diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7821 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.7037037037037037,
+  "eval_steps": 100,
+  "global_step": 1100,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.003367003367003367,
+      "grad_norm": 190.2997283935547,
+      "learning_rate": 6.711409395973154e-07,
+      "loss": 13.9272,
+      "step": 1
+    },
+    {
+      "epoch": 0.006734006734006734,
+      "grad_norm": 196.5933074951172,
+      "learning_rate": 1.3422818791946309e-06,
+      "loss": 14.3753,
+      "step": 2
+    },
+    {
+      "epoch": 0.010101010101010102,
+      "grad_norm": 198.02767944335938,
+      "learning_rate": 2.013422818791946e-06,
+      "loss": 14.4143,
+      "step": 3
+    },
+    {
+      "epoch": 0.013468013468013467,
+      "grad_norm": 186.30801391601562,
+      "learning_rate": 2.6845637583892617e-06,
+      "loss": 13.7729,
+      "step": 4
+    },
+    {
+      "epoch": 0.016835016835016835,
+      "grad_norm": 129.32237243652344,
+      "learning_rate": 3.3557046979865773e-06,
+      "loss": 11.4082,
+      "step": 5
+    },
+    {
+      "epoch": 0.020202020202020204,
+      "grad_norm": 123.9930191040039,
+      "learning_rate": 4.026845637583892e-06,
+      "loss": 11.5581,
+      "step": 6
+    },
+    {
+      "epoch": 0.02356902356902357,
+      "grad_norm": 102.4565658569336,
+      "learning_rate": 4.697986577181209e-06,
+      "loss": 9.8311,
+      "step": 7
+    },
+    {
+      "epoch": 0.026936026936026935,
+      "grad_norm": 98.7117919921875,
+      "learning_rate": 5.3691275167785235e-06,
+      "loss": 9.825,
+      "step": 8
+    },
+    {
+      "epoch": 0.030303030303030304,
+      "grad_norm": 121.9065170288086,
+      "learning_rate": 6.04026845637584e-06,
+      "loss": 8.5157,
+      "step": 9
+    },
+    {
+      "epoch": 0.03367003367003367,
+      "grad_norm": 93.352294921875,
+      "learning_rate": 6.7114093959731546e-06,
+      "loss": 7.6328,
+      "step": 10
+    },
+    {
+      "epoch": 0.037037037037037035,
+      "grad_norm": 108.89420318603516,
+      "learning_rate": 7.382550335570471e-06,
+      "loss": 7.1598,
+      "step": 11
+    },
+    {
+      "epoch": 0.04040404040404041,
+      "grad_norm": 191.65274047851562,
+      "learning_rate": 8.053691275167785e-06,
+      "loss": 6.237,
+      "step": 12
+    },
+    {
+      "epoch": 0.04377104377104377,
+      "grad_norm": 150.62646484375,
+      "learning_rate": 8.724832214765101e-06,
+      "loss": 5.7063,
+      "step": 13
+    },
+    {
+      "epoch": 0.04713804713804714,
+      "grad_norm": 185.48080444335938,
+      "learning_rate": 9.395973154362418e-06,
+      "loss": 5.093,
+      "step": 14
+    },
+    {
+      "epoch": 0.050505050505050504,
+      "grad_norm": 1576.556640625,
+      "learning_rate": 1.006711409395973e-05,
+      "loss": 8.3575,
+      "step": 15
+    },
+    {
+      "epoch": 0.05387205387205387,
+      "grad_norm": 441.4505310058594,
+      "learning_rate": 1.0738255033557047e-05,
+      "loss": 4.679,
+      "step": 16
+    },
+    {
+      "epoch": 0.05723905723905724,
+      "grad_norm": 499.8016357421875,
+      "learning_rate": 1.1409395973154363e-05,
+      "loss": 3.1432,
+      "step": 17
+    },
+    {
+      "epoch": 0.06060606060606061,
+      "grad_norm": 472.59747314453125,
+      "learning_rate": 1.208053691275168e-05,
+      "loss": 2.9237,
+      "step": 18
+    },
+    {
+      "epoch": 0.06397306397306397,
+      "grad_norm": 506.6687927246094,
+      "learning_rate": 1.2751677852348994e-05,
+      "loss": 2.6882,
+      "step": 19
+    },
+    {
+      "epoch": 0.06734006734006734,
+      "grad_norm": 494.16949462890625,
+      "learning_rate": 1.3422818791946309e-05,
+      "loss": 2.4807,
+      "step": 20
+    },
+    {
+      "epoch": 0.0707070707070707,
+      "grad_norm": 463.3478698730469,
+      "learning_rate": 1.4093959731543624e-05,
+      "loss": 2.2508,
+      "step": 21
+    },
+    {
+      "epoch": 0.07407407407407407,
+      "grad_norm": 422.92401123046875,
+      "learning_rate": 1.4765100671140942e-05,
+      "loss": 1.9202,
+      "step": 22
+    },
+    {
+      "epoch": 0.07744107744107744,
+      "grad_norm": 417.1321105957031,
+      "learning_rate": 1.5436241610738255e-05,
+      "loss": 1.6106,
+      "step": 23
+    },
+    {
+      "epoch": 0.08080808080808081,
+      "grad_norm": 360.2781677246094,
+      "learning_rate": 1.610738255033557e-05,
+      "loss": 1.2741,
+      "step": 24
+    },
+    {
+      "epoch": 0.08417508417508418,
+      "grad_norm": 297.3291015625,
+      "learning_rate": 1.6778523489932888e-05,
+      "loss": 1.0282,
+      "step": 25
+    },
+    {
+      "epoch": 0.08754208754208755,
+      "grad_norm": 195.75958251953125,
+      "learning_rate": 1.7449664429530202e-05,
+      "loss": 0.799,
+      "step": 26
+    },
+    {
+      "epoch": 0.09090909090909091,
+      "grad_norm": 116.36829376220703,
+      "learning_rate": 1.8120805369127517e-05,
+      "loss": 0.6593,
+      "step": 27
+    },
+    {
+      "epoch": 0.09427609427609428,
+      "grad_norm": 70.56578063964844,
+      "learning_rate": 1.8791946308724835e-05,
+      "loss": 0.5787,
+      "step": 28
+    },
+    {
+      "epoch": 0.09764309764309764,
+      "grad_norm": 45.22296905517578,
+      "learning_rate": 1.946308724832215e-05,
+      "loss": 0.5196,
+      "step": 29
+    },
+    {
+      "epoch": 0.10101010101010101,
+      "grad_norm": 20.37734603881836,
+      "learning_rate": 2.013422818791946e-05,
+      "loss": 0.4681,
+      "step": 30
+    },
+    {
+      "epoch": 0.10437710437710437,
+      "grad_norm": 7.735367298126221,
+      "learning_rate": 2.080536912751678e-05,
+      "loss": 0.4318,
+      "step": 31
+    },
+    {
+      "epoch": 0.10774410774410774,
+      "grad_norm": 4.360243797302246,
+      "learning_rate": 2.1476510067114094e-05,
+      "loss": 0.4276,
+      "step": 32
+    },
+    {
+      "epoch": 0.1111111111111111,
+      "grad_norm": 4.440345287322998,
+      "learning_rate": 2.2147651006711412e-05,
+      "loss": 0.4463,
+      "step": 33
+    },
+    {
+      "epoch": 0.11447811447811448,
+      "grad_norm": 26.992700576782227,
+      "learning_rate": 2.2818791946308727e-05,
+      "loss": 0.4394,
+      "step": 34
+    },
+    {
+      "epoch": 0.11784511784511785,
+      "grad_norm": 33.81399917602539,
+      "learning_rate": 2.348993288590604e-05,
+      "loss": 0.6005,
+      "step": 35
+    },
+    {
+      "epoch": 0.12121212121212122,
+      "grad_norm": 7.8905029296875,
+      "learning_rate": 2.416107382550336e-05,
+      "loss": 0.4963,
+      "step": 36
+    },
+    {
+      "epoch": 0.12457912457912458,
+      "grad_norm": 2.6311209201812744,
+      "learning_rate": 2.4832214765100674e-05,
+      "loss": 0.39,
+      "step": 37
+    },
+    {
+      "epoch": 0.12794612794612795,
+      "grad_norm": 2.389883041381836,
+      "learning_rate": 2.550335570469799e-05,
+      "loss": 0.3782,
+      "step": 38
+    },
+    {
+      "epoch": 0.13131313131313133,
+      "grad_norm": 2.070525646209717,
+      "learning_rate": 2.6174496644295304e-05,
+      "loss": 0.3592,
+      "step": 39
+    },
+    {
+      "epoch": 0.13468013468013468,
+      "grad_norm": 5.955089569091797,
+      "learning_rate": 2.6845637583892618e-05,
+      "loss": 0.3777,
+      "step": 40
+    },
+    {
+      "epoch": 0.13804713804713806,
+      "grad_norm": 6.50673770904541,
+      "learning_rate": 2.7516778523489933e-05,
+      "loss": 0.389,
+      "step": 41
+    },
+    {
+      "epoch": 0.1414141414141414,
+      "grad_norm": 2.0794308185577393,
+      "learning_rate": 2.8187919463087248e-05,
+      "loss": 0.3618,
+      "step": 42
+    },
+    {
+      "epoch": 0.1447811447811448,
+      "grad_norm": 1.5477614402770996,
+      "learning_rate": 2.885906040268457e-05,
+      "loss": 0.3593,
+      "step": 43
+    },
+    {
+      "epoch": 0.14814814814814814,
+      "grad_norm": 10.740438461303711,
+      "learning_rate": 2.9530201342281884e-05,
+      "loss": 0.3805,
+      "step": 44
+    },
+    {
+      "epoch": 0.15151515151515152,
+      "grad_norm": 2.993213176727295,
+      "learning_rate": 3.02013422818792e-05,
+      "loss": 0.3673,
+      "step": 45
+    },
+    {
+      "epoch": 0.15488215488215487,
+      "grad_norm": 17.512208938598633,
+      "learning_rate": 3.087248322147651e-05,
+      "loss": 0.3922,
+      "step": 46
+    },
+    {
+      "epoch": 0.15824915824915825,
+      "grad_norm": 2.5222012996673584,
+      "learning_rate": 3.1543624161073825e-05,
+      "loss": 0.3873,
+      "step": 47
+    },
+    {
+      "epoch": 0.16161616161616163,
+      "grad_norm": 0.8730729222297668,
+      "learning_rate": 3.221476510067114e-05,
+      "loss": 0.3593,
+      "step": 48
+    },
+    {
+      "epoch": 0.16498316498316498,
+      "grad_norm": 0.8050268292427063,
+      "learning_rate": 3.288590604026846e-05,
+      "loss": 0.3491,
+      "step": 49
+    },
+    {
+      "epoch": 0.16835016835016836,
+      "grad_norm": 0.7536938190460205,
+      "learning_rate": 3.3557046979865775e-05,
+      "loss": 0.3469,
+      "step": 50
+    },
+    {
+      "epoch": 0.1717171717171717,
+      "grad_norm": 0.9090268015861511,
+      "learning_rate": 3.422818791946309e-05,
+      "loss": 0.3663,
+      "step": 51
+    },
+    {
+      "epoch": 0.1750841750841751,
+      "grad_norm": 0.8775368928909302,
+      "learning_rate": 3.4899328859060405e-05,
+      "loss": 0.3489,
+      "step": 52
+    },
+    {
+      "epoch": 0.17845117845117844,
+      "grad_norm": 0.5326427221298218,
+      "learning_rate": 3.557046979865772e-05,
+      "loss": 0.3466,
+      "step": 53
+    },
+    {
+      "epoch": 0.18181818181818182,
+      "grad_norm": 0.561137318611145,
+      "learning_rate": 3.6241610738255034e-05,
+      "loss": 0.3393,
+      "step": 54
+    },
+    {
+      "epoch": 0.18518518518518517,
+      "grad_norm": 0.8053128123283386,
+      "learning_rate": 3.6912751677852356e-05,
+      "loss": 0.352,
+      "step": 55
+    },
+    {
+      "epoch": 0.18855218855218855,
+      "grad_norm": 0.5964087843894958,
+      "learning_rate": 3.758389261744967e-05,
+      "loss": 0.3507,
+      "step": 56
+    },
+    {
+      "epoch": 0.1919191919191919,
+      "grad_norm": 0.5998376607894897,
+      "learning_rate": 3.8255033557046985e-05,
+      "loss": 0.3504,
+      "step": 57
+    },
+    {
+      "epoch": 0.19528619528619529,
+      "grad_norm": 1.2634875774383545,
+      "learning_rate": 3.89261744966443e-05,
+      "loss": 0.337,
+      "step": 58
+    },
+    {
+      "epoch": 0.19865319865319866,
+      "grad_norm": 0.5703901648521423,
+      "learning_rate": 3.959731543624161e-05,
+      "loss": 0.3408,
+      "step": 59
+    },
+    {
+      "epoch": 0.20202020202020202,
+      "grad_norm": 0.7656762003898621,
+      "learning_rate": 4.026845637583892e-05,
+      "loss": 0.3206,
+      "step": 60
+    },
+    {
+      "epoch": 0.2053872053872054,
+      "grad_norm": 0.6210582852363586,
+      "learning_rate": 4.0939597315436244e-05,
+      "loss": 0.354,
+      "step": 61
+    },
+    {
+      "epoch": 0.20875420875420875,
+      "grad_norm": 0.6622840166091919,
+      "learning_rate": 4.161073825503356e-05,
+      "loss": 0.3439,
+      "step": 62
+    },
+    {
+      "epoch": 0.21212121212121213,
+      "grad_norm": 0.46426376700401306,
+      "learning_rate": 4.228187919463087e-05,
+      "loss": 0.3434,
+      "step": 63
+    },
+    {
+      "epoch": 0.21548821548821548,
+      "grad_norm": 0.38662126660346985,
+      "learning_rate": 4.295302013422819e-05,
+      "loss": 0.3362,
+      "step": 64
+    },
+    {
+      "epoch": 0.21885521885521886,
+      "grad_norm": 0.5812459588050842,
+      "learning_rate": 4.36241610738255e-05,
+      "loss": 0.323,
+      "step": 65
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 0.626932680606842,
+      "learning_rate": 4.4295302013422824e-05,
+      "loss": 0.3427,
+      "step": 66
+    },
+    {
+      "epoch": 0.2255892255892256,
+      "grad_norm": 0.5491658449172974,
+      "learning_rate": 4.496644295302014e-05,
+      "loss": 0.3406,
+      "step": 67
+    },
+    {
+      "epoch": 0.22895622895622897,
+      "grad_norm": 0.4023520052433014,
+      "learning_rate": 4.5637583892617453e-05,
+      "loss": 0.3328,
+      "step": 68
+    },
+    {
+      "epoch": 0.23232323232323232,
+      "grad_norm": 0.478535532951355,
+      "learning_rate": 4.630872483221477e-05,
+      "loss": 0.3402,
+      "step": 69
+    },
+    {
+      "epoch": 0.2356902356902357,
+      "grad_norm": 0.44869011640548706,
+      "learning_rate": 4.697986577181208e-05,
+      "loss": 0.3516,
+      "step": 70
+    },
+    {
+      "epoch": 0.23905723905723905,
+      "grad_norm": 0.4810108244419098,
+      "learning_rate": 4.76510067114094e-05,
+      "loss": 0.3411,
+      "step": 71
+    },
+    {
+      "epoch": 0.24242424242424243,
+      "grad_norm": 0.3956281542778015,
+      "learning_rate": 4.832214765100672e-05,
+      "loss": 0.3395,
+      "step": 72
+    },
+    {
+      "epoch": 0.24579124579124578,
+      "grad_norm": 0.40301939845085144,
+      "learning_rate": 4.8993288590604034e-05,
+      "loss": 0.3217,
+      "step": 73
+    },
+    {
+      "epoch": 0.24915824915824916,
+      "grad_norm": 0.44550034403800964,
+      "learning_rate": 4.966442953020135e-05,
+      "loss": 0.3257,
+      "step": 74
+    },
+    {
+      "epoch": 0.25252525252525254,
+      "grad_norm": 0.5890341997146606,
+      "learning_rate": 5.033557046979866e-05,
+      "loss": 0.3335,
+      "step": 75
+    },
+    {
+      "epoch": 0.2558922558922559,
+      "grad_norm": 0.8096022009849548,
+      "learning_rate": 5.100671140939598e-05,
+      "loss": 0.3421,
+      "step": 76
+    },
+    {
+      "epoch": 0.25925925925925924,
+      "grad_norm": 0.6044747829437256,
+      "learning_rate": 5.167785234899329e-05,
+      "loss": 0.3266,
+      "step": 77
+    },
+    {
+      "epoch": 0.26262626262626265,
+      "grad_norm": 0.5191451907157898,
+      "learning_rate": 5.234899328859061e-05,
+      "loss": 0.331,
+      "step": 78
+    },
+    {
+      "epoch": 0.265993265993266,
+      "grad_norm": 1.0799261331558228,
+      "learning_rate": 5.302013422818792e-05,
+      "loss": 0.3243,
+      "step": 79
+    },
+    {
+      "epoch": 0.26936026936026936,
+      "grad_norm": 5.513405799865723,
+      "learning_rate": 5.3691275167785237e-05,
+      "loss": 0.379,
+      "step": 80
+    },
+    {
+      "epoch": 0.2727272727272727,
+      "grad_norm": 0.673650860786438,
+      "learning_rate": 5.436241610738255e-05,
+      "loss": 0.3482,
+      "step": 81
+    },
+    {
+      "epoch": 0.2760942760942761,
+      "grad_norm": 1.1485897302627563,
+      "learning_rate": 5.5033557046979866e-05,
+      "loss": 0.3351,
+      "step": 82
+    },
+    {
+      "epoch": 0.27946127946127947,
+      "grad_norm": 0.5018780827522278,
+      "learning_rate": 5.570469798657718e-05,
+      "loss": 0.3077,
+      "step": 83
+    },
+    {
+      "epoch": 0.2828282828282828,
+      "grad_norm": 4.367802619934082,
+      "learning_rate": 5.6375838926174495e-05,
+      "loss": 0.3284,
+      "step": 84
+    },
+    {
+      "epoch": 0.28619528619528617,
+      "grad_norm": 33.46516036987305,
+      "learning_rate": 5.704697986577181e-05,
+      "loss": 1.0651,
+      "step": 85
+    },
+    {
+      "epoch": 0.2895622895622896,
+      "grad_norm": 91.36512756347656,
+      "learning_rate": 5.771812080536914e-05,
+      "loss": 1.7174,
+      "step": 86
+    },
+    {
+      "epoch": 0.29292929292929293,
+      "grad_norm": 9.666085243225098,
+      "learning_rate": 5.838926174496645e-05,
+      "loss": 0.5601,
+      "step": 87
+    },
+    {
+      "epoch": 0.2962962962962963,
+      "grad_norm": 8.608613967895508,
+      "learning_rate": 5.906040268456377e-05,
+      "loss": 0.3865,
+      "step": 88
+    },
+    {
+      "epoch": 0.2996632996632997,
+      "grad_norm": 3.025059223175049,
+      "learning_rate": 5.973154362416108e-05,
+      "loss": 0.358,
+      "step": 89
+    },
+    {
+      "epoch": 0.30303030303030304,
+      "grad_norm": 9.862916946411133,
+      "learning_rate": 6.04026845637584e-05,
+      "loss": 0.4464,
+      "step": 90
+    },
+    {
+      "epoch": 0.3063973063973064,
+      "grad_norm": 11.05635929107666,
+      "learning_rate": 6.107382550335571e-05,
+      "loss": 0.3977,
+      "step": 91
+    },
+    {
+      "epoch": 0.30976430976430974,
+      "grad_norm": 1.0226973295211792,
+      "learning_rate": 6.174496644295302e-05,
+      "loss": 0.3206,
+      "step": 92
+    },
+    {
+      "epoch": 0.31313131313131315,
+      "grad_norm": 1.007895827293396,
+      "learning_rate": 6.241610738255034e-05,
+      "loss": 0.3355,
+      "step": 93
+    },
+    {
+      "epoch": 0.3164983164983165,
+      "grad_norm": 1.5956454277038574,
+      "learning_rate": 6.308724832214765e-05,
+      "loss": 0.3408,
+      "step": 94
+    },
+    {
+      "epoch": 0.31986531986531985,
+      "grad_norm": 21.75948715209961,
+      "learning_rate": 6.375838926174497e-05,
+      "loss": 0.4627,
+      "step": 95
+    },
+    {
+      "epoch": 0.32323232323232326,
+      "grad_norm": 5.754608154296875,
+      "learning_rate": 6.442953020134228e-05,
+      "loss": 0.3818,
+      "step": 96
+    },
+    {
+      "epoch": 0.3265993265993266,
+      "grad_norm": 3.1888318061828613,
+      "learning_rate": 6.51006711409396e-05,
+      "loss": 0.3713,
+      "step": 97
+    },
+    {
+      "epoch": 0.32996632996632996,
+      "grad_norm": 4.586446762084961,
+      "learning_rate": 6.577181208053692e-05,
+      "loss": 0.3394,
+      "step": 98
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.9332061409950256,
+      "learning_rate": 6.644295302013423e-05,
+      "loss": 0.3267,
+      "step": 99
+    },
+    {
+      "epoch": 0.3367003367003367,
+      "grad_norm": 4.119638442993164,
+      "learning_rate": 6.711409395973155e-05,
+      "loss": 0.3825,
+      "step": 100
+    },
+    {
+      "epoch": 0.3367003367003367,
+      "eval_loss": 0.16432031989097595,
+      "eval_runtime": 33.0116,
+      "eval_samples_per_second": 30.292,
+      "eval_steps_per_second": 1.908,
+      "step": 100
+    },
+    {
+      "epoch": 0.3400673400673401,
+      "grad_norm": 1.244138240814209,
+      "learning_rate": 6.778523489932886e-05,
+      "loss": 0.3191,
+      "step": 101
+    },
+    {
+      "epoch": 0.3434343434343434,
+      "grad_norm": 4.564449310302734,
+      "learning_rate": 6.845637583892618e-05,
+      "loss": 0.3792,
+      "step": 102
+    },
+    {
+      "epoch": 0.3468013468013468,
+      "grad_norm": 71.92516326904297,
+      "learning_rate": 6.912751677852349e-05,
+      "loss": 0.9526,
+      "step": 103
+    },
+    {
+      "epoch": 0.3501683501683502,
+      "grad_norm": 6.8141374588012695,
+      "learning_rate": 6.979865771812081e-05,
+      "loss": 0.4132,
+      "step": 104
+    },
+    {
+      "epoch": 0.35353535353535354,
+      "grad_norm": 4.9158616065979,
+      "learning_rate": 7.046979865771812e-05,
+      "loss": 0.3847,
+      "step": 105
+    },
+    {
+      "epoch": 0.3569023569023569,
+      "grad_norm": 0.9838681221008301,
+      "learning_rate": 7.114093959731544e-05,
+      "loss": 0.335,
+      "step": 106
+    },
+    {
+      "epoch": 0.3602693602693603,
+      "grad_norm": 0.44024720788002014,
+      "learning_rate": 7.181208053691275e-05,
+      "loss": 0.3159,
+      "step": 107
+    },
+    {
+      "epoch": 0.36363636363636365,
+      "grad_norm": 0.5798377394676208,
+      "learning_rate": 7.248322147651007e-05,
+      "loss": 0.3308,
+      "step": 108
+    },
+    {
+      "epoch": 0.367003367003367,
+      "grad_norm": 0.5650081038475037,
+      "learning_rate": 7.315436241610739e-05,
+      "loss": 0.3161,
+      "step": 109
+    },
+    {
+      "epoch": 0.37037037037037035,
+      "grad_norm": 0.5149471163749695,
+      "learning_rate": 7.382550335570471e-05,
+      "loss": 0.3291,
+      "step": 110
+    },
+    {
+      "epoch": 0.37373737373737376,
+      "grad_norm": 0.4448802173137665,
+      "learning_rate": 7.449664429530202e-05,
+      "loss": 0.3145,
+      "step": 111
+    },
+    {
+      "epoch": 0.3771043771043771,
+      "grad_norm": 0.5278413891792297,
+      "learning_rate": 7.516778523489934e-05,
+      "loss": 0.3296,
+      "step": 112
+    },
+    {
+      "epoch": 0.38047138047138046,
+      "grad_norm": 0.455289363861084,
+      "learning_rate": 7.583892617449665e-05,
+      "loss": 0.318,
+      "step": 113
+    },
+    {
+      "epoch": 0.3838383838383838,
+      "grad_norm": 0.5316647291183472,
+      "learning_rate": 7.651006711409397e-05,
+      "loss": 0.3117,
+      "step": 114
+    },
+    {
+      "epoch": 0.3872053872053872,
+      "grad_norm": 0.43862929940223694,
+      "learning_rate": 7.718120805369128e-05,
+      "loss": 0.3062,
+      "step": 115
+    },
+    {
+      "epoch": 0.39057239057239057,
+      "grad_norm": 12.535127639770508,
+      "learning_rate": 7.78523489932886e-05,
+      "loss": 0.3557,
+      "step": 116
+    },
+    {
+      "epoch": 0.3939393939393939,
+      "grad_norm": 15.351152420043945,
+      "learning_rate": 7.852348993288591e-05,
+      "loss": 0.5181,
+      "step": 117
+    },
+    {
+      "epoch": 0.39730639730639733,
+      "grad_norm": 11.918878555297852,
+      "learning_rate": 7.919463087248322e-05,
+      "loss": 0.4005,
+      "step": 118
+    },
+    {
+      "epoch": 0.4006734006734007,
+      "grad_norm": 9.800668716430664,
+      "learning_rate": 7.986577181208054e-05,
+      "loss": 0.422,
+      "step": 119
+    },
+    {
+      "epoch": 0.40404040404040403,
+      "grad_norm": 16.235355377197266,
+      "learning_rate": 8.053691275167784e-05,
+      "loss": 0.4051,
+      "step": 120
+    },
+    {
+      "epoch": 0.4074074074074074,
+      "grad_norm": 1.8551958799362183,
+      "learning_rate": 8.120805369127518e-05,
+      "loss": 0.3506,
+      "step": 121
+    },
+    {
+      "epoch": 0.4107744107744108,
+      "grad_norm": 3.990302562713623,
+      "learning_rate": 8.187919463087249e-05,
+      "loss": 0.3318,
+      "step": 122
+    },
+    {
+      "epoch": 0.41414141414141414,
+      "grad_norm": 22.28190040588379,
+      "learning_rate": 8.255033557046981e-05,
+      "loss": 0.4316,
+      "step": 123
+    },
+    {
+      "epoch": 0.4175084175084175,
+      "grad_norm": 1.9532949924468994,
+      "learning_rate": 8.322147651006712e-05,
+      "loss": 0.3596,
+      "step": 124
+    },
+    {
+      "epoch": 0.4208754208754209,
+      "grad_norm": 0.8453232645988464,
+      "learning_rate": 8.389261744966444e-05,
+      "loss": 0.3473,
+      "step": 125
+    },
+    {
+      "epoch": 0.42424242424242425,
+      "grad_norm": 3.7085459232330322,
+      "learning_rate": 8.456375838926175e-05,
+      "loss": 0.3527,
+      "step": 126
+    },
+    {
+      "epoch": 0.4276094276094276,
+      "grad_norm": 1.9306743144989014,
+      "learning_rate": 8.523489932885907e-05,
+      "loss": 0.3415,
+      "step": 127
+    },
+    {
+      "epoch": 0.43097643097643096,
+      "grad_norm": 5.023862361907959,
+      "learning_rate": 8.590604026845638e-05,
+      "loss": 0.3644,
+      "step": 128
+    },
+    {
+      "epoch": 0.43434343434343436,
+      "grad_norm": 4.241243362426758,
+      "learning_rate": 8.65771812080537e-05,
+      "loss": 0.4321,
+      "step": 129
+    },
+    {
+      "epoch": 0.4377104377104377,
+      "grad_norm": 1.7396281957626343,
+      "learning_rate": 8.7248322147651e-05,
+      "loss": 0.3334,
+      "step": 130
+    },
+    {
+      "epoch": 0.44107744107744107,
+      "grad_norm": 8.367612838745117,
+      "learning_rate": 8.791946308724833e-05,
+      "loss": 0.3571,
+      "step": 131
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 7.692532539367676,
+      "learning_rate": 8.859060402684565e-05,
+      "loss": 0.5196,
+      "step": 132
+    },
+    {
+      "epoch": 0.4478114478114478,
+      "grad_norm": 12.191128730773926,
+      "learning_rate": 8.926174496644296e-05,
+      "loss": 0.6991,
+      "step": 133
+    },
+    {
+      "epoch": 0.4511784511784512,
+      "grad_norm": 7.570639133453369,
+      "learning_rate": 8.993288590604028e-05,
+      "loss": 0.4818,
+      "step": 134
+    },
+    {
+      "epoch": 0.45454545454545453,
+      "grad_norm": 1.7189193964004517,
+      "learning_rate": 9.060402684563759e-05,
+      "loss": 0.3728,
+      "step": 135
+    },
+    {
+      "epoch": 0.45791245791245794,
+      "grad_norm": 9.100985527038574,
+      "learning_rate": 9.127516778523491e-05,
+      "loss": 0.3869,
+      "step": 136
+    },
+    {
+      "epoch": 0.4612794612794613,
+      "grad_norm": 9.76489543914795,
+      "learning_rate": 9.194630872483221e-05,
+      "loss": 0.3861,
+      "step": 137
+    },
+    {
+      "epoch": 0.46464646464646464,
+      "grad_norm": 3.834136962890625,
+      "learning_rate": 9.261744966442954e-05,
+      "loss": 0.4222,
+      "step": 138
+    },
+    {
+      "epoch": 0.468013468013468,
+      "grad_norm": 22.2440242767334,
+      "learning_rate": 9.328859060402684e-05,
+      "loss": 0.7935,
+      "step": 139
+    },
+    {
+      "epoch": 0.4713804713804714,
+      "grad_norm": 1.4633365869522095,
+      "learning_rate": 9.395973154362417e-05,
+      "loss": 0.347,
+      "step": 140
+    },
+    {
+      "epoch": 0.47474747474747475,
+      "grad_norm": 1.9224159717559814,
+      "learning_rate": 9.463087248322147e-05,
+      "loss": 0.3299,
+      "step": 141
+    },
+    {
+      "epoch": 0.4781144781144781,
+      "grad_norm": 3.4107277393341064,
+      "learning_rate": 9.53020134228188e-05,
+      "loss": 0.4199,
+      "step": 142
+    },
+    {
+      "epoch": 0.48148148148148145,
+      "grad_norm": 1.4255735874176025,
+      "learning_rate": 9.59731543624161e-05,
+      "loss": 0.3559,
+      "step": 143
+    },
+    {
+      "epoch": 0.48484848484848486,
+      "grad_norm": 1.4576934576034546,
+      "learning_rate": 9.664429530201344e-05,
+      "loss": 0.3274,
+      "step": 144
+    },
+    {
+      "epoch": 0.4882154882154882,
+      "grad_norm": 1.3531242609024048,
+      "learning_rate": 9.731543624161075e-05,
+      "loss": 0.3417,
+      "step": 145
+    },
+    {
+      "epoch": 0.49158249158249157,
+      "grad_norm": 13.97393798828125,
+      "learning_rate": 9.798657718120807e-05,
+      "loss": 0.3434,
+      "step": 146
+    },
+    {
+      "epoch": 0.494949494949495,
+      "grad_norm": 0.8413457870483398,
+      "learning_rate": 9.865771812080538e-05,
+      "loss": 0.3224,
+      "step": 147
+    },
+    {
+      "epoch": 0.4983164983164983,
+      "grad_norm": 0.41903650760650635,
+      "learning_rate": 9.93288590604027e-05,
+      "loss": 0.3197,
+      "step": 148
+    },
+    {
+      "epoch": 0.5016835016835017,
+      "grad_norm": 1.3428220748901367,
+      "learning_rate": 0.0001,
+      "loss": 0.3184,
+      "step": 149
+    },
+    {
+      "epoch": 0.5050505050505051,
+      "grad_norm": 0.497494637966156,
+      "learning_rate": 9.9999861762256e-05,
+      "loss": 0.3064,
+      "step": 150
+    },
+    {
+      "epoch": 0.5084175084175084,
+      "grad_norm": 0.5110116600990295,
+      "learning_rate": 9.999944704978836e-05,
+      "loss": 0.3195,
+      "step": 151
+    },
+    {
+      "epoch": 0.5117845117845118,
+      "grad_norm": 0.4883813261985779,
+      "learning_rate": 9.999875586489024e-05,
+      "loss": 0.292,
+      "step": 152
+    },
+    {
+      "epoch": 0.5151515151515151,
+      "grad_norm": 0.44456565380096436,
+      "learning_rate": 9.999778821138357e-05,
+      "loss": 0.3084,
+      "step": 153
+    },
+    {
+      "epoch": 0.5185185185185185,
+      "grad_norm": 0.5006658434867859,
+      "learning_rate": 9.999654409461896e-05,
+      "loss": 0.3031,
+      "step": 154
+    },
+    {
+      "epoch": 0.5218855218855218,
+      "grad_norm": 0.4398713707923889,
+      "learning_rate": 9.999502352147583e-05,
+      "loss": 0.3178,
+      "step": 155
+    },
+    {
+      "epoch": 0.5252525252525253,
+      "grad_norm": 0.4853643476963043,
+      "learning_rate": 9.999322650036214e-05,
+      "loss": 0.3195,
+      "step": 156
+    },
+    {
+      "epoch": 0.5286195286195287,
+      "grad_norm": 0.4636339545249939,
+      "learning_rate": 9.999115304121457e-05,
+      "loss": 0.3052,
+      "step": 157
+    },
+    {
+      "epoch": 0.531986531986532,
+      "grad_norm": 0.525205671787262,
+      "learning_rate": 9.998880315549834e-05,
+      "loss": 0.3133,
+      "step": 158
+    },
+    {
+      "epoch": 0.5353535353535354,
+      "grad_norm": 0.40854206681251526,
+      "learning_rate": 9.998617685620714e-05,
+      "loss": 0.3076,
+      "step": 159
+    },
+    {
+      "epoch": 0.5387205387205387,
+      "grad_norm": 0.5355719327926636,
+      "learning_rate": 9.998327415786315e-05,
+      "loss": 0.3052,
+      "step": 160
+    },
+    {
+      "epoch": 0.5420875420875421,
+      "grad_norm": 0.3861645460128784,
+      "learning_rate": 9.998009507651684e-05,
+      "loss": 0.3099,
+      "step": 161
+    },
+    {
+      "epoch": 0.5454545454545454,
+      "grad_norm": 0.5338487029075623,
+      "learning_rate": 9.997663962974697e-05,
+      "loss": 0.3052,
+      "step": 162
+    },
+    {
+      "epoch": 0.5488215488215489,
+      "grad_norm": 0.45219364762306213,
+      "learning_rate": 9.997290783666049e-05,
+      "loss": 0.2948,
+      "step": 163
+    },
+    {
+      "epoch": 0.5521885521885522,
+      "grad_norm": 0.5037462711334229,
+      "learning_rate": 9.996889971789235e-05,
+      "loss": 0.3019,
+      "step": 164
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 0.3949816823005676,
+      "learning_rate": 9.996461529560553e-05,
+      "loss": 0.3028,
+      "step": 165
+    },
+    {
+      "epoch": 0.5589225589225589,
+      "grad_norm": 0.3921789824962616,
+      "learning_rate": 9.996005459349074e-05,
+      "loss": 0.2982,
+      "step": 166
+    },
+    {
+      "epoch": 0.5622895622895623,
+      "grad_norm": 0.4122919738292694,
+      "learning_rate": 9.995521763676645e-05,
+      "loss": 0.3071,
+      "step": 167
+    },
+    {
+      "epoch": 0.5656565656565656,
+      "grad_norm": 0.4212525188922882,
+      "learning_rate": 9.995010445217867e-05,
+      "loss": 0.3086,
+      "step": 168
+    },
+    {
+      "epoch": 0.569023569023569,
+      "grad_norm": 0.3997049033641815,
+      "learning_rate": 9.994471506800079e-05,
+      "loss": 0.2957,
+      "step": 169
+    },
+    {
+      "epoch": 0.5723905723905723,
+      "grad_norm": 0.34380048513412476,
+      "learning_rate": 9.993904951403344e-05,
+      "loss": 0.3122,
+      "step": 170
+    },
+    {
+      "epoch": 0.5757575757575758,
+      "grad_norm": 0.40532243251800537,
+      "learning_rate": 9.99331078216044e-05,
+      "loss": 0.3055,
+      "step": 171
+    },
+    {
+      "epoch": 0.5791245791245792,
+      "grad_norm": 0.4095707833766937,
+      "learning_rate": 9.992689002356828e-05,
+      "loss": 0.2868,
+      "step": 172
+    },
+    {
+      "epoch": 0.5824915824915825,
+      "grad_norm": 0.41159185767173767,
+      "learning_rate": 9.992039615430648e-05,
+      "loss": 0.318,
+      "step": 173
+    },
+    {
+      "epoch": 0.5858585858585859,
+      "grad_norm": 0.3728049397468567,
+      "learning_rate": 9.991362624972688e-05,
+      "loss": 0.309,
+      "step": 174
+    },
+    {
+      "epoch": 0.5892255892255892,
+      "grad_norm": 0.3249180018901825,
+      "learning_rate": 9.990658034726379e-05,
+      "loss": 0.2818,
+      "step": 175
+    },
+    {
+      "epoch": 0.5925925925925926,
+      "grad_norm": 0.35090282559394836,
+      "learning_rate": 9.989925848587756e-05,
+      "loss": 0.2839,
+      "step": 176
+    },
+    {
+      "epoch": 0.5959595959595959,
+      "grad_norm": 0.3364333212375641,
+      "learning_rate": 9.989166070605447e-05,
+      "loss": 0.3063,
+      "step": 177
+    },
+    {
+      "epoch": 0.5993265993265994,
+      "grad_norm": 0.4135960340499878,
+      "learning_rate": 9.988378704980656e-05,
+      "loss": 0.3085,
+      "step": 178
+    },
+    {
+      "epoch": 0.6026936026936027,
+      "grad_norm": 0.35615649819374084,
+      "learning_rate": 9.987563756067129e-05,
+      "loss": 0.2955,
+      "step": 179
+    },
+    {
+      "epoch": 0.6060606060606061,
+      "grad_norm": 0.3038477897644043,
+      "learning_rate": 9.986721228371129e-05,
+      "loss": 0.291,
+      "step": 180
+    },
+    {
+      "epoch": 0.6094276094276094,
+      "grad_norm": 0.4663616120815277,
+      "learning_rate": 9.985851126551428e-05,
+      "loss": 0.3043,
+      "step": 181
+    },
+    {
+      "epoch": 0.6127946127946128,
+      "grad_norm": 0.42187029123306274,
+      "learning_rate": 9.984953455419258e-05,
+      "loss": 0.2747,
+      "step": 182
+    },
+    {
+      "epoch": 0.6161616161616161,
+      "grad_norm": 0.4150826334953308,
+      "learning_rate": 9.9840282199383e-05,
+      "loss": 0.2854,
+      "step": 183
+    },
+    {
+      "epoch": 0.6195286195286195,
+      "grad_norm": 0.36844050884246826,
+      "learning_rate": 9.983075425224653e-05,
+      "loss": 0.2848,
+      "step": 184
+    },
+    {
+      "epoch": 0.622895622895623,
+      "grad_norm": 0.4171907603740692,
+      "learning_rate": 9.982095076546807e-05,
+      "loss": 0.3003,
+      "step": 185
+    },
+    {
+      "epoch": 0.6262626262626263,
+      "grad_norm": 0.3620002269744873,
+      "learning_rate": 9.981087179325608e-05,
+      "loss": 0.3043,
+      "step": 186
+    },
+    {
+      "epoch": 0.6296296296296297,
+      "grad_norm": 0.40760573744773865,
+      "learning_rate": 9.980051739134233e-05,
+      "loss": 0.3059,
+      "step": 187
+    },
+    {
+      "epoch": 0.632996632996633,
+      "grad_norm": 0.32069963216781616,
+      "learning_rate": 9.978988761698161e-05,
+      "loss": 0.2947,
+      "step": 188
+    },
+    {
+      "epoch": 0.6363636363636364,
+      "grad_norm": 0.3327488303184509,
+      "learning_rate": 9.977898252895134e-05,
+      "loss": 0.2805,
+      "step": 189
+    },
+    {
+      "epoch": 0.6397306397306397,
+      "grad_norm": 0.3968160152435303,
+      "learning_rate": 9.976780218755131e-05,
+      "loss": 0.2891,
+      "step": 190
+    },
+    {
+      "epoch": 0.6430976430976431,
+      "grad_norm": 0.4018626809120178,
+      "learning_rate": 9.975634665460332e-05,
+      "loss": 0.2965,
+      "step": 191
+    },
+    {
+      "epoch": 0.6464646464646465,
+      "grad_norm": 0.37805649638175964,
+      "learning_rate": 9.974461599345088e-05,
+      "loss": 0.3008,
+      "step": 192
+    },
+    {
+      "epoch": 0.6498316498316499,
+      "grad_norm": 0.44425806403160095,
+      "learning_rate": 9.973261026895877e-05,
+      "loss": 0.2921,
+      "step": 193
+    },
+    {
+      "epoch": 0.6531986531986532,
+      "grad_norm": 0.375931054353714,
+      "learning_rate": 9.972032954751279e-05,
+      "loss": 0.296,
+      "step": 194
+    },
+    {
+      "epoch": 0.6565656565656566,
+      "grad_norm": 0.44635701179504395,
+      "learning_rate": 9.970777389701926e-05,
+      "loss": 0.29,
+      "step": 195
+    },
+    {
+      "epoch": 0.6599326599326599,
+      "grad_norm": 0.28897619247436523,
+      "learning_rate": 9.969494338690481e-05,
+      "loss": 0.2895,
+      "step": 196
+    },
+    {
+      "epoch": 0.6632996632996633,
+      "grad_norm": 0.4542882740497589,
+      "learning_rate": 9.968183808811586e-05,
+      "loss": 0.2887,
+      "step": 197
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.3715568780899048,
+      "learning_rate": 9.966845807311829e-05,
+      "loss": 0.3038,
+      "step": 198
+    },
+    {
+      "epoch": 0.67003367003367,
+      "grad_norm": 0.36940261721611023,
+      "learning_rate": 9.965480341589701e-05,
+      "loss": 0.2934,
+      "step": 199
+    },
+    {
+      "epoch": 0.6734006734006734,
+      "grad_norm": 0.43046656250953674,
+      "learning_rate": 9.96408741919556e-05,
+      "loss": 0.2951,
+      "step": 200
+    },
+    {
+      "epoch": 0.6734006734006734,
+      "eval_loss": 0.1412619948387146,
+      "eval_runtime": 32.6268,
+      "eval_samples_per_second": 30.65,
+      "eval_steps_per_second": 1.931,
+      "step": 200
+    },
+    {
+      "epoch": 0.6767676767676768,
+      "grad_norm": 0.37590286135673523,
+      "learning_rate": 9.962667047831584e-05,
+      "loss": 0.2922,
+      "step": 201
+    },
+    {
+      "epoch": 0.6801346801346801,
+      "grad_norm": 0.3418475389480591,
+      "learning_rate": 9.961219235351729e-05,
+      "loss": 0.2732,
+      "step": 202
+    },
+    {
+      "epoch": 0.6835016835016835,
+      "grad_norm": 0.3605377674102783,
+      "learning_rate": 9.95974398976169e-05,
+      "loss": 0.2882,
+      "step": 203
+    },
+    {
+      "epoch": 0.6868686868686869,
+      "grad_norm": 0.40477219223976135,
+      "learning_rate": 9.958241319218848e-05,
+      "loss": 0.2859,
+      "step": 204
+    },
+    {
+      "epoch": 0.6902356902356902,
+      "grad_norm": 0.4034753143787384,
+      "learning_rate": 9.95671123203224e-05,
+      "loss": 0.2984,
+      "step": 205
+    },
+    {
+      "epoch": 0.6936026936026936,
+      "grad_norm": 0.3650234043598175,
+      "learning_rate": 9.955153736662493e-05,
+      "loss": 0.2772,
+      "step": 206
+    },
+    {
+      "epoch": 0.696969696969697,
+      "grad_norm": 0.47222810983657837,
+      "learning_rate": 9.953568841721797e-05,
+      "loss": 0.28,
+      "step": 207
+    },
+    {
+      "epoch": 0.7003367003367004,
+      "grad_norm": 0.3858278691768646,
+      "learning_rate": 9.95195655597384e-05,
+      "loss": 0.2815,
+      "step": 208
+    },
+    {
+      "epoch": 0.7037037037037037,
+      "grad_norm": 0.4259450435638428,
+      "learning_rate": 9.950316888333775e-05,
+      "loss": 0.2965,
+      "step": 209
+    },
+    {
+      "epoch": 0.7070707070707071,
+      "grad_norm": 0.4309611916542053,
+      "learning_rate": 9.948649847868159e-05,
+      "loss": 0.2766,
+      "step": 210
+    },
+    {
+      "epoch": 0.7104377104377104,
+      "grad_norm": 0.4742699861526489,
+      "learning_rate": 9.946955443794908e-05,
+      "loss": 0.2859,
+      "step": 211
+    },
+    {
+      "epoch": 0.7138047138047138,
+      "grad_norm": 0.4079667329788208,
+      "learning_rate": 9.945233685483246e-05,
+      "loss": 0.283,
+      "step": 212
+    },
+    {
+      "epoch": 0.7171717171717171,
+      "grad_norm": 0.42072775959968567,
+      "learning_rate": 9.943484582453653e-05,
+      "loss": 0.298,
+      "step": 213
+    },
+    {
+      "epoch": 0.7205387205387206,
+      "grad_norm": 0.43136894702911377,
+      "learning_rate": 9.941708144377813e-05,
+      "loss": 0.2693,
+      "step": 214
+    },
+    {
+      "epoch": 0.7239057239057239,
+      "grad_norm": 0.42598387598991394,
+      "learning_rate": 9.939904381078553e-05,
+      "loss": 0.2836,
+      "step": 215
+    },
+    {
+      "epoch": 0.7272727272727273,
+      "grad_norm": 0.40432655811309814,
+      "learning_rate": 9.938073302529804e-05,
+      "loss": 0.2844,
+      "step": 216
+    },
+    {
+      "epoch": 0.7306397306397306,
+      "grad_norm": 0.3417808413505554,
+      "learning_rate": 9.93621491885653e-05,
+      "loss": 0.2849,
+      "step": 217
+    },
+    {
+      "epoch": 0.734006734006734,
+      "grad_norm": 0.35036516189575195,
+      "learning_rate": 9.934329240334686e-05,
+      "loss": 0.2619,
+      "step": 218
+    },
+    {
+      "epoch": 0.7373737373737373,
+      "grad_norm": 0.38956964015960693,
+      "learning_rate": 9.932416277391143e-05,
+      "loss": 0.2802,
+      "step": 219
+    },
+    {
+      "epoch": 0.7407407407407407,
+      "grad_norm": 0.36884164810180664,
+      "learning_rate": 9.930476040603653e-05,
+      "loss": 0.2961,
+      "step": 220
+    },
+    {
+      "epoch": 0.7441077441077442,
+      "grad_norm": 0.4145122468471527,
+      "learning_rate": 9.928508540700774e-05,
+      "loss": 0.2789,
+      "step": 221
+    },
+    {
+      "epoch": 0.7474747474747475,
+      "grad_norm": 0.36580273509025574,
+      "learning_rate": 9.926513788561816e-05,
+      "loss": 0.2824,
+      "step": 222
+    },
+    {
+      "epoch": 0.7508417508417509,
+      "grad_norm": 0.2912370264530182,
+      "learning_rate": 9.924491795216777e-05,
+      "loss": 0.2811,
+      "step": 223
+    },
+    {
+      "epoch": 0.7542087542087542,
+      "grad_norm": 0.480868399143219,
+      "learning_rate": 9.922442571846293e-05,
+      "loss": 0.2853,
+      "step": 224
+    },
+    {
+      "epoch": 0.7575757575757576,
+      "grad_norm": 0.3405955135822296,
+      "learning_rate": 9.920366129781564e-05,
+      "loss": 0.2908,
+      "step": 225
+    },
+    {
+      "epoch": 0.7609427609427609,
+      "grad_norm": 0.34814175963401794,
+      "learning_rate": 9.918262480504295e-05,
+      "loss": 0.2923,
+      "step": 226
+    },
+    {
+      "epoch": 0.7643097643097643,
+      "grad_norm": 0.36179670691490173,
+      "learning_rate": 9.916131635646635e-05,
+      "loss": 0.276,
+      "step": 227
+    },
+    {
+      "epoch": 0.7676767676767676,
+      "grad_norm": 0.3848663568496704,
+      "learning_rate": 9.913973606991113e-05,
+      "loss": 0.264,
+      "step": 228
+    },
+    {
+      "epoch": 0.7710437710437711,
+      "grad_norm": 0.4516603648662567,
+      "learning_rate": 9.911788406470569e-05,
+      "loss": 0.2854,
+      "step": 229
+    },
+    {
+      "epoch": 0.7744107744107744,
+      "grad_norm": 0.4367293119430542,
+      "learning_rate": 9.90957604616809e-05,
+      "loss": 0.2802,
+      "step": 230
+    },
+    {
+      "epoch": 0.7777777777777778,
+      "grad_norm": 0.41222095489501953,
+      "learning_rate": 9.907336538316944e-05,
+      "loss": 0.275,
+      "step": 231
+    },
+    {
+      "epoch": 0.7811447811447811,
+      "grad_norm": 0.4176308810710907,
+      "learning_rate": 9.905069895300514e-05,
+      "loss": 0.2854,
+      "step": 232
+    },
+    {
+      "epoch": 0.7845117845117845,
+      "grad_norm": 0.4087597131729126,
+      "learning_rate": 9.902776129652223e-05,
+      "loss": 0.2868,
+      "step": 233
+    },
+    {
+      "epoch": 0.7878787878787878,
+      "grad_norm": 0.41595739126205444,
+      "learning_rate": 9.900455254055467e-05,
+      "loss": 0.2835,
+      "step": 234
+    },
+    {
+      "epoch": 0.7912457912457912,
+      "grad_norm": 0.5036376118659973,
+      "learning_rate": 9.898107281343556e-05,
+      "loss": 0.2775,
+      "step": 235
+    },
+    {
+      "epoch": 0.7946127946127947,
+      "grad_norm": 0.46533098816871643,
+      "learning_rate": 9.895732224499625e-05,
+      "loss": 0.285,
+      "step": 236
+    },
+    {
+      "epoch": 0.797979797979798,
+      "grad_norm": 0.4155175983905792,
+      "learning_rate": 9.893330096656574e-05,
+      "loss": 0.2877,
+      "step": 237
+    },
+    {
+      "epoch": 0.8013468013468014,
+      "grad_norm": 0.34219178557395935,
+      "learning_rate": 9.890900911096992e-05,
+      "loss": 0.2751,
+      "step": 238
+    },
+    {
+      "epoch": 0.8047138047138047,
+      "grad_norm": 0.39359742403030396,
+      "learning_rate": 9.888444681253086e-05,
+      "loss": 0.2758,
+      "step": 239
+    },
+    {
+      "epoch": 0.8080808080808081,
+      "grad_norm": 0.3699426054954529,
+      "learning_rate": 9.885961420706602e-05,
+      "loss": 0.2758,
+      "step": 240
+    },
+    {
+      "epoch": 0.8114478114478114,
+      "grad_norm": 0.3353779911994934,
+      "learning_rate": 9.883451143188753e-05,
+      "loss": 0.2891,
+      "step": 241
+    },
+    {
+      "epoch": 0.8148148148148148,
+      "grad_norm": 0.3476376235485077,
+      "learning_rate": 9.880913862580145e-05,
+      "loss": 0.2699,
+      "step": 242
+    },
+    {
+      "epoch": 0.8181818181818182,
+      "grad_norm": 0.37724611163139343,
+      "learning_rate": 9.878349592910692e-05,
+      "loss": 0.2759,
+      "step": 243
+    },
+    {
+      "epoch": 0.8215488215488216,
+      "grad_norm": 0.3629307150840759,
+      "learning_rate": 9.875758348359552e-05,
+      "loss": 0.2741,
+      "step": 244
+    },
+    {
+      "epoch": 0.8249158249158249,
+      "grad_norm": 0.35653156042099,
+      "learning_rate": 9.873140143255036e-05,
+      "loss": 0.2717,
+      "step": 245
+    },
+    {
+      "epoch": 0.8282828282828283,
+      "grad_norm": 0.37418127059936523,
+      "learning_rate": 9.870494992074533e-05,
+      "loss": 0.2743,
+      "step": 246
+    },
+    {
+      "epoch": 0.8316498316498316,
+      "grad_norm": 0.3299245238304138,
+      "learning_rate": 9.867822909444434e-05,
+      "loss": 0.2751,
+      "step": 247
+    },
+    {
+      "epoch": 0.835016835016835,
+      "grad_norm": 0.3463493585586548,
+      "learning_rate": 9.865123910140046e-05,
+      "loss": 0.2778,
+      "step": 248
+    },
+    {
+      "epoch": 0.8383838383838383,
+      "grad_norm": 0.5460504293441772,
+      "learning_rate": 9.862398009085511e-05,
+      "loss": 0.2799,
+      "step": 249
+    },
+    {
+      "epoch": 0.8417508417508418,
+      "grad_norm": 0.3625568151473999,
+      "learning_rate": 9.859645221353725e-05,
+      "loss": 0.2641,
+      "step": 250
+    },
+    {
+      "epoch": 0.8451178451178452,
+      "grad_norm": 0.37554433941841125,
+      "learning_rate": 9.856865562166256e-05,
+      "loss": 0.2625,
+      "step": 251
+    },
+    {
+      "epoch": 0.8484848484848485,
+      "grad_norm": 0.4169292449951172,
+      "learning_rate": 9.854059046893257e-05,
+      "loss": 0.2754,
+      "step": 252
+    },
+    {
+      "epoch": 0.8518518518518519,
+      "grad_norm": 0.4144760072231293,
+      "learning_rate": 9.85122569105338e-05,
+      "loss": 0.2764,
+      "step": 253
+    },
+    {
+      "epoch": 0.8552188552188552,
+      "grad_norm": 0.3319230079650879,
+      "learning_rate": 9.848365510313695e-05,
+      "loss": 0.2812,
+      "step": 254
+    },
+    {
+      "epoch": 0.8585858585858586,
+      "grad_norm": 0.31481847167015076,
+      "learning_rate": 9.845478520489599e-05,
+      "loss": 0.2534,
+      "step": 255
+    },
+    {
+      "epoch": 0.8619528619528619,
+      "grad_norm": 0.37751275300979614,
+      "learning_rate": 9.842564737544731e-05,
+      "loss": 0.2796,
+      "step": 256
+    },
+    {
+      "epoch": 0.8653198653198653,
+      "grad_norm": 0.4334275424480438,
+      "learning_rate": 9.83962417759088e-05,
+      "loss": 0.2829,
+      "step": 257
+    },
+    {
+      "epoch": 0.8686868686868687,
+      "grad_norm": 0.4017227590084076,
+      "learning_rate": 9.836656856887903e-05,
+      "loss": 0.2667,
+      "step": 258
+    },
+    {
+      "epoch": 0.8720538720538721,
+      "grad_norm": 0.42103585600852966,
+      "learning_rate": 9.833662791843627e-05,
+      "loss": 0.2631,
+      "step": 259
+    },
+    {
+      "epoch": 0.8754208754208754,
+      "grad_norm": 0.34120380878448486,
+      "learning_rate": 9.830641999013768e-05,
+      "loss": 0.2613,
+      "step": 260
+    },
+    {
+      "epoch": 0.8787878787878788,
+      "grad_norm": 0.49804946780204773,
+      "learning_rate": 9.827594495101823e-05,
+      "loss": 0.2675,
+      "step": 261
+    },
+    {
+      "epoch": 0.8821548821548821,
+      "grad_norm": 0.3861115574836731,
+      "learning_rate": 9.824520296959001e-05,
+      "loss": 0.2708,
+      "step": 262
+    },
+    {
+      "epoch": 0.8855218855218855,
+      "grad_norm": 0.39489248394966125,
+      "learning_rate": 9.821419421584107e-05,
+      "loss": 0.2831,
+      "step": 263
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.3506355881690979,
+      "learning_rate": 9.818291886123463e-05,
+      "loss": 0.2784,
+      "step": 264
+    },
+    {
+      "epoch": 0.8922558922558923,
+      "grad_norm": 0.35518354177474976,
+      "learning_rate": 9.815137707870805e-05,
+      "loss": 0.2671,
+      "step": 265
+    },
+    {
+      "epoch": 0.8956228956228957,
+      "grad_norm": 0.35561174154281616,
+      "learning_rate": 9.811956904267195e-05,
+      "loss": 0.2784,
+      "step": 266
+    },
+    {
+      "epoch": 0.898989898989899,
+      "grad_norm": 0.3117510974407196,
+      "learning_rate": 9.808749492900918e-05,
+      "loss": 0.2824,
+      "step": 267
+    },
+    {
+      "epoch": 0.9023569023569024,
+      "grad_norm": 0.34295716881752014,
+      "learning_rate": 9.805515491507382e-05,
+      "loss": 0.2704,
+      "step": 268
+    },
+    {
+      "epoch": 0.9057239057239057,
+      "grad_norm": 0.3531172275543213,
+      "learning_rate": 9.802254917969032e-05,
+      "loss": 0.2712,
+      "step": 269
+    },
+    {
+      "epoch": 0.9090909090909091,
+      "grad_norm": 0.3834570646286011,
+      "learning_rate": 9.798967790315244e-05,
+      "loss": 0.285,
+      "step": 270
+    },
+    {
+      "epoch": 0.9124579124579124,
+      "grad_norm": 0.2960718274116516,
+      "learning_rate": 9.795654126722217e-05,
+      "loss": 0.2786,
+      "step": 271
+    },
+    {
+      "epoch": 0.9158249158249159,
+      "grad_norm": 0.3393447697162628,
+      "learning_rate": 9.79231394551289e-05,
+      "loss": 0.2841,
+      "step": 272
+    },
+    {
+      "epoch": 0.9191919191919192,
+      "grad_norm": 0.313174843788147,
+      "learning_rate": 9.788947265156827e-05,
+      "loss": 0.2605,
+      "step": 273
+    },
+    {
+      "epoch": 0.9225589225589226,
+      "grad_norm": 0.3173663914203644,
+      "learning_rate": 9.785554104270118e-05,
+      "loss": 0.2564,
+      "step": 274
+    },
+    {
+      "epoch": 0.9259259259259259,
+      "grad_norm": 0.36193931102752686,
+      "learning_rate": 9.782134481615281e-05,
+      "loss": 0.2659,
+      "step": 275
+    },
+    {
+      "epoch": 0.9292929292929293,
+      "grad_norm": 0.3565308451652527,
+      "learning_rate": 9.778688416101154e-05,
+      "loss": 0.2734,
+      "step": 276
+    },
+    {
+      "epoch": 0.9326599326599326,
+      "grad_norm": 0.32475653290748596,
+      "learning_rate": 9.775215926782788e-05,
+      "loss": 0.2754,
+      "step": 277
+    },
+    {
+      "epoch": 0.936026936026936,
+      "grad_norm": 0.4006199836730957,
+      "learning_rate": 9.771717032861346e-05,
+      "loss": 0.2662,
+      "step": 278
+    },
+    {
+      "epoch": 0.9393939393939394,
+      "grad_norm": 0.31218966841697693,
+      "learning_rate": 9.768191753683998e-05,
+      "loss": 0.2442,
+      "step": 279
+    },
+    {
+      "epoch": 0.9427609427609428,
+      "grad_norm": 0.3531815707683563,
+      "learning_rate": 9.764640108743808e-05,
+      "loss": 0.2485,
+      "step": 280
+    },
+    {
+      "epoch": 0.9461279461279462,
+      "grad_norm": 0.38060134649276733,
+      "learning_rate": 9.761062117679632e-05,
+      "loss": 0.2797,
+      "step": 281
+    },
+    {
+      "epoch": 0.9494949494949495,
+      "grad_norm": 0.416530966758728,
+      "learning_rate": 9.757457800276006e-05,
+      "loss": 0.2615,
+      "step": 282
+    },
+    {
+      "epoch": 0.9528619528619529,
+      "grad_norm": 0.3815469741821289,
+      "learning_rate": 9.75382717646304e-05,
+      "loss": 0.255,
+      "step": 283
+    },
+    {
+      "epoch": 0.9562289562289562,
+      "grad_norm": 0.41139233112335205,
+      "learning_rate": 9.750170266316303e-05,
+      "loss": 0.2615,
+      "step": 284
+    },
+    {
+      "epoch": 0.9595959595959596,
+      "grad_norm": 0.374959260225296,
+      "learning_rate": 9.746487090056713e-05,
+      "loss": 0.2638,
+      "step": 285
+    },
+    {
+      "epoch": 0.9629629629629629,
+      "grad_norm": 0.35468292236328125,
+      "learning_rate": 9.742777668050434e-05,
+      "loss": 0.2572,
+      "step": 286
+    },
+    {
+      "epoch": 0.9663299663299664,
+      "grad_norm": 0.35659849643707275,
+      "learning_rate": 9.739042020808746e-05,
+      "loss": 0.266,
+      "step": 287
+    },
+    {
+      "epoch": 0.9696969696969697,
+      "grad_norm": 0.37296387553215027,
+      "learning_rate": 9.735280168987949e-05,
+      "loss": 0.2677,
+      "step": 288
+    },
+    {
+      "epoch": 0.9730639730639731,
+      "grad_norm": 0.34908655285835266,
+      "learning_rate": 9.73149213338924e-05,
+      "loss": 0.2732,
+      "step": 289
+    },
+    {
+      "epoch": 0.9764309764309764,
+      "grad_norm": 0.3758234679698944,
+      "learning_rate": 9.727677934958599e-05,
+      "loss": 0.2738,
+      "step": 290
+    },
+    {
+      "epoch": 0.9797979797979798,
+      "grad_norm": 0.35159602761268616,
+      "learning_rate": 9.723837594786672e-05,
+      "loss": 0.2684,
+      "step": 291
+    },
+    {
+      "epoch": 0.9831649831649831,
+      "grad_norm": 0.33813127875328064,
+      "learning_rate": 9.719971134108658e-05,
+      "loss": 0.2682,
+      "step": 292
+    },
+    {
+      "epoch": 0.9865319865319865,
+      "grad_norm": 0.32259315252304077,
+      "learning_rate": 9.716078574304189e-05,
+      "loss": 0.2721,
+      "step": 293
+    },
+    {
+      "epoch": 0.98989898989899,
+      "grad_norm": 0.3467171788215637,
+      "learning_rate": 9.712159936897213e-05,
+      "loss": 0.264,
+      "step": 294
+    },
+    {
+      "epoch": 0.9932659932659933,
+      "grad_norm": 0.33499976992607117,
+      "learning_rate": 9.708215243555875e-05,
+      "loss": 0.2636,
+      "step": 295
+    },
+    {
+      "epoch": 0.9966329966329966,
+      "grad_norm": 0.2971048355102539,
+      "learning_rate": 9.704244516092392e-05,
+      "loss": 0.2644,
+      "step": 296
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3515819013118744,
+      "learning_rate": 9.700247776462943e-05,
+      "loss": 0.2587,
+      "step": 297
+    },
+    {
+      "epoch": 1.0033670033670035,
+      "grad_norm": 0.31730222702026367,
+      "learning_rate": 9.696225046767538e-05,
+      "loss": 0.2544,
+      "step": 298
+    },
+    {
+      "epoch": 1.0067340067340067,
+      "grad_norm": 0.3283955752849579,
+      "learning_rate": 9.6921763492499e-05,
+      "loss": 0.2547,
+      "step": 299
+    },
+    {
+      "epoch": 1.0101010101010102,
+      "grad_norm": 0.3477189242839813,
+      "learning_rate": 9.688101706297341e-05,
+      "loss": 0.2639,
+      "step": 300
+    },
+    {
+      "epoch": 1.0101010101010102,
+      "eval_loss": 0.12977388501167297,
+      "eval_runtime": 33.026,
+      "eval_samples_per_second": 30.279,
+      "eval_steps_per_second": 1.908,
+      "step": 300
+    },
+    {
+      "epoch": 1.0134680134680134,
+      "grad_norm": 0.37081846594810486,
+      "learning_rate": 9.684001140440639e-05,
+      "loss": 0.2451,
+      "step": 301
+    },
+    {
+      "epoch": 1.0168350168350169,
+      "grad_norm": 0.36840128898620605,
+      "learning_rate": 9.679874674353913e-05,
+      "loss": 0.2619,
+      "step": 302
+    },
+    {
+      "epoch": 1.02020202020202,
+      "grad_norm": 0.2802034020423889,
+      "learning_rate": 9.6757223308545e-05,
+      "loss": 0.2617,
+      "step": 303
+    },
+    {
+      "epoch": 1.0235690235690236,
+      "grad_norm": 0.31149744987487793,
+      "learning_rate": 9.67154413290282e-05,
+      "loss": 0.2645,
+      "step": 304
+    },
+    {
+      "epoch": 1.026936026936027,
+      "grad_norm": 0.2791181802749634,
+      "learning_rate": 9.667340103602261e-05,
+      "loss": 0.2485,
+      "step": 305
+    },
+    {
+      "epoch": 1.0303030303030303,
+      "grad_norm": 0.28780505061149597,
+      "learning_rate": 9.663110266199044e-05,
+      "loss": 0.2481,
+      "step": 306
+    },
+    {
+      "epoch": 1.0336700336700337,
+      "grad_norm": 0.3085958659648895,
+      "learning_rate": 9.658854644082098e-05,
+      "loss": 0.259,
+      "step": 307
+    },
+    {
+      "epoch": 1.037037037037037,
+      "grad_norm": 0.3242364227771759,
+      "learning_rate": 9.654573260782924e-05,
+      "loss": 0.2638,
+      "step": 308
+    },
+    {
+      "epoch": 1.0404040404040404,
+      "grad_norm": 0.3183690905570984,
+      "learning_rate": 9.650266139975474e-05,
+      "loss": 0.2682,
+      "step": 309
+    },
+    {
+      "epoch": 1.0437710437710437,
+      "grad_norm": 0.2918904721736908,
+      "learning_rate": 9.645933305476016e-05,
+      "loss": 0.2465,
+      "step": 310
+    },
+    {
+      "epoch": 1.0471380471380471,
+      "grad_norm": 0.3348250091075897,
+      "learning_rate": 9.641574781242999e-05,
+      "loss": 0.2609,
+      "step": 311
+    },
+    {
+      "epoch": 1.0505050505050506,
+      "grad_norm": 0.3023556172847748,
+      "learning_rate": 9.637190591376926e-05,
+      "loss": 0.2496,
+      "step": 312
+    },
+    {
+      "epoch": 1.0538720538720538,
+      "grad_norm": 0.31095656752586365,
+      "learning_rate": 9.632780760120215e-05,
+      "loss": 0.2665,
+      "step": 313
+    },
+    {
+      "epoch": 1.0572390572390573,
+      "grad_norm": 0.2977689802646637,
+      "learning_rate": 9.628345311857075e-05,
+      "loss": 0.2475,
+      "step": 314
+    },
+    {
+      "epoch": 1.0606060606060606,
+      "grad_norm": 0.35640037059783936,
+      "learning_rate": 9.623884271113359e-05,
+      "loss": 0.255,
+      "step": 315
+    },
+    {
+      "epoch": 1.063973063973064,
+      "grad_norm": 0.3532579839229584,
+      "learning_rate": 9.619397662556435e-05,
+      "loss": 0.2587,
+      "step": 316
+    },
+    {
+      "epoch": 1.0673400673400673,
+      "grad_norm": 0.38389119505882263,
+      "learning_rate": 9.614885510995047e-05,
+      "loss": 0.2744,
+      "step": 317
+    },
+    {
+      "epoch": 1.0707070707070707,
+      "grad_norm": 0.3254542648792267,
+      "learning_rate": 9.610347841379184e-05,
+      "loss": 0.2505,
+      "step": 318
+    },
+    {
+      "epoch": 1.074074074074074,
+      "grad_norm": 0.36234763264656067,
+      "learning_rate": 9.605784678799934e-05,
+      "loss": 0.259,
+      "step": 319
+    },
+    {
+      "epoch": 1.0774410774410774,
+      "grad_norm": 0.3419968783855438,
+      "learning_rate": 9.60119604848935e-05,
+      "loss": 0.2428,
+      "step": 320
+    },
+    {
+      "epoch": 1.0808080808080809,
+      "grad_norm": 0.3403080105781555,
+      "learning_rate": 9.596581975820303e-05,
+      "loss": 0.2593,
+      "step": 321
+    },
+    {
+      "epoch": 1.0841750841750841,
+      "grad_norm": 0.3113173246383667,
+      "learning_rate": 9.591942486306358e-05,
+      "loss": 0.2659,
+      "step": 322
+    },
+    {
+      "epoch": 1.0875420875420876,
+      "grad_norm": 0.3434092104434967,
+      "learning_rate": 9.587277605601617e-05,
+      "loss": 0.2601,
+      "step": 323
+    },
+    {
+      "epoch": 1.0909090909090908,
+      "grad_norm": 0.36877501010894775,
+      "learning_rate": 9.582587359500581e-05,
+      "loss": 0.2633,
+      "step": 324
+    },
+    {
+      "epoch": 1.0942760942760943,
+      "grad_norm": 0.3114217519760132,
+      "learning_rate": 9.577871773938011e-05,
+      "loss": 0.2459,
+      "step": 325
+    },
+    {
+      "epoch": 1.0976430976430978,
+      "grad_norm": 0.3608934283256531,
+      "learning_rate": 9.573130874988789e-05,
+      "loss": 0.2553,
+      "step": 326
+    },
+    {
+      "epoch": 1.101010101010101,
+      "grad_norm": 0.3732490539550781,
+      "learning_rate": 9.568364688867757e-05,
+      "loss": 0.2606,
+      "step": 327
+    },
+    {
+      "epoch": 1.1043771043771045,
+      "grad_norm": 0.3304762840270996,
+      "learning_rate": 9.563573241929588e-05,
+      "loss": 0.2525,
+      "step": 328
+    },
+    {
+      "epoch": 1.1077441077441077,
+      "grad_norm": 0.43853023648262024,
+      "learning_rate": 9.558756560668636e-05,
+      "loss": 0.2537,
+      "step": 329
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 0.33033648133277893,
+      "learning_rate": 9.553914671718787e-05,
+      "loss": 0.2479,
+      "step": 330
+    },
+    {
+      "epoch": 1.1144781144781144,
+      "grad_norm": 0.3931716978549957,
+      "learning_rate": 9.549047601853311e-05,
+      "loss": 0.2618,
+      "step": 331
+    },
+    {
+      "epoch": 1.1178451178451179,
+      "grad_norm": 0.35425615310668945,
+      "learning_rate": 9.544155377984722e-05,
+      "loss": 0.264,
+      "step": 332
+    },
+    {
+      "epoch": 1.121212121212121,
+      "grad_norm": 0.3464037775993347,
+      "learning_rate": 9.539238027164619e-05,
+      "loss": 0.2633,
+      "step": 333
+    },
+    {
+      "epoch": 1.1245791245791246,
+      "grad_norm": 0.29783573746681213,
+      "learning_rate": 9.534295576583538e-05,
+      "loss": 0.2713,
+      "step": 334
+    },
+    {
+      "epoch": 1.127946127946128,
+      "grad_norm": 0.3788033127784729,
+      "learning_rate": 9.52932805357081e-05,
+      "loss": 0.2402,
+      "step": 335
+    },
+    {
+      "epoch": 1.1313131313131313,
+      "grad_norm": 0.33695003390312195,
+      "learning_rate": 9.524335485594402e-05,
+      "loss": 0.2645,
+      "step": 336
+    },
+    {
+      "epoch": 1.1346801346801347,
+      "grad_norm": 0.2683010399341583,
+      "learning_rate": 9.51931790026077e-05,
+      "loss": 0.2613,
+      "step": 337
+    },
+    {
+      "epoch": 1.138047138047138,
+      "grad_norm": 0.3648826479911804,
+      "learning_rate": 9.514275325314694e-05,
+      "loss": 0.2497,
+      "step": 338
+    },
+    {
+      "epoch": 1.1414141414141414,
+      "grad_norm": 0.3843471109867096,
+      "learning_rate": 9.509207788639147e-05,
+      "loss": 0.257,
+      "step": 339
+    },
+    {
+      "epoch": 1.144781144781145,
+      "grad_norm": 0.34863507747650146,
+      "learning_rate": 9.504115318255122e-05,
+      "loss": 0.2552,
+      "step": 340
+    },
+    {
+      "epoch": 1.1481481481481481,
+      "grad_norm": 0.292370080947876,
+      "learning_rate": 9.498997942321483e-05,
+      "loss": 0.2599,
+      "step": 341
+    },
+    {
+      "epoch": 1.1515151515151516,
+      "grad_norm": 0.34756267070770264,
+      "learning_rate": 9.493855689134809e-05,
+      "loss": 0.2477,
+      "step": 342
+    },
+    {
+      "epoch": 1.1548821548821548,
+      "grad_norm": 0.3180752098560333,
+      "learning_rate": 9.488688587129242e-05,
+      "loss": 0.2615,
+      "step": 343
+    },
+    {
+      "epoch": 1.1582491582491583,
+      "grad_norm": 0.36670762300491333,
+      "learning_rate": 9.48349666487632e-05,
+      "loss": 0.2648,
+      "step": 344
+    },
+    {
+      "epoch": 1.1616161616161615,
+      "grad_norm": 0.3237256109714508,
+      "learning_rate": 9.47827995108483e-05,
+      "loss": 0.2485,
+      "step": 345
+    },
+    {
+      "epoch": 1.164983164983165,
+      "grad_norm": 0.35194486379623413,
+      "learning_rate": 9.47303847460064e-05,
+      "loss": 0.2414,
+      "step": 346
+    },
+    {
+      "epoch": 1.1683501683501682,
+      "grad_norm": 0.3115532398223877,
+      "learning_rate": 9.467772264406545e-05,
+      "loss": 0.2593,
+      "step": 347
+    },
+    {
+      "epoch": 1.1717171717171717,
+      "grad_norm": 0.42125508189201355,
+      "learning_rate": 9.462481349622108e-05,
+      "loss": 0.254,
+      "step": 348
+    },
+    {
+      "epoch": 1.1750841750841752,
+      "grad_norm": 0.47456079721450806,
+      "learning_rate": 9.457165759503493e-05,
+      "loss": 0.269,
+      "step": 349
+    },
+    {
+      "epoch": 1.1784511784511784,
+      "grad_norm": 0.2964518666267395,
+      "learning_rate": 9.451825523443307e-05,
+      "loss": 0.2601,
+      "step": 350
+    },
+    {
+      "epoch": 1.1818181818181819,
+      "grad_norm": 0.38632655143737793,
+      "learning_rate": 9.446460670970436e-05,
+      "loss": 0.254,
+      "step": 351
+    },
+    {
+      "epoch": 1.1851851851851851,
+      "grad_norm": 0.30016183853149414,
+      "learning_rate": 9.441071231749889e-05,
+      "loss": 0.2625,
+      "step": 352
+    },
+    {
+      "epoch": 1.1885521885521886,
+      "grad_norm": 0.4124687910079956,
+      "learning_rate": 9.435657235582616e-05,
+      "loss": 0.2688,
+      "step": 353
+    },
+    {
+      "epoch": 1.1919191919191918,
+      "grad_norm": 0.33176562190055847,
+      "learning_rate": 9.430218712405367e-05,
+      "loss": 0.2558,
+      "step": 354
+    },
+    {
+      "epoch": 1.1952861952861953,
+      "grad_norm": 0.4196166396141052,
+      "learning_rate": 9.424755692290507e-05,
+      "loss": 0.2527,
+      "step": 355
+    },
+    {
+      "epoch": 1.1986531986531987,
+      "grad_norm": 0.3760242760181427,
+      "learning_rate": 9.419268205445862e-05,
+      "loss": 0.254,
+      "step": 356
+    },
+    {
+      "epoch": 1.202020202020202,
+      "grad_norm": 0.3473700284957886,
+      "learning_rate": 9.413756282214537e-05,
+      "loss": 0.2611,
+      "step": 357
+    },
+    {
+      "epoch": 1.2053872053872055,
+      "grad_norm": 0.3585781753063202,
+      "learning_rate": 9.408219953074772e-05,
+      "loss": 0.2573,
+      "step": 358
+    },
+    {
+      "epoch": 1.2087542087542087,
+      "grad_norm": 0.34065455198287964,
+      "learning_rate": 9.402659248639749e-05,
+      "loss": 0.2553,
+      "step": 359
+    },
+    {
+      "epoch": 1.2121212121212122,
+      "grad_norm": 0.350239634513855,
+      "learning_rate": 9.397074199657441e-05,
+      "loss": 0.2582,
+      "step": 360
+    },
+    {
+      "epoch": 1.2154882154882154,
+      "grad_norm": 0.3311632573604584,
+      "learning_rate": 9.391464837010428e-05,
+      "loss": 0.2551,
+      "step": 361
+    },
+    {
+      "epoch": 1.2188552188552189,
+      "grad_norm": 0.31825152039527893,
+      "learning_rate": 9.385831191715735e-05,
+      "loss": 0.2434,
+      "step": 362
+    },
+    {
+      "epoch": 1.2222222222222223,
+      "grad_norm": 0.3181901276111603,
+      "learning_rate": 9.380173294924662e-05,
+      "loss": 0.2738,
+      "step": 363
+    },
+    {
+      "epoch": 1.2255892255892256,
+      "grad_norm": 0.340090274810791,
+      "learning_rate": 9.374491177922602e-05,
+      "loss": 0.2413,
+      "step": 364
+    },
+    {
+      "epoch": 1.228956228956229,
+      "grad_norm": 0.3769470453262329,
+      "learning_rate": 9.368784872128878e-05,
+      "loss": 0.2511,
+      "step": 365
+    },
+    {
+      "epoch": 1.2323232323232323,
+      "grad_norm": 0.47145962715148926,
+      "learning_rate": 9.363054409096561e-05,
+      "loss": 0.2414,
+      "step": 366
+    },
+    {
+      "epoch": 1.2356902356902357,
+      "grad_norm": 0.3955315053462982,
+      "learning_rate": 9.357299820512304e-05,
+      "loss": 0.257,
+      "step": 367
+    },
+    {
+      "epoch": 1.239057239057239,
+      "grad_norm": 0.3811996281147003,
+      "learning_rate": 9.35152113819616e-05,
+      "loss": 0.2487,
+      "step": 368
+    },
+    {
+      "epoch": 1.2424242424242424,
+      "grad_norm": 0.3641284704208374,
+      "learning_rate": 9.345718394101411e-05,
+      "loss": 0.2511,
+      "step": 369
+    },
+    {
+      "epoch": 1.2457912457912457,
+      "grad_norm": 0.34376075863838196,
+      "learning_rate": 9.339891620314387e-05,
+      "loss": 0.2589,
+      "step": 370
+    },
+    {
+      "epoch": 1.2491582491582491,
+      "grad_norm": 0.400547593832016,
+      "learning_rate": 9.334040849054289e-05,
+      "loss": 0.2497,
+      "step": 371
+    },
+    {
+      "epoch": 1.2525252525252526,
+      "grad_norm": 0.3641180694103241,
+      "learning_rate": 9.328166112673012e-05,
+      "loss": 0.2494,
+      "step": 372
+    },
+    {
+      "epoch": 1.2558922558922558,
+      "grad_norm": 0.34279683232307434,
+      "learning_rate": 9.322267443654972e-05,
+      "loss": 0.2558,
+      "step": 373
+    },
+    {
+      "epoch": 1.2592592592592593,
+      "grad_norm": 0.36419054865837097,
+      "learning_rate": 9.316344874616916e-05,
+      "loss": 0.25,
+      "step": 374
+    },
+    {
+      "epoch": 1.2626262626262625,
+      "grad_norm": 0.34432828426361084,
+      "learning_rate": 9.310398438307746e-05,
+      "loss": 0.2495,
+      "step": 375
+    },
+    {
+      "epoch": 1.265993265993266,
+      "grad_norm": 0.35908621549606323,
+      "learning_rate": 9.304428167608342e-05,
+      "loss": 0.2523,
+      "step": 376
+    },
+    {
+      "epoch": 1.2693602693602695,
+      "grad_norm": 0.3936247229576111,
+      "learning_rate": 9.29843409553137e-05,
+      "loss": 0.2581,
+      "step": 377
+    },
+    {
+      "epoch": 1.2727272727272727,
+      "grad_norm": 0.4111359715461731,
+      "learning_rate": 9.292416255221113e-05,
+      "loss": 0.2383,
+      "step": 378
+    },
+    {
+      "epoch": 1.2760942760942762,
+      "grad_norm": 0.3538724482059479,
+      "learning_rate": 9.286374679953279e-05,
+      "loss": 0.2396,
+      "step": 379
+    },
+    {
+      "epoch": 1.2794612794612794,
+      "grad_norm": 0.34664541482925415,
+      "learning_rate": 9.280309403134812e-05,
+      "loss": 0.2523,
+      "step": 380
+    },
+    {
+      "epoch": 1.2828282828282829,
+      "grad_norm": 0.3238551914691925,
+      "learning_rate": 9.274220458303727e-05,
+      "loss": 0.2406,
+      "step": 381
+    },
+    {
+      "epoch": 1.2861952861952861,
+      "grad_norm": 0.3845024108886719,
+      "learning_rate": 9.268107879128898e-05,
+      "loss": 0.2545,
+      "step": 382
+    },
+    {
+      "epoch": 1.2895622895622896,
+      "grad_norm": 0.3516189754009247,
+      "learning_rate": 9.261971699409893e-05,
+      "loss": 0.2363,
+      "step": 383
+    },
+    {
+      "epoch": 1.2929292929292928,
+      "grad_norm": 0.5296482443809509,
+      "learning_rate": 9.255811953076776e-05,
+      "loss": 0.2629,
+      "step": 384
+    },
+    {
+      "epoch": 1.2962962962962963,
+      "grad_norm": 0.36364006996154785,
+      "learning_rate": 9.249628674189927e-05,
+      "loss": 0.2377,
+      "step": 385
+    },
+    {
+      "epoch": 1.2996632996632997,
+      "grad_norm": 0.315027117729187,
+      "learning_rate": 9.243421896939848e-05,
+      "loss": 0.2411,
+      "step": 386
+    },
+    {
+      "epoch": 1.303030303030303,
+      "grad_norm": 0.3599799871444702,
+      "learning_rate": 9.237191655646972e-05,
+      "loss": 0.2512,
+      "step": 387
+    },
+    {
+      "epoch": 1.3063973063973064,
+      "grad_norm": 0.3463782072067261,
+      "learning_rate": 9.230937984761479e-05,
+      "loss": 0.2477,
+      "step": 388
+    },
+    {
+      "epoch": 1.3097643097643097,
+      "grad_norm": 0.3749505877494812,
+      "learning_rate": 9.224660918863104e-05,
+      "loss": 0.2394,
+      "step": 389
+    },
+    {
+      "epoch": 1.3131313131313131,
+      "grad_norm": 0.39092373847961426,
+      "learning_rate": 9.218360492660942e-05,
+      "loss": 0.2446,
+      "step": 390
+    },
+    {
+      "epoch": 1.3164983164983166,
+      "grad_norm": 0.40045619010925293,
+      "learning_rate": 9.212036740993266e-05,
+      "loss": 0.238,
+      "step": 391
+    },
+    {
+      "epoch": 1.3198653198653199,
+      "grad_norm": 0.38552188873291016,
+      "learning_rate": 9.205689698827318e-05,
+      "loss": 0.2312,
+      "step": 392
+    },
+    {
+      "epoch": 1.3232323232323233,
+      "grad_norm": 0.394648015499115,
+      "learning_rate": 9.199319401259131e-05,
+      "loss": 0.2439,
+      "step": 393
+    },
+    {
+      "epoch": 1.3265993265993266,
+      "grad_norm": 0.3606700897216797,
+      "learning_rate": 9.192925883513328e-05,
+      "loss": 0.2359,
+      "step": 394
+    },
+    {
+      "epoch": 1.32996632996633,
+      "grad_norm": 0.4096723198890686,
+      "learning_rate": 9.186509180942928e-05,
+      "loss": 0.2464,
+      "step": 395
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.4110390543937683,
+      "learning_rate": 9.18006932902915e-05,
+      "loss": 0.2387,
+      "step": 396
+    },
+    {
+      "epoch": 1.3367003367003367,
+      "grad_norm": 0.37714695930480957,
+      "learning_rate": 9.173606363381219e-05,
+      "loss": 0.242,
+      "step": 397
+    },
+    {
+      "epoch": 1.34006734006734,
+      "grad_norm": 0.33103522658348083,
+      "learning_rate": 9.167120319736164e-05,
+      "loss": 0.2551,
+      "step": 398
+    },
+    {
+      "epoch": 1.3434343434343434,
+      "grad_norm": 0.34669333696365356,
+      "learning_rate": 9.160611233958629e-05,
+      "loss": 0.2413,
+      "step": 399
+    },
+    {
+      "epoch": 1.3468013468013469,
+      "grad_norm": 0.34055569767951965,
+      "learning_rate": 9.154079142040668e-05,
+      "loss": 0.2372,
+      "step": 400
+    },
+    {
+      "epoch": 1.3468013468013469,
+      "eval_loss": 0.11807812005281448,
+      "eval_runtime": 32.9267,
+      "eval_samples_per_second": 30.37,
+      "eval_steps_per_second": 1.913,
+      "step": 400
+    },
+    {
+      "epoch": 1.3501683501683501,
+      "grad_norm": 0.3376658260822296,
+      "learning_rate": 9.147524080101544e-05,
+      "loss": 0.2428,
+      "step": 401
+    },
+    {
+      "epoch": 1.3535353535353536,
+      "grad_norm": 0.40585628151893616,
+      "learning_rate": 9.140946084387538e-05,
+      "loss": 0.2352,
+      "step": 402
+    },
+    {
+      "epoch": 1.3569023569023568,
+      "grad_norm": 0.34705591201782227,
+      "learning_rate": 9.134345191271742e-05,
+      "loss": 0.231,
+      "step": 403
+    },
+    {
+      "epoch": 1.3602693602693603,
+      "grad_norm": 0.3753274381160736,
+      "learning_rate": 9.127721437253857e-05,
+      "loss": 0.2366,
+      "step": 404
+    },
+    {
+      "epoch": 1.3636363636363638,
+      "grad_norm": 0.31101006269454956,
+      "learning_rate": 9.121074858959997e-05,
+      "loss": 0.2331,
+      "step": 405
+    },
+    {
+      "epoch": 1.367003367003367,
+      "grad_norm": 0.31866899132728577,
+      "learning_rate": 9.114405493142481e-05,
+      "loss": 0.2243,
+      "step": 406
+    },
+    {
+      "epoch": 1.3703703703703702,
+      "grad_norm": 0.4138029217720032,
+      "learning_rate": 9.107713376679634e-05,
+      "loss": 0.2406,
+      "step": 407
+    },
+    {
+      "epoch": 1.3737373737373737,
+      "grad_norm": 0.32557931542396545,
+      "learning_rate": 9.100998546575576e-05,
+      "loss": 0.2405,
+      "step": 408
+    },
+    {
+      "epoch": 1.3771043771043772,
+      "grad_norm": 0.29002058506011963,
+      "learning_rate": 9.094261039960027e-05,
+      "loss": 0.2364,
+      "step": 409
+    },
+    {
+      "epoch": 1.3804713804713804,
+      "grad_norm": 0.390799880027771,
+      "learning_rate": 9.0875008940881e-05,
+      "loss": 0.2358,
+      "step": 410
+    },
+    {
+      "epoch": 1.3838383838383839,
+      "grad_norm": 0.4212111830711365,
+      "learning_rate": 9.08071814634008e-05,
+      "loss": 0.2369,
+      "step": 411
+    },
+    {
+      "epoch": 1.387205387205387,
+      "grad_norm": 0.32682353258132935,
+      "learning_rate": 9.073912834221241e-05,
+      "loss": 0.2418,
+      "step": 412
+    },
+    {
+      "epoch": 1.3905723905723906,
+      "grad_norm": 0.29507341980934143,
+      "learning_rate": 9.067084995361623e-05,
+      "loss": 0.2418,
+      "step": 413
+    },
+    {
+      "epoch": 1.393939393939394,
+      "grad_norm": 0.36445966362953186,
+      "learning_rate": 9.060234667515826e-05,
+      "loss": 0.2354,
+      "step": 414
+    },
+    {
+      "epoch": 1.3973063973063973,
+      "grad_norm": 0.3630370795726776,
+      "learning_rate": 9.053361888562807e-05,
+      "loss": 0.2352,
+      "step": 415
+    },
+    {
+      "epoch": 1.4006734006734007,
+      "grad_norm": 0.3828338384628296,
+      "learning_rate": 9.046466696505662e-05,
+      "loss": 0.2287,
+      "step": 416
+    },
+    {
+      "epoch": 1.404040404040404,
+      "grad_norm": 0.43802979588508606,
+      "learning_rate": 9.039549129471423e-05,
+      "loss": 0.2327,
+      "step": 417
+    },
+    {
+      "epoch": 1.4074074074074074,
+      "grad_norm": 0.374553382396698,
+      "learning_rate": 9.032609225710846e-05,
+      "loss": 0.2369,
+      "step": 418
+    },
+    {
+      "epoch": 1.410774410774411,
+      "grad_norm": 0.4133577346801758,
+      "learning_rate": 9.025647023598196e-05,
+      "loss": 0.2309,
+      "step": 419
+    },
+    {
+      "epoch": 1.4141414141414141,
+      "grad_norm": 0.39232614636421204,
+      "learning_rate": 9.018662561631036e-05,
+      "loss": 0.227,
+      "step": 420
+    },
+    {
+      "epoch": 1.4175084175084174,
+      "grad_norm": 0.3980022370815277,
+      "learning_rate": 9.011655878430019e-05,
+      "loss": 0.2319,
+      "step": 421
+    },
+    {
+      "epoch": 1.4208754208754208,
+      "grad_norm": 0.4843936860561371,
+      "learning_rate": 9.004627012738667e-05,
+      "loss": 0.2291,
+      "step": 422
+    },
+    {
+      "epoch": 1.4242424242424243,
+      "grad_norm": 0.47197064757347107,
+      "learning_rate": 8.997576003423159e-05,
+      "loss": 0.2459,
+      "step": 423
+    },
+    {
+      "epoch": 1.4276094276094276,
+      "grad_norm": 0.48999035358428955,
+      "learning_rate": 8.990502889472124e-05,
+      "loss": 0.2397,
+      "step": 424
+    },
+    {
+      "epoch": 1.430976430976431,
+      "grad_norm": 0.3578794598579407,
+      "learning_rate": 8.983407709996414e-05,
+      "loss": 0.2268,
+      "step": 425
+    },
+    {
+      "epoch": 1.4343434343434343,
+      "grad_norm": 0.37743687629699707,
+      "learning_rate": 8.97629050422889e-05,
+      "loss": 0.2364,
+      "step": 426
+    },
+    {
+      "epoch": 1.4377104377104377,
+      "grad_norm": 0.39339321851730347,
+      "learning_rate": 8.969151311524214e-05,
+      "loss": 0.2433,
+      "step": 427
+    },
+    {
+      "epoch": 1.4410774410774412,
+      "grad_norm": 0.3382370173931122,
+      "learning_rate": 8.96199017135862e-05,
+      "loss": 0.2383,
+      "step": 428
+    },
+    {
+      "epoch": 1.4444444444444444,
+      "grad_norm": 0.4073336720466614,
+      "learning_rate": 8.954807123329704e-05,
+      "loss": 0.2383,
+      "step": 429
+    },
+    {
+      "epoch": 1.4478114478114479,
+      "grad_norm": 0.3423174023628235,
+      "learning_rate": 8.947602207156198e-05,
+      "loss": 0.2399,
+      "step": 430
+    },
+    {
+      "epoch": 1.4511784511784511,
+      "grad_norm": 0.29171693325042725,
+      "learning_rate": 8.940375462677757e-05,
+      "loss": 0.248,
+      "step": 431
+    },
+    {
+      "epoch": 1.4545454545454546,
+      "grad_norm": 0.3864852786064148,
+      "learning_rate": 8.933126929854737e-05,
+      "loss": 0.2533,
+      "step": 432
+    },
+    {
+      "epoch": 1.457912457912458,
+      "grad_norm": 0.36704379320144653,
+      "learning_rate": 8.92585664876797e-05,
+      "loss": 0.2319,
+      "step": 433
+    },
+    {
+      "epoch": 1.4612794612794613,
+      "grad_norm": 0.37201452255249023,
+      "learning_rate": 8.918564659618544e-05,
+      "loss": 0.2263,
+      "step": 434
+    },
+    {
+      "epoch": 1.4646464646464645,
+      "grad_norm": 0.39355430006980896,
+      "learning_rate": 8.911251002727588e-05,
+      "loss": 0.2463,
+      "step": 435
+    },
+    {
+      "epoch": 1.468013468013468,
+      "grad_norm": 0.34966838359832764,
+      "learning_rate": 8.903915718536036e-05,
+      "loss": 0.2319,
+      "step": 436
+    },
+    {
+      "epoch": 1.4713804713804715,
+      "grad_norm": 0.3407142162322998,
+      "learning_rate": 8.896558847604414e-05,
+      "loss": 0.2316,
+      "step": 437
+    },
+    {
+      "epoch": 1.4747474747474747,
+      "grad_norm": 0.32372531294822693,
+      "learning_rate": 8.889180430612612e-05,
+      "loss": 0.235,
+      "step": 438
+    },
+    {
+      "epoch": 1.4781144781144782,
+      "grad_norm": 0.315107136964798,
+      "learning_rate": 8.88178050835966e-05,
+      "loss": 0.2458,
+      "step": 439
+    },
+    {
+      "epoch": 1.4814814814814814,
+      "grad_norm": 0.32456645369529724,
+      "learning_rate": 8.8743591217635e-05,
+      "loss": 0.2274,
+      "step": 440
+    },
+    {
+      "epoch": 1.4848484848484849,
+      "grad_norm": 0.39389902353286743,
+      "learning_rate": 8.86691631186076e-05,
+      "loss": 0.2337,
+      "step": 441
+    },
+    {
+      "epoch": 1.4882154882154883,
+      "grad_norm": 0.334153413772583,
+      "learning_rate": 8.859452119806532e-05,
+      "loss": 0.2387,
+      "step": 442
+    },
+    {
+      "epoch": 1.4915824915824916,
+      "grad_norm": 0.3052537739276886,
+      "learning_rate": 8.851966586874138e-05,
+      "loss": 0.2394,
+      "step": 443
+    },
+    {
+      "epoch": 1.494949494949495,
+      "grad_norm": 0.3574841022491455,
+      "learning_rate": 8.844459754454903e-05,
+      "loss": 0.2351,
+      "step": 444
+    },
+    {
+      "epoch": 1.4983164983164983,
+      "grad_norm": 0.450410395860672,
+      "learning_rate": 8.836931664057935e-05,
+      "loss": 0.233,
+      "step": 445
+    },
+    {
+      "epoch": 1.5016835016835017,
+      "grad_norm": 0.46233242750167847,
+      "learning_rate": 8.82938235730988e-05,
+      "loss": 0.2497,
+      "step": 446
+    },
+    {
+      "epoch": 1.5050505050505052,
+      "grad_norm": 0.3989076316356659,
+      "learning_rate": 8.821811875954704e-05,
+      "loss": 0.227,
+      "step": 447
+    },
+    {
+      "epoch": 1.5084175084175084,
+      "grad_norm": 0.3404209315776825,
+      "learning_rate": 8.814220261853456e-05,
+      "loss": 0.2391,
+      "step": 448
+    },
+    {
+      "epoch": 1.5117845117845117,
+      "grad_norm": 0.4542754590511322,
+      "learning_rate": 8.806607556984044e-05,
+      "loss": 0.2229,
+      "step": 449
+    },
+    {
+      "epoch": 1.5151515151515151,
+      "grad_norm": 0.334740549325943,
+      "learning_rate": 8.798973803440991e-05,
+      "loss": 0.2319,
+      "step": 450
+    },
+    {
+      "epoch": 1.5185185185185186,
+      "grad_norm": 0.3982342779636383,
+      "learning_rate": 8.791319043435214e-05,
+      "loss": 0.2295,
+      "step": 451
+    },
+    {
+      "epoch": 1.5218855218855218,
+      "grad_norm": 0.298423171043396,
+      "learning_rate": 8.78364331929378e-05,
+      "loss": 0.2255,
+      "step": 452
+    },
+    {
+      "epoch": 1.5252525252525253,
+      "grad_norm": 0.4019949734210968,
+      "learning_rate": 8.775946673459681e-05,
+      "loss": 0.2311,
+      "step": 453
+    },
+    {
+      "epoch": 1.5286195286195285,
+      "grad_norm": 0.33627232909202576,
+      "learning_rate": 8.768229148491598e-05,
+      "loss": 0.2086,
+      "step": 454
+    },
+    {
+      "epoch": 1.531986531986532,
+      "grad_norm": 0.335822731256485,
+      "learning_rate": 8.760490787063659e-05,
+      "loss": 0.2272,
+      "step": 455
+    },
+    {
+      "epoch": 1.5353535353535355,
+      "grad_norm": 0.29926082491874695,
+      "learning_rate": 8.75273163196521e-05,
+      "loss": 0.2349,
+      "step": 456
+    },
+    {
+      "epoch": 1.5387205387205387,
+      "grad_norm": 0.31576985120773315,
+      "learning_rate": 8.744951726100573e-05,
+      "loss": 0.2224,
+      "step": 457
+    },
+    {
+      "epoch": 1.542087542087542,
+      "grad_norm": 0.3465363681316376,
+      "learning_rate": 8.737151112488813e-05,
+      "loss": 0.2153,
+      "step": 458
+    },
+    {
+      "epoch": 1.5454545454545454,
+      "grad_norm": 0.370771586894989,
+      "learning_rate": 8.729329834263503e-05,
+      "loss": 0.2303,
+      "step": 459
+    },
+    {
+      "epoch": 1.5488215488215489,
+      "grad_norm": 0.3440144956111908,
+      "learning_rate": 8.721487934672473e-05,
+      "loss": 0.2203,
+      "step": 460
+    },
+    {
+      "epoch": 1.5521885521885523,
+      "grad_norm": 0.4985616207122803,
+      "learning_rate": 8.713625457077585e-05,
+      "loss": 0.2334,
+      "step": 461
+    },
+    {
+      "epoch": 1.5555555555555556,
+      "grad_norm": 0.36030417680740356,
+      "learning_rate": 8.705742444954488e-05,
+      "loss": 0.2216,
+      "step": 462
+    },
+    {
+      "epoch": 1.5589225589225588,
+      "grad_norm": 0.3390983045101166,
+      "learning_rate": 8.69783894189237e-05,
+      "loss": 0.2336,
+      "step": 463
+    },
+    {
+      "epoch": 1.5622895622895623,
+      "grad_norm": 0.2929445505142212,
+      "learning_rate": 8.68991499159373e-05,
+      "loss": 0.2307,
+      "step": 464
+    },
+    {
+      "epoch": 1.5656565656565657,
+      "grad_norm": 0.3234863579273224,
+      "learning_rate": 8.681970637874132e-05,
+      "loss": 0.2249,
+      "step": 465
+    },
+    {
+      "epoch": 1.569023569023569,
+      "grad_norm": 0.3130185306072235,
+      "learning_rate": 8.674005924661952e-05,
+      "loss": 0.2341,
+      "step": 466
+    },
+    {
+      "epoch": 1.5723905723905722,
+      "grad_norm": 0.27616122364997864,
+      "learning_rate": 8.666020895998153e-05,
+      "loss": 0.229,
+      "step": 467
+    },
+    {
+      "epoch": 1.5757575757575757,
+      "grad_norm": 0.33888301253318787,
+      "learning_rate": 8.658015596036028e-05,
+      "loss": 0.2307,
+      "step": 468
+    },
+    {
+      "epoch": 1.5791245791245792,
+      "grad_norm": 0.31470242142677307,
+      "learning_rate": 8.649990069040961e-05,
+      "loss": 0.2195,
+      "step": 469
+    },
+    {
+      "epoch": 1.5824915824915826,
+      "grad_norm": 0.2968938946723938,
+      "learning_rate": 8.641944359390182e-05,
+      "loss": 0.2193,
+      "step": 470
+    },
+    {
+      "epoch": 1.5858585858585859,
+      "grad_norm": 0.3504616320133209,
+      "learning_rate": 8.63387851157252e-05,
+      "loss": 0.2286,
+      "step": 471
+    },
+    {
+      "epoch": 1.589225589225589,
+      "grad_norm": 0.3612460196018219,
+      "learning_rate": 8.62579257018816e-05,
+      "loss": 0.2183,
+      "step": 472
+    },
+    {
+      "epoch": 1.5925925925925926,
+      "grad_norm": 0.4213543236255646,
+      "learning_rate": 8.617686579948397e-05,
+      "loss": 0.22,
+      "step": 473
+    },
+    {
+      "epoch": 1.595959595959596,
+      "grad_norm": 0.3147771656513214,
+      "learning_rate": 8.609560585675379e-05,
+      "loss": 0.2121,
+      "step": 474
+    },
+    {
+      "epoch": 1.5993265993265995,
+      "grad_norm": 0.39270174503326416,
+      "learning_rate": 8.601414632301869e-05,
+      "loss": 0.2521,
+      "step": 475
+    },
+    {
+      "epoch": 1.6026936026936027,
+      "grad_norm": 0.4034063220024109,
+      "learning_rate": 8.593248764871e-05,
+      "loss": 0.2255,
+      "step": 476
+    },
+    {
+      "epoch": 1.606060606060606,
+      "grad_norm": 0.29727184772491455,
+      "learning_rate": 8.585063028536016e-05,
+      "loss": 0.2247,
+      "step": 477
+    },
+    {
+      "epoch": 1.6094276094276094,
+      "grad_norm": 0.38096803426742554,
+      "learning_rate": 8.576857468560022e-05,
+      "loss": 0.2255,
+      "step": 478
+    },
+    {
+      "epoch": 1.612794612794613,
+      "grad_norm": 0.4196261465549469,
+      "learning_rate": 8.568632130315745e-05,
+      "loss": 0.2241,
+      "step": 479
+    },
+    {
+      "epoch": 1.6161616161616161,
+      "grad_norm": 0.28777551651000977,
+      "learning_rate": 8.560387059285273e-05,
+      "loss": 0.2232,
+      "step": 480
+    },
+    {
+      "epoch": 1.6195286195286194,
+      "grad_norm": 0.37960875034332275,
+      "learning_rate": 8.552122301059806e-05,
+      "loss": 0.2206,
+      "step": 481
+    },
+    {
+      "epoch": 1.6228956228956228,
+      "grad_norm": 0.3538387417793274,
+      "learning_rate": 8.543837901339404e-05,
+      "loss": 0.2317,
+      "step": 482
+    },
+    {
+      "epoch": 1.6262626262626263,
+      "grad_norm": 0.31286677718162537,
+      "learning_rate": 8.535533905932738e-05,
+      "loss": 0.2178,
+      "step": 483
+    },
+    {
+      "epoch": 1.6296296296296298,
+      "grad_norm": 0.42340192198753357,
+      "learning_rate": 8.52721036075683e-05,
+      "loss": 0.2245,
+      "step": 484
+    },
+    {
+      "epoch": 1.632996632996633,
+      "grad_norm": 0.37731871008872986,
+      "learning_rate": 8.518867311836808e-05,
+      "loss": 0.2237,
+      "step": 485
+    },
+    {
+      "epoch": 1.6363636363636362,
+      "grad_norm": 0.3666956126689911,
+      "learning_rate": 8.510504805305639e-05,
+      "loss": 0.2243,
+      "step": 486
+    },
+    {
+      "epoch": 1.6397306397306397,
+      "grad_norm": 0.32843998074531555,
+      "learning_rate": 8.502122887403883e-05,
+      "loss": 0.2109,
+      "step": 487
+    },
+    {
+      "epoch": 1.6430976430976432,
+      "grad_norm": 0.3238717019557953,
+      "learning_rate": 8.49372160447944e-05,
+      "loss": 0.2181,
+      "step": 488
+    },
+    {
+      "epoch": 1.6464646464646466,
+      "grad_norm": 0.38621988892555237,
+      "learning_rate": 8.485301002987284e-05,
+      "loss": 0.2228,
+      "step": 489
+    },
+    {
+      "epoch": 1.6498316498316499,
+      "grad_norm": 0.3646789491176605,
+      "learning_rate": 8.476861129489217e-05,
+      "loss": 0.2248,
+      "step": 490
+    },
+    {
+      "epoch": 1.6531986531986531,
+      "grad_norm": 0.323712557554245,
+      "learning_rate": 8.468402030653597e-05,
+      "loss": 0.2119,
+      "step": 491
+    },
+    {
+      "epoch": 1.6565656565656566,
+      "grad_norm": 0.3144836723804474,
+      "learning_rate": 8.459923753255097e-05,
+      "loss": 0.2108,
+      "step": 492
+    },
+    {
+      "epoch": 1.65993265993266,
+      "grad_norm": 0.40645068883895874,
+      "learning_rate": 8.451426344174433e-05,
+      "loss": 0.2182,
+      "step": 493
+    },
+    {
+      "epoch": 1.6632996632996633,
+      "grad_norm": 0.41164469718933105,
+      "learning_rate": 8.44290985039811e-05,
+      "loss": 0.2176,
+      "step": 494
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.3617885708808899,
+      "learning_rate": 8.434374319018165e-05,
+      "loss": 0.2439,
+      "step": 495
+    },
+    {
+      "epoch": 1.67003367003367,
+      "grad_norm": 0.5637879371643066,
+      "learning_rate": 8.425819797231904e-05,
+      "loss": 0.216,
+      "step": 496
+    },
+    {
+      "epoch": 1.6734006734006734,
+      "grad_norm": 0.5605575442314148,
+      "learning_rate": 8.417246332341637e-05,
+      "loss": 0.2173,
+      "step": 497
+    },
+    {
+      "epoch": 1.676767676767677,
+      "grad_norm": 0.4026080369949341,
+      "learning_rate": 8.408653971754421e-05,
+      "loss": 0.2186,
+      "step": 498
+    },
+    {
+      "epoch": 1.6801346801346801,
+      "grad_norm": 0.40690577030181885,
+      "learning_rate": 8.400042762981799e-05,
+      "loss": 0.2142,
+      "step": 499
+    },
+    {
+      "epoch": 1.6835016835016834,
+      "grad_norm": 0.37610065937042236,
+      "learning_rate": 8.391412753639534e-05,
+      "loss": 0.2167,
+      "step": 500
+    },
+    {
+      "epoch": 1.6835016835016834,
+      "eval_loss": 0.10807202011346817,
+      "eval_runtime": 33.2575,
+      "eval_samples_per_second": 30.068,
+      "eval_steps_per_second": 1.894,
+      "step": 500
+    },
+    {
+      "epoch": 1.6868686868686869,
+      "grad_norm": 0.3314391076564789,
+      "learning_rate": 8.382763991447344e-05,
+      "loss": 0.2269,
+      "step": 501
+    },
+    {
+      "epoch": 1.6902356902356903,
+      "grad_norm": 0.36875322461128235,
+      "learning_rate": 8.374096524228647e-05,
+      "loss": 0.2017,
+      "step": 502
+    },
+    {
+      "epoch": 1.6936026936026936,
+      "grad_norm": 0.4028531610965729,
+      "learning_rate": 8.365410399910288e-05,
+      "loss": 0.2184,
+      "step": 503
+    },
+    {
+      "epoch": 1.696969696969697,
+      "grad_norm": 0.3435775637626648,
+      "learning_rate": 8.356705666522275e-05,
+      "loss": 0.227,
+      "step": 504
+    },
+    {
+      "epoch": 1.7003367003367003,
+      "grad_norm": 0.3759492039680481,
+      "learning_rate": 8.347982372197514e-05,
+      "loss": 0.2087,
+      "step": 505
+    },
+    {
+      "epoch": 1.7037037037037037,
+      "grad_norm": 0.3366954028606415,
+      "learning_rate": 8.339240565171551e-05,
+      "loss": 0.2337,
+      "step": 506
+    },
+    {
+      "epoch": 1.7070707070707072,
+      "grad_norm": 0.34926289319992065,
+      "learning_rate": 8.33048029378229e-05,
+      "loss": 0.2097,
+      "step": 507
+    },
+    {
+      "epoch": 1.7104377104377104,
+      "grad_norm": 0.36330562829971313,
+      "learning_rate": 8.321701606469736e-05,
+      "loss": 0.2278,
+      "step": 508
+    },
+    {
+      "epoch": 1.7138047138047137,
+      "grad_norm": 0.3466736376285553,
+      "learning_rate": 8.312904551775731e-05,
+      "loss": 0.2247,
+      "step": 509
+    },
+    {
+      "epoch": 1.7171717171717171,
+      "grad_norm": 0.46628567576408386,
+      "learning_rate": 8.30408917834367e-05,
+      "loss": 0.2239,
+      "step": 510
+    },
+    {
+      "epoch": 1.7205387205387206,
+      "grad_norm": 0.34879177808761597,
+      "learning_rate": 8.295255534918248e-05,
+      "loss": 0.2125,
+      "step": 511
+    },
+    {
+      "epoch": 1.723905723905724,
+      "grad_norm": 0.35456812381744385,
+      "learning_rate": 8.286403670345184e-05,
+      "loss": 0.2171,
+      "step": 512
+    },
+    {
+      "epoch": 1.7272727272727273,
+      "grad_norm": 0.31053483486175537,
+      "learning_rate": 8.277533633570948e-05,
+      "loss": 0.2023,
+      "step": 513
+    },
+    {
+      "epoch": 1.7306397306397305,
+      "grad_norm": 0.4170824885368347,
+      "learning_rate": 8.268645473642493e-05,
+      "loss": 0.2221,
+      "step": 514
+    },
+    {
+      "epoch": 1.734006734006734,
+      "grad_norm": 0.46031665802001953,
+      "learning_rate": 8.259739239706991e-05,
+      "loss": 0.2156,
+      "step": 515
+    },
+    {
+      "epoch": 1.7373737373737375,
+      "grad_norm": 0.5810372829437256,
+      "learning_rate": 8.250814981011545e-05,
+      "loss": 0.2172,
+      "step": 516
+    },
+    {
+      "epoch": 1.7407407407407407,
+      "grad_norm": 0.42679181694984436,
+      "learning_rate": 8.241872746902935e-05,
+      "loss": 0.2097,
+      "step": 517
+    },
+    {
+      "epoch": 1.7441077441077442,
+      "grad_norm": 0.3723827600479126,
+      "learning_rate": 8.232912586827326e-05,
+      "loss": 0.207,
+      "step": 518
+    },
+    {
+      "epoch": 1.7474747474747474,
+      "grad_norm": 0.39440205693244934,
+      "learning_rate": 8.223934550330015e-05,
+      "loss": 0.2155,
+      "step": 519
+    },
+    {
+      "epoch": 1.7508417508417509,
+      "grad_norm": 0.36913472414016724,
+      "learning_rate": 8.21493868705514e-05,
+      "loss": 0.2217,
+      "step": 520
+    },
+    {
+      "epoch": 1.7542087542087543,
+      "grad_norm": 0.45470842719078064,
+      "learning_rate": 8.205925046745419e-05,
+      "loss": 0.2105,
+      "step": 521
+    },
+    {
+      "epoch": 1.7575757575757576,
+      "grad_norm": 0.4667101204395294,
+      "learning_rate": 8.196893679241858e-05,
+      "loss": 0.2184,
+      "step": 522
+    },
+    {
+      "epoch": 1.7609427609427608,
+      "grad_norm": 0.3914218544960022,
+      "learning_rate": 8.187844634483496e-05,
+      "loss": 0.2203,
+      "step": 523
+    },
+    {
+      "epoch": 1.7643097643097643,
+      "grad_norm": 0.41950666904449463,
+      "learning_rate": 8.178777962507112e-05,
+      "loss": 0.2235,
+      "step": 524
+    },
+    {
+      "epoch": 1.7676767676767677,
+      "grad_norm": 0.46251434087753296,
+      "learning_rate": 8.169693713446959e-05,
+      "loss": 0.2118,
+      "step": 525
+    },
+    {
+      "epoch": 1.7710437710437712,
+      "grad_norm": 0.3726736009120941,
+      "learning_rate": 8.16059193753448e-05,
+      "loss": 0.2084,
+      "step": 526
+    },
+    {
+      "epoch": 1.7744107744107744,
+      "grad_norm": 0.3658943474292755,
+      "learning_rate": 8.151472685098036e-05,
+      "loss": 0.2119,
+      "step": 527
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "grad_norm": 0.3458796739578247,
+      "learning_rate": 8.142336006562618e-05,
+      "loss": 0.2217,
+      "step": 528
+    },
+    {
+      "epoch": 1.7811447811447811,
+      "grad_norm": 0.4397581219673157,
+      "learning_rate": 8.133181952449582e-05,
+      "loss": 0.2067,
+      "step": 529
+    },
+    {
+      "epoch": 1.7845117845117846,
+      "grad_norm": 0.3916018009185791,
+      "learning_rate": 8.124010573376357e-05,
+      "loss": 0.2233,
+      "step": 530
+    },
+    {
+      "epoch": 1.7878787878787878,
+      "grad_norm": 0.33343255519866943,
+      "learning_rate": 8.114821920056177e-05,
+      "loss": 0.2164,
+      "step": 531
+    },
+    {
+      "epoch": 1.791245791245791,
+      "grad_norm": 0.3748774826526642,
+      "learning_rate": 8.105616043297787e-05,
+      "loss": 0.2115,
+      "step": 532
+    },
+    {
+      "epoch": 1.7946127946127945,
+      "grad_norm": 0.3121494948863983,
+      "learning_rate": 8.096392994005177e-05,
+      "loss": 0.218,
+      "step": 533
+    },
+    {
+      "epoch": 1.797979797979798,
+      "grad_norm": 0.5338030457496643,
+      "learning_rate": 8.087152823177281e-05,
+      "loss": 0.2121,
+      "step": 534
+    },
+    {
+      "epoch": 1.8013468013468015,
+      "grad_norm": 0.5407494902610779,
+      "learning_rate": 8.077895581907718e-05,
+      "loss": 0.2184,
+      "step": 535
+    },
+    {
+      "epoch": 1.8047138047138047,
+      "grad_norm": 0.334550142288208,
+      "learning_rate": 8.068621321384496e-05,
+      "loss": 0.2119,
+      "step": 536
+    },
+    {
+      "epoch": 1.808080808080808,
+      "grad_norm": 0.4660499095916748,
+      "learning_rate": 8.059330092889723e-05,
+      "loss": 0.2172,
+      "step": 537
+    },
+    {
+      "epoch": 1.8114478114478114,
+      "grad_norm": 0.30550065636634827,
+      "learning_rate": 8.050021947799342e-05,
+      "loss": 0.1999,
+      "step": 538
+    },
+    {
+      "epoch": 1.8148148148148149,
+      "grad_norm": 0.33308038115501404,
+      "learning_rate": 8.040696937582832e-05,
+      "loss": 0.2176,
+      "step": 539
+    },
+    {
+      "epoch": 1.8181818181818183,
+      "grad_norm": 0.34143635630607605,
+      "learning_rate": 8.031355113802927e-05,
+      "loss": 0.2068,
+      "step": 540
+    },
+    {
+      "epoch": 1.8215488215488216,
+      "grad_norm": 0.398154079914093,
+      "learning_rate": 8.021996528115335e-05,
+      "loss": 0.2188,
+      "step": 541
+    },
+    {
+      "epoch": 1.8249158249158248,
+      "grad_norm": 0.3586583137512207,
+      "learning_rate": 8.012621232268443e-05,
+      "loss": 0.1982,
+      "step": 542
+    },
+    {
+      "epoch": 1.8282828282828283,
+      "grad_norm": 0.3652949035167694,
+      "learning_rate": 8.003229278103043e-05,
+      "loss": 0.2047,
+      "step": 543
+    },
+    {
+      "epoch": 1.8316498316498318,
+      "grad_norm": 0.42600250244140625,
+      "learning_rate": 7.993820717552037e-05,
+      "loss": 0.1962,
+      "step": 544
+    },
+    {
+      "epoch": 1.835016835016835,
+      "grad_norm": 0.43108507990837097,
+      "learning_rate": 7.984395602640153e-05,
+      "loss": 0.2144,
+      "step": 545
+    },
+    {
+      "epoch": 1.8383838383838382,
+      "grad_norm": 0.3814637362957001,
+      "learning_rate": 7.974953985483655e-05,
+      "loss": 0.2074,
+      "step": 546
+    },
+    {
+      "epoch": 1.8417508417508417,
+      "grad_norm": 0.3342503607273102,
+      "learning_rate": 7.96549591829006e-05,
+      "loss": 0.2034,
+      "step": 547
+    },
+    {
+      "epoch": 1.8451178451178452,
+      "grad_norm": 0.3812967836856842,
+      "learning_rate": 7.956021453357838e-05,
+      "loss": 0.2067,
+      "step": 548
+    },
+    {
+      "epoch": 1.8484848484848486,
+      "grad_norm": 0.3528459370136261,
+      "learning_rate": 7.946530643076138e-05,
+      "loss": 0.2072,
+      "step": 549
+    },
+    {
+      "epoch": 1.8518518518518519,
+      "grad_norm": 0.29267165064811707,
+      "learning_rate": 7.937023539924486e-05,
+      "loss": 0.2093,
+      "step": 550
+    },
+    {
+      "epoch": 1.855218855218855,
+      "grad_norm": 0.3306651711463928,
+      "learning_rate": 7.927500196472506e-05,
+      "loss": 0.1972,
+      "step": 551
+    },
+    {
+      "epoch": 1.8585858585858586,
+      "grad_norm": 0.38230523467063904,
+      "learning_rate": 7.917960665379616e-05,
+      "loss": 0.2082,
+      "step": 552
+    },
+    {
+      "epoch": 1.861952861952862,
+      "grad_norm": 0.3726579546928406,
+      "learning_rate": 7.908404999394746e-05,
+      "loss": 0.2176,
+      "step": 553
+    },
+    {
+      "epoch": 1.8653198653198653,
+      "grad_norm": 0.38537707924842834,
+      "learning_rate": 7.898833251356045e-05,
+      "loss": 0.2144,
+      "step": 554
+    },
+    {
+      "epoch": 1.8686868686868687,
+      "grad_norm": 0.39219149947166443,
+      "learning_rate": 7.889245474190588e-05,
+      "loss": 0.2153,
+      "step": 555
+    },
+    {
+      "epoch": 1.872053872053872,
+      "grad_norm": 0.30300959944725037,
+      "learning_rate": 7.879641720914079e-05,
+      "loss": 0.2022,
+      "step": 556
+    },
+    {
+      "epoch": 1.8754208754208754,
+      "grad_norm": 0.3848994970321655,
+      "learning_rate": 7.870022044630569e-05,
+      "loss": 0.213,
+      "step": 557
+    },
+    {
+      "epoch": 1.878787878787879,
+      "grad_norm": 0.24942323565483093,
+      "learning_rate": 7.860386498532151e-05,
+      "loss": 0.2037,
+      "step": 558
+    },
+    {
+      "epoch": 1.8821548821548821,
+      "grad_norm": 0.3252245783805847,
+      "learning_rate": 7.85073513589867e-05,
+      "loss": 0.2084,
+      "step": 559
+    },
+    {
+      "epoch": 1.8855218855218854,
+      "grad_norm": 0.37132528424263,
+      "learning_rate": 7.84106801009743e-05,
+      "loss": 0.209,
+      "step": 560
+    },
+    {
+      "epoch": 1.8888888888888888,
+      "grad_norm": 0.3328373432159424,
+      "learning_rate": 7.831385174582901e-05,
+      "loss": 0.1987,
+      "step": 561
+    },
+    {
+      "epoch": 1.8922558922558923,
+      "grad_norm": 0.41517242789268494,
+      "learning_rate": 7.821686682896412e-05,
+      "loss": 0.204,
+      "step": 562
+    },
+    {
+      "epoch": 1.8956228956228958,
+      "grad_norm": 0.3530445694923401,
+      "learning_rate": 7.81197258866587e-05,
+      "loss": 0.189,
+      "step": 563
+    },
+    {
+      "epoch": 1.898989898989899,
+      "grad_norm": 0.380113422870636,
+      "learning_rate": 7.802242945605452e-05,
+      "loss": 0.2092,
+      "step": 564
+    },
+    {
+      "epoch": 1.9023569023569022,
+      "grad_norm": 0.35453280806541443,
+      "learning_rate": 7.792497807515317e-05,
+      "loss": 0.2088,
+      "step": 565
+    },
+    {
+      "epoch": 1.9057239057239057,
+      "grad_norm": 0.3634714186191559,
+      "learning_rate": 7.782737228281299e-05,
+      "loss": 0.2189,
+      "step": 566
+    },
+    {
+      "epoch": 1.9090909090909092,
+      "grad_norm": 0.3585050702095032,
+      "learning_rate": 7.772961261874615e-05,
+      "loss": 0.2085,
+      "step": 567
+    },
+    {
+      "epoch": 1.9124579124579124,
+      "grad_norm": 0.3739174008369446,
+      "learning_rate": 7.763169962351571e-05,
+      "loss": 0.1955,
+      "step": 568
+    },
+    {
+      "epoch": 1.9158249158249159,
+      "grad_norm": 0.4537617862224579,
+      "learning_rate": 7.753363383853249e-05,
+      "loss": 0.2135,
+      "step": 569
+    },
+    {
+      "epoch": 1.9191919191919191,
+      "grad_norm": 0.49453261494636536,
+      "learning_rate": 7.743541580605221e-05,
+      "loss": 0.2129,
+      "step": 570
+    },
+    {
+      "epoch": 1.9225589225589226,
+      "grad_norm": 0.3620549142360687,
+      "learning_rate": 7.733704606917247e-05,
+      "loss": 0.1981,
+      "step": 571
+    },
+    {
+      "epoch": 1.925925925925926,
+      "grad_norm": 0.4411380887031555,
+      "learning_rate": 7.723852517182966e-05,
+      "loss": 0.2001,
+      "step": 572
+    },
+    {
+      "epoch": 1.9292929292929293,
+      "grad_norm": 0.5122190713882446,
+      "learning_rate": 7.713985365879606e-05,
+      "loss": 0.2153,
+      "step": 573
+    },
+    {
+      "epoch": 1.9326599326599325,
+      "grad_norm": 0.44053155183792114,
+      "learning_rate": 7.704103207567676e-05,
+      "loss": 0.2161,
+      "step": 574
+    },
+    {
+      "epoch": 1.936026936026936,
+      "grad_norm": 0.3606555461883545,
+      "learning_rate": 7.694206096890666e-05,
+      "loss": 0.207,
+      "step": 575
+    },
+    {
+      "epoch": 1.9393939393939394,
+      "grad_norm": 0.47892236709594727,
+      "learning_rate": 7.684294088574748e-05,
+      "loss": 0.2023,
+      "step": 576
+    },
+    {
+      "epoch": 1.942760942760943,
+      "grad_norm": 0.4973437488079071,
+      "learning_rate": 7.674367237428466e-05,
+      "loss": 0.2071,
+      "step": 577
+    },
+    {
+      "epoch": 1.9461279461279462,
+      "grad_norm": 0.344251811504364,
+      "learning_rate": 7.664425598342442e-05,
+      "loss": 0.207,
+      "step": 578
+    },
+    {
+      "epoch": 1.9494949494949494,
+      "grad_norm": 0.4177689850330353,
+      "learning_rate": 7.654469226289067e-05,
+      "loss": 0.2012,
+      "step": 579
+    },
+    {
+      "epoch": 1.9528619528619529,
+      "grad_norm": 0.38338756561279297,
+      "learning_rate": 7.644498176322196e-05,
+      "loss": 0.2109,
+      "step": 580
+    },
+    {
+      "epoch": 1.9562289562289563,
+      "grad_norm": 0.4217558801174164,
+      "learning_rate": 7.63451250357685e-05,
+      "loss": 0.2104,
+      "step": 581
+    },
+    {
+      "epoch": 1.9595959595959596,
+      "grad_norm": 0.3722991943359375,
+      "learning_rate": 7.6245122632689e-05,
+      "loss": 0.1995,
+      "step": 582
+    },
+    {
+      "epoch": 1.9629629629629628,
+      "grad_norm": 0.3889348804950714,
+      "learning_rate": 7.614497510694774e-05,
+      "loss": 0.201,
+      "step": 583
+    },
+    {
+      "epoch": 1.9663299663299663,
+      "grad_norm": 0.3091539740562439,
+      "learning_rate": 7.604468301231143e-05,
+      "loss": 0.1914,
+      "step": 584
+    },
+    {
+      "epoch": 1.9696969696969697,
+      "grad_norm": 0.38198480010032654,
+      "learning_rate": 7.59442469033462e-05,
+      "loss": 0.1928,
+      "step": 585
+    },
+    {
+      "epoch": 1.9730639730639732,
+      "grad_norm": 0.4082641303539276,
+      "learning_rate": 7.58436673354145e-05,
+      "loss": 0.1995,
+      "step": 586
+    },
+    {
+      "epoch": 1.9764309764309764,
+      "grad_norm": 0.4138404130935669,
+      "learning_rate": 7.574294486467204e-05,
+      "loss": 0.2016,
+      "step": 587
+    },
+    {
+      "epoch": 1.9797979797979797,
+      "grad_norm": 0.32085567712783813,
+      "learning_rate": 7.564208004806467e-05,
+      "loss": 0.2048,
+      "step": 588
+    },
+    {
+      "epoch": 1.9831649831649831,
+      "grad_norm": 0.30701011419296265,
+      "learning_rate": 7.55410734433254e-05,
+      "loss": 0.1922,
+      "step": 589
+    },
+    {
+      "epoch": 1.9865319865319866,
+      "grad_norm": 0.33269554376602173,
+      "learning_rate": 7.543992560897124e-05,
+      "loss": 0.2134,
+      "step": 590
+    },
+    {
+      "epoch": 1.98989898989899,
+      "grad_norm": 0.3297766447067261,
+      "learning_rate": 7.533863710430012e-05,
+      "loss": 0.1952,
+      "step": 591
+    },
+    {
+      "epoch": 1.9932659932659933,
+      "grad_norm": 0.3603309690952301,
+      "learning_rate": 7.523720848938781e-05,
+      "loss": 0.2022,
+      "step": 592
+    },
+    {
+      "epoch": 1.9966329966329965,
+      "grad_norm": 0.40898165106773376,
+      "learning_rate": 7.513564032508484e-05,
+      "loss": 0.2075,
+      "step": 593
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.39114904403686523,
+      "learning_rate": 7.503393317301337e-05,
+      "loss": 0.1897,
+      "step": 594
+    },
+    {
+      "epoch": 2.0033670033670035,
+      "grad_norm": 0.35284826159477234,
+      "learning_rate": 7.493208759556406e-05,
+      "loss": 0.1964,
+      "step": 595
+    },
+    {
+      "epoch": 2.006734006734007,
+      "grad_norm": 0.42649611830711365,
+      "learning_rate": 7.483010415589306e-05,
+      "loss": 0.1973,
+      "step": 596
+    },
+    {
+      "epoch": 2.01010101010101,
+      "grad_norm": 0.32770025730133057,
+      "learning_rate": 7.472798341791877e-05,
+      "loss": 0.1949,
+      "step": 597
+    },
+    {
+      "epoch": 2.0134680134680134,
+      "grad_norm": 0.39396432042121887,
+      "learning_rate": 7.462572594631882e-05,
+      "loss": 0.2072,
+      "step": 598
+    },
+    {
+      "epoch": 2.016835016835017,
+      "grad_norm": 0.38193821907043457,
+      "learning_rate": 7.452333230652688e-05,
+      "loss": 0.1888,
+      "step": 599
+    },
+    {
+      "epoch": 2.0202020202020203,
+      "grad_norm": 0.41221076250076294,
+      "learning_rate": 7.442080306472962e-05,
+      "loss": 0.1993,
+      "step": 600
+    },
+    {
+      "epoch": 2.0202020202020203,
+      "eval_loss": 0.09800498932600021,
+      "eval_runtime": 32.7986,
+      "eval_samples_per_second": 30.489,
+      "eval_steps_per_second": 1.921,
+      "step": 600
+    },
+    {
+      "epoch": 2.0235690235690234,
+      "grad_norm": 0.34845221042633057,
+      "learning_rate": 7.431813878786343e-05,
+      "loss": 0.1866,
+      "step": 601
+    },
+    {
+      "epoch": 2.026936026936027,
+      "grad_norm": 0.398992121219635,
+      "learning_rate": 7.421534004361148e-05,
+      "loss": 0.2037,
+      "step": 602
+    },
+    {
+      "epoch": 2.0303030303030303,
+      "grad_norm": 0.4208246171474457,
+      "learning_rate": 7.41124074004004e-05,
+      "loss": 0.2019,
+      "step": 603
+    },
+    {
+      "epoch": 2.0336700336700337,
+      "grad_norm": 0.3810597062110901,
+      "learning_rate": 7.400934142739725e-05,
+      "loss": 0.2012,
+      "step": 604
+    },
+    {
+      "epoch": 2.037037037037037,
+      "grad_norm": 0.4348938465118408,
+      "learning_rate": 7.390614269450634e-05,
+      "loss": 0.1936,
+      "step": 605
+    },
+    {
+      "epoch": 2.04040404040404,
+      "grad_norm": 0.43880605697631836,
+      "learning_rate": 7.380281177236608e-05,
+      "loss": 0.1999,
+      "step": 606
+    },
+    {
+      "epoch": 2.0437710437710437,
+      "grad_norm": 0.32157278060913086,
+      "learning_rate": 7.369934923234577e-05,
+      "loss": 0.1835,
+      "step": 607
+    },
+    {
+      "epoch": 2.047138047138047,
+      "grad_norm": 0.40847641229629517,
+      "learning_rate": 7.359575564654259e-05,
+      "loss": 0.1873,
+      "step": 608
+    },
+    {
+      "epoch": 2.0505050505050506,
+      "grad_norm": 0.3321189880371094,
+      "learning_rate": 7.349203158777826e-05,
+      "loss": 0.1874,
+      "step": 609
+    },
+    {
+      "epoch": 2.053872053872054,
+      "grad_norm": 0.4147774875164032,
+      "learning_rate": 7.338817762959595e-05,
+      "loss": 0.1899,
+      "step": 610
+    },
+    {
+      "epoch": 2.057239057239057,
+      "grad_norm": 0.391746461391449,
+      "learning_rate": 7.32841943462572e-05,
+      "loss": 0.1871,
+      "step": 611
+    },
+    {
+      "epoch": 2.0606060606060606,
+      "grad_norm": 0.41342151165008545,
+      "learning_rate": 7.318008231273851e-05,
+      "loss": 0.1887,
+      "step": 612
+    },
+    {
+      "epoch": 2.063973063973064,
+      "grad_norm": 0.4507507383823395,
+      "learning_rate": 7.307584210472844e-05,
+      "loss": 0.1862,
+      "step": 613
+    },
+    {
+      "epoch": 2.0673400673400675,
+      "grad_norm": 0.37700155377388,
+      "learning_rate": 7.297147429862424e-05,
+      "loss": 0.1895,
+      "step": 614
+    },
+    {
+      "epoch": 2.0707070707070705,
+      "grad_norm": 0.46107348799705505,
+      "learning_rate": 7.286697947152867e-05,
+      "loss": 0.1911,
+      "step": 615
+    },
+    {
+      "epoch": 2.074074074074074,
+      "grad_norm": 0.3295687139034271,
+      "learning_rate": 7.276235820124694e-05,
+      "loss": 0.2021,
+      "step": 616
+    },
+    {
+      "epoch": 2.0774410774410774,
+      "grad_norm": 0.34536072611808777,
+      "learning_rate": 7.265761106628337e-05,
+      "loss": 0.1944,
+      "step": 617
+    },
+    {
+      "epoch": 2.080808080808081,
+      "grad_norm": 0.3414456248283386,
+      "learning_rate": 7.255273864583825e-05,
+      "loss": 0.1827,
+      "step": 618
+    },
+    {
+      "epoch": 2.0841750841750843,
+      "grad_norm": 0.3441462516784668,
+      "learning_rate": 7.244774151980466e-05,
+      "loss": 0.1802,
+      "step": 619
+    },
+    {
+      "epoch": 2.0875420875420874,
+      "grad_norm": 0.3795068562030792,
+      "learning_rate": 7.234262026876524e-05,
+      "loss": 0.1868,
+      "step": 620
+    },
+    {
+      "epoch": 2.090909090909091,
+      "grad_norm": 0.3172759711742401,
+      "learning_rate": 7.223737547398898e-05,
+      "loss": 0.1891,
+      "step": 621
+    },
+    {
+      "epoch": 2.0942760942760943,
+      "grad_norm": 0.34847378730773926,
+      "learning_rate": 7.213200771742798e-05,
+      "loss": 0.1887,
+      "step": 622
+    },
+    {
+      "epoch": 2.0976430976430978,
+      "grad_norm": 0.37406644225120544,
+      "learning_rate": 7.202651758171431e-05,
+      "loss": 0.1818,
+      "step": 623
+    },
+    {
+      "epoch": 2.101010101010101,
+      "grad_norm": 0.40993303060531616,
+      "learning_rate": 7.192090565015668e-05,
+      "loss": 0.175,
+      "step": 624
+    },
+    {
+      "epoch": 2.1043771043771042,
+      "grad_norm": 0.3788850009441376,
+      "learning_rate": 7.181517250673728e-05,
+      "loss": 0.1869,
+      "step": 625
+    },
+    {
+      "epoch": 2.1077441077441077,
+      "grad_norm": 0.4065912961959839,
+      "learning_rate": 7.170931873610859e-05,
+      "loss": 0.1849,
+      "step": 626
+    },
+    {
+      "epoch": 2.111111111111111,
+      "grad_norm": 0.32020193338394165,
+      "learning_rate": 7.160334492359007e-05,
+      "loss": 0.1962,
+      "step": 627
+    },
+    {
+      "epoch": 2.1144781144781146,
+      "grad_norm": 0.3998248875141144,
+      "learning_rate": 7.149725165516494e-05,
+      "loss": 0.1973,
+      "step": 628
+    },
+    {
+      "epoch": 2.1178451178451176,
+      "grad_norm": 0.35018956661224365,
+      "learning_rate": 7.139103951747695e-05,
+      "loss": 0.1826,
+      "step": 629
+    },
+    {
+      "epoch": 2.121212121212121,
+      "grad_norm": 0.4554973542690277,
+      "learning_rate": 7.128470909782716e-05,
+      "loss": 0.1991,
+      "step": 630
+    },
+    {
+      "epoch": 2.1245791245791246,
+      "grad_norm": 0.3799019753932953,
+      "learning_rate": 7.117826098417068e-05,
+      "loss": 0.1966,
+      "step": 631
+    },
+    {
+      "epoch": 2.127946127946128,
+      "grad_norm": 0.4118320345878601,
+      "learning_rate": 7.107169576511338e-05,
+      "loss": 0.2056,
+      "step": 632
+    },
+    {
+      "epoch": 2.1313131313131315,
+      "grad_norm": 0.34512877464294434,
+      "learning_rate": 7.096501402990865e-05,
+      "loss": 0.1769,
+      "step": 633
+    },
+    {
+      "epoch": 2.1346801346801345,
+      "grad_norm": 0.4270040690898895,
+      "learning_rate": 7.085821636845425e-05,
+      "loss": 0.1944,
+      "step": 634
+    },
+    {
+      "epoch": 2.138047138047138,
+      "grad_norm": 0.3624259829521179,
+      "learning_rate": 7.075130337128884e-05,
+      "loss": 0.1817,
+      "step": 635
+    },
+    {
+      "epoch": 2.1414141414141414,
+      "grad_norm": 0.447008341550827,
+      "learning_rate": 7.064427562958889e-05,
+      "loss": 0.1832,
+      "step": 636
+    },
+    {
+      "epoch": 2.144781144781145,
+      "grad_norm": 0.3551187813282013,
+      "learning_rate": 7.053713373516538e-05,
+      "loss": 0.189,
+      "step": 637
+    },
+    {
+      "epoch": 2.148148148148148,
+      "grad_norm": 0.426107794046402,
+      "learning_rate": 7.04298782804604e-05,
+      "loss": 0.1996,
+      "step": 638
+    },
+    {
+      "epoch": 2.1515151515151514,
+      "grad_norm": 0.4078080356121063,
+      "learning_rate": 7.032250985854409e-05,
+      "loss": 0.1813,
+      "step": 639
+    },
+    {
+      "epoch": 2.154882154882155,
+      "grad_norm": 0.35285472869873047,
+      "learning_rate": 7.021502906311113e-05,
+      "loss": 0.1851,
+      "step": 640
+    },
+    {
+      "epoch": 2.1582491582491583,
+      "grad_norm": 0.44015195965766907,
+      "learning_rate": 7.01074364884777e-05,
+      "loss": 0.2016,
+      "step": 641
+    },
+    {
+      "epoch": 2.1616161616161618,
+      "grad_norm": 0.31542298197746277,
+      "learning_rate": 6.999973272957793e-05,
+      "loss": 0.1854,
+      "step": 642
+    },
+    {
+      "epoch": 2.164983164983165,
+      "grad_norm": 0.5345585942268372,
+      "learning_rate": 6.989191838196082e-05,
+      "loss": 0.1873,
+      "step": 643
+    },
+    {
+      "epoch": 2.1683501683501682,
+      "grad_norm": 0.3206551969051361,
+      "learning_rate": 6.978399404178688e-05,
+      "loss": 0.177,
+      "step": 644
+    },
+    {
+      "epoch": 2.1717171717171717,
+      "grad_norm": 0.46700677275657654,
+      "learning_rate": 6.967596030582478e-05,
+      "loss": 0.1805,
+      "step": 645
+    },
+    {
+      "epoch": 2.175084175084175,
+      "grad_norm": 0.3999563455581665,
+      "learning_rate": 6.956781777144813e-05,
+      "loss": 0.1904,
+      "step": 646
+    },
+    {
+      "epoch": 2.1784511784511786,
+      "grad_norm": 0.383768767118454,
+      "learning_rate": 6.945956703663211e-05,
+      "loss": 0.1852,
+      "step": 647
+    },
+    {
+      "epoch": 2.1818181818181817,
+      "grad_norm": 0.42654767632484436,
+      "learning_rate": 6.935120869995023e-05,
+      "loss": 0.2006,
+      "step": 648
+    },
+    {
+      "epoch": 2.185185185185185,
+      "grad_norm": 0.3127792775630951,
+      "learning_rate": 6.924274336057099e-05,
+      "loss": 0.1908,
+      "step": 649
+    },
+    {
+      "epoch": 2.1885521885521886,
+      "grad_norm": 0.4523990750312805,
+      "learning_rate": 6.91341716182545e-05,
+      "loss": 0.1943,
+      "step": 650
+    },
+    {
+      "epoch": 2.191919191919192,
+      "grad_norm": 0.3176407217979431,
+      "learning_rate": 6.902549407334929e-05,
+      "loss": 0.1707,
+      "step": 651
+    },
+    {
+      "epoch": 2.1952861952861955,
+      "grad_norm": 0.4285120666027069,
+      "learning_rate": 6.891671132678892e-05,
+      "loss": 0.1879,
+      "step": 652
+    },
+    {
+      "epoch": 2.1986531986531985,
+      "grad_norm": 0.3872413635253906,
+      "learning_rate": 6.880782398008862e-05,
+      "loss": 0.1979,
+      "step": 653
+    },
+    {
+      "epoch": 2.202020202020202,
+      "grad_norm": 0.49398812651634216,
+      "learning_rate": 6.869883263534205e-05,
+      "loss": 0.1923,
+      "step": 654
+    },
+    {
+      "epoch": 2.2053872053872055,
+      "grad_norm": 0.43145304918289185,
+      "learning_rate": 6.858973789521793e-05,
+      "loss": 0.1964,
+      "step": 655
+    },
+    {
+      "epoch": 2.208754208754209,
+      "grad_norm": 0.4020119309425354,
+      "learning_rate": 6.848054036295666e-05,
+      "loss": 0.1822,
+      "step": 656
+    },
+    {
+      "epoch": 2.212121212121212,
+      "grad_norm": 0.29532957077026367,
+      "learning_rate": 6.837124064236709e-05,
+      "loss": 0.1928,
+      "step": 657
+    },
+    {
+      "epoch": 2.2154882154882154,
+      "grad_norm": 0.3485638201236725,
+      "learning_rate": 6.826183933782307e-05,
+      "loss": 0.2078,
+      "step": 658
+    },
+    {
+      "epoch": 2.218855218855219,
+      "grad_norm": 0.3406836986541748,
+      "learning_rate": 6.815233705426019e-05,
+      "loss": 0.1844,
+      "step": 659
+    },
+    {
+      "epoch": 2.2222222222222223,
+      "grad_norm": 0.2849566638469696,
+      "learning_rate": 6.80427343971724e-05,
+      "loss": 0.1914,
+      "step": 660
+    },
+    {
+      "epoch": 2.225589225589226,
+      "grad_norm": 0.34194567799568176,
+      "learning_rate": 6.793303197260864e-05,
+      "loss": 0.1879,
+      "step": 661
+    },
+    {
+      "epoch": 2.228956228956229,
+      "grad_norm": 0.39901259541511536,
+      "learning_rate": 6.782323038716956e-05,
+      "loss": 0.1963,
+      "step": 662
+    },
+    {
+      "epoch": 2.2323232323232323,
+      "grad_norm": 0.37613964080810547,
+      "learning_rate": 6.771333024800411e-05,
+      "loss": 0.1827,
+      "step": 663
+    },
+    {
+      "epoch": 2.2356902356902357,
+      "grad_norm": 0.3915823996067047,
+      "learning_rate": 6.760333216280617e-05,
+      "loss": 0.1988,
+      "step": 664
+    },
+    {
+      "epoch": 2.239057239057239,
+      "grad_norm": 0.3280683755874634,
+      "learning_rate": 6.74932367398112e-05,
+      "loss": 0.1798,
+      "step": 665
+    },
+    {
+      "epoch": 2.242424242424242,
+      "grad_norm": 0.4062442481517792,
+      "learning_rate": 6.738304458779293e-05,
+      "loss": 0.1935,
+      "step": 666
+    },
+    {
+      "epoch": 2.2457912457912457,
+      "grad_norm": 0.3807975947856903,
+      "learning_rate": 6.727275631605995e-05,
+      "loss": 0.19,
+      "step": 667
+    },
+    {
+      "epoch": 2.249158249158249,
+      "grad_norm": 0.33653566241264343,
+      "learning_rate": 6.716237253445235e-05,
+      "loss": 0.1783,
+      "step": 668
+    },
+    {
+      "epoch": 2.2525252525252526,
+      "grad_norm": 0.33660075068473816,
+      "learning_rate": 6.70518938533383e-05,
+      "loss": 0.1798,
+      "step": 669
+    },
+    {
+      "epoch": 2.255892255892256,
+      "grad_norm": 0.33384770154953003,
+      "learning_rate": 6.694132088361075e-05,
+      "loss": 0.1796,
+      "step": 670
+    },
+    {
+      "epoch": 2.259259259259259,
+      "grad_norm": 0.3794649839401245,
+      "learning_rate": 6.683065423668403e-05,
+      "loss": 0.194,
+      "step": 671
+    },
+    {
+      "epoch": 2.2626262626262625,
+      "grad_norm": 0.3536298871040344,
+      "learning_rate": 6.671989452449043e-05,
+      "loss": 0.1814,
+      "step": 672
+    },
+    {
+      "epoch": 2.265993265993266,
+      "grad_norm": 0.33530497550964355,
+      "learning_rate": 6.660904235947687e-05,
+      "loss": 0.1766,
+      "step": 673
+    },
+    {
+      "epoch": 2.2693602693602695,
+      "grad_norm": 0.38430067896842957,
+      "learning_rate": 6.649809835460147e-05,
+      "loss": 0.1834,
+      "step": 674
+    },
+    {
+      "epoch": 2.2727272727272725,
+      "grad_norm": 0.4088013470172882,
+      "learning_rate": 6.638706312333018e-05,
+      "loss": 0.184,
+      "step": 675
+    },
+    {
+      "epoch": 2.276094276094276,
+      "grad_norm": 0.4313696324825287,
+      "learning_rate": 6.627593727963342e-05,
+      "loss": 0.195,
+      "step": 676
+    },
+    {
+      "epoch": 2.2794612794612794,
+      "grad_norm": 0.3910023272037506,
+      "learning_rate": 6.616472143798261e-05,
+      "loss": 0.1955,
+      "step": 677
+    },
+    {
+      "epoch": 2.282828282828283,
+      "grad_norm": 0.3856333792209625,
+      "learning_rate": 6.605341621334683e-05,
+      "loss": 0.1825,
+      "step": 678
+    },
+    {
+      "epoch": 2.2861952861952863,
+      "grad_norm": 0.3617812395095825,
+      "learning_rate": 6.594202222118942e-05,
+      "loss": 0.1866,
+      "step": 679
+    },
+    {
+      "epoch": 2.28956228956229,
+      "grad_norm": 0.38682255148887634,
+      "learning_rate": 6.583054007746452e-05,
+      "loss": 0.1892,
+      "step": 680
+    },
+    {
+      "epoch": 2.292929292929293,
+      "grad_norm": 0.40745124220848083,
+      "learning_rate": 6.571897039861377e-05,
+      "loss": 0.1917,
+      "step": 681
+    },
+    {
+      "epoch": 2.2962962962962963,
+      "grad_norm": 0.3059500455856323,
+      "learning_rate": 6.560731380156275e-05,
+      "loss": 0.1807,
+      "step": 682
+    },
+    {
+      "epoch": 2.2996632996632997,
+      "grad_norm": 0.3456449806690216,
+      "learning_rate": 6.549557090371776e-05,
+      "loss": 0.1798,
+      "step": 683
+    },
+    {
+      "epoch": 2.303030303030303,
+      "grad_norm": 0.4222232699394226,
+      "learning_rate": 6.538374232296221e-05,
+      "loss": 0.1939,
+      "step": 684
+    },
+    {
+      "epoch": 2.3063973063973062,
+      "grad_norm": 0.32959309220314026,
+      "learning_rate": 6.527182867765332e-05,
+      "loss": 0.1839,
+      "step": 685
+    },
+    {
+      "epoch": 2.3097643097643097,
+      "grad_norm": 0.30231377482414246,
+      "learning_rate": 6.515983058661872e-05,
+      "loss": 0.1772,
+      "step": 686
+    },
+    {
+      "epoch": 2.313131313131313,
+      "grad_norm": 0.42110297083854675,
+      "learning_rate": 6.50477486691529e-05,
+      "loss": 0.1911,
+      "step": 687
+    },
+    {
+      "epoch": 2.3164983164983166,
+      "grad_norm": 0.39979901909828186,
+      "learning_rate": 6.493558354501397e-05,
+      "loss": 0.1828,
+      "step": 688
+    },
+    {
+      "epoch": 2.31986531986532,
+      "grad_norm": 0.3712976276874542,
+      "learning_rate": 6.482333583442002e-05,
+      "loss": 0.1841,
+      "step": 689
+    },
+    {
+      "epoch": 2.323232323232323,
+      "grad_norm": 0.4817717373371124,
+      "learning_rate": 6.471100615804591e-05,
+      "loss": 0.1844,
+      "step": 690
+    },
+    {
+      "epoch": 2.3265993265993266,
+      "grad_norm": 0.4232378304004669,
+      "learning_rate": 6.459859513701967e-05,
+      "loss": 0.1748,
+      "step": 691
+    },
+    {
+      "epoch": 2.32996632996633,
+      "grad_norm": 0.3409404158592224,
+      "learning_rate": 6.448610339291912e-05,
+      "loss": 0.1944,
+      "step": 692
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 0.3255225121974945,
+      "learning_rate": 6.437353154776849e-05,
+      "loss": 0.196,
+      "step": 693
+    },
+    {
+      "epoch": 2.3367003367003365,
+      "grad_norm": 0.3025372624397278,
+      "learning_rate": 6.426088022403485e-05,
+      "loss": 0.1706,
+      "step": 694
+    },
+    {
+      "epoch": 2.34006734006734,
+      "grad_norm": 0.3823414146900177,
+      "learning_rate": 6.414815004462483e-05,
+      "loss": 0.1814,
+      "step": 695
+    },
+    {
+      "epoch": 2.3434343434343434,
+      "grad_norm": 0.4448075294494629,
+      "learning_rate": 6.403534163288105e-05,
+      "loss": 0.1966,
+      "step": 696
+    },
+    {
+      "epoch": 2.346801346801347,
+      "grad_norm": 0.36923545598983765,
+      "learning_rate": 6.392245561257871e-05,
+      "loss": 0.1896,
+      "step": 697
+    },
+    {
+      "epoch": 2.3501683501683504,
+      "grad_norm": 0.3588995933532715,
+      "learning_rate": 6.380949260792217e-05,
+      "loss": 0.1817,
+      "step": 698
+    },
+    {
+      "epoch": 2.3535353535353534,
+      "grad_norm": 0.4320705235004425,
+      "learning_rate": 6.369645324354149e-05,
+      "loss": 0.1918,
+      "step": 699
+    },
+    {
+      "epoch": 2.356902356902357,
+      "grad_norm": 0.4778941869735718,
+      "learning_rate": 6.358333814448891e-05,
+      "loss": 0.1923,
+      "step": 700
+    },
+    {
+      "epoch": 2.356902356902357,
+      "eval_loss": 0.09319520741701126,
+      "eval_runtime": 32.7439,
+      "eval_samples_per_second": 30.54,
+      "eval_steps_per_second": 1.924,
+      "step": 700
+    },
+    {
+      "epoch": 2.3602693602693603,
+      "grad_norm": 0.4958076775074005,
+      "learning_rate": 6.347014793623547e-05,
+      "loss": 0.1977,
+      "step": 701
+    },
+    {
+      "epoch": 2.3636363636363638,
+      "grad_norm": 0.3545849025249481,
+      "learning_rate": 6.335688324466756e-05,
+      "loss": 0.1878,
+      "step": 702
+    },
+    {
+      "epoch": 2.3670033670033668,
+      "grad_norm": 0.40290120244026184,
+      "learning_rate": 6.324354469608335e-05,
+      "loss": 0.1789,
+      "step": 703
+    },
+    {
+      "epoch": 2.3703703703703702,
+      "grad_norm": 0.3689594864845276,
+      "learning_rate": 6.313013291718952e-05,
+      "loss": 0.1869,
+      "step": 704
+    },
+    {
+      "epoch": 2.3737373737373737,
+      "grad_norm": 0.34308475255966187,
+      "learning_rate": 6.301664853509754e-05,
+      "loss": 0.1877,
+      "step": 705
+    },
+    {
+      "epoch": 2.377104377104377,
+      "grad_norm": 0.3406205475330353,
+      "learning_rate": 6.290309217732045e-05,
+      "loss": 0.1835,
+      "step": 706
+    },
+    {
+      "epoch": 2.3804713804713806,
+      "grad_norm": 0.39080610871315,
+      "learning_rate": 6.278946447176923e-05,
+      "loss": 0.1968,
+      "step": 707
+    },
+    {
+      "epoch": 2.3838383838383836,
+      "grad_norm": 0.3915717303752899,
+      "learning_rate": 6.267576604674938e-05,
+      "loss": 0.1887,
+      "step": 708
+    },
+    {
+      "epoch": 2.387205387205387,
+      "grad_norm": 0.3860926628112793,
+      "learning_rate": 6.256199753095745e-05,
+      "loss": 0.1856,
+      "step": 709
+    },
+    {
+      "epoch": 2.3905723905723906,
+      "grad_norm": 0.3959527313709259,
+      "learning_rate": 6.244815955347756e-05,
+      "loss": 0.2006,
+      "step": 710
+    },
+    {
+      "epoch": 2.393939393939394,
+      "grad_norm": 0.3251626491546631,
+      "learning_rate": 6.233425274377794e-05,
+      "loss": 0.1742,
+      "step": 711
+    },
+    {
+      "epoch": 2.3973063973063975,
+      "grad_norm": 0.34222376346588135,
+      "learning_rate": 6.222027773170737e-05,
+      "loss": 0.1938,
+      "step": 712
+    },
+    {
+      "epoch": 2.4006734006734005,
+      "grad_norm": 0.2963486313819885,
+      "learning_rate": 6.21062351474918e-05,
+      "loss": 0.1775,
+      "step": 713
+    },
+    {
+      "epoch": 2.404040404040404,
+      "grad_norm": 0.33488693833351135,
+      "learning_rate": 6.199212562173084e-05,
+      "loss": 0.1859,
+      "step": 714
+    },
+    {
+      "epoch": 2.4074074074074074,
+      "grad_norm": 0.37009695172309875,
+      "learning_rate": 6.187794978539419e-05,
+      "loss": 0.1854,
+      "step": 715
+    },
+    {
+      "epoch": 2.410774410774411,
+      "grad_norm": 0.29789450764656067,
+      "learning_rate": 6.176370826981828e-05,
+      "loss": 0.1782,
+      "step": 716
+    },
+    {
+      "epoch": 2.4141414141414144,
+      "grad_norm": 0.3971637189388275,
+      "learning_rate": 6.164940170670266e-05,
+      "loss": 0.1846,
+      "step": 717
+    },
+    {
+      "epoch": 2.4175084175084174,
+      "grad_norm": 0.3055313229560852,
+      "learning_rate": 6.153503072810662e-05,
+      "loss": 0.1892,
+      "step": 718
+    },
+    {
+      "epoch": 2.420875420875421,
+      "grad_norm": 0.3349987864494324,
+      "learning_rate": 6.142059596644558e-05,
+      "loss": 0.1864,
+      "step": 719
+    },
+    {
+      "epoch": 2.4242424242424243,
+      "grad_norm": 0.3594989478588104,
+      "learning_rate": 6.130609805448766e-05,
+      "loss": 0.1745,
+      "step": 720
+    },
+    {
+      "epoch": 2.4276094276094278,
+      "grad_norm": 0.32809680700302124,
+      "learning_rate": 6.119153762535021e-05,
+      "loss": 0.1874,
+      "step": 721
+    },
+    {
+      "epoch": 2.430976430976431,
+      "grad_norm": 0.39594766497612,
+      "learning_rate": 6.107691531249623e-05,
+      "loss": 0.1838,
+      "step": 722
+    },
+    {
+      "epoch": 2.4343434343434343,
+      "grad_norm": 0.31413084268569946,
+      "learning_rate": 6.09622317497309e-05,
+      "loss": 0.1872,
+      "step": 723
+    },
+    {
+      "epoch": 2.4377104377104377,
+      "grad_norm": 0.33545947074890137,
+      "learning_rate": 6.084748757119811e-05,
+      "loss": 0.1831,
+      "step": 724
+    },
+    {
+      "epoch": 2.441077441077441,
+      "grad_norm": 0.3555618226528168,
+      "learning_rate": 6.0732683411376935e-05,
+      "loss": 0.1773,
+      "step": 725
+    },
+    {
+      "epoch": 2.4444444444444446,
+      "grad_norm": 0.31497722864151,
+      "learning_rate": 6.0617819905078075e-05,
+      "loss": 0.1957,
+      "step": 726
+    },
+    {
+      "epoch": 2.4478114478114477,
+      "grad_norm": 0.3945048451423645,
+      "learning_rate": 6.050289768744042e-05,
+      "loss": 0.1844,
+      "step": 727
+    },
+    {
+      "epoch": 2.451178451178451,
+      "grad_norm": 0.3777752220630646,
+      "learning_rate": 6.038791739392748e-05,
+      "loss": 0.1765,
+      "step": 728
+    },
+    {
+      "epoch": 2.4545454545454546,
+      "grad_norm": 0.3376706540584564,
+      "learning_rate": 6.0272879660323934e-05,
+      "loss": 0.1857,
+      "step": 729
+    },
+    {
+      "epoch": 2.457912457912458,
+      "grad_norm": 0.33173221349716187,
+      "learning_rate": 6.015778512273203e-05,
+      "loss": 0.1877,
+      "step": 730
+    },
+    {
+      "epoch": 2.461279461279461,
+      "grad_norm": 0.26840299367904663,
+      "learning_rate": 6.004263441756815e-05,
+      "loss": 0.1898,
+      "step": 731
+    },
+    {
+      "epoch": 2.4646464646464645,
+      "grad_norm": 0.3570701479911804,
+      "learning_rate": 5.992742818155923e-05,
+      "loss": 0.1794,
+      "step": 732
+    },
+    {
+      "epoch": 2.468013468013468,
+      "grad_norm": 0.2803226113319397,
+      "learning_rate": 5.98121670517393e-05,
+      "loss": 0.1826,
+      "step": 733
+    },
+    {
+      "epoch": 2.4713804713804715,
+      "grad_norm": 0.36934715509414673,
+      "learning_rate": 5.9696851665445875e-05,
+      "loss": 0.1855,
+      "step": 734
+    },
+    {
+      "epoch": 2.474747474747475,
+      "grad_norm": 0.38208091259002686,
+      "learning_rate": 5.958148266031654e-05,
+      "loss": 0.1872,
+      "step": 735
+    },
+    {
+      "epoch": 2.478114478114478,
+      "grad_norm": 0.2921079099178314,
+      "learning_rate": 5.9466060674285294e-05,
+      "loss": 0.1887,
+      "step": 736
+    },
+    {
+      "epoch": 2.4814814814814814,
+      "grad_norm": 0.4000777006149292,
+      "learning_rate": 5.9350586345579165e-05,
+      "loss": 0.1932,
+      "step": 737
+    },
+    {
+      "epoch": 2.484848484848485,
+      "grad_norm": 0.3915836215019226,
+      "learning_rate": 5.9235060312714564e-05,
+      "loss": 0.1905,
+      "step": 738
+    },
+    {
+      "epoch": 2.4882154882154883,
+      "grad_norm": 0.3251096308231354,
+      "learning_rate": 5.9119483214493844e-05,
+      "loss": 0.1763,
+      "step": 739
+    },
+    {
+      "epoch": 2.4915824915824913,
+      "grad_norm": 0.3224780261516571,
+      "learning_rate": 5.900385569000166e-05,
+      "loss": 0.1734,
+      "step": 740
+    },
+    {
+      "epoch": 2.494949494949495,
+      "grad_norm": 0.35873523354530334,
+      "learning_rate": 5.8888178378601565e-05,
+      "loss": 0.1821,
+      "step": 741
+    },
+    {
+      "epoch": 2.4983164983164983,
+      "grad_norm": 0.31311190128326416,
+      "learning_rate": 5.877245191993239e-05,
+      "loss": 0.1913,
+      "step": 742
+    },
+    {
+      "epoch": 2.5016835016835017,
+      "grad_norm": 0.30053961277008057,
+      "learning_rate": 5.865667695390468e-05,
+      "loss": 0.175,
+      "step": 743
+    },
+    {
+      "epoch": 2.505050505050505,
+      "grad_norm": 0.3176406919956207,
+      "learning_rate": 5.854085412069726e-05,
+      "loss": 0.1776,
+      "step": 744
+    },
+    {
+      "epoch": 2.5084175084175087,
+      "grad_norm": 0.31768059730529785,
+      "learning_rate": 5.842498406075363e-05,
+      "loss": 0.1767,
+      "step": 745
+    },
+    {
+      "epoch": 2.5117845117845117,
+      "grad_norm": 0.32042112946510315,
+      "learning_rate": 5.8309067414778404e-05,
+      "loss": 0.1783,
+      "step": 746
+    },
+    {
+      "epoch": 2.515151515151515,
+      "grad_norm": 0.2806006669998169,
+      "learning_rate": 5.81931048237338e-05,
+      "loss": 0.1839,
+      "step": 747
+    },
+    {
+      "epoch": 2.5185185185185186,
+      "grad_norm": 0.3236968517303467,
+      "learning_rate": 5.807709692883612e-05,
+      "loss": 0.1798,
+      "step": 748
+    },
+    {
+      "epoch": 2.5218855218855216,
+      "grad_norm": 0.3542146682739258,
+      "learning_rate": 5.796104437155213e-05,
+      "loss": 0.1768,
+      "step": 749
+    },
+    {
+      "epoch": 2.525252525252525,
+      "grad_norm": 0.3283354938030243,
+      "learning_rate": 5.7844947793595584e-05,
+      "loss": 0.1817,
+      "step": 750
+    },
+    {
+      "epoch": 2.5286195286195285,
+      "grad_norm": 0.35863611102104187,
+      "learning_rate": 5.7728807836923624e-05,
+      "loss": 0.1886,
+      "step": 751
+    },
+    {
+      "epoch": 2.531986531986532,
+      "grad_norm": 0.32639971375465393,
+      "learning_rate": 5.7612625143733325e-05,
+      "loss": 0.1835,
+      "step": 752
+    },
+    {
+      "epoch": 2.5353535353535355,
+      "grad_norm": 0.36184388399124146,
+      "learning_rate": 5.749640035645798e-05,
+      "loss": 0.1858,
+      "step": 753
+    },
+    {
+      "epoch": 2.538720538720539,
+      "grad_norm": 0.35223841667175293,
+      "learning_rate": 5.73801341177637e-05,
+      "loss": 0.1745,
+      "step": 754
+    },
+    {
+      "epoch": 2.542087542087542,
+      "grad_norm": 0.40928134322166443,
+      "learning_rate": 5.7263827070545775e-05,
+      "loss": 0.1889,
+      "step": 755
+    },
+    {
+      "epoch": 2.5454545454545454,
+      "grad_norm": 0.33686789870262146,
+      "learning_rate": 5.714747985792516e-05,
+      "loss": 0.1782,
+      "step": 756
+    },
+    {
+      "epoch": 2.548821548821549,
+      "grad_norm": 0.36779022216796875,
+      "learning_rate": 5.7031093123244925e-05,
+      "loss": 0.1824,
+      "step": 757
+    },
+    {
+      "epoch": 2.5521885521885523,
+      "grad_norm": 0.412527471780777,
+      "learning_rate": 5.691466751006662e-05,
+      "loss": 0.1761,
+      "step": 758
+    },
+    {
+      "epoch": 2.5555555555555554,
+      "grad_norm": 0.49104923009872437,
+      "learning_rate": 5.679820366216684e-05,
+      "loss": 0.1861,
+      "step": 759
+    },
+    {
+      "epoch": 2.558922558922559,
+      "grad_norm": 0.3897760808467865,
+      "learning_rate": 5.668170222353355e-05,
+      "loss": 0.188,
+      "step": 760
+    },
+    {
+      "epoch": 2.5622895622895623,
+      "grad_norm": 0.37621814012527466,
+      "learning_rate": 5.656516383836262e-05,
+      "loss": 0.1682,
+      "step": 761
+    },
+    {
+      "epoch": 2.5656565656565657,
+      "grad_norm": 0.4133719801902771,
+      "learning_rate": 5.644858915105414e-05,
+      "loss": 0.189,
+      "step": 762
+    },
+    {
+      "epoch": 2.569023569023569,
+      "grad_norm": 0.4049244225025177,
+      "learning_rate": 5.633197880620904e-05,
+      "loss": 0.1889,
+      "step": 763
+    },
+    {
+      "epoch": 2.5723905723905722,
+      "grad_norm": 0.46882614493370056,
+      "learning_rate": 5.621533344862531e-05,
+      "loss": 0.1825,
+      "step": 764
+    },
+    {
+      "epoch": 2.5757575757575757,
+      "grad_norm": 0.3228312134742737,
+      "learning_rate": 5.6098653723294604e-05,
+      "loss": 0.1775,
+      "step": 765
+    },
+    {
+      "epoch": 2.579124579124579,
+      "grad_norm": 0.4066309630870819,
+      "learning_rate": 5.598194027539862e-05,
+      "loss": 0.1818,
+      "step": 766
+    },
+    {
+      "epoch": 2.5824915824915826,
+      "grad_norm": 0.32752707600593567,
+      "learning_rate": 5.586519375030549e-05,
+      "loss": 0.1806,
+      "step": 767
+    },
+    {
+      "epoch": 2.5858585858585856,
+      "grad_norm": 0.35238194465637207,
+      "learning_rate": 5.574841479356627e-05,
+      "loss": 0.1719,
+      "step": 768
+    },
+    {
+      "epoch": 2.589225589225589,
+      "grad_norm": 0.3612888753414154,
+      "learning_rate": 5.563160405091136e-05,
+      "loss": 0.1646,
+      "step": 769
+    },
+    {
+      "epoch": 2.5925925925925926,
+      "grad_norm": 0.34333643317222595,
+      "learning_rate": 5.551476216824687e-05,
+      "loss": 0.1727,
+      "step": 770
+    },
+    {
+      "epoch": 2.595959595959596,
+      "grad_norm": 0.47696343064308167,
+      "learning_rate": 5.5397889791651145e-05,
+      "loss": 0.1844,
+      "step": 771
+    },
+    {
+      "epoch": 2.5993265993265995,
+      "grad_norm": 0.3781977891921997,
+      "learning_rate": 5.5280987567371134e-05,
+      "loss": 0.1753,
+      "step": 772
+    },
+    {
+      "epoch": 2.602693602693603,
+      "grad_norm": 0.4676324129104614,
+      "learning_rate": 5.516405614181883e-05,
+      "loss": 0.1839,
+      "step": 773
+    },
+    {
+      "epoch": 2.606060606060606,
+      "grad_norm": 0.3599601686000824,
+      "learning_rate": 5.504709616156768e-05,
+      "loss": 0.1786,
+      "step": 774
+    },
+    {
+      "epoch": 2.6094276094276094,
+      "grad_norm": 0.3672841191291809,
+      "learning_rate": 5.493010827334904e-05,
+      "loss": 0.185,
+      "step": 775
+    },
+    {
+      "epoch": 2.612794612794613,
+      "grad_norm": 0.44677501916885376,
+      "learning_rate": 5.4813093124048585e-05,
+      "loss": 0.1951,
+      "step": 776
+    },
+    {
+      "epoch": 2.616161616161616,
+      "grad_norm": 0.3104342818260193,
+      "learning_rate": 5.4696051360702725e-05,
+      "loss": 0.1767,
+      "step": 777
+    },
+    {
+      "epoch": 2.6195286195286194,
+      "grad_norm": 0.3832162618637085,
+      "learning_rate": 5.457898363049504e-05,
+      "loss": 0.1808,
+      "step": 778
+    },
+    {
+      "epoch": 2.622895622895623,
+      "grad_norm": 0.4450579583644867,
+      "learning_rate": 5.446189058075265e-05,
+      "loss": 0.1829,
+      "step": 779
+    },
+    {
+      "epoch": 2.6262626262626263,
+      "grad_norm": 0.31873807311058044,
+      "learning_rate": 5.434477285894277e-05,
+      "loss": 0.182,
+      "step": 780
+    },
+    {
+      "epoch": 2.6296296296296298,
+      "grad_norm": 0.42338353395462036,
+      "learning_rate": 5.4227631112668955e-05,
+      "loss": 0.1931,
+      "step": 781
+    },
+    {
+      "epoch": 2.6329966329966332,
+      "grad_norm": 0.2841234803199768,
+      "learning_rate": 5.411046598966764e-05,
+      "loss": 0.1716,
+      "step": 782
+    },
+    {
+      "epoch": 2.6363636363636362,
+      "grad_norm": 0.4316200315952301,
+      "learning_rate": 5.3993278137804505e-05,
+      "loss": 0.1821,
+      "step": 783
+    },
+    {
+      "epoch": 2.6397306397306397,
+      "grad_norm": 0.3666791021823883,
+      "learning_rate": 5.387606820507095e-05,
+      "loss": 0.1828,
+      "step": 784
+    },
+    {
+      "epoch": 2.643097643097643,
+      "grad_norm": 0.37707844376564026,
+      "learning_rate": 5.375883683958041e-05,
+      "loss": 0.1731,
+      "step": 785
+    },
+    {
+      "epoch": 2.6464646464646466,
+      "grad_norm": 0.41289880871772766,
+      "learning_rate": 5.3641584689564875e-05,
+      "loss": 0.188,
+      "step": 786
+    },
+    {
+      "epoch": 2.6498316498316496,
+      "grad_norm": 0.3792349696159363,
+      "learning_rate": 5.3524312403371257e-05,
+      "loss": 0.1859,
+      "step": 787
+    },
+    {
+      "epoch": 2.653198653198653,
+      "grad_norm": 0.33042725920677185,
+      "learning_rate": 5.34070206294578e-05,
+      "loss": 0.1802,
+      "step": 788
+    },
+    {
+      "epoch": 2.6565656565656566,
+      "grad_norm": 0.423698365688324,
+      "learning_rate": 5.3289710016390535e-05,
+      "loss": 0.1786,
+      "step": 789
+    },
+    {
+      "epoch": 2.65993265993266,
+      "grad_norm": 0.3488188683986664,
+      "learning_rate": 5.3172381212839614e-05,
+      "loss": 0.174,
+      "step": 790
+    },
+    {
+      "epoch": 2.6632996632996635,
+      "grad_norm": 0.4271824359893799,
+      "learning_rate": 5.3055034867575826e-05,
+      "loss": 0.182,
+      "step": 791
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.3261590003967285,
+      "learning_rate": 5.293767162946691e-05,
+      "loss": 0.1556,
+      "step": 792
+    },
+    {
+      "epoch": 2.67003367003367,
+      "grad_norm": 0.35868170857429504,
+      "learning_rate": 5.282029214747404e-05,
+      "loss": 0.1693,
+      "step": 793
+    },
+    {
+      "epoch": 2.6734006734006734,
+      "grad_norm": 0.44755569100379944,
+      "learning_rate": 5.270289707064822e-05,
+      "loss": 0.1895,
+      "step": 794
+    },
+    {
+      "epoch": 2.676767676767677,
+      "grad_norm": 0.42198193073272705,
+      "learning_rate": 5.258548704812667e-05,
+      "loss": 0.1873,
+      "step": 795
+    },
+    {
+      "epoch": 2.68013468013468,
+      "grad_norm": 0.518511176109314,
+      "learning_rate": 5.246806272912925e-05,
+      "loss": 0.1769,
+      "step": 796
+    },
+    {
+      "epoch": 2.6835016835016834,
+      "grad_norm": 0.38407737016677856,
+      "learning_rate": 5.2350624762954884e-05,
+      "loss": 0.1836,
+      "step": 797
+    },
+    {
+      "epoch": 2.686868686868687,
+      "grad_norm": 0.3742847740650177,
+      "learning_rate": 5.223317379897794e-05,
+      "loss": 0.1726,
+      "step": 798
+    },
+    {
+      "epoch": 2.6902356902356903,
+      "grad_norm": 0.43547365069389343,
+      "learning_rate": 5.211571048664469e-05,
+      "loss": 0.1772,
+      "step": 799
+    },
+    {
+      "epoch": 2.6936026936026938,
+      "grad_norm": 0.31277012825012207,
+      "learning_rate": 5.199823547546963e-05,
+      "loss": 0.179,
+      "step": 800
+    },
+    {
+      "epoch": 2.6936026936026938,
+      "eval_loss": 0.0905195027589798,
+      "eval_runtime": 33.1678,
+      "eval_samples_per_second": 30.15,
+      "eval_steps_per_second": 1.899,
+      "step": 800
+    },
+    {
+      "epoch": 2.6969696969696972,
+      "grad_norm": 0.4289379119873047,
+      "learning_rate": 5.188074941503203e-05,
+      "loss": 0.1824,
+      "step": 801
+    },
+    {
+      "epoch": 2.7003367003367003,
+      "grad_norm": 0.3402915596961975,
+      "learning_rate": 5.1763252954972167e-05,
+      "loss": 0.1723,
+      "step": 802
+    },
+    {
+      "epoch": 2.7037037037037037,
+      "grad_norm": 0.36200371384620667,
+      "learning_rate": 5.164574674498788e-05,
+      "loss": 0.1871,
+      "step": 803
+    },
+    {
+      "epoch": 2.707070707070707,
+      "grad_norm": 0.39011457562446594,
+      "learning_rate": 5.152823143483092e-05,
+      "loss": 0.1686,
+      "step": 804
+    },
+    {
+      "epoch": 2.71043771043771,
+      "grad_norm": 0.2998808026313782,
+      "learning_rate": 5.14107076743033e-05,
+      "loss": 0.1728,
+      "step": 805
+    },
+    {
+      "epoch": 2.7138047138047137,
+      "grad_norm": 0.3582398593425751,
+      "learning_rate": 5.129317611325385e-05,
+      "loss": 0.1764,
+      "step": 806
+    },
+    {
+      "epoch": 2.717171717171717,
+      "grad_norm": 0.36014533042907715,
+      "learning_rate": 5.117563740157444e-05,
+      "loss": 0.1752,
+      "step": 807
+    },
+    {
+      "epoch": 2.7205387205387206,
+      "grad_norm": 0.38851645588874817,
+      "learning_rate": 5.105809218919656e-05,
+      "loss": 0.1873,
+      "step": 808
+    },
+    {
+      "epoch": 2.723905723905724,
+      "grad_norm": 0.40050944685935974,
+      "learning_rate": 5.094054112608758e-05,
+      "loss": 0.1839,
+      "step": 809
+    },
+    {
+      "epoch": 2.7272727272727275,
+      "grad_norm": 0.4239293038845062,
+      "learning_rate": 5.082298486224728e-05,
+      "loss": 0.174,
+      "step": 810
+    },
+    {
+      "epoch": 2.7306397306397305,
+      "grad_norm": 0.34228378534317017,
+      "learning_rate": 5.070542404770413e-05,
+      "loss": 0.1751,
+      "step": 811
+    },
+    {
+      "epoch": 2.734006734006734,
+      "grad_norm": 0.3773192763328552,
+      "learning_rate": 5.058785933251184e-05,
+      "loss": 0.1814,
+      "step": 812
+    },
+    {
+      "epoch": 2.7373737373737375,
+      "grad_norm": 0.3601987659931183,
+      "learning_rate": 5.047029136674563e-05,
+      "loss": 0.1854,
+      "step": 813
+    },
+    {
+      "epoch": 2.7407407407407405,
+      "grad_norm": 0.33943113684654236,
+      "learning_rate": 5.035272080049871e-05,
+      "loss": 0.1849,
+      "step": 814
+    },
+    {
+      "epoch": 2.744107744107744,
+      "grad_norm": 0.4696902334690094,
+      "learning_rate": 5.0235148283878675e-05,
+      "loss": 0.1879,
+      "step": 815
+    },
+    {
+      "epoch": 2.7474747474747474,
+      "grad_norm": 0.3241429626941681,
+      "learning_rate": 5.011757446700392e-05,
+      "loss": 0.1746,
+      "step": 816
+    },
+    {
+      "epoch": 2.750841750841751,
+      "grad_norm": 0.4660000205039978,
+      "learning_rate": 5e-05,
+      "loss": 0.1808,
+      "step": 817
+    },
+    {
+      "epoch": 2.7542087542087543,
+      "grad_norm": 0.3961378335952759,
+      "learning_rate": 4.988242553299608e-05,
+      "loss": 0.1779,
+      "step": 818
+    },
+    {
+      "epoch": 2.757575757575758,
+      "grad_norm": 0.41490015387535095,
+      "learning_rate": 4.9764851716121337e-05,
+      "loss": 0.1785,
+      "step": 819
+    },
+    {
+      "epoch": 2.760942760942761,
+      "grad_norm": 0.4009009003639221,
+      "learning_rate": 4.964727919950131e-05,
+      "loss": 0.1722,
+      "step": 820
+    },
+    {
+      "epoch": 2.7643097643097643,
+      "grad_norm": 0.3752220869064331,
+      "learning_rate": 4.95297086332544e-05,
+      "loss": 0.1829,
+      "step": 821
+    },
+    {
+      "epoch": 2.7676767676767677,
+      "grad_norm": 0.33247944712638855,
+      "learning_rate": 4.941214066748818e-05,
+      "loss": 0.1805,
+      "step": 822
+    },
+    {
+      "epoch": 2.771043771043771,
+      "grad_norm": 0.36337146162986755,
+      "learning_rate": 4.929457595229589e-05,
+      "loss": 0.1823,
+      "step": 823
+    },
+    {
+      "epoch": 2.774410774410774,
+      "grad_norm": 0.3234810531139374,
+      "learning_rate": 4.9177015137752726e-05,
+      "loss": 0.1699,
+      "step": 824
+    },
+    {
+      "epoch": 2.7777777777777777,
+      "grad_norm": 0.37526530027389526,
+      "learning_rate": 4.905945887391242e-05,
+      "loss": 0.1706,
+      "step": 825
+    },
+    {
+      "epoch": 2.781144781144781,
+      "grad_norm": 0.3150089383125305,
+      "learning_rate": 4.8941907810803436e-05,
+      "loss": 0.1742,
+      "step": 826
+    },
+    {
+      "epoch": 2.7845117845117846,
+      "grad_norm": 0.441749632358551,
+      "learning_rate": 4.882436259842556e-05,
+      "loss": 0.1766,
+      "step": 827
+    },
+    {
+      "epoch": 2.787878787878788,
+      "grad_norm": 0.3168331980705261,
+      "learning_rate": 4.870682388674616e-05,
+      "loss": 0.1777,
+      "step": 828
+    },
+    {
+      "epoch": 2.791245791245791,
+      "grad_norm": 0.3491107225418091,
+      "learning_rate": 4.858929232569671e-05,
+      "loss": 0.1798,
+      "step": 829
+    },
+    {
+      "epoch": 2.7946127946127945,
+      "grad_norm": 0.32505708932876587,
+      "learning_rate": 4.84717685651691e-05,
+      "loss": 0.1691,
+      "step": 830
+    },
+    {
+      "epoch": 2.797979797979798,
+      "grad_norm": 0.3730446994304657,
+      "learning_rate": 4.8354253255012134e-05,
+      "loss": 0.1851,
+      "step": 831
+    },
+    {
+      "epoch": 2.8013468013468015,
+      "grad_norm": 0.3792212903499603,
+      "learning_rate": 4.8236747045027845e-05,
+      "loss": 0.1855,
+      "step": 832
+    },
+    {
+      "epoch": 2.8047138047138045,
+      "grad_norm": 0.35931986570358276,
+      "learning_rate": 4.811925058496798e-05,
+      "loss": 0.1814,
+      "step": 833
+    },
+    {
+      "epoch": 2.808080808080808,
+      "grad_norm": 0.3025000989437103,
+      "learning_rate": 4.800176452453038e-05,
+      "loss": 0.1808,
+      "step": 834
+    },
+    {
+      "epoch": 2.8114478114478114,
+      "grad_norm": 0.35131269693374634,
+      "learning_rate": 4.788428951335534e-05,
+      "loss": 0.1779,
+      "step": 835
+    },
+    {
+      "epoch": 2.814814814814815,
+      "grad_norm": 0.3271830379962921,
+      "learning_rate": 4.776682620102208e-05,
+      "loss": 0.1872,
+      "step": 836
+    },
+    {
+      "epoch": 2.8181818181818183,
+      "grad_norm": 0.33299174904823303,
+      "learning_rate": 4.7649375237045135e-05,
+      "loss": 0.1696,
+      "step": 837
+    },
+    {
+      "epoch": 2.821548821548822,
+      "grad_norm": 0.36610648036003113,
+      "learning_rate": 4.7531937270870755e-05,
+      "loss": 0.1811,
+      "step": 838
+    },
+    {
+      "epoch": 2.824915824915825,
+      "grad_norm": 0.3686729371547699,
+      "learning_rate": 4.741451295187332e-05,
+      "loss": 0.183,
+      "step": 839
+    },
+    {
+      "epoch": 2.8282828282828283,
+      "grad_norm": 0.40744999051094055,
+      "learning_rate": 4.729710292935178e-05,
+      "loss": 0.1658,
+      "step": 840
+    },
+    {
+      "epoch": 2.8316498316498318,
+      "grad_norm": 0.3003140985965729,
+      "learning_rate": 4.717970785252595e-05,
+      "loss": 0.1709,
+      "step": 841
+    },
+    {
+      "epoch": 2.8350168350168348,
+      "grad_norm": 0.3362986147403717,
+      "learning_rate": 4.706232837053311e-05,
+      "loss": 0.1538,
+      "step": 842
+    },
+    {
+      "epoch": 2.8383838383838382,
+      "grad_norm": 0.3673784136772156,
+      "learning_rate": 4.6944965132424185e-05,
+      "loss": 0.1718,
+      "step": 843
+    },
+    {
+      "epoch": 2.8417508417508417,
+      "grad_norm": 0.37941762804985046,
+      "learning_rate": 4.682761878716039e-05,
+      "loss": 0.1683,
+      "step": 844
+    },
+    {
+      "epoch": 2.845117845117845,
+      "grad_norm": 0.4599365293979645,
+      "learning_rate": 4.671028998360947e-05,
+      "loss": 0.1746,
+      "step": 845
+    },
+    {
+      "epoch": 2.8484848484848486,
+      "grad_norm": 0.40842747688293457,
+      "learning_rate": 4.659297937054221e-05,
+      "loss": 0.1743,
+      "step": 846
+    },
+    {
+      "epoch": 2.851851851851852,
+      "grad_norm": 0.44149237871170044,
+      "learning_rate": 4.647568759662876e-05,
+      "loss": 0.1658,
+      "step": 847
+    },
+    {
+      "epoch": 2.855218855218855,
+      "grad_norm": 0.33786505460739136,
+      "learning_rate": 4.635841531043514e-05,
+      "loss": 0.1715,
+      "step": 848
+    },
+    {
+      "epoch": 2.8585858585858586,
+      "grad_norm": 0.38703563809394836,
+      "learning_rate": 4.6241163160419616e-05,
+      "loss": 0.1885,
+      "step": 849
+    },
+    {
+      "epoch": 2.861952861952862,
+      "grad_norm": 0.33466836810112,
+      "learning_rate": 4.612393179492907e-05,
+      "loss": 0.1727,
+      "step": 850
+    },
+    {
+      "epoch": 2.865319865319865,
+      "grad_norm": 0.3569240868091583,
+      "learning_rate": 4.600672186219551e-05,
+      "loss": 0.1594,
+      "step": 851
+    },
+    {
+      "epoch": 2.8686868686868685,
+      "grad_norm": 0.3640967905521393,
+      "learning_rate": 4.588953401033237e-05,
+      "loss": 0.1716,
+      "step": 852
+    },
+    {
+      "epoch": 2.872053872053872,
+      "grad_norm": 0.3225606679916382,
+      "learning_rate": 4.577236888733105e-05,
+      "loss": 0.1737,
+      "step": 853
+    },
+    {
+      "epoch": 2.8754208754208754,
+      "grad_norm": 0.4322353005409241,
+      "learning_rate": 4.565522714105723e-05,
+      "loss": 0.1818,
+      "step": 854
+    },
+    {
+      "epoch": 2.878787878787879,
+      "grad_norm": 0.43442046642303467,
+      "learning_rate": 4.553810941924735e-05,
+      "loss": 0.1707,
+      "step": 855
+    },
+    {
+      "epoch": 2.8821548821548824,
+      "grad_norm": 0.4079204797744751,
+      "learning_rate": 4.542101636950497e-05,
+      "loss": 0.1686,
+      "step": 856
+    },
+    {
+      "epoch": 2.8855218855218854,
+      "grad_norm": 0.4360505938529968,
+      "learning_rate": 4.5303948639297287e-05,
+      "loss": 0.1755,
+      "step": 857
+    },
+    {
+      "epoch": 2.888888888888889,
+      "grad_norm": 0.3284655213356018,
+      "learning_rate": 4.518690687595142e-05,
+      "loss": 0.1743,
+      "step": 858
+    },
+    {
+      "epoch": 2.8922558922558923,
+      "grad_norm": 0.41207629442214966,
+      "learning_rate": 4.5069891726650974e-05,
+      "loss": 0.1805,
+      "step": 859
+    },
+    {
+      "epoch": 2.8956228956228958,
+      "grad_norm": 0.42025890946388245,
+      "learning_rate": 4.495290383843233e-05,
+      "loss": 0.1646,
+      "step": 860
+    },
+    {
+      "epoch": 2.898989898989899,
+      "grad_norm": 0.2703207731246948,
+      "learning_rate": 4.483594385818118e-05,
+      "loss": 0.1772,
+      "step": 861
+    },
+    {
+      "epoch": 2.9023569023569022,
+      "grad_norm": 0.4024050831794739,
+      "learning_rate": 4.4719012432628884e-05,
+      "loss": 0.1713,
+      "step": 862
+    },
+    {
+      "epoch": 2.9057239057239057,
+      "grad_norm": 0.32182058691978455,
+      "learning_rate": 4.460211020834887e-05,
+      "loss": 0.164,
+      "step": 863
+    },
+    {
+      "epoch": 2.909090909090909,
+      "grad_norm": 0.3172072172164917,
+      "learning_rate": 4.448523783175315e-05,
+      "loss": 0.1803,
+      "step": 864
+    },
+    {
+      "epoch": 2.9124579124579126,
+      "grad_norm": 0.3312835395336151,
+      "learning_rate": 4.436839594908866e-05,
+      "loss": 0.1772,
+      "step": 865
+    },
+    {
+      "epoch": 2.915824915824916,
+      "grad_norm": 0.32331568002700806,
+      "learning_rate": 4.425158520643372e-05,
+      "loss": 0.1831,
+      "step": 866
+    },
+    {
+      "epoch": 2.919191919191919,
+      "grad_norm": 0.38694921135902405,
+      "learning_rate": 4.413480624969452e-05,
+      "loss": 0.1845,
+      "step": 867
+    },
+    {
+      "epoch": 2.9225589225589226,
+      "grad_norm": 0.34721943736076355,
+      "learning_rate": 4.401805972460139e-05,
+      "loss": 0.1687,
+      "step": 868
+    },
+    {
+      "epoch": 2.925925925925926,
+      "grad_norm": 0.3954653739929199,
+      "learning_rate": 4.39013462767054e-05,
+      "loss": 0.1742,
+      "step": 869
+    },
+    {
+      "epoch": 2.929292929292929,
+      "grad_norm": 0.4040646553039551,
+      "learning_rate": 4.378466655137471e-05,
+      "loss": 0.1741,
+      "step": 870
+    },
+    {
+      "epoch": 2.9326599326599325,
+      "grad_norm": 0.38577964901924133,
+      "learning_rate": 4.3668021193790974e-05,
+      "loss": 0.1679,
+      "step": 871
+    },
+    {
+      "epoch": 2.936026936026936,
+      "grad_norm": 0.42223554849624634,
+      "learning_rate": 4.3551410848945865e-05,
+      "loss": 0.1729,
+      "step": 872
+    },
+    {
+      "epoch": 2.9393939393939394,
+      "grad_norm": 0.32160666584968567,
+      "learning_rate": 4.343483616163739e-05,
+      "loss": 0.179,
+      "step": 873
+    },
+    {
+      "epoch": 2.942760942760943,
+      "grad_norm": 0.4601185917854309,
+      "learning_rate": 4.3318297776466456e-05,
+      "loss": 0.1827,
+      "step": 874
+    },
+    {
+      "epoch": 2.9461279461279464,
+      "grad_norm": 0.3302989900112152,
+      "learning_rate": 4.320179633783317e-05,
+      "loss": 0.1769,
+      "step": 875
+    },
+    {
+      "epoch": 2.9494949494949494,
+      "grad_norm": 0.36257827281951904,
+      "learning_rate": 4.3085332489933384e-05,
+      "loss": 0.1715,
+      "step": 876
+    },
+    {
+      "epoch": 2.952861952861953,
+      "grad_norm": 0.33515387773513794,
+      "learning_rate": 4.29689068767551e-05,
+      "loss": 0.1802,
+      "step": 877
+    },
+    {
+      "epoch": 2.9562289562289563,
+      "grad_norm": 0.41977766156196594,
+      "learning_rate": 4.285252014207485e-05,
+      "loss": 0.1766,
+      "step": 878
+    },
+    {
+      "epoch": 2.9595959595959593,
+      "grad_norm": 0.2881879210472107,
+      "learning_rate": 4.273617292945425e-05,
+      "loss": 0.1807,
+      "step": 879
+    },
+    {
+      "epoch": 2.962962962962963,
+      "grad_norm": 0.42628830671310425,
+      "learning_rate": 4.2619865882236313e-05,
+      "loss": 0.1803,
+      "step": 880
+    },
+    {
+      "epoch": 2.9663299663299663,
+      "grad_norm": 0.35057729482650757,
+      "learning_rate": 4.2503599643542024e-05,
+      "loss": 0.1603,
+      "step": 881
+    },
+    {
+      "epoch": 2.9696969696969697,
+      "grad_norm": 0.3465060889720917,
+      "learning_rate": 4.2387374856266686e-05,
+      "loss": 0.1803,
+      "step": 882
+    },
+    {
+      "epoch": 2.973063973063973,
+      "grad_norm": 0.33382153511047363,
+      "learning_rate": 4.227119216307637e-05,
+      "loss": 0.1637,
+      "step": 883
+    },
+    {
+      "epoch": 2.9764309764309766,
+      "grad_norm": 0.3177237808704376,
+      "learning_rate": 4.215505220640442e-05,
+      "loss": 0.17,
+      "step": 884
+    },
+    {
+      "epoch": 2.9797979797979797,
+      "grad_norm": 0.3879055678844452,
+      "learning_rate": 4.203895562844789e-05,
+      "loss": 0.1637,
+      "step": 885
+    },
+    {
+      "epoch": 2.983164983164983,
+      "grad_norm": 0.367330402135849,
+      "learning_rate": 4.1922903071163886e-05,
+      "loss": 0.1722,
+      "step": 886
+    },
+    {
+      "epoch": 2.9865319865319866,
+      "grad_norm": 0.38031652569770813,
+      "learning_rate": 4.18068951762662e-05,
+      "loss": 0.162,
+      "step": 887
+    },
+    {
+      "epoch": 2.98989898989899,
+      "grad_norm": 0.2925989031791687,
+      "learning_rate": 4.16909325852216e-05,
+      "loss": 0.1675,
+      "step": 888
+    },
+    {
+      "epoch": 2.993265993265993,
+      "grad_norm": 0.4009501338005066,
+      "learning_rate": 4.1575015939246384e-05,
+      "loss": 0.1783,
+      "step": 889
+    },
+    {
+      "epoch": 2.9966329966329965,
+      "grad_norm": 0.32625624537467957,
+      "learning_rate": 4.145914587930275e-05,
+      "loss": 0.1718,
+      "step": 890
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.4165663421154022,
+      "learning_rate": 4.134332304609533e-05,
+      "loss": 0.1755,
+      "step": 891
+    },
+    {
+      "epoch": 3.0033670033670035,
+      "grad_norm": 0.2938865125179291,
+      "learning_rate": 4.1227548080067635e-05,
+      "loss": 0.1674,
+      "step": 892
+    },
+    {
+      "epoch": 3.006734006734007,
+      "grad_norm": 0.3969976305961609,
+      "learning_rate": 4.1111821621398446e-05,
+      "loss": 0.1654,
+      "step": 893
+    },
+    {
+      "epoch": 3.01010101010101,
+      "grad_norm": 0.3079482316970825,
+      "learning_rate": 4.099614430999834e-05,
+      "loss": 0.1669,
+      "step": 894
+    },
+    {
+      "epoch": 3.0134680134680134,
+      "grad_norm": 0.3691355288028717,
+      "learning_rate": 4.088051678550617e-05,
+      "loss": 0.1634,
+      "step": 895
+    },
+    {
+      "epoch": 3.016835016835017,
+      "grad_norm": 0.30337515473365784,
+      "learning_rate": 4.0764939687285434e-05,
+      "loss": 0.1673,
+      "step": 896
+    },
+    {
+      "epoch": 3.0202020202020203,
+      "grad_norm": 0.3505813181400299,
+      "learning_rate": 4.064941365442084e-05,
+      "loss": 0.1639,
+      "step": 897
+    },
+    {
+      "epoch": 3.0235690235690234,
+      "grad_norm": 0.46245038509368896,
+      "learning_rate": 4.053393932571472e-05,
+      "loss": 0.1533,
+      "step": 898
+    },
+    {
+      "epoch": 3.026936026936027,
+      "grad_norm": 0.3629229962825775,
+      "learning_rate": 4.0418517339683474e-05,
+      "loss": 0.1758,
+      "step": 899
+    },
+    {
+      "epoch": 3.0303030303030303,
+      "grad_norm": 0.3285175859928131,
+      "learning_rate": 4.030314833455413e-05,
+      "loss": 0.1636,
+      "step": 900
+    },
+    {
+      "epoch": 3.0303030303030303,
+      "eval_loss": 0.0878610908985138,
+      "eval_runtime": 33.3117,
+      "eval_samples_per_second": 30.019,
+      "eval_steps_per_second": 1.891,
+      "step": 900
+    },
+    {
+      "epoch": 3.0336700336700337,
+      "grad_norm": 0.3374682366847992,
+      "learning_rate": 4.0187832948260705e-05,
+      "loss": 0.1574,
+      "step": 901
+    },
+    {
+      "epoch": 3.037037037037037,
+      "grad_norm": 0.337251752614975,
+      "learning_rate": 4.007257181844078e-05,
+      "loss": 0.1621,
+      "step": 902
+    },
+    {
+      "epoch": 3.04040404040404,
+      "grad_norm": 0.3177763521671295,
+      "learning_rate": 3.995736558243186e-05,
+      "loss": 0.1577,
+      "step": 903
+    },
+    {
+      "epoch": 3.0437710437710437,
+      "grad_norm": 0.32198965549468994,
+      "learning_rate": 3.9842214877267986e-05,
+      "loss": 0.1523,
+      "step": 904
+    },
+    {
+      "epoch": 3.047138047138047,
+      "grad_norm": 0.27595996856689453,
+      "learning_rate": 3.9727120339676084e-05,
+      "loss": 0.1611,
+      "step": 905
+    },
+    {
+      "epoch": 3.0505050505050506,
+      "grad_norm": 0.30028364062309265,
+      "learning_rate": 3.961208260607253e-05,
+      "loss": 0.1419,
+      "step": 906
+    },
+    {
+      "epoch": 3.053872053872054,
+      "grad_norm": 0.34587720036506653,
+      "learning_rate": 3.94971023125596e-05,
+      "loss": 0.1682,
+      "step": 907
+    },
+    {
+      "epoch": 3.057239057239057,
+      "grad_norm": 0.35993483662605286,
+      "learning_rate": 3.938218009492193e-05,
+      "loss": 0.1607,
+      "step": 908
+    },
+    {
+      "epoch": 3.0606060606060606,
+      "grad_norm": 0.390819787979126,
+      "learning_rate": 3.926731658862307e-05,
+      "loss": 0.1628,
+      "step": 909
+    },
+    {
+      "epoch": 3.063973063973064,
+      "grad_norm": 0.40493595600128174,
+      "learning_rate": 3.91525124288019e-05,
+      "loss": 0.1388,
+      "step": 910
+    },
+    {
+      "epoch": 3.0673400673400675,
+      "grad_norm": 0.4477592408657074,
+      "learning_rate": 3.903776825026911e-05,
+      "loss": 0.1616,
+      "step": 911
+    },
+    {
+      "epoch": 3.0707070707070705,
+      "grad_norm": 0.42329877614974976,
+      "learning_rate": 3.892308468750379e-05,
+      "loss": 0.1538,
+      "step": 912
+    },
+    {
+      "epoch": 3.074074074074074,
+      "grad_norm": 0.40350863337516785,
+      "learning_rate": 3.8808462374649803e-05,
+      "loss": 0.1525,
+      "step": 913
+    },
+    {
+      "epoch": 3.0774410774410774,
+      "grad_norm": 0.41857612133026123,
+      "learning_rate": 3.869390194551235e-05,
+      "loss": 0.1557,
+      "step": 914
+    },
+    {
+      "epoch": 3.080808080808081,
+      "grad_norm": 0.35566312074661255,
+      "learning_rate": 3.857940403355444e-05,
+      "loss": 0.1592,
+      "step": 915
+    },
+    {
+      "epoch": 3.0841750841750843,
+      "grad_norm": 0.4137890636920929,
+      "learning_rate": 3.846496927189339e-05,
+      "loss": 0.15,
+      "step": 916
+    },
+    {
+      "epoch": 3.0875420875420874,
+      "grad_norm": 0.3321918845176697,
+      "learning_rate": 3.835059829329735e-05,
+      "loss": 0.1572,
+      "step": 917
+    },
+    {
+      "epoch": 3.090909090909091,
+      "grad_norm": 0.31146201491355896,
+      "learning_rate": 3.823629173018173e-05,
+      "loss": 0.1598,
+      "step": 918
+    },
+    {
+      "epoch": 3.0942760942760943,
+      "grad_norm": 0.3829994797706604,
+      "learning_rate": 3.812205021460582e-05,
+      "loss": 0.1709,
+      "step": 919
+    },
+    {
+      "epoch": 3.0976430976430978,
+      "grad_norm": 0.3082868158817291,
+      "learning_rate": 3.800787437826918e-05,
+      "loss": 0.1711,
+      "step": 920
+    },
+    {
+      "epoch": 3.101010101010101,
+      "grad_norm": 0.3517824709415436,
+      "learning_rate": 3.789376485250821e-05,
+      "loss": 0.1628,
+      "step": 921
+    },
+    {
+      "epoch": 3.1043771043771042,
+      "grad_norm": 0.4294639825820923,
+      "learning_rate": 3.777972226829264e-05,
+      "loss": 0.1722,
+      "step": 922
+    },
+    {
+      "epoch": 3.1077441077441077,
+      "grad_norm": 0.4236588478088379,
+      "learning_rate": 3.7665747256222075e-05,
+      "loss": 0.1624,
+      "step": 923
+    },
+    {
+      "epoch": 3.111111111111111,
+      "grad_norm": 0.4085109829902649,
+      "learning_rate": 3.755184044652244e-05,
+      "loss": 0.1568,
+      "step": 924
+    },
+    {
+      "epoch": 3.1144781144781146,
+      "grad_norm": 0.48974359035491943,
+      "learning_rate": 3.7438002469042565e-05,
+      "loss": 0.1492,
+      "step": 925
+    },
+    {
+      "epoch": 3.1178451178451176,
+      "grad_norm": 0.4154088497161865,
+      "learning_rate": 3.732423395325063e-05,
+      "loss": 0.1571,
+      "step": 926
+    },
+    {
+      "epoch": 3.121212121212121,
+      "grad_norm": 0.3798883259296417,
+      "learning_rate": 3.721053552823078e-05,
+      "loss": 0.1578,
+      "step": 927
+    },
+    {
+      "epoch": 3.1245791245791246,
+      "grad_norm": 0.40717121958732605,
+      "learning_rate": 3.7096907822679564e-05,
+      "loss": 0.1617,
+      "step": 928
+    },
+    {
+      "epoch": 3.127946127946128,
+      "grad_norm": 0.3714147210121155,
+      "learning_rate": 3.698335146490246e-05,
+      "loss": 0.1618,
+      "step": 929
+    },
+    {
+      "epoch": 3.1313131313131315,
+      "grad_norm": 0.47017040848731995,
+      "learning_rate": 3.68698670828105e-05,
+      "loss": 0.1665,
+      "step": 930
+    },
+    {
+      "epoch": 3.1346801346801345,
+      "grad_norm": 0.3854427635669708,
+      "learning_rate": 3.675645530391665e-05,
+      "loss": 0.1585,
+      "step": 931
+    },
+    {
+      "epoch": 3.138047138047138,
+      "grad_norm": 0.36705338954925537,
+      "learning_rate": 3.664311675533247e-05,
+      "loss": 0.1648,
+      "step": 932
+    },
+    {
+      "epoch": 3.1414141414141414,
+      "grad_norm": 0.3947654366493225,
+      "learning_rate": 3.6529852063764545e-05,
+      "loss": 0.1763,
+      "step": 933
+    },
+    {
+      "epoch": 3.144781144781145,
+      "grad_norm": 0.4430708885192871,
+      "learning_rate": 3.641666185551111e-05,
+      "loss": 0.173,
+      "step": 934
+    },
+    {
+      "epoch": 3.148148148148148,
+      "grad_norm": 0.3862123191356659,
+      "learning_rate": 3.630354675645853e-05,
+      "loss": 0.1509,
+      "step": 935
+    },
+    {
+      "epoch": 3.1515151515151514,
+      "grad_norm": 0.4604793190956116,
+      "learning_rate": 3.619050739207782e-05,
+      "loss": 0.1625,
+      "step": 936
+    },
+    {
+      "epoch": 3.154882154882155,
+      "grad_norm": 0.37315207719802856,
+      "learning_rate": 3.607754438742129e-05,
+      "loss": 0.1561,
+      "step": 937
+    },
+    {
+      "epoch": 3.1582491582491583,
+      "grad_norm": 0.4637494385242462,
+      "learning_rate": 3.596465836711896e-05,
+      "loss": 0.1523,
+      "step": 938
+    },
+    {
+      "epoch": 3.1616161616161618,
+      "grad_norm": 0.45701348781585693,
+      "learning_rate": 3.585184995537518e-05,
+      "loss": 0.1577,
+      "step": 939
+    },
+    {
+      "epoch": 3.164983164983165,
+      "grad_norm": 0.41691821813583374,
+      "learning_rate": 3.5739119775965165e-05,
+      "loss": 0.1571,
+      "step": 940
+    },
+    {
+      "epoch": 3.1683501683501682,
+      "grad_norm": 0.3542643189430237,
+      "learning_rate": 3.562646845223153e-05,
+      "loss": 0.1487,
+      "step": 941
+    },
+    {
+      "epoch": 3.1717171717171717,
+      "grad_norm": 0.42012447118759155,
+      "learning_rate": 3.551389660708088e-05,
+      "loss": 0.157,
+      "step": 942
+    },
+    {
+      "epoch": 3.175084175084175,
+      "grad_norm": 0.3715340793132782,
+      "learning_rate": 3.540140486298035e-05,
+      "loss": 0.1562,
+      "step": 943
+    },
+    {
+      "epoch": 3.1784511784511786,
+      "grad_norm": 0.4073311388492584,
+      "learning_rate": 3.528899384195409e-05,
+      "loss": 0.1602,
+      "step": 944
+    },
+    {
+      "epoch": 3.1818181818181817,
+      "grad_norm": 0.3359941244125366,
+      "learning_rate": 3.5176664165579986e-05,
+      "loss": 0.1635,
+      "step": 945
+    },
+    {
+      "epoch": 3.185185185185185,
+      "grad_norm": 0.3938975930213928,
+      "learning_rate": 3.506441645498605e-05,
+      "loss": 0.159,
+      "step": 946
+    },
+    {
+      "epoch": 3.1885521885521886,
+      "grad_norm": 0.42345231771469116,
+      "learning_rate": 3.495225133084712e-05,
+      "loss": 0.1541,
+      "step": 947
+    },
+    {
+      "epoch": 3.191919191919192,
+      "grad_norm": 0.36092641949653625,
+      "learning_rate": 3.4840169413381305e-05,
+      "loss": 0.158,
+      "step": 948
+    },
+    {
+      "epoch": 3.1952861952861955,
+      "grad_norm": 0.3029501438140869,
+      "learning_rate": 3.4728171322346694e-05,
+      "loss": 0.1599,
+      "step": 949
+    },
+    {
+      "epoch": 3.1986531986531985,
+      "grad_norm": 0.3518286645412445,
+      "learning_rate": 3.46162576770378e-05,
+      "loss": 0.1656,
+      "step": 950
+    },
+    {
+      "epoch": 3.202020202020202,
+      "grad_norm": 0.3297809362411499,
+      "learning_rate": 3.450442909628224e-05,
+      "loss": 0.1712,
+      "step": 951
+    },
+    {
+      "epoch": 3.2053872053872055,
+      "grad_norm": 0.3591551184654236,
+      "learning_rate": 3.439268619843724e-05,
+      "loss": 0.1665,
+      "step": 952
+    },
+    {
+      "epoch": 3.208754208754209,
+      "grad_norm": 0.31120550632476807,
+      "learning_rate": 3.428102960138625e-05,
+      "loss": 0.1604,
+      "step": 953
+    },
+    {
+      "epoch": 3.212121212121212,
+      "grad_norm": 0.3053344488143921,
+      "learning_rate": 3.4169459922535485e-05,
+      "loss": 0.1567,
+      "step": 954
+    },
+    {
+      "epoch": 3.2154882154882154,
+      "grad_norm": 0.3286115229129791,
+      "learning_rate": 3.405797777881059e-05,
+      "loss": 0.1516,
+      "step": 955
+    },
+    {
+      "epoch": 3.218855218855219,
+      "grad_norm": 0.313930869102478,
+      "learning_rate": 3.3946583786653184e-05,
+      "loss": 0.1624,
+      "step": 956
+    },
+    {
+      "epoch": 3.2222222222222223,
+      "grad_norm": 0.3518214821815491,
+      "learning_rate": 3.38352785620174e-05,
+      "loss": 0.1501,
+      "step": 957
+    },
+    {
+      "epoch": 3.225589225589226,
+      "grad_norm": 0.4037775993347168,
+      "learning_rate": 3.37240627203666e-05,
+      "loss": 0.1508,
+      "step": 958
+    },
+    {
+      "epoch": 3.228956228956229,
+      "grad_norm": 0.3002485930919647,
+      "learning_rate": 3.3612936876669834e-05,
+      "loss": 0.1667,
+      "step": 959
+    },
+    {
+      "epoch": 3.2323232323232323,
+      "grad_norm": 0.35778674483299255,
+      "learning_rate": 3.350190164539855e-05,
+      "loss": 0.1659,
+      "step": 960
+    },
+    {
+      "epoch": 3.2356902356902357,
+      "grad_norm": 0.3302362263202667,
+      "learning_rate": 3.3390957640523145e-05,
+      "loss": 0.161,
+      "step": 961
+    },
+    {
+      "epoch": 3.239057239057239,
+      "grad_norm": 0.3861044943332672,
+      "learning_rate": 3.328010547550959e-05,
+      "loss": 0.1678,
+      "step": 962
+    },
+    {
+      "epoch": 3.242424242424242,
+      "grad_norm": 0.3668946921825409,
+      "learning_rate": 3.316934576331598e-05,
+      "loss": 0.1487,
+      "step": 963
+    },
+    {
+      "epoch": 3.2457912457912457,
+      "grad_norm": 0.3028615117073059,
+      "learning_rate": 3.3058679116389244e-05,
+      "loss": 0.1576,
+      "step": 964
+    },
+    {
+      "epoch": 3.249158249158249,
+      "grad_norm": 0.4418306350708008,
+      "learning_rate": 3.29481061466617e-05,
+      "loss": 0.1471,
+      "step": 965
+    },
+    {
+      "epoch": 3.2525252525252526,
+      "grad_norm": 0.3218114674091339,
+      "learning_rate": 3.283762746554766e-05,
+      "loss": 0.1595,
+      "step": 966
+    },
+    {
+      "epoch": 3.255892255892256,
+      "grad_norm": 0.3931596875190735,
+      "learning_rate": 3.2727243683940045e-05,
+      "loss": 0.1651,
+      "step": 967
+    },
+    {
+      "epoch": 3.259259259259259,
+      "grad_norm": 0.38177919387817383,
+      "learning_rate": 3.2616955412207086e-05,
+      "loss": 0.1512,
+      "step": 968
+    },
+    {
+      "epoch": 3.2626262626262625,
+      "grad_norm": 0.3416683077812195,
+      "learning_rate": 3.250676326018882e-05,
+      "loss": 0.1479,
+      "step": 969
+    },
+    {
+      "epoch": 3.265993265993266,
+      "grad_norm": 0.478951632976532,
+      "learning_rate": 3.239666783719385e-05,
+      "loss": 0.1668,
+      "step": 970
+    },
+    {
+      "epoch": 3.2693602693602695,
+      "grad_norm": 0.3602767586708069,
+      "learning_rate": 3.2286669751995904e-05,
+      "loss": 0.1482,
+      "step": 971
+    },
+    {
+      "epoch": 3.2727272727272725,
+      "grad_norm": 0.4089204668998718,
+      "learning_rate": 3.217676961283044e-05,
+      "loss": 0.1549,
+      "step": 972
+    },
+    {
+      "epoch": 3.276094276094276,
+      "grad_norm": 0.40424418449401855,
+      "learning_rate": 3.2066968027391374e-05,
+      "loss": 0.1688,
+      "step": 973
+    },
+    {
+      "epoch": 3.2794612794612794,
+      "grad_norm": 0.31489500403404236,
+      "learning_rate": 3.195726560282763e-05,
+      "loss": 0.1583,
+      "step": 974
+    },
+    {
+      "epoch": 3.282828282828283,
+      "grad_norm": 0.4403363764286041,
+      "learning_rate": 3.184766294573983e-05,
+      "loss": 0.1546,
+      "step": 975
+    },
+    {
+      "epoch": 3.2861952861952863,
+      "grad_norm": 0.29121315479278564,
+      "learning_rate": 3.1738160662176956e-05,
+      "loss": 0.1608,
+      "step": 976
+    },
+    {
+      "epoch": 3.28956228956229,
+      "grad_norm": 0.4291994571685791,
+      "learning_rate": 3.162875935763294e-05,
+      "loss": 0.1569,
+      "step": 977
+    },
+    {
+      "epoch": 3.292929292929293,
+      "grad_norm": 0.3442822992801666,
+      "learning_rate": 3.151945963704334e-05,
+      "loss": 0.1695,
+      "step": 978
+    },
+    {
+      "epoch": 3.2962962962962963,
+      "grad_norm": 0.5204855799674988,
+      "learning_rate": 3.1410262104782085e-05,
+      "loss": 0.1514,
+      "step": 979
+    },
+    {
+      "epoch": 3.2996632996632997,
+      "grad_norm": 0.6250646710395813,
+      "learning_rate": 3.130116736465795e-05,
+      "loss": 0.1695,
+      "step": 980
+    },
+    {
+      "epoch": 3.303030303030303,
+      "grad_norm": 0.3862772285938263,
+      "learning_rate": 3.119217601991139e-05,
+      "loss": 0.1487,
+      "step": 981
+    },
+    {
+      "epoch": 3.3063973063973062,
+      "grad_norm": 0.5631003975868225,
+      "learning_rate": 3.1083288673211095e-05,
+      "loss": 0.1699,
+      "step": 982
+    },
+    {
+      "epoch": 3.3097643097643097,
+      "grad_norm": 0.39467087388038635,
+      "learning_rate": 3.0974505926650724e-05,
+      "loss": 0.1582,
+      "step": 983
+    },
+    {
+      "epoch": 3.313131313131313,
+      "grad_norm": 0.5696914196014404,
+      "learning_rate": 3.086582838174551e-05,
+      "loss": 0.1658,
+      "step": 984
+    },
+    {
+      "epoch": 3.3164983164983166,
+      "grad_norm": 0.37162837386131287,
+      "learning_rate": 3.0757256639429025e-05,
+      "loss": 0.1472,
+      "step": 985
+    },
+    {
+      "epoch": 3.31986531986532,
+      "grad_norm": 0.3216390609741211,
+      "learning_rate": 3.064879130004978e-05,
+      "loss": 0.1607,
+      "step": 986
+    },
+    {
+      "epoch": 3.323232323232323,
+      "grad_norm": 0.6373658180236816,
+      "learning_rate": 3.0540432963367905e-05,
+      "loss": 0.1627,
+      "step": 987
+    },
+    {
+      "epoch": 3.3265993265993266,
+      "grad_norm": 0.2934598922729492,
+      "learning_rate": 3.04321822285519e-05,
+      "loss": 0.1593,
+      "step": 988
+    },
+    {
+      "epoch": 3.32996632996633,
+      "grad_norm": 0.5632359981536865,
+      "learning_rate": 3.0324039694175233e-05,
+      "loss": 0.1612,
+      "step": 989
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.5589274764060974,
+      "learning_rate": 3.0216005958213144e-05,
+      "loss": 0.1473,
+      "step": 990
+    },
+    {
+      "epoch": 3.3367003367003365,
+      "grad_norm": 0.4226700961589813,
+      "learning_rate": 3.0108081618039167e-05,
+      "loss": 0.1575,
+      "step": 991
+    },
+    {
+      "epoch": 3.34006734006734,
+      "grad_norm": 0.4167967438697815,
+      "learning_rate": 3.0000267270422076e-05,
+      "loss": 0.1575,
+      "step": 992
+    },
+    {
+      "epoch": 3.3434343434343434,
+      "grad_norm": 0.3462510406970978,
+      "learning_rate": 2.9892563511522304e-05,
+      "loss": 0.1445,
+      "step": 993
+    },
+    {
+      "epoch": 3.346801346801347,
+      "grad_norm": 0.40887391567230225,
+      "learning_rate": 2.9784970936888857e-05,
+      "loss": 0.1665,
+      "step": 994
+    },
+    {
+      "epoch": 3.3501683501683504,
+      "grad_norm": 0.44167807698249817,
+      "learning_rate": 2.9677490141455916e-05,
+      "loss": 0.1659,
+      "step": 995
+    },
+    {
+      "epoch": 3.3535353535353534,
+      "grad_norm": 0.423782080411911,
+      "learning_rate": 2.9570121719539602e-05,
+      "loss": 0.1613,
+      "step": 996
+    },
+    {
+      "epoch": 3.356902356902357,
+      "grad_norm": 0.3622366487979889,
+      "learning_rate": 2.946286626483463e-05,
+      "loss": 0.1549,
+      "step": 997
+    },
+    {
+      "epoch": 3.3602693602693603,
+      "grad_norm": 0.3742324113845825,
+      "learning_rate": 2.935572437041111e-05,
+      "loss": 0.166,
+      "step": 998
+    },
+    {
+      "epoch": 3.3636363636363638,
+      "grad_norm": 0.37756288051605225,
+      "learning_rate": 2.924869662871117e-05,
+      "loss": 0.1701,
+      "step": 999
+    },
+    {
+      "epoch": 3.3670033670033668,
+      "grad_norm": 0.3715890645980835,
+      "learning_rate": 2.9141783631545773e-05,
+      "loss": 0.1576,
+      "step": 1000
+    },
+    {
+      "epoch": 3.3670033670033668,
+      "eval_loss": 0.08614125102758408,
+      "eval_runtime": 32.8393,
+      "eval_samples_per_second": 30.451,
+      "eval_steps_per_second": 1.918,
+      "step": 1000
+    },
+    {
+      "epoch": 3.3703703703703702,
+      "grad_norm": 0.34478187561035156,
+      "learning_rate": 2.9034985970091355e-05,
+      "loss": 0.1575,
+      "step": 1001
+    },
+    {
+      "epoch": 3.3737373737373737,
+      "grad_norm": 0.3495587706565857,
+      "learning_rate": 2.8928304234886644e-05,
+      "loss": 0.1606,
+      "step": 1002
+    },
+    {
+      "epoch": 3.377104377104377,
+      "grad_norm": 0.3603619337081909,
+      "learning_rate": 2.8821739015829337e-05,
+      "loss": 0.1589,
+      "step": 1003
+    },
+    {
+      "epoch": 3.3804713804713806,
+      "grad_norm": 0.3719029128551483,
+      "learning_rate": 2.8715290902172863e-05,
+      "loss": 0.159,
+      "step": 1004
+    },
+    {
+      "epoch": 3.3838383838383836,
+      "grad_norm": 0.36092978715896606,
+      "learning_rate": 2.8608960482523056e-05,
+      "loss": 0.1618,
+      "step": 1005
+    },
+    {
+      "epoch": 3.387205387205387,
+      "grad_norm": 0.3863452672958374,
+      "learning_rate": 2.8502748344835063e-05,
+      "loss": 0.1644,
+      "step": 1006
+    },
+    {
+      "epoch": 3.3905723905723906,
+      "grad_norm": 0.3570246398448944,
+      "learning_rate": 2.8396655076409923e-05,
+      "loss": 0.1556,
+      "step": 1007
+    },
+    {
+      "epoch": 3.393939393939394,
+      "grad_norm": 0.4031378924846649,
+      "learning_rate": 2.8290681263891412e-05,
+      "loss": 0.1538,
+      "step": 1008
+    },
+    {
+      "epoch": 3.3973063973063975,
+      "grad_norm": 0.35926946997642517,
+      "learning_rate": 2.818482749326272e-05,
+      "loss": 0.1669,
+      "step": 1009
+    },
+    {
+      "epoch": 3.4006734006734005,
+      "grad_norm": 0.46241873502731323,
+      "learning_rate": 2.8079094349843334e-05,
+      "loss": 0.1601,
+      "step": 1010
+    },
+    {
+      "epoch": 3.404040404040404,
+      "grad_norm": 0.32303571701049805,
+      "learning_rate": 2.797348241828569e-05,
+      "loss": 0.1596,
+      "step": 1011
+    },
+    {
+      "epoch": 3.4074074074074074,
+      "grad_norm": 0.3822338879108429,
+      "learning_rate": 2.786799228257203e-05,
+      "loss": 0.1578,
+      "step": 1012
+    },
+    {
+      "epoch": 3.410774410774411,
+      "grad_norm": 0.3558747172355652,
+      "learning_rate": 2.7762624526011038e-05,
+      "loss": 0.1588,
+      "step": 1013
+    },
+    {
+      "epoch": 3.4141414141414144,
+      "grad_norm": 0.33147427439689636,
+      "learning_rate": 2.7657379731234768e-05,
+      "loss": 0.1606,
+      "step": 1014
+    },
+    {
+      "epoch": 3.4175084175084174,
+      "grad_norm": 0.4226522743701935,
+      "learning_rate": 2.7552258480195347e-05,
+      "loss": 0.1589,
+      "step": 1015
+    },
+    {
+      "epoch": 3.420875420875421,
+      "grad_norm": 0.33748775720596313,
+      "learning_rate": 2.744726135416177e-05,
+      "loss": 0.1559,
+      "step": 1016
+    },
+    {
+      "epoch": 3.4242424242424243,
+      "grad_norm": 0.3910854160785675,
+      "learning_rate": 2.7342388933716668e-05,
+      "loss": 0.1573,
+      "step": 1017
+    },
+    {
+      "epoch": 3.4276094276094278,
+      "grad_norm": 0.32681888341903687,
+      "learning_rate": 2.7237641798753083e-05,
+      "loss": 0.1524,
+      "step": 1018
+    },
+    {
+      "epoch": 3.430976430976431,
+      "grad_norm": 0.4060154855251312,
+      "learning_rate": 2.713302052847132e-05,
+      "loss": 0.155,
+      "step": 1019
+    },
+    {
+      "epoch": 3.4343434343434343,
+      "grad_norm": 0.271331787109375,
+      "learning_rate": 2.702852570137576e-05,
+      "loss": 0.1557,
+      "step": 1020
+    },
+    {
+      "epoch": 3.4377104377104377,
+      "grad_norm": 0.3912033438682556,
+      "learning_rate": 2.6924157895271563e-05,
+      "loss": 0.1625,
+      "step": 1021
+    },
+    {
+      "epoch": 3.441077441077441,
+      "grad_norm": 0.3229612708091736,
+      "learning_rate": 2.6819917687261493e-05,
+      "loss": 0.1614,
+      "step": 1022
+    },
+    {
+      "epoch": 3.4444444444444446,
+      "grad_norm": 0.36763250827789307,
+      "learning_rate": 2.671580565374282e-05,
+      "loss": 0.1677,
+      "step": 1023
+    },
+    {
+      "epoch": 3.4478114478114477,
+      "grad_norm": 0.3430008888244629,
+      "learning_rate": 2.6611822370404037e-05,
+      "loss": 0.1617,
+      "step": 1024
+    },
+    {
+      "epoch": 3.451178451178451,
+      "grad_norm": 0.33642029762268066,
+      "learning_rate": 2.650796841222176e-05,
+      "loss": 0.1556,
+      "step": 1025
+    },
+    {
+      "epoch": 3.4545454545454546,
+      "grad_norm": 0.4070379137992859,
+      "learning_rate": 2.6404244353457424e-05,
+      "loss": 0.1642,
+      "step": 1026
+    },
+    {
+      "epoch": 3.457912457912458,
+      "grad_norm": 0.33227312564849854,
+      "learning_rate": 2.6300650767654234e-05,
+      "loss": 0.1437,
+      "step": 1027
+    },
+    {
+      "epoch": 3.461279461279461,
+      "grad_norm": 0.38948434591293335,
+      "learning_rate": 2.619718822763394e-05,
+      "loss": 0.1569,
+      "step": 1028
+    },
+    {
+      "epoch": 3.4646464646464645,
+      "grad_norm": 0.4101947844028473,
+      "learning_rate": 2.6093857305493664e-05,
+      "loss": 0.158,
+      "step": 1029
+    },
+    {
+      "epoch": 3.468013468013468,
+      "grad_norm": 0.44636788964271545,
+      "learning_rate": 2.5990658572602773e-05,
+      "loss": 0.1587,
+      "step": 1030
+    },
+    {
+      "epoch": 3.4713804713804715,
+      "grad_norm": 0.32242777943611145,
+      "learning_rate": 2.5887592599599618e-05,
+      "loss": 0.1661,
+      "step": 1031
+    },
+    {
+      "epoch": 3.474747474747475,
+      "grad_norm": 0.4262060821056366,
+      "learning_rate": 2.5784659956388535e-05,
+      "loss": 0.1476,
+      "step": 1032
+    },
+    {
+      "epoch": 3.478114478114478,
+      "grad_norm": 0.32246485352516174,
+      "learning_rate": 2.5681861212136578e-05,
+      "loss": 0.1547,
+      "step": 1033
+    },
+    {
+      "epoch": 3.4814814814814814,
+      "grad_norm": 0.3994286060333252,
+      "learning_rate": 2.55791969352704e-05,
+      "loss": 0.157,
+      "step": 1034
+    },
+    {
+      "epoch": 3.484848484848485,
+      "grad_norm": 0.33445990085601807,
+      "learning_rate": 2.547666769347312e-05,
+      "loss": 0.1557,
+      "step": 1035
+    },
+    {
+      "epoch": 3.4882154882154883,
+      "grad_norm": 0.3687197268009186,
+      "learning_rate": 2.537427405368119e-05,
+      "loss": 0.158,
+      "step": 1036
+    },
+    {
+      "epoch": 3.4915824915824913,
+      "grad_norm": 0.3803139925003052,
+      "learning_rate": 2.5272016582081236e-05,
+      "loss": 0.1501,
+      "step": 1037
+    },
+    {
+      "epoch": 3.494949494949495,
+      "grad_norm": 0.3825187087059021,
+      "learning_rate": 2.516989584410696e-05,
+      "loss": 0.154,
+      "step": 1038
+    },
+    {
+      "epoch": 3.4983164983164983,
+      "grad_norm": 0.3470314145088196,
+      "learning_rate": 2.506791240443595e-05,
+      "loss": 0.1538,
+      "step": 1039
+    },
+    {
+      "epoch": 3.5016835016835017,
+      "grad_norm": 0.33760249614715576,
+      "learning_rate": 2.4966066826986644e-05,
+      "loss": 0.1584,
+      "step": 1040
+    },
+    {
+      "epoch": 3.505050505050505,
+      "grad_norm": 0.3067149817943573,
+      "learning_rate": 2.486435967491516e-05,
+      "loss": 0.1458,
+      "step": 1041
+    },
+    {
+      "epoch": 3.5084175084175087,
+      "grad_norm": 0.3700965642929077,
+      "learning_rate": 2.476279151061221e-05,
+      "loss": 0.1578,
+      "step": 1042
+    },
+    {
+      "epoch": 3.5117845117845117,
+      "grad_norm": 0.30760011076927185,
+      "learning_rate": 2.4661362895699903e-05,
+      "loss": 0.1584,
+      "step": 1043
+    },
+    {
+      "epoch": 3.515151515151515,
+      "grad_norm": 0.3704482316970825,
+      "learning_rate": 2.456007439102878e-05,
+      "loss": 0.1498,
+      "step": 1044
+    },
+    {
+      "epoch": 3.5185185185185186,
+      "grad_norm": 0.3368256688117981,
+      "learning_rate": 2.4458926556674615e-05,
+      "loss": 0.1587,
+      "step": 1045
+    },
+    {
+      "epoch": 3.5218855218855216,
+      "grad_norm": 0.33141207695007324,
+      "learning_rate": 2.4357919951935343e-05,
+      "loss": 0.1626,
+      "step": 1046
+    },
+    {
+      "epoch": 3.525252525252525,
+      "grad_norm": 0.4236656725406647,
+      "learning_rate": 2.4257055135327976e-05,
+      "loss": 0.1594,
+      "step": 1047
+    },
+    {
+      "epoch": 3.5286195286195285,
+      "grad_norm": 0.4164685308933258,
+      "learning_rate": 2.4156332664585495e-05,
+      "loss": 0.1531,
+      "step": 1048
+    },
+    {
+      "epoch": 3.531986531986532,
+      "grad_norm": 0.379982590675354,
+      "learning_rate": 2.4055753096653794e-05,
+      "loss": 0.1511,
+      "step": 1049
+    },
+    {
+      "epoch": 3.5353535353535355,
+      "grad_norm": 0.41391053795814514,
+      "learning_rate": 2.395531698768857e-05,
+      "loss": 0.15,
+      "step": 1050
+    },
+    {
+      "epoch": 3.538720538720539,
+      "grad_norm": 0.40213122963905334,
+      "learning_rate": 2.3855024893052285e-05,
+      "loss": 0.1444,
+      "step": 1051
+    },
+    {
+      "epoch": 3.542087542087542,
+      "grad_norm": 0.38797199726104736,
+      "learning_rate": 2.375487736731102e-05,
+      "loss": 0.1476,
+      "step": 1052
+    },
+    {
+      "epoch": 3.5454545454545454,
+      "grad_norm": 0.38671404123306274,
+      "learning_rate": 2.3654874964231518e-05,
+      "loss": 0.1594,
+      "step": 1053
+    },
+    {
+      "epoch": 3.548821548821549,
+      "grad_norm": 0.40808653831481934,
+      "learning_rate": 2.355501823677803e-05,
+      "loss": 0.1529,
+      "step": 1054
+    },
+    {
+      "epoch": 3.5521885521885523,
+      "grad_norm": 0.44210532307624817,
+      "learning_rate": 2.345530773710934e-05,
+      "loss": 0.1681,
+      "step": 1055
+    },
+    {
+      "epoch": 3.5555555555555554,
+      "grad_norm": 0.3211992681026459,
+      "learning_rate": 2.3355744016575585e-05,
+      "loss": 0.1552,
+      "step": 1056
+    },
+    {
+      "epoch": 3.558922558922559,
+      "grad_norm": 0.4301251769065857,
+      "learning_rate": 2.3256327625715347e-05,
+      "loss": 0.1549,
+      "step": 1057
+    },
+    {
+      "epoch": 3.5622895622895623,
+      "grad_norm": 0.3025391697883606,
+      "learning_rate": 2.3157059114252534e-05,
+      "loss": 0.1501,
+      "step": 1058
+    },
+    {
+      "epoch": 3.5656565656565657,
+      "grad_norm": 0.34864532947540283,
+      "learning_rate": 2.3057939031093344e-05,
+      "loss": 0.154,
+      "step": 1059
+    },
+    {
+      "epoch": 3.569023569023569,
+      "grad_norm": 0.3500463366508484,
+      "learning_rate": 2.295896792432326e-05,
+      "loss": 0.1564,
+      "step": 1060
+    },
+    {
+      "epoch": 3.5723905723905722,
+      "grad_norm": 0.3177832067012787,
+      "learning_rate": 2.2860146341203937e-05,
+      "loss": 0.1471,
+      "step": 1061
+    },
+    {
+      "epoch": 3.5757575757575757,
+      "grad_norm": 0.3671930134296417,
+      "learning_rate": 2.2761474828170338e-05,
+      "loss": 0.1634,
+      "step": 1062
+    },
+    {
+      "epoch": 3.579124579124579,
+      "grad_norm": 0.36658650636672974,
+      "learning_rate": 2.2662953930827546e-05,
+      "loss": 0.1585,
+      "step": 1063
+    },
+    {
+      "epoch": 3.5824915824915826,
+      "grad_norm": 0.40321019291877747,
+      "learning_rate": 2.2564584193947797e-05,
+      "loss": 0.154,
+      "step": 1064
+    },
+    {
+      "epoch": 3.5858585858585856,
+      "grad_norm": 0.3584071099758148,
+      "learning_rate": 2.246636616146753e-05,
+      "loss": 0.153,
+      "step": 1065
+    },
+    {
+      "epoch": 3.589225589225589,
+      "grad_norm": 0.4597926139831543,
+      "learning_rate": 2.2368300376484302e-05,
+      "loss": 0.1477,
+      "step": 1066
+    },
+    {
+      "epoch": 3.5925925925925926,
+      "grad_norm": 0.34057462215423584,
+      "learning_rate": 2.227038738125385e-05,
+      "loss": 0.152,
+      "step": 1067
+    },
+    {
+      "epoch": 3.595959595959596,
+      "grad_norm": 0.4538710117340088,
+      "learning_rate": 2.2172627717187034e-05,
+      "loss": 0.1692,
+      "step": 1068
+    },
+    {
+      "epoch": 3.5993265993265995,
+      "grad_norm": 0.45110630989074707,
+      "learning_rate": 2.207502192484685e-05,
+      "loss": 0.1612,
+      "step": 1069
+    },
+    {
+      "epoch": 3.602693602693603,
+      "grad_norm": 0.4063069224357605,
+      "learning_rate": 2.1977570543945486e-05,
+      "loss": 0.1532,
+      "step": 1070
+    },
+    {
+      "epoch": 3.606060606060606,
+      "grad_norm": 0.32669079303741455,
+      "learning_rate": 2.188027411334131e-05,
+      "loss": 0.156,
+      "step": 1071
+    },
+    {
+      "epoch": 3.6094276094276094,
+      "grad_norm": 0.42182454466819763,
+      "learning_rate": 2.1783133171035886e-05,
+      "loss": 0.1615,
+      "step": 1072
+    },
+    {
+      "epoch": 3.612794612794613,
+      "grad_norm": 0.36189764738082886,
+      "learning_rate": 2.1686148254171013e-05,
+      "loss": 0.1623,
+      "step": 1073
+    },
+    {
+      "epoch": 3.616161616161616,
+      "grad_norm": 0.391677588224411,
+      "learning_rate": 2.1589319899025705e-05,
+      "loss": 0.1584,
+      "step": 1074
+    },
+    {
+      "epoch": 3.6195286195286194,
+      "grad_norm": 0.32365673780441284,
+      "learning_rate": 2.14926486410133e-05,
+      "loss": 0.1572,
+      "step": 1075
+    },
+    {
+      "epoch": 3.622895622895623,
+      "grad_norm": 0.4265686273574829,
+      "learning_rate": 2.1396135014678507e-05,
+      "loss": 0.164,
+      "step": 1076
+    },
+    {
+      "epoch": 3.6262626262626263,
+      "grad_norm": 0.40638405084609985,
+      "learning_rate": 2.1299779553694323e-05,
+      "loss": 0.1634,
+      "step": 1077
+    },
+    {
+      "epoch": 3.6296296296296298,
+      "grad_norm": 0.4131862223148346,
+      "learning_rate": 2.1203582790859222e-05,
+      "loss": 0.1608,
+      "step": 1078
+    },
+    {
+      "epoch": 3.6329966329966332,
+      "grad_norm": 0.36556342244148254,
+      "learning_rate": 2.1107545258094135e-05,
+      "loss": 0.1606,
+      "step": 1079
+    },
+    {
+      "epoch": 3.6363636363636362,
+      "grad_norm": 0.44753018021583557,
+      "learning_rate": 2.101166748643955e-05,
+      "loss": 0.1562,
+      "step": 1080
+    },
+    {
+      "epoch": 3.6397306397306397,
+      "grad_norm": 0.2866370379924774,
+      "learning_rate": 2.0915950006052553e-05,
+      "loss": 0.1569,
+      "step": 1081
+    },
+    {
+      "epoch": 3.643097643097643,
+      "grad_norm": 0.3483537435531616,
+      "learning_rate": 2.0820393346203852e-05,
+      "loss": 0.1494,
+      "step": 1082
+    },
+    {
+      "epoch": 3.6464646464646466,
+      "grad_norm": 0.35355475544929504,
+      "learning_rate": 2.0724998035274945e-05,
+      "loss": 0.1478,
+      "step": 1083
+    },
+    {
+      "epoch": 3.6498316498316496,
+      "grad_norm": 0.3692944049835205,
+      "learning_rate": 2.0629764600755135e-05,
+      "loss": 0.1469,
+      "step": 1084
+    },
+    {
+      "epoch": 3.653198653198653,
+      "grad_norm": 0.3002285361289978,
+      "learning_rate": 2.053469356923865e-05,
+      "loss": 0.1543,
+      "step": 1085
+    },
+    {
+      "epoch": 3.6565656565656566,
+      "grad_norm": 0.3803352117538452,
+      "learning_rate": 2.0439785466421653e-05,
+      "loss": 0.1662,
+      "step": 1086
+    },
+    {
+      "epoch": 3.65993265993266,
+      "grad_norm": 0.4002717137336731,
+      "learning_rate": 2.034504081709943e-05,
+      "loss": 0.1421,
+      "step": 1087
+    },
+    {
+      "epoch": 3.6632996632996635,
+      "grad_norm": 0.3593309223651886,
+      "learning_rate": 2.025046014516346e-05,
+      "loss": 0.1538,
+      "step": 1088
+    },
+    {
+      "epoch": 3.6666666666666665,
+      "grad_norm": 0.34992143511772156,
+      "learning_rate": 2.0156043973598476e-05,
+      "loss": 0.1519,
+      "step": 1089
+    },
+    {
+      "epoch": 3.67003367003367,
+      "grad_norm": 0.29695653915405273,
+      "learning_rate": 2.006179282447963e-05,
+      "loss": 0.1375,
+      "step": 1090
+    },
+    {
+      "epoch": 3.6734006734006734,
+      "grad_norm": 0.40330347418785095,
+      "learning_rate": 1.996770721896957e-05,
+      "loss": 0.1492,
+      "step": 1091
+    },
+    {
+      "epoch": 3.676767676767677,
+      "grad_norm": 0.3325386643409729,
+      "learning_rate": 1.987378767731557e-05,
+      "loss": 0.1532,
+      "step": 1092
+    },
+    {
+      "epoch": 3.68013468013468,
+      "grad_norm": 0.36293134093284607,
+      "learning_rate": 1.978003471884665e-05,
+      "loss": 0.1432,
+      "step": 1093
+    },
+    {
+      "epoch": 3.6835016835016834,
+      "grad_norm": 0.31937187910079956,
+      "learning_rate": 1.968644886197073e-05,
+      "loss": 0.1445,
+      "step": 1094
+    },
+    {
+      "epoch": 3.686868686868687,
+      "grad_norm": 0.3382173776626587,
+      "learning_rate": 1.9593030624171683e-05,
+      "loss": 0.152,
+      "step": 1095
+    },
+    {
+      "epoch": 3.6902356902356903,
+      "grad_norm": 0.4315609335899353,
+      "learning_rate": 1.949978052200658e-05,
+      "loss": 0.1544,
+      "step": 1096
+    },
+    {
+      "epoch": 3.6936026936026938,
+      "grad_norm": 0.33818742632865906,
+      "learning_rate": 1.9406699071102774e-05,
+      "loss": 0.1487,
+      "step": 1097
+    },
+    {
+      "epoch": 3.6969696969696972,
+      "grad_norm": 0.40022012591362,
+      "learning_rate": 1.9313786786155076e-05,
+      "loss": 0.1453,
+      "step": 1098
+    },
+    {
+      "epoch": 3.7003367003367003,
+      "grad_norm": 0.37208884954452515,
+      "learning_rate": 1.922104418092283e-05,
+      "loss": 0.1578,
+      "step": 1099
+    },
+    {
+      "epoch": 3.7037037037037037,
+      "grad_norm": 0.3757951855659485,
+      "learning_rate": 1.9128471768227203e-05,
+      "loss": 0.1545,
+      "step": 1100
+    },
+    {
+      "epoch": 3.7037037037037037,
+      "eval_loss": 0.0845775306224823,
+      "eval_runtime": 32.7177,
+      "eval_samples_per_second": 30.564,
+      "eval_steps_per_second": 1.926,
+      "step": 1100
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1485,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.4735886679130767e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}