diff --git "a/checkpoint-675/trainer_state.json" "b/checkpoint-675/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-675/trainer_state.json"
@@ -0,0 +1,4902 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.2727272727272725,
+  "eval_steps": 38,
+  "global_step": 675,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.003367003367003367,
+      "grad_norm": 2.0,
+      "learning_rate": 2.2727272727272729e-07,
+      "loss": 1.0833,
+      "step": 1
+    },
+    {
+      "epoch": 0.003367003367003367,
+      "eval_loss": 1.032954454421997,
+      "eval_runtime": 8.7985,
+      "eval_samples_per_second": 56.828,
+      "eval_steps_per_second": 3.637,
+      "step": 1
+    },
+    {
+      "epoch": 0.006734006734006734,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.5454545454545457e-07,
+      "loss": 0.996,
+      "step": 2
+    },
+    {
+      "epoch": 0.010101010101010102,
+      "grad_norm": 1.640625,
+      "learning_rate": 6.818181818181818e-07,
+      "loss": 1.0489,
+      "step": 3
+    },
+    {
+      "epoch": 0.013468013468013467,
+      "grad_norm": 2.0625,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 1.0546,
+      "step": 4
+    },
+    {
+      "epoch": 0.016835016835016835,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.1363636363636364e-06,
+      "loss": 1.0341,
+      "step": 5
+    },
+    {
+      "epoch": 0.020202020202020204,
+      "grad_norm": 1.9921875,
+      "learning_rate": 1.3636363636363636e-06,
+      "loss": 1.0136,
+      "step": 6
+    },
+    {
+      "epoch": 0.02356902356902357,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.590909090909091e-06,
+      "loss": 1.0305,
+      "step": 7
+    },
+    {
+      "epoch": 0.026936026936026935,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 1.0226,
+      "step": 8
+    },
+    {
+      "epoch": 0.030303030303030304,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.0454545454545457e-06,
+      "loss": 1.0745,
+      "step": 9
+    },
+    {
+      "epoch": 0.03367003367003367,
+      "grad_norm": 2.359375,
+      "learning_rate": 2.2727272727272728e-06,
+      "loss": 1.0232,
+      "step": 10
+    },
+    {
+      "epoch": 0.037037037037037035,
+      "grad_norm": 2.53125,
+      "learning_rate": 2.5e-06,
+      "loss": 1.0106,
+      "step": 11
+    },
+    {
+      "epoch": 0.04040404040404041,
+      "grad_norm": 2.125,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 1.0541,
+      "step": 12
+    },
+    {
+      "epoch": 0.04377104377104377,
+      "grad_norm": 2.15625,
+      "learning_rate": 2.954545454545455e-06,
+      "loss": 1.0693,
+      "step": 13
+    },
+    {
+      "epoch": 0.04713804713804714,
+      "grad_norm": 1.8359375,
+      "learning_rate": 3.181818181818182e-06,
+      "loss": 1.0338,
+      "step": 14
+    },
+    {
+      "epoch": 0.050505050505050504,
+      "grad_norm": 2.0,
+      "learning_rate": 3.409090909090909e-06,
+      "loss": 1.0464,
+      "step": 15
+    },
+    {
+      "epoch": 0.05387205387205387,
+      "grad_norm": 2.34375,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 1.0684,
+      "step": 16
+    },
+    {
+      "epoch": 0.05723905723905724,
+      "grad_norm": 1.8125,
+      "learning_rate": 3.863636363636364e-06,
+      "loss": 1.0117,
+      "step": 17
+    },
+    {
+      "epoch": 0.06060606060606061,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.0909090909090915e-06,
+      "loss": 1.0138,
+      "step": 18
+    },
+    {
+      "epoch": 0.06397306397306397,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.3181818181818185e-06,
+      "loss": 0.9873,
+      "step": 19
+    },
+    {
+      "epoch": 0.06734006734006734,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 1.0524,
+      "step": 20
+    },
+    {
+      "epoch": 0.0707070707070707,
+      "grad_norm": 1.3984375,
+      "learning_rate": 4.772727272727273e-06,
+      "loss": 1.0479,
+      "step": 21
+    },
+    {
+      "epoch": 0.07407407407407407,
+      "grad_norm": 1.609375,
+      "learning_rate": 5e-06,
+      "loss": 0.9894,
+      "step": 22
+    },
+    {
+      "epoch": 0.07744107744107744,
+      "grad_norm": 1.3671875,
+      "learning_rate": 5.2272727272727274e-06,
+      "loss": 0.983,
+      "step": 23
+    },
+    {
+      "epoch": 0.08080808080808081,
+      "grad_norm": 1.6328125,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 1.0563,
+      "step": 24
+    },
+    {
+      "epoch": 0.08417508417508418,
+      "grad_norm": 1.34375,
+      "learning_rate": 5.681818181818183e-06,
+      "loss": 0.9966,
+      "step": 25
+    },
+    {
+      "epoch": 0.08754208754208755,
+      "grad_norm": 1.25,
+      "learning_rate": 5.90909090909091e-06,
+      "loss": 0.9922,
+      "step": 26
+    },
+    {
+      "epoch": 0.09090909090909091,
+      "grad_norm": 1.3984375,
+      "learning_rate": 6.136363636363637e-06,
+      "loss": 1.0677,
+      "step": 27
+    },
+    {
+      "epoch": 0.09427609427609428,
+      "grad_norm": 1.3828125,
+      "learning_rate": 6.363636363636364e-06,
+      "loss": 1.0325,
+      "step": 28
+    },
+    {
+      "epoch": 0.09764309764309764,
+      "grad_norm": 1.1953125,
+      "learning_rate": 6.590909090909091e-06,
+      "loss": 1.0159,
+      "step": 29
+    },
+    {
+      "epoch": 0.10101010101010101,
+      "grad_norm": 1.3125,
+      "learning_rate": 6.818181818181818e-06,
+      "loss": 1.0384,
+      "step": 30
+    },
+    {
+      "epoch": 0.10437710437710437,
+      "grad_norm": 1.1171875,
+      "learning_rate": 7.045454545454546e-06,
+      "loss": 0.9949,
+      "step": 31
+    },
+    {
+      "epoch": 0.10774410774410774,
+      "grad_norm": 1.2421875,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 1.0284,
+      "step": 32
+    },
+    {
+      "epoch": 0.1111111111111111,
+      "grad_norm": 1.15625,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.0245,
+      "step": 33
+    },
+    {
+      "epoch": 0.11447811447811448,
+      "grad_norm": 1.1875,
+      "learning_rate": 7.727272727272727e-06,
+      "loss": 1.0103,
+      "step": 34
+    },
+    {
+      "epoch": 0.11784511784511785,
+      "grad_norm": 1.125,
+      "learning_rate": 7.954545454545455e-06,
+      "loss": 1.0129,
+      "step": 35
+    },
+    {
+      "epoch": 0.12121212121212122,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.181818181818183e-06,
+      "loss": 0.965,
+      "step": 36
+    },
+    {
+      "epoch": 0.12457912457912458,
+      "grad_norm": 1.109375,
+      "learning_rate": 8.40909090909091e-06,
+      "loss": 1.0027,
+      "step": 37
+    },
+    {
+      "epoch": 0.12794612794612795,
+      "grad_norm": 1.0859375,
+      "learning_rate": 8.636363636363637e-06,
+      "loss": 1.0093,
+      "step": 38
+    },
+    {
+      "epoch": 0.12794612794612795,
+      "eval_loss": 0.9909602999687195,
+      "eval_runtime": 8.9064,
+      "eval_samples_per_second": 56.139,
+      "eval_steps_per_second": 3.593,
+      "step": 38
+    },
+    {
+      "epoch": 0.13131313131313133,
+      "grad_norm": 0.96484375,
+      "learning_rate": 8.863636363636365e-06,
+      "loss": 0.98,
+      "step": 39
+    },
+    {
+      "epoch": 0.13468013468013468,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 1.0,
+      "step": 40
+    },
+    {
+      "epoch": 0.13804713804713806,
+      "grad_norm": 1.109375,
+      "learning_rate": 9.318181818181819e-06,
+      "loss": 0.9658,
+      "step": 41
+    },
+    {
+      "epoch": 0.1414141414141414,
+      "grad_norm": 0.95703125,
+      "learning_rate": 9.545454545454547e-06,
+      "loss": 0.9656,
+      "step": 42
+    },
+    {
+      "epoch": 0.1447811447811448,
+      "grad_norm": 1.0,
+      "learning_rate": 9.772727272727273e-06,
+      "loss": 0.9528,
+      "step": 43
+    },
+    {
+      "epoch": 0.14814814814814814,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1e-05,
+      "loss": 0.9546,
+      "step": 44
+    },
+    {
+      "epoch": 0.15151515151515152,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.999965606811854e-06,
+      "loss": 0.9406,
+      "step": 45
+    },
+    {
+      "epoch": 0.15488215488215487,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.99986242772057e-06,
+      "loss": 0.9975,
+      "step": 46
+    },
+    {
+      "epoch": 0.15824915824915825,
+      "grad_norm": 0.94140625,
+      "learning_rate": 9.99969046414561e-06,
+      "loss": 0.9567,
+      "step": 47
+    },
+    {
+      "epoch": 0.16161616161616163,
+      "grad_norm": 0.96484375,
+      "learning_rate": 9.999449718452724e-06,
+      "loss": 0.9707,
+      "step": 48
+    },
+    {
+      "epoch": 0.16498316498316498,
+      "grad_norm": 1.0,
+      "learning_rate": 9.999140193953923e-06,
+      "loss": 0.9559,
+      "step": 49
+    },
+    {
+      "epoch": 0.16835016835016836,
+      "grad_norm": 0.99609375,
+      "learning_rate": 9.998761894907414e-06,
+      "loss": 0.9398,
+      "step": 50
+    },
+    {
+      "epoch": 0.1717171717171717,
+      "grad_norm": 0.83203125,
+      "learning_rate": 9.998314826517564e-06,
+      "loss": 0.9329,
+      "step": 51
+    },
+    {
+      "epoch": 0.1750841750841751,
+      "grad_norm": 0.91015625,
+      "learning_rate": 9.997798994934812e-06,
+      "loss": 0.9478,
+      "step": 52
+    },
+    {
+      "epoch": 0.17845117845117844,
+      "grad_norm": 0.94140625,
+      "learning_rate": 9.997214407255602e-06,
+      "loss": 0.9062,
+      "step": 53
+    },
+    {
+      "epoch": 0.18181818181818182,
+      "grad_norm": 0.83203125,
+      "learning_rate": 9.996561071522264e-06,
+      "loss": 0.9089,
+      "step": 54
+    },
+    {
+      "epoch": 0.18518518518518517,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.995838996722916e-06,
+      "loss": 0.9561,
+      "step": 55
+    },
+    {
+      "epoch": 0.18855218855218855,
+      "grad_norm": 0.796875,
+      "learning_rate": 9.995048192791341e-06,
+      "loss": 0.8948,
+      "step": 56
+    },
+    {
+      "epoch": 0.1919191919191919,
+      "grad_norm": 0.91796875,
+      "learning_rate": 9.994188670606845e-06,
+      "loss": 0.9095,
+      "step": 57
+    },
+    {
+      "epoch": 0.19528619528619529,
+      "grad_norm": 0.8515625,
+      "learning_rate": 9.993260441994116e-06,
+      "loss": 0.9137,
+      "step": 58
+    },
+    {
+      "epoch": 0.19865319865319866,
+      "grad_norm": 0.8671875,
+      "learning_rate": 9.992263519723046e-06,
+      "loss": 0.9012,
+      "step": 59
+    },
+    {
+      "epoch": 0.20202020202020202,
+      "grad_norm": 0.8046875,
+      "learning_rate": 9.99119791750857e-06,
+      "loss": 0.9074,
+      "step": 60
+    },
+    {
+      "epoch": 0.2053872053872054,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.990063650010473e-06,
+      "loss": 0.9106,
+      "step": 61
+    },
+    {
+      "epoch": 0.20875420875420875,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.988860732833183e-06,
+      "loss": 0.9012,
+      "step": 62
+    },
+    {
+      "epoch": 0.21212121212121213,
+      "grad_norm": 0.93359375,
+      "learning_rate": 9.987589182525562e-06,
+      "loss": 0.8928,
+      "step": 63
+    },
+    {
+      "epoch": 0.21548821548821548,
+      "grad_norm": 0.921875,
+      "learning_rate": 9.98624901658068e-06,
+      "loss": 0.9487,
+      "step": 64
+    },
+    {
+      "epoch": 0.21885521885521886,
+      "grad_norm": 0.92578125,
+      "learning_rate": 9.984840253435569e-06,
+      "loss": 0.8887,
+      "step": 65
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 0.8671875,
+      "learning_rate": 9.983362912470967e-06,
+      "loss": 0.9152,
+      "step": 66
+    },
+    {
+      "epoch": 0.2255892255892256,
+      "grad_norm": 0.83984375,
+      "learning_rate": 9.981817014011066e-06,
+      "loss": 0.8849,
+      "step": 67
+    },
+    {
+      "epoch": 0.22895622895622897,
+      "grad_norm": 0.94921875,
+      "learning_rate": 9.980202579323212e-06,
+      "loss": 0.9386,
+      "step": 68
+    },
+    {
+      "epoch": 0.23232323232323232,
+      "grad_norm": 0.8203125,
+      "learning_rate": 9.978519630617632e-06,
+      "loss": 0.8651,
+      "step": 69
+    },
+    {
+      "epoch": 0.2356902356902357,
+      "grad_norm": 0.7109375,
+      "learning_rate": 9.976768191047109e-06,
+      "loss": 0.9044,
+      "step": 70
+    },
+    {
+      "epoch": 0.23905723905723905,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.974948284706684e-06,
+      "loss": 0.8702,
+      "step": 71
+    },
+    {
+      "epoch": 0.24242424242424243,
+      "grad_norm": 0.91796875,
+      "learning_rate": 9.973059936633308e-06,
+      "loss": 0.8748,
+      "step": 72
+    },
+    {
+      "epoch": 0.24579124579124578,
+      "grad_norm": 0.66796875,
+      "learning_rate": 9.971103172805504e-06,
+      "loss": 0.8443,
+      "step": 73
+    },
+    {
+      "epoch": 0.24915824915824916,
+      "grad_norm": 0.671875,
+      "learning_rate": 9.969078020143013e-06,
+      "loss": 0.8605,
+      "step": 74
+    },
+    {
+      "epoch": 0.25252525252525254,
+      "grad_norm": 0.6796875,
+      "learning_rate": 9.966984506506415e-06,
+      "loss": 0.8661,
+      "step": 75
+    },
+    {
+      "epoch": 0.2558922558922559,
+      "grad_norm": 0.953125,
+      "learning_rate": 9.964822660696753e-06,
+      "loss": 0.9169,
+      "step": 76
+    },
+    {
+      "epoch": 0.2558922558922559,
+      "eval_loss": 0.8667835593223572,
+      "eval_runtime": 8.6819,
+      "eval_samples_per_second": 57.591,
+      "eval_steps_per_second": 3.686,
+      "step": 76
+    },
+    {
+      "epoch": 0.25925925925925924,
+      "grad_norm": 0.86328125,
+      "learning_rate": 9.96259251245514e-06,
+      "loss": 0.8719,
+      "step": 77
+    },
+    {
+      "epoch": 0.26262626262626265,
+      "grad_norm": 0.66796875,
+      "learning_rate": 9.960294092462332e-06,
+      "loss": 0.8438,
+      "step": 78
+    },
+    {
+      "epoch": 0.265993265993266,
+      "grad_norm": 0.66796875,
+      "learning_rate": 9.957927432338332e-06,
+      "loss": 0.8723,
+      "step": 79
+    },
+    {
+      "epoch": 0.26936026936026936,
+      "grad_norm": 1.3828125,
+      "learning_rate": 9.95549256464193e-06,
+      "loss": 0.8987,
+      "step": 80
+    },
+    {
+      "epoch": 0.2727272727272727,
+      "grad_norm": 0.69921875,
+      "learning_rate": 9.952989522870275e-06,
+      "loss": 0.8188,
+      "step": 81
+    },
+    {
+      "epoch": 0.2760942760942761,
+      "grad_norm": 0.9375,
+      "learning_rate": 9.950418341458398e-06,
+      "loss": 0.8521,
+      "step": 82
+    },
+    {
+      "epoch": 0.27946127946127947,
+      "grad_norm": 0.828125,
+      "learning_rate": 9.947779055778752e-06,
+      "loss": 0.8305,
+      "step": 83
+    },
+    {
+      "epoch": 0.2828282828282828,
+      "grad_norm": 0.9453125,
+      "learning_rate": 9.945071702140716e-06,
+      "loss": 0.856,
+      "step": 84
+    },
+    {
+      "epoch": 0.28619528619528617,
+      "grad_norm": 0.609375,
+      "learning_rate": 9.9422963177901e-06,
+      "loss": 0.8512,
+      "step": 85
+    },
+    {
+      "epoch": 0.2895622895622896,
+      "grad_norm": 1.546875,
+      "learning_rate": 9.939452940908627e-06,
+      "loss": 0.818,
+      "step": 86
+    },
+    {
+      "epoch": 0.29292929292929293,
+      "grad_norm": 0.640625,
+      "learning_rate": 9.936541610613417e-06,
+      "loss": 0.8071,
+      "step": 87
+    },
+    {
+      "epoch": 0.2962962962962963,
+      "grad_norm": 0.82421875,
+      "learning_rate": 9.933562366956445e-06,
+      "loss": 0.8559,
+      "step": 88
+    },
+    {
+      "epoch": 0.2996632996632997,
+      "grad_norm": 1.046875,
+      "learning_rate": 9.930515250923985e-06,
+      "loss": 0.8646,
+      "step": 89
+    },
+    {
+      "epoch": 0.30303030303030304,
+      "grad_norm": 0.76953125,
+      "learning_rate": 9.927400304436048e-06,
+      "loss": 0.8179,
+      "step": 90
+    },
+    {
+      "epoch": 0.3063973063973064,
+      "grad_norm": 1.09375,
+      "learning_rate": 9.924217570345814e-06,
+      "loss": 0.8093,
+      "step": 91
+    },
+    {
+      "epoch": 0.30976430976430974,
+      "grad_norm": 0.6015625,
+      "learning_rate": 9.920967092439028e-06,
+      "loss": 0.8073,
+      "step": 92
+    },
+    {
+      "epoch": 0.31313131313131315,
+      "grad_norm": 0.84375,
+      "learning_rate": 9.917648915433413e-06,
+      "loss": 0.9031,
+      "step": 93
+    },
+    {
+      "epoch": 0.3164983164983165,
+      "grad_norm": 0.76171875,
+      "learning_rate": 9.914263084978041e-06,
+      "loss": 0.8568,
+      "step": 94
+    },
+    {
+      "epoch": 0.31986531986531985,
+      "grad_norm": 0.8671875,
+      "learning_rate": 9.910809647652716e-06,
+      "loss": 0.7819,
+      "step": 95
+    },
+    {
+      "epoch": 0.32323232323232326,
+      "grad_norm": 0.84765625,
+      "learning_rate": 9.907288650967324e-06,
+      "loss": 0.8236,
+      "step": 96
+    },
+    {
+      "epoch": 0.3265993265993266,
+      "grad_norm": 0.70703125,
+      "learning_rate": 9.903700143361185e-06,
+      "loss": 0.8418,
+      "step": 97
+    },
+    {
+      "epoch": 0.32996632996632996,
+      "grad_norm": 0.94921875,
+      "learning_rate": 9.900044174202389e-06,
+      "loss": 0.8313,
+      "step": 98
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.8125,
+      "learning_rate": 9.896320793787106e-06,
+      "loss": 0.8158,
+      "step": 99
+    },
+    {
+      "epoch": 0.3367003367003367,
+      "grad_norm": 0.76171875,
+      "learning_rate": 9.892530053338909e-06,
+      "loss": 0.7956,
+      "step": 100
+    },
+    {
+      "epoch": 0.3400673400673401,
+      "grad_norm": 0.94921875,
+      "learning_rate": 9.888672005008054e-06,
+      "loss": 0.8154,
+      "step": 101
+    },
+    {
+      "epoch": 0.3434343434343434,
+      "grad_norm": 0.6328125,
+      "learning_rate": 9.884746701870778e-06,
+      "loss": 0.8322,
+      "step": 102
+    },
+    {
+      "epoch": 0.3468013468013468,
+      "grad_norm": 0.62890625,
+      "learning_rate": 9.880754197928553e-06,
+      "loss": 0.8133,
+      "step": 103
+    },
+    {
+      "epoch": 0.3501683501683502,
+      "grad_norm": 0.76953125,
+      "learning_rate": 9.876694548107358e-06,
+      "loss": 0.7882,
+      "step": 104
+    },
+    {
+      "epoch": 0.35353535353535354,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.87256780825691e-06,
+      "loss": 0.8264,
+      "step": 105
+    },
+    {
+      "epoch": 0.3569023569023569,
+      "grad_norm": 0.7421875,
+      "learning_rate": 9.868374035149906e-06,
+      "loss": 0.7894,
+      "step": 106
+    },
+    {
+      "epoch": 0.3602693602693603,
+      "grad_norm": 0.6171875,
+      "learning_rate": 9.864113286481237e-06,
+      "loss": 0.7807,
+      "step": 107
+    },
+    {
+      "epoch": 0.36363636363636365,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.859785620867197e-06,
+      "loss": 0.7895,
+      "step": 108
+    },
+    {
+      "epoch": 0.367003367003367,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.855391097844671e-06,
+      "loss": 0.7936,
+      "step": 109
+    },
+    {
+      "epoch": 0.37037037037037035,
+      "grad_norm": 0.8984375,
+      "learning_rate": 9.850929777870324e-06,
+      "loss": 0.8024,
+      "step": 110
+    },
+    {
+      "epoch": 0.37373737373737376,
+      "grad_norm": 0.6640625,
+      "learning_rate": 9.846401722319759e-06,
+      "loss": 0.7542,
+      "step": 111
+    },
+    {
+      "epoch": 0.3771043771043771,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.841806993486686e-06,
+      "loss": 0.7689,
+      "step": 112
+    },
+    {
+      "epoch": 0.38047138047138046,
+      "grad_norm": 0.765625,
+      "learning_rate": 9.837145654582055e-06,
+      "loss": 0.7133,
+      "step": 113
+    },
+    {
+      "epoch": 0.3838383838383838,
+      "grad_norm": 0.87109375,
+      "learning_rate": 9.832417769733185e-06,
+      "loss": 0.795,
+      "step": 114
+    },
+    {
+      "epoch": 0.3838383838383838,
+      "eval_loss": 0.7676474452018738,
+      "eval_runtime": 8.7789,
+      "eval_samples_per_second": 56.955,
+      "eval_steps_per_second": 3.645,
+      "step": 114
+    },
+    {
+      "epoch": 0.3872053872053872,
+      "grad_norm": 0.86328125,
+      "learning_rate": 9.827623403982893e-06,
+      "loss": 0.7992,
+      "step": 115
+    },
+    {
+      "epoch": 0.39057239057239057,
+      "grad_norm": 0.7421875,
+      "learning_rate": 9.822762623288584e-06,
+      "loss": 0.7678,
+      "step": 116
+    },
+    {
+      "epoch": 0.3939393939393939,
+      "grad_norm": 0.85546875,
+      "learning_rate": 9.81783549452136e-06,
+      "loss": 0.7861,
+      "step": 117
+    },
+    {
+      "epoch": 0.39730639730639733,
+      "grad_norm": 0.71484375,
+      "learning_rate": 9.812842085465086e-06,
+      "loss": 0.7447,
+      "step": 118
+    },
+    {
+      "epoch": 0.4006734006734007,
+      "grad_norm": 0.5703125,
+      "learning_rate": 9.807782464815463e-06,
+      "loss": 0.7218,
+      "step": 119
+    },
+    {
+      "epoch": 0.40404040404040403,
+      "grad_norm": 0.6171875,
+      "learning_rate": 9.802656702179088e-06,
+      "loss": 0.7637,
+      "step": 120
+    },
+    {
+      "epoch": 0.4074074074074074,
+      "grad_norm": 0.54296875,
+      "learning_rate": 9.797464868072489e-06,
+      "loss": 0.7588,
+      "step": 121
+    },
+    {
+      "epoch": 0.4107744107744108,
+      "grad_norm": 0.6171875,
+      "learning_rate": 9.792207033921152e-06,
+      "loss": 0.7539,
+      "step": 122
+    },
+    {
+      "epoch": 0.41414141414141414,
+      "grad_norm": 0.8046875,
+      "learning_rate": 9.786883272058554e-06,
+      "loss": 0.7524,
+      "step": 123
+    },
+    {
+      "epoch": 0.4175084175084175,
+      "grad_norm": 0.6953125,
+      "learning_rate": 9.781493655725149e-06,
+      "loss": 0.731,
+      "step": 124
+    },
+    {
+      "epoch": 0.4208754208754209,
+      "grad_norm": 0.64453125,
+      "learning_rate": 9.776038259067375e-06,
+      "loss": 0.8022,
+      "step": 125
+    },
+    {
+      "epoch": 0.42424242424242425,
+      "grad_norm": 0.48046875,
+      "learning_rate": 9.770517157136625e-06,
+      "loss": 0.7528,
+      "step": 126
+    },
+    {
+      "epoch": 0.4276094276094276,
+      "grad_norm": 0.5546875,
+      "learning_rate": 9.764930425888216e-06,
+      "loss": 0.7037,
+      "step": 127
+    },
+    {
+      "epoch": 0.43097643097643096,
+      "grad_norm": 0.546875,
+      "learning_rate": 9.759278142180348e-06,
+      "loss": 0.7519,
+      "step": 128
+    },
+    {
+      "epoch": 0.43434343434343436,
+      "grad_norm": 0.61328125,
+      "learning_rate": 9.753560383773046e-06,
+      "loss": 0.8163,
+      "step": 129
+    },
+    {
+      "epoch": 0.4377104377104377,
+      "grad_norm": 0.53125,
+      "learning_rate": 9.747777229327085e-06,
+      "loss": 0.7292,
+      "step": 130
+    },
+    {
+      "epoch": 0.44107744107744107,
+      "grad_norm": 0.58203125,
+      "learning_rate": 9.741928758402912e-06,
+      "loss": 0.7351,
+      "step": 131
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.5234375,
+      "learning_rate": 9.736015051459551e-06,
+      "loss": 0.7429,
+      "step": 132
+    },
+    {
+      "epoch": 0.4478114478114478,
+      "grad_norm": 0.65234375,
+      "learning_rate": 9.730036189853497e-06,
+      "loss": 0.7616,
+      "step": 133
+    },
+    {
+      "epoch": 0.4511784511784512,
+      "grad_norm": 0.53125,
+      "learning_rate": 9.723992255837596e-06,
+      "loss": 0.7263,
+      "step": 134
+    },
+    {
+      "epoch": 0.45454545454545453,
+      "grad_norm": 0.5546875,
+      "learning_rate": 9.717883332559911e-06,
+      "loss": 0.7427,
+      "step": 135
+    },
+    {
+      "epoch": 0.45791245791245794,
+      "grad_norm": 0.66796875,
+      "learning_rate": 9.71170950406258e-06,
+      "loss": 0.7261,
+      "step": 136
+    },
+    {
+      "epoch": 0.4612794612794613,
+      "grad_norm": 0.58984375,
+      "learning_rate": 9.705470855280661e-06,
+      "loss": 0.754,
+      "step": 137
+    },
+    {
+      "epoch": 0.46464646464646464,
+      "grad_norm": 0.6015625,
+      "learning_rate": 9.699167472040965e-06,
+      "loss": 0.7554,
+      "step": 138
+    },
+    {
+      "epoch": 0.468013468013468,
+      "grad_norm": 0.5546875,
+      "learning_rate": 9.692799441060868e-06,
+      "loss": 0.7488,
+      "step": 139
+    },
+    {
+      "epoch": 0.4713804713804714,
+      "grad_norm": 0.53125,
+      "learning_rate": 9.686366849947126e-06,
+      "loss": 0.7066,
+      "step": 140
+    },
+    {
+      "epoch": 0.47474747474747475,
+      "grad_norm": 0.55078125,
+      "learning_rate": 9.679869787194664e-06,
+      "loss": 0.7818,
+      "step": 141
+    },
+    {
+      "epoch": 0.4781144781144781,
+      "grad_norm": 0.4765625,
+      "learning_rate": 9.673308342185366e-06,
+      "loss": 0.7289,
+      "step": 142
+    },
+    {
+      "epoch": 0.48148148148148145,
+      "grad_norm": 0.62109375,
+      "learning_rate": 9.666682605186834e-06,
+      "loss": 0.7301,
+      "step": 143
+    },
+    {
+      "epoch": 0.48484848484848486,
+      "grad_norm": 0.54296875,
+      "learning_rate": 9.659992667351157e-06,
+      "loss": 0.7673,
+      "step": 144
+    },
+    {
+      "epoch": 0.4882154882154882,
+      "grad_norm": 0.486328125,
+      "learning_rate": 9.653238620713652e-06,
+      "loss": 0.6884,
+      "step": 145
+    },
+    {
+      "epoch": 0.49158249158249157,
+      "grad_norm": 0.6484375,
+      "learning_rate": 9.646420558191596e-06,
+      "loss": 0.7043,
+      "step": 146
+    },
+    {
+      "epoch": 0.494949494949495,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.639538573582952e-06,
+      "loss": 0.7122,
+      "step": 147
+    },
+    {
+      "epoch": 0.4983164983164983,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.632592761565078e-06,
+      "loss": 0.7232,
+      "step": 148
+    },
+    {
+      "epoch": 0.5016835016835017,
+      "grad_norm": 0.5625,
+      "learning_rate": 9.625583217693419e-06,
+      "loss": 0.7311,
+      "step": 149
+    },
+    {
+      "epoch": 0.5050505050505051,
+      "grad_norm": 0.63671875,
+      "learning_rate": 9.618510038400203e-06,
+      "loss": 0.7322,
+      "step": 150
+    },
+    {
+      "epoch": 0.5084175084175084,
+      "grad_norm": 0.75390625,
+      "learning_rate": 9.611373320993104e-06,
+      "loss": 0.6983,
+      "step": 151
+    },
+    {
+      "epoch": 0.5117845117845118,
+      "grad_norm": 0.53125,
+      "learning_rate": 9.604173163653906e-06,
+      "loss": 0.6999,
+      "step": 152
+    },
+    {
+      "epoch": 0.5117845117845118,
+      "eval_loss": 0.7242563366889954,
+      "eval_runtime": 8.7081,
+      "eval_samples_per_second": 57.418,
+      "eval_steps_per_second": 3.675,
+      "step": 152
+    },
+    {
+      "epoch": 0.5151515151515151,
+      "grad_norm": 0.55859375,
+      "learning_rate": 9.596909665437155e-06,
+      "loss": 0.7469,
+      "step": 153
+    },
+    {
+      "epoch": 0.5185185185185185,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.589582926268798e-06,
+      "loss": 0.7176,
+      "step": 154
+    },
+    {
+      "epoch": 0.5218855218855218,
+      "grad_norm": 0.4921875,
+      "learning_rate": 9.582193046944799e-06,
+      "loss": 0.7455,
+      "step": 155
+    },
+    {
+      "epoch": 0.5252525252525253,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.574740129129767e-06,
+      "loss": 0.7204,
+      "step": 156
+    },
+    {
+      "epoch": 0.5286195286195287,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.567224275355538e-06,
+      "loss": 0.7321,
+      "step": 157
+    },
+    {
+      "epoch": 0.531986531986532,
+      "grad_norm": 0.53125,
+      "learning_rate": 9.559645589019787e-06,
+      "loss": 0.7455,
+      "step": 158
+    },
+    {
+      "epoch": 0.5353535353535354,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.552004174384582e-06,
+      "loss": 0.7464,
+      "step": 159
+    },
+    {
+      "epoch": 0.5387205387205387,
+      "grad_norm": 0.5234375,
+      "learning_rate": 9.544300136574973e-06,
+      "loss": 0.7138,
+      "step": 160
+    },
+    {
+      "epoch": 0.5420875420875421,
+      "grad_norm": 0.59375,
+      "learning_rate": 9.536533581577526e-06,
+      "loss": 0.7321,
+      "step": 161
+    },
+    {
+      "epoch": 0.5454545454545454,
+      "grad_norm": 0.57421875,
+      "learning_rate": 9.528704616238875e-06,
+      "loss": 0.7205,
+      "step": 162
+    },
+    {
+      "epoch": 0.5488215488215489,
+      "grad_norm": 0.484375,
+      "learning_rate": 9.520813348264252e-06,
+      "loss": 0.7107,
+      "step": 163
+    },
+    {
+      "epoch": 0.5521885521885522,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.512859886216003e-06,
+      "loss": 0.6975,
+      "step": 164
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 0.46875,
+      "learning_rate": 9.504844339512096e-06,
+      "loss": 0.7209,
+      "step": 165
+    },
+    {
+      "epoch": 0.5589225589225589,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.496766818424612e-06,
+      "loss": 0.6932,
+      "step": 166
+    },
+    {
+      "epoch": 0.5622895622895623,
+      "grad_norm": 0.69140625,
+      "learning_rate": 9.488627434078232e-06,
+      "loss": 0.6676,
+      "step": 167
+    },
+    {
+      "epoch": 0.5656565656565656,
+      "grad_norm": 0.443359375,
+      "learning_rate": 9.480426298448706e-06,
+      "loss": 0.6909,
+      "step": 168
+    },
+    {
+      "epoch": 0.569023569023569,
+      "grad_norm": 0.4609375,
+      "learning_rate": 9.472163524361317e-06,
+      "loss": 0.7142,
+      "step": 169
+    },
+    {
+      "epoch": 0.5723905723905723,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.46383922548932e-06,
+      "loss": 0.7168,
+      "step": 170
+    },
+    {
+      "epoch": 0.5757575757575758,
+      "grad_norm": 0.435546875,
+      "learning_rate": 9.455453516352385e-06,
+      "loss": 0.6618,
+      "step": 171
+    },
+    {
+      "epoch": 0.5791245791245792,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.447006512315025e-06,
+      "loss": 0.694,
+      "step": 172
+    },
+    {
+      "epoch": 0.5824915824915825,
+      "grad_norm": 0.59765625,
+      "learning_rate": 9.438498329584995e-06,
+      "loss": 0.6858,
+      "step": 173
+    },
+    {
+      "epoch": 0.5858585858585859,
+      "grad_norm": 0.48046875,
+      "learning_rate": 9.429929085211711e-06,
+      "loss": 0.7419,
+      "step": 174
+    },
+    {
+      "epoch": 0.5892255892255892,
+      "grad_norm": 0.4375,
+      "learning_rate": 9.421298897084622e-06,
+      "loss": 0.6734,
+      "step": 175
+    },
+    {
+      "epoch": 0.5925925925925926,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.412607883931608e-06,
+      "loss": 0.7188,
+      "step": 176
+    },
+    {
+      "epoch": 0.5959595959595959,
+      "grad_norm": 0.41796875,
+      "learning_rate": 9.403856165317322e-06,
+      "loss": 0.6732,
+      "step": 177
+    },
+    {
+      "epoch": 0.5993265993265994,
+      "grad_norm": 0.59765625,
+      "learning_rate": 9.39504386164157e-06,
+      "loss": 0.6594,
+      "step": 178
+    },
+    {
+      "epoch": 0.6026936026936027,
+      "grad_norm": 0.490234375,
+      "learning_rate": 9.38617109413764e-06,
+      "loss": 0.7153,
+      "step": 179
+    },
+    {
+      "epoch": 0.6060606060606061,
+      "grad_norm": 0.60546875,
+      "learning_rate": 9.377237984870636e-06,
+      "loss": 0.6762,
+      "step": 180
+    },
+    {
+      "epoch": 0.6094276094276094,
+      "grad_norm": 0.53125,
+      "learning_rate": 9.3682446567358e-06,
+      "loss": 0.6839,
+      "step": 181
+    },
+    {
+      "epoch": 0.6127946127946128,
+      "grad_norm": 0.515625,
+      "learning_rate": 9.359191233456821e-06,
+      "loss": 0.6949,
+      "step": 182
+    },
+    {
+      "epoch": 0.6161616161616161,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.350077839584139e-06,
+      "loss": 0.7045,
+      "step": 183
+    },
+    {
+      "epoch": 0.6195286195286195,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.34090460049322e-06,
+      "loss": 0.668,
+      "step": 184
+    },
+    {
+      "epoch": 0.622895622895623,
+      "grad_norm": 0.54296875,
+      "learning_rate": 9.33167164238284e-06,
+      "loss": 0.7144,
+      "step": 185
+    },
+    {
+      "epoch": 0.6262626262626263,
+      "grad_norm": 0.44921875,
+      "learning_rate": 9.322379092273345e-06,
+      "loss": 0.7084,
+      "step": 186
+    },
+    {
+      "epoch": 0.6296296296296297,
+      "grad_norm": 0.427734375,
+      "learning_rate": 9.313027078004903e-06,
+      "loss": 0.7555,
+      "step": 187
+    },
+    {
+      "epoch": 0.632996632996633,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.303615728235753e-06,
+      "loss": 0.6963,
+      "step": 188
+    },
+    {
+      "epoch": 0.6363636363636364,
+      "grad_norm": 0.515625,
+      "learning_rate": 9.29414517244042e-06,
+      "loss": 0.6785,
+      "step": 189
+    },
+    {
+      "epoch": 0.6397306397306397,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.284615540907947e-06,
+      "loss": 0.7246,
+      "step": 190
+    },
+    {
+      "epoch": 0.6397306397306397,
+      "eval_loss": 0.6989265084266663,
+      "eval_runtime": 8.6934,
+      "eval_samples_per_second": 57.515,
+      "eval_steps_per_second": 3.681,
+      "step": 190
+    },
+    {
+      "epoch": 0.6430976430976431,
+      "grad_norm": 0.54296875,
+      "learning_rate": 9.275026964740101e-06,
+      "loss": 0.6995,
+      "step": 191
+    },
+    {
+      "epoch": 0.6464646464646465,
+      "grad_norm": 0.427734375,
+      "learning_rate": 9.265379575849561e-06,
+      "loss": 0.6988,
+      "step": 192
+    },
+    {
+      "epoch": 0.6498316498316499,
+      "grad_norm": 0.48046875,
+      "learning_rate": 9.255673506958114e-06,
+      "loss": 0.6569,
+      "step": 193
+    },
+    {
+      "epoch": 0.6531986531986532,
+      "grad_norm": 0.486328125,
+      "learning_rate": 9.245908891594818e-06,
+      "loss": 0.6645,
+      "step": 194
+    },
+    {
+      "epoch": 0.6565656565656566,
+      "grad_norm": 0.58984375,
+      "learning_rate": 9.236085864094178e-06,
+      "loss": 0.6745,
+      "step": 195
+    },
+    {
+      "epoch": 0.6599326599326599,
+      "grad_norm": 0.609375,
+      "learning_rate": 9.226204559594284e-06,
+      "loss": 0.7157,
+      "step": 196
+    },
+    {
+      "epoch": 0.6632996632996633,
+      "grad_norm": 0.59375,
+      "learning_rate": 9.216265114034964e-06,
+      "loss": 0.7298,
+      "step": 197
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.451171875,
+      "learning_rate": 9.206267664155906e-06,
+      "loss": 0.6439,
+      "step": 198
+    },
+    {
+      "epoch": 0.67003367003367,
+      "grad_norm": 0.447265625,
+      "learning_rate": 9.19621234749478e-06,
+      "loss": 0.6719,
+      "step": 199
+    },
+    {
+      "epoch": 0.6734006734006734,
+      "grad_norm": 0.41796875,
+      "learning_rate": 9.186099302385345e-06,
+      "loss": 0.6496,
+      "step": 200
+    },
+    {
+      "epoch": 0.6767676767676768,
+      "grad_norm": 0.482421875,
+      "learning_rate": 9.175928667955547e-06,
+      "loss": 0.709,
+      "step": 201
+    },
+    {
+      "epoch": 0.6801346801346801,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.1657005841256e-06,
+      "loss": 0.7054,
+      "step": 202
+    },
+    {
+      "epoch": 0.6835016835016835,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.155415191606074e-06,
+      "loss": 0.6941,
+      "step": 203
+    },
+    {
+      "epoch": 0.6868686868686869,
+      "grad_norm": 0.53515625,
+      "learning_rate": 9.145072631895942e-06,
+      "loss": 0.6919,
+      "step": 204
+    },
+    {
+      "epoch": 0.6902356902356902,
+      "grad_norm": 0.62890625,
+      "learning_rate": 9.134673047280644e-06,
+      "loss": 0.7098,
+      "step": 205
+    },
+    {
+      "epoch": 0.6936026936026936,
+      "grad_norm": 0.546875,
+      "learning_rate": 9.12421658083013e-06,
+      "loss": 0.6902,
+      "step": 206
+    },
+    {
+      "epoch": 0.696969696969697,
+      "grad_norm": 0.58984375,
+      "learning_rate": 9.113703376396885e-06,
+      "loss": 0.7228,
+      "step": 207
+    },
+    {
+      "epoch": 0.7003367003367004,
+      "grad_norm": 0.484375,
+      "learning_rate": 9.103133578613959e-06,
+      "loss": 0.6961,
+      "step": 208
+    },
+    {
+      "epoch": 0.7037037037037037,
+      "grad_norm": 0.515625,
+      "learning_rate": 9.092507332892968e-06,
+      "loss": 0.6689,
+      "step": 209
+    },
+    {
+      "epoch": 0.7070707070707071,
+      "grad_norm": 0.5390625,
+      "learning_rate": 9.081824785422099e-06,
+      "loss": 0.6552,
+      "step": 210
+    },
+    {
+      "epoch": 0.7104377104377104,
+      "grad_norm": 0.462890625,
+      "learning_rate": 9.071086083164099e-06,
+      "loss": 0.6739,
+      "step": 211
+    },
+    {
+      "epoch": 0.7138047138047138,
+      "grad_norm": 0.435546875,
+      "learning_rate": 9.060291373854252e-06,
+      "loss": 0.699,
+      "step": 212
+    },
+    {
+      "epoch": 0.7171717171717171,
+      "grad_norm": 0.49609375,
+      "learning_rate": 9.04944080599834e-06,
+      "loss": 0.6964,
+      "step": 213
+    },
+    {
+      "epoch": 0.7205387205387206,
+      "grad_norm": 0.47265625,
+      "learning_rate": 9.038534528870618e-06,
+      "loss": 0.678,
+      "step": 214
+    },
+    {
+      "epoch": 0.7239057239057239,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.027572692511739e-06,
+      "loss": 0.6847,
+      "step": 215
+    },
+    {
+      "epoch": 0.7272727272727273,
+      "grad_norm": 0.53515625,
+      "learning_rate": 9.016555447726704e-06,
+      "loss": 0.6811,
+      "step": 216
+    },
+    {
+      "epoch": 0.7306397306397306,
+      "grad_norm": 0.482421875,
+      "learning_rate": 9.005482946082784e-06,
+      "loss": 0.6831,
+      "step": 217
+    },
+    {
+      "epoch": 0.734006734006734,
+      "grad_norm": 0.474609375,
+      "learning_rate": 8.994355339907429e-06,
+      "loss": 0.6803,
+      "step": 218
+    },
+    {
+      "epoch": 0.7373737373737373,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.98317278228618e-06,
+      "loss": 0.6639,
+      "step": 219
+    },
+    {
+      "epoch": 0.7407407407407407,
+      "grad_norm": 0.474609375,
+      "learning_rate": 8.971935427060563e-06,
+      "loss": 0.6509,
+      "step": 220
+    },
+    {
+      "epoch": 0.7441077441077442,
+      "grad_norm": 0.47265625,
+      "learning_rate": 8.960643428825965e-06,
+      "loss": 0.647,
+      "step": 221
+    },
+    {
+      "epoch": 0.7474747474747475,
+      "grad_norm": 0.498046875,
+      "learning_rate": 8.949296942929515e-06,
+      "loss": 0.6686,
+      "step": 222
+    },
+    {
+      "epoch": 0.7508417508417509,
+      "grad_norm": 0.5078125,
+      "learning_rate": 8.937896125467941e-06,
+      "loss": 0.6512,
+      "step": 223
+    },
+    {
+      "epoch": 0.7542087542087542,
+      "grad_norm": 0.60546875,
+      "learning_rate": 8.92644113328543e-06,
+      "loss": 0.6544,
+      "step": 224
+    },
+    {
+      "epoch": 0.7575757575757576,
+      "grad_norm": 0.474609375,
+      "learning_rate": 8.91493212397146e-06,
+      "loss": 0.7091,
+      "step": 225
+    },
+    {
+      "epoch": 0.7609427609427609,
+      "grad_norm": 0.515625,
+      "learning_rate": 8.90336925585864e-06,
+      "loss": 0.6522,
+      "step": 226
+    },
+    {
+      "epoch": 0.7643097643097643,
+      "grad_norm": 0.4375,
+      "learning_rate": 8.891752688020532e-06,
+      "loss": 0.6923,
+      "step": 227
+    },
+    {
+      "epoch": 0.7676767676767676,
+      "grad_norm": 0.47265625,
+      "learning_rate": 8.880082580269454e-06,
+      "loss": 0.6873,
+      "step": 228
+    },
+    {
+      "epoch": 0.7676767676767676,
+      "eval_loss": 0.6816474795341492,
+      "eval_runtime": 8.7387,
+      "eval_samples_per_second": 57.217,
+      "eval_steps_per_second": 3.662,
+      "step": 228
+    },
+    {
+      "epoch": 0.7710437710437711,
+      "grad_norm": 0.466796875,
+      "learning_rate": 8.868359093154292e-06,
+      "loss": 0.6989,
+      "step": 229
+    },
+    {
+      "epoch": 0.7744107744107744,
+      "grad_norm": 0.4921875,
+      "learning_rate": 8.856582387958286e-06,
+      "loss": 0.6424,
+      "step": 230
+    },
+    {
+      "epoch": 0.7777777777777778,
+      "grad_norm": 0.546875,
+      "learning_rate": 8.84475262669681e-06,
+      "loss": 0.6477,
+      "step": 231
+    },
+    {
+      "epoch": 0.7811447811447811,
+      "grad_norm": 0.44921875,
+      "learning_rate": 8.832869972115148e-06,
+      "loss": 0.6623,
+      "step": 232
+    },
+    {
+      "epoch": 0.7845117845117845,
+      "grad_norm": 0.5625,
+      "learning_rate": 8.820934587686247e-06,
+      "loss": 0.6642,
+      "step": 233
+    },
+    {
+      "epoch": 0.7878787878787878,
+      "grad_norm": 0.470703125,
+      "learning_rate": 8.808946637608476e-06,
+      "loss": 0.6819,
+      "step": 234
+    },
+    {
+      "epoch": 0.7912457912457912,
+      "grad_norm": 0.470703125,
+      "learning_rate": 8.796906286803366e-06,
+      "loss": 0.7039,
+      "step": 235
+    },
+    {
+      "epoch": 0.7946127946127947,
+      "grad_norm": 0.51171875,
+      "learning_rate": 8.784813700913336e-06,
+      "loss": 0.7232,
+      "step": 236
+    },
+    {
+      "epoch": 0.797979797979798,
+      "grad_norm": 0.44140625,
+      "learning_rate": 8.77266904629942e-06,
+      "loss": 0.6687,
+      "step": 237
+    },
+    {
+      "epoch": 0.8013468013468014,
+      "grad_norm": 0.462890625,
+      "learning_rate": 8.760472490038974e-06,
+      "loss": 0.6664,
+      "step": 238
+    },
+    {
+      "epoch": 0.8047138047138047,
+      "grad_norm": 0.5234375,
+      "learning_rate": 8.748224199923378e-06,
+      "loss": 0.6752,
+      "step": 239
+    },
+    {
+      "epoch": 0.8080808080808081,
+      "grad_norm": 0.5078125,
+      "learning_rate": 8.735924344455732e-06,
+      "loss": 0.6373,
+      "step": 240
+    },
+    {
+      "epoch": 0.8114478114478114,
+      "grad_norm": 0.46484375,
+      "learning_rate": 8.723573092848535e-06,
+      "loss": 0.656,
+      "step": 241
+    },
+    {
+      "epoch": 0.8148148148148148,
+      "grad_norm": 0.46875,
+      "learning_rate": 8.71117061502135e-06,
+      "loss": 0.6747,
+      "step": 242
+    },
+    {
+      "epoch": 0.8181818181818182,
+      "grad_norm": 0.39453125,
+      "learning_rate": 8.698717081598484e-06,
+      "loss": 0.6878,
+      "step": 243
+    },
+    {
+      "epoch": 0.8215488215488216,
+      "grad_norm": 0.3984375,
+      "learning_rate": 8.68621266390662e-06,
+      "loss": 0.6381,
+      "step": 244
+    },
+    {
+      "epoch": 0.8249158249158249,
+      "grad_norm": 0.44140625,
+      "learning_rate": 8.673657533972477e-06,
+      "loss": 0.7127,
+      "step": 245
+    },
+    {
+      "epoch": 0.8282828282828283,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.661051864520434e-06,
+      "loss": 0.6888,
+      "step": 246
+    },
+    {
+      "epoch": 0.8316498316498316,
+      "grad_norm": 0.5078125,
+      "learning_rate": 8.64839582897015e-06,
+      "loss": 0.6444,
+      "step": 247
+    },
+    {
+      "epoch": 0.835016835016835,
+      "grad_norm": 0.42578125,
+      "learning_rate": 8.635689601434197e-06,
+      "loss": 0.6876,
+      "step": 248
+    },
+    {
+      "epoch": 0.8383838383838383,
+      "grad_norm": 0.6015625,
+      "learning_rate": 8.622933356715639e-06,
+      "loss": 0.7125,
+      "step": 249
+    },
+    {
+      "epoch": 0.8417508417508418,
+      "grad_norm": 0.51171875,
+      "learning_rate": 8.61012727030565e-06,
+      "loss": 0.6826,
+      "step": 250
+    },
+    {
+      "epoch": 0.8451178451178452,
+      "grad_norm": 0.435546875,
+      "learning_rate": 8.597271518381083e-06,
+      "loss": 0.6719,
+      "step": 251
+    },
+    {
+      "epoch": 0.8484848484848485,
+      "grad_norm": 0.609375,
+      "learning_rate": 8.584366277802057e-06,
+      "loss": 0.6723,
+      "step": 252
+    },
+    {
+      "epoch": 0.8518518518518519,
+      "grad_norm": 0.5546875,
+      "learning_rate": 8.571411726109518e-06,
+      "loss": 0.6899,
+      "step": 253
+    },
+    {
+      "epoch": 0.8552188552188552,
+      "grad_norm": 0.5078125,
+      "learning_rate": 8.558408041522801e-06,
+      "loss": 0.7042,
+      "step": 254
+    },
+    {
+      "epoch": 0.8585858585858586,
+      "grad_norm": 0.455078125,
+      "learning_rate": 8.545355402937174e-06,
+      "loss": 0.6429,
+      "step": 255
+    },
+    {
+      "epoch": 0.8619528619528619,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.532253989921377e-06,
+      "loss": 0.6586,
+      "step": 256
+    },
+    {
+      "epoch": 0.8653198653198653,
+      "grad_norm": 0.435546875,
+      "learning_rate": 8.519103982715158e-06,
+      "loss": 0.6212,
+      "step": 257
+    },
+    {
+      "epoch": 0.8686868686868687,
+      "grad_norm": 0.486328125,
+      "learning_rate": 8.505905562226784e-06,
+      "loss": 0.6421,
+      "step": 258
+    },
+    {
+      "epoch": 0.8720538720538721,
+      "grad_norm": 0.50390625,
+      "learning_rate": 8.492658910030558e-06,
+      "loss": 0.6051,
+      "step": 259
+    },
+    {
+      "epoch": 0.8754208754208754,
+      "grad_norm": 0.482421875,
+      "learning_rate": 8.479364208364322e-06,
+      "loss": 0.6454,
+      "step": 260
+    },
+    {
+      "epoch": 0.8787878787878788,
+      "grad_norm": 0.48046875,
+      "learning_rate": 8.466021640126946e-06,
+      "loss": 0.6579,
+      "step": 261
+    },
+    {
+      "epoch": 0.8821548821548821,
+      "grad_norm": 0.41015625,
+      "learning_rate": 8.452631388875814e-06,
+      "loss": 0.6046,
+      "step": 262
+    },
+    {
+      "epoch": 0.8855218855218855,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.439193638824296e-06,
+      "loss": 0.6481,
+      "step": 263
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.41796875,
+      "learning_rate": 8.425708574839221e-06,
+      "loss": 0.6418,
+      "step": 264
+    },
+    {
+      "epoch": 0.8922558922558923,
+      "grad_norm": 0.48046875,
+      "learning_rate": 8.412176382438327e-06,
+      "loss": 0.7079,
+      "step": 265
+    },
+    {
+      "epoch": 0.8956228956228957,
+      "grad_norm": 0.5078125,
+      "learning_rate": 8.398597247787706e-06,
+      "loss": 0.7014,
+      "step": 266
+    },
+    {
+      "epoch": 0.8956228956228957,
+      "eval_loss": 0.66867995262146,
+      "eval_runtime": 8.7476,
+      "eval_samples_per_second": 57.158,
+      "eval_steps_per_second": 3.658,
+      "step": 266
+    },
+    {
+      "epoch": 0.898989898989899,
+      "grad_norm": 0.404296875,
+      "learning_rate": 8.384971357699255e-06,
+      "loss": 0.631,
+      "step": 267
+    },
+    {
+      "epoch": 0.9023569023569024,
+      "grad_norm": 0.484375,
+      "learning_rate": 8.371298899628091e-06,
+      "loss": 0.691,
+      "step": 268
+    },
+    {
+      "epoch": 0.9057239057239057,
+      "grad_norm": 0.421875,
+      "learning_rate": 8.357580061669985e-06,
+      "loss": 0.6443,
+      "step": 269
+    },
+    {
+      "epoch": 0.9090909090909091,
+      "grad_norm": 0.4296875,
+      "learning_rate": 8.343815032558768e-06,
+      "loss": 0.6683,
+      "step": 270
+    },
+    {
+      "epoch": 0.9124579124579124,
+      "grad_norm": 0.43359375,
+      "learning_rate": 8.330004001663734e-06,
+      "loss": 0.6208,
+      "step": 271
+    },
+    {
+      "epoch": 0.9158249158249159,
+      "grad_norm": 0.5078125,
+      "learning_rate": 8.316147158987036e-06,
+      "loss": 0.6554,
+      "step": 272
+    },
+    {
+      "epoch": 0.9191919191919192,
+      "grad_norm": 0.484375,
+      "learning_rate": 8.30224469516107e-06,
+      "loss": 0.6884,
+      "step": 273
+    },
+    {
+      "epoch": 0.9225589225589226,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.288296801445863e-06,
+      "loss": 0.6633,
+      "step": 274
+    },
+    {
+      "epoch": 0.9259259259259259,
+      "grad_norm": 0.462890625,
+      "learning_rate": 8.274303669726427e-06,
+      "loss": 0.6521,
+      "step": 275
+    },
+    {
+      "epoch": 0.9292929292929293,
+      "grad_norm": 0.462890625,
+      "learning_rate": 8.260265492510121e-06,
+      "loss": 0.6429,
+      "step": 276
+    },
+    {
+      "epoch": 0.9326599326599326,
+      "grad_norm": 0.4765625,
+      "learning_rate": 8.246182462924022e-06,
+      "loss": 0.6604,
+      "step": 277
+    },
+    {
+      "epoch": 0.936026936026936,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.232054774712239e-06,
+      "loss": 0.6438,
+      "step": 278
+    },
+    {
+      "epoch": 0.9393939393939394,
+      "grad_norm": 0.4375,
+      "learning_rate": 8.217882622233268e-06,
+      "loss": 0.6319,
+      "step": 279
+    },
+    {
+      "epoch": 0.9427609427609428,
+      "grad_norm": 0.5,
+      "learning_rate": 8.203666200457312e-06,
+      "loss": 0.6349,
+      "step": 280
+    },
+    {
+      "epoch": 0.9461279461279462,
+      "grad_norm": 0.45703125,
+      "learning_rate": 8.1894057049636e-06,
+      "loss": 0.6657,
+      "step": 281
+    },
+    {
+      "epoch": 0.9494949494949495,
+      "grad_norm": 0.423828125,
+      "learning_rate": 8.175101331937692e-06,
+      "loss": 0.6866,
+      "step": 282
+    },
+    {
+      "epoch": 0.9528619528619529,
+      "grad_norm": 0.4296875,
+      "learning_rate": 8.160753278168787e-06,
+      "loss": 0.6544,
+      "step": 283
+    },
+    {
+      "epoch": 0.9562289562289562,
+      "grad_norm": 0.482421875,
+      "learning_rate": 8.146361741047007e-06,
+      "loss": 0.676,
+      "step": 284
+    },
+    {
+      "epoch": 0.9595959595959596,
+      "grad_norm": 0.416015625,
+      "learning_rate": 8.131926918560692e-06,
+      "loss": 0.6102,
+      "step": 285
+    },
+    {
+      "epoch": 0.9629629629629629,
+      "grad_norm": 0.474609375,
+      "learning_rate": 8.117449009293668e-06,
+      "loss": 0.6865,
+      "step": 286
+    },
+    {
+      "epoch": 0.9663299663299664,
+      "grad_norm": 0.423828125,
+      "learning_rate": 8.102928212422519e-06,
+      "loss": 0.6298,
+      "step": 287
+    },
+    {
+      "epoch": 0.9696969696969697,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.08836472771384e-06,
+      "loss": 0.6789,
+      "step": 288
+    },
+    {
+      "epoch": 0.9730639730639731,
+      "grad_norm": 0.4453125,
+      "learning_rate": 8.073758755521506e-06,
+      "loss": 0.6157,
+      "step": 289
+    },
+    {
+      "epoch": 0.9764309764309764,
+      "grad_norm": 0.404296875,
+      "learning_rate": 8.059110496783891e-06,
+      "loss": 0.6269,
+      "step": 290
+    },
+    {
+      "epoch": 0.9797979797979798,
+      "grad_norm": 0.427734375,
+      "learning_rate": 8.044420153021123e-06,
+      "loss": 0.6405,
+      "step": 291
+    },
+    {
+      "epoch": 0.9831649831649831,
+      "grad_norm": 0.46875,
+      "learning_rate": 8.02968792633231e-06,
+      "loss": 0.6657,
+      "step": 292
+    },
+    {
+      "epoch": 0.9865319865319865,
+      "grad_norm": 0.466796875,
+      "learning_rate": 8.014914019392743e-06,
+      "loss": 0.6373,
+      "step": 293
+    },
+    {
+      "epoch": 0.98989898989899,
+      "grad_norm": 0.59765625,
+      "learning_rate": 8.00009863545113e-06,
+      "loss": 0.6495,
+      "step": 294
+    },
+    {
+      "epoch": 0.9932659932659933,
+      "grad_norm": 0.478515625,
+      "learning_rate": 7.985241978326786e-06,
+      "loss": 0.6003,
+      "step": 295
+    },
+    {
+      "epoch": 0.9966329966329966,
+      "grad_norm": 0.40625,
+      "learning_rate": 7.970344252406832e-06,
+      "loss": 0.6543,
+      "step": 296
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6953125,
+      "learning_rate": 7.955405662643384e-06,
+      "loss": 0.6625,
+      "step": 297
+    },
+    {
+      "epoch": 1.0033670033670035,
+      "grad_norm": 0.412109375,
+      "learning_rate": 7.940426414550732e-06,
+      "loss": 0.6776,
+      "step": 298
+    },
+    {
+      "epoch": 1.0067340067340067,
+      "grad_norm": 0.419921875,
+      "learning_rate": 7.925406714202518e-06,
+      "loss": 0.6661,
+      "step": 299
+    },
+    {
+      "epoch": 1.0101010101010102,
+      "grad_norm": 0.4609375,
+      "learning_rate": 7.910346768228893e-06,
+      "loss": 0.6743,
+      "step": 300
+    },
+    {
+      "epoch": 1.0134680134680134,
+      "grad_norm": 0.41015625,
+      "learning_rate": 7.895246783813679e-06,
+      "loss": 0.6527,
+      "step": 301
+    },
+    {
+      "epoch": 1.0168350168350169,
+      "grad_norm": 0.462890625,
+      "learning_rate": 7.880106968691516e-06,
+      "loss": 0.6195,
+      "step": 302
+    },
+    {
+      "epoch": 1.02020202020202,
+      "grad_norm": 0.53125,
+      "learning_rate": 7.864927531145012e-06,
+      "loss": 0.5992,
+      "step": 303
+    },
+    {
+      "epoch": 1.0235690235690236,
+      "grad_norm": 0.376953125,
+      "learning_rate": 7.849708680001862e-06,
+      "loss": 0.6586,
+      "step": 304
+    },
+    {
+      "epoch": 1.0235690235690236,
+      "eval_loss": 0.6584768295288086,
+      "eval_runtime": 8.7375,
+      "eval_samples_per_second": 57.225,
+      "eval_steps_per_second": 3.662,
+      "step": 304
+    },
+    {
+      "epoch": 1.026936026936027,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.834450624631998e-06,
+      "loss": 0.6381,
+      "step": 305
+    },
+    {
+      "epoch": 1.0303030303030303,
+      "grad_norm": 0.388671875,
+      "learning_rate": 7.81915357494468e-06,
+      "loss": 0.6492,
+      "step": 306
+    },
+    {
+      "epoch": 1.0336700336700337,
+      "grad_norm": 0.419921875,
+      "learning_rate": 7.803817741385636e-06,
+      "loss": 0.6632,
+      "step": 307
+    },
+    {
+      "epoch": 1.037037037037037,
+      "grad_norm": 0.39453125,
+      "learning_rate": 7.788443334934148e-06,
+      "loss": 0.6424,
+      "step": 308
+    },
+    {
+      "epoch": 1.0404040404040404,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.773030567100158e-06,
+      "loss": 0.6309,
+      "step": 309
+    },
+    {
+      "epoch": 1.0437710437710437,
+      "grad_norm": 0.431640625,
+      "learning_rate": 7.757579649921354e-06,
+      "loss": 0.6469,
+      "step": 310
+    },
+    {
+      "epoch": 1.0471380471380471,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.742090795960258e-06,
+      "loss": 0.6216,
+      "step": 311
+    },
+    {
+      "epoch": 1.0505050505050506,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.726564218301298e-06,
+      "loss": 0.6249,
+      "step": 312
+    },
+    {
+      "epoch": 1.0538720538720538,
+      "grad_norm": 0.462890625,
+      "learning_rate": 7.711000130547875e-06,
+      "loss": 0.671,
+      "step": 313
+    },
+    {
+      "epoch": 1.0572390572390573,
+      "grad_norm": 0.451171875,
+      "learning_rate": 7.695398746819431e-06,
+      "loss": 0.6604,
+      "step": 314
+    },
+    {
+      "epoch": 1.0606060606060606,
+      "grad_norm": 0.40234375,
+      "learning_rate": 7.679760281748492e-06,
+      "loss": 0.626,
+      "step": 315
+    },
+    {
+      "epoch": 1.063973063973064,
+      "grad_norm": 0.369140625,
+      "learning_rate": 7.664084950477732e-06,
+      "loss": 0.6336,
+      "step": 316
+    },
+    {
+      "epoch": 1.0673400673400673,
+      "grad_norm": 0.421875,
+      "learning_rate": 7.648372968656995e-06,
+      "loss": 0.6813,
+      "step": 317
+    },
+    {
+      "epoch": 1.0707070707070707,
+      "grad_norm": 0.515625,
+      "learning_rate": 7.632624552440337e-06,
+      "loss": 0.6539,
+      "step": 318
+    },
+    {
+      "epoch": 1.074074074074074,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.616839918483061e-06,
+      "loss": 0.6378,
+      "step": 319
+    },
+    {
+      "epoch": 1.0774410774410774,
+      "grad_norm": 0.421875,
+      "learning_rate": 7.601019283938718e-06,
+      "loss": 0.7115,
+      "step": 320
+    },
+    {
+      "epoch": 1.0808080808080809,
+      "grad_norm": 0.443359375,
+      "learning_rate": 7.58516286645613e-06,
+      "loss": 0.6628,
+      "step": 321
+    },
+    {
+      "epoch": 1.0841750841750841,
+      "grad_norm": 0.52734375,
+      "learning_rate": 7.569270884176401e-06,
+      "loss": 0.633,
+      "step": 322
+    },
+    {
+      "epoch": 1.0875420875420876,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.553343555729904e-06,
+      "loss": 0.6339,
+      "step": 323
+    },
+    {
+      "epoch": 1.0909090909090908,
+      "grad_norm": 0.40625,
+      "learning_rate": 7.5373811002332785e-06,
+      "loss": 0.6557,
+      "step": 324
+    },
+    {
+      "epoch": 1.0942760942760943,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.521383737286423e-06,
+      "loss": 0.663,
+      "step": 325
+    },
+    {
+      "epoch": 1.0976430976430978,
+      "grad_norm": 0.451171875,
+      "learning_rate": 7.505351686969458e-06,
+      "loss": 0.6663,
+      "step": 326
+    },
+    {
+      "epoch": 1.101010101010101,
+      "grad_norm": 0.400390625,
+      "learning_rate": 7.4892851698397174e-06,
+      "loss": 0.6393,
+      "step": 327
+    },
+    {
+      "epoch": 1.1043771043771045,
+      "grad_norm": 0.435546875,
+      "learning_rate": 7.4731844069286965e-06,
+      "loss": 0.6323,
+      "step": 328
+    },
+    {
+      "epoch": 1.1077441077441077,
+      "grad_norm": 0.458984375,
+      "learning_rate": 7.457049619739023e-06,
+      "loss": 0.6084,
+      "step": 329
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.440881030241407e-06,
+      "loss": 0.6615,
+      "step": 330
+    },
+    {
+      "epoch": 1.1144781144781144,
+      "grad_norm": 0.51171875,
+      "learning_rate": 7.424678860871584e-06,
+      "loss": 0.6438,
+      "step": 331
+    },
+    {
+      "epoch": 1.1178451178451179,
+      "grad_norm": 0.462890625,
+      "learning_rate": 7.408443334527257e-06,
+      "loss": 0.6653,
+      "step": 332
+    },
+    {
+      "epoch": 1.121212121212121,
+      "grad_norm": 0.4140625,
+      "learning_rate": 7.3921746745650315e-06,
+      "loss": 0.6286,
+      "step": 333
+    },
+    {
+      "epoch": 1.1245791245791246,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.3758731047973416e-06,
+      "loss": 0.6356,
+      "step": 334
+    },
+    {
+      "epoch": 1.127946127946128,
+      "grad_norm": 0.416015625,
+      "learning_rate": 7.3595388494893684e-06,
+      "loss": 0.6423,
+      "step": 335
+    },
+    {
+      "epoch": 1.1313131313131313,
+      "grad_norm": 0.412109375,
+      "learning_rate": 7.343172133355958e-06,
+      "loss": 0.6313,
+      "step": 336
+    },
+    {
+      "epoch": 1.1346801346801347,
+      "grad_norm": 0.41015625,
+      "learning_rate": 7.326773181558532e-06,
+      "loss": 0.637,
+      "step": 337
+    },
+    {
+      "epoch": 1.138047138047138,
+      "grad_norm": 0.361328125,
+      "learning_rate": 7.310342219701981e-06,
+      "loss": 0.6153,
+      "step": 338
+    },
+    {
+      "epoch": 1.1414141414141414,
+      "grad_norm": 0.388671875,
+      "learning_rate": 7.293879473831573e-06,
+      "loss": 0.6582,
+      "step": 339
+    },
+    {
+      "epoch": 1.144781144781145,
+      "grad_norm": 0.45703125,
+      "learning_rate": 7.2773851704298304e-06,
+      "loss": 0.6364,
+      "step": 340
+    },
+    {
+      "epoch": 1.1481481481481481,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.260859536413429e-06,
+      "loss": 0.6126,
+      "step": 341
+    },
+    {
+      "epoch": 1.1515151515151516,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.244302799130064e-06,
+      "loss": 0.6532,
+      "step": 342
+    },
+    {
+      "epoch": 1.1515151515151516,
+      "eval_loss": 0.6510886549949646,
+      "eval_runtime": 8.7236,
+      "eval_samples_per_second": 57.316,
+      "eval_steps_per_second": 3.668,
+      "step": 342
+    },
+    {
+      "epoch": 1.1548821548821548,
+      "grad_norm": 0.38671875,
+      "learning_rate": 7.227715186355326e-06,
+      "loss": 0.6067,
+      "step": 343
+    },
+    {
+      "epoch": 1.1582491582491583,
+      "grad_norm": 0.458984375,
+      "learning_rate": 7.211096926289572e-06,
+      "loss": 0.6796,
+      "step": 344
+    },
+    {
+      "epoch": 1.1616161616161615,
+      "grad_norm": 0.443359375,
+      "learning_rate": 7.19444824755478e-06,
+      "loss": 0.6242,
+      "step": 345
+    },
+    {
+      "epoch": 1.164983164983165,
+      "grad_norm": 0.4296875,
+      "learning_rate": 7.1777693791914025e-06,
+      "loss": 0.6288,
+      "step": 346
+    },
+    {
+      "epoch": 1.1683501683501682,
+      "grad_norm": 0.447265625,
+      "learning_rate": 7.161060550655227e-06,
+      "loss": 0.6908,
+      "step": 347
+    },
+    {
+      "epoch": 1.1717171717171717,
+      "grad_norm": 0.45703125,
+      "learning_rate": 7.144321991814205e-06,
+      "loss": 0.6809,
+      "step": 348
+    },
+    {
+      "epoch": 1.1750841750841752,
+      "grad_norm": 0.4140625,
+      "learning_rate": 7.1275539329452984e-06,
+      "loss": 0.6421,
+      "step": 349
+    },
+    {
+      "epoch": 1.1784511784511784,
+      "grad_norm": 0.400390625,
+      "learning_rate": 7.110756604731307e-06,
+      "loss": 0.6613,
+      "step": 350
+    },
+    {
+      "epoch": 1.1818181818181819,
+      "grad_norm": 0.419921875,
+      "learning_rate": 7.0939302382577e-06,
+      "loss": 0.6596,
+      "step": 351
+    },
+    {
+      "epoch": 1.1851851851851851,
+      "grad_norm": 0.40234375,
+      "learning_rate": 7.0770750650094335e-06,
+      "loss": 0.6539,
+      "step": 352
+    },
+    {
+      "epoch": 1.1885521885521886,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.060191316867764e-06,
+      "loss": 0.6451,
+      "step": 353
+    },
+    {
+      "epoch": 1.1919191919191918,
+      "grad_norm": 0.466796875,
+      "learning_rate": 7.0432792261070605e-06,
+      "loss": 0.6211,
+      "step": 354
+    },
+    {
+      "epoch": 1.1952861952861953,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.026339025391615e-06,
+      "loss": 0.6458,
+      "step": 355
+    },
+    {
+      "epoch": 1.1986531986531987,
+      "grad_norm": 0.486328125,
+      "learning_rate": 7.00937094777243e-06,
+      "loss": 0.6579,
+      "step": 356
+    },
+    {
+      "epoch": 1.202020202020202,
+      "grad_norm": 0.4140625,
+      "learning_rate": 6.992375226684016e-06,
+      "loss": 0.6337,
+      "step": 357
+    },
+    {
+      "epoch": 1.2053872053872055,
+      "grad_norm": 0.466796875,
+      "learning_rate": 6.975352095941194e-06,
+      "loss": 0.671,
+      "step": 358
+    },
+    {
+      "epoch": 1.2087542087542087,
+      "grad_norm": 0.455078125,
+      "learning_rate": 6.958301789735853e-06,
+      "loss": 0.6332,
+      "step": 359
+    },
+    {
+      "epoch": 1.2121212121212122,
+      "grad_norm": 0.484375,
+      "learning_rate": 6.941224542633748e-06,
+      "loss": 0.6147,
+      "step": 360
+    },
+    {
+      "epoch": 1.2154882154882154,
+      "grad_norm": 0.3984375,
+      "learning_rate": 6.924120589571275e-06,
+      "loss": 0.6628,
+      "step": 361
+    },
+    {
+      "epoch": 1.2188552188552189,
+      "grad_norm": 0.375,
+      "learning_rate": 6.906990165852218e-06,
+      "loss": 0.6444,
+      "step": 362
+    },
+    {
+      "epoch": 1.2222222222222223,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.889833507144534e-06,
+      "loss": 0.6607,
+      "step": 363
+    },
+    {
+      "epoch": 1.2255892255892256,
+      "grad_norm": 0.4765625,
+      "learning_rate": 6.872650849477098e-06,
+      "loss": 0.669,
+      "step": 364
+    },
+    {
+      "epoch": 1.228956228956229,
+      "grad_norm": 0.462890625,
+      "learning_rate": 6.855442429236463e-06,
+      "loss": 0.6476,
+      "step": 365
+    },
+    {
+      "epoch": 1.2323232323232323,
+      "grad_norm": 0.4140625,
+      "learning_rate": 6.838208483163601e-06,
+      "loss": 0.6087,
+      "step": 366
+    },
+    {
+      "epoch": 1.2356902356902357,
+      "grad_norm": 0.50390625,
+      "learning_rate": 6.820949248350653e-06,
+      "loss": 0.6628,
+      "step": 367
+    },
+    {
+      "epoch": 1.239057239057239,
+      "grad_norm": 0.5859375,
+      "learning_rate": 6.803664962237665e-06,
+      "loss": 0.6354,
+      "step": 368
+    },
+    {
+      "epoch": 1.2424242424242424,
+      "grad_norm": 0.4296875,
+      "learning_rate": 6.786355862609315e-06,
+      "loss": 0.6096,
+      "step": 369
+    },
+    {
+      "epoch": 1.2457912457912457,
+      "grad_norm": 0.375,
+      "learning_rate": 6.769022187591654e-06,
+      "loss": 0.5832,
+      "step": 370
+    },
+    {
+      "epoch": 1.2491582491582491,
+      "grad_norm": 0.390625,
+      "learning_rate": 6.751664175648817e-06,
+      "loss": 0.6322,
+      "step": 371
+    },
+    {
+      "epoch": 1.2525252525252526,
+      "grad_norm": 0.43359375,
+      "learning_rate": 6.734282065579757e-06,
+      "loss": 0.6183,
+      "step": 372
+    },
+    {
+      "epoch": 1.2558922558922558,
+      "grad_norm": 0.41796875,
+      "learning_rate": 6.716876096514944e-06,
+      "loss": 0.6339,
+      "step": 373
+    },
+    {
+      "epoch": 1.2592592592592593,
+      "grad_norm": 0.439453125,
+      "learning_rate": 6.699446507913083e-06,
+      "loss": 0.6411,
+      "step": 374
+    },
+    {
+      "epoch": 1.2626262626262625,
+      "grad_norm": 0.40625,
+      "learning_rate": 6.681993539557828e-06,
+      "loss": 0.6566,
+      "step": 375
+    },
+    {
+      "epoch": 1.265993265993266,
+      "grad_norm": 0.392578125,
+      "learning_rate": 6.664517431554466e-06,
+      "loss": 0.6109,
+      "step": 376
+    },
+    {
+      "epoch": 1.2693602693602695,
+      "grad_norm": 0.37109375,
+      "learning_rate": 6.6470184243266235e-06,
+      "loss": 0.6523,
+      "step": 377
+    },
+    {
+      "epoch": 1.2727272727272727,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.6294967586129614e-06,
+      "loss": 0.6139,
+      "step": 378
+    },
+    {
+      "epoch": 1.2760942760942762,
+      "grad_norm": 0.42578125,
+      "learning_rate": 6.611952675463859e-06,
+      "loss": 0.6616,
+      "step": 379
+    },
+    {
+      "epoch": 1.2794612794612794,
+      "grad_norm": 0.455078125,
+      "learning_rate": 6.594386416238095e-06,
+      "loss": 0.6334,
+      "step": 380
+    },
+    {
+      "epoch": 1.2794612794612794,
+      "eval_loss": 0.6462748050689697,
+      "eval_runtime": 8.7236,
+      "eval_samples_per_second": 57.316,
+      "eval_steps_per_second": 3.668,
+      "step": 380
+    },
+    {
+      "epoch": 1.2828282828282829,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.576798222599532e-06,
+      "loss": 0.6525,
+      "step": 381
+    },
+    {
+      "epoch": 1.2861952861952861,
+      "grad_norm": 0.412109375,
+      "learning_rate": 6.559188336513794e-06,
+      "loss": 0.6356,
+      "step": 382
+    },
+    {
+      "epoch": 1.2895622895622896,
+      "grad_norm": 0.396484375,
+      "learning_rate": 6.5415570002449315e-06,
+      "loss": 0.6291,
+      "step": 383
+    },
+    {
+      "epoch": 1.2929292929292928,
+      "grad_norm": 0.5546875,
+      "learning_rate": 6.523904456352089e-06,
+      "loss": 0.6585,
+      "step": 384
+    },
+    {
+      "epoch": 1.2962962962962963,
+      "grad_norm": 0.37890625,
+      "learning_rate": 6.5062309476861714e-06,
+      "loss": 0.632,
+      "step": 385
+    },
+    {
+      "epoch": 1.2996632996632997,
+      "grad_norm": 0.349609375,
+      "learning_rate": 6.488536717386504e-06,
+      "loss": 0.6063,
+      "step": 386
+    },
+    {
+      "epoch": 1.303030303030303,
+      "grad_norm": 0.46484375,
+      "learning_rate": 6.470822008877482e-06,
+      "loss": 0.6473,
+      "step": 387
+    },
+    {
+      "epoch": 1.3063973063973064,
+      "grad_norm": 0.396484375,
+      "learning_rate": 6.453087065865229e-06,
+      "loss": 0.6384,
+      "step": 388
+    },
+    {
+      "epoch": 1.3097643097643097,
+      "grad_norm": 0.423828125,
+      "learning_rate": 6.435332132334233e-06,
+      "loss": 0.6375,
+      "step": 389
+    },
+    {
+      "epoch": 1.3131313131313131,
+      "grad_norm": 0.421875,
+      "learning_rate": 6.4175574525440065e-06,
+      "loss": 0.6372,
+      "step": 390
+    },
+    {
+      "epoch": 1.3164983164983166,
+      "grad_norm": 0.447265625,
+      "learning_rate": 6.3997632710257095e-06,
+      "loss": 0.6345,
+      "step": 391
+    },
+    {
+      "epoch": 1.3198653198653199,
+      "grad_norm": 0.490234375,
+      "learning_rate": 6.381949832578796e-06,
+      "loss": 0.6516,
+      "step": 392
+    },
+    {
+      "epoch": 1.3232323232323233,
+      "grad_norm": 0.416015625,
+      "learning_rate": 6.364117382267642e-06,
+      "loss": 0.6354,
+      "step": 393
+    },
+    {
+      "epoch": 1.3265993265993266,
+      "grad_norm": 0.390625,
+      "learning_rate": 6.346266165418173e-06,
+      "loss": 0.615,
+      "step": 394
+    },
+    {
+      "epoch": 1.32996632996633,
+      "grad_norm": 0.46484375,
+      "learning_rate": 6.328396427614496e-06,
+      "loss": 0.6414,
+      "step": 395
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.37890625,
+      "learning_rate": 6.310508414695511e-06,
+      "loss": 0.6424,
+      "step": 396
+    },
+    {
+      "epoch": 1.3367003367003367,
+      "grad_norm": 0.4140625,
+      "learning_rate": 6.292602372751536e-06,
+      "loss": 0.6381,
+      "step": 397
+    },
+    {
+      "epoch": 1.34006734006734,
+      "grad_norm": 0.369140625,
+      "learning_rate": 6.274678548120921e-06,
+      "loss": 0.6542,
+      "step": 398
+    },
+    {
+      "epoch": 1.3434343434343434,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.25673718738665e-06,
+      "loss": 0.6461,
+      "step": 399
+    },
+    {
+      "epoch": 1.3468013468013469,
+      "grad_norm": 0.37109375,
+      "learning_rate": 6.238778537372966e-06,
+      "loss": 0.6367,
+      "step": 400
+    },
+    {
+      "epoch": 1.3501683501683501,
+      "grad_norm": 0.369140625,
+      "learning_rate": 6.2208028451419575e-06,
+      "loss": 0.6461,
+      "step": 401
+    },
+    {
+      "epoch": 1.3535353535353536,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.2028103579901725e-06,
+      "loss": 0.6532,
+      "step": 402
+    },
+    {
+      "epoch": 1.3569023569023568,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.184801323445208e-06,
+      "loss": 0.629,
+      "step": 403
+    },
+    {
+      "epoch": 1.3602693602693603,
+      "grad_norm": 0.390625,
+      "learning_rate": 6.166775989262309e-06,
+      "loss": 0.6723,
+      "step": 404
+    },
+    {
+      "epoch": 1.3636363636363638,
+      "grad_norm": 0.4296875,
+      "learning_rate": 6.1487346034209605e-06,
+      "loss": 0.6587,
+      "step": 405
+    },
+    {
+      "epoch": 1.367003367003367,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.130677414121474e-06,
+      "loss": 0.6294,
+      "step": 406
+    },
+    {
+      "epoch": 1.3703703703703702,
+      "grad_norm": 0.439453125,
+      "learning_rate": 6.112604669781572e-06,
+      "loss": 0.6628,
+      "step": 407
+    },
+    {
+      "epoch": 1.3737373737373737,
+      "grad_norm": 0.439453125,
+      "learning_rate": 6.094516619032975e-06,
+      "loss": 0.6444,
+      "step": 408
+    },
+    {
+      "epoch": 1.3771043771043772,
+      "grad_norm": 0.4609375,
+      "learning_rate": 6.076413510717973e-06,
+      "loss": 0.6869,
+      "step": 409
+    },
+    {
+      "epoch": 1.3804713804713804,
+      "grad_norm": 0.421875,
+      "learning_rate": 6.058295593886013e-06,
+      "loss": 0.638,
+      "step": 410
+    },
+    {
+      "epoch": 1.3838383838383839,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.040163117790261e-06,
+      "loss": 0.6067,
+      "step": 411
+    },
+    {
+      "epoch": 1.387205387205387,
+      "grad_norm": 0.60546875,
+      "learning_rate": 6.022016331884185e-06,
+      "loss": 0.6231,
+      "step": 412
+    },
+    {
+      "epoch": 1.3905723905723906,
+      "grad_norm": 0.5,
+      "learning_rate": 6.003855485818111e-06,
+      "loss": 0.6267,
+      "step": 413
+    },
+    {
+      "epoch": 1.393939393939394,
+      "grad_norm": 0.4375,
+      "learning_rate": 5.985680829435798e-06,
+      "loss": 0.6307,
+      "step": 414
+    },
+    {
+      "epoch": 1.3973063973063973,
+      "grad_norm": 0.421875,
+      "learning_rate": 5.967492612770999e-06,
+      "loss": 0.647,
+      "step": 415
+    },
+    {
+      "epoch": 1.4006734006734007,
+      "grad_norm": 0.458984375,
+      "learning_rate": 5.9492910860440135e-06,
+      "loss": 0.6344,
+      "step": 416
+    },
+    {
+      "epoch": 1.404040404040404,
+      "grad_norm": 0.419921875,
+      "learning_rate": 5.931076499658258e-06,
+      "loss": 0.646,
+      "step": 417
+    },
+    {
+      "epoch": 1.4074074074074074,
+      "grad_norm": 0.3984375,
+      "learning_rate": 5.91284910419681e-06,
+      "loss": 0.5968,
+      "step": 418
+    },
+    {
+      "epoch": 1.4074074074074074,
+      "eval_loss": 0.6433593034744263,
+      "eval_runtime": 8.8065,
+      "eval_samples_per_second": 56.776,
+      "eval_steps_per_second": 3.634,
+      "step": 418
+    },
+    {
+      "epoch": 1.410774410774411,
+      "grad_norm": 0.40625,
+      "learning_rate": 5.894609150418964e-06,
+      "loss": 0.6348,
+      "step": 419
+    },
+    {
+      "epoch": 1.4141414141414141,
+      "grad_norm": 0.439453125,
+      "learning_rate": 5.876356889256788e-06,
+      "loss": 0.6255,
+      "step": 420
+    },
+    {
+      "epoch": 1.4175084175084174,
+      "grad_norm": 0.359375,
+      "learning_rate": 5.858092571811661e-06,
+      "loss": 0.6316,
+      "step": 421
+    },
+    {
+      "epoch": 1.4208754208754208,
+      "grad_norm": 0.36328125,
+      "learning_rate": 5.839816449350824e-06,
+      "loss": 0.627,
+      "step": 422
+    },
+    {
+      "epoch": 1.4242424242424243,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.821528773303929e-06,
+      "loss": 0.6571,
+      "step": 423
+    },
+    {
+      "epoch": 1.4276094276094276,
+      "grad_norm": 0.37109375,
+      "learning_rate": 5.803229795259566e-06,
+      "loss": 0.6146,
+      "step": 424
+    },
+    {
+      "epoch": 1.430976430976431,
+      "grad_norm": 0.39453125,
+      "learning_rate": 5.784919766961813e-06,
+      "loss": 0.6388,
+      "step": 425
+    },
+    {
+      "epoch": 1.4343434343434343,
+      "grad_norm": 0.431640625,
+      "learning_rate": 5.76659894030677e-06,
+      "loss": 0.6474,
+      "step": 426
+    },
+    {
+      "epoch": 1.4377104377104377,
+      "grad_norm": 0.4296875,
+      "learning_rate": 5.748267567339093e-06,
+      "loss": 0.6553,
+      "step": 427
+    },
+    {
+      "epoch": 1.4410774410774412,
+      "grad_norm": 0.361328125,
+      "learning_rate": 5.729925900248524e-06,
+      "loss": 0.6037,
+      "step": 428
+    },
+    {
+      "epoch": 1.4444444444444444,
+      "grad_norm": 0.373046875,
+      "learning_rate": 5.711574191366427e-06,
+      "loss": 0.5933,
+      "step": 429
+    },
+    {
+      "epoch": 1.4478114478114479,
+      "grad_norm": 0.46484375,
+      "learning_rate": 5.693212693162311e-06,
+      "loss": 0.676,
+      "step": 430
+    },
+    {
+      "epoch": 1.4511784511784511,
+      "grad_norm": 0.369140625,
+      "learning_rate": 5.674841658240365e-06,
+      "loss": 0.6578,
+      "step": 431
+    },
+    {
+      "epoch": 1.4545454545454546,
+      "grad_norm": 0.380859375,
+      "learning_rate": 5.656461339335968e-06,
+      "loss": 0.6329,
+      "step": 432
+    },
+    {
+      "epoch": 1.457912457912458,
+      "grad_norm": 0.392578125,
+      "learning_rate": 5.638071989312228e-06,
+      "loss": 0.6635,
+      "step": 433
+    },
+    {
+      "epoch": 1.4612794612794613,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.619673861156498e-06,
+      "loss": 0.6239,
+      "step": 434
+    },
+    {
+      "epoch": 1.4646464646464645,
+      "grad_norm": 0.34765625,
+      "learning_rate": 5.60126720797689e-06,
+      "loss": 0.6189,
+      "step": 435
+    },
+    {
+      "epoch": 1.468013468013468,
+      "grad_norm": 0.4296875,
+      "learning_rate": 5.5828522829987965e-06,
+      "loss": 0.6171,
+      "step": 436
+    },
+    {
+      "epoch": 1.4713804713804715,
+      "grad_norm": 0.53125,
+      "learning_rate": 5.564429339561411e-06,
+      "loss": 0.6071,
+      "step": 437
+    },
+    {
+      "epoch": 1.4747474747474747,
+      "grad_norm": 0.54296875,
+      "learning_rate": 5.5459986311142365e-06,
+      "loss": 0.6652,
+      "step": 438
+    },
+    {
+      "epoch": 1.4781144781144782,
+      "grad_norm": 0.3671875,
+      "learning_rate": 5.527560411213604e-06,
+      "loss": 0.5801,
+      "step": 439
+    },
+    {
+      "epoch": 1.4814814814814814,
+      "grad_norm": 0.412109375,
+      "learning_rate": 5.509114933519179e-06,
+      "loss": 0.6077,
+      "step": 440
+    },
+    {
+      "epoch": 1.4848484848484849,
+      "grad_norm": 0.3515625,
+      "learning_rate": 5.490662451790472e-06,
+      "loss": 0.6046,
+      "step": 441
+    },
+    {
+      "epoch": 1.4882154882154883,
+      "grad_norm": 0.4609375,
+      "learning_rate": 5.4722032198833595e-06,
+      "loss": 0.6483,
+      "step": 442
+    },
+    {
+      "epoch": 1.4915824915824916,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.453737491746572e-06,
+      "loss": 0.6694,
+      "step": 443
+    },
+    {
+      "epoch": 1.494949494949495,
+      "grad_norm": 0.490234375,
+      "learning_rate": 5.435265521418215e-06,
+      "loss": 0.6048,
+      "step": 444
+    },
+    {
+      "epoch": 1.4983164983164983,
+      "grad_norm": 0.369140625,
+      "learning_rate": 5.41678756302227e-06,
+      "loss": 0.6167,
+      "step": 445
+    },
+    {
+      "epoch": 1.5016835016835017,
+      "grad_norm": 0.38671875,
+      "learning_rate": 5.3983038707650955e-06,
+      "loss": 0.6296,
+      "step": 446
+    },
+    {
+      "epoch": 1.5050505050505052,
+      "grad_norm": 0.392578125,
+      "learning_rate": 5.379814698931935e-06,
+      "loss": 0.6524,
+      "step": 447
+    },
+    {
+      "epoch": 1.5084175084175084,
+      "grad_norm": 0.4375,
+      "learning_rate": 5.361320301883413e-06,
+      "loss": 0.6326,
+      "step": 448
+    },
+    {
+      "epoch": 1.5117845117845117,
+      "grad_norm": 0.54296875,
+      "learning_rate": 5.342820934052042e-06,
+      "loss": 0.6675,
+      "step": 449
+    },
+    {
+      "epoch": 1.5151515151515151,
+      "grad_norm": 0.451171875,
+      "learning_rate": 5.324316849938715e-06,
+      "loss": 0.6346,
+      "step": 450
+    },
+    {
+      "epoch": 1.5185185185185186,
+      "grad_norm": 0.416015625,
+      "learning_rate": 5.3058083041092145e-06,
+      "loss": 0.6381,
+      "step": 451
+    },
+    {
+      "epoch": 1.5218855218855218,
+      "grad_norm": 0.37109375,
+      "learning_rate": 5.2872955511906974e-06,
+      "loss": 0.6551,
+      "step": 452
+    },
+    {
+      "epoch": 1.5252525252525253,
+      "grad_norm": 0.384765625,
+      "learning_rate": 5.268778845868201e-06,
+      "loss": 0.6186,
+      "step": 453
+    },
+    {
+      "epoch": 1.5286195286195285,
+      "grad_norm": 0.5,
+      "learning_rate": 5.250258442881139e-06,
+      "loss": 0.6483,
+      "step": 454
+    },
+    {
+      "epoch": 1.531986531986532,
+      "grad_norm": 0.42578125,
+      "learning_rate": 5.231734597019792e-06,
+      "loss": 0.6584,
+      "step": 455
+    },
+    {
+      "epoch": 1.5353535353535355,
+      "grad_norm": 0.36328125,
+      "learning_rate": 5.2132075631218075e-06,
+      "loss": 0.6366,
+      "step": 456
+    },
+    {
+      "epoch": 1.5353535353535355,
+      "eval_loss": 0.6413553357124329,
+      "eval_runtime": 8.8625,
+      "eval_samples_per_second": 56.418,
+      "eval_steps_per_second": 3.611,
+      "step": 456
+    },
+    {
+      "epoch": 1.5387205387205387,
+      "grad_norm": 0.421875,
+      "learning_rate": 5.194677596068689e-06,
+      "loss": 0.6116,
+      "step": 457
+    },
+    {
+      "epoch": 1.542087542087542,
+      "grad_norm": 0.359375,
+      "learning_rate": 5.176144950782296e-06,
+      "loss": 0.6505,
+      "step": 458
+    },
+    {
+      "epoch": 1.5454545454545454,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.1576098822213285e-06,
+      "loss": 0.6524,
+      "step": 459
+    },
+    {
+      "epoch": 1.5488215488215489,
+      "grad_norm": 0.361328125,
+      "learning_rate": 5.139072645377829e-06,
+      "loss": 0.5802,
+      "step": 460
+    },
+    {
+      "epoch": 1.5521885521885523,
+      "grad_norm": 0.384765625,
+      "learning_rate": 5.120533495273664e-06,
+      "loss": 0.6071,
+      "step": 461
+    },
+    {
+      "epoch": 1.5555555555555556,
+      "grad_norm": 0.3671875,
+      "learning_rate": 5.101992686957028e-06,
+      "loss": 0.6279,
+      "step": 462
+    },
+    {
+      "epoch": 1.5589225589225588,
+      "grad_norm": 0.57421875,
+      "learning_rate": 5.0834504754989225e-06,
+      "loss": 0.6595,
+      "step": 463
+    },
+    {
+      "epoch": 1.5622895622895623,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.064907115989655e-06,
+      "loss": 0.6813,
+      "step": 464
+    },
+    {
+      "epoch": 1.5656565656565657,
+      "grad_norm": 0.46484375,
+      "learning_rate": 5.046362863535327e-06,
+      "loss": 0.6255,
+      "step": 465
+    },
+    {
+      "epoch": 1.569023569023569,
+      "grad_norm": 0.412109375,
+      "learning_rate": 5.0278179732543244e-06,
+      "loss": 0.6627,
+      "step": 466
+    },
+    {
+      "epoch": 1.5723905723905722,
+      "grad_norm": 0.44921875,
+      "learning_rate": 5.009272700273804e-06,
+      "loss": 0.6293,
+      "step": 467
+    },
+    {
+      "epoch": 1.5757575757575757,
+      "grad_norm": 0.373046875,
+      "learning_rate": 4.990727299726196e-06,
+      "loss": 0.6171,
+      "step": 468
+    },
+    {
+      "epoch": 1.5791245791245792,
+      "grad_norm": 0.39453125,
+      "learning_rate": 4.972182026745678e-06,
+      "loss": 0.6408,
+      "step": 469
+    },
+    {
+      "epoch": 1.5824915824915826,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.953637136464675e-06,
+      "loss": 0.597,
+      "step": 470
+    },
+    {
+      "epoch": 1.5858585858585859,
+      "grad_norm": 0.375,
+      "learning_rate": 4.935092884010347e-06,
+      "loss": 0.6075,
+      "step": 471
+    },
+    {
+      "epoch": 1.589225589225589,
+      "grad_norm": 0.390625,
+      "learning_rate": 4.91654952450108e-06,
+      "loss": 0.6153,
+      "step": 472
+    },
+    {
+      "epoch": 1.5925925925925926,
+      "grad_norm": 0.3515625,
+      "learning_rate": 4.898007313042975e-06,
+      "loss": 0.6299,
+      "step": 473
+    },
+    {
+      "epoch": 1.595959595959596,
+      "grad_norm": 0.359375,
+      "learning_rate": 4.879466504726337e-06,
+      "loss": 0.6415,
+      "step": 474
+    },
+    {
+      "epoch": 1.5993265993265995,
+      "grad_norm": 0.40625,
+      "learning_rate": 4.860927354622173e-06,
+      "loss": 0.646,
+      "step": 475
+    },
+    {
+      "epoch": 1.6026936026936027,
+      "grad_norm": 0.396484375,
+      "learning_rate": 4.842390117778673e-06,
+      "loss": 0.6473,
+      "step": 476
+    },
+    {
+      "epoch": 1.606060606060606,
+      "grad_norm": 0.419921875,
+      "learning_rate": 4.8238550492177065e-06,
+      "loss": 0.6569,
+      "step": 477
+    },
+    {
+      "epoch": 1.6094276094276094,
+      "grad_norm": 0.50390625,
+      "learning_rate": 4.805322403931312e-06,
+      "loss": 0.6664,
+      "step": 478
+    },
+    {
+      "epoch": 1.612794612794613,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.786792436878193e-06,
+      "loss": 0.6481,
+      "step": 479
+    },
+    {
+      "epoch": 1.6161616161616161,
+      "grad_norm": 0.3671875,
+      "learning_rate": 4.768265402980209e-06,
+      "loss": 0.5946,
+      "step": 480
+    },
+    {
+      "epoch": 1.6195286195286194,
+      "grad_norm": 0.427734375,
+      "learning_rate": 4.749741557118863e-06,
+      "loss": 0.6426,
+      "step": 481
+    },
+    {
+      "epoch": 1.6228956228956228,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.7312211541318e-06,
+      "loss": 0.6235,
+      "step": 482
+    },
+    {
+      "epoch": 1.6262626262626263,
+      "grad_norm": 0.359375,
+      "learning_rate": 4.712704448809305e-06,
+      "loss": 0.6413,
+      "step": 483
+    },
+    {
+      "epoch": 1.6296296296296298,
+      "grad_norm": 0.470703125,
+      "learning_rate": 4.694191695890788e-06,
+      "loss": 0.6307,
+      "step": 484
+    },
+    {
+      "epoch": 1.632996632996633,
+      "grad_norm": 0.365234375,
+      "learning_rate": 4.6756831500612846e-06,
+      "loss": 0.6458,
+      "step": 485
+    },
+    {
+      "epoch": 1.6363636363636362,
+      "grad_norm": 0.427734375,
+      "learning_rate": 4.657179065947959e-06,
+      "loss": 0.6619,
+      "step": 486
+    },
+    {
+      "epoch": 1.6397306397306397,
+      "grad_norm": 0.453125,
+      "learning_rate": 4.638679698116588e-06,
+      "loss": 0.6408,
+      "step": 487
+    },
+    {
+      "epoch": 1.6430976430976432,
+      "grad_norm": 0.4453125,
+      "learning_rate": 4.620185301068067e-06,
+      "loss": 0.6118,
+      "step": 488
+    },
+    {
+      "epoch": 1.6464646464646466,
+      "grad_norm": 0.37109375,
+      "learning_rate": 4.601696129234907e-06,
+      "loss": 0.6229,
+      "step": 489
+    },
+    {
+      "epoch": 1.6498316498316499,
+      "grad_norm": 0.396484375,
+      "learning_rate": 4.583212436977731e-06,
+      "loss": 0.6485,
+      "step": 490
+    },
+    {
+      "epoch": 1.6531986531986531,
+      "grad_norm": 0.4453125,
+      "learning_rate": 4.564734478581785e-06,
+      "loss": 0.6682,
+      "step": 491
+    },
+    {
+      "epoch": 1.6565656565656566,
+      "grad_norm": 0.3984375,
+      "learning_rate": 4.546262508253429e-06,
+      "loss": 0.6672,
+      "step": 492
+    },
+    {
+      "epoch": 1.65993265993266,
+      "grad_norm": 0.41796875,
+      "learning_rate": 4.527796780116642e-06,
+      "loss": 0.6653,
+      "step": 493
+    },
+    {
+      "epoch": 1.6632996632996633,
+      "grad_norm": 0.408203125,
+      "learning_rate": 4.509337548209529e-06,
+      "loss": 0.6126,
+      "step": 494
+    },
+    {
+      "epoch": 1.6632996632996633,
+      "eval_loss": 0.6399661302566528,
+      "eval_runtime": 8.7141,
+      "eval_samples_per_second": 57.378,
+      "eval_steps_per_second": 3.672,
+      "step": 494
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.37109375,
+      "learning_rate": 4.4908850664808245e-06,
+      "loss": 0.6047,
+      "step": 495
+    },
+    {
+      "epoch": 1.67003367003367,
+      "grad_norm": 0.376953125,
+      "learning_rate": 4.472439588786398e-06,
+      "loss": 0.6404,
+      "step": 496
+    },
+    {
+      "epoch": 1.6734006734006734,
+      "grad_norm": 0.3671875,
+      "learning_rate": 4.454001368885764e-06,
+      "loss": 0.6162,
+      "step": 497
+    },
+    {
+      "epoch": 1.676767676767677,
+      "grad_norm": 0.359375,
+      "learning_rate": 4.4355706604385905e-06,
+      "loss": 0.6247,
+      "step": 498
+    },
+    {
+      "epoch": 1.6801346801346801,
+      "grad_norm": 0.451171875,
+      "learning_rate": 4.417147717001205e-06,
+      "loss": 0.6434,
+      "step": 499
+    },
+    {
+      "epoch": 1.6835016835016834,
+      "grad_norm": 0.37890625,
+      "learning_rate": 4.3987327920231124e-06,
+      "loss": 0.621,
+      "step": 500
+    },
+    {
+      "epoch": 1.6868686868686869,
+      "grad_norm": 0.412109375,
+      "learning_rate": 4.380326138843503e-06,
+      "loss": 0.6744,
+      "step": 501
+    },
+    {
+      "epoch": 1.6902356902356903,
+      "grad_norm": 0.51171875,
+      "learning_rate": 4.3619280106877716e-06,
+      "loss": 0.6396,
+      "step": 502
+    },
+    {
+      "epoch": 1.6936026936026936,
+      "grad_norm": 0.369140625,
+      "learning_rate": 4.343538660664033e-06,
+      "loss": 0.6274,
+      "step": 503
+    },
+    {
+      "epoch": 1.696969696969697,
+      "grad_norm": 0.43359375,
+      "learning_rate": 4.325158341759638e-06,
+      "loss": 0.649,
+      "step": 504
+    },
+    {
+      "epoch": 1.7003367003367003,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.30678730683769e-06,
+      "loss": 0.6166,
+      "step": 505
+    },
+    {
+      "epoch": 1.7037037037037037,
+      "grad_norm": 0.359375,
+      "learning_rate": 4.2884258086335755e-06,
+      "loss": 0.6649,
+      "step": 506
+    },
+    {
+      "epoch": 1.7070707070707072,
+      "grad_norm": 0.33203125,
+      "learning_rate": 4.270074099751478e-06,
+      "loss": 0.6178,
+      "step": 507
+    },
+    {
+      "epoch": 1.7104377104377104,
+      "grad_norm": 0.451171875,
+      "learning_rate": 4.251732432660909e-06,
+      "loss": 0.6422,
+      "step": 508
+    },
+    {
+      "epoch": 1.7138047138047137,
+      "grad_norm": 0.37109375,
+      "learning_rate": 4.233401059693231e-06,
+      "loss": 0.6151,
+      "step": 509
+    },
+    {
+      "epoch": 1.7171717171717171,
+      "grad_norm": 0.37890625,
+      "learning_rate": 4.215080233038189e-06,
+      "loss": 0.6273,
+      "step": 510
+    },
+    {
+      "epoch": 1.7205387205387206,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.1967702047404365e-06,
+      "loss": 0.6463,
+      "step": 511
+    },
+    {
+      "epoch": 1.723905723905724,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.178471226696073e-06,
+      "loss": 0.6308,
+      "step": 512
+    },
+    {
+      "epoch": 1.7272727272727273,
+      "grad_norm": 0.453125,
+      "learning_rate": 4.160183550649176e-06,
+      "loss": 0.6747,
+      "step": 513
+    },
+    {
+      "epoch": 1.7306397306397305,
+      "grad_norm": 0.41796875,
+      "learning_rate": 4.141907428188341e-06,
+      "loss": 0.6368,
+      "step": 514
+    },
+    {
+      "epoch": 1.734006734006734,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.123643110743214e-06,
+      "loss": 0.6121,
+      "step": 515
+    },
+    {
+      "epoch": 1.7373737373737375,
+      "grad_norm": 0.4375,
+      "learning_rate": 4.105390849581037e-06,
+      "loss": 0.5883,
+      "step": 516
+    },
+    {
+      "epoch": 1.7407407407407407,
+      "grad_norm": 0.392578125,
+      "learning_rate": 4.087150895803192e-06,
+      "loss": 0.6139,
+      "step": 517
+    },
+    {
+      "epoch": 1.7441077441077442,
+      "grad_norm": 0.41015625,
+      "learning_rate": 4.0689235003417425e-06,
+      "loss": 0.6331,
+      "step": 518
+    },
+    {
+      "epoch": 1.7474747474747474,
+      "grad_norm": 0.455078125,
+      "learning_rate": 4.050708913955987e-06,
+      "loss": 0.6562,
+      "step": 519
+    },
+    {
+      "epoch": 1.7508417508417509,
+      "grad_norm": 0.54296875,
+      "learning_rate": 4.032507387229002e-06,
+      "loss": 0.6649,
+      "step": 520
+    },
+    {
+      "epoch": 1.7542087542087543,
+      "grad_norm": 0.4453125,
+      "learning_rate": 4.014319170564202e-06,
+      "loss": 0.6331,
+      "step": 521
+    },
+    {
+      "epoch": 1.7575757575757576,
+      "grad_norm": 0.388671875,
+      "learning_rate": 3.996144514181891e-06,
+      "loss": 0.6033,
+      "step": 522
+    },
+    {
+      "epoch": 1.7609427609427608,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.977983668115818e-06,
+      "loss": 0.5911,
+      "step": 523
+    },
+    {
+      "epoch": 1.7643097643097643,
+      "grad_norm": 0.50390625,
+      "learning_rate": 3.959836882209739e-06,
+      "loss": 0.5901,
+      "step": 524
+    },
+    {
+      "epoch": 1.7676767676767677,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.941704406113989e-06,
+      "loss": 0.6544,
+      "step": 525
+    },
+    {
+      "epoch": 1.7710437710437712,
+      "grad_norm": 0.345703125,
+      "learning_rate": 3.923586489282028e-06,
+      "loss": 0.5829,
+      "step": 526
+    },
+    {
+      "epoch": 1.7744107744107744,
+      "grad_norm": 0.421875,
+      "learning_rate": 3.905483380967027e-06,
+      "loss": 0.6231,
+      "step": 527
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "grad_norm": 0.390625,
+      "learning_rate": 3.887395330218429e-06,
+      "loss": 0.625,
+      "step": 528
+    },
+    {
+      "epoch": 1.7811447811447811,
+      "grad_norm": 0.419921875,
+      "learning_rate": 3.869322585878527e-06,
+      "loss": 0.6534,
+      "step": 529
+    },
+    {
+      "epoch": 1.7845117845117846,
+      "grad_norm": 0.328125,
+      "learning_rate": 3.8512653965790395e-06,
+      "loss": 0.6285,
+      "step": 530
+    },
+    {
+      "epoch": 1.7878787878787878,
+      "grad_norm": 0.376953125,
+      "learning_rate": 3.8332240107376925e-06,
+      "loss": 0.6146,
+      "step": 531
+    },
+    {
+      "epoch": 1.791245791245791,
+      "grad_norm": 0.365234375,
+      "learning_rate": 3.815198676554794e-06,
+      "loss": 0.6564,
+      "step": 532
+    },
+    {
+      "epoch": 1.791245791245791,
+      "eval_loss": 0.6390766501426697,
+      "eval_runtime": 8.8208,
+      "eval_samples_per_second": 56.684,
+      "eval_steps_per_second": 3.628,
+      "step": 532
+    },
+    {
+      "epoch": 1.7946127946127945,
+      "grad_norm": 0.38671875,
+      "learning_rate": 3.7971896420098296e-06,
+      "loss": 0.6665,
+      "step": 533
+    },
+    {
+      "epoch": 1.797979797979798,
+      "grad_norm": 0.376953125,
+      "learning_rate": 3.779197154858044e-06,
+      "loss": 0.5799,
+      "step": 534
+    },
+    {
+      "epoch": 1.8013468013468015,
+      "grad_norm": 0.353515625,
+      "learning_rate": 3.761221462627034e-06,
+      "loss": 0.6358,
+      "step": 535
+    },
+    {
+      "epoch": 1.8047138047138047,
+      "grad_norm": 0.380859375,
+      "learning_rate": 3.7432628126133497e-06,
+      "loss": 0.6194,
+      "step": 536
+    },
+    {
+      "epoch": 1.808080808080808,
+      "grad_norm": 0.384765625,
+      "learning_rate": 3.7253214518790814e-06,
+      "loss": 0.6313,
+      "step": 537
+    },
+    {
+      "epoch": 1.8114478114478114,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.7073976272484647e-06,
+      "loss": 0.6127,
+      "step": 538
+    },
+    {
+      "epoch": 1.8148148148148149,
+      "grad_norm": 0.349609375,
+      "learning_rate": 3.689491585304491e-06,
+      "loss": 0.617,
+      "step": 539
+    },
+    {
+      "epoch": 1.8181818181818183,
+      "grad_norm": 0.40625,
+      "learning_rate": 3.671603572385506e-06,
+      "loss": 0.5871,
+      "step": 540
+    },
+    {
+      "epoch": 1.8215488215488216,
+      "grad_norm": 0.37109375,
+      "learning_rate": 3.6537338345818273e-06,
+      "loss": 0.667,
+      "step": 541
+    },
+    {
+      "epoch": 1.8249158249158248,
+      "grad_norm": 0.380859375,
+      "learning_rate": 3.63588261773236e-06,
+      "loss": 0.6431,
+      "step": 542
+    },
+    {
+      "epoch": 1.8282828282828283,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.6180501674212052e-06,
+      "loss": 0.6173,
+      "step": 543
+    },
+    {
+      "epoch": 1.8316498316498318,
+      "grad_norm": 0.375,
+      "learning_rate": 3.600236728974292e-06,
+      "loss": 0.6238,
+      "step": 544
+    },
+    {
+      "epoch": 1.835016835016835,
+      "grad_norm": 0.353515625,
+      "learning_rate": 3.5824425474559956e-06,
+      "loss": 0.6155,
+      "step": 545
+    },
+    {
+      "epoch": 1.8383838383838382,
+      "grad_norm": 0.3671875,
+      "learning_rate": 3.5646678676657687e-06,
+      "loss": 0.655,
+      "step": 546
+    },
+    {
+      "epoch": 1.8417508417508417,
+      "grad_norm": 0.40625,
+      "learning_rate": 3.546912934134773e-06,
+      "loss": 0.63,
+      "step": 547
+    },
+    {
+      "epoch": 1.8451178451178452,
+      "grad_norm": 0.353515625,
+      "learning_rate": 3.529177991122519e-06,
+      "loss": 0.6469,
+      "step": 548
+    },
+    {
+      "epoch": 1.8484848484848486,
+      "grad_norm": 0.400390625,
+      "learning_rate": 3.5114632826134977e-06,
+      "loss": 0.6211,
+      "step": 549
+    },
+    {
+      "epoch": 1.8518518518518519,
+      "grad_norm": 0.373046875,
+      "learning_rate": 3.4937690523138302e-06,
+      "loss": 0.6184,
+      "step": 550
+    },
+    {
+      "epoch": 1.855218855218855,
+      "grad_norm": 0.390625,
+      "learning_rate": 3.4760955436479134e-06,
+      "loss": 0.637,
+      "step": 551
+    },
+    {
+      "epoch": 1.8585858585858586,
+      "grad_norm": 0.39453125,
+      "learning_rate": 3.4584429997550685e-06,
+      "loss": 0.6394,
+      "step": 552
+    },
+    {
+      "epoch": 1.861952861952862,
+      "grad_norm": 0.443359375,
+      "learning_rate": 3.440811663486206e-06,
+      "loss": 0.6373,
+      "step": 553
+    },
+    {
+      "epoch": 1.8653198653198653,
+      "grad_norm": 0.376953125,
+      "learning_rate": 3.423201777400469e-06,
+      "loss": 0.6505,
+      "step": 554
+    },
+    {
+      "epoch": 1.8686868686868687,
+      "grad_norm": 0.3671875,
+      "learning_rate": 3.4056135837619077e-06,
+      "loss": 0.6813,
+      "step": 555
+    },
+    {
+      "epoch": 1.872053872053872,
+      "grad_norm": 0.50390625,
+      "learning_rate": 3.388047324536143e-06,
+      "loss": 0.6306,
+      "step": 556
+    },
+    {
+      "epoch": 1.8754208754208754,
+      "grad_norm": 0.4296875,
+      "learning_rate": 3.3705032413870402e-06,
+      "loss": 0.6565,
+      "step": 557
+    },
+    {
+      "epoch": 1.878787878787879,
+      "grad_norm": 0.390625,
+      "learning_rate": 3.3529815756733773e-06,
+      "loss": 0.6375,
+      "step": 558
+    },
+    {
+      "epoch": 1.8821548821548821,
+      "grad_norm": 0.3984375,
+      "learning_rate": 3.335482568445536e-06,
+      "loss": 0.623,
+      "step": 559
+    },
+    {
+      "epoch": 1.8855218855218854,
+      "grad_norm": 0.45703125,
+      "learning_rate": 3.318006460442173e-06,
+      "loss": 0.6253,
+      "step": 560
+    },
+    {
+      "epoch": 1.8888888888888888,
+      "grad_norm": 0.330078125,
+      "learning_rate": 3.3005534920869175e-06,
+      "loss": 0.6161,
+      "step": 561
+    },
+    {
+      "epoch": 1.8922558922558923,
+      "grad_norm": 0.369140625,
+      "learning_rate": 3.2831239034850593e-06,
+      "loss": 0.627,
+      "step": 562
+    },
+    {
+      "epoch": 1.8956228956228958,
+      "grad_norm": 0.3984375,
+      "learning_rate": 3.265717934420244e-06,
+      "loss": 0.6625,
+      "step": 563
+    },
+    {
+      "epoch": 1.898989898989899,
+      "grad_norm": 0.380859375,
+      "learning_rate": 3.248335824351183e-06,
+      "loss": 0.642,
+      "step": 564
+    },
+    {
+      "epoch": 1.9023569023569022,
+      "grad_norm": 0.380859375,
+      "learning_rate": 3.230977812408348e-06,
+      "loss": 0.618,
+      "step": 565
+    },
+    {
+      "epoch": 1.9057239057239057,
+      "grad_norm": 0.359375,
+      "learning_rate": 3.213644137390686e-06,
+      "loss": 0.6347,
+      "step": 566
+    },
+    {
+      "epoch": 1.9090909090909092,
+      "grad_norm": 0.392578125,
+      "learning_rate": 3.196335037762337e-06,
+      "loss": 0.6344,
+      "step": 567
+    },
+    {
+      "epoch": 1.9124579124579124,
+      "grad_norm": 0.3984375,
+      "learning_rate": 3.1790507516493473e-06,
+      "loss": 0.6328,
+      "step": 568
+    },
+    {
+      "epoch": 1.9158249158249159,
+      "grad_norm": 0.3671875,
+      "learning_rate": 3.1617915168363994e-06,
+      "loss": 0.6214,
+      "step": 569
+    },
+    {
+      "epoch": 1.9191919191919191,
+      "grad_norm": 0.416015625,
+      "learning_rate": 3.1445575707635383e-06,
+      "loss": 0.6296,
+      "step": 570
+    },
+    {
+      "epoch": 1.9191919191919191,
+      "eval_loss": 0.6386533379554749,
+      "eval_runtime": 8.8263,
+      "eval_samples_per_second": 56.649,
+      "eval_steps_per_second": 3.626,
+      "step": 570
+    },
+    {
+      "epoch": 1.9225589225589226,
+      "grad_norm": 0.33984375,
+      "learning_rate": 3.127349150522903e-06,
+      "loss": 0.6003,
+      "step": 571
+    },
+    {
+      "epoch": 1.925925925925926,
+      "grad_norm": 0.427734375,
+      "learning_rate": 3.110166492855468e-06,
+      "loss": 0.636,
+      "step": 572
+    },
+    {
+      "epoch": 1.9292929292929293,
+      "grad_norm": 0.404296875,
+      "learning_rate": 3.093009834147784e-06,
+      "loss": 0.6082,
+      "step": 573
+    },
+    {
+      "epoch": 1.9326599326599325,
+      "grad_norm": 0.421875,
+      "learning_rate": 3.0758794104287275e-06,
+      "loss": 0.6572,
+      "step": 574
+    },
+    {
+      "epoch": 1.936026936026936,
+      "grad_norm": 0.384765625,
+      "learning_rate": 3.0587754573662508e-06,
+      "loss": 0.6485,
+      "step": 575
+    },
+    {
+      "epoch": 1.9393939393939394,
+      "grad_norm": 0.388671875,
+      "learning_rate": 3.041698210264149e-06,
+      "loss": 0.6575,
+      "step": 576
+    },
+    {
+      "epoch": 1.942760942760943,
+      "grad_norm": 0.388671875,
+      "learning_rate": 3.0246479040588077e-06,
+      "loss": 0.6596,
+      "step": 577
+    },
+    {
+      "epoch": 1.9461279461279462,
+      "grad_norm": 0.3671875,
+      "learning_rate": 3.0076247733159846e-06,
+      "loss": 0.6281,
+      "step": 578
+    },
+    {
+      "epoch": 1.9494949494949494,
+      "grad_norm": 0.40625,
+      "learning_rate": 2.9906290522275738e-06,
+      "loss": 0.6262,
+      "step": 579
+    },
+    {
+      "epoch": 1.9528619528619529,
+      "grad_norm": 0.39453125,
+      "learning_rate": 2.9736609746083853e-06,
+      "loss": 0.6568,
+      "step": 580
+    },
+    {
+      "epoch": 1.9562289562289563,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.9567207738929395e-06,
+      "loss": 0.5815,
+      "step": 581
+    },
+    {
+      "epoch": 1.9595959595959596,
+      "grad_norm": 0.4765625,
+      "learning_rate": 2.939808683132238e-06,
+      "loss": 0.643,
+      "step": 582
+    },
+    {
+      "epoch": 1.9629629629629628,
+      "grad_norm": 0.361328125,
+      "learning_rate": 2.9229249349905686e-06,
+      "loss": 0.6048,
+      "step": 583
+    },
+    {
+      "epoch": 1.9663299663299663,
+      "grad_norm": 0.39453125,
+      "learning_rate": 2.906069761742302e-06,
+      "loss": 0.6503,
+      "step": 584
+    },
+    {
+      "epoch": 1.9696969696969697,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.889243395268696e-06,
+      "loss": 0.63,
+      "step": 585
+    },
+    {
+      "epoch": 1.9730639730639732,
+      "grad_norm": 0.36328125,
+      "learning_rate": 2.872446067054703e-06,
+      "loss": 0.6258,
+      "step": 586
+    },
+    {
+      "epoch": 1.9764309764309764,
+      "grad_norm": 0.408203125,
+      "learning_rate": 2.8556780081857966e-06,
+      "loss": 0.6202,
+      "step": 587
+    },
+    {
+      "epoch": 1.9797979797979797,
+      "grad_norm": 0.36328125,
+      "learning_rate": 2.8389394493447732e-06,
+      "loss": 0.626,
+      "step": 588
+    },
+    {
+      "epoch": 1.9831649831649831,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.8222306208085988e-06,
+      "loss": 0.6197,
+      "step": 589
+    },
+    {
+      "epoch": 1.9865319865319866,
+      "grad_norm": 0.412109375,
+      "learning_rate": 2.805551752445222e-06,
+      "loss": 0.6703,
+      "step": 590
+    },
+    {
+      "epoch": 1.98989898989899,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.7889030737104283e-06,
+      "loss": 0.6288,
+      "step": 591
+    },
+    {
+      "epoch": 1.9932659932659933,
+      "grad_norm": 0.451171875,
+      "learning_rate": 2.772284813644675e-06,
+      "loss": 0.612,
+      "step": 592
+    },
+    {
+      "epoch": 1.9966329966329965,
+      "grad_norm": 0.384765625,
+      "learning_rate": 2.755697200869937e-06,
+      "loss": 0.6372,
+      "step": 593
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.37890625,
+      "learning_rate": 2.7391404635865725e-06,
+      "loss": 0.6565,
+      "step": 594
+    },
+    {
+      "epoch": 2.0033670033670035,
+      "grad_norm": 0.380859375,
+      "learning_rate": 2.7226148295701704e-06,
+      "loss": 0.6427,
+      "step": 595
+    },
+    {
+      "epoch": 2.006734006734007,
+      "grad_norm": 0.427734375,
+      "learning_rate": 2.7061205261684307e-06,
+      "loss": 0.6403,
+      "step": 596
+    },
+    {
+      "epoch": 2.01010101010101,
+      "grad_norm": 0.5546875,
+      "learning_rate": 2.689657780298019e-06,
+      "loss": 0.6027,
+      "step": 597
+    },
+    {
+      "epoch": 2.0134680134680134,
+      "grad_norm": 0.345703125,
+      "learning_rate": 2.6732268184414695e-06,
+      "loss": 0.5689,
+      "step": 598
+    },
+    {
+      "epoch": 2.016835016835017,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.6568278666440417e-06,
+      "loss": 0.6573,
+      "step": 599
+    },
+    {
+      "epoch": 2.0202020202020203,
+      "grad_norm": 0.404296875,
+      "learning_rate": 2.6404611505106337e-06,
+      "loss": 0.6421,
+      "step": 600
+    },
+    {
+      "epoch": 2.0235690235690234,
+      "grad_norm": 0.57421875,
+      "learning_rate": 2.6241268952026592e-06,
+      "loss": 0.6733,
+      "step": 601
+    },
+    {
+      "epoch": 2.026936026936027,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.6078253254349706e-06,
+      "loss": 0.6076,
+      "step": 602
+    },
+    {
+      "epoch": 2.0303030303030303,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.591556665472744e-06,
+      "loss": 0.6978,
+      "step": 603
+    },
+    {
+      "epoch": 2.0336700336700337,
+      "grad_norm": 0.3515625,
+      "learning_rate": 2.5753211391284172e-06,
+      "loss": 0.6213,
+      "step": 604
+    },
+    {
+      "epoch": 2.037037037037037,
+      "grad_norm": 0.357421875,
+      "learning_rate": 2.559118969758595e-06,
+      "loss": 0.6379,
+      "step": 605
+    },
+    {
+      "epoch": 2.04040404040404,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.5429503802609777e-06,
+      "loss": 0.6019,
+      "step": 606
+    },
+    {
+      "epoch": 2.0437710437710437,
+      "grad_norm": 0.341796875,
+      "learning_rate": 2.526815593071306e-06,
+      "loss": 0.6515,
+      "step": 607
+    },
+    {
+      "epoch": 2.047138047138047,
+      "grad_norm": 0.333984375,
+      "learning_rate": 2.5107148301602825e-06,
+      "loss": 0.6225,
+      "step": 608
+    },
+    {
+      "epoch": 2.047138047138047,
+      "eval_loss": 0.6382743716239929,
+      "eval_runtime": 8.8157,
+      "eval_samples_per_second": 56.717,
+      "eval_steps_per_second": 3.63,
+      "step": 608
+    },
+    {
+      "epoch": 2.0505050505050506,
+      "grad_norm": 0.376953125,
+      "learning_rate": 2.494648313030542e-06,
+      "loss": 0.66,
+      "step": 609
+    },
+    {
+      "epoch": 2.053872053872054,
+      "grad_norm": 0.4140625,
+      "learning_rate": 2.478616262713578e-06,
+      "loss": 0.6744,
+      "step": 610
+    },
+    {
+      "epoch": 2.057239057239057,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.4626188997667224e-06,
+      "loss": 0.6055,
+      "step": 611
+    },
+    {
+      "epoch": 2.0606060606060606,
+      "grad_norm": 0.40234375,
+      "learning_rate": 2.4466564442700974e-06,
+      "loss": 0.6622,
+      "step": 612
+    },
+    {
+      "epoch": 2.063973063973064,
+      "grad_norm": 0.408203125,
+      "learning_rate": 2.430729115823601e-06,
+      "loss": 0.6355,
+      "step": 613
+    },
+    {
+      "epoch": 2.0673400673400675,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.414837133543871e-06,
+      "loss": 0.6622,
+      "step": 614
+    },
+    {
+      "epoch": 2.0707070707070705,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.398980716061284e-06,
+      "loss": 0.6248,
+      "step": 615
+    },
+    {
+      "epoch": 2.074074074074074,
+      "grad_norm": 0.380859375,
+      "learning_rate": 2.383160081516941e-06,
+      "loss": 0.6185,
+      "step": 616
+    },
+    {
+      "epoch": 2.0774410774410774,
+      "grad_norm": 0.35546875,
+      "learning_rate": 2.3673754475596634e-06,
+      "loss": 0.6792,
+      "step": 617
+    },
+    {
+      "epoch": 2.080808080808081,
+      "grad_norm": 0.353515625,
+      "learning_rate": 2.3516270313430085e-06,
+      "loss": 0.659,
+      "step": 618
+    },
+    {
+      "epoch": 2.0841750841750843,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.335915049522268e-06,
+      "loss": 0.6598,
+      "step": 619
+    },
+    {
+      "epoch": 2.0875420875420874,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.320239718251508e-06,
+      "loss": 0.6267,
+      "step": 620
+    },
+    {
+      "epoch": 2.090909090909091,
+      "grad_norm": 0.333984375,
+      "learning_rate": 2.30460125318057e-06,
+      "loss": 0.5984,
+      "step": 621
+    },
+    {
+      "epoch": 2.0942760942760943,
+      "grad_norm": 0.4453125,
+      "learning_rate": 2.2889998694521257e-06,
+      "loss": 0.6532,
+      "step": 622
+    },
+    {
+      "epoch": 2.0976430976430978,
+      "grad_norm": 0.388671875,
+      "learning_rate": 2.2734357816987046e-06,
+      "loss": 0.6447,
+      "step": 623
+    },
+    {
+      "epoch": 2.101010101010101,
+      "grad_norm": 0.3671875,
+      "learning_rate": 2.2579092040397436e-06,
+      "loss": 0.6367,
+      "step": 624
+    },
+    {
+      "epoch": 2.1043771043771042,
+      "grad_norm": 0.345703125,
+      "learning_rate": 2.2424203500786473e-06,
+      "loss": 0.5905,
+      "step": 625
+    },
+    {
+      "epoch": 2.1077441077441077,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.226969432899843e-06,
+      "loss": 0.6587,
+      "step": 626
+    },
+    {
+      "epoch": 2.111111111111111,
+      "grad_norm": 0.48828125,
+      "learning_rate": 2.211556665065854e-06,
+      "loss": 0.6198,
+      "step": 627
+    },
+    {
+      "epoch": 2.1144781144781146,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.196182258614365e-06,
+      "loss": 0.6058,
+      "step": 628
+    },
+    {
+      "epoch": 2.1178451178451176,
+      "grad_norm": 0.337890625,
+      "learning_rate": 2.180846425055322e-06,
+      "loss": 0.5772,
+      "step": 629
+    },
+    {
+      "epoch": 2.121212121212121,
+      "grad_norm": 0.396484375,
+      "learning_rate": 2.165549375368004e-06,
+      "loss": 0.6245,
+      "step": 630
+    },
+    {
+      "epoch": 2.1245791245791246,
+      "grad_norm": 0.33984375,
+      "learning_rate": 2.150291319998137e-06,
+      "loss": 0.6194,
+      "step": 631
+    },
+    {
+      "epoch": 2.127946127946128,
+      "grad_norm": 0.3984375,
+      "learning_rate": 2.1350724688549906e-06,
+      "loss": 0.67,
+      "step": 632
+    },
+    {
+      "epoch": 2.1313131313131315,
+      "grad_norm": 0.38671875,
+      "learning_rate": 2.119893031308485e-06,
+      "loss": 0.5599,
+      "step": 633
+    },
+    {
+      "epoch": 2.1346801346801345,
+      "grad_norm": 0.375,
+      "learning_rate": 2.1047532161863242e-06,
+      "loss": 0.6605,
+      "step": 634
+    },
+    {
+      "epoch": 2.138047138047138,
+      "grad_norm": 0.353515625,
+      "learning_rate": 2.0896532317711087e-06,
+      "loss": 0.6119,
+      "step": 635
+    },
+    {
+      "epoch": 2.1414141414141414,
+      "grad_norm": 0.3515625,
+      "learning_rate": 2.0745932857974825e-06,
+      "loss": 0.6513,
+      "step": 636
+    },
+    {
+      "epoch": 2.144781144781145,
+      "grad_norm": 0.36328125,
+      "learning_rate": 2.0595735854492675e-06,
+      "loss": 0.6447,
+      "step": 637
+    },
+    {
+      "epoch": 2.148148148148148,
+      "grad_norm": 0.390625,
+      "learning_rate": 2.0445943373566178e-06,
+      "loss": 0.5849,
+      "step": 638
+    },
+    {
+      "epoch": 2.1515151515151514,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.029655747593169e-06,
+      "loss": 0.6214,
+      "step": 639
+    },
+    {
+      "epoch": 2.154882154882155,
+      "grad_norm": 0.392578125,
+      "learning_rate": 2.014758021673216e-06,
+      "loss": 0.6174,
+      "step": 640
+    },
+    {
+      "epoch": 2.1582491582491583,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.9999013645488714e-06,
+      "loss": 0.6442,
+      "step": 641
+    },
+    {
+      "epoch": 2.1616161616161618,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.9850859806072576e-06,
+      "loss": 0.6485,
+      "step": 642
+    },
+    {
+      "epoch": 2.164983164983165,
+      "grad_norm": 0.353515625,
+      "learning_rate": 1.970312073667693e-06,
+      "loss": 0.6031,
+      "step": 643
+    },
+    {
+      "epoch": 2.1683501683501682,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.9555798469788773e-06,
+      "loss": 0.6336,
+      "step": 644
+    },
+    {
+      "epoch": 2.1717171717171717,
+      "grad_norm": 0.404296875,
+      "learning_rate": 1.9408895032161116e-06,
+      "loss": 0.6509,
+      "step": 645
+    },
+    {
+      "epoch": 2.175084175084175,
+      "grad_norm": 0.40234375,
+      "learning_rate": 1.926241244478496e-06,
+      "loss": 0.6354,
+      "step": 646
+    },
+    {
+      "epoch": 2.175084175084175,
+      "eval_loss": 0.6380563378334045,
+      "eval_runtime": 8.6394,
+      "eval_samples_per_second": 57.874,
+      "eval_steps_per_second": 3.704,
+      "step": 646
+    },
+    {
+      "epoch": 2.1784511784511786,
+      "grad_norm": 0.41015625,
+      "learning_rate": 1.9116352722861596e-06,
+      "loss": 0.6661,
+      "step": 647
+    },
+    {
+      "epoch": 2.1818181818181817,
+      "grad_norm": 0.3984375,
+      "learning_rate": 1.897071787577482e-06,
+      "loss": 0.6442,
+      "step": 648
+    },
+    {
+      "epoch": 2.185185185185185,
+      "grad_norm": 0.396484375,
+      "learning_rate": 1.8825509907063328e-06,
+      "loss": 0.6244,
+      "step": 649
+    },
+    {
+      "epoch": 2.1885521885521886,
+      "grad_norm": 0.38671875,
+      "learning_rate": 1.8680730814393084e-06,
+      "loss": 0.602,
+      "step": 650
+    },
+    {
+      "epoch": 2.191919191919192,
+      "grad_norm": 0.3984375,
+      "learning_rate": 1.8536382589529944e-06,
+      "loss": 0.6499,
+      "step": 651
+    },
+    {
+      "epoch": 2.1952861952861955,
+      "grad_norm": 0.375,
+      "learning_rate": 1.839246721831215e-06,
+      "loss": 0.6436,
+      "step": 652
+    },
+    {
+      "epoch": 2.1986531986531985,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.8248986680623077e-06,
+      "loss": 0.6245,
+      "step": 653
+    },
+    {
+      "epoch": 2.202020202020202,
+      "grad_norm": 0.48828125,
+      "learning_rate": 1.8105942950364008e-06,
+      "loss": 0.6275,
+      "step": 654
+    },
+    {
+      "epoch": 2.2053872053872055,
+      "grad_norm": 0.44921875,
+      "learning_rate": 1.7963337995426876e-06,
+      "loss": 0.6592,
+      "step": 655
+    },
+    {
+      "epoch": 2.208754208754209,
+      "grad_norm": 0.44921875,
+      "learning_rate": 1.7821173777667333e-06,
+      "loss": 0.6618,
+      "step": 656
+    },
+    {
+      "epoch": 2.212121212121212,
+      "grad_norm": 0.419921875,
+      "learning_rate": 1.7679452252877622e-06,
+      "loss": 0.646,
+      "step": 657
+    },
+    {
+      "epoch": 2.2154882154882154,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.7538175370759797e-06,
+      "loss": 0.6202,
+      "step": 658
+    },
+    {
+      "epoch": 2.218855218855219,
+      "grad_norm": 0.33984375,
+      "learning_rate": 1.739734507489877e-06,
+      "loss": 0.6234,
+      "step": 659
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 0.38671875,
+      "learning_rate": 1.7256963302735752e-06,
+      "loss": 0.624,
+      "step": 660
+    },
+    {
+      "epoch": 2.225589225589226,
+      "grad_norm": 0.490234375,
+      "learning_rate": 1.7117031985541366e-06,
+      "loss": 0.6385,
+      "step": 661
+    },
+    {
+      "epoch": 2.228956228956229,
+      "grad_norm": 0.349609375,
+      "learning_rate": 1.6977553048389306e-06,
+      "loss": 0.6177,
+      "step": 662
+    },
+    {
+      "epoch": 2.2323232323232323,
+      "grad_norm": 0.419921875,
+      "learning_rate": 1.6838528410129662e-06,
+      "loss": 0.6581,
+      "step": 663
+    },
+    {
+      "epoch": 2.2356902356902357,
+      "grad_norm": 0.353515625,
+      "learning_rate": 1.6699959983362668e-06,
+      "loss": 0.599,
+      "step": 664
+    },
+    {
+      "epoch": 2.239057239057239,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.6561849674412327e-06,
+      "loss": 0.6155,
+      "step": 665
+    },
+    {
+      "epoch": 2.242424242424242,
+      "grad_norm": 0.3984375,
+      "learning_rate": 1.6424199383300149e-06,
+      "loss": 0.6474,
+      "step": 666
+    },
+    {
+      "epoch": 2.2457912457912457,
+      "grad_norm": 0.376953125,
+      "learning_rate": 1.6287011003719105e-06,
+      "loss": 0.607,
+      "step": 667
+    },
+    {
+      "epoch": 2.249158249158249,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.6150286423007472e-06,
+      "loss": 0.623,
+      "step": 668
+    },
+    {
+      "epoch": 2.2525252525252526,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.601402752212296e-06,
+      "loss": 0.5932,
+      "step": 669
+    },
+    {
+      "epoch": 2.255892255892256,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.5878236175616736e-06,
+      "loss": 0.5884,
+      "step": 670
+    },
+    {
+      "epoch": 2.259259259259259,
+      "grad_norm": 0.380859375,
+      "learning_rate": 1.5742914251607794e-06,
+      "loss": 0.6458,
+      "step": 671
+    },
+    {
+      "epoch": 2.2626262626262625,
+      "grad_norm": 0.419921875,
+      "learning_rate": 1.5608063611757058e-06,
+      "loss": 0.6524,
+      "step": 672
+    },
+    {
+      "epoch": 2.265993265993266,
+      "grad_norm": 0.353515625,
+      "learning_rate": 1.5473686111241881e-06,
+      "loss": 0.6053,
+      "step": 673
+    },
+    {
+      "epoch": 2.2693602693602695,
+      "grad_norm": 0.390625,
+      "learning_rate": 1.5339783598730568e-06,
+      "loss": 0.6409,
+      "step": 674
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 0.400390625,
+      "learning_rate": 1.5206357916356796e-06,
+      "loss": 0.6323,
+      "step": 675
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 891,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 75,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.227939051995136e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}