diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12307 @@
+{
+  "best_metric": 0.5938104448742747,
+  "best_model_checkpoint": "/experiments/spider/557ccb7b-aa17-467d-95d5-80d61ba2c4cf/checkpoint-7488",
+  "epoch": 2495.8777142857143,
+  "global_step": 7488,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001,
+      "loss": 3.7561,
+      "step": 1
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001,
+      "loss": 3.3206,
+      "step": 4
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001,
+      "loss": 2.2708,
+      "step": 8
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0001,
+      "loss": 1.5884,
+      "step": 12
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0001,
+      "loss": 1.2895,
+      "step": 16
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0001,
+      "loss": 0.9622,
+      "step": 20
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.0001,
+      "loss": 0.8593,
+      "step": 24
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 0.0001,
+      "loss": 0.8293,
+      "step": 28
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 0.0001,
+      "loss": 0.6515,
+      "step": 32
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 0.0001,
+      "loss": 0.6408,
+      "step": 36
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 0.0001,
+      "loss": 0.6034,
+      "step": 40
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 0.0001,
+      "loss": 0.5351,
+      "step": 44
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 0.0001,
+      "loss": 0.4827,
+      "step": 48
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 0.0001,
+      "loss": 0.4953,
+      "step": 52
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 0.0001,
+      "loss": 0.4265,
+      "step": 56
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 0.0001,
+      "loss": 0.4181,
+      "step": 60
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 0.0001,
+      "loss": 0.4214,
+      "step": 64
+    },
+    {
+      "epoch": 21.29,
+      "eval_exact_match": 0.28916827852998067,
+      "eval_exec": 0.2911025145067698,
+      "eval_loss": 0.2972656488418579,
+      "eval_runtime": 1033.1993,
+      "eval_samples_per_second": 1.001,
+      "step": 64
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 0.0001,
+      "loss": 0.3768,
+      "step": 68
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 0.0001,
+      "loss": 0.3647,
+      "step": 72
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 0.0001,
+      "loss": 0.3637,
+      "step": 76
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 0.0001,
+      "loss": 0.3269,
+      "step": 80
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 0.0001,
+      "loss": 0.3012,
+      "step": 84
+    },
+    {
+      "epoch": 29.29,
+      "learning_rate": 0.0001,
+      "loss": 0.316,
+      "step": 88
+    },
+    {
+      "epoch": 30.59,
+      "learning_rate": 0.0001,
+      "loss": 0.2855,
+      "step": 92
+    },
+    {
+      "epoch": 31.88,
+      "learning_rate": 0.0001,
+      "loss": 0.2687,
+      "step": 96
+    },
+    {
+      "epoch": 33.29,
+      "learning_rate": 0.0001,
+      "loss": 0.2737,
+      "step": 100
+    },
+    {
+      "epoch": 34.59,
+      "learning_rate": 0.0001,
+      "loss": 0.2494,
+      "step": 104
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 0.0001,
+      "loss": 0.2376,
+      "step": 108
+    },
+    {
+      "epoch": 37.29,
+      "learning_rate": 0.0001,
+      "loss": 0.2512,
+      "step": 112
+    },
+    {
+      "epoch": 38.59,
+      "learning_rate": 0.0001,
+      "loss": 0.2265,
+      "step": 116
+    },
+    {
+      "epoch": 39.88,
+      "learning_rate": 0.0001,
+      "loss": 0.2214,
+      "step": 120
+    },
+    {
+      "epoch": 41.29,
+      "learning_rate": 0.0001,
+      "loss": 0.2245,
+      "step": 124
+    },
+    {
+      "epoch": 42.59,
+      "learning_rate": 0.0001,
+      "loss": 0.2053,
+      "step": 128
+    },
+    {
+      "epoch": 42.59,
+      "eval_exact_match": 0.4032882011605416,
+      "eval_exec": 0.3965183752417795,
+      "eval_loss": 0.21602891385555267,
+      "eval_runtime": 1061.2137,
+      "eval_samples_per_second": 0.974,
+      "step": 128
+    },
+    {
+      "epoch": 43.88,
+      "learning_rate": 0.0001,
+      "loss": 0.2026,
+      "step": 132
+    },
+    {
+      "epoch": 45.29,
+      "learning_rate": 0.0001,
+      "loss": 0.2159,
+      "step": 136
+    },
+    {
+      "epoch": 46.59,
+      "learning_rate": 0.0001,
+      "loss": 0.1857,
+      "step": 140
+    },
+    {
+      "epoch": 47.88,
+      "learning_rate": 0.0001,
+      "loss": 0.1828,
+      "step": 144
+    },
+    {
+      "epoch": 49.29,
+      "learning_rate": 0.0001,
+      "loss": 0.1944,
+      "step": 148
+    },
+    {
+      "epoch": 50.59,
+      "learning_rate": 0.0001,
+      "loss": 0.182,
+      "step": 152
+    },
+    {
+      "epoch": 51.88,
+      "learning_rate": 0.0001,
+      "loss": 0.1716,
+      "step": 156
+    },
+    {
+      "epoch": 53.29,
+      "learning_rate": 0.0001,
+      "loss": 0.189,
+      "step": 160
+    },
+    {
+      "epoch": 54.59,
+      "learning_rate": 0.0001,
+      "loss": 0.1636,
+      "step": 164
+    },
+    {
+      "epoch": 55.88,
+      "learning_rate": 0.0001,
+      "loss": 0.158,
+      "step": 168
+    },
+    {
+      "epoch": 57.29,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 172
+    },
+    {
+      "epoch": 58.59,
+      "learning_rate": 0.0001,
+      "loss": 0.1525,
+      "step": 176
+    },
+    {
+      "epoch": 59.88,
+      "learning_rate": 0.0001,
+      "loss": 0.1485,
+      "step": 180
+    },
+    {
+      "epoch": 61.29,
+      "learning_rate": 0.0001,
+      "loss": 0.1543,
+      "step": 184
+    },
+    {
+      "epoch": 62.59,
+      "learning_rate": 0.0001,
+      "loss": 0.1405,
+      "step": 188
+    },
+    {
+      "epoch": 63.88,
+      "learning_rate": 0.0001,
+      "loss": 0.1406,
+      "step": 192
+    },
+    {
+      "epoch": 63.88,
+      "eval_exact_match": 0.46808510638297873,
+      "eval_exec": 0.45551257253384914,
+      "eval_loss": 0.2108125239610672,
+      "eval_runtime": 1232.503,
+      "eval_samples_per_second": 0.839,
+      "step": 192
+    },
+    {
+      "epoch": 65.29,
+      "learning_rate": 0.0001,
+      "loss": 0.1461,
+      "step": 196
+    },
+    {
+      "epoch": 66.59,
+      "learning_rate": 0.0001,
+      "loss": 0.1352,
+      "step": 200
+    },
+    {
+      "epoch": 67.88,
+      "learning_rate": 0.0001,
+      "loss": 0.1237,
+      "step": 204
+    },
+    {
+      "epoch": 69.29,
+      "learning_rate": 0.0001,
+      "loss": 0.1412,
+      "step": 208
+    },
+    {
+      "epoch": 70.59,
+      "learning_rate": 0.0001,
+      "loss": 0.1199,
+      "step": 212
+    },
+    {
+      "epoch": 71.88,
+      "learning_rate": 0.0001,
+      "loss": 0.1223,
+      "step": 216
+    },
+    {
+      "epoch": 73.29,
+      "learning_rate": 0.0001,
+      "loss": 0.1254,
+      "step": 220
+    },
+    {
+      "epoch": 74.59,
+      "learning_rate": 0.0001,
+      "loss": 0.1208,
+      "step": 224
+    },
+    {
+      "epoch": 75.88,
+      "learning_rate": 0.0001,
+      "loss": 0.1152,
+      "step": 228
+    },
+    {
+      "epoch": 77.29,
+      "learning_rate": 0.0001,
+      "loss": 0.1213,
+      "step": 232
+    },
+    {
+      "epoch": 78.59,
+      "learning_rate": 0.0001,
+      "loss": 0.1127,
+      "step": 236
+    },
+    {
+      "epoch": 79.88,
+      "learning_rate": 0.0001,
+      "loss": 0.1089,
+      "step": 240
+    },
+    {
+      "epoch": 81.29,
+      "learning_rate": 0.0001,
+      "loss": 0.1169,
+      "step": 244
+    },
+    {
+      "epoch": 82.59,
+      "learning_rate": 0.0001,
+      "loss": 0.1,
+      "step": 248
+    },
+    {
+      "epoch": 83.88,
+      "learning_rate": 0.0001,
+      "loss": 0.1022,
+      "step": 252
+    },
+    {
+      "epoch": 85.29,
+      "learning_rate": 0.0001,
+      "loss": 0.1093,
+      "step": 256
+    },
+    {
+      "epoch": 85.29,
+      "eval_exact_match": 0.45938104448742745,
+      "eval_exec": 0.47678916827852996,
+      "eval_loss": 0.20979681611061096,
+      "eval_runtime": 1364.3846,
+      "eval_samples_per_second": 0.758,
+      "step": 256
+    },
+    {
+      "epoch": 86.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0979,
+      "step": 260
+    },
+    {
+      "epoch": 87.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0959,
+      "step": 264
+    },
+    {
+      "epoch": 89.29,
+      "learning_rate": 0.0001,
+      "loss": 0.104,
+      "step": 268
+    },
+    {
+      "epoch": 90.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0909,
+      "step": 272
+    },
+    {
+      "epoch": 91.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0876,
+      "step": 276
+    },
+    {
+      "epoch": 93.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0926,
+      "step": 280
+    },
+    {
+      "epoch": 94.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0836,
+      "step": 284
+    },
+    {
+      "epoch": 95.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0898,
+      "step": 288
+    },
+    {
+      "epoch": 97.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0933,
+      "step": 292
+    },
+    {
+      "epoch": 98.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0848,
+      "step": 296
+    },
+    {
+      "epoch": 99.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0813,
+      "step": 300
+    },
+    {
+      "epoch": 101.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0838,
+      "step": 304
+    },
+    {
+      "epoch": 102.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0784,
+      "step": 308
+    },
+    {
+      "epoch": 103.88,
+      "learning_rate": 0.0001,
+      "loss": 0.077,
+      "step": 312
+    },
+    {
+      "epoch": 105.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0826,
+      "step": 316
+    },
+    {
+      "epoch": 106.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0752,
+      "step": 320
+    },
+    {
+      "epoch": 106.59,
+      "eval_exact_match": 0.5077369439071566,
+      "eval_exec": 0.504835589941973,
+      "eval_loss": 0.20611998438835144,
+      "eval_runtime": 1729.5727,
+      "eval_samples_per_second": 0.598,
+      "step": 320
+    },
+    {
+      "epoch": 107.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0736,
+      "step": 324
+    },
+    {
+      "epoch": 109.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0762,
+      "step": 328
+    },
+    {
+      "epoch": 110.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0676,
+      "step": 332
+    },
+    {
+      "epoch": 111.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0701,
+      "step": 336
+    },
+    {
+      "epoch": 113.29,
+      "learning_rate": 0.0001,
+      "loss": 0.072,
+      "step": 340
+    },
+    {
+      "epoch": 114.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0644,
+      "step": 344
+    },
+    {
+      "epoch": 115.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0654,
+      "step": 348
+    },
+    {
+      "epoch": 117.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0704,
+      "step": 352
+    },
+    {
+      "epoch": 118.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0634,
+      "step": 356
+    },
+    {
+      "epoch": 119.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0609,
+      "step": 360
+    },
+    {
+      "epoch": 121.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0674,
+      "step": 364
+    },
+    {
+      "epoch": 122.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0605,
+      "step": 368
+    },
+    {
+      "epoch": 123.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0592,
+      "step": 372
+    },
+    {
+      "epoch": 125.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0606,
+      "step": 376
+    },
+    {
+      "epoch": 126.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0616,
+      "step": 380
+    },
+    {
+      "epoch": 127.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0592,
+      "step": 384
+    },
+    {
+      "epoch": 127.88,
+      "eval_exact_match": 0.5212765957446809,
+      "eval_exec": 0.5174081237911026,
+      "eval_loss": 0.2202136367559433,
+      "eval_runtime": 1391.0459,
+      "eval_samples_per_second": 0.743,
+      "step": 384
+    },
+    {
+      "epoch": 129.29,
+      "learning_rate": 0.0001,
+      "loss": 0.059,
+      "step": 388
+    },
+    {
+      "epoch": 130.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0565,
+      "step": 392
+    },
+    {
+      "epoch": 131.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0524,
+      "step": 396
+    },
+    {
+      "epoch": 133.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0578,
+      "step": 400
+    },
+    {
+      "epoch": 134.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0534,
+      "step": 404
+    },
+    {
+      "epoch": 135.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0494,
+      "step": 408
+    },
+    {
+      "epoch": 137.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0547,
+      "step": 412
+    },
+    {
+      "epoch": 138.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0471,
+      "step": 416
+    },
+    {
+      "epoch": 139.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0453,
+      "step": 420
+    },
+    {
+      "epoch": 141.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0484,
+      "step": 424
+    },
+    {
+      "epoch": 142.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0468,
+      "step": 428
+    },
+    {
+      "epoch": 143.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0447,
+      "step": 432
+    },
+    {
+      "epoch": 145.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0473,
+      "step": 436
+    },
+    {
+      "epoch": 146.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0436,
+      "step": 440
+    },
+    {
+      "epoch": 147.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0451,
+      "step": 444
+    },
+    {
+      "epoch": 149.29,
+      "learning_rate": 0.0001,
+      "loss": 0.05,
+      "step": 448
+    },
+    {
+      "epoch": 149.29,
+      "eval_exact_match": 0.4922630560928433,
+      "eval_exec": 0.5125725338491296,
+      "eval_loss": 0.25774919986724854,
+      "eval_runtime": 1244.2562,
+      "eval_samples_per_second": 0.831,
+      "step": 448
+    },
+    {
+      "epoch": 150.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0444,
+      "step": 452
+    },
+    {
+      "epoch": 151.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0388,
+      "step": 456
+    },
+    {
+      "epoch": 153.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0441,
+      "step": 460
+    },
+    {
+      "epoch": 154.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0399,
+      "step": 464
+    },
+    {
+      "epoch": 155.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 468
+    },
+    {
+      "epoch": 157.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0406,
+      "step": 472
+    },
+    {
+      "epoch": 158.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0369,
+      "step": 476
+    },
+    {
+      "epoch": 159.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0343,
+      "step": 480
+    },
+    {
+      "epoch": 161.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0421,
+      "step": 484
+    },
+    {
+      "epoch": 162.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0393,
+      "step": 488
+    },
+    {
+      "epoch": 163.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0376,
+      "step": 492
+    },
+    {
+      "epoch": 165.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0391,
+      "step": 496
+    },
+    {
+      "epoch": 166.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0358,
+      "step": 500
+    },
+    {
+      "epoch": 167.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 504
+    },
+    {
+      "epoch": 169.29,
+      "learning_rate": 0.0001,
+      "loss": 0.036,
+      "step": 508
+    },
+    {
+      "epoch": 170.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0333,
+      "step": 512
+    },
+    {
+      "epoch": 170.59,
+      "eval_exact_match": 0.5377176015473888,
+      "eval_exec": 0.5328820116054158,
+      "eval_loss": 0.2692297399044037,
+      "eval_runtime": 1006.9413,
+      "eval_samples_per_second": 1.027,
+      "step": 512
+    },
+    {
+      "epoch": 171.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 516
+    },
+    {
+      "epoch": 173.29,
+      "learning_rate": 0.0001,
+      "loss": 0.034,
+      "step": 520
+    },
+    {
+      "epoch": 174.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0312,
+      "step": 524
+    },
+    {
+      "epoch": 175.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0306,
+      "step": 528
+    },
+    {
+      "epoch": 177.29,
+      "learning_rate": 0.0001,
+      "loss": 0.031,
+      "step": 532
+    },
+    {
+      "epoch": 178.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0316,
+      "step": 536
+    },
+    {
+      "epoch": 179.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0294,
+      "step": 540
+    },
+    {
+      "epoch": 181.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0344,
+      "step": 544
+    },
+    {
+      "epoch": 182.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0295,
+      "step": 548
+    },
+    {
+      "epoch": 183.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0305,
+      "step": 552
+    },
+    {
+      "epoch": 185.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0315,
+      "step": 556
+    },
+    {
+      "epoch": 186.59,
+      "learning_rate": 0.0001,
+      "loss": 0.028,
+      "step": 560
+    },
+    {
+      "epoch": 187.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 564
+    },
+    {
+      "epoch": 189.29,
+      "learning_rate": 0.0001,
+      "loss": 0.029,
+      "step": 568
+    },
+    {
+      "epoch": 190.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0256,
+      "step": 572
+    },
+    {
+      "epoch": 191.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0241,
+      "step": 576
+    },
+    {
+      "epoch": 191.88,
+      "eval_exact_match": 0.5483558994197292,
+      "eval_exec": 0.5319148936170213,
+      "eval_loss": 0.2877974212169647,
+      "eval_runtime": 813.3555,
+      "eval_samples_per_second": 1.271,
+      "step": 576
+    },
+    {
+      "epoch": 193.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0286,
+      "step": 580
+    },
+    {
+      "epoch": 194.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0252,
+      "step": 584
+    },
+    {
+      "epoch": 195.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0253,
+      "step": 588
+    },
+    {
+      "epoch": 197.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0274,
+      "step": 592
+    },
+    {
+      "epoch": 198.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0232,
+      "step": 596
+    },
+    {
+      "epoch": 199.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0281,
+      "step": 600
+    },
+    {
+      "epoch": 201.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0265,
+      "step": 604
+    },
+    {
+      "epoch": 202.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0243,
+      "step": 608
+    },
+    {
+      "epoch": 203.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 612
+    },
+    {
+      "epoch": 205.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0248,
+      "step": 616
+    },
+    {
+      "epoch": 206.59,
+      "learning_rate": 0.0001,
+      "loss": 0.023,
+      "step": 620
+    },
+    {
+      "epoch": 207.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0228,
+      "step": 624
+    },
+    {
+      "epoch": 209.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0228,
+      "step": 628
+    },
+    {
+      "epoch": 210.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0205,
+      "step": 632
+    },
+    {
+      "epoch": 211.88,
+      "learning_rate": 0.0001,
+      "loss": 0.021,
+      "step": 636
+    },
+    {
+      "epoch": 213.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0235,
+      "step": 640
+    },
+    {
+      "epoch": 213.29,
+      "eval_exact_match": 0.5502901353965184,
+      "eval_exec": 0.5328820116054158,
+      "eval_loss": 0.30348333716392517,
+      "eval_runtime": 804.5631,
+      "eval_samples_per_second": 1.285,
+      "step": 640
+    },
+    {
+      "epoch": 214.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0194,
+      "step": 644
+    },
+    {
+      "epoch": 215.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 648
+    },
+    {
+      "epoch": 217.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0206,
+      "step": 652
+    },
+    {
+      "epoch": 218.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0187,
+      "step": 656
+    },
+    {
+      "epoch": 219.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0209,
+      "step": 660
+    },
+    {
+      "epoch": 221.29,
+      "learning_rate": 0.0001,
+      "loss": 0.02,
+      "step": 664
+    },
+    {
+      "epoch": 222.59,
+      "learning_rate": 0.0001,
+      "loss": 0.019,
+      "step": 668
+    },
+    {
+      "epoch": 223.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0195,
+      "step": 672
+    },
+    {
+      "epoch": 225.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 676
+    },
+    {
+      "epoch": 226.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0216,
+      "step": 680
+    },
+    {
+      "epoch": 227.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0228,
+      "step": 684
+    },
+    {
+      "epoch": 229.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0207,
+      "step": 688
+    },
+    {
+      "epoch": 230.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0178,
+      "step": 692
+    },
+    {
+      "epoch": 231.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0165,
+      "step": 696
+    },
+    {
+      "epoch": 233.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0184,
+      "step": 700
+    },
+    {
+      "epoch": 234.59,
+      "learning_rate": 0.0001,
+      "loss": 0.016,
+      "step": 704
+    },
+    {
+      "epoch": 234.59,
+      "eval_exact_match": 0.5541586073500967,
+      "eval_exec": 0.5415860735009671,
+      "eval_loss": 0.30872371792793274,
+      "eval_runtime": 765.822,
+      "eval_samples_per_second": 1.35,
+      "step": 704
+    },
+    {
+      "epoch": 235.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0171,
+      "step": 708
+    },
+    {
+      "epoch": 237.29,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 712
+    },
+    {
+      "epoch": 238.59,
+      "learning_rate": 0.0001,
+      "loss": 0.017,
+      "step": 716
+    },
+    {
+      "epoch": 239.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0167,
+      "step": 720
+    },
+    {
+      "epoch": 241.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0166,
+      "step": 724
+    },
+    {
+      "epoch": 242.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 728
+    },
+    {
+      "epoch": 243.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 732
+    },
+    {
+      "epoch": 245.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 736
+    },
+    {
+      "epoch": 246.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 740
+    },
+    {
+      "epoch": 247.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0148,
+      "step": 744
+    },
+    {
+      "epoch": 249.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0144,
+      "step": 748
+    },
+    {
+      "epoch": 250.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0136,
+      "step": 752
+    },
+    {
+      "epoch": 251.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0137,
+      "step": 756
+    },
+    {
+      "epoch": 253.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0152,
+      "step": 760
+    },
+    {
+      "epoch": 254.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0139,
+      "step": 764
+    },
+    {
+      "epoch": 255.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0141,
+      "step": 768
+    },
+    {
+      "epoch": 255.88,
+      "eval_exact_match": 0.5512572533849129,
+      "eval_exec": 0.528046421663443,
+      "eval_loss": 0.32061684131622314,
+      "eval_runtime": 795.9733,
+      "eval_samples_per_second": 1.299,
+      "step": 768
+    },
+    {
+      "epoch": 257.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 772
+    },
+    {
+      "epoch": 258.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0133,
+      "step": 776
+    },
+    {
+      "epoch": 259.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 780
+    },
+    {
+      "epoch": 261.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0132,
+      "step": 784
+    },
+    {
+      "epoch": 262.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0129,
+      "step": 788
+    },
+    {
+      "epoch": 263.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0122,
+      "step": 792
+    },
+    {
+      "epoch": 265.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0147,
+      "step": 796
+    },
+    {
+      "epoch": 266.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0176,
+      "step": 800
+    },
+    {
+      "epoch": 267.88,
+      "learning_rate": 0.0001,
+      "loss": 0.013,
+      "step": 804
+    },
+    {
+      "epoch": 269.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 808
+    },
+    {
+      "epoch": 270.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 812
+    },
+    {
+      "epoch": 271.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 816
+    },
+    {
+      "epoch": 273.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0123,
+      "step": 820
+    },
+    {
+      "epoch": 274.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0117,
+      "step": 824
+    },
+    {
+      "epoch": 275.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 828
+    },
+    {
+      "epoch": 277.29,
+      "learning_rate": 0.0001,
+      "loss": 0.012,
+      "step": 832
+    },
+    {
+      "epoch": 277.29,
+      "eval_exact_match": 0.5609284332688588,
+      "eval_exec": 0.5386847195357833,
+      "eval_loss": 0.33844655752182007,
+      "eval_runtime": 805.2139,
+      "eval_samples_per_second": 1.284,
+      "step": 832
+    },
+    {
+      "epoch": 278.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0109,
+      "step": 836
+    },
+    {
+      "epoch": 279.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 840
+    },
+    {
+      "epoch": 281.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 844
+    },
+    {
+      "epoch": 282.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0093,
+      "step": 848
+    },
+    {
+      "epoch": 283.88,
+      "learning_rate": 0.0001,
+      "loss": 0.011,
+      "step": 852
+    },
+    {
+      "epoch": 285.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0115,
+      "step": 856
+    },
+    {
+      "epoch": 286.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 860
+    },
+    {
+      "epoch": 287.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 864
+    },
+    {
+      "epoch": 289.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0108,
+      "step": 868
+    },
+    {
+      "epoch": 290.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 872
+    },
+    {
+      "epoch": 291.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0102,
+      "step": 876
+    },
+    {
+      "epoch": 293.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0096,
+      "step": 880
+    },
+    {
+      "epoch": 294.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0103,
+      "step": 884
+    },
+    {
+      "epoch": 295.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 888
+    },
+    {
+      "epoch": 297.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 892
+    },
+    {
+      "epoch": 298.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 896
+    },
+    {
+      "epoch": 298.59,
+      "eval_exact_match": 0.5464216634429401,
+      "eval_exec": 0.5328820116054158,
+      "eval_loss": 0.35882410407066345,
+      "eval_runtime": 806.695,
+      "eval_samples_per_second": 1.282,
+      "step": 896
+    },
+    {
+      "epoch": 299.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0097,
+      "step": 900
+    },
+    {
+      "epoch": 301.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0125,
+      "step": 904
+    },
+    {
+      "epoch": 302.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 908
+    },
+    {
+      "epoch": 303.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0084,
+      "step": 912
+    },
+    {
+      "epoch": 305.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0087,
+      "step": 916
+    },
+    {
+      "epoch": 306.59,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 920
+    },
+    {
+      "epoch": 307.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0081,
+      "step": 924
+    },
+    {
+      "epoch": 309.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0089,
+      "step": 928
+    },
+    {
+      "epoch": 310.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0079,
+      "step": 932
+    },
+    {
+      "epoch": 311.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0069,
+      "step": 936
+    },
+    {
+      "epoch": 313.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0086,
+      "step": 940
+    },
+    {
+      "epoch": 314.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 944
+    },
+    {
+      "epoch": 315.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 948
+    },
+    {
+      "epoch": 317.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0088,
+      "step": 952
+    },
+    {
+      "epoch": 318.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0085,
+      "step": 956
+    },
+    {
+      "epoch": 319.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "step": 960
+    },
+    {
+      "epoch": 319.88,
+      "eval_exact_match": 0.5522243713733076,
+      "eval_exec": 0.5270793036750484,
+      "eval_loss": 0.3698420524597168,
+      "eval_runtime": 782.9496,
+      "eval_samples_per_second": 1.321,
+      "step": 960
+    },
+    {
+      "epoch": 321.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0076,
+      "step": 964
+    },
+    {
+      "epoch": 322.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0066,
+      "step": 968
+    },
+    {
+      "epoch": 323.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0067,
+      "step": 972
+    },
+    {
+      "epoch": 325.29,
+      "learning_rate": 0.0001,
+      "loss": 0.008,
+      "step": 976
+    },
+    {
+      "epoch": 326.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0062,
+      "step": 980
+    },
+    {
+      "epoch": 327.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0065,
+      "step": 984
+    },
+    {
+      "epoch": 329.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 988
+    },
+    {
+      "epoch": 330.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "step": 992
+    },
+    {
+      "epoch": 331.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 996
+    },
+    {
+      "epoch": 333.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "step": 1000
+    },
+    {
+      "epoch": 334.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0069,
+      "step": 1004
+    },
+    {
+      "epoch": 335.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0054,
+      "step": 1008
+    },
+    {
+      "epoch": 337.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0068,
+      "step": 1012
+    },
+    {
+      "epoch": 338.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0107,
+      "step": 1016
+    },
+    {
+      "epoch": 339.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0078,
+      "step": 1020
+    },
+    {
+      "epoch": 341.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0073,
+      "step": 1024
+    },
+    {
+      "epoch": 341.29,
+      "eval_exact_match": 0.5541586073500967,
+      "eval_exec": 0.5367504835589942,
+      "eval_loss": 0.3737126886844635,
+      "eval_runtime": 804.0571,
+      "eval_samples_per_second": 1.286,
+      "step": 1024
+    },
+    {
+      "epoch": 342.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0069,
+      "step": 1028
+    },
+    {
+      "epoch": 343.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0056,
+      "step": 1032
+    },
+    {
+      "epoch": 345.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0065,
+      "step": 1036
+    },
+    {
+      "epoch": 346.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0068,
+      "step": 1040
+    },
+    {
+      "epoch": 347.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0072,
+      "step": 1044
+    },
+    {
+      "epoch": 349.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0059,
+      "step": 1048
+    },
+    {
+      "epoch": 350.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0058,
+      "step": 1052
+    },
+    {
+      "epoch": 351.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0058,
+      "step": 1056
+    },
+    {
+      "epoch": 353.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0071,
+      "step": 1060
+    },
+    {
+      "epoch": 354.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0055,
+      "step": 1064
+    },
+    {
+      "epoch": 355.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0053,
+      "step": 1068
+    },
+    {
+      "epoch": 357.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0064,
+      "step": 1072
+    },
+    {
+      "epoch": 358.59,
+      "learning_rate": 0.0001,
+      "loss": 0.005,
+      "step": 1076
+    },
+    {
+      "epoch": 359.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0047,
+      "step": 1080
+    },
+    {
+      "epoch": 361.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0061,
+      "step": 1084
+    },
+    {
+      "epoch": 362.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0047,
+      "step": 1088
+    },
+    {
+      "epoch": 362.59,
+      "eval_exact_match": 0.5531914893617021,
+      "eval_exec": 0.5367504835589942,
+      "eval_loss": 0.38038715720176697,
+      "eval_runtime": 802.0481,
+      "eval_samples_per_second": 1.289,
+      "step": 1088
+    },
+    {
+      "epoch": 363.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0054,
+      "step": 1092
+    },
+    {
+      "epoch": 365.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0053,
+      "step": 1096
+    },
+    {
+      "epoch": 366.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0055,
+      "step": 1100
+    },
+    {
+      "epoch": 367.88,
+      "learning_rate": 0.0001,
+      "loss": 0.005,
+      "step": 1104
+    },
+    {
+      "epoch": 369.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0055,
+      "step": 1108
+    },
+    {
+      "epoch": 370.59,
+      "learning_rate": 0.0001,
+      "loss": 0.005,
+      "step": 1112
+    },
+    {
+      "epoch": 371.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0042,
+      "step": 1116
+    },
+    {
+      "epoch": 373.29,
+      "learning_rate": 0.0001,
+      "loss": 0.005,
+      "step": 1120
+    },
+    {
+      "epoch": 374.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0049,
+      "step": 1124
+    },
+    {
+      "epoch": 375.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0049,
+      "step": 1128
+    },
+    {
+      "epoch": 377.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0051,
+      "step": 1132
+    },
+    {
+      "epoch": 378.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0049,
+      "step": 1136
+    },
+    {
+      "epoch": 379.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0046,
+      "step": 1140
+    },
+    {
+      "epoch": 381.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0046,
+      "step": 1144
+    },
+    {
+      "epoch": 382.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0045,
+      "step": 1148
+    },
+    {
+      "epoch": 383.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0041,
+      "step": 1152
+    },
+    {
+      "epoch": 383.88,
+      "eval_exact_match": 0.5609284332688588,
+      "eval_exec": 0.5406189555125726,
+      "eval_loss": 0.4036901295185089,
+      "eval_runtime": 785.1498,
+      "eval_samples_per_second": 1.317,
+      "step": 1152
+    },
+    {
+      "epoch": 385.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0046,
+      "step": 1156
+    },
+    {
+      "epoch": 386.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0039,
+      "step": 1160
+    },
+    {
+      "epoch": 387.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0045,
+      "step": 1164
+    },
+    {
+      "epoch": 389.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0046,
+      "step": 1168
+    },
+    {
+      "epoch": 390.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0047,
+      "step": 1172
+    },
+    {
+      "epoch": 391.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0038,
+      "step": 1176
+    },
+    {
+      "epoch": 393.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0047,
+      "step": 1180
+    },
+    {
+      "epoch": 394.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0046,
+      "step": 1184
+    },
+    {
+      "epoch": 395.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0041,
+      "step": 1188
+    },
+    {
+      "epoch": 397.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0046,
+      "step": 1192
+    },
+    {
+      "epoch": 398.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0038,
+      "step": 1196
+    },
+    {
+      "epoch": 399.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0041,
+      "step": 1200
+    },
+    {
+      "epoch": 401.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0037,
+      "step": 1204
+    },
+    {
+      "epoch": 402.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0037,
+      "step": 1208
+    },
+    {
+      "epoch": 403.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0039,
+      "step": 1212
+    },
+    {
+      "epoch": 405.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0044,
+      "step": 1216
+    },
+    {
+      "epoch": 405.29,
+      "eval_exact_match": 0.5618955512572534,
+      "eval_exec": 0.5444874274661509,
+      "eval_loss": 0.4164343774318695,
+      "eval_runtime": 781.2432,
+      "eval_samples_per_second": 1.324,
+      "step": 1216
+    },
+    {
+      "epoch": 406.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0037,
+      "step": 1220
+    },
+    {
+      "epoch": 407.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0036,
+      "step": 1224
+    },
+    {
+      "epoch": 409.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0043,
+      "step": 1228
+    },
+    {
+      "epoch": 410.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0041,
+      "step": 1232
+    },
+    {
+      "epoch": 411.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0039,
+      "step": 1236
+    },
+    {
+      "epoch": 413.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0043,
+      "step": 1240
+    },
+    {
+      "epoch": 414.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0035,
+      "step": 1244
+    },
+    {
+      "epoch": 415.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0035,
+      "step": 1248
+    },
+    {
+      "epoch": 417.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0033,
+      "step": 1252
+    },
+    {
+      "epoch": 418.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0042,
+      "step": 1256
+    },
+    {
+      "epoch": 419.88,
+      "learning_rate": 0.0001,
+      "loss": 0.003,
+      "step": 1260
+    },
+    {
+      "epoch": 421.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0037,
+      "step": 1264
+    },
+    {
+      "epoch": 422.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0031,
+      "step": 1268
+    },
+    {
+      "epoch": 423.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0031,
+      "step": 1272
+    },
+    {
+      "epoch": 425.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0037,
+      "step": 1276
+    },
+    {
+      "epoch": 426.59,
+      "learning_rate": 0.0001,
+      "loss": 0.003,
+      "step": 1280
+    },
+    {
+      "epoch": 426.59,
+      "eval_exact_match": 0.5715667311411993,
+      "eval_exec": 0.5493230174081238,
+      "eval_loss": 0.4233261048793793,
+      "eval_runtime": 788.3912,
+      "eval_samples_per_second": 1.312,
+      "step": 1280
+    },
+    {
+      "epoch": 427.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0032,
+      "step": 1284
+    },
+    {
+      "epoch": 429.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0031,
+      "step": 1288
+    },
+    {
+      "epoch": 430.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0035,
+      "step": 1292
+    },
+    {
+      "epoch": 431.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0035,
+      "step": 1296
+    },
+    {
+      "epoch": 433.29,
+      "learning_rate": 0.0001,
+      "loss": 0.004,
+      "step": 1300
+    },
+    {
+      "epoch": 434.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0032,
+      "step": 1304
+    },
+    {
+      "epoch": 435.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0036,
+      "step": 1308
+    },
+    {
+      "epoch": 437.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0031,
+      "step": 1312
+    },
+    {
+      "epoch": 438.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0031,
+      "step": 1316
+    },
+    {
+      "epoch": 439.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0029,
+      "step": 1320
+    },
+    {
+      "epoch": 441.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0032,
+      "step": 1324
+    },
+    {
+      "epoch": 442.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0025,
+      "step": 1328
+    },
+    {
+      "epoch": 443.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0029,
+      "step": 1332
+    },
+    {
+      "epoch": 445.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0033,
+      "step": 1336
+    },
+    {
+      "epoch": 446.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0026,
+      "step": 1340
+    },
+    {
+      "epoch": 447.88,
+      "learning_rate": 0.0001,
+      "loss": 0.003,
+      "step": 1344
+    },
+    {
+      "epoch": 447.88,
+      "eval_exact_match": 0.5406189555125726,
+      "eval_exec": 0.5309477756286267,
+      "eval_loss": 0.4268878102302551,
+      "eval_runtime": 784.731,
+      "eval_samples_per_second": 1.318,
+      "step": 1344
+    },
+    {
+      "epoch": 449.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0036,
+      "step": 1348
+    },
+    {
+      "epoch": 450.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0031,
+      "step": 1352
+    },
+    {
+      "epoch": 451.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0028,
+      "step": 1356
+    },
+    {
+      "epoch": 453.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0033,
+      "step": 1360
+    },
+    {
+      "epoch": 454.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0028,
+      "step": 1364
+    },
+    {
+      "epoch": 455.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0028,
+      "step": 1368
+    },
+    {
+      "epoch": 457.29,
+      "learning_rate": 0.0001,
+      "loss": 0.003,
+      "step": 1372
+    },
+    {
+      "epoch": 458.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0032,
+      "step": 1376
+    },
+    {
+      "epoch": 459.88,
+      "learning_rate": 0.0001,
+      "loss": 0.003,
+      "step": 1380
+    },
+    {
+      "epoch": 461.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0034,
+      "step": 1384
+    },
+    {
+      "epoch": 462.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0026,
+      "step": 1388
+    },
+    {
+      "epoch": 463.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0028,
+      "step": 1392
+    },
+    {
+      "epoch": 465.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0028,
+      "step": 1396
+    },
+    {
+      "epoch": 466.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0029,
+      "step": 1400
+    },
+    {
+      "epoch": 467.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0027,
+      "step": 1404
+    },
+    {
+      "epoch": 469.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0025,
+      "step": 1408
+    },
+    {
+      "epoch": 469.29,
+      "eval_exact_match": 0.5696324951644101,
+      "eval_exec": 0.5454545454545454,
+      "eval_loss": 0.4336416721343994,
+      "eval_runtime": 790.3318,
+      "eval_samples_per_second": 1.308,
+      "step": 1408
+    },
+    {
+      "epoch": 470.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0027,
+      "step": 1412
+    },
+    {
+      "epoch": 471.88,
+      "learning_rate": 0.0001,
+      "loss": 0.003,
+      "step": 1416
+    },
+    {
+      "epoch": 473.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0024,
+      "step": 1420
+    },
+    {
+      "epoch": 474.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0029,
+      "step": 1424
+    },
+    {
+      "epoch": 475.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0025,
+      "step": 1428
+    },
+    {
+      "epoch": 477.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0028,
+      "step": 1432
+    },
+    {
+      "epoch": 478.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0024,
+      "step": 1436
+    },
+    {
+      "epoch": 479.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0021,
+      "step": 1440
+    },
+    {
+      "epoch": 481.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0025,
+      "step": 1444
+    },
+    {
+      "epoch": 482.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0022,
+      "step": 1448
+    },
+    {
+      "epoch": 483.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0027,
+      "step": 1452
+    },
+    {
+      "epoch": 485.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0022,
+      "step": 1456
+    },
+    {
+      "epoch": 486.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0029,
+      "step": 1460
+    },
+    {
+      "epoch": 487.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0025,
+      "step": 1464
+    },
+    {
+      "epoch": 489.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0023,
+      "step": 1468
+    },
+    {
+      "epoch": 490.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0022,
+      "step": 1472
+    },
+    {
+      "epoch": 490.59,
+      "eval_exact_match": 0.5638297872340425,
+      "eval_exec": 0.5473887814313346,
+      "eval_loss": 0.4378665089607239,
+      "eval_runtime": 798.3477,
+      "eval_samples_per_second": 1.295,
+      "step": 1472
+    },
+    {
+      "epoch": 491.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0025,
+      "step": 1476
+    },
+    {
+      "epoch": 493.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0029,
+      "step": 1480
+    },
+    {
+      "epoch": 494.59,
+      "learning_rate": 0.0001,
+      "loss": 0.002,
+      "step": 1484
+    },
+    {
+      "epoch": 495.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0022,
+      "step": 1488
+    },
+    {
+      "epoch": 497.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0029,
+      "step": 1492
+    },
+    {
+      "epoch": 498.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0022,
+      "step": 1496
+    },
+    {
+      "epoch": 499.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0022,
+      "step": 1500
+    },
+    {
+      "epoch": 501.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0026,
+      "step": 1504
+    },
+    {
+      "epoch": 502.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0027,
+      "step": 1508
+    },
+    {
+      "epoch": 503.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0018,
+      "step": 1512
+    },
+    {
+      "epoch": 505.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0033,
+      "step": 1516
+    },
+    {
+      "epoch": 506.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0023,
+      "step": 1520
+    },
+    {
+      "epoch": 507.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0026,
+      "step": 1524
+    },
+    {
+      "epoch": 509.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0067,
+      "step": 1528
+    },
+    {
+      "epoch": 510.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0045,
+      "step": 1532
+    },
+    {
+      "epoch": 511.88,
+      "learning_rate": 0.0001,
+      "loss": 0.002,
+      "step": 1536
+    },
+    {
+      "epoch": 511.88,
+      "eval_exact_match": 0.5676982591876208,
+      "eval_exec": 0.5493230174081238,
+      "eval_loss": 0.4264160394668579,
+      "eval_runtime": 821.4716,
+      "eval_samples_per_second": 1.259,
+      "step": 1536
+    },
+    {
+      "epoch": 513.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0019,
+      "step": 1540
+    },
+    {
+      "epoch": 514.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0022,
+      "step": 1544
+    },
+    {
+      "epoch": 515.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0021,
+      "step": 1548
+    },
+    {
+      "epoch": 517.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0025,
+      "step": 1552
+    },
+    {
+      "epoch": 518.59,
+      "learning_rate": 0.0001,
+      "loss": 0.002,
+      "step": 1556
+    },
+    {
+      "epoch": 519.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0021,
+      "step": 1560
+    },
+    {
+      "epoch": 521.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0029,
+      "step": 1564
+    },
+    {
+      "epoch": 522.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0032,
+      "step": 1568
+    },
+    {
+      "epoch": 523.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0023,
+      "step": 1572
+    },
+    {
+      "epoch": 525.29,
+      "learning_rate": 0.0001,
+      "loss": 0.002,
+      "step": 1576
+    },
+    {
+      "epoch": 526.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0024,
+      "step": 1580
+    },
+    {
+      "epoch": 527.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0021,
+      "step": 1584
+    },
+    {
+      "epoch": 529.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0018,
+      "step": 1588
+    },
+    {
+      "epoch": 530.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0019,
+      "step": 1592
+    },
+    {
+      "epoch": 531.88,
+      "learning_rate": 0.0001,
+      "loss": 0.002,
+      "step": 1596
+    },
+    {
+      "epoch": 533.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0023,
+      "step": 1600
+    },
+    {
+      "epoch": 533.29,
+      "eval_exact_match": 0.5657640232108317,
+      "eval_exec": 0.5473887814313346,
+      "eval_loss": 0.456968754529953,
+      "eval_runtime": 833.296,
+      "eval_samples_per_second": 1.241,
+      "step": 1600
+    },
+    {
+      "epoch": 534.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0074,
+      "step": 1604
+    },
+    {
+      "epoch": 535.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0024,
+      "step": 1608
+    },
+    {
+      "epoch": 537.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0021,
+      "step": 1612
+    },
+    {
+      "epoch": 538.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 1616
+    },
+    {
+      "epoch": 539.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "step": 1620
+    },
+    {
+      "epoch": 541.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0021,
+      "step": 1624
+    },
+    {
+      "epoch": 542.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "step": 1628
+    },
+    {
+      "epoch": 543.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 1632
+    },
+    {
+      "epoch": 545.29,
+      "learning_rate": 0.0001,
+      "loss": 0.002,
+      "step": 1636
+    },
+    {
+      "epoch": 546.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0023,
+      "step": 1640
+    },
+    {
+      "epoch": 547.88,
+      "learning_rate": 0.0001,
+      "loss": 0.002,
+      "step": 1644
+    },
+    {
+      "epoch": 549.29,
+      "learning_rate": 0.0001,
+      "loss": 0.002,
+      "step": 1648
+    },
+    {
+      "epoch": 550.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0018,
+      "step": 1652
+    },
+    {
+      "epoch": 551.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 1656
+    },
+    {
+      "epoch": 553.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0022,
+      "step": 1660
+    },
+    {
+      "epoch": 554.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "step": 1664
+    },
+    {
+      "epoch": 554.59,
+      "eval_exact_match": 0.5715667311411993,
+      "eval_exec": 0.5531914893617021,
+      "eval_loss": 0.4596085846424103,
+      "eval_runtime": 801.1711,
+      "eval_samples_per_second": 1.291,
+      "step": 1664
+    },
+    {
+      "epoch": 555.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0014,
+      "step": 1668
+    },
+    {
+      "epoch": 557.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0016,
+      "step": 1672
+    },
+    {
+      "epoch": 558.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0022,
+      "step": 1676
+    },
+    {
+      "epoch": 559.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 1680
+    },
+    {
+      "epoch": 561.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0016,
+      "step": 1684
+    },
+    {
+      "epoch": 562.59,
+      "learning_rate": 0.0001,
+      "loss": 0.002,
+      "step": 1688
+    },
+    {
+      "epoch": 563.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "step": 1692
+    },
+    {
+      "epoch": 565.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "step": 1696
+    },
+    {
+      "epoch": 566.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "step": 1700
+    },
+    {
+      "epoch": 567.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "step": 1704
+    },
+    {
+      "epoch": 569.29,
+      "learning_rate": 0.0001,
+      "loss": 0.002,
+      "step": 1708
+    },
+    {
+      "epoch": 570.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 1712
+    },
+    {
+      "epoch": 571.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0019,
+      "step": 1716
+    },
+    {
+      "epoch": 573.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0019,
+      "step": 1720
+    },
+    {
+      "epoch": 574.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "step": 1724
+    },
+    {
+      "epoch": 575.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0019,
+      "step": 1728
+    },
+    {
+      "epoch": 575.88,
+      "eval_exact_match": 0.5647969052224371,
+      "eval_exec": 0.5454545454545454,
+      "eval_loss": 0.44897839426994324,
+      "eval_runtime": 797.981,
+      "eval_samples_per_second": 1.296,
+      "step": 1728
+    },
+    {
+      "epoch": 577.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0021,
+      "step": 1732
+    },
+    {
+      "epoch": 578.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0021,
+      "step": 1736
+    },
+    {
+      "epoch": 579.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 1740
+    },
+    {
+      "epoch": 581.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "step": 1744
+    },
+    {
+      "epoch": 582.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 1748
+    },
+    {
+      "epoch": 583.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0016,
+      "step": 1752
+    },
+    {
+      "epoch": 585.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "step": 1756
+    },
+    {
+      "epoch": 586.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0018,
+      "step": 1760
+    },
+    {
+      "epoch": 587.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 1764
+    },
+    {
+      "epoch": 589.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "step": 1768
+    },
+    {
+      "epoch": 590.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "step": 1772
+    },
+    {
+      "epoch": 591.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0019,
+      "step": 1776
+    },
+    {
+      "epoch": 593.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 1780
+    },
+    {
+      "epoch": 594.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0018,
+      "step": 1784
+    },
+    {
+      "epoch": 595.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "step": 1788
+    },
+    {
+      "epoch": 597.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0018,
+      "step": 1792
+    },
+    {
+      "epoch": 597.29,
+      "eval_exact_match": 0.5599613152804642,
+      "eval_exec": 0.5415860735009671,
+      "eval_loss": 0.4542723298072815,
+      "eval_runtime": 763.9026,
+      "eval_samples_per_second": 1.354,
+      "step": 1792
+    },
+    {
+      "epoch": 598.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 1796
+    },
+    {
+      "epoch": 599.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 1800
+    },
+    {
+      "epoch": 601.29,
+      "learning_rate": 0.0001,
+      "loss": 0.002,
+      "step": 1804
+    },
+    {
+      "epoch": 602.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 1808
+    },
+    {
+      "epoch": 603.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 1812
+    },
+    {
+      "epoch": 605.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0014,
+      "step": 1816
+    },
+    {
+      "epoch": 606.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0016,
+      "step": 1820
+    },
+    {
+      "epoch": 607.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 1824
+    },
+    {
+      "epoch": 609.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0023,
+      "step": 1828
+    },
+    {
+      "epoch": 610.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0043,
+      "step": 1832
+    },
+    {
+      "epoch": 611.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 1836
+    },
+    {
+      "epoch": 613.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 1840
+    },
+    {
+      "epoch": 614.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 1844
+    },
+    {
+      "epoch": 615.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 1848
+    },
+    {
+      "epoch": 617.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0019,
+      "step": 1852
+    },
+    {
+      "epoch": 618.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0018,
+      "step": 1856
+    },
+    {
+      "epoch": 618.59,
+      "eval_exact_match": 0.5570599613152805,
+      "eval_exec": 0.5357833655705996,
+      "eval_loss": 0.45874252915382385,
+      "eval_runtime": 776.7688,
+      "eval_samples_per_second": 1.331,
+      "step": 1856
+    },
+    {
+      "epoch": 619.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 1860
+    },
+    {
+      "epoch": 621.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "step": 1864
+    },
+    {
+      "epoch": 622.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 1868
+    },
+    {
+      "epoch": 623.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 1872
+    },
+    {
+      "epoch": 625.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0016,
+      "step": 1876
+    },
+    {
+      "epoch": 626.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 1880
+    },
+    {
+      "epoch": 627.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 1884
+    },
+    {
+      "epoch": 629.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 1888
+    },
+    {
+      "epoch": 630.59,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 1892
+    },
+    {
+      "epoch": 631.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 1896
+    },
+    {
+      "epoch": 633.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 1900
+    },
+    {
+      "epoch": 634.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 1904
+    },
+    {
+      "epoch": 635.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 1908
+    },
+    {
+      "epoch": 637.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0021,
+      "step": 1912
+    },
+    {
+      "epoch": 638.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0014,
+      "step": 1916
+    },
+    {
+      "epoch": 639.88,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 1920
+    },
+    {
+      "epoch": 639.88,
+      "eval_exact_match": 0.5696324951644101,
+      "eval_exec": 0.5444874274661509,
+      "eval_loss": 0.46859657764434814,
+      "eval_runtime": 796.1162,
+      "eval_samples_per_second": 1.299,
+      "step": 1920
+    },
+    {
+      "epoch": 641.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 1924
+    },
+    {
+      "epoch": 642.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 1928
+    },
+    {
+      "epoch": 643.88,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 1932
+    },
+    {
+      "epoch": 645.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 1936
+    },
+    {
+      "epoch": 646.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 1940
+    },
+    {
+      "epoch": 647.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 1944
+    },
+    {
+      "epoch": 649.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0014,
+      "step": 1948
+    },
+    {
+      "epoch": 650.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0014,
+      "step": 1952
+    },
+    {
+      "epoch": 651.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 1956
+    },
+    {
+      "epoch": 653.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 1960
+    },
+    {
+      "epoch": 654.59,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 1964
+    },
+    {
+      "epoch": 655.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 1968
+    },
+    {
+      "epoch": 657.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0105,
+      "step": 1972
+    },
+    {
+      "epoch": 658.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 1976
+    },
+    {
+      "epoch": 659.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 1980
+    },
+    {
+      "epoch": 661.29,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 1984
+    },
+    {
+      "epoch": 661.29,
+      "eval_exact_match": 0.5793036750483559,
+      "eval_exec": 0.5531914893617021,
+      "eval_loss": 0.46243324875831604,
+      "eval_runtime": 799.6383,
+      "eval_samples_per_second": 1.293,
+      "step": 1984
+    },
+    {
+      "epoch": 662.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 1988
+    },
+    {
+      "epoch": 663.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 1992
+    },
+    {
+      "epoch": 665.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0014,
+      "step": 1996
+    },
+    {
+      "epoch": 666.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2000
+    },
+    {
+      "epoch": 667.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 2004
+    },
+    {
+      "epoch": 669.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 2008
+    },
+    {
+      "epoch": 670.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 2012
+    },
+    {
+      "epoch": 671.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2016
+    },
+    {
+      "epoch": 673.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2020
+    },
+    {
+      "epoch": 674.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 2024
+    },
+    {
+      "epoch": 675.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 2028
+    },
+    {
+      "epoch": 677.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 2032
+    },
+    {
+      "epoch": 678.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 2036
+    },
+    {
+      "epoch": 679.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 2040
+    },
+    {
+      "epoch": 681.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 2044
+    },
+    {
+      "epoch": 682.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2048
+    },
+    {
+      "epoch": 682.59,
+      "eval_exact_match": 0.5783365570599613,
+      "eval_exec": 0.5531914893617021,
+      "eval_loss": 0.47705477476119995,
+      "eval_runtime": 775.0862,
+      "eval_samples_per_second": 1.334,
+      "step": 2048
+    },
+    {
+      "epoch": 683.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2052
+    },
+    {
+      "epoch": 685.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 2056
+    },
+    {
+      "epoch": 686.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 2060
+    },
+    {
+      "epoch": 687.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2064
+    },
+    {
+      "epoch": 689.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2068
+    },
+    {
+      "epoch": 690.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2072
+    },
+    {
+      "epoch": 691.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2076
+    },
+    {
+      "epoch": 693.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 2080
+    },
+    {
+      "epoch": 694.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2084
+    },
+    {
+      "epoch": 695.88,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2088
+    },
+    {
+      "epoch": 697.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 2092
+    },
+    {
+      "epoch": 698.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2096
+    },
+    {
+      "epoch": 699.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2100
+    },
+    {
+      "epoch": 701.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0014,
+      "step": 2104
+    },
+    {
+      "epoch": 702.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2108
+    },
+    {
+      "epoch": 703.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "step": 2112
+    },
+    {
+      "epoch": 703.88,
+      "eval_exact_match": 0.5822050290135397,
+      "eval_exec": 0.5599613152804642,
+      "eval_loss": 0.4848947525024414,
+      "eval_runtime": 810.582,
+      "eval_samples_per_second": 1.276,
+      "step": 2112
+    },
+    {
+      "epoch": 705.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 2116
+    },
+    {
+      "epoch": 706.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2120
+    },
+    {
+      "epoch": 707.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2124
+    },
+    {
+      "epoch": 709.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2128
+    },
+    {
+      "epoch": 710.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2132
+    },
+    {
+      "epoch": 711.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2136
+    },
+    {
+      "epoch": 713.29,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2140
+    },
+    {
+      "epoch": 714.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2144
+    },
+    {
+      "epoch": 715.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2148
+    },
+    {
+      "epoch": 717.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2152
+    },
+    {
+      "epoch": 718.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2156
+    },
+    {
+      "epoch": 719.88,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2160
+    },
+    {
+      "epoch": 721.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2164
+    },
+    {
+      "epoch": 722.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2168
+    },
+    {
+      "epoch": 723.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2172
+    },
+    {
+      "epoch": 725.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 2176
+    },
+    {
+      "epoch": 725.29,
+      "eval_exact_match": 0.5696324951644101,
+      "eval_exec": 0.5464216634429401,
+      "eval_loss": 0.48570743203163147,
+      "eval_runtime": 780.2548,
+      "eval_samples_per_second": 1.325,
+      "step": 2176
+    },
+    {
+      "epoch": 726.59,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2180
+    },
+    {
+      "epoch": 727.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2184
+    },
+    {
+      "epoch": 729.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0023,
+      "step": 2188
+    },
+    {
+      "epoch": 730.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0018,
+      "step": 2192
+    },
+    {
+      "epoch": 731.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2196
+    },
+    {
+      "epoch": 733.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0017,
+      "step": 2200
+    },
+    {
+      "epoch": 734.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 2204
+    },
+    {
+      "epoch": 735.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2208
+    },
+    {
+      "epoch": 737.29,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2212
+    },
+    {
+      "epoch": 738.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 2216
+    },
+    {
+      "epoch": 739.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2220
+    },
+    {
+      "epoch": 741.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2224
+    },
+    {
+      "epoch": 742.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2228
+    },
+    {
+      "epoch": 743.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2232
+    },
+    {
+      "epoch": 745.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2236
+    },
+    {
+      "epoch": 746.59,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2240
+    },
+    {
+      "epoch": 746.59,
+      "eval_exact_match": 0.5754352030947776,
+      "eval_exec": 0.5551257253384912,
+      "eval_loss": 0.48890140652656555,
+      "eval_runtime": 792.9162,
+      "eval_samples_per_second": 1.304,
+      "step": 2240
+    },
+    {
+      "epoch": 747.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2244
+    },
+    {
+      "epoch": 749.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2248
+    },
+    {
+      "epoch": 750.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2252
+    },
+    {
+      "epoch": 751.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2256
+    },
+    {
+      "epoch": 753.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2260
+    },
+    {
+      "epoch": 754.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2264
+    },
+    {
+      "epoch": 755.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2268
+    },
+    {
+      "epoch": 757.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2272
+    },
+    {
+      "epoch": 758.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2276
+    },
+    {
+      "epoch": 759.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2280
+    },
+    {
+      "epoch": 761.29,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2284
+    },
+    {
+      "epoch": 762.59,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2288
+    },
+    {
+      "epoch": 763.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2292
+    },
+    {
+      "epoch": 765.29,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2296
+    },
+    {
+      "epoch": 766.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2300
+    },
+    {
+      "epoch": 767.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2304
+    },
+    {
+      "epoch": 767.88,
+      "eval_exact_match": 0.5715667311411993,
+      "eval_exec": 0.5444874274661509,
+      "eval_loss": 0.5018870830535889,
+      "eval_runtime": 784.4908,
+      "eval_samples_per_second": 1.318,
+      "step": 2304
+    },
+    {
+      "epoch": 769.29,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2308
+    },
+    {
+      "epoch": 770.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 2312
+    },
+    {
+      "epoch": 771.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2316
+    },
+    {
+      "epoch": 773.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2320
+    },
+    {
+      "epoch": 774.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2324
+    },
+    {
+      "epoch": 775.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2328
+    },
+    {
+      "epoch": 777.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2332
+    },
+    {
+      "epoch": 778.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2336
+    },
+    {
+      "epoch": 779.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2340
+    },
+    {
+      "epoch": 781.29,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2344
+    },
+    {
+      "epoch": 782.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2348
+    },
+    {
+      "epoch": 783.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2352
+    },
+    {
+      "epoch": 785.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2356
+    },
+    {
+      "epoch": 786.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2360
+    },
+    {
+      "epoch": 787.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2364
+    },
+    {
+      "epoch": 789.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2368
+    },
+    {
+      "epoch": 789.29,
+      "eval_exact_match": 0.5609284332688588,
+      "eval_exec": 0.5444874274661509,
+      "eval_loss": 0.5034850835800171,
+      "eval_runtime": 769.6664,
+      "eval_samples_per_second": 1.343,
+      "step": 2368
+    },
+    {
+      "epoch": 790.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2372
+    },
+    {
+      "epoch": 791.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2376
+    },
+    {
+      "epoch": 793.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2380
+    },
+    {
+      "epoch": 794.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2384
+    },
+    {
+      "epoch": 795.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2388
+    },
+    {
+      "epoch": 797.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2392
+    },
+    {
+      "epoch": 798.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2396
+    },
+    {
+      "epoch": 799.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2400
+    },
+    {
+      "epoch": 801.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2404
+    },
+    {
+      "epoch": 802.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2408
+    },
+    {
+      "epoch": 803.88,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2412
+    },
+    {
+      "epoch": 805.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2416
+    },
+    {
+      "epoch": 806.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2420
+    },
+    {
+      "epoch": 807.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2424
+    },
+    {
+      "epoch": 809.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2428
+    },
+    {
+      "epoch": 810.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2432
+    },
+    {
+      "epoch": 810.59,
+      "eval_exact_match": 0.5725338491295938,
+      "eval_exec": 0.5531914893617021,
+      "eval_loss": 0.5050273537635803,
+      "eval_runtime": 780.7218,
+      "eval_samples_per_second": 1.324,
+      "step": 2432
+    },
+    {
+      "epoch": 811.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2436
+    },
+    {
+      "epoch": 813.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2440
+    },
+    {
+      "epoch": 814.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2444
+    },
+    {
+      "epoch": 815.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2448
+    },
+    {
+      "epoch": 817.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2452
+    },
+    {
+      "epoch": 818.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2456
+    },
+    {
+      "epoch": 819.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2460
+    },
+    {
+      "epoch": 821.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2464
+    },
+    {
+      "epoch": 822.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2468
+    },
+    {
+      "epoch": 823.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2472
+    },
+    {
+      "epoch": 825.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2476
+    },
+    {
+      "epoch": 826.59,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2480
+    },
+    {
+      "epoch": 827.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2484
+    },
+    {
+      "epoch": 829.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0038,
+      "step": 2488
+    },
+    {
+      "epoch": 830.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2492
+    },
+    {
+      "epoch": 831.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2496
+    },
+    {
+      "epoch": 831.88,
+      "eval_exact_match": 0.5754352030947776,
+      "eval_exec": 0.5522243713733076,
+      "eval_loss": 0.49086302518844604,
+      "eval_runtime": 785.9929,
+      "eval_samples_per_second": 1.316,
+      "step": 2496
+    },
+    {
+      "epoch": 833.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2500
+    },
+    {
+      "epoch": 834.59,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2504
+    },
+    {
+      "epoch": 835.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2508
+    },
+    {
+      "epoch": 837.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2512
+    },
+    {
+      "epoch": 838.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2516
+    },
+    {
+      "epoch": 839.88,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2520
+    },
+    {
+      "epoch": 841.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 2524
+    },
+    {
+      "epoch": 842.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2528
+    },
+    {
+      "epoch": 843.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2532
+    },
+    {
+      "epoch": 845.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2536
+    },
+    {
+      "epoch": 846.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2540
+    },
+    {
+      "epoch": 847.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2544
+    },
+    {
+      "epoch": 849.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2548
+    },
+    {
+      "epoch": 850.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2552
+    },
+    {
+      "epoch": 851.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2556
+    },
+    {
+      "epoch": 853.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0012,
+      "step": 2560
+    },
+    {
+      "epoch": 853.29,
+      "eval_exact_match": 0.5705996131528046,
+      "eval_exec": 0.5560928433268859,
+      "eval_loss": 0.5105037689208984,
+      "eval_runtime": 771.5445,
+      "eval_samples_per_second": 1.34,
+      "step": 2560
+    },
+    {
+      "epoch": 854.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2564
+    },
+    {
+      "epoch": 855.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2568
+    },
+    {
+      "epoch": 857.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0013,
+      "step": 2572
+    },
+    {
+      "epoch": 858.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2576
+    },
+    {
+      "epoch": 859.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2580
+    },
+    {
+      "epoch": 861.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2584
+    },
+    {
+      "epoch": 862.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2588
+    },
+    {
+      "epoch": 863.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2592
+    },
+    {
+      "epoch": 865.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2596
+    },
+    {
+      "epoch": 866.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2600
+    },
+    {
+      "epoch": 867.88,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2604
+    },
+    {
+      "epoch": 869.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2608
+    },
+    {
+      "epoch": 870.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2612
+    },
+    {
+      "epoch": 871.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2616
+    },
+    {
+      "epoch": 873.29,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2620
+    },
+    {
+      "epoch": 874.59,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2624
+    },
+    {
+      "epoch": 874.59,
+      "eval_exact_match": 0.5735009671179884,
+      "eval_exec": 0.5531914893617021,
+      "eval_loss": 0.5072650909423828,
+      "eval_runtime": 804.9263,
+      "eval_samples_per_second": 1.285,
+      "step": 2624
+    },
+    {
+      "epoch": 875.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2628
+    },
+    {
+      "epoch": 877.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2632
+    },
+    {
+      "epoch": 878.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2636
+    },
+    {
+      "epoch": 879.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2640
+    },
+    {
+      "epoch": 881.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2644
+    },
+    {
+      "epoch": 882.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2648
+    },
+    {
+      "epoch": 883.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2652
+    },
+    {
+      "epoch": 885.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2656
+    },
+    {
+      "epoch": 886.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2660
+    },
+    {
+      "epoch": 887.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2664
+    },
+    {
+      "epoch": 889.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2668
+    },
+    {
+      "epoch": 890.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 2672
+    },
+    {
+      "epoch": 891.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2676
+    },
+    {
+      "epoch": 893.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2680
+    },
+    {
+      "epoch": 894.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2684
+    },
+    {
+      "epoch": 895.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2688
+    },
+    {
+      "epoch": 895.88,
+      "eval_exact_match": 0.5841392649903289,
+      "eval_exec": 0.5599613152804642,
+      "eval_loss": 0.498661607503891,
+      "eval_runtime": 823.9748,
+      "eval_samples_per_second": 1.255,
+      "step": 2688
+    },
+    {
+      "epoch": 897.29,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2692
+    },
+    {
+      "epoch": 898.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2696
+    },
+    {
+      "epoch": 899.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2700
+    },
+    {
+      "epoch": 901.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2704
+    },
+    {
+      "epoch": 902.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2708
+    },
+    {
+      "epoch": 903.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2712
+    },
+    {
+      "epoch": 905.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2716
+    },
+    {
+      "epoch": 906.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2720
+    },
+    {
+      "epoch": 907.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2724
+    },
+    {
+      "epoch": 909.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2728
+    },
+    {
+      "epoch": 910.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2732
+    },
+    {
+      "epoch": 911.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2736
+    },
+    {
+      "epoch": 913.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2740
+    },
+    {
+      "epoch": 914.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2744
+    },
+    {
+      "epoch": 915.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2748
+    },
+    {
+      "epoch": 917.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2752
+    },
+    {
+      "epoch": 917.29,
+      "eval_exact_match": 0.5822050290135397,
+      "eval_exec": 0.5589941972920697,
+      "eval_loss": 0.49509045481681824,
+      "eval_runtime": 1542.8924,
+      "eval_samples_per_second": 0.67,
+      "step": 2752
+    },
+    {
+      "epoch": 918.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2756
+    },
+    {
+      "epoch": 919.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2760
+    },
+    {
+      "epoch": 921.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2764
+    },
+    {
+      "epoch": 922.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 2768
+    },
+    {
+      "epoch": 923.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2772
+    },
+    {
+      "epoch": 925.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2776
+    },
+    {
+      "epoch": 926.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2780
+    },
+    {
+      "epoch": 927.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2784
+    },
+    {
+      "epoch": 929.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2788
+    },
+    {
+      "epoch": 930.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2792
+    },
+    {
+      "epoch": 931.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0016,
+      "step": 2796
+    },
+    {
+      "epoch": 933.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 2800
+    },
+    {
+      "epoch": 934.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2804
+    },
+    {
+      "epoch": 935.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2808
+    },
+    {
+      "epoch": 937.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2812
+    },
+    {
+      "epoch": 938.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 2816
+    },
+    {
+      "epoch": 938.59,
+      "eval_exact_match": 0.574468085106383,
+      "eval_exec": 0.5560928433268859,
+      "eval_loss": 0.5093162059783936,
+      "eval_runtime": 1522.1173,
+      "eval_samples_per_second": 0.679,
+      "step": 2816
+    },
+    {
+      "epoch": 939.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2820
+    },
+    {
+      "epoch": 941.29,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 2824
+    },
+    {
+      "epoch": 942.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2828
+    },
+    {
+      "epoch": 943.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2832
+    },
+    {
+      "epoch": 945.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2836
+    },
+    {
+      "epoch": 946.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2840
+    },
+    {
+      "epoch": 947.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2844
+    },
+    {
+      "epoch": 949.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2848
+    },
+    {
+      "epoch": 950.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2852
+    },
+    {
+      "epoch": 951.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2856
+    },
+    {
+      "epoch": 953.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2860
+    },
+    {
+      "epoch": 954.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2864
+    },
+    {
+      "epoch": 955.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2868
+    },
+    {
+      "epoch": 957.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2872
+    },
+    {
+      "epoch": 958.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2876
+    },
+    {
+      "epoch": 959.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2880
+    },
+    {
+      "epoch": 959.88,
+      "eval_exact_match": 0.5764023210831721,
+      "eval_exec": 0.5609284332688588,
+      "eval_loss": 0.52325040102005,
+      "eval_runtime": 1114.7015,
+      "eval_samples_per_second": 0.928,
+      "step": 2880
+    },
+    {
+      "epoch": 961.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 2884
+    },
+    {
+      "epoch": 962.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2888
+    },
+    {
+      "epoch": 963.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 2892
+    },
+    {
+      "epoch": 965.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2896
+    },
+    {
+      "epoch": 966.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2900
+    },
+    {
+      "epoch": 967.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2904
+    },
+    {
+      "epoch": 969.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2908
+    },
+    {
+      "epoch": 970.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2912
+    },
+    {
+      "epoch": 971.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2916
+    },
+    {
+      "epoch": 973.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2920
+    },
+    {
+      "epoch": 974.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2924
+    },
+    {
+      "epoch": 975.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2928
+    },
+    {
+      "epoch": 977.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2932
+    },
+    {
+      "epoch": 978.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2936
+    },
+    {
+      "epoch": 979.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2940
+    },
+    {
+      "epoch": 981.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2944
+    },
+    {
+      "epoch": 981.29,
+      "eval_exact_match": 0.5802707930367504,
+      "eval_exec": 0.5609284332688588,
+      "eval_loss": 0.5160704255104065,
+      "eval_runtime": 1079.9947,
+      "eval_samples_per_second": 0.957,
+      "step": 2944
+    },
+    {
+      "epoch": 982.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2948
+    },
+    {
+      "epoch": 983.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2952
+    },
+    {
+      "epoch": 985.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2956
+    },
+    {
+      "epoch": 986.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2960
+    },
+    {
+      "epoch": 987.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 2964
+    },
+    {
+      "epoch": 989.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2968
+    },
+    {
+      "epoch": 990.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 2972
+    },
+    {
+      "epoch": 991.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2976
+    },
+    {
+      "epoch": 993.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2980
+    },
+    {
+      "epoch": 994.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2984
+    },
+    {
+      "epoch": 995.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2988
+    },
+    {
+      "epoch": 997.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 2992
+    },
+    {
+      "epoch": 998.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 2996
+    },
+    {
+      "epoch": 999.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3000
+    },
+    {
+      "epoch": 1001.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3004
+    },
+    {
+      "epoch": 1002.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3008
+    },
+    {
+      "epoch": 1002.59,
+      "eval_exact_match": 0.574468085106383,
+      "eval_exec": 0.558027079303675,
+      "eval_loss": 0.5089902877807617,
+      "eval_runtime": 1087.4284,
+      "eval_samples_per_second": 0.951,
+      "step": 3008
+    },
+    {
+      "epoch": 1003.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 3012
+    },
+    {
+      "epoch": 1005.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3016
+    },
+    {
+      "epoch": 1006.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 3020
+    },
+    {
+      "epoch": 1007.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3024
+    },
+    {
+      "epoch": 1009.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3028
+    },
+    {
+      "epoch": 1010.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3032
+    },
+    {
+      "epoch": 1011.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3036
+    },
+    {
+      "epoch": 1013.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3040
+    },
+    {
+      "epoch": 1014.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3044
+    },
+    {
+      "epoch": 1015.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3048
+    },
+    {
+      "epoch": 1017.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3052
+    },
+    {
+      "epoch": 1018.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3056
+    },
+    {
+      "epoch": 1019.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3060
+    },
+    {
+      "epoch": 1021.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3064
+    },
+    {
+      "epoch": 1022.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3068
+    },
+    {
+      "epoch": 1023.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3072
+    },
+    {
+      "epoch": 1023.88,
+      "eval_exact_match": 0.5696324951644101,
+      "eval_exec": 0.5483558994197292,
+      "eval_loss": 0.5173510313034058,
+      "eval_runtime": 1085.3183,
+      "eval_samples_per_second": 0.953,
+      "step": 3072
+    },
+    {
+      "epoch": 1025.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3076
+    },
+    {
+      "epoch": 1026.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3080
+    },
+    {
+      "epoch": 1027.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3084
+    },
+    {
+      "epoch": 1029.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 3088
+    },
+    {
+      "epoch": 1030.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3092
+    },
+    {
+      "epoch": 1031.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3096
+    },
+    {
+      "epoch": 1033.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3100
+    },
+    {
+      "epoch": 1034.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3104
+    },
+    {
+      "epoch": 1035.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3108
+    },
+    {
+      "epoch": 1037.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3112
+    },
+    {
+      "epoch": 1038.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3116
+    },
+    {
+      "epoch": 1039.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3120
+    },
+    {
+      "epoch": 1041.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3124
+    },
+    {
+      "epoch": 1042.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3128
+    },
+    {
+      "epoch": 1043.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3132
+    },
+    {
+      "epoch": 1045.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3136
+    },
+    {
+      "epoch": 1045.29,
+      "eval_exact_match": 0.5793036750483559,
+      "eval_exec": 0.5589941972920697,
+      "eval_loss": 0.5162827968597412,
+      "eval_runtime": 1238.641,
+      "eval_samples_per_second": 0.835,
+      "step": 3136
+    },
+    {
+      "epoch": 1046.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3140
+    },
+    {
+      "epoch": 1047.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3144
+    },
+    {
+      "epoch": 1049.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3148
+    },
+    {
+      "epoch": 1050.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3152
+    },
+    {
+      "epoch": 1051.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3156
+    },
+    {
+      "epoch": 1053.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3160
+    },
+    {
+      "epoch": 1054.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3164
+    },
+    {
+      "epoch": 1055.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3168
+    },
+    {
+      "epoch": 1057.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 3172
+    },
+    {
+      "epoch": 1058.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3176
+    },
+    {
+      "epoch": 1059.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3180
+    },
+    {
+      "epoch": 1061.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 3184
+    },
+    {
+      "epoch": 1062.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3188
+    },
+    {
+      "epoch": 1063.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3192
+    },
+    {
+      "epoch": 1065.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3196
+    },
+    {
+      "epoch": 1066.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3200
+    },
+    {
+      "epoch": 1066.59,
+      "eval_exact_match": 0.5773694390715667,
+      "eval_exec": 0.5570599613152805,
+      "eval_loss": 0.535306990146637,
+      "eval_runtime": 1229.1064,
+      "eval_samples_per_second": 0.841,
+      "step": 3200
+    },
+    {
+      "epoch": 1067.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3204
+    },
+    {
+      "epoch": 1069.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3208
+    },
+    {
+      "epoch": 1070.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3212
+    },
+    {
+      "epoch": 1071.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3216
+    },
+    {
+      "epoch": 1073.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3220
+    },
+    {
+      "epoch": 1074.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3224
+    },
+    {
+      "epoch": 1075.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0009,
+      "step": 3228
+    },
+    {
+      "epoch": 1077.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3232
+    },
+    {
+      "epoch": 1078.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3236
+    },
+    {
+      "epoch": 1079.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3240
+    },
+    {
+      "epoch": 1081.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3244
+    },
+    {
+      "epoch": 1082.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3248
+    },
+    {
+      "epoch": 1083.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3252
+    },
+    {
+      "epoch": 1085.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3256
+    },
+    {
+      "epoch": 1086.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3260
+    },
+    {
+      "epoch": 1087.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3264
+    },
+    {
+      "epoch": 1087.88,
+      "eval_exact_match": 0.5754352030947776,
+      "eval_exec": 0.5531914893617021,
+      "eval_loss": 0.524728536605835,
+      "eval_runtime": 1228.713,
+      "eval_samples_per_second": 0.842,
+      "step": 3264
+    },
+    {
+      "epoch": 1089.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3268
+    },
+    {
+      "epoch": 1090.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3272
+    },
+    {
+      "epoch": 1091.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3276
+    },
+    {
+      "epoch": 1093.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3280
+    },
+    {
+      "epoch": 1094.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 3284
+    },
+    {
+      "epoch": 1095.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3288
+    },
+    {
+      "epoch": 1097.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3292
+    },
+    {
+      "epoch": 1098.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3296
+    },
+    {
+      "epoch": 1099.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3300
+    },
+    {
+      "epoch": 1101.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3304
+    },
+    {
+      "epoch": 1102.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3308
+    },
+    {
+      "epoch": 1103.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3312
+    },
+    {
+      "epoch": 1105.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3316
+    },
+    {
+      "epoch": 1106.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 3320
+    },
+    {
+      "epoch": 1107.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3324
+    },
+    {
+      "epoch": 1109.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3328
+    },
+    {
+      "epoch": 1109.29,
+      "eval_exact_match": 0.5705996131528046,
+      "eval_exec": 0.5541586073500967,
+      "eval_loss": 0.5230495929718018,
+      "eval_runtime": 1220.3723,
+      "eval_samples_per_second": 0.847,
+      "step": 3328
+    },
+    {
+      "epoch": 1110.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3332
+    },
+    {
+      "epoch": 1111.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3336
+    },
+    {
+      "epoch": 1113.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3340
+    },
+    {
+      "epoch": 1114.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3344
+    },
+    {
+      "epoch": 1115.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3348
+    },
+    {
+      "epoch": 1117.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3352
+    },
+    {
+      "epoch": 1118.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3356
+    },
+    {
+      "epoch": 1119.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3360
+    },
+    {
+      "epoch": 1121.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3364
+    },
+    {
+      "epoch": 1122.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3368
+    },
+    {
+      "epoch": 1123.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3372
+    },
+    {
+      "epoch": 1125.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3376
+    },
+    {
+      "epoch": 1126.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3380
+    },
+    {
+      "epoch": 1127.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3384
+    },
+    {
+      "epoch": 1129.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3388
+    },
+    {
+      "epoch": 1130.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3392
+    },
+    {
+      "epoch": 1130.59,
+      "eval_exact_match": 0.5812379110251451,
+      "eval_exec": 0.5551257253384912,
+      "eval_loss": 0.5162366628646851,
+      "eval_runtime": 1203.0611,
+      "eval_samples_per_second": 0.859,
+      "step": 3392
+    },
+    {
+      "epoch": 1131.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3396
+    },
+    {
+      "epoch": 1133.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3400
+    },
+    {
+      "epoch": 1134.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3404
+    },
+    {
+      "epoch": 1135.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3408
+    },
+    {
+      "epoch": 1137.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3412
+    },
+    {
+      "epoch": 1138.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3416
+    },
+    {
+      "epoch": 1139.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3420
+    },
+    {
+      "epoch": 1141.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3424
+    },
+    {
+      "epoch": 1142.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3428
+    },
+    {
+      "epoch": 1143.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3432
+    },
+    {
+      "epoch": 1145.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3436
+    },
+    {
+      "epoch": 1146.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3440
+    },
+    {
+      "epoch": 1147.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3444
+    },
+    {
+      "epoch": 1149.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3448
+    },
+    {
+      "epoch": 1150.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3452
+    },
+    {
+      "epoch": 1151.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 3456
+    },
+    {
+      "epoch": 1151.88,
+      "eval_exact_match": 0.5812379110251451,
+      "eval_exec": 0.5560928433268859,
+      "eval_loss": 0.5390752553939819,
+      "eval_runtime": 790.5575,
+      "eval_samples_per_second": 1.308,
+      "step": 3456
+    },
+    {
+      "epoch": 1153.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3460
+    },
+    {
+      "epoch": 1154.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3464
+    },
+    {
+      "epoch": 1155.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 3468
+    },
+    {
+      "epoch": 1157.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3472
+    },
+    {
+      "epoch": 1158.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3476
+    },
+    {
+      "epoch": 1159.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3480
+    },
+    {
+      "epoch": 1161.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3484
+    },
+    {
+      "epoch": 1162.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3488
+    },
+    {
+      "epoch": 1163.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3492
+    },
+    {
+      "epoch": 1165.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3496
+    },
+    {
+      "epoch": 1166.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3500
+    },
+    {
+      "epoch": 1167.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3504
+    },
+    {
+      "epoch": 1169.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3508
+    },
+    {
+      "epoch": 1170.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3512
+    },
+    {
+      "epoch": 1171.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3516
+    },
+    {
+      "epoch": 1173.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3520
+    },
+    {
+      "epoch": 1173.29,
+      "eval_exact_match": 0.574468085106383,
+      "eval_exec": 0.5531914893617021,
+      "eval_loss": 0.528677761554718,
+      "eval_runtime": 782.4267,
+      "eval_samples_per_second": 1.322,
+      "step": 3520
+    },
+    {
+      "epoch": 1174.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3524
+    },
+    {
+      "epoch": 1175.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3528
+    },
+    {
+      "epoch": 1177.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3532
+    },
+    {
+      "epoch": 1178.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3536
+    },
+    {
+      "epoch": 1179.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3540
+    },
+    {
+      "epoch": 1181.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3544
+    },
+    {
+      "epoch": 1182.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3548
+    },
+    {
+      "epoch": 1183.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3552
+    },
+    {
+      "epoch": 1185.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3556
+    },
+    {
+      "epoch": 1186.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3560
+    },
+    {
+      "epoch": 1187.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3564
+    },
+    {
+      "epoch": 1189.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3568
+    },
+    {
+      "epoch": 1190.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3572
+    },
+    {
+      "epoch": 1191.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3576
+    },
+    {
+      "epoch": 1193.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3580
+    },
+    {
+      "epoch": 1194.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3584
+    },
+    {
+      "epoch": 1194.59,
+      "eval_exact_match": 0.5686653771760155,
+      "eval_exec": 0.5493230174081238,
+      "eval_loss": 0.5310149788856506,
+      "eval_runtime": 794.635,
+      "eval_samples_per_second": 1.301,
+      "step": 3584
+    },
+    {
+      "epoch": 1195.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3588
+    },
+    {
+      "epoch": 1197.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3592
+    },
+    {
+      "epoch": 1198.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3596
+    },
+    {
+      "epoch": 1199.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3600
+    },
+    {
+      "epoch": 1201.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3604
+    },
+    {
+      "epoch": 1202.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 3608
+    },
+    {
+      "epoch": 1203.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 3612
+    },
+    {
+      "epoch": 1205.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3616
+    },
+    {
+      "epoch": 1206.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3620
+    },
+    {
+      "epoch": 1207.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3624
+    },
+    {
+      "epoch": 1209.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3628
+    },
+    {
+      "epoch": 1210.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3632
+    },
+    {
+      "epoch": 1211.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3636
+    },
+    {
+      "epoch": 1213.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3640
+    },
+    {
+      "epoch": 1214.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3644
+    },
+    {
+      "epoch": 1215.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3648
+    },
+    {
+      "epoch": 1215.88,
+      "eval_exact_match": 0.5773694390715667,
+      "eval_exec": 0.5589941972920697,
+      "eval_loss": 0.541464626789093,
+      "eval_runtime": 821.1144,
+      "eval_samples_per_second": 1.259,
+      "step": 3648
+    },
+    {
+      "epoch": 1217.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3652
+    },
+    {
+      "epoch": 1218.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3656
+    },
+    {
+      "epoch": 1219.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3660
+    },
+    {
+      "epoch": 1221.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3664
+    },
+    {
+      "epoch": 1222.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3668
+    },
+    {
+      "epoch": 1223.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3672
+    },
+    {
+      "epoch": 1225.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3676
+    },
+    {
+      "epoch": 1226.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3680
+    },
+    {
+      "epoch": 1227.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3684
+    },
+    {
+      "epoch": 1229.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3688
+    },
+    {
+      "epoch": 1230.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3692
+    },
+    {
+      "epoch": 1231.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3696
+    },
+    {
+      "epoch": 1233.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3700
+    },
+    {
+      "epoch": 1234.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3704
+    },
+    {
+      "epoch": 1235.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3708
+    },
+    {
+      "epoch": 1237.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3712
+    },
+    {
+      "epoch": 1237.29,
+      "eval_exact_match": 0.5793036750483559,
+      "eval_exec": 0.5531914893617021,
+      "eval_loss": 0.5400662422180176,
+      "eval_runtime": 781.6381,
+      "eval_samples_per_second": 1.323,
+      "step": 3712
+    },
+    {
+      "epoch": 1238.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3716
+    },
+    {
+      "epoch": 1239.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3720
+    },
+    {
+      "epoch": 1241.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3724
+    },
+    {
+      "epoch": 1242.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3728
+    },
+    {
+      "epoch": 1243.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3732
+    },
+    {
+      "epoch": 1245.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3736
+    },
+    {
+      "epoch": 1246.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3740
+    },
+    {
+      "epoch": 1247.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3744
+    },
+    {
+      "epoch": 1249.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3748
+    },
+    {
+      "epoch": 1250.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 3752
+    },
+    {
+      "epoch": 1251.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3756
+    },
+    {
+      "epoch": 1253.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3760
+    },
+    {
+      "epoch": 1254.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3764
+    },
+    {
+      "epoch": 1255.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3768
+    },
+    {
+      "epoch": 1257.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 3772
+    },
+    {
+      "epoch": 1258.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3776
+    },
+    {
+      "epoch": 1258.59,
+      "eval_exact_match": 0.5851063829787234,
+      "eval_exec": 0.5638297872340425,
+      "eval_loss": 0.5534989237785339,
+      "eval_runtime": 817.9796,
+      "eval_samples_per_second": 1.264,
+      "step": 3776
+    },
+    {
+      "epoch": 1259.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3780
+    },
+    {
+      "epoch": 1261.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3784
+    },
+    {
+      "epoch": 1262.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3788
+    },
+    {
+      "epoch": 1263.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 3792
+    },
+    {
+      "epoch": 1265.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3796
+    },
+    {
+      "epoch": 1266.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3800
+    },
+    {
+      "epoch": 1267.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3804
+    },
+    {
+      "epoch": 1269.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3808
+    },
+    {
+      "epoch": 1270.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3812
+    },
+    {
+      "epoch": 1271.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3816
+    },
+    {
+      "epoch": 1273.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3820
+    },
+    {
+      "epoch": 1274.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3824
+    },
+    {
+      "epoch": 1275.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 3828
+    },
+    {
+      "epoch": 1277.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3832
+    },
+    {
+      "epoch": 1278.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 3836
+    },
+    {
+      "epoch": 1279.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3840
+    },
+    {
+      "epoch": 1279.88,
+      "eval_exact_match": 0.5725338491295938,
+      "eval_exec": 0.5522243713733076,
+      "eval_loss": 0.5276786684989929,
+      "eval_runtime": 790.6261,
+      "eval_samples_per_second": 1.308,
+      "step": 3840
+    },
+    {
+      "epoch": 1281.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3844
+    },
+    {
+      "epoch": 1282.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 3848
+    },
+    {
+      "epoch": 1283.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3852
+    },
+    {
+      "epoch": 1285.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3856
+    },
+    {
+      "epoch": 1286.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3860
+    },
+    {
+      "epoch": 1287.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3864
+    },
+    {
+      "epoch": 1289.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3868
+    },
+    {
+      "epoch": 1290.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3872
+    },
+    {
+      "epoch": 1291.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3876
+    },
+    {
+      "epoch": 1293.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3880
+    },
+    {
+      "epoch": 1294.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3884
+    },
+    {
+      "epoch": 1295.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3888
+    },
+    {
+      "epoch": 1297.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3892
+    },
+    {
+      "epoch": 1298.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3896
+    },
+    {
+      "epoch": 1299.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3900
+    },
+    {
+      "epoch": 1301.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3904
+    },
+    {
+      "epoch": 1301.29,
+      "eval_exact_match": 0.5705996131528046,
+      "eval_exec": 0.5502901353965184,
+      "eval_loss": 0.5508450269699097,
+      "eval_runtime": 774.5817,
+      "eval_samples_per_second": 1.335,
+      "step": 3904
+    },
+    {
+      "epoch": 1302.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3908
+    },
+    {
+      "epoch": 1303.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3912
+    },
+    {
+      "epoch": 1305.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3916
+    },
+    {
+      "epoch": 1306.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3920
+    },
+    {
+      "epoch": 1307.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3924
+    },
+    {
+      "epoch": 1309.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 3928
+    },
+    {
+      "epoch": 1310.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3932
+    },
+    {
+      "epoch": 1311.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3936
+    },
+    {
+      "epoch": 1313.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3940
+    },
+    {
+      "epoch": 1314.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 3944
+    },
+    {
+      "epoch": 1315.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3948
+    },
+    {
+      "epoch": 1317.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 3952
+    },
+    {
+      "epoch": 1318.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 3956
+    },
+    {
+      "epoch": 1319.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 3960
+    },
+    {
+      "epoch": 1321.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3964
+    },
+    {
+      "epoch": 1322.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3968
+    },
+    {
+      "epoch": 1322.59,
+      "eval_exact_match": 0.5705996131528046,
+      "eval_exec": 0.5473887814313346,
+      "eval_loss": 0.553875744342804,
+      "eval_runtime": 779.9737,
+      "eval_samples_per_second": 1.326,
+      "step": 3968
+    },
+    {
+      "epoch": 1323.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3972
+    },
+    {
+      "epoch": 1325.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3976
+    },
+    {
+      "epoch": 1326.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3980
+    },
+    {
+      "epoch": 1327.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3984
+    },
+    {
+      "epoch": 1329.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 3988
+    },
+    {
+      "epoch": 1330.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3992
+    },
+    {
+      "epoch": 1331.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 3996
+    },
+    {
+      "epoch": 1333.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4000
+    },
+    {
+      "epoch": 1334.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4004
+    },
+    {
+      "epoch": 1335.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4008
+    },
+    {
+      "epoch": 1337.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4012
+    },
+    {
+      "epoch": 1338.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4016
+    },
+    {
+      "epoch": 1339.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4020
+    },
+    {
+      "epoch": 1341.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4024
+    },
+    {
+      "epoch": 1342.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4028
+    },
+    {
+      "epoch": 1343.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4032
+    },
+    {
+      "epoch": 1343.88,
+      "eval_exact_match": 0.5783365570599613,
+      "eval_exec": 0.5570599613152805,
+      "eval_loss": 0.5482780933380127,
+      "eval_runtime": 1357.0193,
+      "eval_samples_per_second": 0.762,
+      "step": 4032
+    },
+    {
+      "epoch": 1345.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4036
+    },
+    {
+      "epoch": 1346.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4040
+    },
+    {
+      "epoch": 1347.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4044
+    },
+    {
+      "epoch": 1349.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4048
+    },
+    {
+      "epoch": 1350.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4052
+    },
+    {
+      "epoch": 1351.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4056
+    },
+    {
+      "epoch": 1353.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4060
+    },
+    {
+      "epoch": 1354.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4064
+    },
+    {
+      "epoch": 1355.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 4068
+    },
+    {
+      "epoch": 1357.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4072
+    },
+    {
+      "epoch": 1358.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0032,
+      "step": 4076
+    },
+    {
+      "epoch": 1359.88,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 4080
+    },
+    {
+      "epoch": 1361.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 4084
+    },
+    {
+      "epoch": 1362.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 4088
+    },
+    {
+      "epoch": 1363.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4092
+    },
+    {
+      "epoch": 1365.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4096
+    },
+    {
+      "epoch": 1365.29,
+      "eval_exact_match": 0.5696324951644101,
+      "eval_exec": 0.5512572533849129,
+      "eval_loss": 0.5236480236053467,
+      "eval_runtime": 1332.7882,
+      "eval_samples_per_second": 0.776,
+      "step": 4096
+    },
+    {
+      "epoch": 1366.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4100
+    },
+    {
+      "epoch": 1367.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4104
+    },
+    {
+      "epoch": 1369.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4108
+    },
+    {
+      "epoch": 1370.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4112
+    },
+    {
+      "epoch": 1371.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4116
+    },
+    {
+      "epoch": 1373.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4120
+    },
+    {
+      "epoch": 1374.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4124
+    },
+    {
+      "epoch": 1375.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4128
+    },
+    {
+      "epoch": 1377.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4132
+    },
+    {
+      "epoch": 1378.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4136
+    },
+    {
+      "epoch": 1379.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4140
+    },
+    {
+      "epoch": 1381.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4144
+    },
+    {
+      "epoch": 1382.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4148
+    },
+    {
+      "epoch": 1383.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4152
+    },
+    {
+      "epoch": 1385.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4156
+    },
+    {
+      "epoch": 1386.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4160
+    },
+    {
+      "epoch": 1386.59,
+      "eval_exact_match": 0.5812379110251451,
+      "eval_exec": 0.5541586073500967,
+      "eval_loss": 0.5603498816490173,
+      "eval_runtime": 1324.2306,
+      "eval_samples_per_second": 0.781,
+      "step": 4160
+    },
+    {
+      "epoch": 1387.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4164
+    },
+    {
+      "epoch": 1389.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4168
+    },
+    {
+      "epoch": 1390.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4172
+    },
+    {
+      "epoch": 1391.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4176
+    },
+    {
+      "epoch": 1393.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4180
+    },
+    {
+      "epoch": 1394.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4184
+    },
+    {
+      "epoch": 1395.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4188
+    },
+    {
+      "epoch": 1397.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4192
+    },
+    {
+      "epoch": 1398.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4196
+    },
+    {
+      "epoch": 1399.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4200
+    },
+    {
+      "epoch": 1401.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4204
+    },
+    {
+      "epoch": 1402.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4208
+    },
+    {
+      "epoch": 1403.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4212
+    },
+    {
+      "epoch": 1405.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4216
+    },
+    {
+      "epoch": 1406.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4220
+    },
+    {
+      "epoch": 1407.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4224
+    },
+    {
+      "epoch": 1407.88,
+      "eval_exact_match": 0.574468085106383,
+      "eval_exec": 0.5589941972920697,
+      "eval_loss": 0.5609294772148132,
+      "eval_runtime": 799.2223,
+      "eval_samples_per_second": 1.294,
+      "step": 4224
+    },
+    {
+      "epoch": 1409.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4228
+    },
+    {
+      "epoch": 1410.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4232
+    },
+    {
+      "epoch": 1411.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4236
+    },
+    {
+      "epoch": 1413.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4240
+    },
+    {
+      "epoch": 1414.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4244
+    },
+    {
+      "epoch": 1415.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4248
+    },
+    {
+      "epoch": 1417.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4252
+    },
+    {
+      "epoch": 1418.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4256
+    },
+    {
+      "epoch": 1419.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4260
+    },
+    {
+      "epoch": 1421.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4264
+    },
+    {
+      "epoch": 1422.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4268
+    },
+    {
+      "epoch": 1423.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4272
+    },
+    {
+      "epoch": 1425.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4276
+    },
+    {
+      "epoch": 1426.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4280
+    },
+    {
+      "epoch": 1427.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4284
+    },
+    {
+      "epoch": 1429.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4288
+    },
+    {
+      "epoch": 1429.29,
+      "eval_exact_match": 0.5696324951644101,
+      "eval_exec": 0.5522243713733076,
+      "eval_loss": 0.5556755065917969,
+      "eval_runtime": 793.4016,
+      "eval_samples_per_second": 1.303,
+      "step": 4288
+    },
+    {
+      "epoch": 1430.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4292
+    },
+    {
+      "epoch": 1431.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4296
+    },
+    {
+      "epoch": 1433.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4300
+    },
+    {
+      "epoch": 1434.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4304
+    },
+    {
+      "epoch": 1435.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4308
+    },
+    {
+      "epoch": 1437.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4312
+    },
+    {
+      "epoch": 1438.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4316
+    },
+    {
+      "epoch": 1439.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4320
+    },
+    {
+      "epoch": 1441.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4324
+    },
+    {
+      "epoch": 1442.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4328
+    },
+    {
+      "epoch": 1443.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4332
+    },
+    {
+      "epoch": 1445.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4336
+    },
+    {
+      "epoch": 1446.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4340
+    },
+    {
+      "epoch": 1447.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4344
+    },
+    {
+      "epoch": 1449.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4348
+    },
+    {
+      "epoch": 1450.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4352
+    },
+    {
+      "epoch": 1450.59,
+      "eval_exact_match": 0.5764023210831721,
+      "eval_exec": 0.5512572533849129,
+      "eval_loss": 0.5522686243057251,
+      "eval_runtime": 779.4291,
+      "eval_samples_per_second": 1.327,
+      "step": 4352
+    },
+    {
+      "epoch": 1451.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4356
+    },
+    {
+      "epoch": 1453.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4360
+    },
+    {
+      "epoch": 1454.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4364
+    },
+    {
+      "epoch": 1455.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4368
+    },
+    {
+      "epoch": 1457.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4372
+    },
+    {
+      "epoch": 1458.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4376
+    },
+    {
+      "epoch": 1459.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4380
+    },
+    {
+      "epoch": 1461.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4384
+    },
+    {
+      "epoch": 1462.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4388
+    },
+    {
+      "epoch": 1463.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4392
+    },
+    {
+      "epoch": 1465.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4396
+    },
+    {
+      "epoch": 1466.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4400
+    },
+    {
+      "epoch": 1467.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4404
+    },
+    {
+      "epoch": 1469.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4408
+    },
+    {
+      "epoch": 1470.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4412
+    },
+    {
+      "epoch": 1471.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4416
+    },
+    {
+      "epoch": 1471.88,
+      "eval_exact_match": 0.5696324951644101,
+      "eval_exec": 0.5464216634429401,
+      "eval_loss": 0.5721571445465088,
+      "eval_runtime": 777.7109,
+      "eval_samples_per_second": 1.33,
+      "step": 4416
+    },
+    {
+      "epoch": 1473.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 4420
+    },
+    {
+      "epoch": 1474.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4424
+    },
+    {
+      "epoch": 1475.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4428
+    },
+    {
+      "epoch": 1477.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4432
+    },
+    {
+      "epoch": 1478.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4436
+    },
+    {
+      "epoch": 1479.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4440
+    },
+    {
+      "epoch": 1481.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4444
+    },
+    {
+      "epoch": 1482.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4448
+    },
+    {
+      "epoch": 1483.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4452
+    },
+    {
+      "epoch": 1485.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4456
+    },
+    {
+      "epoch": 1486.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4460
+    },
+    {
+      "epoch": 1487.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4464
+    },
+    {
+      "epoch": 1489.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4468
+    },
+    {
+      "epoch": 1490.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4472
+    },
+    {
+      "epoch": 1491.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4476
+    },
+    {
+      "epoch": 1493.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4480
+    },
+    {
+      "epoch": 1493.29,
+      "eval_exact_match": 0.574468085106383,
+      "eval_exec": 0.5531914893617021,
+      "eval_loss": 0.5571283102035522,
+      "eval_runtime": 802.2815,
+      "eval_samples_per_second": 1.289,
+      "step": 4480
+    },
+    {
+      "epoch": 1494.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4484
+    },
+    {
+      "epoch": 1495.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4488
+    },
+    {
+      "epoch": 1497.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4492
+    },
+    {
+      "epoch": 1498.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4496
+    },
+    {
+      "epoch": 1499.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4500
+    },
+    {
+      "epoch": 1501.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4504
+    },
+    {
+      "epoch": 1502.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4508
+    },
+    {
+      "epoch": 1503.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 4512
+    },
+    {
+      "epoch": 1505.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4516
+    },
+    {
+      "epoch": 1506.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4520
+    },
+    {
+      "epoch": 1507.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4524
+    },
+    {
+      "epoch": 1509.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4528
+    },
+    {
+      "epoch": 1510.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4532
+    },
+    {
+      "epoch": 1511.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4536
+    },
+    {
+      "epoch": 1513.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4540
+    },
+    {
+      "epoch": 1514.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4544
+    },
+    {
+      "epoch": 1514.59,
+      "eval_exact_match": 0.5793036750483559,
+      "eval_exec": 0.5599613152804642,
+      "eval_loss": 0.5782627463340759,
+      "eval_runtime": 805.4733,
+      "eval_samples_per_second": 1.284,
+      "step": 4544
+    },
+    {
+      "epoch": 1515.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4548
+    },
+    {
+      "epoch": 1517.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4552
+    },
+    {
+      "epoch": 1518.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4556
+    },
+    {
+      "epoch": 1519.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4560
+    },
+    {
+      "epoch": 1521.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4564
+    },
+    {
+      "epoch": 1522.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4568
+    },
+    {
+      "epoch": 1523.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4572
+    },
+    {
+      "epoch": 1525.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4576
+    },
+    {
+      "epoch": 1526.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4580
+    },
+    {
+      "epoch": 1527.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4584
+    },
+    {
+      "epoch": 1529.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4588
+    },
+    {
+      "epoch": 1530.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4592
+    },
+    {
+      "epoch": 1531.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4596
+    },
+    {
+      "epoch": 1533.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4600
+    },
+    {
+      "epoch": 1534.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4604
+    },
+    {
+      "epoch": 1535.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4608
+    },
+    {
+      "epoch": 1535.88,
+      "eval_exact_match": 0.5851063829787234,
+      "eval_exec": 0.5647969052224371,
+      "eval_loss": 0.5510526299476624,
+      "eval_runtime": 825.4245,
+      "eval_samples_per_second": 1.253,
+      "step": 4608
+    },
+    {
+      "epoch": 1537.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4612
+    },
+    {
+      "epoch": 1538.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4616
+    },
+    {
+      "epoch": 1539.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4620
+    },
+    {
+      "epoch": 1541.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4624
+    },
+    {
+      "epoch": 1542.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4628
+    },
+    {
+      "epoch": 1543.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4632
+    },
+    {
+      "epoch": 1545.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4636
+    },
+    {
+      "epoch": 1546.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4640
+    },
+    {
+      "epoch": 1547.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4644
+    },
+    {
+      "epoch": 1549.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4648
+    },
+    {
+      "epoch": 1550.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4652
+    },
+    {
+      "epoch": 1551.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4656
+    },
+    {
+      "epoch": 1553.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4660
+    },
+    {
+      "epoch": 1554.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4664
+    },
+    {
+      "epoch": 1555.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4668
+    },
+    {
+      "epoch": 1557.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4672
+    },
+    {
+      "epoch": 1557.29,
+      "eval_exact_match": 0.5705996131528046,
+      "eval_exec": 0.5502901353965184,
+      "eval_loss": 0.5632638931274414,
+      "eval_runtime": 780.1472,
+      "eval_samples_per_second": 1.325,
+      "step": 4672
+    },
+    {
+      "epoch": 1558.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4676
+    },
+    {
+      "epoch": 1559.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4680
+    },
+    {
+      "epoch": 1561.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4684
+    },
+    {
+      "epoch": 1562.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4688
+    },
+    {
+      "epoch": 1563.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4692
+    },
+    {
+      "epoch": 1565.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4696
+    },
+    {
+      "epoch": 1566.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4700
+    },
+    {
+      "epoch": 1567.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 4704
+    },
+    {
+      "epoch": 1569.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4708
+    },
+    {
+      "epoch": 1570.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4712
+    },
+    {
+      "epoch": 1571.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4716
+    },
+    {
+      "epoch": 1573.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4720
+    },
+    {
+      "epoch": 1574.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4724
+    },
+    {
+      "epoch": 1575.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4728
+    },
+    {
+      "epoch": 1577.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4732
+    },
+    {
+      "epoch": 1578.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4736
+    },
+    {
+      "epoch": 1578.59,
+      "eval_exact_match": 0.5647969052224371,
+      "eval_exec": 0.5444874274661509,
+      "eval_loss": 0.5800946950912476,
+      "eval_runtime": 786.4833,
+      "eval_samples_per_second": 1.315,
+      "step": 4736
+    },
+    {
+      "epoch": 1579.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4740
+    },
+    {
+      "epoch": 1581.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4744
+    },
+    {
+      "epoch": 1582.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4748
+    },
+    {
+      "epoch": 1583.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4752
+    },
+    {
+      "epoch": 1585.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4756
+    },
+    {
+      "epoch": 1586.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4760
+    },
+    {
+      "epoch": 1587.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4764
+    },
+    {
+      "epoch": 1589.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4768
+    },
+    {
+      "epoch": 1590.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4772
+    },
+    {
+      "epoch": 1591.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4776
+    },
+    {
+      "epoch": 1593.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4780
+    },
+    {
+      "epoch": 1594.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4784
+    },
+    {
+      "epoch": 1595.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4788
+    },
+    {
+      "epoch": 1597.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4792
+    },
+    {
+      "epoch": 1598.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4796
+    },
+    {
+      "epoch": 1599.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4800
+    },
+    {
+      "epoch": 1599.88,
+      "eval_exact_match": 0.5725338491295938,
+      "eval_exec": 0.5512572533849129,
+      "eval_loss": 0.5803279876708984,
+      "eval_runtime": 776.3884,
+      "eval_samples_per_second": 1.332,
+      "step": 4800
+    },
+    {
+      "epoch": 1601.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4804
+    },
+    {
+      "epoch": 1602.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4808
+    },
+    {
+      "epoch": 1603.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4812
+    },
+    {
+      "epoch": 1605.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4816
+    },
+    {
+      "epoch": 1606.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4820
+    },
+    {
+      "epoch": 1607.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4824
+    },
+    {
+      "epoch": 1609.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4828
+    },
+    {
+      "epoch": 1610.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4832
+    },
+    {
+      "epoch": 1611.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4836
+    },
+    {
+      "epoch": 1613.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4840
+    },
+    {
+      "epoch": 1614.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4844
+    },
+    {
+      "epoch": 1615.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4848
+    },
+    {
+      "epoch": 1617.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4852
+    },
+    {
+      "epoch": 1618.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4856
+    },
+    {
+      "epoch": 1619.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4860
+    },
+    {
+      "epoch": 1621.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4864
+    },
+    {
+      "epoch": 1621.29,
+      "eval_exact_match": 0.5676982591876208,
+      "eval_exec": 0.5473887814313346,
+      "eval_loss": 0.5743498206138611,
+      "eval_runtime": 750.997,
+      "eval_samples_per_second": 1.377,
+      "step": 4864
+    },
+    {
+      "epoch": 1622.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 4868
+    },
+    {
+      "epoch": 1623.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4872
+    },
+    {
+      "epoch": 1625.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4876
+    },
+    {
+      "epoch": 1626.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4880
+    },
+    {
+      "epoch": 1627.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4884
+    },
+    {
+      "epoch": 1629.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4888
+    },
+    {
+      "epoch": 1630.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4892
+    },
+    {
+      "epoch": 1631.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4896
+    },
+    {
+      "epoch": 1633.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4900
+    },
+    {
+      "epoch": 1634.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4904
+    },
+    {
+      "epoch": 1635.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4908
+    },
+    {
+      "epoch": 1637.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4912
+    },
+    {
+      "epoch": 1638.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4916
+    },
+    {
+      "epoch": 1639.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4920
+    },
+    {
+      "epoch": 1641.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4924
+    },
+    {
+      "epoch": 1642.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 4928
+    },
+    {
+      "epoch": 1642.59,
+      "eval_exact_match": 0.5773694390715667,
+      "eval_exec": 0.5531914893617021,
+      "eval_loss": 0.5624192953109741,
+      "eval_runtime": 736.5225,
+      "eval_samples_per_second": 1.404,
+      "step": 4928
+    },
+    {
+      "epoch": 1643.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4932
+    },
+    {
+      "epoch": 1645.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4936
+    },
+    {
+      "epoch": 1646.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4940
+    },
+    {
+      "epoch": 1647.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4944
+    },
+    {
+      "epoch": 1649.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4948
+    },
+    {
+      "epoch": 1650.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4952
+    },
+    {
+      "epoch": 1651.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4956
+    },
+    {
+      "epoch": 1653.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4960
+    },
+    {
+      "epoch": 1654.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4964
+    },
+    {
+      "epoch": 1655.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4968
+    },
+    {
+      "epoch": 1657.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4972
+    },
+    {
+      "epoch": 1658.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4976
+    },
+    {
+      "epoch": 1659.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4980
+    },
+    {
+      "epoch": 1661.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 4984
+    },
+    {
+      "epoch": 1662.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 4988
+    },
+    {
+      "epoch": 1663.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 4992
+    },
+    {
+      "epoch": 1663.88,
+      "eval_exact_match": 0.5783365570599613,
+      "eval_exec": 0.5531914893617021,
+      "eval_loss": 0.5575723648071289,
+      "eval_runtime": 760.8089,
+      "eval_samples_per_second": 1.359,
+      "step": 4992
+    },
+    {
+      "epoch": 1665.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 4996
+    },
+    {
+      "epoch": 1666.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5000
+    },
+    {
+      "epoch": 1667.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5004
+    },
+    {
+      "epoch": 1669.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5008
+    },
+    {
+      "epoch": 1670.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5012
+    },
+    {
+      "epoch": 1671.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5016
+    },
+    {
+      "epoch": 1673.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5020
+    },
+    {
+      "epoch": 1674.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5024
+    },
+    {
+      "epoch": 1675.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5028
+    },
+    {
+      "epoch": 1677.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5032
+    },
+    {
+      "epoch": 1678.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5036
+    },
+    {
+      "epoch": 1679.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 5040
+    },
+    {
+      "epoch": 1681.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5044
+    },
+    {
+      "epoch": 1682.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5048
+    },
+    {
+      "epoch": 1683.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5052
+    },
+    {
+      "epoch": 1685.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5056
+    },
+    {
+      "epoch": 1685.29,
+      "eval_exact_match": 0.5735009671179884,
+      "eval_exec": 0.5551257253384912,
+      "eval_loss": 0.5752196907997131,
+      "eval_runtime": 742.7504,
+      "eval_samples_per_second": 1.392,
+      "step": 5056
+    },
+    {
+      "epoch": 1686.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5060
+    },
+    {
+      "epoch": 1687.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5064
+    },
+    {
+      "epoch": 1689.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5068
+    },
+    {
+      "epoch": 1690.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5072
+    },
+    {
+      "epoch": 1691.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5076
+    },
+    {
+      "epoch": 1693.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5080
+    },
+    {
+      "epoch": 1694.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0014,
+      "step": 5084
+    },
+    {
+      "epoch": 1695.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5088
+    },
+    {
+      "epoch": 1697.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5092
+    },
+    {
+      "epoch": 1698.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5096
+    },
+    {
+      "epoch": 1699.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 5100
+    },
+    {
+      "epoch": 1701.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5104
+    },
+    {
+      "epoch": 1702.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5108
+    },
+    {
+      "epoch": 1703.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5112
+    },
+    {
+      "epoch": 1705.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5116
+    },
+    {
+      "epoch": 1706.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5120
+    },
+    {
+      "epoch": 1706.59,
+      "eval_exact_match": 0.5764023210831721,
+      "eval_exec": 0.5599613152804642,
+      "eval_loss": 0.5517346858978271,
+      "eval_runtime": 767.1181,
+      "eval_samples_per_second": 1.348,
+      "step": 5120
+    },
+    {
+      "epoch": 1707.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5124
+    },
+    {
+      "epoch": 1709.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5128
+    },
+    {
+      "epoch": 1710.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5132
+    },
+    {
+      "epoch": 1711.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5136
+    },
+    {
+      "epoch": 1713.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 5140
+    },
+    {
+      "epoch": 1714.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5144
+    },
+    {
+      "epoch": 1715.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5148
+    },
+    {
+      "epoch": 1717.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5152
+    },
+    {
+      "epoch": 1718.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5156
+    },
+    {
+      "epoch": 1719.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5160
+    },
+    {
+      "epoch": 1721.29,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 5164
+    },
+    {
+      "epoch": 1722.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5168
+    },
+    {
+      "epoch": 1723.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5172
+    },
+    {
+      "epoch": 1725.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5176
+    },
+    {
+      "epoch": 1726.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5180
+    },
+    {
+      "epoch": 1727.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 5184
+    },
+    {
+      "epoch": 1727.88,
+      "eval_exact_match": 0.5783365570599613,
+      "eval_exec": 0.5531914893617021,
+      "eval_loss": 0.5575510859489441,
+      "eval_runtime": 747.6192,
+      "eval_samples_per_second": 1.383,
+      "step": 5184
+    },
+    {
+      "epoch": 1729.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5188
+    },
+    {
+      "epoch": 1730.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5192
+    },
+    {
+      "epoch": 1731.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5196
+    },
+    {
+      "epoch": 1733.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5200
+    },
+    {
+      "epoch": 1734.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5204
+    },
+    {
+      "epoch": 1735.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5208
+    },
+    {
+      "epoch": 1737.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5212
+    },
+    {
+      "epoch": 1738.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0036,
+      "step": 5216
+    },
+    {
+      "epoch": 1739.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 5220
+    },
+    {
+      "epoch": 1741.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5224
+    },
+    {
+      "epoch": 1742.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5228
+    },
+    {
+      "epoch": 1743.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5232
+    },
+    {
+      "epoch": 1745.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5236
+    },
+    {
+      "epoch": 1746.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 5240
+    },
+    {
+      "epoch": 1747.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5244
+    },
+    {
+      "epoch": 1749.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5248
+    },
+    {
+      "epoch": 1749.29,
+      "eval_exact_match": 0.5783365570599613,
+      "eval_exec": 0.5551257253384912,
+      "eval_loss": 0.54725581407547,
+      "eval_runtime": 742.8841,
+      "eval_samples_per_second": 1.392,
+      "step": 5248
+    },
+    {
+      "epoch": 1750.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5252
+    },
+    {
+      "epoch": 1751.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5256
+    },
+    {
+      "epoch": 1753.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5260
+    },
+    {
+      "epoch": 1754.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5264
+    },
+    {
+      "epoch": 1755.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5268
+    },
+    {
+      "epoch": 1757.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5272
+    },
+    {
+      "epoch": 1758.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5276
+    },
+    {
+      "epoch": 1759.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 5280
+    },
+    {
+      "epoch": 1761.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 5284
+    },
+    {
+      "epoch": 1762.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5288
+    },
+    {
+      "epoch": 1763.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5292
+    },
+    {
+      "epoch": 1765.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5296
+    },
+    {
+      "epoch": 1766.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5300
+    },
+    {
+      "epoch": 1767.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5304
+    },
+    {
+      "epoch": 1769.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5308
+    },
+    {
+      "epoch": 1770.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5312
+    },
+    {
+      "epoch": 1770.59,
+      "eval_exact_match": 0.5676982591876208,
+      "eval_exec": 0.5473887814313346,
+      "eval_loss": 0.5591778755187988,
+      "eval_runtime": 743.6516,
+      "eval_samples_per_second": 1.39,
+      "step": 5312
+    },
+    {
+      "epoch": 1771.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5316
+    },
+    {
+      "epoch": 1773.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5320
+    },
+    {
+      "epoch": 1774.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5324
+    },
+    {
+      "epoch": 1775.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5328
+    },
+    {
+      "epoch": 1777.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 5332
+    },
+    {
+      "epoch": 1778.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5336
+    },
+    {
+      "epoch": 1779.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5340
+    },
+    {
+      "epoch": 1781.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5344
+    },
+    {
+      "epoch": 1782.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5348
+    },
+    {
+      "epoch": 1783.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5352
+    },
+    {
+      "epoch": 1785.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5356
+    },
+    {
+      "epoch": 1786.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5360
+    },
+    {
+      "epoch": 1787.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5364
+    },
+    {
+      "epoch": 1789.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5368
+    },
+    {
+      "epoch": 1790.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5372
+    },
+    {
+      "epoch": 1791.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5376
+    },
+    {
+      "epoch": 1791.88,
+      "eval_exact_match": 0.5647969052224371,
+      "eval_exec": 0.5493230174081238,
+      "eval_loss": 0.552025318145752,
+      "eval_runtime": 743.4945,
+      "eval_samples_per_second": 1.391,
+      "step": 5376
+    },
+    {
+      "epoch": 1793.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5380
+    },
+    {
+      "epoch": 1794.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5384
+    },
+    {
+      "epoch": 1795.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5388
+    },
+    {
+      "epoch": 1797.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5392
+    },
+    {
+      "epoch": 1798.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 5396
+    },
+    {
+      "epoch": 1799.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5400
+    },
+    {
+      "epoch": 1801.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5404
+    },
+    {
+      "epoch": 1802.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5408
+    },
+    {
+      "epoch": 1803.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5412
+    },
+    {
+      "epoch": 1805.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5416
+    },
+    {
+      "epoch": 1806.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5420
+    },
+    {
+      "epoch": 1807.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5424
+    },
+    {
+      "epoch": 1809.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5428
+    },
+    {
+      "epoch": 1810.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5432
+    },
+    {
+      "epoch": 1811.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5436
+    },
+    {
+      "epoch": 1813.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5440
+    },
+    {
+      "epoch": 1813.29,
+      "eval_exact_match": 0.5657640232108317,
+      "eval_exec": 0.5444874274661509,
+      "eval_loss": 0.5696110129356384,
+      "eval_runtime": 752.5784,
+      "eval_samples_per_second": 1.374,
+      "step": 5440
+    },
+    {
+      "epoch": 1814.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5444
+    },
+    {
+      "epoch": 1815.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5448
+    },
+    {
+      "epoch": 1817.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5452
+    },
+    {
+      "epoch": 1818.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5456
+    },
+    {
+      "epoch": 1819.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5460
+    },
+    {
+      "epoch": 1821.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5464
+    },
+    {
+      "epoch": 1822.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5468
+    },
+    {
+      "epoch": 1823.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5472
+    },
+    {
+      "epoch": 1825.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5476
+    },
+    {
+      "epoch": 1826.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5480
+    },
+    {
+      "epoch": 1827.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5484
+    },
+    {
+      "epoch": 1829.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5488
+    },
+    {
+      "epoch": 1830.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 5492
+    },
+    {
+      "epoch": 1831.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5496
+    },
+    {
+      "epoch": 1833.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5500
+    },
+    {
+      "epoch": 1834.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5504
+    },
+    {
+      "epoch": 1834.59,
+      "eval_exact_match": 0.5822050290135397,
+      "eval_exec": 0.5647969052224371,
+      "eval_loss": 0.5593258142471313,
+      "eval_runtime": 767.7315,
+      "eval_samples_per_second": 1.347,
+      "step": 5504
+    },
+    {
+      "epoch": 1835.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5508
+    },
+    {
+      "epoch": 1837.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5512
+    },
+    {
+      "epoch": 1838.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5516
+    },
+    {
+      "epoch": 1839.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5520
+    },
+    {
+      "epoch": 1841.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5524
+    },
+    {
+      "epoch": 1842.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5528
+    },
+    {
+      "epoch": 1843.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5532
+    },
+    {
+      "epoch": 1845.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5536
+    },
+    {
+      "epoch": 1846.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5540
+    },
+    {
+      "epoch": 1847.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 5544
+    },
+    {
+      "epoch": 1849.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5548
+    },
+    {
+      "epoch": 1850.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5552
+    },
+    {
+      "epoch": 1851.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5556
+    },
+    {
+      "epoch": 1853.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5560
+    },
+    {
+      "epoch": 1854.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5564
+    },
+    {
+      "epoch": 1855.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5568
+    },
+    {
+      "epoch": 1855.88,
+      "eval_exact_match": 0.5764023210831721,
+      "eval_exec": 0.5609284332688588,
+      "eval_loss": 0.5532220005989075,
+      "eval_runtime": 752.4589,
+      "eval_samples_per_second": 1.374,
+      "step": 5568
+    },
+    {
+      "epoch": 1857.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5572
+    },
+    {
+      "epoch": 1858.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5576
+    },
+    {
+      "epoch": 1859.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5580
+    },
+    {
+      "epoch": 1861.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5584
+    },
+    {
+      "epoch": 1862.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5588
+    },
+    {
+      "epoch": 1863.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5592
+    },
+    {
+      "epoch": 1865.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5596
+    },
+    {
+      "epoch": 1866.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5600
+    },
+    {
+      "epoch": 1867.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5604
+    },
+    {
+      "epoch": 1869.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5608
+    },
+    {
+      "epoch": 1870.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5612
+    },
+    {
+      "epoch": 1871.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5616
+    },
+    {
+      "epoch": 1873.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5620
+    },
+    {
+      "epoch": 1874.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5624
+    },
+    {
+      "epoch": 1875.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5628
+    },
+    {
+      "epoch": 1877.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5632
+    },
+    {
+      "epoch": 1877.29,
+      "eval_exact_match": 0.5764023210831721,
+      "eval_exec": 0.5531914893617021,
+      "eval_loss": 0.5749230980873108,
+      "eval_runtime": 749.982,
+      "eval_samples_per_second": 1.379,
+      "step": 5632
+    },
+    {
+      "epoch": 1878.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5636
+    },
+    {
+      "epoch": 1879.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5640
+    },
+    {
+      "epoch": 1881.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5644
+    },
+    {
+      "epoch": 1882.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5648
+    },
+    {
+      "epoch": 1883.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5652
+    },
+    {
+      "epoch": 1885.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5656
+    },
+    {
+      "epoch": 1886.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5660
+    },
+    {
+      "epoch": 1887.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5664
+    },
+    {
+      "epoch": 1889.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5668
+    },
+    {
+      "epoch": 1890.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5672
+    },
+    {
+      "epoch": 1891.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5676
+    },
+    {
+      "epoch": 1893.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5680
+    },
+    {
+      "epoch": 1894.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5684
+    },
+    {
+      "epoch": 1895.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5688
+    },
+    {
+      "epoch": 1897.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5692
+    },
+    {
+      "epoch": 1898.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5696
+    },
+    {
+      "epoch": 1898.59,
+      "eval_exact_match": 0.5812379110251451,
+      "eval_exec": 0.5647969052224371,
+      "eval_loss": 0.577684760093689,
+      "eval_runtime": 771.8541,
+      "eval_samples_per_second": 1.34,
+      "step": 5696
+    },
+    {
+      "epoch": 1899.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5700
+    },
+    {
+      "epoch": 1901.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5704
+    },
+    {
+      "epoch": 1902.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5708
+    },
+    {
+      "epoch": 1903.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5712
+    },
+    {
+      "epoch": 1905.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5716
+    },
+    {
+      "epoch": 1906.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5720
+    },
+    {
+      "epoch": 1907.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5724
+    },
+    {
+      "epoch": 1909.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5728
+    },
+    {
+      "epoch": 1910.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5732
+    },
+    {
+      "epoch": 1911.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5736
+    },
+    {
+      "epoch": 1913.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5740
+    },
+    {
+      "epoch": 1914.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5744
+    },
+    {
+      "epoch": 1915.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5748
+    },
+    {
+      "epoch": 1917.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5752
+    },
+    {
+      "epoch": 1918.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5756
+    },
+    {
+      "epoch": 1919.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5760
+    },
+    {
+      "epoch": 1919.88,
+      "eval_exact_match": 0.5696324951644101,
+      "eval_exec": 0.5541586073500967,
+      "eval_loss": 0.5722802877426147,
+      "eval_runtime": 778.1483,
+      "eval_samples_per_second": 1.329,
+      "step": 5760
+    },
+    {
+      "epoch": 1921.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5764
+    },
+    {
+      "epoch": 1922.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5768
+    },
+    {
+      "epoch": 1923.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5772
+    },
+    {
+      "epoch": 1925.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5776
+    },
+    {
+      "epoch": 1926.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5780
+    },
+    {
+      "epoch": 1927.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5784
+    },
+    {
+      "epoch": 1929.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5788
+    },
+    {
+      "epoch": 1930.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5792
+    },
+    {
+      "epoch": 1931.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5796
+    },
+    {
+      "epoch": 1933.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5800
+    },
+    {
+      "epoch": 1934.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5804
+    },
+    {
+      "epoch": 1935.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5808
+    },
+    {
+      "epoch": 1937.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5812
+    },
+    {
+      "epoch": 1938.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5816
+    },
+    {
+      "epoch": 1939.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5820
+    },
+    {
+      "epoch": 1941.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5824
+    },
+    {
+      "epoch": 1941.29,
+      "eval_exact_match": 0.5851063829787234,
+      "eval_exec": 0.5599613152804642,
+      "eval_loss": 0.5796244144439697,
+      "eval_runtime": 777.1272,
+      "eval_samples_per_second": 1.331,
+      "step": 5824
+    },
+    {
+      "epoch": 1942.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5828
+    },
+    {
+      "epoch": 1943.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5832
+    },
+    {
+      "epoch": 1945.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5836
+    },
+    {
+      "epoch": 1946.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5840
+    },
+    {
+      "epoch": 1947.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5844
+    },
+    {
+      "epoch": 1949.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5848
+    },
+    {
+      "epoch": 1950.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5852
+    },
+    {
+      "epoch": 1951.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5856
+    },
+    {
+      "epoch": 1953.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5860
+    },
+    {
+      "epoch": 1954.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5864
+    },
+    {
+      "epoch": 1955.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5868
+    },
+    {
+      "epoch": 1957.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5872
+    },
+    {
+      "epoch": 1958.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 5876
+    },
+    {
+      "epoch": 1959.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5880
+    },
+    {
+      "epoch": 1961.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5884
+    },
+    {
+      "epoch": 1962.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5888
+    },
+    {
+      "epoch": 1962.59,
+      "eval_exact_match": 0.5783365570599613,
+      "eval_exec": 0.5609284332688588,
+      "eval_loss": 0.5814169645309448,
+      "eval_runtime": 773.9935,
+      "eval_samples_per_second": 1.336,
+      "step": 5888
+    },
+    {
+      "epoch": 1963.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5892
+    },
+    {
+      "epoch": 1965.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5896
+    },
+    {
+      "epoch": 1966.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5900
+    },
+    {
+      "epoch": 1967.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5904
+    },
+    {
+      "epoch": 1969.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5908
+    },
+    {
+      "epoch": 1970.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5912
+    },
+    {
+      "epoch": 1971.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5916
+    },
+    {
+      "epoch": 1973.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 5920
+    },
+    {
+      "epoch": 1974.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5924
+    },
+    {
+      "epoch": 1975.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5928
+    },
+    {
+      "epoch": 1977.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5932
+    },
+    {
+      "epoch": 1978.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5936
+    },
+    {
+      "epoch": 1979.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5940
+    },
+    {
+      "epoch": 1981.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5944
+    },
+    {
+      "epoch": 1982.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5948
+    },
+    {
+      "epoch": 1983.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0008,
+      "step": 5952
+    },
+    {
+      "epoch": 1983.88,
+      "eval_exact_match": 0.5686653771760155,
+      "eval_exec": 0.5483558994197292,
+      "eval_loss": 0.5712096095085144,
+      "eval_runtime": 777.5973,
+      "eval_samples_per_second": 1.33,
+      "step": 5952
+    },
+    {
+      "epoch": 1985.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5956
+    },
+    {
+      "epoch": 1986.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5960
+    },
+    {
+      "epoch": 1987.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5964
+    },
+    {
+      "epoch": 1989.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5968
+    },
+    {
+      "epoch": 1990.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 5972
+    },
+    {
+      "epoch": 1991.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5976
+    },
+    {
+      "epoch": 1993.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5980
+    },
+    {
+      "epoch": 1994.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5984
+    },
+    {
+      "epoch": 1995.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 5988
+    },
+    {
+      "epoch": 1997.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 5992
+    },
+    {
+      "epoch": 1998.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 5996
+    },
+    {
+      "epoch": 1999.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6000
+    },
+    {
+      "epoch": 2001.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6004
+    },
+    {
+      "epoch": 2002.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6008
+    },
+    {
+      "epoch": 2003.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6012
+    },
+    {
+      "epoch": 2005.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6016
+    },
+    {
+      "epoch": 2005.29,
+      "eval_exact_match": 0.5870406189555126,
+      "eval_exec": 0.558027079303675,
+      "eval_loss": 0.5804425477981567,
+      "eval_runtime": 756.6009,
+      "eval_samples_per_second": 1.367,
+      "step": 6016
+    },
+    {
+      "epoch": 2006.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6020
+    },
+    {
+      "epoch": 2007.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6024
+    },
+    {
+      "epoch": 2009.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 6028
+    },
+    {
+      "epoch": 2010.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6032
+    },
+    {
+      "epoch": 2011.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6036
+    },
+    {
+      "epoch": 2013.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6040
+    },
+    {
+      "epoch": 2014.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6044
+    },
+    {
+      "epoch": 2015.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 6048
+    },
+    {
+      "epoch": 2017.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6052
+    },
+    {
+      "epoch": 2018.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6056
+    },
+    {
+      "epoch": 2019.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6060
+    },
+    {
+      "epoch": 2021.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6064
+    },
+    {
+      "epoch": 2022.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6068
+    },
+    {
+      "epoch": 2023.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6072
+    },
+    {
+      "epoch": 2025.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6076
+    },
+    {
+      "epoch": 2026.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6080
+    },
+    {
+      "epoch": 2026.59,
+      "eval_exact_match": 0.5686653771760155,
+      "eval_exec": 0.5493230174081238,
+      "eval_loss": 0.5960844159126282,
+      "eval_runtime": 736.162,
+      "eval_samples_per_second": 1.405,
+      "step": 6080
+    },
+    {
+      "epoch": 2027.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6084
+    },
+    {
+      "epoch": 2029.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6088
+    },
+    {
+      "epoch": 2030.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6092
+    },
+    {
+      "epoch": 2031.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6096
+    },
+    {
+      "epoch": 2033.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6100
+    },
+    {
+      "epoch": 2034.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6104
+    },
+    {
+      "epoch": 2035.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6108
+    },
+    {
+      "epoch": 2037.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6112
+    },
+    {
+      "epoch": 2038.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6116
+    },
+    {
+      "epoch": 2039.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6120
+    },
+    {
+      "epoch": 2041.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6124
+    },
+    {
+      "epoch": 2042.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6128
+    },
+    {
+      "epoch": 2043.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6132
+    },
+    {
+      "epoch": 2045.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6136
+    },
+    {
+      "epoch": 2046.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6140
+    },
+    {
+      "epoch": 2047.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6144
+    },
+    {
+      "epoch": 2047.88,
+      "eval_exact_match": 0.5725338491295938,
+      "eval_exec": 0.5599613152804642,
+      "eval_loss": 0.5896542072296143,
+      "eval_runtime": 750.4016,
+      "eval_samples_per_second": 1.378,
+      "step": 6144
+    },
+    {
+      "epoch": 2049.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6148
+    },
+    {
+      "epoch": 2050.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6152
+    },
+    {
+      "epoch": 2051.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6156
+    },
+    {
+      "epoch": 2053.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6160
+    },
+    {
+      "epoch": 2054.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6164
+    },
+    {
+      "epoch": 2055.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6168
+    },
+    {
+      "epoch": 2057.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6172
+    },
+    {
+      "epoch": 2058.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6176
+    },
+    {
+      "epoch": 2059.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6180
+    },
+    {
+      "epoch": 2061.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6184
+    },
+    {
+      "epoch": 2062.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6188
+    },
+    {
+      "epoch": 2063.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6192
+    },
+    {
+      "epoch": 2065.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6196
+    },
+    {
+      "epoch": 2066.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6200
+    },
+    {
+      "epoch": 2067.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6204
+    },
+    {
+      "epoch": 2069.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6208
+    },
+    {
+      "epoch": 2069.29,
+      "eval_exact_match": 0.5764023210831721,
+      "eval_exec": 0.5551257253384912,
+      "eval_loss": 0.5848955512046814,
+      "eval_runtime": 758.9748,
+      "eval_samples_per_second": 1.362,
+      "step": 6208
+    },
+    {
+      "epoch": 2070.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6212
+    },
+    {
+      "epoch": 2071.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6216
+    },
+    {
+      "epoch": 2073.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6220
+    },
+    {
+      "epoch": 2074.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6224
+    },
+    {
+      "epoch": 2075.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6228
+    },
+    {
+      "epoch": 2077.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6232
+    },
+    {
+      "epoch": 2078.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6236
+    },
+    {
+      "epoch": 2079.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6240
+    },
+    {
+      "epoch": 2081.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6244
+    },
+    {
+      "epoch": 2082.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6248
+    },
+    {
+      "epoch": 2083.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0,
+      "step": 6252
+    },
+    {
+      "epoch": 2085.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6256
+    },
+    {
+      "epoch": 2086.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6260
+    },
+    {
+      "epoch": 2087.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6264
+    },
+    {
+      "epoch": 2089.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6268
+    },
+    {
+      "epoch": 2090.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6272
+    },
+    {
+      "epoch": 2090.59,
+      "eval_exact_match": 0.5783365570599613,
+      "eval_exec": 0.558027079303675,
+      "eval_loss": 0.6036255955696106,
+      "eval_runtime": 767.4245,
+      "eval_samples_per_second": 1.347,
+      "step": 6272
+    },
+    {
+      "epoch": 2091.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6276
+    },
+    {
+      "epoch": 2093.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6280
+    },
+    {
+      "epoch": 2094.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6284
+    },
+    {
+      "epoch": 2095.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6288
+    },
+    {
+      "epoch": 2097.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6292
+    },
+    {
+      "epoch": 2098.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6296
+    },
+    {
+      "epoch": 2099.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6300
+    },
+    {
+      "epoch": 2101.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6304
+    },
+    {
+      "epoch": 2102.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6308
+    },
+    {
+      "epoch": 2103.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6312
+    },
+    {
+      "epoch": 2105.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6316
+    },
+    {
+      "epoch": 2106.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 6320
+    },
+    {
+      "epoch": 2107.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6324
+    },
+    {
+      "epoch": 2109.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6328
+    },
+    {
+      "epoch": 2110.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6332
+    },
+    {
+      "epoch": 2111.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6336
+    },
+    {
+      "epoch": 2111.88,
+      "eval_exact_match": 0.5647969052224371,
+      "eval_exec": 0.5473887814313346,
+      "eval_loss": 0.5777551531791687,
+      "eval_runtime": 780.0584,
+      "eval_samples_per_second": 1.326,
+      "step": 6336
+    },
+    {
+      "epoch": 2113.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6340
+    },
+    {
+      "epoch": 2114.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6344
+    },
+    {
+      "epoch": 2115.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6348
+    },
+    {
+      "epoch": 2117.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6352
+    },
+    {
+      "epoch": 2118.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6356
+    },
+    {
+      "epoch": 2119.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6360
+    },
+    {
+      "epoch": 2121.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6364
+    },
+    {
+      "epoch": 2122.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6368
+    },
+    {
+      "epoch": 2123.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6372
+    },
+    {
+      "epoch": 2125.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6376
+    },
+    {
+      "epoch": 2126.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6380
+    },
+    {
+      "epoch": 2127.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6384
+    },
+    {
+      "epoch": 2129.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6388
+    },
+    {
+      "epoch": 2130.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6392
+    },
+    {
+      "epoch": 2131.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6396
+    },
+    {
+      "epoch": 2133.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6400
+    },
+    {
+      "epoch": 2133.29,
+      "eval_exact_match": 0.5735009671179884,
+      "eval_exec": 0.5560928433268859,
+      "eval_loss": 0.5904595255851746,
+      "eval_runtime": 764.6436,
+      "eval_samples_per_second": 1.352,
+      "step": 6400
+    },
+    {
+      "epoch": 2134.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6404
+    },
+    {
+      "epoch": 2135.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6408
+    },
+    {
+      "epoch": 2137.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0,
+      "step": 6412
+    },
+    {
+      "epoch": 2138.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0,
+      "step": 6416
+    },
+    {
+      "epoch": 2139.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6420
+    },
+    {
+      "epoch": 2141.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6424
+    },
+    {
+      "epoch": 2142.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6428
+    },
+    {
+      "epoch": 2143.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6432
+    },
+    {
+      "epoch": 2145.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6436
+    },
+    {
+      "epoch": 2146.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6440
+    },
+    {
+      "epoch": 2147.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6444
+    },
+    {
+      "epoch": 2149.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6448
+    },
+    {
+      "epoch": 2150.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0,
+      "step": 6452
+    },
+    {
+      "epoch": 2151.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6456
+    },
+    {
+      "epoch": 2153.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6460
+    },
+    {
+      "epoch": 2154.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6464
+    },
+    {
+      "epoch": 2154.59,
+      "eval_exact_match": 0.5715667311411993,
+      "eval_exec": 0.5541586073500967,
+      "eval_loss": 0.5959080457687378,
+      "eval_runtime": 803.0113,
+      "eval_samples_per_second": 1.288,
+      "step": 6464
+    },
+    {
+      "epoch": 2155.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6468
+    },
+    {
+      "epoch": 2157.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 6472
+    },
+    {
+      "epoch": 2158.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6476
+    },
+    {
+      "epoch": 2159.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6480
+    },
+    {
+      "epoch": 2161.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6484
+    },
+    {
+      "epoch": 2162.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6488
+    },
+    {
+      "epoch": 2163.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6492
+    },
+    {
+      "epoch": 2165.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6496
+    },
+    {
+      "epoch": 2166.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6500
+    },
+    {
+      "epoch": 2167.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6504
+    },
+    {
+      "epoch": 2169.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6508
+    },
+    {
+      "epoch": 2170.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6512
+    },
+    {
+      "epoch": 2171.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6516
+    },
+    {
+      "epoch": 2173.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6520
+    },
+    {
+      "epoch": 2174.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6524
+    },
+    {
+      "epoch": 2175.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6528
+    },
+    {
+      "epoch": 2175.88,
+      "eval_exact_match": 0.5802707930367504,
+      "eval_exec": 0.5599613152804642,
+      "eval_loss": 0.589346706867218,
+      "eval_runtime": 846.0447,
+      "eval_samples_per_second": 1.222,
+      "step": 6528
+    },
+    {
+      "epoch": 2177.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6532
+    },
+    {
+      "epoch": 2178.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6536
+    },
+    {
+      "epoch": 2179.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6540
+    },
+    {
+      "epoch": 2181.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 6544
+    },
+    {
+      "epoch": 2182.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6548
+    },
+    {
+      "epoch": 2183.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6552
+    },
+    {
+      "epoch": 2185.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6556
+    },
+    {
+      "epoch": 2186.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6560
+    },
+    {
+      "epoch": 2187.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6564
+    },
+    {
+      "epoch": 2189.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6568
+    },
+    {
+      "epoch": 2190.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6572
+    },
+    {
+      "epoch": 2191.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6576
+    },
+    {
+      "epoch": 2193.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6580
+    },
+    {
+      "epoch": 2194.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6584
+    },
+    {
+      "epoch": 2195.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6588
+    },
+    {
+      "epoch": 2197.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6592
+    },
+    {
+      "epoch": 2197.29,
+      "eval_exact_match": 0.5754352030947776,
+      "eval_exec": 0.5522243713733076,
+      "eval_loss": 0.5941418409347534,
+      "eval_runtime": 854.5308,
+      "eval_samples_per_second": 1.21,
+      "step": 6592
+    },
+    {
+      "epoch": 2198.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6596
+    },
+    {
+      "epoch": 2199.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6600
+    },
+    {
+      "epoch": 2201.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6604
+    },
+    {
+      "epoch": 2202.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6608
+    },
+    {
+      "epoch": 2203.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6612
+    },
+    {
+      "epoch": 2205.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6616
+    },
+    {
+      "epoch": 2206.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6620
+    },
+    {
+      "epoch": 2207.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6624
+    },
+    {
+      "epoch": 2209.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6628
+    },
+    {
+      "epoch": 2210.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0,
+      "step": 6632
+    },
+    {
+      "epoch": 2211.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6636
+    },
+    {
+      "epoch": 2213.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6640
+    },
+    {
+      "epoch": 2214.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0,
+      "step": 6644
+    },
+    {
+      "epoch": 2215.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6648
+    },
+    {
+      "epoch": 2217.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6652
+    },
+    {
+      "epoch": 2218.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6656
+    },
+    {
+      "epoch": 2218.59,
+      "eval_exact_match": 0.5705996131528046,
+      "eval_exec": 0.5531914893617021,
+      "eval_loss": 0.6073092222213745,
+      "eval_runtime": 824.5809,
+      "eval_samples_per_second": 1.254,
+      "step": 6656
+    },
+    {
+      "epoch": 2219.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6660
+    },
+    {
+      "epoch": 2221.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 6664
+    },
+    {
+      "epoch": 2222.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6668
+    },
+    {
+      "epoch": 2223.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6672
+    },
+    {
+      "epoch": 2225.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6676
+    },
+    {
+      "epoch": 2226.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6680
+    },
+    {
+      "epoch": 2227.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6684
+    },
+    {
+      "epoch": 2229.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6688
+    },
+    {
+      "epoch": 2230.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6692
+    },
+    {
+      "epoch": 2231.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6696
+    },
+    {
+      "epoch": 2233.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6700
+    },
+    {
+      "epoch": 2234.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6704
+    },
+    {
+      "epoch": 2235.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6708
+    },
+    {
+      "epoch": 2237.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6712
+    },
+    {
+      "epoch": 2238.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6716
+    },
+    {
+      "epoch": 2239.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 6720
+    },
+    {
+      "epoch": 2239.88,
+      "eval_exact_match": 0.5647969052224371,
+      "eval_exec": 0.5473887814313346,
+      "eval_loss": 0.58829265832901,
+      "eval_runtime": 846.314,
+      "eval_samples_per_second": 1.222,
+      "step": 6720
+    },
+    {
+      "epoch": 2241.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6724
+    },
+    {
+      "epoch": 2242.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6728
+    },
+    {
+      "epoch": 2243.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6732
+    },
+    {
+      "epoch": 2245.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6736
+    },
+    {
+      "epoch": 2246.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6740
+    },
+    {
+      "epoch": 2247.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6744
+    },
+    {
+      "epoch": 2249.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6748
+    },
+    {
+      "epoch": 2250.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6752
+    },
+    {
+      "epoch": 2251.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 6756
+    },
+    {
+      "epoch": 2253.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6760
+    },
+    {
+      "epoch": 2254.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6764
+    },
+    {
+      "epoch": 2255.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6768
+    },
+    {
+      "epoch": 2257.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6772
+    },
+    {
+      "epoch": 2258.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6776
+    },
+    {
+      "epoch": 2259.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6780
+    },
+    {
+      "epoch": 2261.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6784
+    },
+    {
+      "epoch": 2261.29,
+      "eval_exact_match": 0.5686653771760155,
+      "eval_exec": 0.5541586073500967,
+      "eval_loss": 0.5778570175170898,
+      "eval_runtime": 820.0682,
+      "eval_samples_per_second": 1.261,
+      "step": 6784
+    },
+    {
+      "epoch": 2262.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6788
+    },
+    {
+      "epoch": 2263.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6792
+    },
+    {
+      "epoch": 2265.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6796
+    },
+    {
+      "epoch": 2266.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6800
+    },
+    {
+      "epoch": 2267.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6804
+    },
+    {
+      "epoch": 2269.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6808
+    },
+    {
+      "epoch": 2270.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6812
+    },
+    {
+      "epoch": 2271.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6816
+    },
+    {
+      "epoch": 2273.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6820
+    },
+    {
+      "epoch": 2274.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6824
+    },
+    {
+      "epoch": 2275.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6828
+    },
+    {
+      "epoch": 2277.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6832
+    },
+    {
+      "epoch": 2278.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6836
+    },
+    {
+      "epoch": 2279.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6840
+    },
+    {
+      "epoch": 2281.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6844
+    },
+    {
+      "epoch": 2282.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6848
+    },
+    {
+      "epoch": 2282.59,
+      "eval_exact_match": 0.574468085106383,
+      "eval_exec": 0.5551257253384912,
+      "eval_loss": 0.5766767859458923,
+      "eval_runtime": 827.9144,
+      "eval_samples_per_second": 1.249,
+      "step": 6848
+    },
+    {
+      "epoch": 2283.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6852
+    },
+    {
+      "epoch": 2285.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6856
+    },
+    {
+      "epoch": 2286.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6860
+    },
+    {
+      "epoch": 2287.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6864
+    },
+    {
+      "epoch": 2289.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6868
+    },
+    {
+      "epoch": 2290.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6872
+    },
+    {
+      "epoch": 2291.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6876
+    },
+    {
+      "epoch": 2293.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6880
+    },
+    {
+      "epoch": 2294.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6884
+    },
+    {
+      "epoch": 2295.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6888
+    },
+    {
+      "epoch": 2297.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 6892
+    },
+    {
+      "epoch": 2298.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6896
+    },
+    {
+      "epoch": 2299.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6900
+    },
+    {
+      "epoch": 2301.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6904
+    },
+    {
+      "epoch": 2302.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6908
+    },
+    {
+      "epoch": 2303.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6912
+    },
+    {
+      "epoch": 2303.88,
+      "eval_exact_match": 0.5841392649903289,
+      "eval_exec": 0.562862669245648,
+      "eval_loss": 0.5999452471733093,
+      "eval_runtime": 815.8266,
+      "eval_samples_per_second": 1.267,
+      "step": 6912
+    },
+    {
+      "epoch": 2305.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6916
+    },
+    {
+      "epoch": 2306.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6920
+    },
+    {
+      "epoch": 2307.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6924
+    },
+    {
+      "epoch": 2309.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 6928
+    },
+    {
+      "epoch": 2310.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 6932
+    },
+    {
+      "epoch": 2311.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6936
+    },
+    {
+      "epoch": 2313.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6940
+    },
+    {
+      "epoch": 2314.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0,
+      "step": 6944
+    },
+    {
+      "epoch": 2315.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6948
+    },
+    {
+      "epoch": 2317.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6952
+    },
+    {
+      "epoch": 2318.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6956
+    },
+    {
+      "epoch": 2319.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0,
+      "step": 6960
+    },
+    {
+      "epoch": 2321.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6964
+    },
+    {
+      "epoch": 2322.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6968
+    },
+    {
+      "epoch": 2323.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6972
+    },
+    {
+      "epoch": 2325.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6976
+    },
+    {
+      "epoch": 2325.29,
+      "eval_exact_match": 0.5851063829787234,
+      "eval_exec": 0.5589941972920697,
+      "eval_loss": 0.6149199604988098,
+      "eval_runtime": 862.7841,
+      "eval_samples_per_second": 1.198,
+      "step": 6976
+    },
+    {
+      "epoch": 2326.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 6980
+    },
+    {
+      "epoch": 2327.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6984
+    },
+    {
+      "epoch": 2329.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6988
+    },
+    {
+      "epoch": 2330.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6992
+    },
+    {
+      "epoch": 2331.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 6996
+    },
+    {
+      "epoch": 2333.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7000
+    },
+    {
+      "epoch": 2334.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7004
+    },
+    {
+      "epoch": 2335.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7008
+    },
+    {
+      "epoch": 2337.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7012
+    },
+    {
+      "epoch": 2338.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7016
+    },
+    {
+      "epoch": 2339.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7020
+    },
+    {
+      "epoch": 2341.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7024
+    },
+    {
+      "epoch": 2342.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7028
+    },
+    {
+      "epoch": 2343.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7032
+    },
+    {
+      "epoch": 2345.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 7036
+    },
+    {
+      "epoch": 2346.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7040
+    },
+    {
+      "epoch": 2346.59,
+      "eval_exact_match": 0.5783365570599613,
+      "eval_exec": 0.5570599613152805,
+      "eval_loss": 0.5974252223968506,
+      "eval_runtime": 849.1866,
+      "eval_samples_per_second": 1.218,
+      "step": 7040
+    },
+    {
+      "epoch": 2347.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7044
+    },
+    {
+      "epoch": 2349.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7048
+    },
+    {
+      "epoch": 2350.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7052
+    },
+    {
+      "epoch": 2351.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7056
+    },
+    {
+      "epoch": 2353.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7060
+    },
+    {
+      "epoch": 2354.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7064
+    },
+    {
+      "epoch": 2355.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7068
+    },
+    {
+      "epoch": 2357.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7072
+    },
+    {
+      "epoch": 2358.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7076
+    },
+    {
+      "epoch": 2359.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7080
+    },
+    {
+      "epoch": 2361.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7084
+    },
+    {
+      "epoch": 2362.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7088
+    },
+    {
+      "epoch": 2363.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 7092
+    },
+    {
+      "epoch": 2365.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 7096
+    },
+    {
+      "epoch": 2366.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0011,
+      "step": 7100
+    },
+    {
+      "epoch": 2367.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0193,
+      "step": 7104
+    },
+    {
+      "epoch": 2367.88,
+      "eval_exact_match": 0.5589941972920697,
+      "eval_exec": 0.534816247582205,
+      "eval_loss": 0.5628390312194824,
+      "eval_runtime": 760.6994,
+      "eval_samples_per_second": 1.359,
+      "step": 7104
+    },
+    {
+      "epoch": 2369.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0047,
+      "step": 7108
+    },
+    {
+      "epoch": 2370.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 7112
+    },
+    {
+      "epoch": 2371.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0015,
+      "step": 7116
+    },
+    {
+      "epoch": 2373.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 7120
+    },
+    {
+      "epoch": 2374.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0016,
+      "step": 7124
+    },
+    {
+      "epoch": 2375.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 7128
+    },
+    {
+      "epoch": 2377.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 7132
+    },
+    {
+      "epoch": 2378.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 7136
+    },
+    {
+      "epoch": 2379.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 7140
+    },
+    {
+      "epoch": 2381.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 7144
+    },
+    {
+      "epoch": 2382.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0006,
+      "step": 7148
+    },
+    {
+      "epoch": 2383.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0007,
+      "step": 7152
+    },
+    {
+      "epoch": 2385.29,
+      "learning_rate": 0.0001,
+      "loss": 0.001,
+      "step": 7156
+    },
+    {
+      "epoch": 2386.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7160
+    },
+    {
+      "epoch": 2387.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 7164
+    },
+    {
+      "epoch": 2389.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7168
+    },
+    {
+      "epoch": 2389.29,
+      "eval_exact_match": 0.5735009671179884,
+      "eval_exec": 0.5531914893617021,
+      "eval_loss": 0.5860015749931335,
+      "eval_runtime": 737.6008,
+      "eval_samples_per_second": 1.402,
+      "step": 7168
+    },
+    {
+      "epoch": 2390.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7172
+    },
+    {
+      "epoch": 2391.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7176
+    },
+    {
+      "epoch": 2393.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7180
+    },
+    {
+      "epoch": 2394.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7184
+    },
+    {
+      "epoch": 2395.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7188
+    },
+    {
+      "epoch": 2397.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7192
+    },
+    {
+      "epoch": 2398.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7196
+    },
+    {
+      "epoch": 2399.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7200
+    },
+    {
+      "epoch": 2401.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7204
+    },
+    {
+      "epoch": 2402.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7208
+    },
+    {
+      "epoch": 2403.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7212
+    },
+    {
+      "epoch": 2405.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 7216
+    },
+    {
+      "epoch": 2406.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 7220
+    },
+    {
+      "epoch": 2407.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 7224
+    },
+    {
+      "epoch": 2409.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7228
+    },
+    {
+      "epoch": 2410.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7232
+    },
+    {
+      "epoch": 2410.59,
+      "eval_exact_match": 0.5793036750483559,
+      "eval_exec": 0.5638297872340425,
+      "eval_loss": 0.6152721643447876,
+      "eval_runtime": 803.2558,
+      "eval_samples_per_second": 1.287,
+      "step": 7232
+    },
+    {
+      "epoch": 2411.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7236
+    },
+    {
+      "epoch": 2413.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7240
+    },
+    {
+      "epoch": 2414.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7244
+    },
+    {
+      "epoch": 2415.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 7248
+    },
+    {
+      "epoch": 2417.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7252
+    },
+    {
+      "epoch": 2418.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0005,
+      "step": 7256
+    },
+    {
+      "epoch": 2419.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 7260
+    },
+    {
+      "epoch": 2421.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7264
+    },
+    {
+      "epoch": 2422.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7268
+    },
+    {
+      "epoch": 2423.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7272
+    },
+    {
+      "epoch": 2425.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 7276
+    },
+    {
+      "epoch": 2426.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7280
+    },
+    {
+      "epoch": 2427.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7284
+    },
+    {
+      "epoch": 2429.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7288
+    },
+    {
+      "epoch": 2430.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7292
+    },
+    {
+      "epoch": 2431.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7296
+    },
+    {
+      "epoch": 2431.88,
+      "eval_exact_match": 0.5754352030947776,
+      "eval_exec": 0.5570599613152805,
+      "eval_loss": 0.6106190085411072,
+      "eval_runtime": 791.7109,
+      "eval_samples_per_second": 1.306,
+      "step": 7296
+    },
+    {
+      "epoch": 2433.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7300
+    },
+    {
+      "epoch": 2434.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7304
+    },
+    {
+      "epoch": 2435.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7308
+    },
+    {
+      "epoch": 2437.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 7312
+    },
+    {
+      "epoch": 2438.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0004,
+      "step": 7316
+    },
+    {
+      "epoch": 2439.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7320
+    },
+    {
+      "epoch": 2441.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7324
+    },
+    {
+      "epoch": 2442.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7328
+    },
+    {
+      "epoch": 2443.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7332
+    },
+    {
+      "epoch": 2445.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7336
+    },
+    {
+      "epoch": 2446.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7340
+    },
+    {
+      "epoch": 2447.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7344
+    },
+    {
+      "epoch": 2449.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7348
+    },
+    {
+      "epoch": 2450.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 7352
+    },
+    {
+      "epoch": 2451.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7356
+    },
+    {
+      "epoch": 2453.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7360
+    },
+    {
+      "epoch": 2453.29,
+      "eval_exact_match": 0.5841392649903289,
+      "eval_exec": 0.5589941972920697,
+      "eval_loss": 0.6141391396522522,
+      "eval_runtime": 764.5355,
+      "eval_samples_per_second": 1.352,
+      "step": 7360
+    },
+    {
+      "epoch": 2454.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7364
+    },
+    {
+      "epoch": 2455.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7368
+    },
+    {
+      "epoch": 2457.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7372
+    },
+    {
+      "epoch": 2458.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7376
+    },
+    {
+      "epoch": 2459.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7380
+    },
+    {
+      "epoch": 2461.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7384
+    },
+    {
+      "epoch": 2462.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0,
+      "step": 7388
+    },
+    {
+      "epoch": 2463.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7392
+    },
+    {
+      "epoch": 2465.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 7396
+    },
+    {
+      "epoch": 2466.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0,
+      "step": 7400
+    },
+    {
+      "epoch": 2467.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7404
+    },
+    {
+      "epoch": 2469.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7408
+    },
+    {
+      "epoch": 2470.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7412
+    },
+    {
+      "epoch": 2471.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7416
+    },
+    {
+      "epoch": 2473.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7420
+    },
+    {
+      "epoch": 2474.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7424
+    },
+    {
+      "epoch": 2474.59,
+      "eval_exact_match": 0.5831721470019342,
+      "eval_exec": 0.5599613152804642,
+      "eval_loss": 0.6056718826293945,
+      "eval_runtime": 750.5989,
+      "eval_samples_per_second": 1.378,
+      "step": 7424
+    },
+    {
+      "epoch": 2475.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7428
+    },
+    {
+      "epoch": 2477.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7432
+    },
+    {
+      "epoch": 2478.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7436
+    },
+    {
+      "epoch": 2479.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7440
+    },
+    {
+      "epoch": 2481.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7444
+    },
+    {
+      "epoch": 2482.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0002,
+      "step": 7448
+    },
+    {
+      "epoch": 2483.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7452
+    },
+    {
+      "epoch": 2485.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7456
+    },
+    {
+      "epoch": 2486.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7460
+    },
+    {
+      "epoch": 2487.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7464
+    },
+    {
+      "epoch": 2489.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7468
+    },
+    {
+      "epoch": 2490.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0003,
+      "step": 7472
+    },
+    {
+      "epoch": 2491.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7476
+    },
+    {
+      "epoch": 2493.29,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7480
+    },
+    {
+      "epoch": 2494.59,
+      "learning_rate": 0.0001,
+      "loss": 0.0001,
+      "step": 7484
+    },
+    {
+      "epoch": 2495.88,
+      "learning_rate": 0.0001,
+      "loss": 0.0,
+      "step": 7488
+    },
+    {
+      "epoch": 2495.88,
+      "eval_exact_match": 0.5938104448742747,
+      "eval_exec": 0.5647969052224371,
+      "eval_loss": 0.620333731174469,
+      "eval_runtime": 727.1648,
+      "eval_samples_per_second": 1.422,
+      "step": 7488
+    }
+  ],
+  "max_steps": 9216,
+  "num_train_epochs": 3072,
+  "total_flos": 8.548781744243868e+18,
+  "trial_name": null,
+  "trial_params": null
+}