0.0005,
+      "step": 23400
+    },
+    {
+      "epoch": 100.02,
+      "learning_rate": 7.63945002425844e-05,
+      "loss": 0.0,
+      "step": 23404
+    },
+    {
+      "epoch": 100.03,
+      "learning_rate": 7.638460137611349e-05,
+      "loss": 0.0005,
+      "step": 23408
+    },
+    {
+      "epoch": 100.05,
+      "learning_rate": 7.637470107619021e-05,
+      "loss": 0.0,
+      "step": 23412
+    },
+    {
+      "epoch": 100.07,
+      "learning_rate": 7.636479934335243e-05,
+      "loss": 0.0002,
+      "step": 23416
+    },
+    {
+      "epoch": 100.09,
+      "learning_rate": 7.635489617813814e-05,
+      "loss": 0.0007,
+      "step": 23420
+    },
+    {
+      "epoch": 100.1,
+      "learning_rate": 7.634499158108532e-05,
+      "loss": 0.0,
+      "step": 23424
+    },
+    {
+      "epoch": 100.12,
+      "learning_rate": 7.633508555273213e-05,
+      "loss": 0.0,
+      "step": 23428
+    },
+    {
+      "epoch": 100.14,
+      "learning_rate": 7.63251780936167e-05,
+      "loss": 0.0,
+      "step": 23432
+    },
+    {
+      "epoch": 100.15,
+      "learning_rate": 7.631526920427732e-05,
+      "loss": 0.0,
+      "step": 23436
+    },
+    {
+      "epoch": 100.17,
+      "learning_rate": 7.630535888525232e-05,
+      "loss": 0.0001,
+      "step": 23440
+    },
+    {
+      "epoch": 100.19,
+      "learning_rate": 7.629544713708016e-05,
+      "loss": 0.0002,
+      "step": 23444
+    },
+    {
+      "epoch": 100.21,
+      "learning_rate": 7.62855339602993e-05,
+      "loss": 0.0001,
+      "step": 23448
+    },
+    {
+      "epoch": 100.22,
+      "learning_rate": 7.627561935544831e-05,
+      "loss": 0.0,
+      "step": 23452
+    },
+    {
+      "epoch": 100.24,
+      "learning_rate": 7.626570332306585e-05,
+      "loss": 0.0,
+      "step": 23456
+    },
+    {
+      "epoch": 100.26,
+      "learning_rate": 7.625578586369067e-05,
+      "loss": 0.0,
+      "step": 23460
+    },
+    {
+      "epoch": 100.27,
+      "learning_rate": 7.624586697786156e-05,
+      "loss": 0.0,
+      "step": 23464
+    },
+    {
+      "epoch": 100.29,
+      "learning_rate": 7.623594666611738e-05,
+      "loss": 0.0,
+      "step": 23468
+    },
+    {
+      "epoch": 100.31,
+      "learning_rate": 7.622602492899714e-05,
+      "loss": 0.0,
+      "step": 23472
+    },
+    {
+      "epoch": 100.32,
+      "learning_rate": 7.621610176703984e-05,
+      "loss": 0.0,
+      "step": 23476
+    },
+    {
+      "epoch": 100.34,
+      "learning_rate": 7.620617718078462e-05,
+      "loss": 0.0001,
+      "step": 23480
+    },
+    {
+      "epoch": 100.36,
+      "learning_rate": 7.619625117077067e-05,
+      "loss": 0.0001,
+      "step": 23484
+    },
+    {
+      "epoch": 100.38,
+      "learning_rate": 7.618632373753724e-05,
+      "loss": 0.0,
+      "step": 23488
+    },
+    {
+      "epoch": 100.39,
+      "learning_rate": 7.617639488162371e-05,
+      "loss": 0.0002,
+      "step": 23492
+    },
+    {
+      "epoch": 100.41,
+      "learning_rate": 7.616646460356949e-05,
+      "loss": 0.0,
+      "step": 23496
+    },
+    {
+      "epoch": 100.43,
+      "learning_rate": 7.615653290391409e-05,
+      "loss": 0.0001,
+      "step": 23500
+    },
+    {
+      "epoch": 100.44,
+      "learning_rate": 7.614659978319709e-05,
+      "loss": 0.0022,
+      "step": 23504
+    },
+    {
+      "epoch": 100.46,
+      "learning_rate": 7.613666524195814e-05,
+      "loss": 0.0002,
+      "step": 23508
+    },
+    {
+      "epoch": 100.48,
+      "learning_rate": 7.612672928073698e-05,
+      "loss": 0.0004,
+      "step": 23512
+    },
+    {
+      "epoch": 100.5,
+      "learning_rate": 7.611679190007342e-05,
+      "loss": 0.001,
+      "step": 23516
+    },
+    {
+      "epoch": 100.51,
+      "learning_rate": 7.610685310050733e-05,
+      "loss": 0.0,
+      "step": 23520
+    },
+    {
+      "epoch": 100.53,
+      "learning_rate": 7.609691288257872e-05,
+      "loss": 0.0001,
+      "step": 23524
+    },
+    {
+      "epoch": 100.55,
+      "learning_rate": 7.608697124682761e-05,
+      "loss": 0.0009,
+      "step": 23528
+    },
+    {
+      "epoch": 100.56,
+      "learning_rate": 7.607702819379413e-05,
+      "loss": 0.0,
+      "step": 23532
+    },
+    {
+      "epoch": 100.58,
+      "learning_rate": 7.606708372401843e-05,
+      "loss": 0.0,
+      "step": 23536
+    },
+    {
+      "epoch": 100.6,
+      "learning_rate": 7.605713783804085e-05,
+      "loss": 0.0002,
+      "step": 23540
+    },
+    {
+      "epoch": 100.62,
+      "learning_rate": 7.604719053640172e-05,
+      "loss": 0.0,
+      "step": 23544
+    },
+    {
+      "epoch": 100.63,
+      "learning_rate": 7.603724181964144e-05,
+      "loss": 0.0,
+      "step": 23548
+    },
+    {
+      "epoch": 100.65,
+      "learning_rate": 7.602729168830056e-05,
+      "loss": 0.0014,
+      "step": 23552
+    },
+    {
+      "epoch": 100.67,
+      "learning_rate": 7.601734014291961e-05,
+      "loss": 0.0008,
+      "step": 23556
+    },
+    {
+      "epoch": 100.68,
+      "learning_rate": 7.60073871840393e-05,
+      "loss": 0.0003,
+      "step": 23560
+    },
+    {
+      "epoch": 100.7,
+      "learning_rate": 7.599743281220033e-05,
+      "loss": 0.0001,
+      "step": 23564
+    },
+    {
+      "epoch": 100.72,
+      "learning_rate": 7.598747702794353e-05,
+      "loss": 0.0,
+      "step": 23568
+    },
+    {
+      "epoch": 100.74,
+      "learning_rate": 7.59775198318098e-05,
+      "loss": 0.0,
+      "step": 23572
+    },
+    {
+      "epoch": 100.75,
+      "learning_rate": 7.596756122434007e-05,
+      "loss": 0.0,
+      "step": 23576
+    },
+    {
+      "epoch": 100.77,
+      "learning_rate": 7.59576012060754e-05,
+      "loss": 0.0023,
+      "step": 23580
+    },
+    {
+      "epoch": 100.79,
+      "learning_rate": 7.594763977755692e-05,
+      "loss": 0.0008,
+      "step": 23584
+    },
+    {
+      "epoch": 100.8,
+      "learning_rate": 7.593767693932583e-05,
+      "loss": 0.0,
+      "step": 23588
+    },
+    {
+      "epoch": 100.82,
+      "learning_rate": 7.592771269192339e-05,
+      "loss": 0.0,
+      "step": 23592
+    },
+    {
+      "epoch": 100.84,
+      "learning_rate": 7.591774703589092e-05,
+      "loss": 0.0003,
+      "step": 23596
+    },
+    {
+      "epoch": 100.85,
+      "learning_rate": 7.59077799717699e-05,
+      "loss": 0.0,
+      "step": 23600
+    },
+    {
+      "epoch": 100.87,
+      "learning_rate": 7.58978115001018e-05,
+      "loss": 0.0,
+      "step": 23604
+    },
+    {
+      "epoch": 100.89,
+      "learning_rate": 7.588784162142819e-05,
+      "loss": 0.0,
+      "step": 23608
+    },
+    {
+      "epoch": 100.91,
+      "learning_rate": 7.587787033629075e-05,
+      "loss": 0.0,
+      "step": 23612
+    },
+    {
+      "epoch": 100.92,
+      "learning_rate": 7.586789764523121e-05,
+      "loss": 0.0001,
+      "step": 23616
+    },
+    {
+      "epoch": 100.94,
+      "learning_rate": 7.585792354879136e-05,
+      "loss": 0.0,
+      "step": 23620
+    },
+    {
+      "epoch": 100.96,
+      "learning_rate": 7.584794804751309e-05,
+      "loss": 0.0001,
+      "step": 23624
+    },
+    {
+      "epoch": 100.97,
+      "learning_rate": 7.583797114193836e-05,
+      "loss": 0.0001,
+      "step": 23628
+    },
+    {
+      "epoch": 100.99,
+      "learning_rate": 7.582799283260921e-05,
+      "loss": 0.0,
+      "step": 23632
+    },
+    {
+      "epoch": 101.01,
+      "learning_rate": 7.581801312006776e-05,
+      "loss": 0.0,
+      "step": 23636
+    },
+    {
+      "epoch": 101.03,
+      "learning_rate": 7.580803200485618e-05,
+      "loss": 0.0,
+      "step": 23640
+    },
+    {
+      "epoch": 101.04,
+      "learning_rate": 7.579804948751675e-05,
+      "loss": 0.0,
+      "step": 23644
+    },
+    {
+      "epoch": 101.06,
+      "learning_rate": 7.57880655685918e-05,
+      "loss": 0.0001,
+      "step": 23648
+    },
+    {
+      "epoch": 101.08,
+      "learning_rate": 7.577808024862378e-05,
+      "loss": 0.0,
+      "step": 23652
+    },
+    {
+      "epoch": 101.09,
+      "learning_rate": 7.576809352815512e-05,
+      "loss": 0.0,
+      "step": 23656
+    },
+    {
+      "epoch": 101.11,
+      "learning_rate": 7.575810540772846e-05,
+      "loss": 0.0,
+      "step": 23660
+    },
+    {
+      "epoch": 101.13,
+      "learning_rate": 7.574811588788641e-05,
+      "loss": 0.0,
+      "step": 23664
+    },
+    {
+      "epoch": 101.15,
+      "learning_rate": 7.57381249691717e-05,
+      "loss": 0.0001,
+      "step": 23668
+    },
+    {
+      "epoch": 101.16,
+      "learning_rate": 7.572813265212711e-05,
+      "loss": 0.0001,
+      "step": 23672
+    },
+    {
+      "epoch": 101.18,
+      "learning_rate": 7.571813893729552e-05,
+      "loss": 0.0002,
+      "step": 23676
+    },
+    {
+      "epoch": 101.2,
+      "learning_rate": 7.570814382521991e-05,
+      "loss": 0.0,
+      "step": 23680
+    },
+    {
+      "epoch": 101.21,
+      "learning_rate": 7.569814731644327e-05,
+      "loss": 0.0013,
+      "step": 23684
+    },
+    {
+      "epoch": 101.23,
+      "learning_rate": 7.568814941150873e-05,
+      "loss": 0.0008,
+      "step": 23688
+    },
+    {
+      "epoch": 101.25,
+      "learning_rate": 7.567815011095945e-05,
+      "loss": 0.0,
+      "step": 23692
+    },
+    {
+      "epoch": 101.26,
+      "learning_rate": 7.566814941533866e-05,
+      "loss": 0.0,
+      "step": 23696
+    },
+    {
+      "epoch": 101.28,
+      "learning_rate": 7.565814732518974e-05,
+      "loss": 0.0017,
+      "step": 23700
+    },
+    {
+      "epoch": 101.3,
+      "learning_rate": 7.564814384105607e-05,
+      "loss": 0.0,
+      "step": 23704
+    },
+    {
+      "epoch": 101.32,
+      "learning_rate": 7.563813896348113e-05,
+      "loss": 0.0,
+      "step": 23708
+    },
+    {
+      "epoch": 101.33,
+      "learning_rate": 7.562813269300845e-05,
+      "loss": 0.0001,
+      "step": 23712
+    },
+    {
+      "epoch": 101.35,
+      "learning_rate": 7.561812503018172e-05,
+      "loss": 0.0,
+      "step": 23716
+    },
+    {
+      "epoch": 101.37,
+      "learning_rate": 7.56081159755446e-05,
+      "loss": 0.0006,
+      "step": 23720
+    },
+    {
+      "epoch": 101.38,
+      "learning_rate": 7.559810552964091e-05,
+      "loss": 0.0,
+      "step": 23724
+    },
+    {
+      "epoch": 101.4,
+      "learning_rate": 7.558809369301447e-05,
+      "loss": 0.0,
+      "step": 23728
+    },
+    {
+      "epoch": 101.42,
+      "learning_rate": 7.557808046620922e-05,
+      "loss": 0.0006,
+      "step": 23732
+    },
+    {
+      "epoch": 101.44,
+      "learning_rate": 7.556806584976919e-05,
+      "loss": 0.0,
+      "step": 23736
+    },
+    {
+      "epoch": 101.45,
+      "learning_rate": 7.555804984423849e-05,
+      "loss": 0.0003,
+      "step": 23740
+    },
+    {
+      "epoch": 101.47,
+      "learning_rate": 7.554803245016123e-05,
+      "loss": 0.0,
+      "step": 23744
+    },
+    {
+      "epoch": 101.49,
+      "learning_rate": 7.553801366808165e-05,
+      "loss": 0.0,
+      "step": 23748
+    },
+    {
+      "epoch": 101.5,
+      "learning_rate": 7.552799349854408e-05,
+      "loss": 0.0009,
+      "step": 23752
+    },
+    {
+      "epoch": 101.52,
+      "learning_rate": 7.551797194209292e-05,
+      "loss": 0.0,
+      "step": 23756
+    },
+    {
+      "epoch": 101.54,
+      "learning_rate": 7.550794899927258e-05,
+      "loss": 0.0,
+      "step": 23760
+    },
+    {
+      "epoch": 101.56,
+      "learning_rate": 7.549792467062768e-05,
+      "loss": 0.0,
+      "step": 23764
+    },
+    {
+      "epoch": 101.57,
+      "learning_rate": 7.548789895670277e-05,
+      "loss": 0.0,
+      "step": 23768
+    },
+    {
+      "epoch": 101.59,
+      "learning_rate": 7.547787185804253e-05,
+      "loss": 0.0,
+      "step": 23772
+    },
+    {
+      "epoch": 101.61,
+      "learning_rate": 7.546784337519177e-05,
+      "loss": 0.0,
+      "step": 23776
+    },
+    {
+      "epoch": 101.62,
+      "learning_rate": 7.545781350869532e-05,
+      "loss": 0.0,
+      "step": 23780
+    },
+    {
+      "epoch": 101.64,
+      "learning_rate": 7.544778225909807e-05,
+      "loss": 0.0,
+      "step": 23784
+    },
+    {
+      "epoch": 101.66,
+      "learning_rate": 7.543774962694501e-05,
+      "loss": 0.0,
+      "step": 23788
+    },
+    {
+      "epoch": 101.68,
+      "learning_rate": 7.542771561278123e-05,
+      "loss": 0.0,
+      "step": 23792
+    },
+    {
+      "epoch": 101.69,
+      "learning_rate": 7.541768021715184e-05,
+      "loss": 0.0001,
+      "step": 23796
+    },
+    {
+      "epoch": 101.71,
+      "learning_rate": 7.540764344060207e-05,
+      "loss": 0.0,
+      "step": 23800
+    },
+    {
+      "epoch": 101.73,
+      "learning_rate": 7.539760528367721e-05,
+      "loss": 0.0,
+      "step": 23804
+    },
+    {
+      "epoch": 101.74,
+      "learning_rate": 7.538756574692263e-05,
+      "loss": 0.0,
+      "step": 23808
+    },
+    {
+      "epoch": 101.76,
+      "learning_rate": 7.537752483088376e-05,
+      "loss": 0.0,
+      "step": 23812
+    },
+    {
+      "epoch": 101.78,
+      "learning_rate": 7.536748253610611e-05,
+      "loss": 0.0,
+      "step": 23816
+    },
+    {
+      "epoch": 101.79,
+      "learning_rate": 7.535743886313526e-05,
+      "loss": 0.0001,
+      "step": 23820
+    },
+    {
+      "epoch": 101.81,
+      "learning_rate": 7.534739381251691e-05,
+      "loss": 0.0011,
+      "step": 23824
+    },
+    {
+      "epoch": 101.83,
+      "learning_rate": 7.533734738479676e-05,
+      "loss": 0.0001,
+      "step": 23828
+    },
+    {
+      "epoch": 101.85,
+      "learning_rate": 7.532729958052065e-05,
+      "loss": 0.0,
+      "step": 23832
+    },
+    {
+      "epoch": 101.86,
+      "learning_rate": 7.531725040023445e-05,
+      "loss": 0.0,
+      "step": 23836
+    },
+    {
+      "epoch": 101.88,
+      "learning_rate": 7.530719984448416e-05,
+      "loss": 0.0,
+      "step": 23840
+    },
+    {
+      "epoch": 101.9,
+      "learning_rate": 7.529714791381575e-05,
+      "loss": 0.0,
+      "step": 23844
+    },
+    {
+      "epoch": 101.91,
+      "learning_rate": 7.528709460877542e-05,
+      "loss": 0.0,
+      "step": 23848
+    },
+    {
+      "epoch": 101.93,
+      "learning_rate": 7.527703992990926e-05,
+      "loss": 0.0,
+      "step": 23852
+    },
+    {
+      "epoch": 101.95,
+      "learning_rate": 7.526698387776361e-05,
+      "loss": 0.0004,
+      "step": 23856
+    },
+    {
+      "epoch": 101.97,
+      "learning_rate": 7.525692645288477e-05,
+      "loss": 0.0,
+      "step": 23860
+    },
+    {
+      "epoch": 101.98,
+      "learning_rate": 7.524686765581916e-05,
+      "loss": 0.0005,
+      "step": 23864
+    },
+    {
+      "epoch": 102.0,
+      "learning_rate": 7.523680748711328e-05,
+      "loss": 0.0001,
+      "step": 23868
+    },
+    {
+      "epoch": 102.02,
+      "learning_rate": 7.522674594731366e-05,
+      "loss": 0.0,
+      "step": 23872
+    },
+    {
+      "epoch": 102.03,
+      "learning_rate": 7.521668303696694e-05,
+      "loss": 0.0,
+      "step": 23876
+    },
+    {
+      "epoch": 102.05,
+      "learning_rate": 7.520661875661987e-05,
+      "loss": 0.0,
+      "step": 23880
+    },
+    {
+      "epoch": 102.07,
+      "learning_rate": 7.519655310681919e-05,
+      "loss": 0.0001,
+      "step": 23884
+    },
+    {
+      "epoch": 102.09,
+      "learning_rate": 7.518648608811176e-05,
+      "loss": 0.0001,
+      "step": 23888
+    },
+    {
+      "epoch": 102.1,
+      "learning_rate": 7.517641770104453e-05,
+      "loss": 0.0001,
+      "step": 23892
+    },
+    {
+      "epoch": 102.12,
+      "learning_rate": 7.516634794616451e-05,
+      "loss": 0.0,
+      "step": 23896
+    },
+    {
+      "epoch": 102.14,
+      "learning_rate": 7.515627682401874e-05,
+      "loss": 0.0,
+      "step": 23900
+    },
+    {
+      "epoch": 102.15,
+      "learning_rate": 7.514620433515443e-05,
+      "loss": 0.0003,
+      "step": 23904
+    },
+    {
+      "epoch": 102.17,
+      "learning_rate": 7.51361304801188e-05,
+      "loss": 0.0,
+      "step": 23908
+    },
+    {
+      "epoch": 102.19,
+      "learning_rate": 7.51260552594591e-05,
+      "loss": 0.0001,
+      "step": 23912
+    },
+    {
+      "epoch": 102.21,
+      "learning_rate": 7.511597867372278e-05,
+      "loss": 0.0,
+      "step": 23916
+    },
+    {
+      "epoch": 102.22,
+      "learning_rate": 7.510590072345724e-05,
+      "loss": 0.0,
+      "step": 23920
+    },
+    {
+      "epoch": 102.24,
+      "learning_rate": 7.509582140921006e-05,
+      "loss": 0.0001,
+      "step": 23924
+    },
+    {
+      "epoch": 102.26,
+      "learning_rate": 7.508574073152876e-05,
+      "loss": 0.0004,
+      "step": 23928
+    },
+    {
+      "epoch": 102.27,
+      "learning_rate": 7.507565869096109e-05,
+      "loss": 0.0,
+      "step": 23932
+    },
+    {
+      "epoch": 102.29,
+      "learning_rate": 7.506557528805476e-05,
+      "loss": 0.0,
+      "step": 23936
+    },
+    {
+      "epoch": 102.31,
+      "learning_rate": 7.50554905233576e-05,
+      "loss": 0.0,
+      "step": 23940
+    },
+    {
+      "epoch": 102.32,
+      "learning_rate": 7.50454043974175e-05,
+      "loss": 0.0,
+      "step": 23944
+    },
+    {
+      "epoch": 102.34,
+      "learning_rate": 7.503531691078245e-05,
+      "loss": 0.0001,
+      "step": 23948
+    },
+    {
+      "epoch": 102.36,
+      "learning_rate": 7.50252280640005e-05,
+      "loss": 0.0,
+      "step": 23952
+    },
+    {
+      "epoch": 102.38,
+      "learning_rate": 7.501513785761971e-05,
+      "loss": 0.0001,
+      "step": 23956
+    },
+    {
+      "epoch": 102.39,
+      "learning_rate": 7.500504629218834e-05,
+      "loss": 0.0,
+      "step": 23960
+    },
+    {
+      "epoch": 102.41,
+      "learning_rate": 7.499495336825463e-05,
+      "loss": 0.0006,
+      "step": 23964
+    },
+    {
+      "epoch": 102.43,
+      "learning_rate": 7.49848590863669e-05,
+      "loss": 0.0,
+      "step": 23968
+    },
+    {
+      "epoch": 102.44,
+      "learning_rate": 7.497476344707359e-05,
+      "loss": 0.0,
+      "step": 23972
+    },
+    {
+      "epoch": 102.46,
+      "learning_rate": 7.49646664509232e-05,
+      "loss": 0.0,
+      "step": 23976
+    },
+    {
+      "epoch": 102.48,
+      "learning_rate": 7.495456809846424e-05,
+      "loss": 0.0002,
+      "step": 23980
+    },
+    {
+      "epoch": 102.5,
+      "learning_rate": 7.494446839024539e-05,
+      "loss": 0.0,
+      "step": 23984
+    },
+    {
+      "epoch": 102.51,
+      "learning_rate": 7.493436732681535e-05,
+      "loss": 0.0001,
+      "step": 23988
+    },
+    {
+      "epoch": 102.53,
+      "learning_rate": 7.492426490872289e-05,
+      "loss": 0.0001,
+      "step": 23992
+    },
+    {
+      "epoch": 102.55,
+      "learning_rate": 7.491416113651686e-05,
+      "loss": 0.0001,
+      "step": 23996
+    },
+    {
+      "epoch": 102.56,
+      "learning_rate": 7.49040560107462e-05,
+      "loss": 0.0,
+      "step": 24000
+    },
+    {
+      "epoch": 102.56,
+      "eval_exact_match": 0.5343035343035343,
+      "eval_loss": 0.9552536010742188,
+      "eval_runtime": 133.8111,
+      "eval_samples_per_second": 7.189,
+      "step": 24000
+    }
+  ],
+  "max_steps": 59904,
+  "num_train_epochs": 256,
+  "total_flos": 2.0936137837879296e+17,
+  "trial_name": null,
+  "trial_params": null
+}