diff --git "a/checkpoint-1854/trainer_state.json" "b/checkpoint-1854/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1854/trainer_state.json"
@@ -0,0 +1,13131 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.998787878787879,
+  "eval_steps": 123,
+  "global_step": 1854,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0016161616161616162,
+      "grad_norm": 31.875,
+      "learning_rate": 1.6129032258064518e-07,
+      "loss": 3.549,
+      "step": 1
+    },
+    {
+      "epoch": 0.0032323232323232323,
+      "grad_norm": 31.5,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 3.6934,
+      "step": 2
+    },
+    {
+      "epoch": 0.0048484848484848485,
+      "grad_norm": 31.75,
+      "learning_rate": 4.838709677419355e-07,
+      "loss": 3.458,
+      "step": 3
+    },
+    {
+      "epoch": 0.006464646464646465,
+      "grad_norm": 31.0,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 3.6786,
+      "step": 4
+    },
+    {
+      "epoch": 0.00808080808080808,
+      "grad_norm": 30.625,
+      "learning_rate": 8.064516129032258e-07,
+      "loss": 3.4513,
+      "step": 5
+    },
+    {
+      "epoch": 0.009696969696969697,
+      "grad_norm": 30.75,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 3.5122,
+      "step": 6
+    },
+    {
+      "epoch": 0.011313131313131313,
+      "grad_norm": 30.875,
+      "learning_rate": 1.1290322580645162e-06,
+      "loss": 3.4923,
+      "step": 7
+    },
+    {
+      "epoch": 0.01292929292929293,
+      "grad_norm": 30.5,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 3.4448,
+      "step": 8
+    },
+    {
+      "epoch": 0.014545454545454545,
+      "grad_norm": 30.625,
+      "learning_rate": 1.4516129032258066e-06,
+      "loss": 3.3285,
+      "step": 9
+    },
+    {
+      "epoch": 0.01616161616161616,
+      "grad_norm": 33.25,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 3.5214,
+      "step": 10
+    },
+    {
+      "epoch": 0.017777777777777778,
+      "grad_norm": 29.25,
+      "learning_rate": 1.774193548387097e-06,
+      "loss": 3.2033,
+      "step": 11
+    },
+    {
+      "epoch": 0.019393939393939394,
+      "grad_norm": 28.0,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 3.3047,
+      "step": 12
+    },
+    {
+      "epoch": 0.02101010101010101,
+      "grad_norm": 27.625,
+      "learning_rate": 2.096774193548387e-06,
+      "loss": 3.1362,
+      "step": 13
+    },
+    {
+      "epoch": 0.022626262626262626,
+      "grad_norm": 25.625,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 3.4485,
+      "step": 14
+    },
+    {
+      "epoch": 0.024242424242424242,
+      "grad_norm": 24.125,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 3.1889,
+      "step": 15
+    },
+    {
+      "epoch": 0.02585858585858586,
+      "grad_norm": 22.625,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 3.2893,
+      "step": 16
+    },
+    {
+      "epoch": 0.027474747474747475,
+      "grad_norm": 20.875,
+      "learning_rate": 2.7419354838709676e-06,
+      "loss": 3.1222,
+      "step": 17
+    },
+    {
+      "epoch": 0.02909090909090909,
+      "grad_norm": 19.625,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 3.0179,
+      "step": 18
+    },
+    {
+      "epoch": 0.030707070707070707,
+      "grad_norm": 18.375,
+      "learning_rate": 3.0645161290322584e-06,
+      "loss": 3.2402,
+      "step": 19
+    },
+    {
+      "epoch": 0.03232323232323232,
+      "grad_norm": 18.375,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 2.9451,
+      "step": 20
+    },
+    {
+      "epoch": 0.03393939393939394,
+      "grad_norm": 17.125,
+      "learning_rate": 3.3870967741935484e-06,
+      "loss": 2.8062,
+      "step": 21
+    },
+    {
+      "epoch": 0.035555555555555556,
+      "grad_norm": 16.25,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 2.653,
+      "step": 22
+    },
+    {
+      "epoch": 0.037171717171717175,
+      "grad_norm": 14.4375,
+      "learning_rate": 3.7096774193548392e-06,
+      "loss": 2.7289,
+      "step": 23
+    },
+    {
+      "epoch": 0.03878787878787879,
+      "grad_norm": 14.5625,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 2.4543,
+      "step": 24
+    },
+    {
+      "epoch": 0.04040404040404041,
+      "grad_norm": 13.375,
+      "learning_rate": 4.032258064516129e-06,
+      "loss": 2.2324,
+      "step": 25
+    },
+    {
+      "epoch": 0.04202020202020202,
+      "grad_norm": 12.5625,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 2.5554,
+      "step": 26
+    },
+    {
+      "epoch": 0.04363636363636364,
+      "grad_norm": 11.1875,
+      "learning_rate": 4.35483870967742e-06,
+      "loss": 2.2655,
+      "step": 27
+    },
+    {
+      "epoch": 0.04525252525252525,
+      "grad_norm": 10.625,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 2.2051,
+      "step": 28
+    },
+    {
+      "epoch": 0.04686868686868687,
+      "grad_norm": 10.125,
+      "learning_rate": 4.67741935483871e-06,
+      "loss": 2.1744,
+      "step": 29
+    },
+    {
+      "epoch": 0.048484848484848485,
+      "grad_norm": 9.8125,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 2.1542,
+      "step": 30
+    },
+    {
+      "epoch": 0.050101010101010104,
+      "grad_norm": 8.1875,
+      "learning_rate": 5e-06,
+      "loss": 2.067,
+      "step": 31
+    },
+    {
+      "epoch": 0.05171717171717172,
+      "grad_norm": 7.8125,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 1.7957,
+      "step": 32
+    },
+    {
+      "epoch": 0.05333333333333334,
+      "grad_norm": 7.34375,
+      "learning_rate": 5.322580645161291e-06,
+      "loss": 1.8623,
+      "step": 33
+    },
+    {
+      "epoch": 0.05494949494949495,
+      "grad_norm": 7.03125,
+      "learning_rate": 5.483870967741935e-06,
+      "loss": 1.6657,
+      "step": 34
+    },
+    {
+      "epoch": 0.05656565656565657,
+      "grad_norm": 6.3125,
+      "learning_rate": 5.645161290322582e-06,
+      "loss": 1.8778,
+      "step": 35
+    },
+    {
+      "epoch": 0.05818181818181818,
+      "grad_norm": 6.875,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 1.5248,
+      "step": 36
+    },
+    {
+      "epoch": 0.0597979797979798,
+      "grad_norm": 5.84375,
+      "learning_rate": 5.967741935483872e-06,
+      "loss": 1.5545,
+      "step": 37
+    },
+    {
+      "epoch": 0.061414141414141414,
+      "grad_norm": 5.5625,
+      "learning_rate": 6.129032258064517e-06,
+      "loss": 1.7116,
+      "step": 38
+    },
+    {
+      "epoch": 0.06303030303030303,
+      "grad_norm": 5.25,
+      "learning_rate": 6.290322580645162e-06,
+      "loss": 1.684,
+      "step": 39
+    },
+    {
+      "epoch": 0.06464646464646465,
+      "grad_norm": 4.9375,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 1.4803,
+      "step": 40
+    },
+    {
+      "epoch": 0.06626262626262626,
+      "grad_norm": 5.1875,
+      "learning_rate": 6.612903225806452e-06,
+      "loss": 1.5409,
+      "step": 41
+    },
+    {
+      "epoch": 0.06787878787878789,
+      "grad_norm": 4.71875,
+      "learning_rate": 6.774193548387097e-06,
+      "loss": 1.428,
+      "step": 42
+    },
+    {
+      "epoch": 0.0694949494949495,
+      "grad_norm": 4.8125,
+      "learning_rate": 6.935483870967743e-06,
+      "loss": 1.3183,
+      "step": 43
+    },
+    {
+      "epoch": 0.07111111111111111,
+      "grad_norm": 4.59375,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 1.5286,
+      "step": 44
+    },
+    {
+      "epoch": 0.07272727272727272,
+      "grad_norm": 4.3125,
+      "learning_rate": 7.258064516129033e-06,
+      "loss": 1.6481,
+      "step": 45
+    },
+    {
+      "epoch": 0.07434343434343435,
+      "grad_norm": 4.71875,
+      "learning_rate": 7.4193548387096784e-06,
+      "loss": 1.3227,
+      "step": 46
+    },
+    {
+      "epoch": 0.07595959595959596,
+      "grad_norm": 4.1875,
+      "learning_rate": 7.580645161290323e-06,
+      "loss": 1.5541,
+      "step": 47
+    },
+    {
+      "epoch": 0.07757575757575758,
+      "grad_norm": 4.5,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 1.6115,
+      "step": 48
+    },
+    {
+      "epoch": 0.07919191919191919,
+      "grad_norm": 4.375,
+      "learning_rate": 7.903225806451613e-06,
+      "loss": 1.5377,
+      "step": 49
+    },
+    {
+      "epoch": 0.08080808080808081,
+      "grad_norm": 4.46875,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 1.4701,
+      "step": 50
+    },
+    {
+      "epoch": 0.08242424242424243,
+      "grad_norm": 4.09375,
+      "learning_rate": 8.225806451612904e-06,
+      "loss": 1.3279,
+      "step": 51
+    },
+    {
+      "epoch": 0.08404040404040404,
+      "grad_norm": 4.1875,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 1.3193,
+      "step": 52
+    },
+    {
+      "epoch": 0.08565656565656565,
+      "grad_norm": 4.21875,
+      "learning_rate": 8.548387096774194e-06,
+      "loss": 1.3744,
+      "step": 53
+    },
+    {
+      "epoch": 0.08727272727272728,
+      "grad_norm": 4.4375,
+      "learning_rate": 8.70967741935484e-06,
+      "loss": 1.284,
+      "step": 54
+    },
+    {
+      "epoch": 0.08888888888888889,
+      "grad_norm": 4.59375,
+      "learning_rate": 8.870967741935484e-06,
+      "loss": 1.4204,
+      "step": 55
+    },
+    {
+      "epoch": 0.0905050505050505,
+      "grad_norm": 4.71875,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 1.097,
+      "step": 56
+    },
+    {
+      "epoch": 0.09212121212121212,
+      "grad_norm": 4.8125,
+      "learning_rate": 9.193548387096775e-06,
+      "loss": 1.6384,
+      "step": 57
+    },
+    {
+      "epoch": 0.09373737373737374,
+      "grad_norm": 4.1875,
+      "learning_rate": 9.35483870967742e-06,
+      "loss": 1.5198,
+      "step": 58
+    },
+    {
+      "epoch": 0.09535353535353536,
+      "grad_norm": 4.21875,
+      "learning_rate": 9.516129032258065e-06,
+      "loss": 0.9927,
+      "step": 59
+    },
+    {
+      "epoch": 0.09696969696969697,
+      "grad_norm": 4.28125,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 1.2467,
+      "step": 60
+    },
+    {
+      "epoch": 0.09858585858585858,
+      "grad_norm": 4.1875,
+      "learning_rate": 9.838709677419356e-06,
+      "loss": 1.4106,
+      "step": 61
+    },
+    {
+      "epoch": 0.10020202020202021,
+      "grad_norm": 3.890625,
+      "learning_rate": 1e-05,
+      "loss": 1.1905,
+      "step": 62
+    },
+    {
+      "epoch": 0.10181818181818182,
+      "grad_norm": 3.53125,
+      "learning_rate": 9.999920184173449e-06,
+      "loss": 1.2691,
+      "step": 63
+    },
+    {
+      "epoch": 0.10343434343434343,
+      "grad_norm": 3.78125,
+      "learning_rate": 9.999680739242022e-06,
+      "loss": 1.3019,
+      "step": 64
+    },
+    {
+      "epoch": 0.10505050505050505,
+      "grad_norm": 4.0,
+      "learning_rate": 9.999281672850317e-06,
+      "loss": 1.2566,
+      "step": 65
+    },
+    {
+      "epoch": 0.10666666666666667,
+      "grad_norm": 4.125,
+      "learning_rate": 9.99872299773906e-06,
+      "loss": 1.0568,
+      "step": 66
+    },
+    {
+      "epoch": 0.10828282828282829,
+      "grad_norm": 3.515625,
+      "learning_rate": 9.998004731744696e-06,
+      "loss": 1.0483,
+      "step": 67
+    },
+    {
+      "epoch": 0.1098989898989899,
+      "grad_norm": 3.421875,
+      "learning_rate": 9.997126897798826e-06,
+      "loss": 1.0828,
+      "step": 68
+    },
+    {
+      "epoch": 0.11151515151515151,
+      "grad_norm": 3.375,
+      "learning_rate": 9.996089523927461e-06,
+      "loss": 1.0364,
+      "step": 69
+    },
+    {
+      "epoch": 0.11313131313131314,
+      "grad_norm": 3.6875,
+      "learning_rate": 9.994892643250147e-06,
+      "loss": 1.4205,
+      "step": 70
+    },
+    {
+      "epoch": 0.11474747474747475,
+      "grad_norm": 3.453125,
+      "learning_rate": 9.993536293978892e-06,
+      "loss": 0.9768,
+      "step": 71
+    },
+    {
+      "epoch": 0.11636363636363636,
+      "grad_norm": 3.375,
+      "learning_rate": 9.99202051941695e-06,
+      "loss": 1.1508,
+      "step": 72
+    },
+    {
+      "epoch": 0.11797979797979798,
+      "grad_norm": 4.09375,
+      "learning_rate": 9.99034536795744e-06,
+      "loss": 1.1358,
+      "step": 73
+    },
+    {
+      "epoch": 0.1195959595959596,
+      "grad_norm": 3.140625,
+      "learning_rate": 9.9885108930818e-06,
+      "loss": 1.0681,
+      "step": 74
+    },
+    {
+      "epoch": 0.12121212121212122,
+      "grad_norm": 3.375,
+      "learning_rate": 9.986517153358086e-06,
+      "loss": 0.9067,
+      "step": 75
+    },
+    {
+      "epoch": 0.12282828282828283,
+      "grad_norm": 3.0625,
+      "learning_rate": 9.984364212439089e-06,
+      "loss": 1.1379,
+      "step": 76
+    },
+    {
+      "epoch": 0.12444444444444444,
+      "grad_norm": 3.234375,
+      "learning_rate": 9.982052139060312e-06,
+      "loss": 0.9532,
+      "step": 77
+    },
+    {
+      "epoch": 0.12606060606060607,
+      "grad_norm": 3.34375,
+      "learning_rate": 9.979581007037776e-06,
+      "loss": 1.1585,
+      "step": 78
+    },
+    {
+      "epoch": 0.12767676767676767,
+      "grad_norm": 3.34375,
+      "learning_rate": 9.976950895265657e-06,
+      "loss": 1.1824,
+      "step": 79
+    },
+    {
+      "epoch": 0.1292929292929293,
+      "grad_norm": 6.21875,
+      "learning_rate": 9.974161887713775e-06,
+      "loss": 1.2959,
+      "step": 80
+    },
+    {
+      "epoch": 0.13090909090909092,
+      "grad_norm": 3.375,
+      "learning_rate": 9.971214073424906e-06,
+      "loss": 1.6093,
+      "step": 81
+    },
+    {
+      "epoch": 0.13252525252525252,
+      "grad_norm": 3.453125,
+      "learning_rate": 9.968107546511942e-06,
+      "loss": 1.2891,
+      "step": 82
+    },
+    {
+      "epoch": 0.13414141414141414,
+      "grad_norm": 3.765625,
+      "learning_rate": 9.964842406154892e-06,
+      "loss": 1.0043,
+      "step": 83
+    },
+    {
+      "epoch": 0.13575757575757577,
+      "grad_norm": 3.03125,
+      "learning_rate": 9.961418756597703e-06,
+      "loss": 0.8925,
+      "step": 84
+    },
+    {
+      "epoch": 0.13737373737373737,
+      "grad_norm": 3.15625,
+      "learning_rate": 9.957836707144943e-06,
+      "loss": 1.0153,
+      "step": 85
+    },
+    {
+      "epoch": 0.138989898989899,
+      "grad_norm": 3.140625,
+      "learning_rate": 9.95409637215831e-06,
+      "loss": 1.1331,
+      "step": 86
+    },
+    {
+      "epoch": 0.1406060606060606,
+      "grad_norm": 3.234375,
+      "learning_rate": 9.950197871052974e-06,
+      "loss": 0.7181,
+      "step": 87
+    },
+    {
+      "epoch": 0.14222222222222222,
+      "grad_norm": 3.125,
+      "learning_rate": 9.94614132829377e-06,
+      "loss": 1.1873,
+      "step": 88
+    },
+    {
+      "epoch": 0.14383838383838385,
+      "grad_norm": 3.15625,
+      "learning_rate": 9.941926873391223e-06,
+      "loss": 0.8637,
+      "step": 89
+    },
+    {
+      "epoch": 0.14545454545454545,
+      "grad_norm": 3.375,
+      "learning_rate": 9.937554640897414e-06,
+      "loss": 1.1883,
+      "step": 90
+    },
+    {
+      "epoch": 0.14707070707070707,
+      "grad_norm": 3.359375,
+      "learning_rate": 9.933024770401682e-06,
+      "loss": 1.032,
+      "step": 91
+    },
+    {
+      "epoch": 0.1486868686868687,
+      "grad_norm": 3.21875,
+      "learning_rate": 9.928337406526172e-06,
+      "loss": 1.0899,
+      "step": 92
+    },
+    {
+      "epoch": 0.1503030303030303,
+      "grad_norm": 2.859375,
+      "learning_rate": 9.923492698921214e-06,
+      "loss": 0.7693,
+      "step": 93
+    },
+    {
+      "epoch": 0.15191919191919193,
+      "grad_norm": 3.0625,
+      "learning_rate": 9.918490802260538e-06,
+      "loss": 1.2708,
+      "step": 94
+    },
+    {
+      "epoch": 0.15353535353535352,
+      "grad_norm": 3.03125,
+      "learning_rate": 9.913331876236358e-06,
+      "loss": 1.2416,
+      "step": 95
+    },
+    {
+      "epoch": 0.15515151515151515,
+      "grad_norm": 3.0625,
+      "learning_rate": 9.90801608555425e-06,
+      "loss": 1.0996,
+      "step": 96
+    },
+    {
+      "epoch": 0.15676767676767678,
+      "grad_norm": 2.921875,
+      "learning_rate": 9.902543599927903e-06,
+      "loss": 0.8051,
+      "step": 97
+    },
+    {
+      "epoch": 0.15838383838383838,
+      "grad_norm": 3.09375,
+      "learning_rate": 9.896914594073703e-06,
+      "loss": 0.9512,
+      "step": 98
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.0,
+      "learning_rate": 9.891129247705153e-06,
+      "loss": 1.027,
+      "step": 99
+    },
+    {
+      "epoch": 0.16161616161616163,
+      "grad_norm": 3.734375,
+      "learning_rate": 9.885187745527132e-06,
+      "loss": 0.9311,
+      "step": 100
+    },
+    {
+      "epoch": 0.16323232323232323,
+      "grad_norm": 3.046875,
+      "learning_rate": 9.879090277230005e-06,
+      "loss": 1.0291,
+      "step": 101
+    },
+    {
+      "epoch": 0.16484848484848486,
+      "grad_norm": 3.28125,
+      "learning_rate": 9.87283703748356e-06,
+      "loss": 1.2054,
+      "step": 102
+    },
+    {
+      "epoch": 0.16646464646464645,
+      "grad_norm": 3.40625,
+      "learning_rate": 9.866428225930798e-06,
+      "loss": 0.9991,
+      "step": 103
+    },
+    {
+      "epoch": 0.16808080808080808,
+      "grad_norm": 3.046875,
+      "learning_rate": 9.859864047181551e-06,
+      "loss": 1.0588,
+      "step": 104
+    },
+    {
+      "epoch": 0.1696969696969697,
+      "grad_norm": 3.140625,
+      "learning_rate": 9.853144710805966e-06,
+      "loss": 1.1424,
+      "step": 105
+    },
+    {
+      "epoch": 0.1713131313131313,
+      "grad_norm": 2.796875,
+      "learning_rate": 9.846270431327793e-06,
+      "loss": 1.0388,
+      "step": 106
+    },
+    {
+      "epoch": 0.17292929292929293,
+      "grad_norm": 2.984375,
+      "learning_rate": 9.839241428217553e-06,
+      "loss": 0.8593,
+      "step": 107
+    },
+    {
+      "epoch": 0.17454545454545456,
+      "grad_norm": 3.21875,
+      "learning_rate": 9.832057925885526e-06,
+      "loss": 1.045,
+      "step": 108
+    },
+    {
+      "epoch": 0.17616161616161616,
+      "grad_norm": 3.171875,
+      "learning_rate": 9.824720153674578e-06,
+      "loss": 0.9642,
+      "step": 109
+    },
+    {
+      "epoch": 0.17777777777777778,
+      "grad_norm": 3.28125,
+      "learning_rate": 9.817228345852853e-06,
+      "loss": 1.0778,
+      "step": 110
+    },
+    {
+      "epoch": 0.17939393939393938,
+      "grad_norm": 2.953125,
+      "learning_rate": 9.809582741606283e-06,
+      "loss": 0.9812,
+      "step": 111
+    },
+    {
+      "epoch": 0.181010101010101,
+      "grad_norm": 3.125,
+      "learning_rate": 9.801783585030959e-06,
+      "loss": 0.9692,
+      "step": 112
+    },
+    {
+      "epoch": 0.18262626262626264,
+      "grad_norm": 3.421875,
+      "learning_rate": 9.79383112512533e-06,
+      "loss": 1.547,
+      "step": 113
+    },
+    {
+      "epoch": 0.18424242424242424,
+      "grad_norm": 3.140625,
+      "learning_rate": 9.785725615782262e-06,
+      "loss": 0.7568,
+      "step": 114
+    },
+    {
+      "epoch": 0.18585858585858586,
+      "grad_norm": 3.3125,
+      "learning_rate": 9.777467315780926e-06,
+      "loss": 1.137,
+      "step": 115
+    },
+    {
+      "epoch": 0.1874747474747475,
+      "grad_norm": 3.859375,
+      "learning_rate": 9.769056488778538e-06,
+      "loss": 0.9443,
+      "step": 116
+    },
+    {
+      "epoch": 0.1890909090909091,
+      "grad_norm": 3.046875,
+      "learning_rate": 9.760493403301941e-06,
+      "loss": 1.1996,
+      "step": 117
+    },
+    {
+      "epoch": 0.1907070707070707,
+      "grad_norm": 2.953125,
+      "learning_rate": 9.751778332739033e-06,
+      "loss": 0.7473,
+      "step": 118
+    },
+    {
+      "epoch": 0.1923232323232323,
+      "grad_norm": 3.25,
+      "learning_rate": 9.74291155533004e-06,
+      "loss": 1.2059,
+      "step": 119
+    },
+    {
+      "epoch": 0.19393939393939394,
+      "grad_norm": 2.84375,
+      "learning_rate": 9.733893354158628e-06,
+      "loss": 1.0131,
+      "step": 120
+    },
+    {
+      "epoch": 0.19555555555555557,
+      "grad_norm": 2.8125,
+      "learning_rate": 9.724724017142869e-06,
+      "loss": 0.9501,
+      "step": 121
+    },
+    {
+      "epoch": 0.19717171717171716,
+      "grad_norm": 3.0,
+      "learning_rate": 9.715403837026046e-06,
+      "loss": 1.0739,
+      "step": 122
+    },
+    {
+      "epoch": 0.1987878787878788,
+      "grad_norm": 3.078125,
+      "learning_rate": 9.705933111367314e-06,
+      "loss": 1.0345,
+      "step": 123
+    },
+    {
+      "epoch": 0.1987878787878788,
+      "eval_loss": 0.9800001978874207,
+      "eval_runtime": 33.66,
+      "eval_samples_per_second": 32.68,
+      "eval_steps_per_second": 4.1,
+      "step": 123
+    },
+    {
+      "epoch": 0.20040404040404042,
+      "grad_norm": 3.0,
+      "learning_rate": 9.69631214253219e-06,
+      "loss": 1.2008,
+      "step": 124
+    },
+    {
+      "epoch": 0.20202020202020202,
+      "grad_norm": 2.828125,
+      "learning_rate": 9.68654123768291e-06,
+      "loss": 0.8149,
+      "step": 125
+    },
+    {
+      "epoch": 0.20363636363636364,
+      "grad_norm": 3.03125,
+      "learning_rate": 9.676620708768608e-06,
+      "loss": 0.9796,
+      "step": 126
+    },
+    {
+      "epoch": 0.20525252525252524,
+      "grad_norm": 3.109375,
+      "learning_rate": 9.666550872515367e-06,
+      "loss": 1.0588,
+      "step": 127
+    },
+    {
+      "epoch": 0.20686868686868687,
+      "grad_norm": 3.375,
+      "learning_rate": 9.656332050416118e-06,
+      "loss": 1.1567,
+      "step": 128
+    },
+    {
+      "epoch": 0.2084848484848485,
+      "grad_norm": 3.640625,
+      "learning_rate": 9.645964568720345e-06,
+      "loss": 0.9878,
+      "step": 129
+    },
+    {
+      "epoch": 0.2101010101010101,
+      "grad_norm": 3.078125,
+      "learning_rate": 9.635448758423703e-06,
+      "loss": 0.8607,
+      "step": 130
+    },
+    {
+      "epoch": 0.21171717171717172,
+      "grad_norm": 2.96875,
+      "learning_rate": 9.624784955257423e-06,
+      "loss": 1.0056,
+      "step": 131
+    },
+    {
+      "epoch": 0.21333333333333335,
+      "grad_norm": 3.40625,
+      "learning_rate": 9.613973499677613e-06,
+      "loss": 1.1434,
+      "step": 132
+    },
+    {
+      "epoch": 0.21494949494949495,
+      "grad_norm": 3.40625,
+      "learning_rate": 9.60301473685438e-06,
+      "loss": 0.9318,
+      "step": 133
+    },
+    {
+      "epoch": 0.21656565656565657,
+      "grad_norm": 3.125,
+      "learning_rate": 9.591909016660806e-06,
+      "loss": 0.9447,
+      "step": 134
+    },
+    {
+      "epoch": 0.21818181818181817,
+      "grad_norm": 2.953125,
+      "learning_rate": 9.580656693661787e-06,
+      "loss": 0.9087,
+      "step": 135
+    },
+    {
+      "epoch": 0.2197979797979798,
+      "grad_norm": 2.84375,
+      "learning_rate": 9.569258127102708e-06,
+      "loss": 0.899,
+      "step": 136
+    },
+    {
+      "epoch": 0.22141414141414142,
+      "grad_norm": 2.984375,
+      "learning_rate": 9.55771368089797e-06,
+      "loss": 1.0117,
+      "step": 137
+    },
+    {
+      "epoch": 0.22303030303030302,
+      "grad_norm": 3.078125,
+      "learning_rate": 9.546023723619387e-06,
+      "loss": 0.8801,
+      "step": 138
+    },
+    {
+      "epoch": 0.22464646464646465,
+      "grad_norm": 2.796875,
+      "learning_rate": 9.534188628484391e-06,
+      "loss": 1.2514,
+      "step": 139
+    },
+    {
+      "epoch": 0.22626262626262628,
+      "grad_norm": 2.859375,
+      "learning_rate": 9.522208773344147e-06,
+      "loss": 1.1562,
+      "step": 140
+    },
+    {
+      "epoch": 0.22787878787878788,
+      "grad_norm": 2.9375,
+      "learning_rate": 9.510084540671471e-06,
+      "loss": 1.0885,
+      "step": 141
+    },
+    {
+      "epoch": 0.2294949494949495,
+      "grad_norm": 2.953125,
+      "learning_rate": 9.497816317548625e-06,
+      "loss": 1.2913,
+      "step": 142
+    },
+    {
+      "epoch": 0.2311111111111111,
+      "grad_norm": 3.015625,
+      "learning_rate": 9.485404495654953e-06,
+      "loss": 0.8103,
+      "step": 143
+    },
+    {
+      "epoch": 0.23272727272727273,
+      "grad_norm": 2.890625,
+      "learning_rate": 9.472849471254386e-06,
+      "loss": 0.7462,
+      "step": 144
+    },
+    {
+      "epoch": 0.23434343434343435,
+      "grad_norm": 3.046875,
+      "learning_rate": 9.460151645182784e-06,
+      "loss": 1.2383,
+      "step": 145
+    },
+    {
+      "epoch": 0.23595959595959595,
+      "grad_norm": 2.75,
+      "learning_rate": 9.447311422835141e-06,
+      "loss": 0.9858,
+      "step": 146
+    },
+    {
+      "epoch": 0.23757575757575758,
+      "grad_norm": 3.015625,
+      "learning_rate": 9.43432921415264e-06,
+      "loss": 1.1112,
+      "step": 147
+    },
+    {
+      "epoch": 0.2391919191919192,
+      "grad_norm": 3.171875,
+      "learning_rate": 9.421205433609568e-06,
+      "loss": 0.7905,
+      "step": 148
+    },
+    {
+      "epoch": 0.2408080808080808,
+      "grad_norm": 2.859375,
+      "learning_rate": 9.407940500200082e-06,
+      "loss": 0.9302,
+      "step": 149
+    },
+    {
+      "epoch": 0.24242424242424243,
+      "grad_norm": 3.046875,
+      "learning_rate": 9.39453483742483e-06,
+      "loss": 0.8402,
+      "step": 150
+    },
+    {
+      "epoch": 0.24404040404040403,
+      "grad_norm": 3.015625,
+      "learning_rate": 9.380988873277436e-06,
+      "loss": 1.0566,
+      "step": 151
+    },
+    {
+      "epoch": 0.24565656565656566,
+      "grad_norm": 3.125,
+      "learning_rate": 9.367303040230828e-06,
+      "loss": 1.0689,
+      "step": 152
+    },
+    {
+      "epoch": 0.24727272727272728,
+      "grad_norm": 2.796875,
+      "learning_rate": 9.35347777522344e-06,
+      "loss": 1.1064,
+      "step": 153
+    },
+    {
+      "epoch": 0.24888888888888888,
+      "grad_norm": 3.0,
+      "learning_rate": 9.33951351964525e-06,
+      "loss": 0.8518,
+      "step": 154
+    },
+    {
+      "epoch": 0.2505050505050505,
+      "grad_norm": 2.890625,
+      "learning_rate": 9.3254107193237e-06,
+      "loss": 0.9333,
+      "step": 155
+    },
+    {
+      "epoch": 0.25212121212121213,
+      "grad_norm": 2.78125,
+      "learning_rate": 9.311169824509454e-06,
+      "loss": 0.7111,
+      "step": 156
+    },
+    {
+      "epoch": 0.25373737373737376,
+      "grad_norm": 2.859375,
+      "learning_rate": 9.296791289862031e-06,
+      "loss": 0.7406,
+      "step": 157
+    },
+    {
+      "epoch": 0.25535353535353533,
+      "grad_norm": 3.234375,
+      "learning_rate": 9.28227557443528e-06,
+      "loss": 1.057,
+      "step": 158
+    },
+    {
+      "epoch": 0.25696969696969696,
+      "grad_norm": 3.46875,
+      "learning_rate": 9.267623141662734e-06,
+      "loss": 1.0833,
+      "step": 159
+    },
+    {
+      "epoch": 0.2585858585858586,
+      "grad_norm": 3.171875,
+      "learning_rate": 9.252834459342801e-06,
+      "loss": 1.2084,
+      "step": 160
+    },
+    {
+      "epoch": 0.2602020202020202,
+      "grad_norm": 2.75,
+      "learning_rate": 9.237909999623847e-06,
+      "loss": 0.9263,
+      "step": 161
+    },
+    {
+      "epoch": 0.26181818181818184,
+      "grad_norm": 3.015625,
+      "learning_rate": 9.222850238989104e-06,
+      "loss": 0.8601,
+      "step": 162
+    },
+    {
+      "epoch": 0.2634343434343434,
+      "grad_norm": 2.921875,
+      "learning_rate": 9.207655658241469e-06,
+      "loss": 0.7775,
+      "step": 163
+    },
+    {
+      "epoch": 0.26505050505050504,
+      "grad_norm": 3.0625,
+      "learning_rate": 9.192326742488153e-06,
+      "loss": 0.9596,
+      "step": 164
+    },
+    {
+      "epoch": 0.26666666666666666,
+      "grad_norm": 2.84375,
+      "learning_rate": 9.176863981125185e-06,
+      "loss": 0.7565,
+      "step": 165
+    },
+    {
+      "epoch": 0.2682828282828283,
+      "grad_norm": 2.796875,
+      "learning_rate": 9.161267867821802e-06,
+      "loss": 1.1118,
+      "step": 166
+    },
+    {
+      "epoch": 0.2698989898989899,
+      "grad_norm": 2.875,
+      "learning_rate": 9.14553890050467e-06,
+      "loss": 0.9555,
+      "step": 167
+    },
+    {
+      "epoch": 0.27151515151515154,
+      "grad_norm": 3.09375,
+      "learning_rate": 9.129677581342e-06,
+      "loss": 1.0553,
+      "step": 168
+    },
+    {
+      "epoch": 0.2731313131313131,
+      "grad_norm": 3.28125,
+      "learning_rate": 9.113684416727511e-06,
+      "loss": 1.0046,
+      "step": 169
+    },
+    {
+      "epoch": 0.27474747474747474,
+      "grad_norm": 3.265625,
+      "learning_rate": 9.097559917264268e-06,
+      "loss": 1.0465,
+      "step": 170
+    },
+    {
+      "epoch": 0.27636363636363637,
+      "grad_norm": 2.984375,
+      "learning_rate": 9.081304597748366e-06,
+      "loss": 0.8378,
+      "step": 171
+    },
+    {
+      "epoch": 0.277979797979798,
+      "grad_norm": 3.234375,
+      "learning_rate": 9.064918977152517e-06,
+      "loss": 0.9942,
+      "step": 172
+    },
+    {
+      "epoch": 0.2795959595959596,
+      "grad_norm": 2.875,
+      "learning_rate": 9.048403578609454e-06,
+      "loss": 1.032,
+      "step": 173
+    },
+    {
+      "epoch": 0.2812121212121212,
+      "grad_norm": 2.953125,
+      "learning_rate": 9.031758929395259e-06,
+      "loss": 1.1305,
+      "step": 174
+    },
+    {
+      "epoch": 0.2828282828282828,
+      "grad_norm": 3.09375,
+      "learning_rate": 9.014985560912499e-06,
+      "loss": 0.7198,
+      "step": 175
+    },
+    {
+      "epoch": 0.28444444444444444,
+      "grad_norm": 3.046875,
+      "learning_rate": 8.998084008673284e-06,
+      "loss": 0.8477,
+      "step": 176
+    },
+    {
+      "epoch": 0.28606060606060607,
+      "grad_norm": 2.8125,
+      "learning_rate": 8.981054812282162e-06,
+      "loss": 0.9012,
+      "step": 177
+    },
+    {
+      "epoch": 0.2876767676767677,
+      "grad_norm": 2.890625,
+      "learning_rate": 8.963898515418885e-06,
+      "loss": 0.8814,
+      "step": 178
+    },
+    {
+      "epoch": 0.28929292929292927,
+      "grad_norm": 3.1875,
+      "learning_rate": 8.946615665821059e-06,
+      "loss": 0.9186,
+      "step": 179
+    },
+    {
+      "epoch": 0.2909090909090909,
+      "grad_norm": 3.0,
+      "learning_rate": 8.929206815266653e-06,
+      "loss": 0.8421,
+      "step": 180
+    },
+    {
+      "epoch": 0.2925252525252525,
+      "grad_norm": 3.171875,
+      "learning_rate": 8.911672519556386e-06,
+      "loss": 0.9127,
+      "step": 181
+    },
+    {
+      "epoch": 0.29414141414141415,
+      "grad_norm": 3.15625,
+      "learning_rate": 8.89401333849598e-06,
+      "loss": 0.8361,
+      "step": 182
+    },
+    {
+      "epoch": 0.2957575757575758,
+      "grad_norm": 3.171875,
+      "learning_rate": 8.87622983587829e-06,
+      "loss": 1.2804,
+      "step": 183
+    },
+    {
+      "epoch": 0.2973737373737374,
+      "grad_norm": 3.21875,
+      "learning_rate": 8.8583225794653e-06,
+      "loss": 1.1745,
+      "step": 184
+    },
+    {
+      "epoch": 0.298989898989899,
+      "grad_norm": 3.0625,
+      "learning_rate": 8.840292140969995e-06,
+      "loss": 0.7905,
+      "step": 185
+    },
+    {
+      "epoch": 0.3006060606060606,
+      "grad_norm": 3.25,
+      "learning_rate": 8.82213909603812e-06,
+      "loss": 1.1923,
+      "step": 186
+    },
+    {
+      "epoch": 0.3022222222222222,
+      "grad_norm": 3.03125,
+      "learning_rate": 8.803864024229786e-06,
+      "loss": 1.0463,
+      "step": 187
+    },
+    {
+      "epoch": 0.30383838383838385,
+      "grad_norm": 3.015625,
+      "learning_rate": 8.78546750900098e-06,
+      "loss": 0.8148,
+      "step": 188
+    },
+    {
+      "epoch": 0.3054545454545455,
+      "grad_norm": 3.171875,
+      "learning_rate": 8.766950137684929e-06,
+      "loss": 0.7807,
+      "step": 189
+    },
+    {
+      "epoch": 0.30707070707070705,
+      "grad_norm": 3.4375,
+      "learning_rate": 8.748312501473351e-06,
+      "loss": 0.9707,
+      "step": 190
+    },
+    {
+      "epoch": 0.3086868686868687,
+      "grad_norm": 3.0,
+      "learning_rate": 8.72955519539758e-06,
+      "loss": 0.7882,
+      "step": 191
+    },
+    {
+      "epoch": 0.3103030303030303,
+      "grad_norm": 2.859375,
+      "learning_rate": 8.710678818309576e-06,
+      "loss": 0.8979,
+      "step": 192
+    },
+    {
+      "epoch": 0.31191919191919193,
+      "grad_norm": 3.046875,
+      "learning_rate": 8.691683972862792e-06,
+      "loss": 0.9243,
+      "step": 193
+    },
+    {
+      "epoch": 0.31353535353535356,
+      "grad_norm": 2.859375,
+      "learning_rate": 8.672571265492944e-06,
+      "loss": 1.473,
+      "step": 194
+    },
+    {
+      "epoch": 0.3151515151515151,
+      "grad_norm": 2.9375,
+      "learning_rate": 8.653341306398645e-06,
+      "loss": 1.0458,
+      "step": 195
+    },
+    {
+      "epoch": 0.31676767676767675,
+      "grad_norm": 2.78125,
+      "learning_rate": 8.63399470952193e-06,
+      "loss": 0.9828,
+      "step": 196
+    },
+    {
+      "epoch": 0.3183838383838384,
+      "grad_norm": 2.765625,
+      "learning_rate": 8.614532092528645e-06,
+      "loss": 0.9094,
+      "step": 197
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.15625,
+      "learning_rate": 8.594954076788736e-06,
+      "loss": 1.0664,
+      "step": 198
+    },
+    {
+      "epoch": 0.32161616161616163,
+      "grad_norm": 3.296875,
+      "learning_rate": 8.575261287356407e-06,
+      "loss": 0.9402,
+      "step": 199
+    },
+    {
+      "epoch": 0.32323232323232326,
+      "grad_norm": 3.140625,
+      "learning_rate": 8.555454352950161e-06,
+      "loss": 0.8442,
+      "step": 200
+    },
+    {
+      "epoch": 0.32484848484848483,
+      "grad_norm": 2.96875,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 1.0386,
+      "step": 201
+    },
+    {
+      "epoch": 0.32646464646464646,
+      "grad_norm": 3.515625,
+      "learning_rate": 8.515500582290914e-06,
+      "loss": 0.8054,
+      "step": 202
+    },
+    {
+      "epoch": 0.3280808080808081,
+      "grad_norm": 3.5625,
+      "learning_rate": 8.495355021615204e-06,
+      "loss": 0.9728,
+      "step": 203
+    },
+    {
+      "epoch": 0.3296969696969697,
+      "grad_norm": 3.234375,
+      "learning_rate": 8.475097867079437e-06,
+      "loss": 1.0753,
+      "step": 204
+    },
+    {
+      "epoch": 0.33131313131313134,
+      "grad_norm": 2.984375,
+      "learning_rate": 8.454729765420228e-06,
+      "loss": 0.8202,
+      "step": 205
+    },
+    {
+      "epoch": 0.3329292929292929,
+      "grad_norm": 2.890625,
+      "learning_rate": 8.434251366916323e-06,
+      "loss": 0.7964,
+      "step": 206
+    },
+    {
+      "epoch": 0.33454545454545453,
+      "grad_norm": 3.125,
+      "learning_rate": 8.413663325367845e-06,
+      "loss": 0.8977,
+      "step": 207
+    },
+    {
+      "epoch": 0.33616161616161616,
+      "grad_norm": 2.90625,
+      "learning_rate": 8.392966298075413e-06,
+      "loss": 0.7363,
+      "step": 208
+    },
+    {
+      "epoch": 0.3377777777777778,
+      "grad_norm": 3.09375,
+      "learning_rate": 8.372160945819164e-06,
+      "loss": 0.7138,
+      "step": 209
+    },
+    {
+      "epoch": 0.3393939393939394,
+      "grad_norm": 3.1875,
+      "learning_rate": 8.351247932837655e-06,
+      "loss": 0.8652,
+      "step": 210
+    },
+    {
+      "epoch": 0.341010101010101,
+      "grad_norm": 2.953125,
+      "learning_rate": 8.330227926806652e-06,
+      "loss": 0.712,
+      "step": 211
+    },
+    {
+      "epoch": 0.3426262626262626,
+      "grad_norm": 3.0,
+      "learning_rate": 8.309101598817812e-06,
+      "loss": 0.9503,
+      "step": 212
+    },
+    {
+      "epoch": 0.34424242424242424,
+      "grad_norm": 3.34375,
+      "learning_rate": 8.287869623357275e-06,
+      "loss": 1.0047,
+      "step": 213
+    },
+    {
+      "epoch": 0.34585858585858587,
+      "grad_norm": 3.203125,
+      "learning_rate": 8.266532678284103e-06,
+      "loss": 0.9647,
+      "step": 214
+    },
+    {
+      "epoch": 0.3474747474747475,
+      "grad_norm": 3.046875,
+      "learning_rate": 8.245091444808663e-06,
+      "loss": 0.9629,
+      "step": 215
+    },
+    {
+      "epoch": 0.3490909090909091,
+      "grad_norm": 2.84375,
+      "learning_rate": 8.223546607470863e-06,
+      "loss": 0.8938,
+      "step": 216
+    },
+    {
+      "epoch": 0.3507070707070707,
+      "grad_norm": 2.859375,
+      "learning_rate": 8.201898854118301e-06,
+      "loss": 0.8341,
+      "step": 217
+    },
+    {
+      "epoch": 0.3523232323232323,
+      "grad_norm": 3.625,
+      "learning_rate": 8.18014887588431e-06,
+      "loss": 0.9081,
+      "step": 218
+    },
+    {
+      "epoch": 0.35393939393939394,
+      "grad_norm": 3.5,
+      "learning_rate": 8.158297367165885e-06,
+      "loss": 1.0218,
+      "step": 219
+    },
+    {
+      "epoch": 0.35555555555555557,
+      "grad_norm": 3.046875,
+      "learning_rate": 8.13634502560152e-06,
+      "loss": 1.053,
+      "step": 220
+    },
+    {
+      "epoch": 0.3571717171717172,
+      "grad_norm": 2.953125,
+      "learning_rate": 8.114292552048925e-06,
+      "loss": 0.7958,
+      "step": 221
+    },
+    {
+      "epoch": 0.35878787878787877,
+      "grad_norm": 3.015625,
+      "learning_rate": 8.092140650562665e-06,
+      "loss": 0.9637,
+      "step": 222
+    },
+    {
+      "epoch": 0.3604040404040404,
+      "grad_norm": 3.34375,
+      "learning_rate": 8.069890028371672e-06,
+      "loss": 1.0193,
+      "step": 223
+    },
+    {
+      "epoch": 0.362020202020202,
+      "grad_norm": 2.953125,
+      "learning_rate": 8.047541395856661e-06,
+      "loss": 0.919,
+      "step": 224
+    },
+    {
+      "epoch": 0.36363636363636365,
+      "grad_norm": 3.0625,
+      "learning_rate": 8.025095466527468e-06,
+      "loss": 1.0924,
+      "step": 225
+    },
+    {
+      "epoch": 0.3652525252525253,
+      "grad_norm": 3.265625,
+      "learning_rate": 8.002552957000254e-06,
+      "loss": 0.9645,
+      "step": 226
+    },
+    {
+      "epoch": 0.36686868686868684,
+      "grad_norm": 2.984375,
+      "learning_rate": 7.979914586974628e-06,
+      "loss": 1.169,
+      "step": 227
+    },
+    {
+      "epoch": 0.36848484848484847,
+      "grad_norm": 3.171875,
+      "learning_rate": 7.957181079210676e-06,
+      "loss": 0.7609,
+      "step": 228
+    },
+    {
+      "epoch": 0.3701010101010101,
+      "grad_norm": 3.453125,
+      "learning_rate": 7.934353159505885e-06,
+      "loss": 0.9479,
+      "step": 229
+    },
+    {
+      "epoch": 0.3717171717171717,
+      "grad_norm": 2.65625,
+      "learning_rate": 7.911431556671967e-06,
+      "loss": 0.6938,
+      "step": 230
+    },
+    {
+      "epoch": 0.37333333333333335,
+      "grad_norm": 3.0625,
+      "learning_rate": 7.888417002511592e-06,
+      "loss": 0.8091,
+      "step": 231
+    },
+    {
+      "epoch": 0.374949494949495,
+      "grad_norm": 2.796875,
+      "learning_rate": 7.865310231795026e-06,
+      "loss": 0.6972,
+      "step": 232
+    },
+    {
+      "epoch": 0.37656565656565655,
+      "grad_norm": 3.234375,
+      "learning_rate": 7.842111982236669e-06,
+      "loss": 0.9092,
+      "step": 233
+    },
+    {
+      "epoch": 0.3781818181818182,
+      "grad_norm": 3.03125,
+      "learning_rate": 7.818822994471504e-06,
+      "loss": 0.7496,
+      "step": 234
+    },
+    {
+      "epoch": 0.3797979797979798,
+      "grad_norm": 3.171875,
+      "learning_rate": 7.79544401203146e-06,
+      "loss": 0.962,
+      "step": 235
+    },
+    {
+      "epoch": 0.3814141414141414,
+      "grad_norm": 2.84375,
+      "learning_rate": 7.771975781321655e-06,
+      "loss": 0.9505,
+      "step": 236
+    },
+    {
+      "epoch": 0.38303030303030305,
+      "grad_norm": 2.90625,
+      "learning_rate": 7.748419051596586e-06,
+      "loss": 0.986,
+      "step": 237
+    },
+    {
+      "epoch": 0.3846464646464646,
+      "grad_norm": 2.828125,
+      "learning_rate": 7.72477457493619e-06,
+      "loss": 0.6717,
+      "step": 238
+    },
+    {
+      "epoch": 0.38626262626262625,
+      "grad_norm": 3.0625,
+      "learning_rate": 7.701043106221847e-06,
+      "loss": 0.6942,
+      "step": 239
+    },
+    {
+      "epoch": 0.3878787878787879,
+      "grad_norm": 2.875,
+      "learning_rate": 7.677225403112277e-06,
+      "loss": 0.8816,
+      "step": 240
+    },
+    {
+      "epoch": 0.3894949494949495,
+      "grad_norm": 3.265625,
+      "learning_rate": 7.653322226019341e-06,
+      "loss": 0.8844,
+      "step": 241
+    },
+    {
+      "epoch": 0.39111111111111113,
+      "grad_norm": 2.984375,
+      "learning_rate": 7.629334338083774e-06,
+      "loss": 0.9657,
+      "step": 242
+    },
+    {
+      "epoch": 0.3927272727272727,
+      "grad_norm": 3.03125,
+      "learning_rate": 7.605262505150819e-06,
+      "loss": 0.9712,
+      "step": 243
+    },
+    {
+      "epoch": 0.39434343434343433,
+      "grad_norm": 4.0625,
+      "learning_rate": 7.58110749574577e-06,
+      "loss": 0.5433,
+      "step": 244
+    },
+    {
+      "epoch": 0.39595959595959596,
+      "grad_norm": 3.203125,
+      "learning_rate": 7.556870081049444e-06,
+      "loss": 0.8649,
+      "step": 245
+    },
+    {
+      "epoch": 0.3975757575757576,
+      "grad_norm": 2.703125,
+      "learning_rate": 7.532551034873558e-06,
+      "loss": 0.8537,
+      "step": 246
+    },
+    {
+      "epoch": 0.3975757575757576,
+      "eval_loss": 0.880824863910675,
+      "eval_runtime": 33.712,
+      "eval_samples_per_second": 32.629,
+      "eval_steps_per_second": 4.094,
+      "step": 246
+    },
+    {
+      "epoch": 0.3991919191919192,
+      "grad_norm": 2.859375,
+      "learning_rate": 7.50815113363602e-06,
+      "loss": 0.8151,
+      "step": 247
+    },
+    {
+      "epoch": 0.40080808080808084,
+      "grad_norm": 3.109375,
+      "learning_rate": 7.483671156336142e-06,
+      "loss": 0.9314,
+      "step": 248
+    },
+    {
+      "epoch": 0.4024242424242424,
+      "grad_norm": 3.25,
+      "learning_rate": 7.459111884529775e-06,
+      "loss": 0.8823,
+      "step": 249
+    },
+    {
+      "epoch": 0.40404040404040403,
+      "grad_norm": 3.046875,
+      "learning_rate": 7.43447410230435e-06,
+      "loss": 0.812,
+      "step": 250
+    },
+    {
+      "epoch": 0.40565656565656566,
+      "grad_norm": 2.859375,
+      "learning_rate": 7.409758596253849e-06,
+      "loss": 0.798,
+      "step": 251
+    },
+    {
+      "epoch": 0.4072727272727273,
+      "grad_norm": 2.75,
+      "learning_rate": 7.384966155453686e-06,
+      "loss": 0.9391,
+      "step": 252
+    },
+    {
+      "epoch": 0.4088888888888889,
+      "grad_norm": 3.578125,
+      "learning_rate": 7.360097571435527e-06,
+      "loss": 0.8732,
+      "step": 253
+    },
+    {
+      "epoch": 0.4105050505050505,
+      "grad_norm": 3.453125,
+      "learning_rate": 7.335153638162005e-06,
+      "loss": 0.9427,
+      "step": 254
+    },
+    {
+      "epoch": 0.4121212121212121,
+      "grad_norm": 2.953125,
+      "learning_rate": 7.310135152001381e-06,
+      "loss": 1.1767,
+      "step": 255
+    },
+    {
+      "epoch": 0.41373737373737374,
+      "grad_norm": 3.609375,
+      "learning_rate": 7.285042911702116e-06,
+      "loss": 1.0432,
+      "step": 256
+    },
+    {
+      "epoch": 0.41535353535353536,
+      "grad_norm": 3.5625,
+      "learning_rate": 7.259877718367372e-06,
+      "loss": 1.1834,
+      "step": 257
+    },
+    {
+      "epoch": 0.416969696969697,
+      "grad_norm": 3.25,
+      "learning_rate": 7.234640375429427e-06,
+      "loss": 0.921,
+      "step": 258
+    },
+    {
+      "epoch": 0.41858585858585856,
+      "grad_norm": 2.96875,
+      "learning_rate": 7.209331688624039e-06,
+      "loss": 0.9534,
+      "step": 259
+    },
+    {
+      "epoch": 0.4202020202020202,
+      "grad_norm": 3.078125,
+      "learning_rate": 7.183952465964711e-06,
+      "loss": 0.6627,
+      "step": 260
+    },
+    {
+      "epoch": 0.4218181818181818,
+      "grad_norm": 3.125,
+      "learning_rate": 7.158503517716894e-06,
+      "loss": 0.8148,
+      "step": 261
+    },
+    {
+      "epoch": 0.42343434343434344,
+      "grad_norm": 3.125,
+      "learning_rate": 7.132985656372126e-06,
+      "loss": 0.8112,
+      "step": 262
+    },
+    {
+      "epoch": 0.42505050505050507,
+      "grad_norm": 2.9375,
+      "learning_rate": 7.1073996966220835e-06,
+      "loss": 0.8821,
+      "step": 263
+    },
+    {
+      "epoch": 0.4266666666666667,
+      "grad_norm": 2.96875,
+      "learning_rate": 7.0817464553325764e-06,
+      "loss": 1.0924,
+      "step": 264
+    },
+    {
+      "epoch": 0.42828282828282827,
+      "grad_norm": 3.015625,
+      "learning_rate": 7.0560267515174685e-06,
+      "loss": 0.8484,
+      "step": 265
+    },
+    {
+      "epoch": 0.4298989898989899,
+      "grad_norm": 3.015625,
+      "learning_rate": 7.030241406312528e-06,
+      "loss": 0.9783,
+      "step": 266
+    },
+    {
+      "epoch": 0.4315151515151515,
+      "grad_norm": 3.1875,
+      "learning_rate": 7.004391242949209e-06,
+      "loss": 1.1527,
+      "step": 267
+    },
+    {
+      "epoch": 0.43313131313131314,
+      "grad_norm": 3.171875,
+      "learning_rate": 6.978477086728375e-06,
+      "loss": 0.8954,
+      "step": 268
+    },
+    {
+      "epoch": 0.43474747474747477,
+      "grad_norm": 3.171875,
+      "learning_rate": 6.952499764993945e-06,
+      "loss": 0.8888,
+      "step": 269
+    },
+    {
+      "epoch": 0.43636363636363634,
+      "grad_norm": 2.96875,
+      "learning_rate": 6.926460107106483e-06,
+      "loss": 0.9581,
+      "step": 270
+    },
+    {
+      "epoch": 0.43797979797979797,
+      "grad_norm": 3.09375,
+      "learning_rate": 6.900358944416712e-06,
+      "loss": 0.8114,
+      "step": 271
+    },
+    {
+      "epoch": 0.4395959595959596,
+      "grad_norm": 2.953125,
+      "learning_rate": 6.874197110238986e-06,
+      "loss": 0.9492,
+      "step": 272
+    },
+    {
+      "epoch": 0.4412121212121212,
+      "grad_norm": 3.0,
+      "learning_rate": 6.847975439824669e-06,
+      "loss": 1.1569,
+      "step": 273
+    },
+    {
+      "epoch": 0.44282828282828285,
+      "grad_norm": 2.90625,
+      "learning_rate": 6.8216947703354815e-06,
+      "loss": 0.9314,
+      "step": 274
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 3.328125,
+      "learning_rate": 6.795355940816768e-06,
+      "loss": 1.1551,
+      "step": 275
+    },
+    {
+      "epoch": 0.44606060606060605,
+      "grad_norm": 3.4375,
+      "learning_rate": 6.7689597921707065e-06,
+      "loss": 1.1893,
+      "step": 276
+    },
+    {
+      "epoch": 0.4476767676767677,
+      "grad_norm": 2.875,
+      "learning_rate": 6.742507167129465e-06,
+      "loss": 0.7809,
+      "step": 277
+    },
+    {
+      "epoch": 0.4492929292929293,
+      "grad_norm": 3.0,
+      "learning_rate": 6.715998910228296e-06,
+      "loss": 0.7677,
+      "step": 278
+    },
+    {
+      "epoch": 0.4509090909090909,
+      "grad_norm": 2.78125,
+      "learning_rate": 6.689435867778578e-06,
+      "loss": 0.8212,
+      "step": 279
+    },
+    {
+      "epoch": 0.45252525252525255,
+      "grad_norm": 3.203125,
+      "learning_rate": 6.6628188878407806e-06,
+      "loss": 0.9297,
+      "step": 280
+    },
+    {
+      "epoch": 0.4541414141414141,
+      "grad_norm": 3.03125,
+      "learning_rate": 6.636148820197409e-06,
+      "loss": 0.6459,
+      "step": 281
+    },
+    {
+      "epoch": 0.45575757575757575,
+      "grad_norm": 3.03125,
+      "learning_rate": 6.609426516325859e-06,
+      "loss": 0.8884,
+      "step": 282
+    },
+    {
+      "epoch": 0.4573737373737374,
+      "grad_norm": 2.859375,
+      "learning_rate": 6.58265282937124e-06,
+      "loss": 0.7032,
+      "step": 283
+    },
+    {
+      "epoch": 0.458989898989899,
+      "grad_norm": 2.8125,
+      "learning_rate": 6.555828614119132e-06,
+      "loss": 1.0391,
+      "step": 284
+    },
+    {
+      "epoch": 0.46060606060606063,
+      "grad_norm": 3.21875,
+      "learning_rate": 6.528954726968302e-06,
+      "loss": 0.7977,
+      "step": 285
+    },
+    {
+      "epoch": 0.4622222222222222,
+      "grad_norm": 2.96875,
+      "learning_rate": 6.502032025903356e-06,
+      "loss": 0.9772,
+      "step": 286
+    },
+    {
+      "epoch": 0.4638383838383838,
+      "grad_norm": 2.90625,
+      "learning_rate": 6.475061370467346e-06,
+      "loss": 0.7837,
+      "step": 287
+    },
+    {
+      "epoch": 0.46545454545454545,
+      "grad_norm": 3.125,
+      "learning_rate": 6.4480436217343366e-06,
+      "loss": 1.2059,
+      "step": 288
+    },
+    {
+      "epoch": 0.4670707070707071,
+      "grad_norm": 2.953125,
+      "learning_rate": 6.420979642281909e-06,
+      "loss": 0.8215,
+      "step": 289
+    },
+    {
+      "epoch": 0.4686868686868687,
+      "grad_norm": 2.84375,
+      "learning_rate": 6.393870296163616e-06,
+      "loss": 0.7734,
+      "step": 290
+    },
+    {
+      "epoch": 0.4703030303030303,
+      "grad_norm": 2.96875,
+      "learning_rate": 6.366716448881407e-06,
+      "loss": 0.7059,
+      "step": 291
+    },
+    {
+      "epoch": 0.4719191919191919,
+      "grad_norm": 3.203125,
+      "learning_rate": 6.339518967357985e-06,
+      "loss": 0.8906,
+      "step": 292
+    },
+    {
+      "epoch": 0.47353535353535353,
+      "grad_norm": 3.28125,
+      "learning_rate": 6.312278719909138e-06,
+      "loss": 0.9409,
+      "step": 293
+    },
+    {
+      "epoch": 0.47515151515151516,
+      "grad_norm": 3.34375,
+      "learning_rate": 6.284996576216014e-06,
+      "loss": 1.285,
+      "step": 294
+    },
+    {
+      "epoch": 0.4767676767676768,
+      "grad_norm": 3.078125,
+      "learning_rate": 6.257673407297352e-06,
+      "loss": 0.9044,
+      "step": 295
+    },
+    {
+      "epoch": 0.4783838383838384,
+      "grad_norm": 3.21875,
+      "learning_rate": 6.230310085481677e-06,
+      "loss": 0.7709,
+      "step": 296
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 3.25,
+      "learning_rate": 6.2029074843794445e-06,
+      "loss": 1.0275,
+      "step": 297
+    },
+    {
+      "epoch": 0.4816161616161616,
+      "grad_norm": 3.125,
+      "learning_rate": 6.175466478855161e-06,
+      "loss": 0.8711,
+      "step": 298
+    },
+    {
+      "epoch": 0.48323232323232324,
+      "grad_norm": 3.171875,
+      "learning_rate": 6.147987944999441e-06,
+      "loss": 1.0903,
+      "step": 299
+    },
+    {
+      "epoch": 0.48484848484848486,
+      "grad_norm": 2.953125,
+      "learning_rate": 6.1204727601010396e-06,
+      "loss": 0.768,
+      "step": 300
+    },
+    {
+      "epoch": 0.4864646464646465,
+      "grad_norm": 3.078125,
+      "learning_rate": 6.092921802618849e-06,
+      "loss": 0.6141,
+      "step": 301
+    },
+    {
+      "epoch": 0.48808080808080806,
+      "grad_norm": 2.625,
+      "learning_rate": 6.065335952153846e-06,
+      "loss": 0.7249,
+      "step": 302
+    },
+    {
+      "epoch": 0.4896969696969697,
+      "grad_norm": 3.125,
+      "learning_rate": 6.037716089421011e-06,
+      "loss": 0.6771,
+      "step": 303
+    },
+    {
+      "epoch": 0.4913131313131313,
+      "grad_norm": 3.71875,
+      "learning_rate": 6.010063096221215e-06,
+      "loss": 0.9863,
+      "step": 304
+    },
+    {
+      "epoch": 0.49292929292929294,
+      "grad_norm": 3.265625,
+      "learning_rate": 5.982377855413063e-06,
+      "loss": 0.784,
+      "step": 305
+    },
+    {
+      "epoch": 0.49454545454545457,
+      "grad_norm": 2.96875,
+      "learning_rate": 5.954661250884704e-06,
+      "loss": 0.801,
+      "step": 306
+    },
+    {
+      "epoch": 0.49616161616161614,
+      "grad_norm": 3.421875,
+      "learning_rate": 5.926914167525618e-06,
+      "loss": 0.9741,
+      "step": 307
+    },
+    {
+      "epoch": 0.49777777777777776,
+      "grad_norm": 3.25,
+      "learning_rate": 5.899137491198364e-06,
+      "loss": 0.6499,
+      "step": 308
+    },
+    {
+      "epoch": 0.4993939393939394,
+      "grad_norm": 2.875,
+      "learning_rate": 5.871332108710292e-06,
+      "loss": 0.7907,
+      "step": 309
+    },
+    {
+      "epoch": 0.501010101010101,
+      "grad_norm": 2.75,
+      "learning_rate": 5.843498907785236e-06,
+      "loss": 0.8293,
+      "step": 310
+    },
+    {
+      "epoch": 0.5026262626262626,
+      "grad_norm": 3.359375,
+      "learning_rate": 5.815638777035175e-06,
+      "loss": 1.2285,
+      "step": 311
+    },
+    {
+      "epoch": 0.5042424242424243,
+      "grad_norm": 3.34375,
+      "learning_rate": 5.78775260593185e-06,
+      "loss": 1.0661,
+      "step": 312
+    },
+    {
+      "epoch": 0.5058585858585859,
+      "grad_norm": 3.125,
+      "learning_rate": 5.759841284778379e-06,
+      "loss": 1.0879,
+      "step": 313
+    },
+    {
+      "epoch": 0.5074747474747475,
+      "grad_norm": 2.90625,
+      "learning_rate": 5.731905704680834e-06,
+      "loss": 0.6887,
+      "step": 314
+    },
+    {
+      "epoch": 0.509090909090909,
+      "grad_norm": 2.96875,
+      "learning_rate": 5.703946757519777e-06,
+      "loss": 0.7939,
+      "step": 315
+    },
+    {
+      "epoch": 0.5107070707070707,
+      "grad_norm": 2.890625,
+      "learning_rate": 5.6759653359218e-06,
+      "loss": 1.0402,
+      "step": 316
+    },
+    {
+      "epoch": 0.5123232323232323,
+      "grad_norm": 3.046875,
+      "learning_rate": 5.647962333231021e-06,
+      "loss": 0.6047,
+      "step": 317
+    },
+    {
+      "epoch": 0.5139393939393939,
+      "grad_norm": 3.09375,
+      "learning_rate": 5.6199386434805615e-06,
+      "loss": 0.7038,
+      "step": 318
+    },
+    {
+      "epoch": 0.5155555555555555,
+      "grad_norm": 3.0,
+      "learning_rate": 5.591895161364006e-06,
+      "loss": 0.8637,
+      "step": 319
+    },
+    {
+      "epoch": 0.5171717171717172,
+      "grad_norm": 3.03125,
+      "learning_rate": 5.563832782206835e-06,
+      "loss": 1.0575,
+      "step": 320
+    },
+    {
+      "epoch": 0.5187878787878788,
+      "grad_norm": 2.765625,
+      "learning_rate": 5.535752401937846e-06,
+      "loss": 0.4904,
+      "step": 321
+    },
+    {
+      "epoch": 0.5204040404040404,
+      "grad_norm": 3.046875,
+      "learning_rate": 5.507654917060541e-06,
+      "loss": 1.0068,
+      "step": 322
+    },
+    {
+      "epoch": 0.522020202020202,
+      "grad_norm": 2.90625,
+      "learning_rate": 5.4795412246245126e-06,
+      "loss": 0.942,
+      "step": 323
+    },
+    {
+      "epoch": 0.5236363636363637,
+      "grad_norm": 3.78125,
+      "learning_rate": 5.451412222196801e-06,
+      "loss": 1.142,
+      "step": 324
+    },
+    {
+      "epoch": 0.5252525252525253,
+      "grad_norm": 3.0625,
+      "learning_rate": 5.4232688078332384e-06,
+      "loss": 0.9834,
+      "step": 325
+    },
+    {
+      "epoch": 0.5268686868686868,
+      "grad_norm": 2.8125,
+      "learning_rate": 5.395111880049775e-06,
+      "loss": 0.5232,
+      "step": 326
+    },
+    {
+      "epoch": 0.5284848484848484,
+      "grad_norm": 3.15625,
+      "learning_rate": 5.366942337793798e-06,
+      "loss": 1.1519,
+      "step": 327
+    },
+    {
+      "epoch": 0.5301010101010101,
+      "grad_norm": 2.8125,
+      "learning_rate": 5.338761080415425e-06,
+      "loss": 0.7669,
+      "step": 328
+    },
+    {
+      "epoch": 0.5317171717171717,
+      "grad_norm": 2.796875,
+      "learning_rate": 5.310569007638795e-06,
+      "loss": 1.0665,
+      "step": 329
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 3.265625,
+      "learning_rate": 5.28236701953335e-06,
+      "loss": 0.9622,
+      "step": 330
+    },
+    {
+      "epoch": 0.534949494949495,
+      "grad_norm": 2.90625,
+      "learning_rate": 5.254156016485082e-06,
+      "loss": 0.7276,
+      "step": 331
+    },
+    {
+      "epoch": 0.5365656565656566,
+      "grad_norm": 2.953125,
+      "learning_rate": 5.225936899167803e-06,
+      "loss": 0.8417,
+      "step": 332
+    },
+    {
+      "epoch": 0.5381818181818182,
+      "grad_norm": 3.015625,
+      "learning_rate": 5.197710568514381e-06,
+      "loss": 1.0137,
+      "step": 333
+    },
+    {
+      "epoch": 0.5397979797979798,
+      "grad_norm": 3.109375,
+      "learning_rate": 5.169477925687981e-06,
+      "loss": 0.8616,
+      "step": 334
+    },
+    {
+      "epoch": 0.5414141414141415,
+      "grad_norm": 2.984375,
+      "learning_rate": 5.141239872053294e-06,
+      "loss": 0.6512,
+      "step": 335
+    },
+    {
+      "epoch": 0.5430303030303031,
+      "grad_norm": 2.890625,
+      "learning_rate": 5.112997309147753e-06,
+      "loss": 1.2423,
+      "step": 336
+    },
+    {
+      "epoch": 0.5446464646464646,
+      "grad_norm": 3.203125,
+      "learning_rate": 5.084751138652764e-06,
+      "loss": 0.7816,
+      "step": 337
+    },
+    {
+      "epoch": 0.5462626262626262,
+      "grad_norm": 3.28125,
+      "learning_rate": 5.0565022623649e-06,
+      "loss": 0.7545,
+      "step": 338
+    },
+    {
+      "epoch": 0.5478787878787879,
+      "grad_norm": 3.09375,
+      "learning_rate": 5.028251582167126e-06,
+      "loss": 0.7782,
+      "step": 339
+    },
+    {
+      "epoch": 0.5494949494949495,
+      "grad_norm": 2.765625,
+      "learning_rate": 5e-06,
+      "loss": 0.8367,
+      "step": 340
+    },
+    {
+      "epoch": 0.5511111111111111,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.971748417832876e-06,
+      "loss": 0.7385,
+      "step": 341
+    },
+    {
+      "epoch": 0.5527272727272727,
+      "grad_norm": 4.03125,
+      "learning_rate": 4.943497737635103e-06,
+      "loss": 1.0,
+      "step": 342
+    },
+    {
+      "epoch": 0.5543434343434344,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.915248861347239e-06,
+      "loss": 0.7811,
+      "step": 343
+    },
+    {
+      "epoch": 0.555959595959596,
+      "grad_norm": 3.375,
+      "learning_rate": 4.887002690852249e-06,
+      "loss": 0.5882,
+      "step": 344
+    },
+    {
+      "epoch": 0.5575757575757576,
+      "grad_norm": 3.0,
+      "learning_rate": 4.858760127946707e-06,
+      "loss": 0.6655,
+      "step": 345
+    },
+    {
+      "epoch": 0.5591919191919192,
+      "grad_norm": 3.203125,
+      "learning_rate": 4.830522074312019e-06,
+      "loss": 1.0647,
+      "step": 346
+    },
+    {
+      "epoch": 0.5608080808080808,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.80228943148562e-06,
+      "loss": 0.9217,
+      "step": 347
+    },
+    {
+      "epoch": 0.5624242424242424,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.774063100832199e-06,
+      "loss": 0.851,
+      "step": 348
+    },
+    {
+      "epoch": 0.564040404040404,
+      "grad_norm": 3.59375,
+      "learning_rate": 4.745843983514919e-06,
+      "loss": 0.6812,
+      "step": 349
+    },
+    {
+      "epoch": 0.5656565656565656,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.717632980466652e-06,
+      "loss": 0.6393,
+      "step": 350
+    },
+    {
+      "epoch": 0.5672727272727273,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.6894309923612055e-06,
+      "loss": 0.9291,
+      "step": 351
+    },
+    {
+      "epoch": 0.5688888888888889,
+      "grad_norm": 2.765625,
+      "learning_rate": 4.661238919584578e-06,
+      "loss": 0.7628,
+      "step": 352
+    },
+    {
+      "epoch": 0.5705050505050505,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.633057662206205e-06,
+      "loss": 1.0961,
+      "step": 353
+    },
+    {
+      "epoch": 0.5721212121212121,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.6048881199502265e-06,
+      "loss": 1.1471,
+      "step": 354
+    },
+    {
+      "epoch": 0.5737373737373738,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.576731192166762e-06,
+      "loss": 0.9697,
+      "step": 355
+    },
+    {
+      "epoch": 0.5753535353535354,
+      "grad_norm": 3.0,
+      "learning_rate": 4.548587777803198e-06,
+      "loss": 0.716,
+      "step": 356
+    },
+    {
+      "epoch": 0.576969696969697,
+      "grad_norm": 2.8125,
+      "learning_rate": 4.520458775375488e-06,
+      "loss": 0.8469,
+      "step": 357
+    },
+    {
+      "epoch": 0.5785858585858585,
+      "grad_norm": 3.109375,
+      "learning_rate": 4.49234508293946e-06,
+      "loss": 0.7453,
+      "step": 358
+    },
+    {
+      "epoch": 0.5802020202020202,
+      "grad_norm": 3.109375,
+      "learning_rate": 4.464247598062156e-06,
+      "loss": 0.7149,
+      "step": 359
+    },
+    {
+      "epoch": 0.5818181818181818,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.436167217793167e-06,
+      "loss": 0.6653,
+      "step": 360
+    },
+    {
+      "epoch": 0.5834343434343434,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.408104838635996e-06,
+      "loss": 0.7812,
+      "step": 361
+    },
+    {
+      "epoch": 0.585050505050505,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.38006135651944e-06,
+      "loss": 0.9034,
+      "step": 362
+    },
+    {
+      "epoch": 0.5866666666666667,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.3520376667689815e-06,
+      "loss": 0.7555,
+      "step": 363
+    },
+    {
+      "epoch": 0.5882828282828283,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.3240346640782014e-06,
+      "loss": 0.7781,
+      "step": 364
+    },
+    {
+      "epoch": 0.5898989898989899,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.2960532424802235e-06,
+      "loss": 0.9856,
+      "step": 365
+    },
+    {
+      "epoch": 0.5915151515151515,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.268094295319167e-06,
+      "loss": 1.1502,
+      "step": 366
+    },
+    {
+      "epoch": 0.5931313131313132,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.240158715221621e-06,
+      "loss": 1.2424,
+      "step": 367
+    },
+    {
+      "epoch": 0.5947474747474748,
+      "grad_norm": 3.0,
+      "learning_rate": 4.212247394068151e-06,
+      "loss": 0.848,
+      "step": 368
+    },
+    {
+      "epoch": 0.5963636363636363,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.184361222964827e-06,
+      "loss": 0.5807,
+      "step": 369
+    },
+    {
+      "epoch": 0.5963636363636363,
+      "eval_loss": 0.8503313064575195,
+      "eval_runtime": 33.7011,
+      "eval_samples_per_second": 32.64,
+      "eval_steps_per_second": 4.095,
+      "step": 369
+    },
+    {
+      "epoch": 0.597979797979798,
+      "grad_norm": 3.265625,
+      "learning_rate": 4.1565010922147644e-06,
+      "loss": 1.0351,
+      "step": 370
+    },
+    {
+      "epoch": 0.5995959595959596,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.1286678912897095e-06,
+      "loss": 0.7701,
+      "step": 371
+    },
+    {
+      "epoch": 0.6012121212121212,
+      "grad_norm": 2.703125,
+      "learning_rate": 4.100862508801639e-06,
+      "loss": 0.7962,
+      "step": 372
+    },
+    {
+      "epoch": 0.6028282828282828,
+      "grad_norm": 3.109375,
+      "learning_rate": 4.0730858324743845e-06,
+      "loss": 0.8057,
+      "step": 373
+    },
+    {
+      "epoch": 0.6044444444444445,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.045338749115299e-06,
+      "loss": 0.9921,
+      "step": 374
+    },
+    {
+      "epoch": 0.6060606060606061,
+      "grad_norm": 3.109375,
+      "learning_rate": 4.017622144586938e-06,
+      "loss": 0.9868,
+      "step": 375
+    },
+    {
+      "epoch": 0.6076767676767677,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.989936903778785e-06,
+      "loss": 0.882,
+      "step": 376
+    },
+    {
+      "epoch": 0.6092929292929293,
+      "grad_norm": 2.796875,
+      "learning_rate": 3.962283910578991e-06,
+      "loss": 0.7169,
+      "step": 377
+    },
+    {
+      "epoch": 0.610909090909091,
+      "grad_norm": 2.890625,
+      "learning_rate": 3.934664047846157e-06,
+      "loss": 0.8103,
+      "step": 378
+    },
+    {
+      "epoch": 0.6125252525252525,
+      "grad_norm": 2.953125,
+      "learning_rate": 3.907078197381153e-06,
+      "loss": 1.0984,
+      "step": 379
+    },
+    {
+      "epoch": 0.6141414141414141,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.879527239898962e-06,
+      "loss": 0.983,
+      "step": 380
+    },
+    {
+      "epoch": 0.6157575757575757,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.85201205500056e-06,
+      "loss": 0.8142,
+      "step": 381
+    },
+    {
+      "epoch": 0.6173737373737374,
+      "grad_norm": 2.640625,
+      "learning_rate": 3.8245335211448404e-06,
+      "loss": 0.5103,
+      "step": 382
+    },
+    {
+      "epoch": 0.618989898989899,
+      "grad_norm": 3.265625,
+      "learning_rate": 3.797092515620557e-06,
+      "loss": 0.7719,
+      "step": 383
+    },
+    {
+      "epoch": 0.6206060606060606,
+      "grad_norm": 2.953125,
+      "learning_rate": 3.769689914518326e-06,
+      "loss": 0.7184,
+      "step": 384
+    },
+    {
+      "epoch": 0.6222222222222222,
+      "grad_norm": 2.828125,
+      "learning_rate": 3.7423265927026473e-06,
+      "loss": 0.4874,
+      "step": 385
+    },
+    {
+      "epoch": 0.6238383838383839,
+      "grad_norm": 3.15625,
+      "learning_rate": 3.715003423783986e-06,
+      "loss": 0.8992,
+      "step": 386
+    },
+    {
+      "epoch": 0.6254545454545455,
+      "grad_norm": 2.953125,
+      "learning_rate": 3.6877212800908625e-06,
+      "loss": 0.9217,
+      "step": 387
+    },
+    {
+      "epoch": 0.6270707070707071,
+      "grad_norm": 4.0,
+      "learning_rate": 3.660481032642016e-06,
+      "loss": 0.8571,
+      "step": 388
+    },
+    {
+      "epoch": 0.6286868686868687,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.633283551118595e-06,
+      "loss": 0.8707,
+      "step": 389
+    },
+    {
+      "epoch": 0.6303030303030303,
+      "grad_norm": 3.59375,
+      "learning_rate": 3.6061297038363853e-06,
+      "loss": 1.032,
+      "step": 390
+    },
+    {
+      "epoch": 0.6319191919191919,
+      "grad_norm": 3.3125,
+      "learning_rate": 3.579020357718092e-06,
+      "loss": 0.9184,
+      "step": 391
+    },
+    {
+      "epoch": 0.6335353535353535,
+      "grad_norm": 2.9375,
+      "learning_rate": 3.5519563782656642e-06,
+      "loss": 0.9731,
+      "step": 392
+    },
+    {
+      "epoch": 0.6351515151515151,
+      "grad_norm": 2.9375,
+      "learning_rate": 3.524938629532656e-06,
+      "loss": 0.6837,
+      "step": 393
+    },
+    {
+      "epoch": 0.6367676767676768,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.497967974096647e-06,
+      "loss": 0.7254,
+      "step": 394
+    },
+    {
+      "epoch": 0.6383838383838384,
+      "grad_norm": 3.5,
+      "learning_rate": 3.4710452730316978e-06,
+      "loss": 0.8514,
+      "step": 395
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.4441713858808684e-06,
+      "loss": 0.8363,
+      "step": 396
+    },
+    {
+      "epoch": 0.6416161616161616,
+      "grad_norm": 3.28125,
+      "learning_rate": 3.4173471706287607e-06,
+      "loss": 0.7557,
+      "step": 397
+    },
+    {
+      "epoch": 0.6432323232323233,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.3905734836741415e-06,
+      "loss": 0.964,
+      "step": 398
+    },
+    {
+      "epoch": 0.6448484848484849,
+      "grad_norm": 3.328125,
+      "learning_rate": 3.3638511798025927e-06,
+      "loss": 0.8774,
+      "step": 399
+    },
+    {
+      "epoch": 0.6464646464646465,
+      "grad_norm": 3.1875,
+      "learning_rate": 3.3371811121592203e-06,
+      "loss": 0.7785,
+      "step": 400
+    },
+    {
+      "epoch": 0.648080808080808,
+      "grad_norm": 3.515625,
+      "learning_rate": 3.3105641322214238e-06,
+      "loss": 0.9186,
+      "step": 401
+    },
+    {
+      "epoch": 0.6496969696969697,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.2840010897717045e-06,
+      "loss": 1.0093,
+      "step": 402
+    },
+    {
+      "epoch": 0.6513131313131313,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.257492832870537e-06,
+      "loss": 1.0655,
+      "step": 403
+    },
+    {
+      "epoch": 0.6529292929292929,
+      "grad_norm": 2.953125,
+      "learning_rate": 3.2310402078292956e-06,
+      "loss": 1.0494,
+      "step": 404
+    },
+    {
+      "epoch": 0.6545454545454545,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.2046440591832323e-06,
+      "loss": 0.6991,
+      "step": 405
+    },
+    {
+      "epoch": 0.6561616161616162,
+      "grad_norm": 2.90625,
+      "learning_rate": 3.178305229664519e-06,
+      "loss": 0.8491,
+      "step": 406
+    },
+    {
+      "epoch": 0.6577777777777778,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.152024560175332e-06,
+      "loss": 0.9829,
+      "step": 407
+    },
+    {
+      "epoch": 0.6593939393939394,
+      "grad_norm": 2.765625,
+      "learning_rate": 3.125802889761016e-06,
+      "loss": 0.6532,
+      "step": 408
+    },
+    {
+      "epoch": 0.661010101010101,
+      "grad_norm": 2.890625,
+      "learning_rate": 3.099641055583289e-06,
+      "loss": 0.5892,
+      "step": 409
+    },
+    {
+      "epoch": 0.6626262626262627,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.073539892893519e-06,
+      "loss": 0.8826,
+      "step": 410
+    },
+    {
+      "epoch": 0.6642424242424242,
+      "grad_norm": 3.28125,
+      "learning_rate": 3.047500235006056e-06,
+      "loss": 0.7701,
+      "step": 411
+    },
+    {
+      "epoch": 0.6658585858585858,
+      "grad_norm": 3.34375,
+      "learning_rate": 3.021522913271627e-06,
+      "loss": 0.7495,
+      "step": 412
+    },
+    {
+      "epoch": 0.6674747474747474,
+      "grad_norm": 3.046875,
+      "learning_rate": 2.995608757050793e-06,
+      "loss": 0.961,
+      "step": 413
+    },
+    {
+      "epoch": 0.6690909090909091,
+      "grad_norm": 3.203125,
+      "learning_rate": 2.969758593687475e-06,
+      "loss": 0.7076,
+      "step": 414
+    },
+    {
+      "epoch": 0.6707070707070707,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.9439732484825323e-06,
+      "loss": 0.7592,
+      "step": 415
+    },
+    {
+      "epoch": 0.6723232323232323,
+      "grad_norm": 2.859375,
+      "learning_rate": 2.9182535446674244e-06,
+      "loss": 0.7731,
+      "step": 416
+    },
+    {
+      "epoch": 0.673939393939394,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.8926003033779194e-06,
+      "loss": 0.7137,
+      "step": 417
+    },
+    {
+      "epoch": 0.6755555555555556,
+      "grad_norm": 3.0625,
+      "learning_rate": 2.8670143436278757e-06,
+      "loss": 1.3734,
+      "step": 418
+    },
+    {
+      "epoch": 0.6771717171717172,
+      "grad_norm": 3.125,
+      "learning_rate": 2.8414964822831063e-06,
+      "loss": 0.8115,
+      "step": 419
+    },
+    {
+      "epoch": 0.6787878787878788,
+      "grad_norm": 2.984375,
+      "learning_rate": 2.8160475340352913e-06,
+      "loss": 0.6125,
+      "step": 420
+    },
+    {
+      "epoch": 0.6804040404040405,
+      "grad_norm": 3.078125,
+      "learning_rate": 2.790668311375962e-06,
+      "loss": 0.9947,
+      "step": 421
+    },
+    {
+      "epoch": 0.682020202020202,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.765359624570574e-06,
+      "loss": 0.7132,
+      "step": 422
+    },
+    {
+      "epoch": 0.6836363636363636,
+      "grad_norm": 3.21875,
+      "learning_rate": 2.7401222816326316e-06,
+      "loss": 0.8839,
+      "step": 423
+    },
+    {
+      "epoch": 0.6852525252525252,
+      "grad_norm": 2.828125,
+      "learning_rate": 2.714957088297886e-06,
+      "loss": 0.5796,
+      "step": 424
+    },
+    {
+      "epoch": 0.6868686868686869,
+      "grad_norm": 3.0,
+      "learning_rate": 2.6898648479986187e-06,
+      "loss": 0.76,
+      "step": 425
+    },
+    {
+      "epoch": 0.6884848484848485,
+      "grad_norm": 3.0,
+      "learning_rate": 2.664846361837997e-06,
+      "loss": 0.9098,
+      "step": 426
+    },
+    {
+      "epoch": 0.6901010101010101,
+      "grad_norm": 3.15625,
+      "learning_rate": 2.639902428564475e-06,
+      "loss": 0.8364,
+      "step": 427
+    },
+    {
+      "epoch": 0.6917171717171717,
+      "grad_norm": 3.0625,
+      "learning_rate": 2.6150338445463146e-06,
+      "loss": 0.7251,
+      "step": 428
+    },
+    {
+      "epoch": 0.6933333333333334,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.5902414037461544e-06,
+      "loss": 0.7137,
+      "step": 429
+    },
+    {
+      "epoch": 0.694949494949495,
+      "grad_norm": 3.1875,
+      "learning_rate": 2.565525897695651e-06,
+      "loss": 1.1286,
+      "step": 430
+    },
+    {
+      "epoch": 0.6965656565656566,
+      "grad_norm": 2.984375,
+      "learning_rate": 2.540888115470226e-06,
+      "loss": 0.8104,
+      "step": 431
+    },
+    {
+      "epoch": 0.6981818181818182,
+      "grad_norm": 3.1875,
+      "learning_rate": 2.51632884366386e-06,
+      "loss": 0.8526,
+      "step": 432
+    },
+    {
+      "epoch": 0.6997979797979798,
+      "grad_norm": 3.46875,
+      "learning_rate": 2.4918488663639824e-06,
+      "loss": 0.789,
+      "step": 433
+    },
+    {
+      "epoch": 0.7014141414141414,
+      "grad_norm": 3.234375,
+      "learning_rate": 2.4674489651264433e-06,
+      "loss": 0.8389,
+      "step": 434
+    },
+    {
+      "epoch": 0.703030303030303,
+      "grad_norm": 3.09375,
+      "learning_rate": 2.4431299189505563e-06,
+      "loss": 0.9738,
+      "step": 435
+    },
+    {
+      "epoch": 0.7046464646464646,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.418892504254231e-06,
+      "loss": 0.7604,
+      "step": 436
+    },
+    {
+      "epoch": 0.7062626262626263,
+      "grad_norm": 3.0625,
+      "learning_rate": 2.394737494849184e-06,
+      "loss": 0.6702,
+      "step": 437
+    },
+    {
+      "epoch": 0.7078787878787879,
+      "grad_norm": 3.359375,
+      "learning_rate": 2.3706656619162278e-06,
+      "loss": 0.9634,
+      "step": 438
+    },
+    {
+      "epoch": 0.7094949494949495,
+      "grad_norm": 3.0625,
+      "learning_rate": 2.34667777398066e-06,
+      "loss": 0.8533,
+      "step": 439
+    },
+    {
+      "epoch": 0.7111111111111111,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.322774596887726e-06,
+      "loss": 0.6953,
+      "step": 440
+    },
+    {
+      "epoch": 0.7127272727272728,
+      "grad_norm": 2.890625,
+      "learning_rate": 2.298956893778154e-06,
+      "loss": 0.7109,
+      "step": 441
+    },
+    {
+      "epoch": 0.7143434343434344,
+      "grad_norm": 3.234375,
+      "learning_rate": 2.275225425063813e-06,
+      "loss": 0.8229,
+      "step": 442
+    },
+    {
+      "epoch": 0.7159595959595959,
+      "grad_norm": 2.90625,
+      "learning_rate": 2.251580948403416e-06,
+      "loss": 0.8026,
+      "step": 443
+    },
+    {
+      "epoch": 0.7175757575757575,
+      "grad_norm": 2.890625,
+      "learning_rate": 2.2280242186783473e-06,
+      "loss": 0.6429,
+      "step": 444
+    },
+    {
+      "epoch": 0.7191919191919192,
+      "grad_norm": 2.96875,
+      "learning_rate": 2.204555987968541e-06,
+      "loss": 0.8036,
+      "step": 445
+    },
+    {
+      "epoch": 0.7208080808080808,
+      "grad_norm": 3.0,
+      "learning_rate": 2.1811770055284968e-06,
+      "loss": 0.6782,
+      "step": 446
+    },
+    {
+      "epoch": 0.7224242424242424,
+      "grad_norm": 3.0,
+      "learning_rate": 2.157888017763333e-06,
+      "loss": 1.0331,
+      "step": 447
+    },
+    {
+      "epoch": 0.724040404040404,
+      "grad_norm": 2.875,
+      "learning_rate": 2.134689768204975e-06,
+      "loss": 0.6081,
+      "step": 448
+    },
+    {
+      "epoch": 0.7256565656565657,
+      "grad_norm": 2.96875,
+      "learning_rate": 2.1115829974884097e-06,
+      "loss": 0.8604,
+      "step": 449
+    },
+    {
+      "epoch": 0.7272727272727273,
+      "grad_norm": 3.046875,
+      "learning_rate": 2.0885684433280336e-06,
+      "loss": 0.7877,
+      "step": 450
+    },
+    {
+      "epoch": 0.7288888888888889,
+      "grad_norm": 3.46875,
+      "learning_rate": 2.065646840494115e-06,
+      "loss": 0.8764,
+      "step": 451
+    },
+    {
+      "epoch": 0.7305050505050505,
+      "grad_norm": 3.21875,
+      "learning_rate": 2.042818920789326e-06,
+      "loss": 0.641,
+      "step": 452
+    },
+    {
+      "epoch": 0.7321212121212122,
+      "grad_norm": 3.125,
+      "learning_rate": 2.020085413025375e-06,
+      "loss": 0.7619,
+      "step": 453
+    },
+    {
+      "epoch": 0.7337373737373737,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9974470429997482e-06,
+      "loss": 1.0079,
+      "step": 454
+    },
+    {
+      "epoch": 0.7353535353535353,
+      "grad_norm": 3.078125,
+      "learning_rate": 1.974904533472532e-06,
+      "loss": 0.8312,
+      "step": 455
+    },
+    {
+      "epoch": 0.7369696969696969,
+      "grad_norm": 2.953125,
+      "learning_rate": 1.9524586041433393e-06,
+      "loss": 0.8404,
+      "step": 456
+    },
+    {
+      "epoch": 0.7385858585858586,
+      "grad_norm": 2.96875,
+      "learning_rate": 1.9301099716283293e-06,
+      "loss": 1.0085,
+      "step": 457
+    },
+    {
+      "epoch": 0.7402020202020202,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.907859349437336e-06,
+      "loss": 1.0982,
+      "step": 458
+    },
+    {
+      "epoch": 0.7418181818181818,
+      "grad_norm": 3.375,
+      "learning_rate": 1.8857074479510761e-06,
+      "loss": 0.7389,
+      "step": 459
+    },
+    {
+      "epoch": 0.7434343434343434,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.8636549743984815e-06,
+      "loss": 0.8347,
+      "step": 460
+    },
+    {
+      "epoch": 0.7450505050505051,
+      "grad_norm": 2.78125,
+      "learning_rate": 1.8417026328341158e-06,
+      "loss": 0.5843,
+      "step": 461
+    },
+    {
+      "epoch": 0.7466666666666667,
+      "grad_norm": 2.671875,
+      "learning_rate": 1.8198511241156902e-06,
+      "loss": 0.6667,
+      "step": 462
+    },
+    {
+      "epoch": 0.7482828282828283,
+      "grad_norm": 3.421875,
+      "learning_rate": 1.7981011458816988e-06,
+      "loss": 1.1586,
+      "step": 463
+    },
+    {
+      "epoch": 0.74989898989899,
+      "grad_norm": 2.78125,
+      "learning_rate": 1.776453392529139e-06,
+      "loss": 0.6758,
+      "step": 464
+    },
+    {
+      "epoch": 0.7515151515151515,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.7549085551913358e-06,
+      "loss": 0.7947,
+      "step": 465
+    },
+    {
+      "epoch": 0.7531313131313131,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.7334673217158976e-06,
+      "loss": 0.9363,
+      "step": 466
+    },
+    {
+      "epoch": 0.7547474747474747,
+      "grad_norm": 3.078125,
+      "learning_rate": 1.7121303766427266e-06,
+      "loss": 0.8628,
+      "step": 467
+    },
+    {
+      "epoch": 0.7563636363636363,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.6908984011821883e-06,
+      "loss": 0.7512,
+      "step": 468
+    },
+    {
+      "epoch": 0.757979797979798,
+      "grad_norm": 3.140625,
+      "learning_rate": 1.669772073193352e-06,
+      "loss": 0.9199,
+      "step": 469
+    },
+    {
+      "epoch": 0.7595959595959596,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.6487520671623469e-06,
+      "loss": 0.8583,
+      "step": 470
+    },
+    {
+      "epoch": 0.7612121212121212,
+      "grad_norm": 2.828125,
+      "learning_rate": 1.6278390541808364e-06,
+      "loss": 0.9503,
+      "step": 471
+    },
+    {
+      "epoch": 0.7628282828282829,
+      "grad_norm": 3.3125,
+      "learning_rate": 1.6070337019245896e-06,
+      "loss": 0.8231,
+      "step": 472
+    },
+    {
+      "epoch": 0.7644444444444445,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.5863366746321578e-06,
+      "loss": 0.9843,
+      "step": 473
+    },
+    {
+      "epoch": 0.7660606060606061,
+      "grad_norm": 3.421875,
+      "learning_rate": 1.5657486330836786e-06,
+      "loss": 1.0286,
+      "step": 474
+    },
+    {
+      "epoch": 0.7676767676767676,
+      "grad_norm": 3.078125,
+      "learning_rate": 1.5452702345797738e-06,
+      "loss": 1.0374,
+      "step": 475
+    },
+    {
+      "epoch": 0.7692929292929293,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.5249021329205638e-06,
+      "loss": 0.9008,
+      "step": 476
+    },
+    {
+      "epoch": 0.7709090909090909,
+      "grad_norm": 2.890625,
+      "learning_rate": 1.5046449783847965e-06,
+      "loss": 0.6363,
+      "step": 477
+    },
+    {
+      "epoch": 0.7725252525252525,
+      "grad_norm": 2.859375,
+      "learning_rate": 1.4844994177090871e-06,
+      "loss": 0.6327,
+      "step": 478
+    },
+    {
+      "epoch": 0.7741414141414141,
+      "grad_norm": 3.21875,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 1.0177,
+      "step": 479
+    },
+    {
+      "epoch": 0.7757575757575758,
+      "grad_norm": 3.34375,
+      "learning_rate": 1.4445456470498392e-06,
+      "loss": 0.7853,
+      "step": 480
+    },
+    {
+      "epoch": 0.7773737373737374,
+      "grad_norm": 3.125,
+      "learning_rate": 1.4247387126435957e-06,
+      "loss": 0.6677,
+      "step": 481
+    },
+    {
+      "epoch": 0.778989898989899,
+      "grad_norm": 2.96875,
+      "learning_rate": 1.4050459232112652e-06,
+      "loss": 1.006,
+      "step": 482
+    },
+    {
+      "epoch": 0.7806060606060606,
+      "grad_norm": 3.046875,
+      "learning_rate": 1.3854679074713557e-06,
+      "loss": 0.8052,
+      "step": 483
+    },
+    {
+      "epoch": 0.7822222222222223,
+      "grad_norm": 3.046875,
+      "learning_rate": 1.3660052904780707e-06,
+      "loss": 0.8768,
+      "step": 484
+    },
+    {
+      "epoch": 0.7838383838383839,
+      "grad_norm": 3.171875,
+      "learning_rate": 1.3466586936013548e-06,
+      "loss": 1.0445,
+      "step": 485
+    },
+    {
+      "epoch": 0.7854545454545454,
+      "grad_norm": 2.9375,
+      "learning_rate": 1.3274287345070564e-06,
+      "loss": 0.8651,
+      "step": 486
+    },
+    {
+      "epoch": 0.787070707070707,
+      "grad_norm": 2.734375,
+      "learning_rate": 1.3083160271372092e-06,
+      "loss": 0.6247,
+      "step": 487
+    },
+    {
+      "epoch": 0.7886868686868687,
+      "grad_norm": 3.046875,
+      "learning_rate": 1.2893211816904243e-06,
+      "loss": 0.7506,
+      "step": 488
+    },
+    {
+      "epoch": 0.7903030303030303,
+      "grad_norm": 3.140625,
+      "learning_rate": 1.2704448046024192e-06,
+      "loss": 0.6766,
+      "step": 489
+    },
+    {
+      "epoch": 0.7919191919191919,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.2516874985266508e-06,
+      "loss": 0.8885,
+      "step": 490
+    },
+    {
+      "epoch": 0.7935353535353535,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.2330498623150722e-06,
+      "loss": 0.8875,
+      "step": 491
+    },
+    {
+      "epoch": 0.7951515151515152,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.2145324909990202e-06,
+      "loss": 0.7419,
+      "step": 492
+    },
+    {
+      "epoch": 0.7951515151515152,
+      "eval_loss": 0.8412812352180481,
+      "eval_runtime": 33.6709,
+      "eval_samples_per_second": 32.669,
+      "eval_steps_per_second": 4.098,
+      "step": 492
+    },
+    {
+      "epoch": 0.7967676767676768,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.1961359757702151e-06,
+      "loss": 0.7417,
+      "step": 493
+    },
+    {
+      "epoch": 0.7983838383838384,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.1778609039618804e-06,
+      "loss": 0.925,
+      "step": 494
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.953125,
+      "learning_rate": 1.1597078590300054e-06,
+      "loss": 1.1692,
+      "step": 495
+    },
+    {
+      "epoch": 0.8016161616161617,
+      "grad_norm": 3.375,
+      "learning_rate": 1.1416774205347015e-06,
+      "loss": 1.0403,
+      "step": 496
+    },
+    {
+      "epoch": 0.8032323232323232,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.1237701641217097e-06,
+      "loss": 0.7989,
+      "step": 497
+    },
+    {
+      "epoch": 0.8048484848484848,
+      "grad_norm": 3.140625,
+      "learning_rate": 1.1059866615040205e-06,
+      "loss": 0.8693,
+      "step": 498
+    },
+    {
+      "epoch": 0.8064646464646464,
+      "grad_norm": 3.03125,
+      "learning_rate": 1.0883274804436155e-06,
+      "loss": 0.9305,
+      "step": 499
+    },
+    {
+      "epoch": 0.8080808080808081,
+      "grad_norm": 3.09375,
+      "learning_rate": 1.0707931847333487e-06,
+      "loss": 0.8911,
+      "step": 500
+    },
+    {
+      "epoch": 0.8096969696969697,
+      "grad_norm": 3.09375,
+      "learning_rate": 1.053384334178944e-06,
+      "loss": 0.8849,
+      "step": 501
+    },
+    {
+      "epoch": 0.8113131313131313,
+      "grad_norm": 3.046875,
+      "learning_rate": 1.036101484581117e-06,
+      "loss": 0.6888,
+      "step": 502
+    },
+    {
+      "epoch": 0.812929292929293,
+      "grad_norm": 2.875,
+      "learning_rate": 1.0189451877178386e-06,
+      "loss": 0.585,
+      "step": 503
+    },
+    {
+      "epoch": 0.8145454545454546,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.0019159913267156e-06,
+      "loss": 0.7421,
+      "step": 504
+    },
+    {
+      "epoch": 0.8161616161616162,
+      "grad_norm": 3.09375,
+      "learning_rate": 9.850144390875022e-07,
+      "loss": 1.0121,
+      "step": 505
+    },
+    {
+      "epoch": 0.8177777777777778,
+      "grad_norm": 3.75,
+      "learning_rate": 9.68241070604743e-07,
+      "loss": 0.8492,
+      "step": 506
+    },
+    {
+      "epoch": 0.8193939393939393,
+      "grad_norm": 3.109375,
+      "learning_rate": 9.51596421390547e-07,
+      "loss": 1.4227,
+      "step": 507
+    },
+    {
+      "epoch": 0.821010101010101,
+      "grad_norm": 3.15625,
+      "learning_rate": 9.350810228474855e-07,
+      "loss": 0.9121,
+      "step": 508
+    },
+    {
+      "epoch": 0.8226262626262626,
+      "grad_norm": 3.15625,
+      "learning_rate": 9.186954022516343e-07,
+      "loss": 0.9289,
+      "step": 509
+    },
+    {
+      "epoch": 0.8242424242424242,
+      "grad_norm": 3.03125,
+      "learning_rate": 9.024400827357344e-07,
+      "loss": 1.0219,
+      "step": 510
+    },
+    {
+      "epoch": 0.8258585858585858,
+      "grad_norm": 3.453125,
+      "learning_rate": 8.863155832724895e-07,
+      "loss": 0.8358,
+      "step": 511
+    },
+    {
+      "epoch": 0.8274747474747475,
+      "grad_norm": 3.1875,
+      "learning_rate": 8.703224186580012e-07,
+      "loss": 0.813,
+      "step": 512
+    },
+    {
+      "epoch": 0.8290909090909091,
+      "grad_norm": 2.953125,
+      "learning_rate": 8.544610994953317e-07,
+      "loss": 0.5463,
+      "step": 513
+    },
+    {
+      "epoch": 0.8307070707070707,
+      "grad_norm": 3.25,
+      "learning_rate": 8.387321321781977e-07,
+      "loss": 0.8594,
+      "step": 514
+    },
+    {
+      "epoch": 0.8323232323232324,
+      "grad_norm": 2.96875,
+      "learning_rate": 8.23136018874815e-07,
+      "loss": 0.6723,
+      "step": 515
+    },
+    {
+      "epoch": 0.833939393939394,
+      "grad_norm": 2.65625,
+      "learning_rate": 8.07673257511849e-07,
+      "loss": 0.7454,
+      "step": 516
+    },
+    {
+      "epoch": 0.8355555555555556,
+      "grad_norm": 2.921875,
+      "learning_rate": 7.923443417585324e-07,
+      "loss": 0.8643,
+      "step": 517
+    },
+    {
+      "epoch": 0.8371717171717171,
+      "grad_norm": 3.109375,
+      "learning_rate": 7.771497610108981e-07,
+      "loss": 0.7421,
+      "step": 518
+    },
+    {
+      "epoch": 0.8387878787878787,
+      "grad_norm": 2.9375,
+      "learning_rate": 7.620900003761561e-07,
+      "loss": 1.2427,
+      "step": 519
+    },
+    {
+      "epoch": 0.8404040404040404,
+      "grad_norm": 2.8125,
+      "learning_rate": 7.471655406572003e-07,
+      "loss": 1.0072,
+      "step": 520
+    },
+    {
+      "epoch": 0.842020202020202,
+      "grad_norm": 3.015625,
+      "learning_rate": 7.323768583372675e-07,
+      "loss": 0.651,
+      "step": 521
+    },
+    {
+      "epoch": 0.8436363636363636,
+      "grad_norm": 3.65625,
+      "learning_rate": 7.177244255647209e-07,
+      "loss": 0.8858,
+      "step": 522
+    },
+    {
+      "epoch": 0.8452525252525253,
+      "grad_norm": 3.15625,
+      "learning_rate": 7.032087101379703e-07,
+      "loss": 0.8872,
+      "step": 523
+    },
+    {
+      "epoch": 0.8468686868686869,
+      "grad_norm": 3.09375,
+      "learning_rate": 6.888301754905469e-07,
+      "loss": 0.9091,
+      "step": 524
+    },
+    {
+      "epoch": 0.8484848484848485,
+      "grad_norm": 2.84375,
+      "learning_rate": 6.745892806763015e-07,
+      "loss": 0.487,
+      "step": 525
+    },
+    {
+      "epoch": 0.8501010101010101,
+      "grad_norm": 2.890625,
+      "learning_rate": 6.604864803547511e-07,
+      "loss": 0.5723,
+      "step": 526
+    },
+    {
+      "epoch": 0.8517171717171718,
+      "grad_norm": 2.8125,
+      "learning_rate": 6.465222247765618e-07,
+      "loss": 0.6302,
+      "step": 527
+    },
+    {
+      "epoch": 0.8533333333333334,
+      "grad_norm": 3.390625,
+      "learning_rate": 6.326969597691724e-07,
+      "loss": 0.9424,
+      "step": 528
+    },
+    {
+      "epoch": 0.8549494949494949,
+      "grad_norm": 3.65625,
+      "learning_rate": 6.190111267225651e-07,
+      "loss": 1.0664,
+      "step": 529
+    },
+    {
+      "epoch": 0.8565656565656565,
+      "grad_norm": 3.171875,
+      "learning_rate": 6.054651625751717e-07,
+      "loss": 0.7773,
+      "step": 530
+    },
+    {
+      "epoch": 0.8581818181818182,
+      "grad_norm": 3.015625,
+      "learning_rate": 5.920594997999202e-07,
+      "loss": 0.9076,
+      "step": 531
+    },
+    {
+      "epoch": 0.8597979797979798,
+      "grad_norm": 3.15625,
+      "learning_rate": 5.787945663904332e-07,
+      "loss": 0.7588,
+      "step": 532
+    },
+    {
+      "epoch": 0.8614141414141414,
+      "grad_norm": 2.921875,
+      "learning_rate": 5.65670785847362e-07,
+      "loss": 0.8204,
+      "step": 533
+    },
+    {
+      "epoch": 0.863030303030303,
+      "grad_norm": 3.125,
+      "learning_rate": 5.526885771648599e-07,
+      "loss": 0.7045,
+      "step": 534
+    },
+    {
+      "epoch": 0.8646464646464647,
+      "grad_norm": 2.796875,
+      "learning_rate": 5.398483548172162e-07,
+      "loss": 0.9205,
+      "step": 535
+    },
+    {
+      "epoch": 0.8662626262626263,
+      "grad_norm": 3.015625,
+      "learning_rate": 5.271505287456153e-07,
+      "loss": 0.6762,
+      "step": 536
+    },
+    {
+      "epoch": 0.8678787878787879,
+      "grad_norm": 2.90625,
+      "learning_rate": 5.145955043450484e-07,
+      "loss": 0.6195,
+      "step": 537
+    },
+    {
+      "epoch": 0.8694949494949495,
+      "grad_norm": 3.015625,
+      "learning_rate": 5.021836824513759e-07,
+      "loss": 1.0143,
+      "step": 538
+    },
+    {
+      "epoch": 0.8711111111111111,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.899154593285294e-07,
+      "loss": 0.7606,
+      "step": 539
+    },
+    {
+      "epoch": 0.8727272727272727,
+      "grad_norm": 3.125,
+      "learning_rate": 4.777912266558532e-07,
+      "loss": 0.8525,
+      "step": 540
+    },
+    {
+      "epoch": 0.8743434343434343,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.658113715156098e-07,
+      "loss": 1.0104,
+      "step": 541
+    },
+    {
+      "epoch": 0.8759595959595959,
+      "grad_norm": 3.0,
+      "learning_rate": 4.5397627638061604e-07,
+      "loss": 0.7297,
+      "step": 542
+    },
+    {
+      "epoch": 0.8775757575757576,
+      "grad_norm": 2.703125,
+      "learning_rate": 4.422863191020299e-07,
+      "loss": 1.0049,
+      "step": 543
+    },
+    {
+      "epoch": 0.8791919191919192,
+      "grad_norm": 2.890625,
+      "learning_rate": 4.307418728972934e-07,
+      "loss": 0.7605,
+      "step": 544
+    },
+    {
+      "epoch": 0.8808080808080808,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.193433063382135e-07,
+      "loss": 0.9836,
+      "step": 545
+    },
+    {
+      "epoch": 0.8824242424242424,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.080909833391944e-07,
+      "loss": 0.9216,
+      "step": 546
+    },
+    {
+      "epoch": 0.8840404040404041,
+      "grad_norm": 2.796875,
+      "learning_rate": 3.9698526314562114e-07,
+      "loss": 0.8258,
+      "step": 547
+    },
+    {
+      "epoch": 0.8856565656565657,
+      "grad_norm": 2.828125,
+      "learning_rate": 3.8602650032238675e-07,
+      "loss": 0.9187,
+      "step": 548
+    },
+    {
+      "epoch": 0.8872727272727273,
+      "grad_norm": 3.125,
+      "learning_rate": 3.752150447425773e-07,
+      "loss": 0.8915,
+      "step": 549
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 2.71875,
+      "learning_rate": 3.6455124157629805e-07,
+      "loss": 0.7028,
+      "step": 550
+    },
+    {
+      "epoch": 0.8905050505050505,
+      "grad_norm": 3.15625,
+      "learning_rate": 3.5403543127965514e-07,
+      "loss": 0.7958,
+      "step": 551
+    },
+    {
+      "epoch": 0.8921212121212121,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.436679495838835e-07,
+      "loss": 1.073,
+      "step": 552
+    },
+    {
+      "epoch": 0.8937373737373737,
+      "grad_norm": 2.921875,
+      "learning_rate": 3.334491274846324e-07,
+      "loss": 0.9868,
+      "step": 553
+    },
+    {
+      "epoch": 0.8953535353535353,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.233792912313943e-07,
+      "loss": 0.8007,
+      "step": 554
+    },
+    {
+      "epoch": 0.896969696969697,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.134587623170909e-07,
+      "loss": 0.8423,
+      "step": 555
+    },
+    {
+      "epoch": 0.8985858585858586,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.0368785746780925e-07,
+      "loss": 0.7721,
+      "step": 556
+    },
+    {
+      "epoch": 0.9002020202020202,
+      "grad_norm": 3.09375,
+      "learning_rate": 2.940668886326864e-07,
+      "loss": 0.862,
+      "step": 557
+    },
+    {
+      "epoch": 0.9018181818181819,
+      "grad_norm": 2.859375,
+      "learning_rate": 2.8459616297395464e-07,
+      "loss": 0.7686,
+      "step": 558
+    },
+    {
+      "epoch": 0.9034343434343435,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.7527598285713387e-07,
+      "loss": 0.925,
+      "step": 559
+    },
+    {
+      "epoch": 0.9050505050505051,
+      "grad_norm": 2.9375,
+      "learning_rate": 2.6610664584137413e-07,
+      "loss": 0.5698,
+      "step": 560
+    },
+    {
+      "epoch": 0.9066666666666666,
+      "grad_norm": 6.875,
+      "learning_rate": 2.570884446699612e-07,
+      "loss": 0.8914,
+      "step": 561
+    },
+    {
+      "epoch": 0.9082828282828282,
+      "grad_norm": 3.015625,
+      "learning_rate": 2.482216672609677e-07,
+      "loss": 1.0094,
+      "step": 562
+    },
+    {
+      "epoch": 0.9098989898989899,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.3950659669806033e-07,
+      "loss": 0.8511,
+      "step": 563
+    },
+    {
+      "epoch": 0.9115151515151515,
+      "grad_norm": 3.0,
+      "learning_rate": 2.3094351122146307e-07,
+      "loss": 0.7152,
+      "step": 564
+    },
+    {
+      "epoch": 0.9131313131313131,
+      "grad_norm": 3.125,
+      "learning_rate": 2.2253268421907503e-07,
+      "loss": 0.7391,
+      "step": 565
+    },
+    {
+      "epoch": 0.9147474747474748,
+      "grad_norm": 2.9375,
+      "learning_rate": 2.142743842177386e-07,
+      "loss": 1.0925,
+      "step": 566
+    },
+    {
+      "epoch": 0.9163636363636364,
+      "grad_norm": 3.203125,
+      "learning_rate": 2.0616887487467096e-07,
+      "loss": 1.0317,
+      "step": 567
+    },
+    {
+      "epoch": 0.917979797979798,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.98216414969043e-07,
+      "loss": 0.7862,
+      "step": 568
+    },
+    {
+      "epoch": 0.9195959595959596,
+      "grad_norm": 2.75,
+      "learning_rate": 1.9041725839371805e-07,
+      "loss": 0.8115,
+      "step": 569
+    },
+    {
+      "epoch": 0.9212121212121213,
+      "grad_norm": 2.96875,
+      "learning_rate": 1.8277165414714858e-07,
+      "loss": 0.8351,
+      "step": 570
+    },
+    {
+      "epoch": 0.9228282828282828,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.752798463254235e-07,
+      "loss": 0.7064,
+      "step": 571
+    },
+    {
+      "epoch": 0.9244444444444444,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.6794207411447548e-07,
+      "loss": 0.9278,
+      "step": 572
+    },
+    {
+      "epoch": 0.926060606060606,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.6075857178244613e-07,
+      "loss": 0.9863,
+      "step": 573
+    },
+    {
+      "epoch": 0.9276767676767677,
+      "grad_norm": 2.765625,
+      "learning_rate": 1.5372956867220678e-07,
+      "loss": 0.8924,
+      "step": 574
+    },
+    {
+      "epoch": 0.9292929292929293,
+      "grad_norm": 2.8125,
+      "learning_rate": 1.4685528919403446e-07,
+      "loss": 0.7934,
+      "step": 575
+    },
+    {
+      "epoch": 0.9309090909090909,
+      "grad_norm": 3.0,
+      "learning_rate": 1.4013595281844872e-07,
+      "loss": 0.8043,
+      "step": 576
+    },
+    {
+      "epoch": 0.9325252525252525,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.335717740692033e-07,
+      "loss": 0.6211,
+      "step": 577
+    },
+    {
+      "epoch": 0.9341414141414142,
+      "grad_norm": 3.59375,
+      "learning_rate": 1.2716296251644e-07,
+      "loss": 0.8752,
+      "step": 578
+    },
+    {
+      "epoch": 0.9357575757575758,
+      "grad_norm": 2.953125,
+      "learning_rate": 1.2090972276999513e-07,
+      "loss": 1.1873,
+      "step": 579
+    },
+    {
+      "epoch": 0.9373737373737374,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.1481225447286803e-07,
+      "loss": 0.966,
+      "step": 580
+    },
+    {
+      "epoch": 0.938989898989899,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.0887075229484789e-07,
+      "loss": 1.1486,
+      "step": 581
+    },
+    {
+      "epoch": 0.9406060606060606,
+      "grad_norm": 3.125,
+      "learning_rate": 1.0308540592629756e-07,
+      "loss": 0.974,
+      "step": 582
+    },
+    {
+      "epoch": 0.9422222222222222,
+      "grad_norm": 3.109375,
+      "learning_rate": 9.745640007209844e-08,
+      "loss": 0.7665,
+      "step": 583
+    },
+    {
+      "epoch": 0.9438383838383838,
+      "grad_norm": 2.734375,
+      "learning_rate": 9.198391444575072e-08,
+      "loss": 0.9478,
+      "step": 584
+    },
+    {
+      "epoch": 0.9454545454545454,
+      "grad_norm": 2.8125,
+      "learning_rate": 8.666812376364187e-08,
+      "loss": 0.6932,
+      "step": 585
+    },
+    {
+      "epoch": 0.9470707070707071,
+      "grad_norm": 2.890625,
+      "learning_rate": 8.150919773946165e-08,
+      "loss": 0.9752,
+      "step": 586
+    },
+    {
+      "epoch": 0.9486868686868687,
+      "grad_norm": 3.109375,
+      "learning_rate": 7.650730107878812e-08,
+      "loss": 0.8468,
+      "step": 587
+    },
+    {
+      "epoch": 0.9503030303030303,
+      "grad_norm": 3.515625,
+      "learning_rate": 7.166259347382854e-08,
+      "loss": 0.8219,
+      "step": 588
+    },
+    {
+      "epoch": 0.9519191919191919,
+      "grad_norm": 3.40625,
+      "learning_rate": 6.697522959831837e-08,
+      "loss": 0.8738,
+      "step": 589
+    },
+    {
+      "epoch": 0.9535353535353536,
+      "grad_norm": 3.09375,
+      "learning_rate": 6.244535910258697e-08,
+      "loss": 0.7416,
+      "step": 590
+    },
+    {
+      "epoch": 0.9551515151515152,
+      "grad_norm": 3.234375,
+      "learning_rate": 5.8073126608778064e-08,
+      "loss": 0.7994,
+      "step": 591
+    },
+    {
+      "epoch": 0.9567676767676768,
+      "grad_norm": 2.84375,
+      "learning_rate": 5.3858671706230605e-08,
+      "loss": 0.6755,
+      "step": 592
+    },
+    {
+      "epoch": 0.9583838383838383,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.98021289470263e-08,
+      "loss": 0.7478,
+      "step": 593
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 3.125,
+      "learning_rate": 4.590362784169022e-08,
+      "loss": 0.8433,
+      "step": 594
+    },
+    {
+      "epoch": 0.9616161616161616,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.2163292855056936e-08,
+      "loss": 0.7783,
+      "step": 595
+    },
+    {
+      "epoch": 0.9632323232323232,
+      "grad_norm": 2.890625,
+      "learning_rate": 3.858124340229863e-08,
+      "loss": 0.6822,
+      "step": 596
+    },
+    {
+      "epoch": 0.9648484848484848,
+      "grad_norm": 2.90625,
+      "learning_rate": 3.515759384510986e-08,
+      "loss": 0.8468,
+      "step": 597
+    },
+    {
+      "epoch": 0.9664646464646465,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.1892453488058803e-08,
+      "loss": 0.9057,
+      "step": 598
+    },
+    {
+      "epoch": 0.9680808080808081,
+      "grad_norm": 3.046875,
+      "learning_rate": 2.87859265750956e-08,
+      "loss": 0.8434,
+      "step": 599
+    },
+    {
+      "epoch": 0.9696969696969697,
+      "grad_norm": 2.90625,
+      "learning_rate": 2.5838112286226123e-08,
+      "loss": 0.8441,
+      "step": 600
+    },
+    {
+      "epoch": 0.9713131313131314,
+      "grad_norm": 3.015625,
+      "learning_rate": 2.304910473434341e-08,
+      "loss": 0.9293,
+      "step": 601
+    },
+    {
+      "epoch": 0.972929292929293,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.0418992962224495e-08,
+      "loss": 0.9303,
+      "step": 602
+    },
+    {
+      "epoch": 0.9745454545454545,
+      "grad_norm": 2.90625,
+      "learning_rate": 1.7947860939688255e-08,
+      "loss": 0.7841,
+      "step": 603
+    },
+    {
+      "epoch": 0.9761616161616161,
+      "grad_norm": 3.140625,
+      "learning_rate": 1.563578756091144e-08,
+      "loss": 0.9497,
+      "step": 604
+    },
+    {
+      "epoch": 0.9777777777777777,
+      "grad_norm": 3.09375,
+      "learning_rate": 1.3482846641914572e-08,
+      "loss": 0.9669,
+      "step": 605
+    },
+    {
+      "epoch": 0.9793939393939394,
+      "grad_norm": 3.359375,
+      "learning_rate": 1.1489106918200487e-08,
+      "loss": 0.7065,
+      "step": 606
+    },
+    {
+      "epoch": 0.981010101010101,
+      "grad_norm": 3.28125,
+      "learning_rate": 9.654632042562229e-09,
+      "loss": 0.8554,
+      "step": 607
+    },
+    {
+      "epoch": 0.9826262626262626,
+      "grad_norm": 3.28125,
+      "learning_rate": 7.979480583052423e-09,
+      "loss": 1.3692,
+      "step": 608
+    },
+    {
+      "epoch": 0.9842424242424243,
+      "grad_norm": 2.875,
+      "learning_rate": 6.4637060211092395e-09,
+      "loss": 0.602,
+      "step": 609
+    },
+    {
+      "epoch": 0.9858585858585859,
+      "grad_norm": 3.15625,
+      "learning_rate": 5.107356749853298e-09,
+      "loss": 0.882,
+      "step": 610
+    },
+    {
+      "epoch": 0.9874747474747475,
+      "grad_norm": 3.109375,
+      "learning_rate": 3.910476072539471e-09,
+      "loss": 0.7065,
+      "step": 611
+    },
+    {
+      "epoch": 0.9890909090909091,
+      "grad_norm": 2.78125,
+      "learning_rate": 2.8731022011757593e-09,
+      "loss": 0.7047,
+      "step": 612
+    },
+    {
+      "epoch": 0.9907070707070708,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.9952682553042722e-09,
+      "loss": 0.9843,
+      "step": 613
+    },
+    {
+      "epoch": 0.9923232323232323,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.2770022609409628e-09,
+      "loss": 0.8378,
+      "step": 614
+    },
+    {
+      "epoch": 0.9939393939393939,
+      "grad_norm": 3.125,
+      "learning_rate": 7.18327149683562e-10,
+      "loss": 0.9967,
+      "step": 615
+    },
+    {
+      "epoch": 0.9939393939393939,
+      "eval_loss": 0.840579628944397,
+      "eval_runtime": 33.6723,
+      "eval_samples_per_second": 32.668,
+      "eval_steps_per_second": 4.098,
+      "step": 615
+    },
+    {
+      "epoch": 0.9955555555555555,
+      "grad_norm": 2.875,
+      "learning_rate": 3.1926075797827914e-10,
+      "loss": 1.1188,
+      "step": 616
+    },
+    {
+      "epoch": 0.9971717171717172,
+      "grad_norm": 2.78125,
+      "learning_rate": 7.9815826551366e-11,
+      "loss": 0.8099,
+      "step": 617
+    },
+    {
+      "epoch": 0.9987878787878788,
+      "grad_norm": 3.03125,
+      "learning_rate": 0.0,
+      "loss": 0.7638,
+      "step": 618
+    },
+    {
+      "epoch": 1.0016161616161616,
+      "grad_norm": 3.140625,
+      "learning_rate": 5.8574189300346215e-06,
+      "loss": 1.0697,
+      "step": 619
+    },
+    {
+      "epoch": 1.0032323232323233,
+      "grad_norm": 3.109375,
+      "learning_rate": 5.843498907785236e-06,
+      "loss": 0.7671,
+      "step": 620
+    },
+    {
+      "epoch": 1.0048484848484849,
+      "grad_norm": 2.859375,
+      "learning_rate": 5.829572153066166e-06,
+      "loss": 0.7977,
+      "step": 621
+    },
+    {
+      "epoch": 1.0064646464646465,
+      "grad_norm": 2.9375,
+      "learning_rate": 5.815638777035175e-06,
+      "loss": 0.6885,
+      "step": 622
+    },
+    {
+      "epoch": 1.0080808080808081,
+      "grad_norm": 3.03125,
+      "learning_rate": 5.801698890902877e-06,
+      "loss": 0.8538,
+      "step": 623
+    },
+    {
+      "epoch": 1.0096969696969698,
+      "grad_norm": 2.71875,
+      "learning_rate": 5.78775260593185e-06,
+      "loss": 1.0268,
+      "step": 624
+    },
+    {
+      "epoch": 1.0113131313131314,
+      "grad_norm": 2.96875,
+      "learning_rate": 5.77380003343574e-06,
+      "loss": 1.0371,
+      "step": 625
+    },
+    {
+      "epoch": 1.012929292929293,
+      "grad_norm": 3.015625,
+      "learning_rate": 5.759841284778379e-06,
+      "loss": 0.816,
+      "step": 626
+    },
+    {
+      "epoch": 1.0145454545454546,
+      "grad_norm": 2.953125,
+      "learning_rate": 5.7458764713729e-06,
+      "loss": 0.5553,
+      "step": 627
+    },
+    {
+      "epoch": 1.0161616161616163,
+      "grad_norm": 3.5,
+      "learning_rate": 5.731905704680834e-06,
+      "loss": 0.7933,
+      "step": 628
+    },
+    {
+      "epoch": 1.0177777777777777,
+      "grad_norm": 3.296875,
+      "learning_rate": 5.717929096211232e-06,
+      "loss": 0.6598,
+      "step": 629
+    },
+    {
+      "epoch": 1.0193939393939393,
+      "grad_norm": 2.875,
+      "learning_rate": 5.703946757519777e-06,
+      "loss": 0.7425,
+      "step": 630
+    },
+    {
+      "epoch": 1.021010101010101,
+      "grad_norm": 3.25,
+      "learning_rate": 5.689958800207879e-06,
+      "loss": 1.0336,
+      "step": 631
+    },
+    {
+      "epoch": 1.0226262626262625,
+      "grad_norm": 2.90625,
+      "learning_rate": 5.6759653359218e-06,
+      "loss": 0.6566,
+      "step": 632
+    },
+    {
+      "epoch": 1.0242424242424242,
+      "grad_norm": 3.125,
+      "learning_rate": 5.661966476351756e-06,
+      "loss": 0.6963,
+      "step": 633
+    },
+    {
+      "epoch": 1.0258585858585858,
+      "grad_norm": 3.078125,
+      "learning_rate": 5.647962333231021e-06,
+      "loss": 1.0396,
+      "step": 634
+    },
+    {
+      "epoch": 1.0274747474747474,
+      "grad_norm": 3.21875,
+      "learning_rate": 5.633953018335047e-06,
+      "loss": 0.7885,
+      "step": 635
+    },
+    {
+      "epoch": 1.029090909090909,
+      "grad_norm": 3.046875,
+      "learning_rate": 5.6199386434805615e-06,
+      "loss": 0.8576,
+      "step": 636
+    },
+    {
+      "epoch": 1.0307070707070707,
+      "grad_norm": 2.921875,
+      "learning_rate": 5.605919320524677e-06,
+      "loss": 0.6605,
+      "step": 637
+    },
+    {
+      "epoch": 1.0323232323232323,
+      "grad_norm": 3.3125,
+      "learning_rate": 5.591895161364006e-06,
+      "loss": 0.6527,
+      "step": 638
+    },
+    {
+      "epoch": 1.033939393939394,
+      "grad_norm": 3.046875,
+      "learning_rate": 5.577866277933752e-06,
+      "loss": 0.6625,
+      "step": 639
+    },
+    {
+      "epoch": 1.0355555555555556,
+      "grad_norm": 2.828125,
+      "learning_rate": 5.563832782206835e-06,
+      "loss": 0.7634,
+      "step": 640
+    },
+    {
+      "epoch": 1.0371717171717172,
+      "grad_norm": 2.84375,
+      "learning_rate": 5.549794786192984e-06,
+      "loss": 0.7123,
+      "step": 641
+    },
+    {
+      "epoch": 1.0387878787878788,
+      "grad_norm": 3.1875,
+      "learning_rate": 5.535752401937846e-06,
+      "loss": 0.787,
+      "step": 642
+    },
+    {
+      "epoch": 1.0404040404040404,
+      "grad_norm": 3.3125,
+      "learning_rate": 5.521705741522096e-06,
+      "loss": 0.7617,
+      "step": 643
+    },
+    {
+      "epoch": 1.042020202020202,
+      "grad_norm": 3.21875,
+      "learning_rate": 5.507654917060541e-06,
+      "loss": 0.8361,
+      "step": 644
+    },
+    {
+      "epoch": 1.0436363636363637,
+      "grad_norm": 2.9375,
+      "learning_rate": 5.49360004070122e-06,
+      "loss": 0.7832,
+      "step": 645
+    },
+    {
+      "epoch": 1.0452525252525253,
+      "grad_norm": 3.078125,
+      "learning_rate": 5.4795412246245126e-06,
+      "loss": 0.7117,
+      "step": 646
+    },
+    {
+      "epoch": 1.046868686868687,
+      "grad_norm": 3.125,
+      "learning_rate": 5.465478581042248e-06,
+      "loss": 0.8191,
+      "step": 647
+    },
+    {
+      "epoch": 1.0484848484848486,
+      "grad_norm": 3.1875,
+      "learning_rate": 5.451412222196801e-06,
+      "loss": 0.9368,
+      "step": 648
+    },
+    {
+      "epoch": 1.0501010101010102,
+      "grad_norm": 3.109375,
+      "learning_rate": 5.4373422603602025e-06,
+      "loss": 0.7972,
+      "step": 649
+    },
+    {
+      "epoch": 1.0517171717171716,
+      "grad_norm": 3.453125,
+      "learning_rate": 5.4232688078332384e-06,
+      "loss": 1.1199,
+      "step": 650
+    },
+    {
+      "epoch": 1.0533333333333332,
+      "grad_norm": 3.125,
+      "learning_rate": 5.409191976944559e-06,
+      "loss": 0.7229,
+      "step": 651
+    },
+    {
+      "epoch": 1.0549494949494949,
+      "grad_norm": 3.125,
+      "learning_rate": 5.395111880049775e-06,
+      "loss": 0.8179,
+      "step": 652
+    },
+    {
+      "epoch": 1.0565656565656565,
+      "grad_norm": 2.71875,
+      "learning_rate": 5.3810286295305715e-06,
+      "loss": 0.7806,
+      "step": 653
+    },
+    {
+      "epoch": 1.0581818181818181,
+      "grad_norm": 4.03125,
+      "learning_rate": 5.366942337793798e-06,
+      "loss": 0.8215,
+      "step": 654
+    },
+    {
+      "epoch": 1.0597979797979797,
+      "grad_norm": 4.03125,
+      "learning_rate": 5.352853117270581e-06,
+      "loss": 1.0786,
+      "step": 655
+    },
+    {
+      "epoch": 1.0614141414141414,
+      "grad_norm": 3.78125,
+      "learning_rate": 5.338761080415425e-06,
+      "loss": 0.6667,
+      "step": 656
+    },
+    {
+      "epoch": 1.063030303030303,
+      "grad_norm": 3.203125,
+      "learning_rate": 5.324666339705307e-06,
+      "loss": 0.9499,
+      "step": 657
+    },
+    {
+      "epoch": 1.0646464646464646,
+      "grad_norm": 3.3125,
+      "learning_rate": 5.310569007638795e-06,
+      "loss": 1.1114,
+      "step": 658
+    },
+    {
+      "epoch": 1.0662626262626262,
+      "grad_norm": 3.828125,
+      "learning_rate": 5.296469196735134e-06,
+      "loss": 0.7428,
+      "step": 659
+    },
+    {
+      "epoch": 1.0678787878787879,
+      "grad_norm": 3.640625,
+      "learning_rate": 5.28236701953335e-06,
+      "loss": 0.7143,
+      "step": 660
+    },
+    {
+      "epoch": 1.0694949494949495,
+      "grad_norm": 3.453125,
+      "learning_rate": 5.268262588591364e-06,
+      "loss": 0.8979,
+      "step": 661
+    },
+    {
+      "epoch": 1.0711111111111111,
+      "grad_norm": 3.09375,
+      "learning_rate": 5.254156016485082e-06,
+      "loss": 0.8049,
+      "step": 662
+    },
+    {
+      "epoch": 1.0727272727272728,
+      "grad_norm": 3.515625,
+      "learning_rate": 5.240047415807499e-06,
+      "loss": 0.9266,
+      "step": 663
+    },
+    {
+      "epoch": 1.0743434343434344,
+      "grad_norm": 3.28125,
+      "learning_rate": 5.225936899167803e-06,
+      "loss": 0.9866,
+      "step": 664
+    },
+    {
+      "epoch": 1.075959595959596,
+      "grad_norm": 3.109375,
+      "learning_rate": 5.211824579190473e-06,
+      "loss": 0.6352,
+      "step": 665
+    },
+    {
+      "epoch": 1.0775757575757576,
+      "grad_norm": 3.5,
+      "learning_rate": 5.197710568514381e-06,
+      "loss": 0.7511,
+      "step": 666
+    },
+    {
+      "epoch": 1.0791919191919193,
+      "grad_norm": 3.03125,
+      "learning_rate": 5.183594979791896e-06,
+      "loss": 0.8406,
+      "step": 667
+    },
+    {
+      "epoch": 1.0808080808080809,
+      "grad_norm": 3.171875,
+      "learning_rate": 5.169477925687981e-06,
+      "loss": 0.6917,
+      "step": 668
+    },
+    {
+      "epoch": 1.0824242424242425,
+      "grad_norm": 3.046875,
+      "learning_rate": 5.155359518879295e-06,
+      "loss": 0.9025,
+      "step": 669
+    },
+    {
+      "epoch": 1.0840404040404041,
+      "grad_norm": 3.046875,
+      "learning_rate": 5.141239872053294e-06,
+      "loss": 0.7685,
+      "step": 670
+    },
+    {
+      "epoch": 1.0856565656565658,
+      "grad_norm": 3.40625,
+      "learning_rate": 5.1271190979073296e-06,
+      "loss": 0.6685,
+      "step": 671
+    },
+    {
+      "epoch": 1.0872727272727274,
+      "grad_norm": 4.0625,
+      "learning_rate": 5.112997309147753e-06,
+      "loss": 0.7217,
+      "step": 672
+    },
+    {
+      "epoch": 1.0888888888888888,
+      "grad_norm": 3.15625,
+      "learning_rate": 5.098874618489017e-06,
+      "loss": 0.6777,
+      "step": 673
+    },
+    {
+      "epoch": 1.0905050505050504,
+      "grad_norm": 3.375,
+      "learning_rate": 5.084751138652764e-06,
+      "loss": 0.915,
+      "step": 674
+    },
+    {
+      "epoch": 1.092121212121212,
+      "grad_norm": 2.921875,
+      "learning_rate": 5.070626982366943e-06,
+      "loss": 0.9366,
+      "step": 675
+    },
+    {
+      "epoch": 1.0937373737373737,
+      "grad_norm": 3.0,
+      "learning_rate": 5.0565022623649e-06,
+      "loss": 0.6551,
+      "step": 676
+    },
+    {
+      "epoch": 1.0953535353535353,
+      "grad_norm": 2.921875,
+      "learning_rate": 5.0423770913844795e-06,
+      "loss": 0.8253,
+      "step": 677
+    },
+    {
+      "epoch": 1.096969696969697,
+      "grad_norm": 3.09375,
+      "learning_rate": 5.028251582167126e-06,
+      "loss": 0.9756,
+      "step": 678
+    },
+    {
+      "epoch": 1.0985858585858586,
+      "grad_norm": 3.234375,
+      "learning_rate": 5.0141258474569855e-06,
+      "loss": 0.9388,
+      "step": 679
+    },
+    {
+      "epoch": 1.1002020202020202,
+      "grad_norm": 2.828125,
+      "learning_rate": 5e-06,
+      "loss": 0.7607,
+      "step": 680
+    },
+    {
+      "epoch": 1.1018181818181818,
+      "grad_norm": 3.4375,
+      "learning_rate": 4.985874152543015e-06,
+      "loss": 0.7243,
+      "step": 681
+    },
+    {
+      "epoch": 1.1034343434343434,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.971748417832876e-06,
+      "loss": 0.73,
+      "step": 682
+    },
+    {
+      "epoch": 1.105050505050505,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.957622908615522e-06,
+      "loss": 0.785,
+      "step": 683
+    },
+    {
+      "epoch": 1.1066666666666667,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.943497737635103e-06,
+      "loss": 0.8805,
+      "step": 684
+    },
+    {
+      "epoch": 1.1082828282828283,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.92937301763306e-06,
+      "loss": 0.6538,
+      "step": 685
+    },
+    {
+      "epoch": 1.10989898989899,
+      "grad_norm": 3.109375,
+      "learning_rate": 4.915248861347239e-06,
+      "loss": 0.8414,
+      "step": 686
+    },
+    {
+      "epoch": 1.1115151515151516,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.901125381510986e-06,
+      "loss": 0.8531,
+      "step": 687
+    },
+    {
+      "epoch": 1.1131313131313132,
+      "grad_norm": 3.359375,
+      "learning_rate": 4.887002690852249e-06,
+      "loss": 0.6754,
+      "step": 688
+    },
+    {
+      "epoch": 1.1147474747474748,
+      "grad_norm": 3.390625,
+      "learning_rate": 4.872880902092673e-06,
+      "loss": 0.8694,
+      "step": 689
+    },
+    {
+      "epoch": 1.1163636363636364,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.858760127946707e-06,
+      "loss": 0.9963,
+      "step": 690
+    },
+    {
+      "epoch": 1.117979797979798,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.8446404811207055e-06,
+      "loss": 0.7919,
+      "step": 691
+    },
+    {
+      "epoch": 1.1195959595959597,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.830522074312019e-06,
+      "loss": 0.8495,
+      "step": 692
+    },
+    {
+      "epoch": 1.121212121212121,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.816405020208105e-06,
+      "loss": 0.6419,
+      "step": 693
+    },
+    {
+      "epoch": 1.1228282828282827,
+      "grad_norm": 3.125,
+      "learning_rate": 4.80228943148562e-06,
+      "loss": 0.7856,
+      "step": 694
+    },
+    {
+      "epoch": 1.1244444444444444,
+      "grad_norm": 3.125,
+      "learning_rate": 4.788175420809528e-06,
+      "loss": 0.6257,
+      "step": 695
+    },
+    {
+      "epoch": 1.126060606060606,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.774063100832199e-06,
+      "loss": 0.6963,
+      "step": 696
+    },
+    {
+      "epoch": 1.1276767676767676,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.759952584192502e-06,
+      "loss": 0.6149,
+      "step": 697
+    },
+    {
+      "epoch": 1.1292929292929292,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.745843983514919e-06,
+      "loss": 0.906,
+      "step": 698
+    },
+    {
+      "epoch": 1.1309090909090909,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.731737411408637e-06,
+      "loss": 0.7671,
+      "step": 699
+    },
+    {
+      "epoch": 1.1325252525252525,
+      "grad_norm": 3.109375,
+      "learning_rate": 4.717632980466652e-06,
+      "loss": 0.8264,
+      "step": 700
+    },
+    {
+      "epoch": 1.1341414141414141,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.703530803264868e-06,
+      "loss": 0.7456,
+      "step": 701
+    },
+    {
+      "epoch": 1.1357575757575757,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.6894309923612055e-06,
+      "loss": 0.6059,
+      "step": 702
+    },
+    {
+      "epoch": 1.1373737373737374,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.675333660294693e-06,
+      "loss": 0.9912,
+      "step": 703
+    },
+    {
+      "epoch": 1.138989898989899,
+      "grad_norm": 3.5625,
+      "learning_rate": 4.661238919584578e-06,
+      "loss": 1.2429,
+      "step": 704
+    },
+    {
+      "epoch": 1.1406060606060606,
+      "grad_norm": 2.875,
+      "learning_rate": 4.647146882729422e-06,
+      "loss": 0.9141,
+      "step": 705
+    },
+    {
+      "epoch": 1.1422222222222222,
+      "grad_norm": 3.046875,
+      "learning_rate": 4.633057662206205e-06,
+      "loss": 0.9058,
+      "step": 706
+    },
+    {
+      "epoch": 1.1438383838383839,
+      "grad_norm": 3.125,
+      "learning_rate": 4.618971370469431e-06,
+      "loss": 0.7651,
+      "step": 707
+    },
+    {
+      "epoch": 1.1454545454545455,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.6048881199502265e-06,
+      "loss": 0.8145,
+      "step": 708
+    },
+    {
+      "epoch": 1.1470707070707071,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.590808023055444e-06,
+      "loss": 0.7758,
+      "step": 709
+    },
+    {
+      "epoch": 1.1486868686868688,
+      "grad_norm": 3.46875,
+      "learning_rate": 4.576731192166762e-06,
+      "loss": 0.8211,
+      "step": 710
+    },
+    {
+      "epoch": 1.1503030303030304,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.562657739639798e-06,
+      "loss": 1.0202,
+      "step": 711
+    },
+    {
+      "epoch": 1.151919191919192,
+      "grad_norm": 3.125,
+      "learning_rate": 4.548587777803198e-06,
+      "loss": 0.7618,
+      "step": 712
+    },
+    {
+      "epoch": 1.1535353535353536,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.534521418957753e-06,
+      "loss": 0.7477,
+      "step": 713
+    },
+    {
+      "epoch": 1.1551515151515153,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.520458775375488e-06,
+      "loss": 1.0355,
+      "step": 714
+    },
+    {
+      "epoch": 1.1567676767676769,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.506399959298781e-06,
+      "loss": 0.7698,
+      "step": 715
+    },
+    {
+      "epoch": 1.1583838383838383,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.49234508293946e-06,
+      "loss": 0.6,
+      "step": 716
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.4782942584779046e-06,
+      "loss": 0.7238,
+      "step": 717
+    },
+    {
+      "epoch": 1.1616161616161615,
+      "grad_norm": 3.15625,
+      "learning_rate": 4.464247598062156e-06,
+      "loss": 0.7954,
+      "step": 718
+    },
+    {
+      "epoch": 1.1632323232323232,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.450205213807018e-06,
+      "loss": 0.5872,
+      "step": 719
+    },
+    {
+      "epoch": 1.1648484848484848,
+      "grad_norm": 3.515625,
+      "learning_rate": 4.436167217793167e-06,
+      "loss": 1.0127,
+      "step": 720
+    },
+    {
+      "epoch": 1.1664646464646464,
+      "grad_norm": 3.453125,
+      "learning_rate": 4.4221337220662495e-06,
+      "loss": 1.1503,
+      "step": 721
+    },
+    {
+      "epoch": 1.168080808080808,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.408104838635996e-06,
+      "loss": 0.8002,
+      "step": 722
+    },
+    {
+      "epoch": 1.1696969696969697,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.394080679475324e-06,
+      "loss": 0.6457,
+      "step": 723
+    },
+    {
+      "epoch": 1.1713131313131313,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.38006135651944e-06,
+      "loss": 1.1433,
+      "step": 724
+    },
+    {
+      "epoch": 1.172929292929293,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.366046981664954e-06,
+      "loss": 0.9229,
+      "step": 725
+    },
+    {
+      "epoch": 1.1745454545454546,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.3520376667689815e-06,
+      "loss": 0.9632,
+      "step": 726
+    },
+    {
+      "epoch": 1.1761616161616162,
+      "grad_norm": 3.0,
+      "learning_rate": 4.338033523648247e-06,
+      "loss": 0.9278,
+      "step": 727
+    },
+    {
+      "epoch": 1.1777777777777778,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.3240346640782014e-06,
+      "loss": 1.0234,
+      "step": 728
+    },
+    {
+      "epoch": 1.1793939393939394,
+      "grad_norm": 3.375,
+      "learning_rate": 4.3100411997921235e-06,
+      "loss": 0.7461,
+      "step": 729
+    },
+    {
+      "epoch": 1.181010101010101,
+      "grad_norm": 2.828125,
+      "learning_rate": 4.2960532424802235e-06,
+      "loss": 0.9571,
+      "step": 730
+    },
+    {
+      "epoch": 1.1826262626262627,
+      "grad_norm": 3.25,
+      "learning_rate": 4.282070903788767e-06,
+      "loss": 0.9227,
+      "step": 731
+    },
+    {
+      "epoch": 1.1842424242424243,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.268094295319167e-06,
+      "loss": 0.8272,
+      "step": 732
+    },
+    {
+      "epoch": 1.185858585858586,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.2541235286271e-06,
+      "loss": 0.8806,
+      "step": 733
+    },
+    {
+      "epoch": 1.1874747474747476,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.240158715221621e-06,
+      "loss": 0.815,
+      "step": 734
+    },
+    {
+      "epoch": 1.189090909090909,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.2261999665642615e-06,
+      "loss": 0.8117,
+      "step": 735
+    },
+    {
+      "epoch": 1.1907070707070706,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.212247394068151e-06,
+      "loss": 0.5376,
+      "step": 736
+    },
+    {
+      "epoch": 1.1923232323232322,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.198301109097124e-06,
+      "loss": 1.1354,
+      "step": 737
+    },
+    {
+      "epoch": 1.1939393939393939,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.184361222964827e-06,
+      "loss": 0.7252,
+      "step": 738
+    },
+    {
+      "epoch": 1.1939393939393939,
+      "eval_loss": 0.8301404118537903,
+      "eval_runtime": 33.6772,
+      "eval_samples_per_second": 32.663,
+      "eval_steps_per_second": 4.098,
+      "step": 738
+    },
+    {
+      "epoch": 1.1955555555555555,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.170427846933835e-06,
+      "loss": 0.5139,
+      "step": 739
+    },
+    {
+      "epoch": 1.197171717171717,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.1565010922147644e-06,
+      "loss": 0.6976,
+      "step": 740
+    },
+    {
+      "epoch": 1.1987878787878787,
+      "grad_norm": 3.09375,
+      "learning_rate": 4.142581069965379e-06,
+      "loss": 0.8952,
+      "step": 741
+    },
+    {
+      "epoch": 1.2004040404040404,
+      "grad_norm": 3.90625,
+      "learning_rate": 4.1286678912897095e-06,
+      "loss": 0.7332,
+      "step": 742
+    },
+    {
+      "epoch": 1.202020202020202,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.114761667237164e-06,
+      "loss": 0.6894,
+      "step": 743
+    },
+    {
+      "epoch": 1.2036363636363636,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.100862508801639e-06,
+      "loss": 0.7328,
+      "step": 744
+    },
+    {
+      "epoch": 1.2052525252525252,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.086970526920638e-06,
+      "loss": 0.5731,
+      "step": 745
+    },
+    {
+      "epoch": 1.2068686868686869,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.0730858324743845e-06,
+      "loss": 0.8963,
+      "step": 746
+    },
+    {
+      "epoch": 1.2084848484848485,
+      "grad_norm": 3.25,
+      "learning_rate": 4.059208536284936e-06,
+      "loss": 0.7759,
+      "step": 747
+    },
+    {
+      "epoch": 1.2101010101010101,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.045338749115299e-06,
+      "loss": 0.854,
+      "step": 748
+    },
+    {
+      "epoch": 1.2117171717171717,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.031476581668549e-06,
+      "loss": 0.7933,
+      "step": 749
+    },
+    {
+      "epoch": 1.2133333333333334,
+      "grad_norm": 3.171875,
+      "learning_rate": 4.017622144586938e-06,
+      "loss": 0.7486,
+      "step": 750
+    },
+    {
+      "epoch": 1.214949494949495,
+      "grad_norm": 2.875,
+      "learning_rate": 4.003775548451028e-06,
+      "loss": 0.7833,
+      "step": 751
+    },
+    {
+      "epoch": 1.2165656565656566,
+      "grad_norm": 3.4375,
+      "learning_rate": 3.989936903778785e-06,
+      "loss": 0.7697,
+      "step": 752
+    },
+    {
+      "epoch": 1.2181818181818183,
+      "grad_norm": 3.578125,
+      "learning_rate": 3.97610632102472e-06,
+      "loss": 0.7627,
+      "step": 753
+    },
+    {
+      "epoch": 1.2197979797979799,
+      "grad_norm": 2.984375,
+      "learning_rate": 3.962283910578991e-06,
+      "loss": 1.0429,
+      "step": 754
+    },
+    {
+      "epoch": 1.2214141414141415,
+      "grad_norm": 3.3125,
+      "learning_rate": 3.948469782766528e-06,
+      "loss": 0.7313,
+      "step": 755
+    },
+    {
+      "epoch": 1.2230303030303031,
+      "grad_norm": 3.515625,
+      "learning_rate": 3.934664047846157e-06,
+      "loss": 0.9511,
+      "step": 756
+    },
+    {
+      "epoch": 1.2246464646464648,
+      "grad_norm": 2.875,
+      "learning_rate": 3.9208668160097095e-06,
+      "loss": 0.7375,
+      "step": 757
+    },
+    {
+      "epoch": 1.2262626262626264,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.907078197381153e-06,
+      "loss": 0.7903,
+      "step": 758
+    },
+    {
+      "epoch": 1.2278787878787878,
+      "grad_norm": 3.5,
+      "learning_rate": 3.893298302015705e-06,
+      "loss": 0.8649,
+      "step": 759
+    },
+    {
+      "epoch": 1.2294949494949494,
+      "grad_norm": 3.328125,
+      "learning_rate": 3.879527239898962e-06,
+      "loss": 0.5326,
+      "step": 760
+    },
+    {
+      "epoch": 1.231111111111111,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.8657651209460104e-06,
+      "loss": 0.7415,
+      "step": 761
+    },
+    {
+      "epoch": 1.2327272727272727,
+      "grad_norm": 2.78125,
+      "learning_rate": 3.85201205500056e-06,
+      "loss": 0.5808,
+      "step": 762
+    },
+    {
+      "epoch": 1.2343434343434343,
+      "grad_norm": 3.375,
+      "learning_rate": 3.838268151834065e-06,
+      "loss": 0.8453,
+      "step": 763
+    },
+    {
+      "epoch": 1.235959595959596,
+      "grad_norm": 3.328125,
+      "learning_rate": 3.8245335211448404e-06,
+      "loss": 0.8512,
+      "step": 764
+    },
+    {
+      "epoch": 1.2375757575757576,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.810808272557196e-06,
+      "loss": 0.5858,
+      "step": 765
+    },
+    {
+      "epoch": 1.2391919191919192,
+      "grad_norm": 3.46875,
+      "learning_rate": 3.797092515620557e-06,
+      "loss": 0.7213,
+      "step": 766
+    },
+    {
+      "epoch": 1.2408080808080808,
+      "grad_norm": 3.46875,
+      "learning_rate": 3.7833863598085897e-06,
+      "loss": 1.1535,
+      "step": 767
+    },
+    {
+      "epoch": 1.2424242424242424,
+      "grad_norm": 2.984375,
+      "learning_rate": 3.769689914518326e-06,
+      "loss": 0.6643,
+      "step": 768
+    },
+    {
+      "epoch": 1.244040404040404,
+      "grad_norm": 3.390625,
+      "learning_rate": 3.7560032890692967e-06,
+      "loss": 1.357,
+      "step": 769
+    },
+    {
+      "epoch": 1.2456565656565657,
+      "grad_norm": 3.328125,
+      "learning_rate": 3.7423265927026473e-06,
+      "loss": 0.8818,
+      "step": 770
+    },
+    {
+      "epoch": 1.2472727272727273,
+      "grad_norm": 3.1875,
+      "learning_rate": 3.728659934580286e-06,
+      "loss": 0.7223,
+      "step": 771
+    },
+    {
+      "epoch": 1.248888888888889,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.715003423783986e-06,
+      "loss": 0.9481,
+      "step": 772
+    },
+    {
+      "epoch": 1.2505050505050506,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.701357169314536e-06,
+      "loss": 1.0692,
+      "step": 773
+    },
+    {
+      "epoch": 1.2521212121212122,
+      "grad_norm": 3.609375,
+      "learning_rate": 3.6877212800908625e-06,
+      "loss": 1.0956,
+      "step": 774
+    },
+    {
+      "epoch": 1.2537373737373738,
+      "grad_norm": 3.46875,
+      "learning_rate": 3.6740958649491575e-06,
+      "loss": 1.1225,
+      "step": 775
+    },
+    {
+      "epoch": 1.2553535353535352,
+      "grad_norm": 3.40625,
+      "learning_rate": 3.660481032642016e-06,
+      "loss": 0.7702,
+      "step": 776
+    },
+    {
+      "epoch": 1.2569696969696968,
+      "grad_norm": 3.359375,
+      "learning_rate": 3.646876891837566e-06,
+      "loss": 0.7712,
+      "step": 777
+    },
+    {
+      "epoch": 1.2585858585858585,
+      "grad_norm": 3.28125,
+      "learning_rate": 3.633283551118595e-06,
+      "loss": 0.8993,
+      "step": 778
+    },
+    {
+      "epoch": 1.26020202020202,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.6197011189816945e-06,
+      "loss": 0.7197,
+      "step": 779
+    },
+    {
+      "epoch": 1.2618181818181817,
+      "grad_norm": 3.125,
+      "learning_rate": 3.6061297038363853e-06,
+      "loss": 0.7741,
+      "step": 780
+    },
+    {
+      "epoch": 1.2634343434343434,
+      "grad_norm": 3.109375,
+      "learning_rate": 3.5925694140042546e-06,
+      "loss": 0.8199,
+      "step": 781
+    },
+    {
+      "epoch": 1.265050505050505,
+      "grad_norm": 2.90625,
+      "learning_rate": 3.579020357718092e-06,
+      "loss": 1.0651,
+      "step": 782
+    },
+    {
+      "epoch": 1.2666666666666666,
+      "grad_norm": 3.3125,
+      "learning_rate": 3.5654826431210273e-06,
+      "loss": 0.933,
+      "step": 783
+    },
+    {
+      "epoch": 1.2682828282828282,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.5519563782656642e-06,
+      "loss": 0.7186,
+      "step": 784
+    },
+    {
+      "epoch": 1.2698989898989899,
+      "grad_norm": 3.1875,
+      "learning_rate": 3.5384416711132173e-06,
+      "loss": 0.7756,
+      "step": 785
+    },
+    {
+      "epoch": 1.2715151515151515,
+      "grad_norm": 3.5,
+      "learning_rate": 3.524938629532656e-06,
+      "loss": 1.0024,
+      "step": 786
+    },
+    {
+      "epoch": 1.2731313131313131,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.5114473612998367e-06,
+      "loss": 0.8499,
+      "step": 787
+    },
+    {
+      "epoch": 1.2747474747474747,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.497967974096647e-06,
+      "loss": 0.7138,
+      "step": 788
+    },
+    {
+      "epoch": 1.2763636363636364,
+      "grad_norm": 3.265625,
+      "learning_rate": 3.4845005755101446e-06,
+      "loss": 0.8185,
+      "step": 789
+    },
+    {
+      "epoch": 1.277979797979798,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.4710452730316978e-06,
+      "loss": 0.6772,
+      "step": 790
+    },
+    {
+      "epoch": 1.2795959595959596,
+      "grad_norm": 3.34375,
+      "learning_rate": 3.457602174056134e-06,
+      "loss": 0.8103,
+      "step": 791
+    },
+    {
+      "epoch": 1.2812121212121212,
+      "grad_norm": 3.3125,
+      "learning_rate": 3.4441713858808684e-06,
+      "loss": 0.8651,
+      "step": 792
+    },
+    {
+      "epoch": 1.2828282828282829,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.4307530157050624e-06,
+      "loss": 0.8169,
+      "step": 793
+    },
+    {
+      "epoch": 1.2844444444444445,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.4173471706287607e-06,
+      "loss": 0.8362,
+      "step": 794
+    },
+    {
+      "epoch": 1.2860606060606061,
+      "grad_norm": 3.109375,
+      "learning_rate": 3.4039539576520376e-06,
+      "loss": 0.8504,
+      "step": 795
+    },
+    {
+      "epoch": 1.2876767676767678,
+      "grad_norm": 3.859375,
+      "learning_rate": 3.3905734836741415e-06,
+      "loss": 0.9376,
+      "step": 796
+    },
+    {
+      "epoch": 1.2892929292929294,
+      "grad_norm": 3.828125,
+      "learning_rate": 3.3772058554926456e-06,
+      "loss": 0.7374,
+      "step": 797
+    },
+    {
+      "epoch": 1.290909090909091,
+      "grad_norm": 3.078125,
+      "learning_rate": 3.3638511798025927e-06,
+      "loss": 0.7132,
+      "step": 798
+    },
+    {
+      "epoch": 1.2925252525252526,
+      "grad_norm": 3.453125,
+      "learning_rate": 3.350509563195641e-06,
+      "loss": 0.9722,
+      "step": 799
+    },
+    {
+      "epoch": 1.2941414141414143,
+      "grad_norm": 3.328125,
+      "learning_rate": 3.3371811121592203e-06,
+      "loss": 0.6305,
+      "step": 800
+    },
+    {
+      "epoch": 1.2957575757575759,
+      "grad_norm": 2.9375,
+      "learning_rate": 3.3238659330756774e-06,
+      "loss": 0.9804,
+      "step": 801
+    },
+    {
+      "epoch": 1.2973737373737375,
+      "grad_norm": 3.78125,
+      "learning_rate": 3.3105641322214238e-06,
+      "loss": 0.8994,
+      "step": 802
+    },
+    {
+      "epoch": 1.298989898989899,
+      "grad_norm": 3.28125,
+      "learning_rate": 3.2972758157660966e-06,
+      "loss": 0.5214,
+      "step": 803
+    },
+    {
+      "epoch": 1.3006060606060605,
+      "grad_norm": 3.15625,
+      "learning_rate": 3.2840010897717045e-06,
+      "loss": 0.827,
+      "step": 804
+    },
+    {
+      "epoch": 1.3022222222222222,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.270740060191779e-06,
+      "loss": 0.8329,
+      "step": 805
+    },
+    {
+      "epoch": 1.3038383838383838,
+      "grad_norm": 3.46875,
+      "learning_rate": 3.257492832870537e-06,
+      "loss": 0.9276,
+      "step": 806
+    },
+    {
+      "epoch": 1.3054545454545454,
+      "grad_norm": 3.28125,
+      "learning_rate": 3.2442595135420287e-06,
+      "loss": 0.5933,
+      "step": 807
+    },
+    {
+      "epoch": 1.307070707070707,
+      "grad_norm": 2.90625,
+      "learning_rate": 3.2310402078292956e-06,
+      "loss": 0.5578,
+      "step": 808
+    },
+    {
+      "epoch": 1.3086868686868687,
+      "grad_norm": 3.25,
+      "learning_rate": 3.2178350212435315e-06,
+      "loss": 0.8982,
+      "step": 809
+    },
+    {
+      "epoch": 1.3103030303030303,
+      "grad_norm": 3.546875,
+      "learning_rate": 3.2046440591832323e-06,
+      "loss": 0.8584,
+      "step": 810
+    },
+    {
+      "epoch": 1.311919191919192,
+      "grad_norm": 3.34375,
+      "learning_rate": 3.191467426933365e-06,
+      "loss": 0.6637,
+      "step": 811
+    },
+    {
+      "epoch": 1.3135353535353536,
+      "grad_norm": 2.625,
+      "learning_rate": 3.178305229664519e-06,
+      "loss": 0.5181,
+      "step": 812
+    },
+    {
+      "epoch": 1.3151515151515152,
+      "grad_norm": 2.9375,
+      "learning_rate": 3.1651575724320665e-06,
+      "loss": 0.5668,
+      "step": 813
+    },
+    {
+      "epoch": 1.3167676767676768,
+      "grad_norm": 3.96875,
+      "learning_rate": 3.152024560175332e-06,
+      "loss": 0.8147,
+      "step": 814
+    },
+    {
+      "epoch": 1.3183838383838384,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.1389062977167474e-06,
+      "loss": 0.8671,
+      "step": 815
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 3.28125,
+      "learning_rate": 3.125802889761016e-06,
+      "loss": 0.6689,
+      "step": 816
+    },
+    {
+      "epoch": 1.3216161616161617,
+      "grad_norm": 3.609375,
+      "learning_rate": 3.112714440894281e-06,
+      "loss": 0.9006,
+      "step": 817
+    },
+    {
+      "epoch": 1.3232323232323233,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.099641055583289e-06,
+      "loss": 0.6475,
+      "step": 818
+    },
+    {
+      "epoch": 1.3248484848484847,
+      "grad_norm": 3.15625,
+      "learning_rate": 3.0865828381745515e-06,
+      "loss": 0.803,
+      "step": 819
+    },
+    {
+      "epoch": 1.3264646464646463,
+      "grad_norm": 3.671875,
+      "learning_rate": 3.073539892893519e-06,
+      "loss": 0.612,
+      "step": 820
+    },
+    {
+      "epoch": 1.328080808080808,
+      "grad_norm": 3.4375,
+      "learning_rate": 3.0605123238437457e-06,
+      "loss": 0.873,
+      "step": 821
+    },
+    {
+      "epoch": 1.3296969696969696,
+      "grad_norm": 3.1875,
+      "learning_rate": 3.047500235006056e-06,
+      "loss": 0.6955,
+      "step": 822
+    },
+    {
+      "epoch": 1.3313131313131312,
+      "grad_norm": 3.28125,
+      "learning_rate": 3.0345037302377216e-06,
+      "loss": 1.0614,
+      "step": 823
+    },
+    {
+      "epoch": 1.3329292929292929,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.021522913271627e-06,
+      "loss": 0.676,
+      "step": 824
+    },
+    {
+      "epoch": 1.3345454545454545,
+      "grad_norm": 3.71875,
+      "learning_rate": 3.008557887715441e-06,
+      "loss": 0.9605,
+      "step": 825
+    },
+    {
+      "epoch": 1.336161616161616,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.995608757050793e-06,
+      "loss": 0.7708,
+      "step": 826
+    },
+    {
+      "epoch": 1.3377777777777777,
+      "grad_norm": 3.484375,
+      "learning_rate": 2.9826756246324477e-06,
+      "loss": 0.905,
+      "step": 827
+    },
+    {
+      "epoch": 1.3393939393939394,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.969758593687475e-06,
+      "loss": 1.6665,
+      "step": 828
+    },
+    {
+      "epoch": 1.341010101010101,
+      "grad_norm": 3.1875,
+      "learning_rate": 2.95685776731443e-06,
+      "loss": 0.7902,
+      "step": 829
+    },
+    {
+      "epoch": 1.3426262626262626,
+      "grad_norm": 3.5,
+      "learning_rate": 2.9439732484825323e-06,
+      "loss": 0.7342,
+      "step": 830
+    },
+    {
+      "epoch": 1.3442424242424242,
+      "grad_norm": 3.671875,
+      "learning_rate": 2.931105140030838e-06,
+      "loss": 0.6227,
+      "step": 831
+    },
+    {
+      "epoch": 1.3458585858585859,
+      "grad_norm": 3.921875,
+      "learning_rate": 2.9182535446674244e-06,
+      "loss": 0.9251,
+      "step": 832
+    },
+    {
+      "epoch": 1.3474747474747475,
+      "grad_norm": 3.1875,
+      "learning_rate": 2.905418564968565e-06,
+      "loss": 1.0284,
+      "step": 833
+    },
+    {
+      "epoch": 1.3490909090909091,
+      "grad_norm": 3.359375,
+      "learning_rate": 2.8926003033779194e-06,
+      "loss": 0.6573,
+      "step": 834
+    },
+    {
+      "epoch": 1.3507070707070707,
+      "grad_norm": 3.46875,
+      "learning_rate": 2.8797988622057005e-06,
+      "loss": 0.8825,
+      "step": 835
+    },
+    {
+      "epoch": 1.3523232323232324,
+      "grad_norm": 3.109375,
+      "learning_rate": 2.8670143436278757e-06,
+      "loss": 0.6492,
+      "step": 836
+    },
+    {
+      "epoch": 1.353939393939394,
+      "grad_norm": 3.515625,
+      "learning_rate": 2.8542468496853404e-06,
+      "loss": 0.9723,
+      "step": 837
+    },
+    {
+      "epoch": 1.3555555555555556,
+      "grad_norm": 3.578125,
+      "learning_rate": 2.8414964822831063e-06,
+      "loss": 1.1992,
+      "step": 838
+    },
+    {
+      "epoch": 1.3571717171717173,
+      "grad_norm": 3.03125,
+      "learning_rate": 2.8287633431894877e-06,
+      "loss": 0.9513,
+      "step": 839
+    },
+    {
+      "epoch": 1.3587878787878789,
+      "grad_norm": 3.125,
+      "learning_rate": 2.8160475340352913e-06,
+      "loss": 0.8701,
+      "step": 840
+    },
+    {
+      "epoch": 1.3604040404040405,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.8033491563129977e-06,
+      "loss": 0.8466,
+      "step": 841
+    },
+    {
+      "epoch": 1.3620202020202021,
+      "grad_norm": 3.4375,
+      "learning_rate": 2.790668311375962e-06,
+      "loss": 0.7244,
+      "step": 842
+    },
+    {
+      "epoch": 1.3636363636363638,
+      "grad_norm": 3.0625,
+      "learning_rate": 2.7780051004375986e-06,
+      "loss": 0.737,
+      "step": 843
+    },
+    {
+      "epoch": 1.3652525252525254,
+      "grad_norm": 3.671875,
+      "learning_rate": 2.765359624570574e-06,
+      "loss": 0.7409,
+      "step": 844
+    },
+    {
+      "epoch": 1.3668686868686868,
+      "grad_norm": 3.234375,
+      "learning_rate": 2.752731984706e-06,
+      "loss": 0.7497,
+      "step": 845
+    },
+    {
+      "epoch": 1.3684848484848484,
+      "grad_norm": 3.40625,
+      "learning_rate": 2.7401222816326316e-06,
+      "loss": 0.8155,
+      "step": 846
+    },
+    {
+      "epoch": 1.37010101010101,
+      "grad_norm": 3.15625,
+      "learning_rate": 2.7275306159960522e-06,
+      "loss": 0.6168,
+      "step": 847
+    },
+    {
+      "epoch": 1.3717171717171717,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.714957088297886e-06,
+      "loss": 0.9487,
+      "step": 848
+    },
+    {
+      "epoch": 1.3733333333333333,
+      "grad_norm": 3.078125,
+      "learning_rate": 2.7024017988949802e-06,
+      "loss": 1.1136,
+      "step": 849
+    },
+    {
+      "epoch": 1.374949494949495,
+      "grad_norm": 2.96875,
+      "learning_rate": 2.6898648479986187e-06,
+      "loss": 0.5994,
+      "step": 850
+    },
+    {
+      "epoch": 1.3765656565656565,
+      "grad_norm": 3.765625,
+      "learning_rate": 2.677346335673713e-06,
+      "loss": 0.839,
+      "step": 851
+    },
+    {
+      "epoch": 1.3781818181818182,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.664846361837997e-06,
+      "loss": 0.791,
+      "step": 852
+    },
+    {
+      "epoch": 1.3797979797979798,
+      "grad_norm": 3.21875,
+      "learning_rate": 2.6523650262612476e-06,
+      "loss": 0.8164,
+      "step": 853
+    },
+    {
+      "epoch": 1.3814141414141414,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.639902428564475e-06,
+      "loss": 0.7807,
+      "step": 854
+    },
+    {
+      "epoch": 1.383030303030303,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.6274586682191306e-06,
+      "loss": 0.7378,
+      "step": 855
+    },
+    {
+      "epoch": 1.3846464646464647,
+      "grad_norm": 3.09375,
+      "learning_rate": 2.6150338445463146e-06,
+      "loss": 0.6805,
+      "step": 856
+    },
+    {
+      "epoch": 1.3862626262626263,
+      "grad_norm": 3.109375,
+      "learning_rate": 2.6026280567159847e-06,
+      "loss": 0.9109,
+      "step": 857
+    },
+    {
+      "epoch": 1.387878787878788,
+      "grad_norm": 3.203125,
+      "learning_rate": 2.5902414037461544e-06,
+      "loss": 0.8796,
+      "step": 858
+    },
+    {
+      "epoch": 1.3894949494949496,
+      "grad_norm": 3.359375,
+      "learning_rate": 2.5778739845021177e-06,
+      "loss": 0.9433,
+      "step": 859
+    },
+    {
+      "epoch": 1.3911111111111112,
+      "grad_norm": 3.5,
+      "learning_rate": 2.565525897695651e-06,
+      "loss": 0.8533,
+      "step": 860
+    },
+    {
+      "epoch": 1.3927272727272726,
+      "grad_norm": 3.109375,
+      "learning_rate": 2.5531972418842266e-06,
+      "loss": 0.9605,
+      "step": 861
+    },
+    {
+      "epoch": 1.3927272727272726,
+      "eval_loss": 0.8214139938354492,
+      "eval_runtime": 33.7125,
+      "eval_samples_per_second": 32.629,
+      "eval_steps_per_second": 4.093,
+      "step": 861
+    },
+    {
+      "epoch": 1.3943434343434342,
+      "grad_norm": 3.078125,
+      "learning_rate": 2.540888115470226e-06,
+      "loss": 0.7973,
+      "step": 862
+    },
+    {
+      "epoch": 1.3959595959595958,
+      "grad_norm": 3.328125,
+      "learning_rate": 2.5285986167001565e-06,
+      "loss": 0.8539,
+      "step": 863
+    },
+    {
+      "epoch": 1.3975757575757575,
+      "grad_norm": 3.34375,
+      "learning_rate": 2.51632884366386e-06,
+      "loss": 0.7262,
+      "step": 864
+    },
+    {
+      "epoch": 1.399191919191919,
+      "grad_norm": 3.671875,
+      "learning_rate": 2.5040788942937423e-06,
+      "loss": 0.6404,
+      "step": 865
+    },
+    {
+      "epoch": 1.4008080808080807,
+      "grad_norm": 2.9375,
+      "learning_rate": 2.4918488663639824e-06,
+      "loss": 0.7307,
+      "step": 866
+    },
+    {
+      "epoch": 1.4024242424242424,
+      "grad_norm": 3.578125,
+      "learning_rate": 2.479638857489753e-06,
+      "loss": 0.8732,
+      "step": 867
+    },
+    {
+      "epoch": 1.404040404040404,
+      "grad_norm": 3.109375,
+      "learning_rate": 2.4674489651264433e-06,
+      "loss": 0.5595,
+      "step": 868
+    },
+    {
+      "epoch": 1.4056565656565656,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.4552792865688814e-06,
+      "loss": 0.8797,
+      "step": 869
+    },
+    {
+      "epoch": 1.4072727272727272,
+      "grad_norm": 3.3125,
+      "learning_rate": 2.4431299189505563e-06,
+      "loss": 0.8571,
+      "step": 870
+    },
+    {
+      "epoch": 1.4088888888888889,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.4310009592428442e-06,
+      "loss": 0.973,
+      "step": 871
+    },
+    {
+      "epoch": 1.4105050505050505,
+      "grad_norm": 3.078125,
+      "learning_rate": 2.418892504254231e-06,
+      "loss": 0.9452,
+      "step": 872
+    },
+    {
+      "epoch": 1.412121212121212,
+      "grad_norm": 3.203125,
+      "learning_rate": 2.4068046506295455e-06,
+      "loss": 0.7929,
+      "step": 873
+    },
+    {
+      "epoch": 1.4137373737373737,
+      "grad_norm": 3.234375,
+      "learning_rate": 2.394737494849184e-06,
+      "loss": 0.9876,
+      "step": 874
+    },
+    {
+      "epoch": 1.4153535353535354,
+      "grad_norm": 3.125,
+      "learning_rate": 2.3826911332283364e-06,
+      "loss": 0.8509,
+      "step": 875
+    },
+    {
+      "epoch": 1.416969696969697,
+      "grad_norm": 2.875,
+      "learning_rate": 2.3706656619162278e-06,
+      "loss": 0.8132,
+      "step": 876
+    },
+    {
+      "epoch": 1.4185858585858586,
+      "grad_norm": 3.015625,
+      "learning_rate": 2.358661176895342e-06,
+      "loss": 0.9839,
+      "step": 877
+    },
+    {
+      "epoch": 1.4202020202020202,
+      "grad_norm": 3.515625,
+      "learning_rate": 2.34667777398066e-06,
+      "loss": 1.0469,
+      "step": 878
+    },
+    {
+      "epoch": 1.4218181818181819,
+      "grad_norm": 2.78125,
+      "learning_rate": 2.3347155488188944e-06,
+      "loss": 0.6276,
+      "step": 879
+    },
+    {
+      "epoch": 1.4234343434343435,
+      "grad_norm": 2.859375,
+      "learning_rate": 2.322774596887726e-06,
+      "loss": 0.6175,
+      "step": 880
+    },
+    {
+      "epoch": 1.4250505050505051,
+      "grad_norm": 3.078125,
+      "learning_rate": 2.3108550134950352e-06,
+      "loss": 0.7977,
+      "step": 881
+    },
+    {
+      "epoch": 1.4266666666666667,
+      "grad_norm": 3.03125,
+      "learning_rate": 2.298956893778154e-06,
+      "loss": 0.8741,
+      "step": 882
+    },
+    {
+      "epoch": 1.4282828282828284,
+      "grad_norm": 3.109375,
+      "learning_rate": 2.2870803327030995e-06,
+      "loss": 0.5821,
+      "step": 883
+    },
+    {
+      "epoch": 1.42989898989899,
+      "grad_norm": 3.15625,
+      "learning_rate": 2.275225425063813e-06,
+      "loss": 0.9132,
+      "step": 884
+    },
+    {
+      "epoch": 1.4315151515151516,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.2633922654814082e-06,
+      "loss": 0.5968,
+      "step": 885
+    },
+    {
+      "epoch": 1.4331313131313133,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.251580948403416e-06,
+      "loss": 0.7606,
+      "step": 886
+    },
+    {
+      "epoch": 1.4347474747474749,
+      "grad_norm": 4.375,
+      "learning_rate": 2.23979156810303e-06,
+      "loss": 0.7595,
+      "step": 887
+    },
+    {
+      "epoch": 1.4363636363636363,
+      "grad_norm": 3.875,
+      "learning_rate": 2.2280242186783473e-06,
+      "loss": 0.7991,
+      "step": 888
+    },
+    {
+      "epoch": 1.437979797979798,
+      "grad_norm": 3.4375,
+      "learning_rate": 2.2162789940516268e-06,
+      "loss": 0.7858,
+      "step": 889
+    },
+    {
+      "epoch": 1.4395959595959595,
+      "grad_norm": 3.5625,
+      "learning_rate": 2.204555987968541e-06,
+      "loss": 0.8497,
+      "step": 890
+    },
+    {
+      "epoch": 1.4412121212121212,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.192855293997418e-06,
+      "loss": 0.9145,
+      "step": 891
+    },
+    {
+      "epoch": 1.4428282828282828,
+      "grad_norm": 3.25,
+      "learning_rate": 2.1811770055284968e-06,
+      "loss": 0.699,
+      "step": 892
+    },
+    {
+      "epoch": 1.4444444444444444,
+      "grad_norm": 3.328125,
+      "learning_rate": 2.1695212157731904e-06,
+      "loss": 0.7465,
+      "step": 893
+    },
+    {
+      "epoch": 1.446060606060606,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.157888017763333e-06,
+      "loss": 0.9405,
+      "step": 894
+    },
+    {
+      "epoch": 1.4476767676767677,
+      "grad_norm": 3.125,
+      "learning_rate": 2.1462775043504423e-06,
+      "loss": 0.572,
+      "step": 895
+    },
+    {
+      "epoch": 1.4492929292929293,
+      "grad_norm": 3.125,
+      "learning_rate": 2.134689768204975e-06,
+      "loss": 0.7691,
+      "step": 896
+    },
+    {
+      "epoch": 1.450909090909091,
+      "grad_norm": 3.90625,
+      "learning_rate": 2.12312490181559e-06,
+      "loss": 0.8271,
+      "step": 897
+    },
+    {
+      "epoch": 1.4525252525252526,
+      "grad_norm": 3.21875,
+      "learning_rate": 2.1115829974884097e-06,
+      "loss": 0.7694,
+      "step": 898
+    },
+    {
+      "epoch": 1.4541414141414142,
+      "grad_norm": 3.609375,
+      "learning_rate": 2.1000641473462773e-06,
+      "loss": 0.8983,
+      "step": 899
+    },
+    {
+      "epoch": 1.4557575757575758,
+      "grad_norm": 3.578125,
+      "learning_rate": 2.0885684433280336e-06,
+      "loss": 0.8814,
+      "step": 900
+    },
+    {
+      "epoch": 1.4573737373737374,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.077095977187773e-06,
+      "loss": 0.6576,
+      "step": 901
+    },
+    {
+      "epoch": 1.458989898989899,
+      "grad_norm": 2.96875,
+      "learning_rate": 2.065646840494115e-06,
+      "loss": 0.5109,
+      "step": 902
+    },
+    {
+      "epoch": 1.4606060606060607,
+      "grad_norm": 3.484375,
+      "learning_rate": 2.054221124629473e-06,
+      "loss": 0.6154,
+      "step": 903
+    },
+    {
+      "epoch": 1.462222222222222,
+      "grad_norm": 3.234375,
+      "learning_rate": 2.042818920789326e-06,
+      "loss": 0.7635,
+      "step": 904
+    },
+    {
+      "epoch": 1.4638383838383837,
+      "grad_norm": 3.578125,
+      "learning_rate": 2.0314403199814846e-06,
+      "loss": 0.8321,
+      "step": 905
+    },
+    {
+      "epoch": 1.4654545454545453,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.020085413025375e-06,
+      "loss": 0.7735,
+      "step": 906
+    },
+    {
+      "epoch": 1.467070707070707,
+      "grad_norm": 3.109375,
+      "learning_rate": 2.008754290551306e-06,
+      "loss": 0.6717,
+      "step": 907
+    },
+    {
+      "epoch": 1.4686868686868686,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.9974470429997482e-06,
+      "loss": 1.1384,
+      "step": 908
+    },
+    {
+      "epoch": 1.4703030303030302,
+      "grad_norm": 3.296875,
+      "learning_rate": 1.9861637606206134e-06,
+      "loss": 0.6889,
+      "step": 909
+    },
+    {
+      "epoch": 1.4719191919191918,
+      "grad_norm": 4.03125,
+      "learning_rate": 1.974904533472532e-06,
+      "loss": 0.99,
+      "step": 910
+    },
+    {
+      "epoch": 1.4735353535353535,
+      "grad_norm": 3.140625,
+      "learning_rate": 1.9636694514221356e-06,
+      "loss": 0.5254,
+      "step": 911
+    },
+    {
+      "epoch": 1.475151515151515,
+      "grad_norm": 3.296875,
+      "learning_rate": 1.9524586041433393e-06,
+      "loss": 0.6115,
+      "step": 912
+    },
+    {
+      "epoch": 1.4767676767676767,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.9412720811166248e-06,
+      "loss": 0.86,
+      "step": 913
+    },
+    {
+      "epoch": 1.4783838383838384,
+      "grad_norm": 4.21875,
+      "learning_rate": 1.9301099716283293e-06,
+      "loss": 0.6302,
+      "step": 914
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 3.4375,
+      "learning_rate": 1.918972364769932e-06,
+      "loss": 0.5905,
+      "step": 915
+    },
+    {
+      "epoch": 1.4816161616161616,
+      "grad_norm": 3.296875,
+      "learning_rate": 1.907859349437336e-06,
+      "loss": 0.7295,
+      "step": 916
+    },
+    {
+      "epoch": 1.4832323232323232,
+      "grad_norm": 3.765625,
+      "learning_rate": 1.896771014330171e-06,
+      "loss": 0.7796,
+      "step": 917
+    },
+    {
+      "epoch": 1.4848484848484849,
+      "grad_norm": 3.21875,
+      "learning_rate": 1.8857074479510761e-06,
+      "loss": 0.7584,
+      "step": 918
+    },
+    {
+      "epoch": 1.4864646464646465,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.8746687386049972e-06,
+      "loss": 0.6619,
+      "step": 919
+    },
+    {
+      "epoch": 1.4880808080808081,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.8636549743984815e-06,
+      "loss": 0.7889,
+      "step": 920
+    },
+    {
+      "epoch": 1.4896969696969697,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.8526662432389752e-06,
+      "loss": 0.6925,
+      "step": 921
+    },
+    {
+      "epoch": 1.4913131313131314,
+      "grad_norm": 3.234375,
+      "learning_rate": 1.8417026328341158e-06,
+      "loss": 0.7704,
+      "step": 922
+    },
+    {
+      "epoch": 1.492929292929293,
+      "grad_norm": 3.796875,
+      "learning_rate": 1.8307642306910423e-06,
+      "loss": 0.6063,
+      "step": 923
+    },
+    {
+      "epoch": 1.4945454545454546,
+      "grad_norm": 3.328125,
+      "learning_rate": 1.8198511241156902e-06,
+      "loss": 0.6971,
+      "step": 924
+    },
+    {
+      "epoch": 1.4961616161616162,
+      "grad_norm": 3.328125,
+      "learning_rate": 1.8089634002120953e-06,
+      "loss": 0.8929,
+      "step": 925
+    },
+    {
+      "epoch": 1.4977777777777779,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.7981011458816988e-06,
+      "loss": 0.6117,
+      "step": 926
+    },
+    {
+      "epoch": 1.4993939393939395,
+      "grad_norm": 3.3125,
+      "learning_rate": 1.7872644478226563e-06,
+      "loss": 0.9486,
+      "step": 927
+    },
+    {
+      "epoch": 1.5010101010101011,
+      "grad_norm": 3.125,
+      "learning_rate": 1.776453392529139e-06,
+      "loss": 0.6738,
+      "step": 928
+    },
+    {
+      "epoch": 1.5026262626262628,
+      "grad_norm": 3.171875,
+      "learning_rate": 1.7656680662906495e-06,
+      "loss": 1.0356,
+      "step": 929
+    },
+    {
+      "epoch": 1.5042424242424244,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.7549085551913358e-06,
+      "loss": 0.6251,
+      "step": 930
+    },
+    {
+      "epoch": 1.505858585858586,
+      "grad_norm": 2.953125,
+      "learning_rate": 1.744174945109296e-06,
+      "loss": 0.8604,
+      "step": 931
+    },
+    {
+      "epoch": 1.5074747474747476,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.7334673217158976e-06,
+      "loss": 0.8673,
+      "step": 932
+    },
+    {
+      "epoch": 1.509090909090909,
+      "grad_norm": 2.875,
+      "learning_rate": 1.7227857704750888e-06,
+      "loss": 0.5662,
+      "step": 933
+    },
+    {
+      "epoch": 1.5107070707070707,
+      "grad_norm": 3.328125,
+      "learning_rate": 1.7121303766427266e-06,
+      "loss": 0.6247,
+      "step": 934
+    },
+    {
+      "epoch": 1.5123232323232323,
+      "grad_norm": 3.234375,
+      "learning_rate": 1.7015012252658864e-06,
+      "loss": 0.6032,
+      "step": 935
+    },
+    {
+      "epoch": 1.513939393939394,
+      "grad_norm": 3.46875,
+      "learning_rate": 1.6908984011821883e-06,
+      "loss": 0.967,
+      "step": 936
+    },
+    {
+      "epoch": 1.5155555555555555,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.6803219890191174e-06,
+      "loss": 0.9265,
+      "step": 937
+    },
+    {
+      "epoch": 1.5171717171717172,
+      "grad_norm": 3.53125,
+      "learning_rate": 1.669772073193352e-06,
+      "loss": 0.8432,
+      "step": 938
+    },
+    {
+      "epoch": 1.5187878787878788,
+      "grad_norm": 3.71875,
+      "learning_rate": 1.6592487379100819e-06,
+      "loss": 0.6279,
+      "step": 939
+    },
+    {
+      "epoch": 1.5204040404040404,
+      "grad_norm": 3.234375,
+      "learning_rate": 1.6487520671623469e-06,
+      "loss": 0.7296,
+      "step": 940
+    },
+    {
+      "epoch": 1.522020202020202,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.6382821447303593e-06,
+      "loss": 0.8051,
+      "step": 941
+    },
+    {
+      "epoch": 1.5236363636363637,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.6278390541808364e-06,
+      "loss": 0.6958,
+      "step": 942
+    },
+    {
+      "epoch": 1.5252525252525253,
+      "grad_norm": 3.0,
+      "learning_rate": 1.6174228788663366e-06,
+      "loss": 0.6439,
+      "step": 943
+    },
+    {
+      "epoch": 1.5268686868686867,
+      "grad_norm": 3.1875,
+      "learning_rate": 1.6070337019245896e-06,
+      "loss": 1.0541,
+      "step": 944
+    },
+    {
+      "epoch": 1.5284848484848483,
+      "grad_norm": 3.078125,
+      "learning_rate": 1.5966716062778336e-06,
+      "loss": 0.7799,
+      "step": 945
+    },
+    {
+      "epoch": 1.53010101010101,
+      "grad_norm": 3.25,
+      "learning_rate": 1.5863366746321578e-06,
+      "loss": 0.7188,
+      "step": 946
+    },
+    {
+      "epoch": 1.5317171717171716,
+      "grad_norm": 3.390625,
+      "learning_rate": 1.5760289894768377e-06,
+      "loss": 0.7883,
+      "step": 947
+    },
+    {
+      "epoch": 1.5333333333333332,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.5657486330836786e-06,
+      "loss": 0.9214,
+      "step": 948
+    },
+    {
+      "epoch": 1.5349494949494948,
+      "grad_norm": 3.453125,
+      "learning_rate": 1.5554956875063587e-06,
+      "loss": 0.7906,
+      "step": 949
+    },
+    {
+      "epoch": 1.5365656565656565,
+      "grad_norm": 3.53125,
+      "learning_rate": 1.5452702345797738e-06,
+      "loss": 0.908,
+      "step": 950
+    },
+    {
+      "epoch": 1.538181818181818,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.5350723559193837e-06,
+      "loss": 0.8251,
+      "step": 951
+    },
+    {
+      "epoch": 1.5397979797979797,
+      "grad_norm": 3.421875,
+      "learning_rate": 1.5249021329205638e-06,
+      "loss": 0.7924,
+      "step": 952
+    },
+    {
+      "epoch": 1.5414141414141413,
+      "grad_norm": 3.296875,
+      "learning_rate": 1.51475964675795e-06,
+      "loss": 0.9427,
+      "step": 953
+    },
+    {
+      "epoch": 1.543030303030303,
+      "grad_norm": 3.75,
+      "learning_rate": 1.5046449783847965e-06,
+      "loss": 0.73,
+      "step": 954
+    },
+    {
+      "epoch": 1.5446464646464646,
+      "grad_norm": 3.078125,
+      "learning_rate": 1.4945582085323275e-06,
+      "loss": 0.992,
+      "step": 955
+    },
+    {
+      "epoch": 1.5462626262626262,
+      "grad_norm": 3.515625,
+      "learning_rate": 1.4844994177090871e-06,
+      "loss": 0.6967,
+      "step": 956
+    },
+    {
+      "epoch": 1.5478787878787879,
+      "grad_norm": 3.171875,
+      "learning_rate": 1.4744686862003082e-06,
+      "loss": 1.0792,
+      "step": 957
+    },
+    {
+      "epoch": 1.5494949494949495,
+      "grad_norm": 3.1875,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 0.9407,
+      "step": 958
+    },
+    {
+      "epoch": 1.551111111111111,
+      "grad_norm": 2.921875,
+      "learning_rate": 1.4544917211466264e-06,
+      "loss": 0.8695,
+      "step": 959
+    },
+    {
+      "epoch": 1.5527272727272727,
+      "grad_norm": 3.1875,
+      "learning_rate": 1.4445456470498392e-06,
+      "loss": 0.7095,
+      "step": 960
+    },
+    {
+      "epoch": 1.5543434343434344,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.4346279511624728e-06,
+      "loss": 0.6074,
+      "step": 961
+    },
+    {
+      "epoch": 1.555959595959596,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.4247387126435957e-06,
+      "loss": 0.7463,
+      "step": 962
+    },
+    {
+      "epoch": 1.5575757575757576,
+      "grad_norm": 3.078125,
+      "learning_rate": 1.414878010425137e-06,
+      "loss": 0.947,
+      "step": 963
+    },
+    {
+      "epoch": 1.5591919191919192,
+      "grad_norm": 3.296875,
+      "learning_rate": 1.4050459232112652e-06,
+      "loss": 0.9163,
+      "step": 964
+    },
+    {
+      "epoch": 1.5608080808080809,
+      "grad_norm": 3.125,
+      "learning_rate": 1.3952425294777533e-06,
+      "loss": 0.7265,
+      "step": 965
+    },
+    {
+      "epoch": 1.5624242424242425,
+      "grad_norm": 3.359375,
+      "learning_rate": 1.3854679074713557e-06,
+      "loss": 0.858,
+      "step": 966
+    },
+    {
+      "epoch": 1.5640404040404041,
+      "grad_norm": 3.25,
+      "learning_rate": 1.3757221352091805e-06,
+      "loss": 0.8308,
+      "step": 967
+    },
+    {
+      "epoch": 1.5656565656565657,
+      "grad_norm": 3.03125,
+      "learning_rate": 1.3660052904780707e-06,
+      "loss": 1.0028,
+      "step": 968
+    },
+    {
+      "epoch": 1.5672727272727274,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.3563174508339793e-06,
+      "loss": 0.7451,
+      "step": 969
+    },
+    {
+      "epoch": 1.568888888888889,
+      "grad_norm": 3.125,
+      "learning_rate": 1.3466586936013548e-06,
+      "loss": 0.8512,
+      "step": 970
+    },
+    {
+      "epoch": 1.5705050505050506,
+      "grad_norm": 3.296875,
+      "learning_rate": 1.337029095872519e-06,
+      "loss": 0.8706,
+      "step": 971
+    },
+    {
+      "epoch": 1.5721212121212123,
+      "grad_norm": 3.078125,
+      "learning_rate": 1.3274287345070564e-06,
+      "loss": 0.7325,
+      "step": 972
+    },
+    {
+      "epoch": 1.5737373737373739,
+      "grad_norm": 3.171875,
+      "learning_rate": 1.3178576861311988e-06,
+      "loss": 0.7431,
+      "step": 973
+    },
+    {
+      "epoch": 1.5753535353535355,
+      "grad_norm": 3.5625,
+      "learning_rate": 1.3083160271372092e-06,
+      "loss": 0.9196,
+      "step": 974
+    },
+    {
+      "epoch": 1.5769696969696971,
+      "grad_norm": 2.953125,
+      "learning_rate": 1.2988038336827814e-06,
+      "loss": 0.56,
+      "step": 975
+    },
+    {
+      "epoch": 1.5785858585858585,
+      "grad_norm": 3.234375,
+      "learning_rate": 1.2893211816904243e-06,
+      "loss": 0.8315,
+      "step": 976
+    },
+    {
+      "epoch": 1.5802020202020202,
+      "grad_norm": 3.390625,
+      "learning_rate": 1.2798681468468605e-06,
+      "loss": 0.7115,
+      "step": 977
+    },
+    {
+      "epoch": 1.5818181818181818,
+      "grad_norm": 2.96875,
+      "learning_rate": 1.2704448046024192e-06,
+      "loss": 0.678,
+      "step": 978
+    },
+    {
+      "epoch": 1.5834343434343434,
+      "grad_norm": 3.328125,
+      "learning_rate": 1.261051230170437e-06,
+      "loss": 0.5541,
+      "step": 979
+    },
+    {
+      "epoch": 1.585050505050505,
+      "grad_norm": 3.125,
+      "learning_rate": 1.2516874985266508e-06,
+      "loss": 0.923,
+      "step": 980
+    },
+    {
+      "epoch": 1.5866666666666667,
+      "grad_norm": 3.1875,
+      "learning_rate": 1.2423536844086105e-06,
+      "loss": 0.7646,
+      "step": 981
+    },
+    {
+      "epoch": 1.5882828282828283,
+      "grad_norm": 3.25,
+      "learning_rate": 1.2330498623150722e-06,
+      "loss": 0.6332,
+      "step": 982
+    },
+    {
+      "epoch": 1.58989898989899,
+      "grad_norm": 3.03125,
+      "learning_rate": 1.2237761065054106e-06,
+      "loss": 0.6711,
+      "step": 983
+    },
+    {
+      "epoch": 1.5915151515151515,
+      "grad_norm": 3.046875,
+      "learning_rate": 1.2145324909990202e-06,
+      "loss": 0.7785,
+      "step": 984
+    },
+    {
+      "epoch": 1.5915151515151515,
+      "eval_loss": 0.8186196088790894,
+      "eval_runtime": 33.6691,
+      "eval_samples_per_second": 32.671,
+      "eval_steps_per_second": 4.099,
+      "step": 984
+    },
+    {
+      "epoch": 1.5931313131313132,
+      "grad_norm": 3.3125,
+      "learning_rate": 1.2053190895747324e-06,
+      "loss": 0.9328,
+      "step": 985
+    },
+    {
+      "epoch": 1.5947474747474748,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.1961359757702151e-06,
+      "loss": 0.8298,
+      "step": 986
+    },
+    {
+      "epoch": 1.5963636363636362,
+      "grad_norm": 3.1875,
+      "learning_rate": 1.1869832228813987e-06,
+      "loss": 0.767,
+      "step": 987
+    },
+    {
+      "epoch": 1.5979797979797978,
+      "grad_norm": 3.171875,
+      "learning_rate": 1.1778609039618804e-06,
+      "loss": 0.8414,
+      "step": 988
+    },
+    {
+      "epoch": 1.5995959595959595,
+      "grad_norm": 2.921875,
+      "learning_rate": 1.168769091822351e-06,
+      "loss": 0.5727,
+      "step": 989
+    },
+    {
+      "epoch": 1.601212121212121,
+      "grad_norm": 3.4375,
+      "learning_rate": 1.1597078590300054e-06,
+      "loss": 0.6862,
+      "step": 990
+    },
+    {
+      "epoch": 1.6028282828282827,
+      "grad_norm": 2.8125,
+      "learning_rate": 1.150677277907964e-06,
+      "loss": 0.7116,
+      "step": 991
+    },
+    {
+      "epoch": 1.6044444444444443,
+      "grad_norm": 2.921875,
+      "learning_rate": 1.1416774205347015e-06,
+      "loss": 0.6659,
+      "step": 992
+    },
+    {
+      "epoch": 1.606060606060606,
+      "grad_norm": 2.953125,
+      "learning_rate": 1.1327083587434662e-06,
+      "loss": 0.5345,
+      "step": 993
+    },
+    {
+      "epoch": 1.6076767676767676,
+      "grad_norm": 3.59375,
+      "learning_rate": 1.1237701641217097e-06,
+      "loss": 0.8891,
+      "step": 994
+    },
+    {
+      "epoch": 1.6092929292929292,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.1148629080105128e-06,
+      "loss": 1.2119,
+      "step": 995
+    },
+    {
+      "epoch": 1.6109090909090908,
+      "grad_norm": 3.578125,
+      "learning_rate": 1.1059866615040205e-06,
+      "loss": 0.9016,
+      "step": 996
+    },
+    {
+      "epoch": 1.6125252525252525,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.0971414954488657e-06,
+      "loss": 0.7799,
+      "step": 997
+    },
+    {
+      "epoch": 1.614141414141414,
+      "grad_norm": 3.078125,
+      "learning_rate": 1.0883274804436155e-06,
+      "loss": 0.7891,
+      "step": 998
+    },
+    {
+      "epoch": 1.6157575757575757,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.0795446868381987e-06,
+      "loss": 0.5814,
+      "step": 999
+    },
+    {
+      "epoch": 1.6173737373737374,
+      "grad_norm": 3.21875,
+      "learning_rate": 1.0707931847333487e-06,
+      "loss": 1.0249,
+      "step": 1000
+    },
+    {
+      "epoch": 1.618989898989899,
+      "grad_norm": 3.765625,
+      "learning_rate": 1.0620730439800424e-06,
+      "loss": 0.9242,
+      "step": 1001
+    },
+    {
+      "epoch": 1.6206060606060606,
+      "grad_norm": 2.875,
+      "learning_rate": 1.053384334178944e-06,
+      "loss": 0.7892,
+      "step": 1002
+    },
+    {
+      "epoch": 1.6222222222222222,
+      "grad_norm": 3.3125,
+      "learning_rate": 1.0447271246798446e-06,
+      "loss": 0.7965,
+      "step": 1003
+    },
+    {
+      "epoch": 1.6238383838383839,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.036101484581117e-06,
+      "loss": 0.9524,
+      "step": 1004
+    },
+    {
+      "epoch": 1.6254545454545455,
+      "grad_norm": 3.34375,
+      "learning_rate": 1.027507482729157e-06,
+      "loss": 1.0069,
+      "step": 1005
+    },
+    {
+      "epoch": 1.6270707070707071,
+      "grad_norm": 3.1875,
+      "learning_rate": 1.0189451877178386e-06,
+      "loss": 0.9097,
+      "step": 1006
+    },
+    {
+      "epoch": 1.6286868686868687,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.0104146678879633e-06,
+      "loss": 1.0461,
+      "step": 1007
+    },
+    {
+      "epoch": 1.6303030303030304,
+      "grad_norm": 3.234375,
+      "learning_rate": 1.0019159913267156e-06,
+      "loss": 0.7216,
+      "step": 1008
+    },
+    {
+      "epoch": 1.631919191919192,
+      "grad_norm": 3.15625,
+      "learning_rate": 9.934492258671214e-07,
+      "loss": 0.5974,
+      "step": 1009
+    },
+    {
+      "epoch": 1.6335353535353536,
+      "grad_norm": 2.921875,
+      "learning_rate": 9.850144390875022e-07,
+      "loss": 0.8383,
+      "step": 1010
+    },
+    {
+      "epoch": 1.6351515151515152,
+      "grad_norm": 3.328125,
+      "learning_rate": 9.76611698310942e-07,
+      "loss": 0.9198,
+      "step": 1011
+    },
+    {
+      "epoch": 1.6367676767676769,
+      "grad_norm": 3.421875,
+      "learning_rate": 9.68241070604743e-07,
+      "loss": 1.0553,
+      "step": 1012
+    },
+    {
+      "epoch": 1.6383838383838385,
+      "grad_norm": 3.421875,
+      "learning_rate": 9.59902622779898e-07,
+      "loss": 0.6781,
+      "step": 1013
+    },
+    {
+      "epoch": 1.6400000000000001,
+      "grad_norm": 3.125,
+      "learning_rate": 9.51596421390547e-07,
+      "loss": 0.7612,
+      "step": 1014
+    },
+    {
+      "epoch": 1.6416161616161618,
+      "grad_norm": 2.96875,
+      "learning_rate": 9.433225327334567e-07,
+      "loss": 0.9693,
+      "step": 1015
+    },
+    {
+      "epoch": 1.6432323232323234,
+      "grad_norm": 4.34375,
+      "learning_rate": 9.350810228474855e-07,
+      "loss": 0.7133,
+      "step": 1016
+    },
+    {
+      "epoch": 1.644848484848485,
+      "grad_norm": 3.0625,
+      "learning_rate": 9.268719575130563e-07,
+      "loss": 0.7443,
+      "step": 1017
+    },
+    {
+      "epoch": 1.6464646464646466,
+      "grad_norm": 3.640625,
+      "learning_rate": 9.186954022516343e-07,
+      "loss": 0.8859,
+      "step": 1018
+    },
+    {
+      "epoch": 1.648080808080808,
+      "grad_norm": 3.625,
+      "learning_rate": 9.105514223252016e-07,
+      "loss": 0.9868,
+      "step": 1019
+    },
+    {
+      "epoch": 1.6496969696969697,
+      "grad_norm": 3.34375,
+      "learning_rate": 9.024400827357344e-07,
+      "loss": 0.9599,
+      "step": 1020
+    },
+    {
+      "epoch": 1.6513131313131313,
+      "grad_norm": 3.265625,
+      "learning_rate": 8.943614482246915e-07,
+      "loss": 0.7457,
+      "step": 1021
+    },
+    {
+      "epoch": 1.652929292929293,
+      "grad_norm": 3.9375,
+      "learning_rate": 8.863155832724895e-07,
+      "loss": 0.8587,
+      "step": 1022
+    },
+    {
+      "epoch": 1.6545454545454545,
+      "grad_norm": 3.15625,
+      "learning_rate": 8.783025520979932e-07,
+      "loss": 0.7854,
+      "step": 1023
+    },
+    {
+      "epoch": 1.6561616161616162,
+      "grad_norm": 3.34375,
+      "learning_rate": 8.703224186580012e-07,
+      "loss": 1.0456,
+      "step": 1024
+    },
+    {
+      "epoch": 1.6577777777777778,
+      "grad_norm": 2.984375,
+      "learning_rate": 8.623752466467344e-07,
+      "loss": 0.8111,
+      "step": 1025
+    },
+    {
+      "epoch": 1.6593939393939394,
+      "grad_norm": 3.1875,
+      "learning_rate": 8.544610994953317e-07,
+      "loss": 0.842,
+      "step": 1026
+    },
+    {
+      "epoch": 1.661010101010101,
+      "grad_norm": 3.4375,
+      "learning_rate": 8.465800403713365e-07,
+      "loss": 0.8471,
+      "step": 1027
+    },
+    {
+      "epoch": 1.6626262626262627,
+      "grad_norm": 3.546875,
+      "learning_rate": 8.387321321781977e-07,
+      "loss": 0.7155,
+      "step": 1028
+    },
+    {
+      "epoch": 1.664242424242424,
+      "grad_norm": 2.984375,
+      "learning_rate": 8.309174375547724e-07,
+      "loss": 0.9536,
+      "step": 1029
+    },
+    {
+      "epoch": 1.6658585858585857,
+      "grad_norm": 2.984375,
+      "learning_rate": 8.23136018874815e-07,
+      "loss": 0.5539,
+      "step": 1030
+    },
+    {
+      "epoch": 1.6674747474747473,
+      "grad_norm": 3.03125,
+      "learning_rate": 8.153879382464836e-07,
+      "loss": 0.7364,
+      "step": 1031
+    },
+    {
+      "epoch": 1.669090909090909,
+      "grad_norm": 3.046875,
+      "learning_rate": 8.07673257511849e-07,
+      "loss": 0.873,
+      "step": 1032
+    },
+    {
+      "epoch": 1.6707070707070706,
+      "grad_norm": 3.578125,
+      "learning_rate": 7.999920382463955e-07,
+      "loss": 0.6453,
+      "step": 1033
+    },
+    {
+      "epoch": 1.6723232323232322,
+      "grad_norm": 3.375,
+      "learning_rate": 7.923443417585324e-07,
+      "loss": 0.7421,
+      "step": 1034
+    },
+    {
+      "epoch": 1.6739393939393938,
+      "grad_norm": 3.359375,
+      "learning_rate": 7.847302290891029e-07,
+      "loss": 1.0101,
+      "step": 1035
+    },
+    {
+      "epoch": 1.6755555555555555,
+      "grad_norm": 3.015625,
+      "learning_rate": 7.771497610108981e-07,
+      "loss": 0.9205,
+      "step": 1036
+    },
+    {
+      "epoch": 1.677171717171717,
+      "grad_norm": 3.6875,
+      "learning_rate": 7.696029980281722e-07,
+      "loss": 1.0818,
+      "step": 1037
+    },
+    {
+      "epoch": 1.6787878787878787,
+      "grad_norm": 3.0,
+      "learning_rate": 7.620900003761561e-07,
+      "loss": 0.8263,
+      "step": 1038
+    },
+    {
+      "epoch": 1.6804040404040403,
+      "grad_norm": 3.15625,
+      "learning_rate": 7.546108280205816e-07,
+      "loss": 0.8513,
+      "step": 1039
+    },
+    {
+      "epoch": 1.682020202020202,
+      "grad_norm": 2.9375,
+      "learning_rate": 7.471655406572003e-07,
+      "loss": 0.9792,
+      "step": 1040
+    },
+    {
+      "epoch": 1.6836363636363636,
+      "grad_norm": 3.21875,
+      "learning_rate": 7.397541977113076e-07,
+      "loss": 0.8123,
+      "step": 1041
+    },
+    {
+      "epoch": 1.6852525252525252,
+      "grad_norm": 2.75,
+      "learning_rate": 7.323768583372675e-07,
+      "loss": 0.9469,
+      "step": 1042
+    },
+    {
+      "epoch": 1.6868686868686869,
+      "grad_norm": 3.34375,
+      "learning_rate": 7.250335814180432e-07,
+      "loss": 0.6794,
+      "step": 1043
+    },
+    {
+      "epoch": 1.6884848484848485,
+      "grad_norm": 3.125,
+      "learning_rate": 7.177244255647209e-07,
+      "loss": 0.8688,
+      "step": 1044
+    },
+    {
+      "epoch": 1.69010101010101,
+      "grad_norm": 3.0,
+      "learning_rate": 7.10449449116049e-07,
+      "loss": 1.0007,
+      "step": 1045
+    },
+    {
+      "epoch": 1.6917171717171717,
+      "grad_norm": 3.125,
+      "learning_rate": 7.032087101379703e-07,
+      "loss": 0.7705,
+      "step": 1046
+    },
+    {
+      "epoch": 1.6933333333333334,
+      "grad_norm": 3.28125,
+      "learning_rate": 6.960022664231558e-07,
+      "loss": 0.8262,
+      "step": 1047
+    },
+    {
+      "epoch": 1.694949494949495,
+      "grad_norm": 3.046875,
+      "learning_rate": 6.888301754905469e-07,
+      "loss": 0.8702,
+      "step": 1048
+    },
+    {
+      "epoch": 1.6965656565656566,
+      "grad_norm": 3.4375,
+      "learning_rate": 6.816924945848941e-07,
+      "loss": 0.9841,
+      "step": 1049
+    },
+    {
+      "epoch": 1.6981818181818182,
+      "grad_norm": 2.9375,
+      "learning_rate": 6.745892806763015e-07,
+      "loss": 0.8487,
+      "step": 1050
+    },
+    {
+      "epoch": 1.6997979797979799,
+      "grad_norm": 3.015625,
+      "learning_rate": 6.675205904597715e-07,
+      "loss": 0.7953,
+      "step": 1051
+    },
+    {
+      "epoch": 1.7014141414141415,
+      "grad_norm": 3.203125,
+      "learning_rate": 6.604864803547511e-07,
+      "loss": 0.71,
+      "step": 1052
+    },
+    {
+      "epoch": 1.7030303030303031,
+      "grad_norm": 3.25,
+      "learning_rate": 6.534870065046839e-07,
+      "loss": 0.6827,
+      "step": 1053
+    },
+    {
+      "epoch": 1.7046464646464647,
+      "grad_norm": 3.578125,
+      "learning_rate": 6.465222247765618e-07,
+      "loss": 0.8225,
+      "step": 1054
+    },
+    {
+      "epoch": 1.7062626262626264,
+      "grad_norm": 3.078125,
+      "learning_rate": 6.395921907604747e-07,
+      "loss": 0.8645,
+      "step": 1055
+    },
+    {
+      "epoch": 1.707878787878788,
+      "grad_norm": 3.109375,
+      "learning_rate": 6.326969597691724e-07,
+      "loss": 0.8962,
+      "step": 1056
+    },
+    {
+      "epoch": 1.7094949494949496,
+      "grad_norm": 2.828125,
+      "learning_rate": 6.258365868376215e-07,
+      "loss": 0.8968,
+      "step": 1057
+    },
+    {
+      "epoch": 1.7111111111111112,
+      "grad_norm": 3.421875,
+      "learning_rate": 6.190111267225651e-07,
+      "loss": 0.7882,
+      "step": 1058
+    },
+    {
+      "epoch": 1.7127272727272729,
+      "grad_norm": 3.09375,
+      "learning_rate": 6.122206339020854e-07,
+      "loss": 0.6881,
+      "step": 1059
+    },
+    {
+      "epoch": 1.7143434343434345,
+      "grad_norm": 3.28125,
+      "learning_rate": 6.054651625751717e-07,
+      "loss": 0.856,
+      "step": 1060
+    },
+    {
+      "epoch": 1.715959595959596,
+      "grad_norm": 3.046875,
+      "learning_rate": 5.987447666612823e-07,
+      "loss": 0.9082,
+      "step": 1061
+    },
+    {
+      "epoch": 1.7175757575757575,
+      "grad_norm": 3.296875,
+      "learning_rate": 5.920594997999202e-07,
+      "loss": 0.7142,
+      "step": 1062
+    },
+    {
+      "epoch": 1.7191919191919192,
+      "grad_norm": 3.0625,
+      "learning_rate": 5.854094153502022e-07,
+      "loss": 0.9069,
+      "step": 1063
+    },
+    {
+      "epoch": 1.7208080808080808,
+      "grad_norm": 2.953125,
+      "learning_rate": 5.787945663904332e-07,
+      "loss": 0.798,
+      "step": 1064
+    },
+    {
+      "epoch": 1.7224242424242424,
+      "grad_norm": 3.578125,
+      "learning_rate": 5.722150057176823e-07,
+      "loss": 0.7055,
+      "step": 1065
+    },
+    {
+      "epoch": 1.724040404040404,
+      "grad_norm": 3.390625,
+      "learning_rate": 5.65670785847362e-07,
+      "loss": 0.7295,
+      "step": 1066
+    },
+    {
+      "epoch": 1.7256565656565657,
+      "grad_norm": 3.09375,
+      "learning_rate": 5.591619590128067e-07,
+      "loss": 0.6732,
+      "step": 1067
+    },
+    {
+      "epoch": 1.7272727272727273,
+      "grad_norm": 3.34375,
+      "learning_rate": 5.526885771648599e-07,
+      "loss": 0.6679,
+      "step": 1068
+    },
+    {
+      "epoch": 1.728888888888889,
+      "grad_norm": 3.078125,
+      "learning_rate": 5.462506919714583e-07,
+      "loss": 0.607,
+      "step": 1069
+    },
+    {
+      "epoch": 1.7305050505050505,
+      "grad_norm": 3.1875,
+      "learning_rate": 5.398483548172162e-07,
+      "loss": 0.921,
+      "step": 1070
+    },
+    {
+      "epoch": 1.7321212121212122,
+      "grad_norm": 3.1875,
+      "learning_rate": 5.334816168030204e-07,
+      "loss": 0.8394,
+      "step": 1071
+    },
+    {
+      "epoch": 1.7337373737373736,
+      "grad_norm": 2.90625,
+      "learning_rate": 5.271505287456153e-07,
+      "loss": 0.7587,
+      "step": 1072
+    },
+    {
+      "epoch": 1.7353535353535352,
+      "grad_norm": 3.125,
+      "learning_rate": 5.208551411772056e-07,
+      "loss": 0.6143,
+      "step": 1073
+    },
+    {
+      "epoch": 1.7369696969696968,
+      "grad_norm": 3.46875,
+      "learning_rate": 5.145955043450484e-07,
+      "loss": 1.1425,
+      "step": 1074
+    },
+    {
+      "epoch": 1.7385858585858585,
+      "grad_norm": 3.546875,
+      "learning_rate": 5.083716682110512e-07,
+      "loss": 0.5865,
+      "step": 1075
+    },
+    {
+      "epoch": 1.74020202020202,
+      "grad_norm": 3.09375,
+      "learning_rate": 5.021836824513759e-07,
+      "loss": 0.6178,
+      "step": 1076
+    },
+    {
+      "epoch": 1.7418181818181817,
+      "grad_norm": 3.375,
+      "learning_rate": 4.960315964560425e-07,
+      "loss": 0.7236,
+      "step": 1077
+    },
+    {
+      "epoch": 1.7434343434343433,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.899154593285294e-07,
+      "loss": 0.5867,
+      "step": 1078
+    },
+    {
+      "epoch": 1.745050505050505,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.838353198853884e-07,
+      "loss": 0.5361,
+      "step": 1079
+    },
+    {
+      "epoch": 1.7466666666666666,
+      "grad_norm": 3.21875,
+      "learning_rate": 4.777912266558532e-07,
+      "loss": 1.073,
+      "step": 1080
+    },
+    {
+      "epoch": 1.7482828282828282,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.717832278814488e-07,
+      "loss": 1.0095,
+      "step": 1081
+    },
+    {
+      "epoch": 1.7498989898989898,
+      "grad_norm": 2.953125,
+      "learning_rate": 4.658113715156098e-07,
+      "loss": 1.0493,
+      "step": 1082
+    },
+    {
+      "epoch": 1.7515151515151515,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.5987570522329773e-07,
+      "loss": 0.7496,
+      "step": 1083
+    },
+    {
+      "epoch": 1.753131313131313,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.5397627638061604e-07,
+      "loss": 0.502,
+      "step": 1084
+    },
+    {
+      "epoch": 1.7547474747474747,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.481131320744386e-07,
+      "loss": 0.8937,
+      "step": 1085
+    },
+    {
+      "epoch": 1.7563636363636363,
+      "grad_norm": 3.203125,
+      "learning_rate": 4.422863191020299e-07,
+      "loss": 0.7344,
+      "step": 1086
+    },
+    {
+      "epoch": 1.757979797979798,
+      "grad_norm": 3.265625,
+      "learning_rate": 4.3649588397067177e-07,
+      "loss": 0.7553,
+      "step": 1087
+    },
+    {
+      "epoch": 1.7595959595959596,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.307418728972934e-07,
+      "loss": 1.0243,
+      "step": 1088
+    },
+    {
+      "epoch": 1.7612121212121212,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.2502433180810153e-07,
+      "loss": 0.5675,
+      "step": 1089
+    },
+    {
+      "epoch": 1.7628282828282829,
+      "grad_norm": 3.25,
+      "learning_rate": 4.193433063382135e-07,
+      "loss": 1.0319,
+      "step": 1090
+    },
+    {
+      "epoch": 1.7644444444444445,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.1369884183129485e-07,
+      "loss": 1.2346,
+      "step": 1091
+    },
+    {
+      "epoch": 1.766060606060606,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.080909833391944e-07,
+      "loss": 0.8378,
+      "step": 1092
+    },
+    {
+      "epoch": 1.7676767676767677,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.0251977562158873e-07,
+      "loss": 1.0959,
+      "step": 1093
+    },
+    {
+      "epoch": 1.7692929292929294,
+      "grad_norm": 3.5,
+      "learning_rate": 3.9698526314562114e-07,
+      "loss": 1.2264,
+      "step": 1094
+    },
+    {
+      "epoch": 1.770909090909091,
+      "grad_norm": 3.359375,
+      "learning_rate": 3.9148749008554807e-07,
+      "loss": 0.9057,
+      "step": 1095
+    },
+    {
+      "epoch": 1.7725252525252526,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.8602650032238675e-07,
+      "loss": 0.8186,
+      "step": 1096
+    },
+    {
+      "epoch": 1.7741414141414142,
+      "grad_norm": 3.40625,
+      "learning_rate": 3.8060233744356634e-07,
+      "loss": 0.9942,
+      "step": 1097
+    },
+    {
+      "epoch": 1.7757575757575759,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.752150447425773e-07,
+      "loss": 0.7508,
+      "step": 1098
+    },
+    {
+      "epoch": 1.7773737373737375,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.698646652186272e-07,
+      "loss": 1.2919,
+      "step": 1099
+    },
+    {
+      "epoch": 1.7789898989898991,
+      "grad_norm": 3.109375,
+      "learning_rate": 3.6455124157629805e-07,
+      "loss": 0.8741,
+      "step": 1100
+    },
+    {
+      "epoch": 1.7806060606060607,
+      "grad_norm": 3.5,
+      "learning_rate": 3.5927481622520463e-07,
+      "loss": 0.7342,
+      "step": 1101
+    },
+    {
+      "epoch": 1.7822222222222224,
+      "grad_norm": 3.03125,
+      "learning_rate": 3.5403543127965514e-07,
+      "loss": 0.994,
+      "step": 1102
+    },
+    {
+      "epoch": 1.783838383838384,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.488331285583174e-07,
+      "loss": 0.7645,
+      "step": 1103
+    },
+    {
+      "epoch": 1.7854545454545454,
+      "grad_norm": 2.984375,
+      "learning_rate": 3.436679495838835e-07,
+      "loss": 0.5207,
+      "step": 1104
+    },
+    {
+      "epoch": 1.787070707070707,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.38539935582739e-07,
+      "loss": 0.5003,
+      "step": 1105
+    },
+    {
+      "epoch": 1.7886868686868687,
+      "grad_norm": 2.828125,
+      "learning_rate": 3.334491274846324e-07,
+      "loss": 0.619,
+      "step": 1106
+    },
+    {
+      "epoch": 1.7903030303030303,
+      "grad_norm": 3.125,
+      "learning_rate": 3.283955659223509e-07,
+      "loss": 0.7233,
+      "step": 1107
+    },
+    {
+      "epoch": 1.7903030303030303,
+      "eval_loss": 0.8177520036697388,
+      "eval_runtime": 33.6579,
+      "eval_samples_per_second": 32.682,
+      "eval_steps_per_second": 4.1,
+      "step": 1107
+    },
+    {
+      "epoch": 1.791919191919192,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.233792912313943e-07,
+      "loss": 0.8822,
+      "step": 1108
+    },
+    {
+      "epoch": 1.7935353535353535,
+      "grad_norm": 3.1875,
+      "learning_rate": 3.1840034344965355e-07,
+      "loss": 0.5735,
+      "step": 1109
+    },
+    {
+      "epoch": 1.7951515151515152,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.134587623170909e-07,
+      "loss": 0.8211,
+      "step": 1110
+    },
+    {
+      "epoch": 1.7967676767676768,
+      "grad_norm": 3.0625,
+      "learning_rate": 3.085545872754242e-07,
+      "loss": 0.7394,
+      "step": 1111
+    },
+    {
+      "epoch": 1.7983838383838384,
+      "grad_norm": 3.671875,
+      "learning_rate": 3.0368785746780925e-07,
+      "loss": 0.654,
+      "step": 1112
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 2.875,
+      "learning_rate": 2.9885861173852946e-07,
+      "loss": 0.8044,
+      "step": 1113
+    },
+    {
+      "epoch": 1.8016161616161617,
+      "grad_norm": 3.234375,
+      "learning_rate": 2.940668886326864e-07,
+      "loss": 0.8558,
+      "step": 1114
+    },
+    {
+      "epoch": 1.803232323232323,
+      "grad_norm": 3.25,
+      "learning_rate": 2.8931272639589005e-07,
+      "loss": 0.6761,
+      "step": 1115
+    },
+    {
+      "epoch": 1.8048484848484847,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.8459616297395464e-07,
+      "loss": 0.6599,
+      "step": 1116
+    },
+    {
+      "epoch": 1.8064646464646463,
+      "grad_norm": 2.9375,
+      "learning_rate": 2.7991723601259677e-07,
+      "loss": 0.7911,
+      "step": 1117
+    },
+    {
+      "epoch": 1.808080808080808,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.7527598285713387e-07,
+      "loss": 0.7821,
+      "step": 1118
+    },
+    {
+      "epoch": 1.8096969696969696,
+      "grad_norm": 3.234375,
+      "learning_rate": 2.706724405521838e-07,
+      "loss": 0.6494,
+      "step": 1119
+    },
+    {
+      "epoch": 1.8113131313131312,
+      "grad_norm": 3.53125,
+      "learning_rate": 2.6610664584137413e-07,
+      "loss": 0.9946,
+      "step": 1120
+    },
+    {
+      "epoch": 1.8129292929292928,
+      "grad_norm": 3.34375,
+      "learning_rate": 2.615786351670452e-07,
+      "loss": 0.94,
+      "step": 1121
+    },
+    {
+      "epoch": 1.8145454545454545,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.570884446699612e-07,
+      "loss": 0.7691,
+      "step": 1122
+    },
+    {
+      "epoch": 1.816161616161616,
+      "grad_norm": 3.03125,
+      "learning_rate": 2.5263611018901957e-07,
+      "loss": 0.7532,
+      "step": 1123
+    },
+    {
+      "epoch": 1.8177777777777777,
+      "grad_norm": 3.015625,
+      "learning_rate": 2.482216672609677e-07,
+      "loss": 0.7615,
+      "step": 1124
+    },
+    {
+      "epoch": 1.8193939393939393,
+      "grad_norm": 3.328125,
+      "learning_rate": 2.438451511201162e-07,
+      "loss": 0.8134,
+      "step": 1125
+    },
+    {
+      "epoch": 1.821010101010101,
+      "grad_norm": 3.125,
+      "learning_rate": 2.3950659669806033e-07,
+      "loss": 0.643,
+      "step": 1126
+    },
+    {
+      "epoch": 1.8226262626262626,
+      "grad_norm": 2.828125,
+      "learning_rate": 2.3520603862339952e-07,
+      "loss": 0.6166,
+      "step": 1127
+    },
+    {
+      "epoch": 1.8242424242424242,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.3094351122146307e-07,
+      "loss": 0.9321,
+      "step": 1128
+    },
+    {
+      "epoch": 1.8258585858585858,
+      "grad_norm": 3.109375,
+      "learning_rate": 2.2671904851403303e-07,
+      "loss": 1.0975,
+      "step": 1129
+    },
+    {
+      "epoch": 1.8274747474747475,
+      "grad_norm": 2.984375,
+      "learning_rate": 2.2253268421907503e-07,
+      "loss": 0.706,
+      "step": 1130
+    },
+    {
+      "epoch": 1.829090909090909,
+      "grad_norm": 3.125,
+      "learning_rate": 2.1838445175046796e-07,
+      "loss": 0.9068,
+      "step": 1131
+    },
+    {
+      "epoch": 1.8307070707070707,
+      "grad_norm": 3.203125,
+      "learning_rate": 2.142743842177386e-07,
+      "loss": 0.9609,
+      "step": 1132
+    },
+    {
+      "epoch": 1.8323232323232324,
+      "grad_norm": 3.203125,
+      "learning_rate": 2.1020251442579632e-07,
+      "loss": 0.8877,
+      "step": 1133
+    },
+    {
+      "epoch": 1.833939393939394,
+      "grad_norm": 3.5,
+      "learning_rate": 2.0616887487467096e-07,
+      "loss": 0.8619,
+      "step": 1134
+    },
+    {
+      "epoch": 1.8355555555555556,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.0217349775925487e-07,
+      "loss": 0.816,
+      "step": 1135
+    },
+    {
+      "epoch": 1.8371717171717172,
+      "grad_norm": 2.953125,
+      "learning_rate": 1.98216414969043e-07,
+      "loss": 0.5237,
+      "step": 1136
+    },
+    {
+      "epoch": 1.8387878787878789,
+      "grad_norm": 3.234375,
+      "learning_rate": 1.942976580878825e-07,
+      "loss": 0.8957,
+      "step": 1137
+    },
+    {
+      "epoch": 1.8404040404040405,
+      "grad_norm": 3.25,
+      "learning_rate": 1.9041725839371805e-07,
+      "loss": 0.585,
+      "step": 1138
+    },
+    {
+      "epoch": 1.8420202020202021,
+      "grad_norm": 3.140625,
+      "learning_rate": 1.8657524685834206e-07,
+      "loss": 0.9216,
+      "step": 1139
+    },
+    {
+      "epoch": 1.8436363636363637,
+      "grad_norm": 3.296875,
+      "learning_rate": 1.8277165414714858e-07,
+      "loss": 0.7426,
+      "step": 1140
+    },
+    {
+      "epoch": 1.8452525252525254,
+      "grad_norm": 3.421875,
+      "learning_rate": 1.7900651061888762e-07,
+      "loss": 0.9956,
+      "step": 1141
+    },
+    {
+      "epoch": 1.846868686868687,
+      "grad_norm": 2.859375,
+      "learning_rate": 1.752798463254235e-07,
+      "loss": 0.7815,
+      "step": 1142
+    },
+    {
+      "epoch": 1.8484848484848486,
+      "grad_norm": 2.828125,
+      "learning_rate": 1.7159169101149398e-07,
+      "loss": 0.8854,
+      "step": 1143
+    },
+    {
+      "epoch": 1.8501010101010102,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.6794207411447548e-07,
+      "loss": 1.0409,
+      "step": 1144
+    },
+    {
+      "epoch": 1.8517171717171719,
+      "grad_norm": 3.046875,
+      "learning_rate": 1.6433102476414386e-07,
+      "loss": 0.8278,
+      "step": 1145
+    },
+    {
+      "epoch": 1.8533333333333335,
+      "grad_norm": 2.890625,
+      "learning_rate": 1.6075857178244613e-07,
+      "loss": 1.1842,
+      "step": 1146
+    },
+    {
+      "epoch": 1.854949494949495,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.5722474368326802e-07,
+      "loss": 0.6928,
+      "step": 1147
+    },
+    {
+      "epoch": 1.8565656565656565,
+      "grad_norm": 3.296875,
+      "learning_rate": 1.5372956867220678e-07,
+      "loss": 1.0738,
+      "step": 1148
+    },
+    {
+      "epoch": 1.8581818181818182,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.5027307464634656e-07,
+      "loss": 0.9691,
+      "step": 1149
+    },
+    {
+      "epoch": 1.8597979797979798,
+      "grad_norm": 2.796875,
+      "learning_rate": 1.4685528919403446e-07,
+      "loss": 0.5858,
+      "step": 1150
+    },
+    {
+      "epoch": 1.8614141414141414,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.434762395946626e-07,
+      "loss": 0.8529,
+      "step": 1151
+    },
+    {
+      "epoch": 1.863030303030303,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.4013595281844872e-07,
+      "loss": 0.5898,
+      "step": 1152
+    },
+    {
+      "epoch": 1.8646464646464647,
+      "grad_norm": 3.171875,
+      "learning_rate": 1.368344555262202e-07,
+      "loss": 0.7817,
+      "step": 1153
+    },
+    {
+      "epoch": 1.8662626262626263,
+      "grad_norm": 3.078125,
+      "learning_rate": 1.335717740692033e-07,
+      "loss": 0.7894,
+      "step": 1154
+    },
+    {
+      "epoch": 1.867878787878788,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.30347934488812e-07,
+      "loss": 0.7156,
+      "step": 1155
+    },
+    {
+      "epoch": 1.8694949494949495,
+      "grad_norm": 3.0,
+      "learning_rate": 1.2716296251644e-07,
+      "loss": 0.648,
+      "step": 1156
+    },
+    {
+      "epoch": 1.871111111111111,
+      "grad_norm": 3.1875,
+      "learning_rate": 1.2401688357325415e-07,
+      "loss": 0.8612,
+      "step": 1157
+    },
+    {
+      "epoch": 1.8727272727272726,
+      "grad_norm": 2.953125,
+      "learning_rate": 1.2090972276999513e-07,
+      "loss": 0.6399,
+      "step": 1158
+    },
+    {
+      "epoch": 1.8743434343434342,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.1784150490677215e-07,
+      "loss": 0.7838,
+      "step": 1159
+    },
+    {
+      "epoch": 1.8759595959595958,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.1481225447286803e-07,
+      "loss": 0.7164,
+      "step": 1160
+    },
+    {
+      "epoch": 1.8775757575757575,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.1182199564654495e-07,
+      "loss": 0.9411,
+      "step": 1161
+    },
+    {
+      "epoch": 1.879191919191919,
+      "grad_norm": 3.375,
+      "learning_rate": 1.0887075229484789e-07,
+      "loss": 0.5863,
+      "step": 1162
+    },
+    {
+      "epoch": 1.8808080808080807,
+      "grad_norm": 3.359375,
+      "learning_rate": 1.0595854797341709e-07,
+      "loss": 0.8106,
+      "step": 1163
+    },
+    {
+      "epoch": 1.8824242424242423,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.0308540592629756e-07,
+      "loss": 0.8153,
+      "step": 1164
+    },
+    {
+      "epoch": 1.884040404040404,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.0025134908575707e-07,
+      "loss": 0.5583,
+      "step": 1165
+    },
+    {
+      "epoch": 1.8856565656565656,
+      "grad_norm": 3.265625,
+      "learning_rate": 9.745640007209844e-08,
+      "loss": 0.8847,
+      "step": 1166
+    },
+    {
+      "epoch": 1.8872727272727272,
+      "grad_norm": 3.1875,
+      "learning_rate": 9.470058119348257e-08,
+      "loss": 0.8772,
+      "step": 1167
+    },
+    {
+      "epoch": 1.8888888888888888,
+      "grad_norm": 3.203125,
+      "learning_rate": 9.198391444575072e-08,
+      "loss": 0.9593,
+      "step": 1168
+    },
+    {
+      "epoch": 1.8905050505050505,
+      "grad_norm": 3.3125,
+      "learning_rate": 8.930642151224634e-08,
+      "loss": 0.6201,
+      "step": 1169
+    },
+    {
+      "epoch": 1.892121212121212,
+      "grad_norm": 3.265625,
+      "learning_rate": 8.666812376364187e-08,
+      "loss": 0.6472,
+      "step": 1170
+    },
+    {
+      "epoch": 1.8937373737373737,
+      "grad_norm": 3.078125,
+      "learning_rate": 8.406904225777223e-08,
+      "loss": 0.8019,
+      "step": 1171
+    },
+    {
+      "epoch": 1.8953535353535353,
+      "grad_norm": 3.125,
+      "learning_rate": 8.150919773946165e-08,
+      "loss": 0.7295,
+      "step": 1172
+    },
+    {
+      "epoch": 1.896969696969697,
+      "grad_norm": 3.59375,
+      "learning_rate": 7.898861064036257e-08,
+      "loss": 1.0313,
+      "step": 1173
+    },
+    {
+      "epoch": 1.8985858585858586,
+      "grad_norm": 3.453125,
+      "learning_rate": 7.650730107878812e-08,
+      "loss": 0.809,
+      "step": 1174
+    },
+    {
+      "epoch": 1.9002020202020202,
+      "grad_norm": 3.265625,
+      "learning_rate": 7.406528885955665e-08,
+      "loss": 0.71,
+      "step": 1175
+    },
+    {
+      "epoch": 1.9018181818181819,
+      "grad_norm": 3.421875,
+      "learning_rate": 7.166259347382854e-08,
+      "loss": 0.6814,
+      "step": 1176
+    },
+    {
+      "epoch": 1.9034343434343435,
+      "grad_norm": 2.96875,
+      "learning_rate": 6.929923409895345e-08,
+      "loss": 0.6194,
+      "step": 1177
+    },
+    {
+      "epoch": 1.905050505050505,
+      "grad_norm": 3.328125,
+      "learning_rate": 6.697522959831837e-08,
+      "loss": 0.7504,
+      "step": 1178
+    },
+    {
+      "epoch": 1.9066666666666667,
+      "grad_norm": 3.046875,
+      "learning_rate": 6.46905985211932e-08,
+      "loss": 0.7204,
+      "step": 1179
+    },
+    {
+      "epoch": 1.9082828282828284,
+      "grad_norm": 3.0,
+      "learning_rate": 6.244535910258697e-08,
+      "loss": 0.98,
+      "step": 1180
+    },
+    {
+      "epoch": 1.90989898989899,
+      "grad_norm": 3.15625,
+      "learning_rate": 6.023952926309917e-08,
+      "loss": 0.8714,
+      "step": 1181
+    },
+    {
+      "epoch": 1.9115151515151516,
+      "grad_norm": 3.015625,
+      "learning_rate": 5.8073126608778064e-08,
+      "loss": 0.6274,
+      "step": 1182
+    },
+    {
+      "epoch": 1.9131313131313132,
+      "grad_norm": 3.46875,
+      "learning_rate": 5.594616843097978e-08,
+      "loss": 0.7374,
+      "step": 1183
+    },
+    {
+      "epoch": 1.9147474747474749,
+      "grad_norm": 3.25,
+      "learning_rate": 5.3858671706230605e-08,
+      "loss": 0.6138,
+      "step": 1184
+    },
+    {
+      "epoch": 1.9163636363636365,
+      "grad_norm": 3.4375,
+      "learning_rate": 5.181065309609212e-08,
+      "loss": 0.6107,
+      "step": 1185
+    },
+    {
+      "epoch": 1.9179797979797981,
+      "grad_norm": 2.859375,
+      "learning_rate": 4.98021289470263e-08,
+      "loss": 0.4929,
+      "step": 1186
+    },
+    {
+      "epoch": 1.9195959595959597,
+      "grad_norm": 2.90625,
+      "learning_rate": 4.7833115290266686e-08,
+      "loss": 0.6894,
+      "step": 1187
+    },
+    {
+      "epoch": 1.9212121212121214,
+      "grad_norm": 2.9375,
+      "learning_rate": 4.590362784169022e-08,
+      "loss": 0.8471,
+      "step": 1188
+    },
+    {
+      "epoch": 1.9228282828282828,
+      "grad_norm": 3.0625,
+      "learning_rate": 4.401368200169009e-08,
+      "loss": 0.9233,
+      "step": 1189
+    },
+    {
+      "epoch": 1.9244444444444444,
+      "grad_norm": 3.140625,
+      "learning_rate": 4.2163292855056936e-08,
+      "loss": 0.7859,
+      "step": 1190
+    },
+    {
+      "epoch": 1.926060606060606,
+      "grad_norm": 3.125,
+      "learning_rate": 4.035247517085283e-08,
+      "loss": 0.9855,
+      "step": 1191
+    },
+    {
+      "epoch": 1.9276767676767677,
+      "grad_norm": 3.46875,
+      "learning_rate": 3.858124340229863e-08,
+      "loss": 0.8916,
+      "step": 1192
+    },
+    {
+      "epoch": 1.9292929292929293,
+      "grad_norm": 2.78125,
+      "learning_rate": 3.684961168665457e-08,
+      "loss": 0.6614,
+      "step": 1193
+    },
+    {
+      "epoch": 1.930909090909091,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.515759384510986e-08,
+      "loss": 0.7658,
+      "step": 1194
+    },
+    {
+      "epoch": 1.9325252525252525,
+      "grad_norm": 3.171875,
+      "learning_rate": 3.350520338267216e-08,
+      "loss": 0.7559,
+      "step": 1195
+    },
+    {
+      "epoch": 1.9341414141414142,
+      "grad_norm": 3.125,
+      "learning_rate": 3.1892453488058803e-08,
+      "loss": 0.8424,
+      "step": 1196
+    },
+    {
+      "epoch": 1.9357575757575758,
+      "grad_norm": 3.25,
+      "learning_rate": 3.0319357033591904e-08,
+      "loss": 0.963,
+      "step": 1197
+    },
+    {
+      "epoch": 1.9373737373737374,
+      "grad_norm": 2.96875,
+      "learning_rate": 2.87859265750956e-08,
+      "loss": 0.7586,
+      "step": 1198
+    },
+    {
+      "epoch": 1.938989898989899,
+      "grad_norm": 3.515625,
+      "learning_rate": 2.7292174351797317e-08,
+      "loss": 1.0173,
+      "step": 1199
+    },
+    {
+      "epoch": 1.9406060606060604,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.5838112286226123e-08,
+      "loss": 1.096,
+      "step": 1200
+    },
+    {
+      "epoch": 1.942222222222222,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.4423751984122835e-08,
+      "loss": 1.0508,
+      "step": 1201
+    },
+    {
+      "epoch": 1.9438383838383837,
+      "grad_norm": 3.34375,
+      "learning_rate": 2.304910473434341e-08,
+      "loss": 0.8316,
+      "step": 1202
+    },
+    {
+      "epoch": 1.9454545454545453,
+      "grad_norm": 3.015625,
+      "learning_rate": 2.1714181508770138e-08,
+      "loss": 0.7511,
+      "step": 1203
+    },
+    {
+      "epoch": 1.947070707070707,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.0418992962224495e-08,
+      "loss": 0.9021,
+      "step": 1204
+    },
+    {
+      "epoch": 1.9486868686868686,
+      "grad_norm": 3.140625,
+      "learning_rate": 1.916354943238219e-08,
+      "loss": 0.8465,
+      "step": 1205
+    },
+    {
+      "epoch": 1.9503030303030302,
+      "grad_norm": 3.140625,
+      "learning_rate": 1.7947860939688255e-08,
+      "loss": 0.7897,
+      "step": 1206
+    },
+    {
+      "epoch": 1.9519191919191918,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.6771937187280986e-08,
+      "loss": 0.8088,
+      "step": 1207
+    },
+    {
+      "epoch": 1.9535353535353535,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.563578756091144e-08,
+      "loss": 0.7406,
+      "step": 1208
+    },
+    {
+      "epoch": 1.955151515151515,
+      "grad_norm": 2.890625,
+      "learning_rate": 1.4539421128870745e-08,
+      "loss": 1.0346,
+      "step": 1209
+    },
+    {
+      "epoch": 1.9567676767676767,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.3482846641914572e-08,
+      "loss": 0.7987,
+      "step": 1210
+    },
+    {
+      "epoch": 1.9583838383838383,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.2466072533197093e-08,
+      "loss": 0.8569,
+      "step": 1211
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 3.734375,
+      "learning_rate": 1.1489106918200487e-08,
+      "loss": 0.8364,
+      "step": 1212
+    },
+    {
+      "epoch": 1.9616161616161616,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.0551957594672201e-08,
+      "loss": 0.7226,
+      "step": 1213
+    },
+    {
+      "epoch": 1.9632323232323232,
+      "grad_norm": 3.21875,
+      "learning_rate": 9.654632042562229e-09,
+      "loss": 0.749,
+      "step": 1214
+    },
+    {
+      "epoch": 1.9648484848484848,
+      "grad_norm": 3.0,
+      "learning_rate": 8.797137423963153e-09,
+      "loss": 0.8142,
+      "step": 1215
+    },
+    {
+      "epoch": 1.9664646464646465,
+      "grad_norm": 3.109375,
+      "learning_rate": 7.979480583052423e-09,
+      "loss": 0.7345,
+      "step": 1216
+    },
+    {
+      "epoch": 1.968080808080808,
+      "grad_norm": 3.015625,
+      "learning_rate": 7.201668046038501e-09,
+      "loss": 0.7713,
+      "step": 1217
+    },
+    {
+      "epoch": 1.9696969696969697,
+      "grad_norm": 3.03125,
+      "learning_rate": 6.4637060211092395e-09,
+      "loss": 0.6869,
+      "step": 1218
+    },
+    {
+      "epoch": 1.9713131313131314,
+      "grad_norm": 3.65625,
+      "learning_rate": 5.765600398381366e-09,
+      "loss": 0.9868,
+      "step": 1219
+    },
+    {
+      "epoch": 1.972929292929293,
+      "grad_norm": 3.1875,
+      "learning_rate": 5.107356749853298e-09,
+      "loss": 0.5491,
+      "step": 1220
+    },
+    {
+      "epoch": 1.9745454545454546,
+      "grad_norm": 3.359375,
+      "learning_rate": 4.488980329361847e-09,
+      "loss": 0.8056,
+      "step": 1221
+    },
+    {
+      "epoch": 1.9761616161616162,
+      "grad_norm": 3.65625,
+      "learning_rate": 3.910476072539471e-09,
+      "loss": 0.7613,
+      "step": 1222
+    },
+    {
+      "epoch": 1.9777777777777779,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.3718485967748626e-09,
+      "loss": 0.8703,
+      "step": 1223
+    },
+    {
+      "epoch": 1.9793939393939395,
+      "grad_norm": 3.046875,
+      "learning_rate": 2.8731022011757593e-09,
+      "loss": 0.8264,
+      "step": 1224
+    },
+    {
+      "epoch": 1.981010101010101,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.4142408665367433e-09,
+      "loss": 0.6388,
+      "step": 1225
+    },
+    {
+      "epoch": 1.9826262626262627,
+      "grad_norm": 3.1875,
+      "learning_rate": 1.9952682553042722e-09,
+      "loss": 0.7838,
+      "step": 1226
+    },
+    {
+      "epoch": 1.9842424242424244,
+      "grad_norm": 3.1875,
+      "learning_rate": 1.6161877115494773e-09,
+      "loss": 0.7301,
+      "step": 1227
+    },
+    {
+      "epoch": 1.985858585858586,
+      "grad_norm": 3.421875,
+      "learning_rate": 1.2770022609409628e-09,
+      "loss": 0.7425,
+      "step": 1228
+    },
+    {
+      "epoch": 1.9874747474747476,
+      "grad_norm": 3.0,
+      "learning_rate": 9.777146107209367e-10,
+      "loss": 0.9393,
+      "step": 1229
+    },
+    {
+      "epoch": 1.9890909090909092,
+      "grad_norm": 3.578125,
+      "learning_rate": 7.18327149683562e-10,
+      "loss": 0.8389,
+      "step": 1230
+    },
+    {
+      "epoch": 1.9890909090909092,
+      "eval_loss": 0.817283034324646,
+      "eval_runtime": 33.6747,
+      "eval_samples_per_second": 32.665,
+      "eval_steps_per_second": 4.098,
+      "step": 1230
+    },
+    {
+      "epoch": 1.9907070707070709,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.988419481549711e-10,
+      "loss": 0.8228,
+      "step": 1231
+    },
+    {
+      "epoch": 1.9923232323232323,
+      "grad_norm": 3.453125,
+      "learning_rate": 3.1926075797827914e-10,
+      "loss": 0.9733,
+      "step": 1232
+    },
+    {
+      "epoch": 1.993939393939394,
+      "grad_norm": 3.171875,
+      "learning_rate": 1.7958501249859538e-10,
+      "loss": 0.9304,
+      "step": 1233
+    },
+    {
+      "epoch": 1.9955555555555555,
+      "grad_norm": 3.421875,
+      "learning_rate": 7.9815826551366e-11,
+      "loss": 0.8094,
+      "step": 1234
+    },
+    {
+      "epoch": 1.9971717171717172,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.9953996454047386e-11,
+      "loss": 0.6743,
+      "step": 1235
+    },
+    {
+      "epoch": 1.9987878787878788,
+      "grad_norm": 3.140625,
+      "learning_rate": 0.0,
+      "loss": 0.8149,
+      "step": 1236
+    },
+    {
+      "epoch": 2.0016161616161616,
+      "grad_norm": 3.328125,
+      "learning_rate": 3.0128778019093774e-06,
+      "loss": 1.0424,
+      "step": 1237
+    },
+    {
+      "epoch": 2.0032323232323233,
+      "grad_norm": 3.203125,
+      "learning_rate": 3.0042397396216076e-06,
+      "loss": 0.7341,
+      "step": 1238
+    },
+    {
+      "epoch": 2.004848484848485,
+      "grad_norm": 3.046875,
+      "learning_rate": 2.995608757050793e-06,
+      "loss": 0.883,
+      "step": 1239
+    },
+    {
+      "epoch": 2.0064646464646465,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.9869848848142957e-06,
+      "loss": 0.8047,
+      "step": 1240
+    },
+    {
+      "epoch": 2.008080808080808,
+      "grad_norm": 2.78125,
+      "learning_rate": 2.978368153504254e-06,
+      "loss": 0.6947,
+      "step": 1241
+    },
+    {
+      "epoch": 2.0096969696969698,
+      "grad_norm": 3.21875,
+      "learning_rate": 2.969758593687475e-06,
+      "loss": 0.5461,
+      "step": 1242
+    },
+    {
+      "epoch": 2.0113131313131314,
+      "grad_norm": 3.3125,
+      "learning_rate": 2.961156235905325e-06,
+      "loss": 0.7415,
+      "step": 1243
+    },
+    {
+      "epoch": 2.012929292929293,
+      "grad_norm": 3.03125,
+      "learning_rate": 2.952561110673623e-06,
+      "loss": 0.7528,
+      "step": 1244
+    },
+    {
+      "epoch": 2.0145454545454546,
+      "grad_norm": 3.078125,
+      "learning_rate": 2.9439732484825323e-06,
+      "loss": 0.765,
+      "step": 1245
+    },
+    {
+      "epoch": 2.0161616161616163,
+      "grad_norm": 3.0,
+      "learning_rate": 2.9353926797964495e-06,
+      "loss": 0.7575,
+      "step": 1246
+    },
+    {
+      "epoch": 2.017777777777778,
+      "grad_norm": 3.75,
+      "learning_rate": 2.9268194350538988e-06,
+      "loss": 1.0869,
+      "step": 1247
+    },
+    {
+      "epoch": 2.0193939393939395,
+      "grad_norm": 3.703125,
+      "learning_rate": 2.9182535446674244e-06,
+      "loss": 0.7021,
+      "step": 1248
+    },
+    {
+      "epoch": 2.021010101010101,
+      "grad_norm": 3.203125,
+      "learning_rate": 2.9096950390234808e-06,
+      "loss": 0.7172,
+      "step": 1249
+    },
+    {
+      "epoch": 2.0226262626262628,
+      "grad_norm": 2.90625,
+      "learning_rate": 2.9011439484823287e-06,
+      "loss": 1.0203,
+      "step": 1250
+    },
+    {
+      "epoch": 2.0242424242424244,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.8926003033779194e-06,
+      "loss": 0.9086,
+      "step": 1251
+    },
+    {
+      "epoch": 2.025858585858586,
+      "grad_norm": 3.65625,
+      "learning_rate": 2.8840641340177955e-06,
+      "loss": 0.8249,
+      "step": 1252
+    },
+    {
+      "epoch": 2.0274747474747477,
+      "grad_norm": 3.3125,
+      "learning_rate": 2.8755354706829825e-06,
+      "loss": 0.7349,
+      "step": 1253
+    },
+    {
+      "epoch": 2.0290909090909093,
+      "grad_norm": 2.984375,
+      "learning_rate": 2.8670143436278757e-06,
+      "loss": 0.791,
+      "step": 1254
+    },
+    {
+      "epoch": 2.030707070707071,
+      "grad_norm": 3.578125,
+      "learning_rate": 2.858500783080138e-06,
+      "loss": 0.9202,
+      "step": 1255
+    },
+    {
+      "epoch": 2.0323232323232325,
+      "grad_norm": 3.359375,
+      "learning_rate": 2.84999481924059e-06,
+      "loss": 0.896,
+      "step": 1256
+    },
+    {
+      "epoch": 2.033939393939394,
+      "grad_norm": 2.90625,
+      "learning_rate": 2.8414964822831063e-06,
+      "loss": 0.9948,
+      "step": 1257
+    },
+    {
+      "epoch": 2.0355555555555553,
+      "grad_norm": 3.46875,
+      "learning_rate": 2.8330058023545027e-06,
+      "loss": 0.6328,
+      "step": 1258
+    },
+    {
+      "epoch": 2.037171717171717,
+      "grad_norm": 3.65625,
+      "learning_rate": 2.824522809574437e-06,
+      "loss": 0.7788,
+      "step": 1259
+    },
+    {
+      "epoch": 2.0387878787878786,
+      "grad_norm": 3.484375,
+      "learning_rate": 2.8160475340352913e-06,
+      "loss": 0.8316,
+      "step": 1260
+    },
+    {
+      "epoch": 2.04040404040404,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.8075800058020765e-06,
+      "loss": 0.5439,
+      "step": 1261
+    },
+    {
+      "epoch": 2.042020202020202,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.799120254912321e-06,
+      "loss": 0.8943,
+      "step": 1262
+    },
+    {
+      "epoch": 2.0436363636363635,
+      "grad_norm": 3.21875,
+      "learning_rate": 2.790668311375962e-06,
+      "loss": 0.7263,
+      "step": 1263
+    },
+    {
+      "epoch": 2.045252525252525,
+      "grad_norm": 3.78125,
+      "learning_rate": 2.7822242051752425e-06,
+      "loss": 1.1025,
+      "step": 1264
+    },
+    {
+      "epoch": 2.0468686868686867,
+      "grad_norm": 3.828125,
+      "learning_rate": 2.773787966264602e-06,
+      "loss": 0.8268,
+      "step": 1265
+    },
+    {
+      "epoch": 2.0484848484848484,
+      "grad_norm": 3.515625,
+      "learning_rate": 2.765359624570574e-06,
+      "loss": 0.6683,
+      "step": 1266
+    },
+    {
+      "epoch": 2.05010101010101,
+      "grad_norm": 3.109375,
+      "learning_rate": 2.7569392099916753e-06,
+      "loss": 0.7588,
+      "step": 1267
+    },
+    {
+      "epoch": 2.0517171717171716,
+      "grad_norm": 3.3125,
+      "learning_rate": 2.7485267523983038e-06,
+      "loss": 0.8358,
+      "step": 1268
+    },
+    {
+      "epoch": 2.0533333333333332,
+      "grad_norm": 3.046875,
+      "learning_rate": 2.7401222816326316e-06,
+      "loss": 0.6513,
+      "step": 1269
+    },
+    {
+      "epoch": 2.054949494949495,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.731725827508494e-06,
+      "loss": 0.8144,
+      "step": 1270
+    },
+    {
+      "epoch": 2.0565656565656565,
+      "grad_norm": 3.6875,
+      "learning_rate": 2.7233374198112934e-06,
+      "loss": 0.7074,
+      "step": 1271
+    },
+    {
+      "epoch": 2.058181818181818,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.714957088297886e-06,
+      "loss": 0.8872,
+      "step": 1272
+    },
+    {
+      "epoch": 2.0597979797979797,
+      "grad_norm": 3.703125,
+      "learning_rate": 2.7065848626964796e-06,
+      "loss": 0.9735,
+      "step": 1273
+    },
+    {
+      "epoch": 2.0614141414141414,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.6982207727065252e-06,
+      "loss": 0.8238,
+      "step": 1274
+    },
+    {
+      "epoch": 2.063030303030303,
+      "grad_norm": 3.25,
+      "learning_rate": 2.6898648479986187e-06,
+      "loss": 0.7669,
+      "step": 1275
+    },
+    {
+      "epoch": 2.0646464646464646,
+      "grad_norm": 3.453125,
+      "learning_rate": 2.681517118214389e-06,
+      "loss": 0.6541,
+      "step": 1276
+    },
+    {
+      "epoch": 2.0662626262626262,
+      "grad_norm": 3.328125,
+      "learning_rate": 2.6731776129663873e-06,
+      "loss": 0.7117,
+      "step": 1277
+    },
+    {
+      "epoch": 2.067878787878788,
+      "grad_norm": 3.609375,
+      "learning_rate": 2.664846361837997e-06,
+      "loss": 0.6556,
+      "step": 1278
+    },
+    {
+      "epoch": 2.0694949494949495,
+      "grad_norm": 3.578125,
+      "learning_rate": 2.6565233943833204e-06,
+      "loss": 0.8226,
+      "step": 1279
+    },
+    {
+      "epoch": 2.071111111111111,
+      "grad_norm": 3.359375,
+      "learning_rate": 2.6482087401270705e-06,
+      "loss": 0.8197,
+      "step": 1280
+    },
+    {
+      "epoch": 2.0727272727272728,
+      "grad_norm": 3.234375,
+      "learning_rate": 2.639902428564475e-06,
+      "loss": 0.8156,
+      "step": 1281
+    },
+    {
+      "epoch": 2.0743434343434344,
+      "grad_norm": 3.078125,
+      "learning_rate": 2.6316044891611633e-06,
+      "loss": 0.6801,
+      "step": 1282
+    },
+    {
+      "epoch": 2.075959595959596,
+      "grad_norm": 3.40625,
+      "learning_rate": 2.6233149513530675e-06,
+      "loss": 1.0043,
+      "step": 1283
+    },
+    {
+      "epoch": 2.0775757575757576,
+      "grad_norm": 3.515625,
+      "learning_rate": 2.6150338445463146e-06,
+      "loss": 0.6945,
+      "step": 1284
+    },
+    {
+      "epoch": 2.0791919191919193,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.6067611981171275e-06,
+      "loss": 0.7657,
+      "step": 1285
+    },
+    {
+      "epoch": 2.080808080808081,
+      "grad_norm": 3.109375,
+      "learning_rate": 2.5984970414117096e-06,
+      "loss": 0.633,
+      "step": 1286
+    },
+    {
+      "epoch": 2.0824242424242425,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.5902414037461544e-06,
+      "loss": 0.923,
+      "step": 1287
+    },
+    {
+      "epoch": 2.084040404040404,
+      "grad_norm": 3.359375,
+      "learning_rate": 2.5819943144063326e-06,
+      "loss": 0.6408,
+      "step": 1288
+    },
+    {
+      "epoch": 2.0856565656565658,
+      "grad_norm": 3.5,
+      "learning_rate": 2.5737558026477915e-06,
+      "loss": 0.8528,
+      "step": 1289
+    },
+    {
+      "epoch": 2.0872727272727274,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.565525897695651e-06,
+      "loss": 0.6908,
+      "step": 1290
+    },
+    {
+      "epoch": 2.088888888888889,
+      "grad_norm": 3.1875,
+      "learning_rate": 2.557304628744498e-06,
+      "loss": 0.5904,
+      "step": 1291
+    },
+    {
+      "epoch": 2.0905050505050506,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.549092024958285e-06,
+      "loss": 0.8887,
+      "step": 1292
+    },
+    {
+      "epoch": 2.0921212121212123,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.540888115470226e-06,
+      "loss": 0.8401,
+      "step": 1293
+    },
+    {
+      "epoch": 2.093737373737374,
+      "grad_norm": 3.453125,
+      "learning_rate": 2.532692929382692e-06,
+      "loss": 0.8209,
+      "step": 1294
+    },
+    {
+      "epoch": 2.0953535353535355,
+      "grad_norm": 2.84375,
+      "learning_rate": 2.5245064957671117e-06,
+      "loss": 0.9172,
+      "step": 1295
+    },
+    {
+      "epoch": 2.096969696969697,
+      "grad_norm": 3.375,
+      "learning_rate": 2.51632884366386e-06,
+      "loss": 0.5987,
+      "step": 1296
+    },
+    {
+      "epoch": 2.0985858585858588,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.5081600020821655e-06,
+      "loss": 0.863,
+      "step": 1297
+    },
+    {
+      "epoch": 2.1002020202020204,
+      "grad_norm": 3.078125,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 0.8251,
+      "step": 1298
+    },
+    {
+      "epoch": 2.101818181818182,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.4918488663639824e-06,
+      "loss": 0.8216,
+      "step": 1299
+    },
+    {
+      "epoch": 2.103434343434343,
+      "grad_norm": 3.34375,
+      "learning_rate": 2.4837066300892647e-06,
+      "loss": 0.7048,
+      "step": 1300
+    },
+    {
+      "epoch": 2.105050505050505,
+      "grad_norm": 3.46875,
+      "learning_rate": 2.475573320059442e-06,
+      "loss": 0.9212,
+      "step": 1301
+    },
+    {
+      "epoch": 2.1066666666666665,
+      "grad_norm": 3.4375,
+      "learning_rate": 2.4674489651264433e-06,
+      "loss": 0.6265,
+      "step": 1302
+    },
+    {
+      "epoch": 2.108282828282828,
+      "grad_norm": 3.234375,
+      "learning_rate": 2.45933359411043e-06,
+      "loss": 0.7448,
+      "step": 1303
+    },
+    {
+      "epoch": 2.1098989898989897,
+      "grad_norm": 3.4375,
+      "learning_rate": 2.4512272357996937e-06,
+      "loss": 0.6336,
+      "step": 1304
+    },
+    {
+      "epoch": 2.1115151515151513,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.4431299189505563e-06,
+      "loss": 0.9352,
+      "step": 1305
+    },
+    {
+      "epoch": 2.113131313131313,
+      "grad_norm": 3.234375,
+      "learning_rate": 2.4350416722872657e-06,
+      "loss": 0.5659,
+      "step": 1306
+    },
+    {
+      "epoch": 2.1147474747474746,
+      "grad_norm": 2.96875,
+      "learning_rate": 2.426962524501892e-06,
+      "loss": 0.675,
+      "step": 1307
+    },
+    {
+      "epoch": 2.1163636363636362,
+      "grad_norm": 3.328125,
+      "learning_rate": 2.418892504254231e-06,
+      "loss": 1.0917,
+      "step": 1308
+    },
+    {
+      "epoch": 2.117979797979798,
+      "grad_norm": 3.375,
+      "learning_rate": 2.410831640171699e-06,
+      "loss": 0.7817,
+      "step": 1309
+    },
+    {
+      "epoch": 2.1195959595959595,
+      "grad_norm": 3.890625,
+      "learning_rate": 2.402779960849232e-06,
+      "loss": 0.8288,
+      "step": 1310
+    },
+    {
+      "epoch": 2.121212121212121,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.394737494849184e-06,
+      "loss": 0.486,
+      "step": 1311
+    },
+    {
+      "epoch": 2.1228282828282827,
+      "grad_norm": 3.09375,
+      "learning_rate": 2.3867042707012234e-06,
+      "loss": 0.8399,
+      "step": 1312
+    },
+    {
+      "epoch": 2.1244444444444444,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.378680316902237e-06,
+      "loss": 0.7501,
+      "step": 1313
+    },
+    {
+      "epoch": 2.126060606060606,
+      "grad_norm": 3.328125,
+      "learning_rate": 2.3706656619162278e-06,
+      "loss": 1.0384,
+      "step": 1314
+    },
+    {
+      "epoch": 2.1276767676767676,
+      "grad_norm": 3.1875,
+      "learning_rate": 2.3626603341742077e-06,
+      "loss": 0.5542,
+      "step": 1315
+    },
+    {
+      "epoch": 2.1292929292929292,
+      "grad_norm": 3.828125,
+      "learning_rate": 2.3546643620741054e-06,
+      "loss": 0.878,
+      "step": 1316
+    },
+    {
+      "epoch": 2.130909090909091,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.34667777398066e-06,
+      "loss": 0.7471,
+      "step": 1317
+    },
+    {
+      "epoch": 2.1325252525252525,
+      "grad_norm": 2.921875,
+      "learning_rate": 2.3387005982253218e-06,
+      "loss": 0.6114,
+      "step": 1318
+    },
+    {
+      "epoch": 2.134141414141414,
+      "grad_norm": 3.328125,
+      "learning_rate": 2.3307328631061526e-06,
+      "loss": 0.7777,
+      "step": 1319
+    },
+    {
+      "epoch": 2.1357575757575757,
+      "grad_norm": 3.171875,
+      "learning_rate": 2.322774596887726e-06,
+      "loss": 0.6017,
+      "step": 1320
+    },
+    {
+      "epoch": 2.1373737373737374,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.314825827801019e-06,
+      "loss": 0.7445,
+      "step": 1321
+    },
+    {
+      "epoch": 2.138989898989899,
+      "grad_norm": 3.453125,
+      "learning_rate": 2.3068865840433286e-06,
+      "loss": 0.7317,
+      "step": 1322
+    },
+    {
+      "epoch": 2.1406060606060606,
+      "grad_norm": 2.90625,
+      "learning_rate": 2.298956893778154e-06,
+      "loss": 1.0594,
+      "step": 1323
+    },
+    {
+      "epoch": 2.1422222222222222,
+      "grad_norm": 3.4375,
+      "learning_rate": 2.29103678513511e-06,
+      "loss": 1.0783,
+      "step": 1324
+    },
+    {
+      "epoch": 2.143838383838384,
+      "grad_norm": 3.625,
+      "learning_rate": 2.2831262862098186e-06,
+      "loss": 0.5743,
+      "step": 1325
+    },
+    {
+      "epoch": 2.1454545454545455,
+      "grad_norm": 3.5,
+      "learning_rate": 2.275225425063813e-06,
+      "loss": 0.9351,
+      "step": 1326
+    },
+    {
+      "epoch": 2.147070707070707,
+      "grad_norm": 3.578125,
+      "learning_rate": 2.267334229724437e-06,
+      "loss": 1.2655,
+      "step": 1327
+    },
+    {
+      "epoch": 2.1486868686868688,
+      "grad_norm": 3.109375,
+      "learning_rate": 2.259452728184749e-06,
+      "loss": 0.4254,
+      "step": 1328
+    },
+    {
+      "epoch": 2.1503030303030304,
+      "grad_norm": 3.328125,
+      "learning_rate": 2.251580948403416e-06,
+      "loss": 0.7853,
+      "step": 1329
+    },
+    {
+      "epoch": 2.151919191919192,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.2437189183046236e-06,
+      "loss": 0.6547,
+      "step": 1330
+    },
+    {
+      "epoch": 2.1535353535353536,
+      "grad_norm": 3.203125,
+      "learning_rate": 2.2358666657779626e-06,
+      "loss": 0.8174,
+      "step": 1331
+    },
+    {
+      "epoch": 2.1551515151515153,
+      "grad_norm": 3.84375,
+      "learning_rate": 2.2280242186783473e-06,
+      "loss": 0.812,
+      "step": 1332
+    },
+    {
+      "epoch": 2.156767676767677,
+      "grad_norm": 3.109375,
+      "learning_rate": 2.2201916048259033e-06,
+      "loss": 0.4584,
+      "step": 1333
+    },
+    {
+      "epoch": 2.1583838383838385,
+      "grad_norm": 3.59375,
+      "learning_rate": 2.21236885200588e-06,
+      "loss": 0.6674,
+      "step": 1334
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 3.8125,
+      "learning_rate": 2.204555987968541e-06,
+      "loss": 0.7239,
+      "step": 1335
+    },
+    {
+      "epoch": 2.1616161616161618,
+      "grad_norm": 3.15625,
+      "learning_rate": 2.1967530404290702e-06,
+      "loss": 0.9386,
+      "step": 1336
+    },
+    {
+      "epoch": 2.1632323232323234,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.1889600370674796e-06,
+      "loss": 0.7385,
+      "step": 1337
+    },
+    {
+      "epoch": 2.164848484848485,
+      "grad_norm": 3.3125,
+      "learning_rate": 2.1811770055284968e-06,
+      "loss": 0.591,
+      "step": 1338
+    },
+    {
+      "epoch": 2.1664646464646466,
+      "grad_norm": 3.5625,
+      "learning_rate": 2.1734039734214816e-06,
+      "loss": 0.8033,
+      "step": 1339
+    },
+    {
+      "epoch": 2.1680808080808083,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.1656409683203216e-06,
+      "loss": 1.1095,
+      "step": 1340
+    },
+    {
+      "epoch": 2.16969696969697,
+      "grad_norm": 3.53125,
+      "learning_rate": 2.157888017763333e-06,
+      "loss": 0.8222,
+      "step": 1341
+    },
+    {
+      "epoch": 2.1713131313131315,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.1501451492531664e-06,
+      "loss": 0.7421,
+      "step": 1342
+    },
+    {
+      "epoch": 2.172929292929293,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.1424123902567063e-06,
+      "loss": 0.7589,
+      "step": 1343
+    },
+    {
+      "epoch": 2.174545454545455,
+      "grad_norm": 2.984375,
+      "learning_rate": 2.134689768204975e-06,
+      "loss": 0.8158,
+      "step": 1344
+    },
+    {
+      "epoch": 2.176161616161616,
+      "grad_norm": 3.109375,
+      "learning_rate": 2.1269773104930365e-06,
+      "loss": 0.8679,
+      "step": 1345
+    },
+    {
+      "epoch": 2.1777777777777776,
+      "grad_norm": 3.328125,
+      "learning_rate": 2.1192750444798982e-06,
+      "loss": 0.9143,
+      "step": 1346
+    },
+    {
+      "epoch": 2.179393939393939,
+      "grad_norm": 3.703125,
+      "learning_rate": 2.1115829974884097e-06,
+      "loss": 0.605,
+      "step": 1347
+    },
+    {
+      "epoch": 2.181010101010101,
+      "grad_norm": 3.125,
+      "learning_rate": 2.103901196805173e-06,
+      "loss": 0.523,
+      "step": 1348
+    },
+    {
+      "epoch": 2.1826262626262625,
+      "grad_norm": 3.5625,
+      "learning_rate": 2.0962296696804447e-06,
+      "loss": 0.7345,
+      "step": 1349
+    },
+    {
+      "epoch": 2.184242424242424,
+      "grad_norm": 3.5,
+      "learning_rate": 2.0885684433280336e-06,
+      "loss": 0.8119,
+      "step": 1350
+    },
+    {
+      "epoch": 2.1858585858585857,
+      "grad_norm": 3.625,
+      "learning_rate": 2.080917544925209e-06,
+      "loss": 0.8786,
+      "step": 1351
+    },
+    {
+      "epoch": 2.1874747474747473,
+      "grad_norm": 2.890625,
+      "learning_rate": 2.073277001612603e-06,
+      "loss": 0.8235,
+      "step": 1352
+    },
+    {
+      "epoch": 2.189090909090909,
+      "grad_norm": 3.1875,
+      "learning_rate": 2.065646840494115e-06,
+      "loss": 0.976,
+      "step": 1353
+    },
+    {
+      "epoch": 2.189090909090909,
+      "eval_loss": 0.8148007988929749,
+      "eval_runtime": 34.0335,
+      "eval_samples_per_second": 32.321,
+      "eval_steps_per_second": 4.055,
+      "step": 1353
+    },
+    {
+      "epoch": 2.1907070707070706,
+      "grad_norm": 3.5,
+      "learning_rate": 2.058027088636814e-06,
+      "loss": 1.2466,
+      "step": 1354
+    },
+    {
+      "epoch": 2.1923232323232322,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.050417773070845e-06,
+      "loss": 1.1437,
+      "step": 1355
+    },
+    {
+      "epoch": 2.193939393939394,
+      "grad_norm": 3.421875,
+      "learning_rate": 2.042818920789326e-06,
+      "loss": 1.0728,
+      "step": 1356
+    },
+    {
+      "epoch": 2.1955555555555555,
+      "grad_norm": 3.234375,
+      "learning_rate": 2.035230558748265e-06,
+      "loss": 0.8289,
+      "step": 1357
+    },
+    {
+      "epoch": 2.197171717171717,
+      "grad_norm": 2.984375,
+      "learning_rate": 2.0276527138664537e-06,
+      "loss": 0.9644,
+      "step": 1358
+    },
+    {
+      "epoch": 2.1987878787878787,
+      "grad_norm": 3.0625,
+      "learning_rate": 2.020085413025375e-06,
+      "loss": 0.7024,
+      "step": 1359
+    },
+    {
+      "epoch": 2.2004040404040404,
+      "grad_norm": 3.3125,
+      "learning_rate": 2.012528683069109e-06,
+      "loss": 0.6661,
+      "step": 1360
+    },
+    {
+      "epoch": 2.202020202020202,
+      "grad_norm": 3.15625,
+      "learning_rate": 2.004982550804238e-06,
+      "loss": 0.6854,
+      "step": 1361
+    },
+    {
+      "epoch": 2.2036363636363636,
+      "grad_norm": 3.375,
+      "learning_rate": 1.9974470429997482e-06,
+      "loss": 0.754,
+      "step": 1362
+    },
+    {
+      "epoch": 2.2052525252525252,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.9899221863869374e-06,
+      "loss": 0.7284,
+      "step": 1363
+    },
+    {
+      "epoch": 2.206868686868687,
+      "grad_norm": 3.140625,
+      "learning_rate": 1.98240800765932e-06,
+      "loss": 0.7513,
+      "step": 1364
+    },
+    {
+      "epoch": 2.2084848484848485,
+      "grad_norm": 3.53125,
+      "learning_rate": 1.974904533472532e-06,
+      "loss": 0.9995,
+      "step": 1365
+    },
+    {
+      "epoch": 2.21010101010101,
+      "grad_norm": 3.53125,
+      "learning_rate": 1.9674117904442364e-06,
+      "loss": 0.7987,
+      "step": 1366
+    },
+    {
+      "epoch": 2.2117171717171717,
+      "grad_norm": 3.09375,
+      "learning_rate": 1.959929805154027e-06,
+      "loss": 0.6981,
+      "step": 1367
+    },
+    {
+      "epoch": 2.2133333333333334,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.9524586041433393e-06,
+      "loss": 0.8207,
+      "step": 1368
+    },
+    {
+      "epoch": 2.214949494949495,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.944998213915349e-06,
+      "loss": 0.855,
+      "step": 1369
+    },
+    {
+      "epoch": 2.2165656565656566,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.9375486609348842e-06,
+      "loss": 0.5397,
+      "step": 1370
+    },
+    {
+      "epoch": 2.2181818181818183,
+      "grad_norm": 4.0,
+      "learning_rate": 1.9301099716283293e-06,
+      "loss": 0.6543,
+      "step": 1371
+    },
+    {
+      "epoch": 2.21979797979798,
+      "grad_norm": 3.25,
+      "learning_rate": 1.9226821723835322e-06,
+      "loss": 0.6552,
+      "step": 1372
+    },
+    {
+      "epoch": 2.2214141414141415,
+      "grad_norm": 3.25,
+      "learning_rate": 1.9152652895497035e-06,
+      "loss": 0.7258,
+      "step": 1373
+    },
+    {
+      "epoch": 2.223030303030303,
+      "grad_norm": 3.421875,
+      "learning_rate": 1.907859349437336e-06,
+      "loss": 0.9764,
+      "step": 1374
+    },
+    {
+      "epoch": 2.2246464646464648,
+      "grad_norm": 3.875,
+      "learning_rate": 1.9004643783181026e-06,
+      "loss": 0.826,
+      "step": 1375
+    },
+    {
+      "epoch": 2.2262626262626264,
+      "grad_norm": 3.328125,
+      "learning_rate": 1.8930804024247635e-06,
+      "loss": 0.7608,
+      "step": 1376
+    },
+    {
+      "epoch": 2.227878787878788,
+      "grad_norm": 3.234375,
+      "learning_rate": 1.8857074479510761e-06,
+      "loss": 0.8576,
+      "step": 1377
+    },
+    {
+      "epoch": 2.2294949494949496,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.8783455410517004e-06,
+      "loss": 0.6145,
+      "step": 1378
+    },
+    {
+      "epoch": 2.2311111111111113,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.8709947078421065e-06,
+      "loss": 0.8222,
+      "step": 1379
+    },
+    {
+      "epoch": 2.232727272727273,
+      "grad_norm": 3.21875,
+      "learning_rate": 1.8636549743984815e-06,
+      "loss": 0.7151,
+      "step": 1380
+    },
+    {
+      "epoch": 2.2343434343434345,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.8563263667576393e-06,
+      "loss": 0.9346,
+      "step": 1381
+    },
+    {
+      "epoch": 2.235959595959596,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.8490089109169218e-06,
+      "loss": 0.7475,
+      "step": 1382
+    },
+    {
+      "epoch": 2.2375757575757578,
+      "grad_norm": 3.21875,
+      "learning_rate": 1.8417026328341158e-06,
+      "loss": 0.7821,
+      "step": 1383
+    },
+    {
+      "epoch": 2.2391919191919194,
+      "grad_norm": 3.171875,
+      "learning_rate": 1.8344075584273547e-06,
+      "loss": 0.851,
+      "step": 1384
+    },
+    {
+      "epoch": 2.2408080808080806,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.827123713575028e-06,
+      "loss": 0.6882,
+      "step": 1385
+    },
+    {
+      "epoch": 2.242424242424242,
+      "grad_norm": 3.390625,
+      "learning_rate": 1.8198511241156902e-06,
+      "loss": 0.9434,
+      "step": 1386
+    },
+    {
+      "epoch": 2.244040404040404,
+      "grad_norm": 3.453125,
+      "learning_rate": 1.8125898158479687e-06,
+      "loss": 0.8263,
+      "step": 1387
+    },
+    {
+      "epoch": 2.2456565656565655,
+      "grad_norm": 3.578125,
+      "learning_rate": 1.8053398145304723e-06,
+      "loss": 0.6343,
+      "step": 1388
+    },
+    {
+      "epoch": 2.247272727272727,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.7981011458816988e-06,
+      "loss": 0.6728,
+      "step": 1389
+    },
+    {
+      "epoch": 2.2488888888888887,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.7908738355799454e-06,
+      "loss": 0.8371,
+      "step": 1390
+    },
+    {
+      "epoch": 2.2505050505050503,
+      "grad_norm": 3.3125,
+      "learning_rate": 1.7836579092632194e-06,
+      "loss": 0.6575,
+      "step": 1391
+    },
+    {
+      "epoch": 2.252121212121212,
+      "grad_norm": 3.390625,
+      "learning_rate": 1.776453392529139e-06,
+      "loss": 0.657,
+      "step": 1392
+    },
+    {
+      "epoch": 2.2537373737373736,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.7692603109348505e-06,
+      "loss": 0.9262,
+      "step": 1393
+    },
+    {
+      "epoch": 2.255353535353535,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.7620786899969412e-06,
+      "loss": 0.8043,
+      "step": 1394
+    },
+    {
+      "epoch": 2.256969696969697,
+      "grad_norm": 3.53125,
+      "learning_rate": 1.7549085551913358e-06,
+      "loss": 0.6849,
+      "step": 1395
+    },
+    {
+      "epoch": 2.2585858585858585,
+      "grad_norm": 3.140625,
+      "learning_rate": 1.747749931953217e-06,
+      "loss": 0.7111,
+      "step": 1396
+    },
+    {
+      "epoch": 2.26020202020202,
+      "grad_norm": 3.671875,
+      "learning_rate": 1.7406028456769303e-06,
+      "loss": 0.7179,
+      "step": 1397
+    },
+    {
+      "epoch": 2.2618181818181817,
+      "grad_norm": 3.875,
+      "learning_rate": 1.7334673217158976e-06,
+      "loss": 0.6348,
+      "step": 1398
+    },
+    {
+      "epoch": 2.2634343434343434,
+      "grad_norm": 3.296875,
+      "learning_rate": 1.7263433853825184e-06,
+      "loss": 0.936,
+      "step": 1399
+    },
+    {
+      "epoch": 2.265050505050505,
+      "grad_norm": 3.59375,
+      "learning_rate": 1.719231061948094e-06,
+      "loss": 0.8846,
+      "step": 1400
+    },
+    {
+      "epoch": 2.2666666666666666,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.7121303766427266e-06,
+      "loss": 0.5793,
+      "step": 1401
+    },
+    {
+      "epoch": 2.2682828282828282,
+      "grad_norm": 2.875,
+      "learning_rate": 1.7050413546552347e-06,
+      "loss": 0.6675,
+      "step": 1402
+    },
+    {
+      "epoch": 2.26989898989899,
+      "grad_norm": 3.375,
+      "learning_rate": 1.6979640211330616e-06,
+      "loss": 0.8148,
+      "step": 1403
+    },
+    {
+      "epoch": 2.2715151515151515,
+      "grad_norm": 3.765625,
+      "learning_rate": 1.6908984011821883e-06,
+      "loss": 0.7067,
+      "step": 1404
+    },
+    {
+      "epoch": 2.273131313131313,
+      "grad_norm": 4.21875,
+      "learning_rate": 1.6838445198670423e-06,
+      "loss": 0.875,
+      "step": 1405
+    },
+    {
+      "epoch": 2.2747474747474747,
+      "grad_norm": 3.171875,
+      "learning_rate": 1.6768024022104106e-06,
+      "loss": 0.6799,
+      "step": 1406
+    },
+    {
+      "epoch": 2.2763636363636364,
+      "grad_norm": 3.9375,
+      "learning_rate": 1.669772073193352e-06,
+      "loss": 0.762,
+      "step": 1407
+    },
+    {
+      "epoch": 2.277979797979798,
+      "grad_norm": 3.734375,
+      "learning_rate": 1.6627535577550996e-06,
+      "loss": 0.8989,
+      "step": 1408
+    },
+    {
+      "epoch": 2.2795959595959596,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.655746880792986e-06,
+      "loss": 0.6661,
+      "step": 1409
+    },
+    {
+      "epoch": 2.2812121212121212,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.6487520671623469e-06,
+      "loss": 0.728,
+      "step": 1410
+    },
+    {
+      "epoch": 2.282828282828283,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.6417691416764325e-06,
+      "loss": 0.8858,
+      "step": 1411
+    },
+    {
+      "epoch": 2.2844444444444445,
+      "grad_norm": 3.296875,
+      "learning_rate": 1.6347981291063224e-06,
+      "loss": 1.0144,
+      "step": 1412
+    },
+    {
+      "epoch": 2.286060606060606,
+      "grad_norm": 3.125,
+      "learning_rate": 1.6278390541808364e-06,
+      "loss": 0.7793,
+      "step": 1413
+    },
+    {
+      "epoch": 2.2876767676767678,
+      "grad_norm": 3.5,
+      "learning_rate": 1.6208919415864476e-06,
+      "loss": 0.6699,
+      "step": 1414
+    },
+    {
+      "epoch": 2.2892929292929294,
+      "grad_norm": 3.796875,
+      "learning_rate": 1.6139568159671931e-06,
+      "loss": 0.7621,
+      "step": 1415
+    },
+    {
+      "epoch": 2.290909090909091,
+      "grad_norm": 3.1875,
+      "learning_rate": 1.6070337019245896e-06,
+      "loss": 0.8945,
+      "step": 1416
+    },
+    {
+      "epoch": 2.2925252525252526,
+      "grad_norm": 3.875,
+      "learning_rate": 1.60012262401754e-06,
+      "loss": 0.8163,
+      "step": 1417
+    },
+    {
+      "epoch": 2.2941414141414143,
+      "grad_norm": 3.328125,
+      "learning_rate": 1.5932236067622542e-06,
+      "loss": 0.8186,
+      "step": 1418
+    },
+    {
+      "epoch": 2.295757575757576,
+      "grad_norm": 3.671875,
+      "learning_rate": 1.5863366746321578e-06,
+      "loss": 1.2321,
+      "step": 1419
+    },
+    {
+      "epoch": 2.2973737373737375,
+      "grad_norm": 3.703125,
+      "learning_rate": 1.5794618520578053e-06,
+      "loss": 0.7436,
+      "step": 1420
+    },
+    {
+      "epoch": 2.298989898989899,
+      "grad_norm": 3.25,
+      "learning_rate": 1.5725991634267945e-06,
+      "loss": 0.6044,
+      "step": 1421
+    },
+    {
+      "epoch": 2.3006060606060608,
+      "grad_norm": 3.390625,
+      "learning_rate": 1.5657486330836786e-06,
+      "loss": 0.5315,
+      "step": 1422
+    },
+    {
+      "epoch": 2.3022222222222224,
+      "grad_norm": 3.03125,
+      "learning_rate": 1.5589102853298803e-06,
+      "loss": 0.4947,
+      "step": 1423
+    },
+    {
+      "epoch": 2.303838383838384,
+      "grad_norm": 3.453125,
+      "learning_rate": 1.5520841444236118e-06,
+      "loss": 0.8543,
+      "step": 1424
+    },
+    {
+      "epoch": 2.3054545454545456,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.5452702345797738e-06,
+      "loss": 0.5381,
+      "step": 1425
+    },
+    {
+      "epoch": 2.3070707070707073,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.5384685799698839e-06,
+      "loss": 0.6309,
+      "step": 1426
+    },
+    {
+      "epoch": 2.308686868686869,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.5316792047219864e-06,
+      "loss": 0.7624,
+      "step": 1427
+    },
+    {
+      "epoch": 2.3103030303030305,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.5249021329205638e-06,
+      "loss": 0.9326,
+      "step": 1428
+    },
+    {
+      "epoch": 2.311919191919192,
+      "grad_norm": 3.09375,
+      "learning_rate": 1.5181373886064549e-06,
+      "loss": 0.6245,
+      "step": 1429
+    },
+    {
+      "epoch": 2.3135353535353538,
+      "grad_norm": 3.53125,
+      "learning_rate": 1.5113849957767685e-06,
+      "loss": 0.6439,
+      "step": 1430
+    },
+    {
+      "epoch": 2.315151515151515,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.5046449783847965e-06,
+      "loss": 0.8069,
+      "step": 1431
+    },
+    {
+      "epoch": 2.3167676767676766,
+      "grad_norm": 3.6875,
+      "learning_rate": 1.4979173603399323e-06,
+      "loss": 0.9888,
+      "step": 1432
+    },
+    {
+      "epoch": 2.318383838383838,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.4912021655075842e-06,
+      "loss": 0.9978,
+      "step": 1433
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 3.515625,
+      "learning_rate": 1.4844994177090871e-06,
+      "loss": 0.9391,
+      "step": 1434
+    },
+    {
+      "epoch": 2.3216161616161615,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.4778091407216255e-06,
+      "loss": 1.0607,
+      "step": 1435
+    },
+    {
+      "epoch": 2.323232323232323,
+      "grad_norm": 3.171875,
+      "learning_rate": 1.4711313582781434e-06,
+      "loss": 1.1193,
+      "step": 1436
+    },
+    {
+      "epoch": 2.3248484848484847,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 0.8976,
+      "step": 1437
+    },
+    {
+      "epoch": 2.3264646464646463,
+      "grad_norm": 3.125,
+      "learning_rate": 1.4578133717331982e-06,
+      "loss": 0.7979,
+      "step": 1438
+    },
+    {
+      "epoch": 2.328080808080808,
+      "grad_norm": 3.421875,
+      "learning_rate": 1.451173214875674e-06,
+      "loss": 0.5247,
+      "step": 1439
+    },
+    {
+      "epoch": 2.3296969696969696,
+      "grad_norm": 3.234375,
+      "learning_rate": 1.4445456470498392e-06,
+      "loss": 0.8337,
+      "step": 1440
+    },
+    {
+      "epoch": 2.3313131313131312,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.437930691766185e-06,
+      "loss": 0.7524,
+      "step": 1441
+    },
+    {
+      "epoch": 2.332929292929293,
+      "grad_norm": 3.296875,
+      "learning_rate": 1.4313283724904632e-06,
+      "loss": 0.6592,
+      "step": 1442
+    },
+    {
+      "epoch": 2.3345454545454545,
+      "grad_norm": 3.359375,
+      "learning_rate": 1.4247387126435957e-06,
+      "loss": 0.9083,
+      "step": 1443
+    },
+    {
+      "epoch": 2.336161616161616,
+      "grad_norm": 3.125,
+      "learning_rate": 1.418161735601601e-06,
+      "loss": 1.2121,
+      "step": 1444
+    },
+    {
+      "epoch": 2.3377777777777777,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.4115974646955067e-06,
+      "loss": 0.5373,
+      "step": 1445
+    },
+    {
+      "epoch": 2.3393939393939394,
+      "grad_norm": 3.296875,
+      "learning_rate": 1.4050459232112652e-06,
+      "loss": 0.9544,
+      "step": 1446
+    },
+    {
+      "epoch": 2.341010101010101,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.398507134389674e-06,
+      "loss": 0.6236,
+      "step": 1447
+    },
+    {
+      "epoch": 2.3426262626262626,
+      "grad_norm": 3.59375,
+      "learning_rate": 1.3919811214262913e-06,
+      "loss": 0.5788,
+      "step": 1448
+    },
+    {
+      "epoch": 2.3442424242424242,
+      "grad_norm": 3.125,
+      "learning_rate": 1.3854679074713557e-06,
+      "loss": 0.656,
+      "step": 1449
+    },
+    {
+      "epoch": 2.345858585858586,
+      "grad_norm": 3.34375,
+      "learning_rate": 1.378967515629701e-06,
+      "loss": 0.8725,
+      "step": 1450
+    },
+    {
+      "epoch": 2.3474747474747475,
+      "grad_norm": 3.734375,
+      "learning_rate": 1.37247996896068e-06,
+      "loss": 0.7228,
+      "step": 1451
+    },
+    {
+      "epoch": 2.349090909090909,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.3660052904780707e-06,
+      "loss": 0.7864,
+      "step": 1452
+    },
+    {
+      "epoch": 2.3507070707070707,
+      "grad_norm": 3.1875,
+      "learning_rate": 1.359543503150012e-06,
+      "loss": 0.8786,
+      "step": 1453
+    },
+    {
+      "epoch": 2.3523232323232324,
+      "grad_norm": 3.078125,
+      "learning_rate": 1.353094629898909e-06,
+      "loss": 0.7266,
+      "step": 1454
+    },
+    {
+      "epoch": 2.353939393939394,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.3466586936013548e-06,
+      "loss": 0.6905,
+      "step": 1455
+    },
+    {
+      "epoch": 2.3555555555555556,
+      "grad_norm": 3.640625,
+      "learning_rate": 1.3402357170880514e-06,
+      "loss": 0.6917,
+      "step": 1456
+    },
+    {
+      "epoch": 2.3571717171717173,
+      "grad_norm": 3.0,
+      "learning_rate": 1.3338257231437267e-06,
+      "loss": 0.7492,
+      "step": 1457
+    },
+    {
+      "epoch": 2.358787878787879,
+      "grad_norm": 3.34375,
+      "learning_rate": 1.3274287345070564e-06,
+      "loss": 1.0009,
+      "step": 1458
+    },
+    {
+      "epoch": 2.3604040404040405,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.3210447738705807e-06,
+      "loss": 0.6621,
+      "step": 1459
+    },
+    {
+      "epoch": 2.362020202020202,
+      "grad_norm": 3.453125,
+      "learning_rate": 1.3146738638806217e-06,
+      "loss": 0.7287,
+      "step": 1460
+    },
+    {
+      "epoch": 2.3636363636363638,
+      "grad_norm": 3.296875,
+      "learning_rate": 1.3083160271372092e-06,
+      "loss": 0.4957,
+      "step": 1461
+    },
+    {
+      "epoch": 2.3652525252525254,
+      "grad_norm": 3.421875,
+      "learning_rate": 1.3019712861939964e-06,
+      "loss": 0.6317,
+      "step": 1462
+    },
+    {
+      "epoch": 2.366868686868687,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.2956396635581809e-06,
+      "loss": 0.715,
+      "step": 1463
+    },
+    {
+      "epoch": 2.3684848484848486,
+      "grad_norm": 3.5,
+      "learning_rate": 1.2893211816904243e-06,
+      "loss": 0.8714,
+      "step": 1464
+    },
+    {
+      "epoch": 2.3701010101010103,
+      "grad_norm": 3.375,
+      "learning_rate": 1.283015863004773e-06,
+      "loss": 0.7633,
+      "step": 1465
+    },
+    {
+      "epoch": 2.371717171717172,
+      "grad_norm": 3.09375,
+      "learning_rate": 1.2767237298685787e-06,
+      "loss": 1.025,
+      "step": 1466
+    },
+    {
+      "epoch": 2.3733333333333335,
+      "grad_norm": 2.953125,
+      "learning_rate": 1.2704448046024192e-06,
+      "loss": 0.5842,
+      "step": 1467
+    },
+    {
+      "epoch": 2.374949494949495,
+      "grad_norm": 3.375,
+      "learning_rate": 1.26417910948002e-06,
+      "loss": 0.9373,
+      "step": 1468
+    },
+    {
+      "epoch": 2.3765656565656563,
+      "grad_norm": 3.1875,
+      "learning_rate": 1.2579266667281697e-06,
+      "loss": 0.8847,
+      "step": 1469
+    },
+    {
+      "epoch": 2.378181818181818,
+      "grad_norm": 3.8125,
+      "learning_rate": 1.2516874985266508e-06,
+      "loss": 0.5596,
+      "step": 1470
+    },
+    {
+      "epoch": 2.3797979797979796,
+      "grad_norm": 3.671875,
+      "learning_rate": 1.2454616270081548e-06,
+      "loss": 1.0815,
+      "step": 1471
+    },
+    {
+      "epoch": 2.381414141414141,
+      "grad_norm": 3.015625,
+      "learning_rate": 1.239249074258203e-06,
+      "loss": 0.5137,
+      "step": 1472
+    },
+    {
+      "epoch": 2.383030303030303,
+      "grad_norm": 3.328125,
+      "learning_rate": 1.2330498623150722e-06,
+      "loss": 0.9805,
+      "step": 1473
+    },
+    {
+      "epoch": 2.3846464646464645,
+      "grad_norm": 3.5,
+      "learning_rate": 1.2268640131697129e-06,
+      "loss": 0.8201,
+      "step": 1474
+    },
+    {
+      "epoch": 2.386262626262626,
+      "grad_norm": 3.328125,
+      "learning_rate": 1.2206915487656734e-06,
+      "loss": 0.6376,
+      "step": 1475
+    },
+    {
+      "epoch": 2.3878787878787877,
+      "grad_norm": 3.1875,
+      "learning_rate": 1.2145324909990202e-06,
+      "loss": 0.6826,
+      "step": 1476
+    },
+    {
+      "epoch": 2.3878787878787877,
+      "eval_loss": 0.8126972913742065,
+      "eval_runtime": 34.0671,
+      "eval_samples_per_second": 32.289,
+      "eval_steps_per_second": 4.051,
+      "step": 1476
+    },
+    {
+      "epoch": 2.3894949494949493,
+      "grad_norm": 3.625,
+      "learning_rate": 1.2083868617182647e-06,
+      "loss": 1.0168,
+      "step": 1477
+    },
+    {
+      "epoch": 2.391111111111111,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.202254682724276e-06,
+      "loss": 0.7223,
+      "step": 1478
+    },
+    {
+      "epoch": 2.3927272727272726,
+      "grad_norm": 3.25,
+      "learning_rate": 1.1961359757702151e-06,
+      "loss": 0.7447,
+      "step": 1479
+    },
+    {
+      "epoch": 2.394343434343434,
+      "grad_norm": 3.640625,
+      "learning_rate": 1.190030762561452e-06,
+      "loss": 0.9484,
+      "step": 1480
+    },
+    {
+      "epoch": 2.395959595959596,
+      "grad_norm": 4.46875,
+      "learning_rate": 1.1839390647554888e-06,
+      "loss": 0.7267,
+      "step": 1481
+    },
+    {
+      "epoch": 2.3975757575757575,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.1778609039618804e-06,
+      "loss": 0.9313,
+      "step": 1482
+    },
+    {
+      "epoch": 2.399191919191919,
+      "grad_norm": 3.5625,
+      "learning_rate": 1.1717963017421669e-06,
+      "loss": 0.5965,
+      "step": 1483
+    },
+    {
+      "epoch": 2.4008080808080807,
+      "grad_norm": 3.4375,
+      "learning_rate": 1.1657452796097879e-06,
+      "loss": 0.6206,
+      "step": 1484
+    },
+    {
+      "epoch": 2.4024242424242424,
+      "grad_norm": 3.59375,
+      "learning_rate": 1.1597078590300054e-06,
+      "loss": 0.8369,
+      "step": 1485
+    },
+    {
+      "epoch": 2.404040404040404,
+      "grad_norm": 3.140625,
+      "learning_rate": 1.1536840614198376e-06,
+      "loss": 0.6965,
+      "step": 1486
+    },
+    {
+      "epoch": 2.4056565656565656,
+      "grad_norm": 3.53125,
+      "learning_rate": 1.147673908147973e-06,
+      "loss": 0.7485,
+      "step": 1487
+    },
+    {
+      "epoch": 2.4072727272727272,
+      "grad_norm": 3.25,
+      "learning_rate": 1.1416774205347015e-06,
+      "loss": 0.8467,
+      "step": 1488
+    },
+    {
+      "epoch": 2.408888888888889,
+      "grad_norm": 3.140625,
+      "learning_rate": 1.1356946198518325e-06,
+      "loss": 1.0472,
+      "step": 1489
+    },
+    {
+      "epoch": 2.4105050505050505,
+      "grad_norm": 3.3125,
+      "learning_rate": 1.1297255273226254e-06,
+      "loss": 0.8043,
+      "step": 1490
+    },
+    {
+      "epoch": 2.412121212121212,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.1237701641217097e-06,
+      "loss": 1.018,
+      "step": 1491
+    },
+    {
+      "epoch": 2.4137373737373737,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.117828551375013e-06,
+      "loss": 0.6631,
+      "step": 1492
+    },
+    {
+      "epoch": 2.4153535353535354,
+      "grad_norm": 3.359375,
+      "learning_rate": 1.111900710159684e-06,
+      "loss": 0.7284,
+      "step": 1493
+    },
+    {
+      "epoch": 2.416969696969697,
+      "grad_norm": 3.640625,
+      "learning_rate": 1.1059866615040205e-06,
+      "loss": 1.0587,
+      "step": 1494
+    },
+    {
+      "epoch": 2.4185858585858586,
+      "grad_norm": 3.21875,
+      "learning_rate": 1.100086426387389e-06,
+      "loss": 0.479,
+      "step": 1495
+    },
+    {
+      "epoch": 2.4202020202020202,
+      "grad_norm": 3.1875,
+      "learning_rate": 1.094200025740157e-06,
+      "loss": 1.0917,
+      "step": 1496
+    },
+    {
+      "epoch": 2.421818181818182,
+      "grad_norm": 3.578125,
+      "learning_rate": 1.0883274804436155e-06,
+      "loss": 0.9734,
+      "step": 1497
+    },
+    {
+      "epoch": 2.4234343434343435,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.0824688113299054e-06,
+      "loss": 1.038,
+      "step": 1498
+    },
+    {
+      "epoch": 2.425050505050505,
+      "grad_norm": 3.703125,
+      "learning_rate": 1.0766240391819438e-06,
+      "loss": 0.7082,
+      "step": 1499
+    },
+    {
+      "epoch": 2.4266666666666667,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.0707931847333487e-06,
+      "loss": 0.8108,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4282828282828284,
+      "grad_norm": 3.4375,
+      "learning_rate": 1.064976268668368e-06,
+      "loss": 1.021,
+      "step": 1501
+    },
+    {
+      "epoch": 2.42989898989899,
+      "grad_norm": 3.359375,
+      "learning_rate": 1.0591733116218046e-06,
+      "loss": 0.7082,
+      "step": 1502
+    },
+    {
+      "epoch": 2.4315151515151516,
+      "grad_norm": 3.34375,
+      "learning_rate": 1.053384334178944e-06,
+      "loss": 0.7176,
+      "step": 1503
+    },
+    {
+      "epoch": 2.4331313131313133,
+      "grad_norm": 3.09375,
+      "learning_rate": 1.0476093568754776e-06,
+      "loss": 1.0279,
+      "step": 1504
+    },
+    {
+      "epoch": 2.434747474747475,
+      "grad_norm": 3.65625,
+      "learning_rate": 1.0418484001974372e-06,
+      "loss": 0.8781,
+      "step": 1505
+    },
+    {
+      "epoch": 2.4363636363636365,
+      "grad_norm": 3.328125,
+      "learning_rate": 1.036101484581117e-06,
+      "loss": 0.6487,
+      "step": 1506
+    },
+    {
+      "epoch": 2.437979797979798,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.0303686304129995e-06,
+      "loss": 0.7354,
+      "step": 1507
+    },
+    {
+      "epoch": 2.4395959595959598,
+      "grad_norm": 3.3125,
+      "learning_rate": 1.0246498580296903e-06,
+      "loss": 0.9746,
+      "step": 1508
+    },
+    {
+      "epoch": 2.4412121212121214,
+      "grad_norm": 3.296875,
+      "learning_rate": 1.0189451877178386e-06,
+      "loss": 0.8807,
+      "step": 1509
+    },
+    {
+      "epoch": 2.442828282828283,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.0132546397140687e-06,
+      "loss": 0.8349,
+      "step": 1510
+    },
+    {
+      "epoch": 2.4444444444444446,
+      "grad_norm": 2.984375,
+      "learning_rate": 1.0075782342049084e-06,
+      "loss": 0.8903,
+      "step": 1511
+    },
+    {
+      "epoch": 2.4460606060606063,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.0019159913267156e-06,
+      "loss": 0.7151,
+      "step": 1512
+    },
+    {
+      "epoch": 2.447676767676768,
+      "grad_norm": 3.21875,
+      "learning_rate": 9.9626793116561e-07,
+      "loss": 0.6079,
+      "step": 1513
+    },
+    {
+      "epoch": 2.4492929292929295,
+      "grad_norm": 3.1875,
+      "learning_rate": 9.90634073757397e-07,
+      "loss": 0.8021,
+      "step": 1514
+    },
+    {
+      "epoch": 2.450909090909091,
+      "grad_norm": 3.25,
+      "learning_rate": 9.850144390875022e-07,
+      "loss": 0.6596,
+      "step": 1515
+    },
+    {
+      "epoch": 2.4525252525252528,
+      "grad_norm": 3.140625,
+      "learning_rate": 9.794090470908962e-07,
+      "loss": 0.6563,
+      "step": 1516
+    },
+    {
+      "epoch": 2.454141414141414,
+      "grad_norm": 3.1875,
+      "learning_rate": 9.738179176520256e-07,
+      "loss": 0.8274,
+      "step": 1517
+    },
+    {
+      "epoch": 2.4557575757575756,
+      "grad_norm": 3.171875,
+      "learning_rate": 9.68241070604743e-07,
+      "loss": 0.6864,
+      "step": 1518
+    },
+    {
+      "epoch": 2.457373737373737,
+      "grad_norm": 3.3125,
+      "learning_rate": 9.626785257322352e-07,
+      "loss": 0.5913,
+      "step": 1519
+    },
+    {
+      "epoch": 2.458989898989899,
+      "grad_norm": 3.015625,
+      "learning_rate": 9.571303027669548e-07,
+      "loss": 0.5952,
+      "step": 1520
+    },
+    {
+      "epoch": 2.4606060606060605,
+      "grad_norm": 2.984375,
+      "learning_rate": 9.51596421390547e-07,
+      "loss": 0.6674,
+      "step": 1521
+    },
+    {
+      "epoch": 2.462222222222222,
+      "grad_norm": 3.484375,
+      "learning_rate": 9.460769012337839e-07,
+      "loss": 0.9879,
+      "step": 1522
+    },
+    {
+      "epoch": 2.4638383838383837,
+      "grad_norm": 3.078125,
+      "learning_rate": 9.405717618764925e-07,
+      "loss": 0.7607,
+      "step": 1523
+    },
+    {
+      "epoch": 2.4654545454545453,
+      "grad_norm": 3.625,
+      "learning_rate": 9.350810228474855e-07,
+      "loss": 0.9173,
+      "step": 1524
+    },
+    {
+      "epoch": 2.467070707070707,
+      "grad_norm": 3.140625,
+      "learning_rate": 9.296047036244926e-07,
+      "loss": 0.8267,
+      "step": 1525
+    },
+    {
+      "epoch": 2.4686868686868686,
+      "grad_norm": 3.453125,
+      "learning_rate": 9.241428236340904e-07,
+      "loss": 0.7018,
+      "step": 1526
+    },
+    {
+      "epoch": 2.4703030303030302,
+      "grad_norm": 3.1875,
+      "learning_rate": 9.186954022516343e-07,
+      "loss": 0.6323,
+      "step": 1527
+    },
+    {
+      "epoch": 2.471919191919192,
+      "grad_norm": 3.40625,
+      "learning_rate": 9.132624588011896e-07,
+      "loss": 0.6411,
+      "step": 1528
+    },
+    {
+      "epoch": 2.4735353535353535,
+      "grad_norm": 3.421875,
+      "learning_rate": 9.078440125554639e-07,
+      "loss": 0.779,
+      "step": 1529
+    },
+    {
+      "epoch": 2.475151515151515,
+      "grad_norm": 3.953125,
+      "learning_rate": 9.024400827357344e-07,
+      "loss": 1.1329,
+      "step": 1530
+    },
+    {
+      "epoch": 2.4767676767676767,
+      "grad_norm": 3.03125,
+      "learning_rate": 8.970506885117868e-07,
+      "loss": 0.8261,
+      "step": 1531
+    },
+    {
+      "epoch": 2.4783838383838384,
+      "grad_norm": 3.515625,
+      "learning_rate": 8.916758490018418e-07,
+      "loss": 0.6912,
+      "step": 1532
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 2.8125,
+      "learning_rate": 8.863155832724895e-07,
+      "loss": 0.7447,
+      "step": 1533
+    },
+    {
+      "epoch": 2.4816161616161616,
+      "grad_norm": 2.890625,
+      "learning_rate": 8.809699103386204e-07,
+      "loss": 0.9137,
+      "step": 1534
+    },
+    {
+      "epoch": 2.4832323232323232,
+      "grad_norm": 3.46875,
+      "learning_rate": 8.756388491633605e-07,
+      "loss": 0.6883,
+      "step": 1535
+    },
+    {
+      "epoch": 2.484848484848485,
+      "grad_norm": 3.359375,
+      "learning_rate": 8.703224186580012e-07,
+      "loss": 0.8061,
+      "step": 1536
+    },
+    {
+      "epoch": 2.4864646464646465,
+      "grad_norm": 3.21875,
+      "learning_rate": 8.650206376819331e-07,
+      "loss": 0.8854,
+      "step": 1537
+    },
+    {
+      "epoch": 2.488080808080808,
+      "grad_norm": 3.25,
+      "learning_rate": 8.597335250425809e-07,
+      "loss": 0.9307,
+      "step": 1538
+    },
+    {
+      "epoch": 2.4896969696969697,
+      "grad_norm": 3.015625,
+      "learning_rate": 8.544610994953317e-07,
+      "loss": 0.9382,
+      "step": 1539
+    },
+    {
+      "epoch": 2.4913131313131314,
+      "grad_norm": 3.34375,
+      "learning_rate": 8.492033797434762e-07,
+      "loss": 0.7716,
+      "step": 1540
+    },
+    {
+      "epoch": 2.492929292929293,
+      "grad_norm": 3.28125,
+      "learning_rate": 8.439603844381361e-07,
+      "loss": 0.5451,
+      "step": 1541
+    },
+    {
+      "epoch": 2.4945454545454546,
+      "grad_norm": 3.203125,
+      "learning_rate": 8.387321321781977e-07,
+      "loss": 0.9585,
+      "step": 1542
+    },
+    {
+      "epoch": 2.4961616161616162,
+      "grad_norm": 3.546875,
+      "learning_rate": 8.335186415102541e-07,
+      "loss": 0.7767,
+      "step": 1543
+    },
+    {
+      "epoch": 2.497777777777778,
+      "grad_norm": 3.21875,
+      "learning_rate": 8.283199309285284e-07,
+      "loss": 0.9783,
+      "step": 1544
+    },
+    {
+      "epoch": 2.4993939393939395,
+      "grad_norm": 3.546875,
+      "learning_rate": 8.23136018874815e-07,
+      "loss": 0.8914,
+      "step": 1545
+    },
+    {
+      "epoch": 2.501010101010101,
+      "grad_norm": 2.84375,
+      "learning_rate": 8.179669237384097e-07,
+      "loss": 0.4491,
+      "step": 1546
+    },
+    {
+      "epoch": 2.5026262626262628,
+      "grad_norm": 3.125,
+      "learning_rate": 8.128126638560507e-07,
+      "loss": 0.7482,
+      "step": 1547
+    },
+    {
+      "epoch": 2.5042424242424244,
+      "grad_norm": 3.34375,
+      "learning_rate": 8.07673257511849e-07,
+      "loss": 0.9324,
+      "step": 1548
+    },
+    {
+      "epoch": 2.505858585858586,
+      "grad_norm": 3.484375,
+      "learning_rate": 8.025487229372225e-07,
+      "loss": 0.9344,
+      "step": 1549
+    },
+    {
+      "epoch": 2.5074747474747476,
+      "grad_norm": 2.984375,
+      "learning_rate": 7.97439078310836e-07,
+      "loss": 0.7426,
+      "step": 1550
+    },
+    {
+      "epoch": 2.509090909090909,
+      "grad_norm": 2.96875,
+      "learning_rate": 7.923443417585324e-07,
+      "loss": 0.5285,
+      "step": 1551
+    },
+    {
+      "epoch": 2.5107070707070704,
+      "grad_norm": 3.15625,
+      "learning_rate": 7.872645313532701e-07,
+      "loss": 0.6484,
+      "step": 1552
+    },
+    {
+      "epoch": 2.512323232323232,
+      "grad_norm": 3.046875,
+      "learning_rate": 7.821996651150598e-07,
+      "loss": 0.5095,
+      "step": 1553
+    },
+    {
+      "epoch": 2.5139393939393937,
+      "grad_norm": 3.171875,
+      "learning_rate": 7.771497610108981e-07,
+      "loss": 0.6595,
+      "step": 1554
+    },
+    {
+      "epoch": 2.5155555555555553,
+      "grad_norm": 3.53125,
+      "learning_rate": 7.721148369547077e-07,
+      "loss": 0.6742,
+      "step": 1555
+    },
+    {
+      "epoch": 2.517171717171717,
+      "grad_norm": 3.140625,
+      "learning_rate": 7.670949108072673e-07,
+      "loss": 0.8343,
+      "step": 1556
+    },
+    {
+      "epoch": 2.5187878787878786,
+      "grad_norm": 3.296875,
+      "learning_rate": 7.620900003761561e-07,
+      "loss": 0.7357,
+      "step": 1557
+    },
+    {
+      "epoch": 2.52040404040404,
+      "grad_norm": 3.234375,
+      "learning_rate": 7.57100123415685e-07,
+      "loss": 1.2226,
+      "step": 1558
+    },
+    {
+      "epoch": 2.522020202020202,
+      "grad_norm": 3.15625,
+      "learning_rate": 7.521252976268362e-07,
+      "loss": 0.6258,
+      "step": 1559
+    },
+    {
+      "epoch": 2.5236363636363635,
+      "grad_norm": 3.21875,
+      "learning_rate": 7.471655406572003e-07,
+      "loss": 0.6993,
+      "step": 1560
+    },
+    {
+      "epoch": 2.525252525252525,
+      "grad_norm": 3.4375,
+      "learning_rate": 7.422208701009114e-07,
+      "loss": 0.7232,
+      "step": 1561
+    },
+    {
+      "epoch": 2.5268686868686867,
+      "grad_norm": 3.296875,
+      "learning_rate": 7.372913034985879e-07,
+      "loss": 0.5537,
+      "step": 1562
+    },
+    {
+      "epoch": 2.5284848484848483,
+      "grad_norm": 3.3125,
+      "learning_rate": 7.323768583372675e-07,
+      "loss": 0.8405,
+      "step": 1563
+    },
+    {
+      "epoch": 2.53010101010101,
+      "grad_norm": 3.390625,
+      "learning_rate": 7.274775520503491e-07,
+      "loss": 0.6211,
+      "step": 1564
+    },
+    {
+      "epoch": 2.5317171717171716,
+      "grad_norm": 3.0625,
+      "learning_rate": 7.225934020175235e-07,
+      "loss": 0.8441,
+      "step": 1565
+    },
+    {
+      "epoch": 2.533333333333333,
+      "grad_norm": 3.21875,
+      "learning_rate": 7.177244255647209e-07,
+      "loss": 0.804,
+      "step": 1566
+    },
+    {
+      "epoch": 2.534949494949495,
+      "grad_norm": 3.375,
+      "learning_rate": 7.128706399640428e-07,
+      "loss": 0.6561,
+      "step": 1567
+    },
+    {
+      "epoch": 2.5365656565656565,
+      "grad_norm": 3.21875,
+      "learning_rate": 7.080320624337039e-07,
+      "loss": 0.6208,
+      "step": 1568
+    },
+    {
+      "epoch": 2.538181818181818,
+      "grad_norm": 3.15625,
+      "learning_rate": 7.032087101379703e-07,
+      "loss": 0.8447,
+      "step": 1569
+    },
+    {
+      "epoch": 2.5397979797979797,
+      "grad_norm": 3.390625,
+      "learning_rate": 6.984006001870974e-07,
+      "loss": 0.896,
+      "step": 1570
+    },
+    {
+      "epoch": 2.5414141414141413,
+      "grad_norm": 3.640625,
+      "learning_rate": 6.936077496372711e-07,
+      "loss": 0.7503,
+      "step": 1571
+    },
+    {
+      "epoch": 2.543030303030303,
+      "grad_norm": 3.25,
+      "learning_rate": 6.888301754905469e-07,
+      "loss": 0.7943,
+      "step": 1572
+    },
+    {
+      "epoch": 2.5446464646464646,
+      "grad_norm": 3.6875,
+      "learning_rate": 6.840678946947871e-07,
+      "loss": 0.8684,
+      "step": 1573
+    },
+    {
+      "epoch": 2.5462626262626262,
+      "grad_norm": 3.109375,
+      "learning_rate": 6.79320924143605e-07,
+      "loss": 0.5932,
+      "step": 1574
+    },
+    {
+      "epoch": 2.547878787878788,
+      "grad_norm": 3.125,
+      "learning_rate": 6.745892806763015e-07,
+      "loss": 0.8376,
+      "step": 1575
+    },
+    {
+      "epoch": 2.5494949494949495,
+      "grad_norm": 3.96875,
+      "learning_rate": 6.698729810778065e-07,
+      "loss": 0.7034,
+      "step": 1576
+    },
+    {
+      "epoch": 2.551111111111111,
+      "grad_norm": 3.734375,
+      "learning_rate": 6.6517204207862e-07,
+      "loss": 0.8909,
+      "step": 1577
+    },
+    {
+      "epoch": 2.5527272727272727,
+      "grad_norm": 2.90625,
+      "learning_rate": 6.604864803547511e-07,
+      "loss": 0.5081,
+      "step": 1578
+    },
+    {
+      "epoch": 2.5543434343434344,
+      "grad_norm": 3.1875,
+      "learning_rate": 6.558163125276612e-07,
+      "loss": 0.5302,
+      "step": 1579
+    },
+    {
+      "epoch": 2.555959595959596,
+      "grad_norm": 3.65625,
+      "learning_rate": 6.51161555164203e-07,
+      "loss": 1.1241,
+      "step": 1580
+    },
+    {
+      "epoch": 2.5575757575757576,
+      "grad_norm": 3.546875,
+      "learning_rate": 6.465222247765618e-07,
+      "loss": 0.8271,
+      "step": 1581
+    },
+    {
+      "epoch": 2.5591919191919192,
+      "grad_norm": 3.40625,
+      "learning_rate": 6.418983378221988e-07,
+      "loss": 0.9064,
+      "step": 1582
+    },
+    {
+      "epoch": 2.560808080808081,
+      "grad_norm": 3.15625,
+      "learning_rate": 6.372899107037905e-07,
+      "loss": 0.624,
+      "step": 1583
+    },
+    {
+      "epoch": 2.5624242424242425,
+      "grad_norm": 3.828125,
+      "learning_rate": 6.326969597691724e-07,
+      "loss": 0.9117,
+      "step": 1584
+    },
+    {
+      "epoch": 2.564040404040404,
+      "grad_norm": 3.390625,
+      "learning_rate": 6.281195013112795e-07,
+      "loss": 0.8046,
+      "step": 1585
+    },
+    {
+      "epoch": 2.5656565656565657,
+      "grad_norm": 3.640625,
+      "learning_rate": 6.235575515680898e-07,
+      "loss": 0.7662,
+      "step": 1586
+    },
+    {
+      "epoch": 2.5672727272727274,
+      "grad_norm": 3.4375,
+      "learning_rate": 6.190111267225651e-07,
+      "loss": 0.6031,
+      "step": 1587
+    },
+    {
+      "epoch": 2.568888888888889,
+      "grad_norm": 3.375,
+      "learning_rate": 6.144802429025948e-07,
+      "loss": 0.6936,
+      "step": 1588
+    },
+    {
+      "epoch": 2.5705050505050506,
+      "grad_norm": 2.9375,
+      "learning_rate": 6.09964916180939e-07,
+      "loss": 0.6361,
+      "step": 1589
+    },
+    {
+      "epoch": 2.5721212121212123,
+      "grad_norm": 3.046875,
+      "learning_rate": 6.054651625751717e-07,
+      "loss": 0.621,
+      "step": 1590
+    },
+    {
+      "epoch": 2.573737373737374,
+      "grad_norm": 3.34375,
+      "learning_rate": 6.009809980476189e-07,
+      "loss": 0.9012,
+      "step": 1591
+    },
+    {
+      "epoch": 2.5753535353535355,
+      "grad_norm": 3.296875,
+      "learning_rate": 5.965124385053112e-07,
+      "loss": 0.8539,
+      "step": 1592
+    },
+    {
+      "epoch": 2.576969696969697,
+      "grad_norm": 3.09375,
+      "learning_rate": 5.920594997999202e-07,
+      "loss": 0.7713,
+      "step": 1593
+    },
+    {
+      "epoch": 2.5785858585858588,
+      "grad_norm": 3.171875,
+      "learning_rate": 5.876221977277042e-07,
+      "loss": 0.412,
+      "step": 1594
+    },
+    {
+      "epoch": 2.5802020202020204,
+      "grad_norm": 3.109375,
+      "learning_rate": 5.832005480294539e-07,
+      "loss": 0.6625,
+      "step": 1595
+    },
+    {
+      "epoch": 2.581818181818182,
+      "grad_norm": 3.6875,
+      "learning_rate": 5.787945663904332e-07,
+      "loss": 0.9274,
+      "step": 1596
+    },
+    {
+      "epoch": 2.5834343434343436,
+      "grad_norm": 3.265625,
+      "learning_rate": 5.744042684403273e-07,
+      "loss": 1.1282,
+      "step": 1597
+    },
+    {
+      "epoch": 2.5850505050505053,
+      "grad_norm": 3.296875,
+      "learning_rate": 5.700296697531843e-07,
+      "loss": 0.8945,
+      "step": 1598
+    },
+    {
+      "epoch": 2.586666666666667,
+      "grad_norm": 3.375,
+      "learning_rate": 5.65670785847362e-07,
+      "loss": 0.7712,
+      "step": 1599
+    },
+    {
+      "epoch": 2.586666666666667,
+      "eval_loss": 0.8117458820343018,
+      "eval_runtime": 34.0478,
+      "eval_samples_per_second": 32.308,
+      "eval_steps_per_second": 4.053,
+      "step": 1599
+    },
+    {
+      "epoch": 2.5882828282828285,
+      "grad_norm": 3.828125,
+      "learning_rate": 5.613276321854699e-07,
+      "loss": 0.7403,
+      "step": 1600
+    },
+    {
+      "epoch": 2.58989898989899,
+      "grad_norm": 3.296875,
+      "learning_rate": 5.570002241743173e-07,
+      "loss": 0.7457,
+      "step": 1601
+    },
+    {
+      "epoch": 2.5915151515151518,
+      "grad_norm": 3.1875,
+      "learning_rate": 5.526885771648599e-07,
+      "loss": 0.7863,
+      "step": 1602
+    },
+    {
+      "epoch": 2.5931313131313134,
+      "grad_norm": 3.046875,
+      "learning_rate": 5.48392706452141e-07,
+      "loss": 0.7826,
+      "step": 1603
+    },
+    {
+      "epoch": 2.594747474747475,
+      "grad_norm": 3.421875,
+      "learning_rate": 5.441126272752395e-07,
+      "loss": 0.9158,
+      "step": 1604
+    },
+    {
+      "epoch": 2.596363636363636,
+      "grad_norm": 3.421875,
+      "learning_rate": 5.398483548172162e-07,
+      "loss": 0.9169,
+      "step": 1605
+    },
+    {
+      "epoch": 2.597979797979798,
+      "grad_norm": 3.296875,
+      "learning_rate": 5.355999042050603e-07,
+      "loss": 0.8839,
+      "step": 1606
+    },
+    {
+      "epoch": 2.5995959595959595,
+      "grad_norm": 3.265625,
+      "learning_rate": 5.313672905096323e-07,
+      "loss": 0.7788,
+      "step": 1607
+    },
+    {
+      "epoch": 2.601212121212121,
+      "grad_norm": 3.125,
+      "learning_rate": 5.271505287456153e-07,
+      "loss": 0.813,
+      "step": 1608
+    },
+    {
+      "epoch": 2.6028282828282827,
+      "grad_norm": 3.59375,
+      "learning_rate": 5.229496338714596e-07,
+      "loss": 0.9984,
+      "step": 1609
+    },
+    {
+      "epoch": 2.6044444444444443,
+      "grad_norm": 3.359375,
+      "learning_rate": 5.187646207893287e-07,
+      "loss": 1.0605,
+      "step": 1610
+    },
+    {
+      "epoch": 2.606060606060606,
+      "grad_norm": 3.125,
+      "learning_rate": 5.145955043450484e-07,
+      "loss": 0.892,
+      "step": 1611
+    },
+    {
+      "epoch": 2.6076767676767676,
+      "grad_norm": 3.359375,
+      "learning_rate": 5.104422993280522e-07,
+      "loss": 0.7316,
+      "step": 1612
+    },
+    {
+      "epoch": 2.609292929292929,
+      "grad_norm": 3.0,
+      "learning_rate": 5.063050204713299e-07,
+      "loss": 0.8063,
+      "step": 1613
+    },
+    {
+      "epoch": 2.610909090909091,
+      "grad_norm": 3.328125,
+      "learning_rate": 5.021836824513759e-07,
+      "loss": 0.6119,
+      "step": 1614
+    },
+    {
+      "epoch": 2.6125252525252525,
+      "grad_norm": 3.390625,
+      "learning_rate": 4.980782998881362e-07,
+      "loss": 0.7325,
+      "step": 1615
+    },
+    {
+      "epoch": 2.614141414141414,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.939888873449567e-07,
+      "loss": 0.7797,
+      "step": 1616
+    },
+    {
+      "epoch": 2.6157575757575757,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.899154593285294e-07,
+      "loss": 0.6084,
+      "step": 1617
+    },
+    {
+      "epoch": 2.6173737373737374,
+      "grad_norm": 3.015625,
+      "learning_rate": 4.858580302888466e-07,
+      "loss": 1.1623,
+      "step": 1618
+    },
+    {
+      "epoch": 2.618989898989899,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.818166146191438e-07,
+      "loss": 0.8682,
+      "step": 1619
+    },
+    {
+      "epoch": 2.6206060606060606,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.777912266558532e-07,
+      "loss": 0.7321,
+      "step": 1620
+    },
+    {
+      "epoch": 2.6222222222222222,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.7378188067854824e-07,
+      "loss": 0.834,
+      "step": 1621
+    },
+    {
+      "epoch": 2.623838383838384,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.6978859090989703e-07,
+      "loss": 0.5543,
+      "step": 1622
+    },
+    {
+      "epoch": 2.6254545454545455,
+      "grad_norm": 2.96875,
+      "learning_rate": 4.658113715156098e-07,
+      "loss": 1.0021,
+      "step": 1623
+    },
+    {
+      "epoch": 2.627070707070707,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.618502366043881e-07,
+      "loss": 0.9864,
+      "step": 1624
+    },
+    {
+      "epoch": 2.6286868686868687,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.5790520022787887e-07,
+      "loss": 0.6257,
+      "step": 1625
+    },
+    {
+      "epoch": 2.6303030303030304,
+      "grad_norm": 3.390625,
+      "learning_rate": 4.5397627638061604e-07,
+      "loss": 0.805,
+      "step": 1626
+    },
+    {
+      "epoch": 2.631919191919192,
+      "grad_norm": 3.1875,
+      "learning_rate": 4.500634789999808e-07,
+      "loss": 1.0041,
+      "step": 1627
+    },
+    {
+      "epoch": 2.6335353535353536,
+      "grad_norm": 3.375,
+      "learning_rate": 4.4616682196614636e-07,
+      "loss": 0.872,
+      "step": 1628
+    },
+    {
+      "epoch": 2.6351515151515152,
+      "grad_norm": 3.234375,
+      "learning_rate": 4.422863191020299e-07,
+      "loss": 0.9557,
+      "step": 1629
+    },
+    {
+      "epoch": 2.636767676767677,
+      "grad_norm": 3.421875,
+      "learning_rate": 4.3842198417324346e-07,
+      "loss": 0.845,
+      "step": 1630
+    },
+    {
+      "epoch": 2.6383838383838385,
+      "grad_norm": 3.34375,
+      "learning_rate": 4.345738308880454e-07,
+      "loss": 0.6979,
+      "step": 1631
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 3.109375,
+      "learning_rate": 4.307418728972934e-07,
+      "loss": 0.5708,
+      "step": 1632
+    },
+    {
+      "epoch": 2.6416161616161618,
+      "grad_norm": 3.53125,
+      "learning_rate": 4.269261237943911e-07,
+      "loss": 1.0834,
+      "step": 1633
+    },
+    {
+      "epoch": 2.6432323232323234,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.2312659711524486e-07,
+      "loss": 0.7123,
+      "step": 1634
+    },
+    {
+      "epoch": 2.644848484848485,
+      "grad_norm": 3.296875,
+      "learning_rate": 4.193433063382135e-07,
+      "loss": 0.9179,
+      "step": 1635
+    },
+    {
+      "epoch": 2.6464646464646466,
+      "grad_norm": 3.484375,
+      "learning_rate": 4.1557626488406223e-07,
+      "loss": 0.9971,
+      "step": 1636
+    },
+    {
+      "epoch": 2.648080808080808,
+      "grad_norm": 3.078125,
+      "learning_rate": 4.118254861159116e-07,
+      "loss": 0.5714,
+      "step": 1637
+    },
+    {
+      "epoch": 2.6496969696969694,
+      "grad_norm": 3.25,
+      "learning_rate": 4.080909833391944e-07,
+      "loss": 0.5749,
+      "step": 1638
+    },
+    {
+      "epoch": 2.651313131313131,
+      "grad_norm": 3.203125,
+      "learning_rate": 4.0437276980160477e-07,
+      "loss": 0.5459,
+      "step": 1639
+    },
+    {
+      "epoch": 2.6529292929292927,
+      "grad_norm": 3.96875,
+      "learning_rate": 4.0067085869305357e-07,
+      "loss": 0.6293,
+      "step": 1640
+    },
+    {
+      "epoch": 2.6545454545454543,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.9698526314562114e-07,
+      "loss": 0.6402,
+      "step": 1641
+    },
+    {
+      "epoch": 2.656161616161616,
+      "grad_norm": 3.109375,
+      "learning_rate": 3.9331599623350815e-07,
+      "loss": 0.7701,
+      "step": 1642
+    },
+    {
+      "epoch": 2.6577777777777776,
+      "grad_norm": 3.5,
+      "learning_rate": 3.896630709729943e-07,
+      "loss": 0.9759,
+      "step": 1643
+    },
+    {
+      "epoch": 2.659393939393939,
+      "grad_norm": 3.609375,
+      "learning_rate": 3.8602650032238675e-07,
+      "loss": 1.1406,
+      "step": 1644
+    },
+    {
+      "epoch": 2.661010101010101,
+      "grad_norm": 3.15625,
+      "learning_rate": 3.824062971819792e-07,
+      "loss": 0.8669,
+      "step": 1645
+    },
+    {
+      "epoch": 2.6626262626262625,
+      "grad_norm": 3.359375,
+      "learning_rate": 3.788024743940016e-07,
+      "loss": 0.9318,
+      "step": 1646
+    },
+    {
+      "epoch": 2.664242424242424,
+      "grad_norm": 3.484375,
+      "learning_rate": 3.752150447425773e-07,
+      "loss": 0.5965,
+      "step": 1647
+    },
+    {
+      "epoch": 2.6658585858585857,
+      "grad_norm": 3.296875,
+      "learning_rate": 3.71644020953677e-07,
+      "loss": 0.7127,
+      "step": 1648
+    },
+    {
+      "epoch": 2.6674747474747473,
+      "grad_norm": 3.109375,
+      "learning_rate": 3.680894156950743e-07,
+      "loss": 0.6395,
+      "step": 1649
+    },
+    {
+      "epoch": 2.669090909090909,
+      "grad_norm": 3.453125,
+      "learning_rate": 3.6455124157629805e-07,
+      "loss": 1.0619,
+      "step": 1650
+    },
+    {
+      "epoch": 2.6707070707070706,
+      "grad_norm": 3.578125,
+      "learning_rate": 3.610295111485923e-07,
+      "loss": 0.8617,
+      "step": 1651
+    },
+    {
+      "epoch": 2.672323232323232,
+      "grad_norm": 3.25,
+      "learning_rate": 3.575242369048665e-07,
+      "loss": 1.0097,
+      "step": 1652
+    },
+    {
+      "epoch": 2.673939393939394,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.5403543127965514e-07,
+      "loss": 0.8187,
+      "step": 1653
+    },
+    {
+      "epoch": 2.6755555555555555,
+      "grad_norm": 3.328125,
+      "learning_rate": 3.505631066490728e-07,
+      "loss": 0.9833,
+      "step": 1654
+    },
+    {
+      "epoch": 2.677171717171717,
+      "grad_norm": 3.25,
+      "learning_rate": 3.4710727533076806e-07,
+      "loss": 0.5944,
+      "step": 1655
+    },
+    {
+      "epoch": 2.6787878787878787,
+      "grad_norm": 3.421875,
+      "learning_rate": 3.436679495838835e-07,
+      "loss": 0.5962,
+      "step": 1656
+    },
+    {
+      "epoch": 2.6804040404040403,
+      "grad_norm": 3.28125,
+      "learning_rate": 3.402451416090091e-07,
+      "loss": 0.8288,
+      "step": 1657
+    },
+    {
+      "epoch": 2.682020202020202,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.3683886354813953e-07,
+      "loss": 0.7182,
+      "step": 1658
+    },
+    {
+      "epoch": 2.6836363636363636,
+      "grad_norm": 3.34375,
+      "learning_rate": 3.334491274846324e-07,
+      "loss": 0.7424,
+      "step": 1659
+    },
+    {
+      "epoch": 2.6852525252525252,
+      "grad_norm": 3.125,
+      "learning_rate": 3.300759454431657e-07,
+      "loss": 0.825,
+      "step": 1660
+    },
+    {
+      "epoch": 2.686868686868687,
+      "grad_norm": 3.234375,
+      "learning_rate": 3.2671932938968974e-07,
+      "loss": 0.6556,
+      "step": 1661
+    },
+    {
+      "epoch": 2.6884848484848485,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.233792912313943e-07,
+      "loss": 0.7843,
+      "step": 1662
+    },
+    {
+      "epoch": 2.69010101010101,
+      "grad_norm": 3.546875,
+      "learning_rate": 3.200558428166589e-07,
+      "loss": 0.7975,
+      "step": 1663
+    },
+    {
+      "epoch": 2.6917171717171717,
+      "grad_norm": 3.046875,
+      "learning_rate": 3.1674899593501175e-07,
+      "loss": 0.6052,
+      "step": 1664
+    },
+    {
+      "epoch": 2.6933333333333334,
+      "grad_norm": 3.953125,
+      "learning_rate": 3.134587623170909e-07,
+      "loss": 0.7822,
+      "step": 1665
+    },
+    {
+      "epoch": 2.694949494949495,
+      "grad_norm": 3.28125,
+      "learning_rate": 3.101851536346007e-07,
+      "loss": 0.6645,
+      "step": 1666
+    },
+    {
+      "epoch": 2.6965656565656566,
+      "grad_norm": 3.296875,
+      "learning_rate": 3.069281815002695e-07,
+      "loss": 0.5885,
+      "step": 1667
+    },
+    {
+      "epoch": 2.6981818181818182,
+      "grad_norm": 3.703125,
+      "learning_rate": 3.0368785746780925e-07,
+      "loss": 1.1617,
+      "step": 1668
+    },
+    {
+      "epoch": 2.69979797979798,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.004641930318769e-07,
+      "loss": 0.8797,
+      "step": 1669
+    },
+    {
+      "epoch": 2.7014141414141415,
+      "grad_norm": 3.234375,
+      "learning_rate": 2.9725719962802936e-07,
+      "loss": 0.8122,
+      "step": 1670
+    },
+    {
+      "epoch": 2.703030303030303,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.940668886326864e-07,
+      "loss": 0.7088,
+      "step": 1671
+    },
+    {
+      "epoch": 2.7046464646464647,
+      "grad_norm": 3.453125,
+      "learning_rate": 2.9089327136308855e-07,
+      "loss": 0.6649,
+      "step": 1672
+    },
+    {
+      "epoch": 2.7062626262626264,
+      "grad_norm": 3.03125,
+      "learning_rate": 2.8773635907725696e-07,
+      "loss": 0.6795,
+      "step": 1673
+    },
+    {
+      "epoch": 2.707878787878788,
+      "grad_norm": 3.15625,
+      "learning_rate": 2.8459616297395464e-07,
+      "loss": 0.785,
+      "step": 1674
+    },
+    {
+      "epoch": 2.7094949494949496,
+      "grad_norm": 3.015625,
+      "learning_rate": 2.8147269419264544e-07,
+      "loss": 0.8491,
+      "step": 1675
+    },
+    {
+      "epoch": 2.7111111111111112,
+      "grad_norm": 3.109375,
+      "learning_rate": 2.7836596381345613e-07,
+      "loss": 0.8183,
+      "step": 1676
+    },
+    {
+      "epoch": 2.712727272727273,
+      "grad_norm": 3.625,
+      "learning_rate": 2.7527598285713387e-07,
+      "loss": 0.9963,
+      "step": 1677
+    },
+    {
+      "epoch": 2.7143434343434345,
+      "grad_norm": 3.140625,
+      "learning_rate": 2.722027622850104e-07,
+      "loss": 1.2384,
+      "step": 1678
+    },
+    {
+      "epoch": 2.715959595959596,
+      "grad_norm": 3.125,
+      "learning_rate": 2.69146312998963e-07,
+      "loss": 0.7577,
+      "step": 1679
+    },
+    {
+      "epoch": 2.7175757575757578,
+      "grad_norm": 3.46875,
+      "learning_rate": 2.6610664584137413e-07,
+      "loss": 0.778,
+      "step": 1680
+    },
+    {
+      "epoch": 2.7191919191919194,
+      "grad_norm": 3.515625,
+      "learning_rate": 2.6308377159509233e-07,
+      "loss": 0.6091,
+      "step": 1681
+    },
+    {
+      "epoch": 2.720808080808081,
+      "grad_norm": 3.078125,
+      "learning_rate": 2.600777009833982e-07,
+      "loss": 0.6888,
+      "step": 1682
+    },
+    {
+      "epoch": 2.7224242424242426,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.570884446699612e-07,
+      "loss": 0.7954,
+      "step": 1683
+    },
+    {
+      "epoch": 2.7240404040404043,
+      "grad_norm": 3.625,
+      "learning_rate": 2.541160132588044e-07,
+      "loss": 0.6764,
+      "step": 1684
+    },
+    {
+      "epoch": 2.725656565656566,
+      "grad_norm": 3.53125,
+      "learning_rate": 2.511604172942672e-07,
+      "loss": 0.8086,
+      "step": 1685
+    },
+    {
+      "epoch": 2.7272727272727275,
+      "grad_norm": 3.078125,
+      "learning_rate": 2.482216672609677e-07,
+      "loss": 0.7137,
+      "step": 1686
+    },
+    {
+      "epoch": 2.728888888888889,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.4529977358376365e-07,
+      "loss": 0.5803,
+      "step": 1687
+    },
+    {
+      "epoch": 2.7305050505050508,
+      "grad_norm": 3.46875,
+      "learning_rate": 2.423947466277177e-07,
+      "loss": 0.7995,
+      "step": 1688
+    },
+    {
+      "epoch": 2.7321212121212124,
+      "grad_norm": 3.1875,
+      "learning_rate": 2.3950659669806033e-07,
+      "loss": 0.9997,
+      "step": 1689
+    },
+    {
+      "epoch": 2.7337373737373736,
+      "grad_norm": 3.5625,
+      "learning_rate": 2.3663533404015227e-07,
+      "loss": 0.8963,
+      "step": 1690
+    },
+    {
+      "epoch": 2.735353535353535,
+      "grad_norm": 3.1875,
+      "learning_rate": 2.3378096883944823e-07,
+      "loss": 0.6283,
+      "step": 1691
+    },
+    {
+      "epoch": 2.736969696969697,
+      "grad_norm": 3.53125,
+      "learning_rate": 2.3094351122146307e-07,
+      "loss": 0.7604,
+      "step": 1692
+    },
+    {
+      "epoch": 2.7385858585858585,
+      "grad_norm": 3.34375,
+      "learning_rate": 2.2812297125173243e-07,
+      "loss": 0.786,
+      "step": 1693
+    },
+    {
+      "epoch": 2.74020202020202,
+      "grad_norm": 3.578125,
+      "learning_rate": 2.2531935893577827e-07,
+      "loss": 0.9648,
+      "step": 1694
+    },
+    {
+      "epoch": 2.7418181818181817,
+      "grad_norm": 3.390625,
+      "learning_rate": 2.2253268421907503e-07,
+      "loss": 0.8194,
+      "step": 1695
+    },
+    {
+      "epoch": 2.7434343434343433,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.1976295698701245e-07,
+      "loss": 0.6365,
+      "step": 1696
+    },
+    {
+      "epoch": 2.745050505050505,
+      "grad_norm": 3.5,
+      "learning_rate": 2.1701018706486165e-07,
+      "loss": 0.7921,
+      "step": 1697
+    },
+    {
+      "epoch": 2.7466666666666666,
+      "grad_norm": 2.9375,
+      "learning_rate": 2.142743842177386e-07,
+      "loss": 1.0496,
+      "step": 1698
+    },
+    {
+      "epoch": 2.748282828282828,
+      "grad_norm": 3.453125,
+      "learning_rate": 2.115555581505724e-07,
+      "loss": 0.7429,
+      "step": 1699
+    },
+    {
+      "epoch": 2.74989898989899,
+      "grad_norm": 3.421875,
+      "learning_rate": 2.0885371850806691e-07,
+      "loss": 0.9672,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7515151515151515,
+      "grad_norm": 3.203125,
+      "learning_rate": 2.0616887487467096e-07,
+      "loss": 0.8057,
+      "step": 1701
+    },
+    {
+      "epoch": 2.753131313131313,
+      "grad_norm": 3.4375,
+      "learning_rate": 2.0350103677454047e-07,
+      "loss": 1.0128,
+      "step": 1702
+    },
+    {
+      "epoch": 2.7547474747474747,
+      "grad_norm": 3.28125,
+      "learning_rate": 2.0085021367150682e-07,
+      "loss": 0.5771,
+      "step": 1703
+    },
+    {
+      "epoch": 2.7563636363636363,
+      "grad_norm": 3.328125,
+      "learning_rate": 1.98216414969043e-07,
+      "loss": 0.9348,
+      "step": 1704
+    },
+    {
+      "epoch": 2.757979797979798,
+      "grad_norm": 3.3125,
+      "learning_rate": 1.9559965001022917e-07,
+      "loss": 0.9639,
+      "step": 1705
+    },
+    {
+      "epoch": 2.7595959595959596,
+      "grad_norm": 3.5,
+      "learning_rate": 1.9299992807772173e-07,
+      "loss": 0.5887,
+      "step": 1706
+    },
+    {
+      "epoch": 2.7612121212121212,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.9041725839371805e-07,
+      "loss": 0.9635,
+      "step": 1707
+    },
+    {
+      "epoch": 2.762828282828283,
+      "grad_norm": 3.0,
+      "learning_rate": 1.8785165011992513e-07,
+      "loss": 1.0032,
+      "step": 1708
+    },
+    {
+      "epoch": 2.7644444444444445,
+      "grad_norm": 3.390625,
+      "learning_rate": 1.8530311235752552e-07,
+      "loss": 0.5771,
+      "step": 1709
+    },
+    {
+      "epoch": 2.766060606060606,
+      "grad_norm": 3.359375,
+      "learning_rate": 1.8277165414714858e-07,
+      "loss": 1.0076,
+      "step": 1710
+    },
+    {
+      "epoch": 2.7676767676767677,
+      "grad_norm": 3.703125,
+      "learning_rate": 1.802572844688333e-07,
+      "loss": 1.0227,
+      "step": 1711
+    },
+    {
+      "epoch": 2.7692929292929294,
+      "grad_norm": 3.390625,
+      "learning_rate": 1.7776001224200257e-07,
+      "loss": 1.2966,
+      "step": 1712
+    },
+    {
+      "epoch": 2.770909090909091,
+      "grad_norm": 3.515625,
+      "learning_rate": 1.752798463254235e-07,
+      "loss": 0.6614,
+      "step": 1713
+    },
+    {
+      "epoch": 2.7725252525252526,
+      "grad_norm": 3.34375,
+      "learning_rate": 1.7281679551718445e-07,
+      "loss": 0.7075,
+      "step": 1714
+    },
+    {
+      "epoch": 2.7741414141414142,
+      "grad_norm": 3.0,
+      "learning_rate": 1.7037086855465902e-07,
+      "loss": 0.585,
+      "step": 1715
+    },
+    {
+      "epoch": 2.775757575757576,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.6794207411447548e-07,
+      "loss": 0.8203,
+      "step": 1716
+    },
+    {
+      "epoch": 2.7773737373737375,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.6553042081248694e-07,
+      "loss": 0.7022,
+      "step": 1717
+    },
+    {
+      "epoch": 2.778989898989899,
+      "grad_norm": 3.375,
+      "learning_rate": 1.6313591720374057e-07,
+      "loss": 0.9707,
+      "step": 1718
+    },
+    {
+      "epoch": 2.7806060606060607,
+      "grad_norm": 3.234375,
+      "learning_rate": 1.6075857178244613e-07,
+      "loss": 0.7645,
+      "step": 1719
+    },
+    {
+      "epoch": 2.7822222222222224,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.583983929819488e-07,
+      "loss": 0.6546,
+      "step": 1720
+    },
+    {
+      "epoch": 2.783838383838384,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.560553891746952e-07,
+      "loss": 0.6775,
+      "step": 1721
+    },
+    {
+      "epoch": 2.785454545454545,
+      "grad_norm": 3.421875,
+      "learning_rate": 1.5372956867220678e-07,
+      "loss": 0.9744,
+      "step": 1722
+    },
+    {
+      "epoch": 2.785454545454545,
+      "eval_loss": 0.8116081357002258,
+      "eval_runtime": 34.0673,
+      "eval_samples_per_second": 32.289,
+      "eval_steps_per_second": 4.051,
+      "step": 1722
+    },
+    {
+      "epoch": 2.787070707070707,
+      "grad_norm": 3.453125,
+      "learning_rate": 1.5142093972504933e-07,
+      "loss": 0.5182,
+      "step": 1723
+    },
+    {
+      "epoch": 2.7886868686868684,
+      "grad_norm": 3.640625,
+      "learning_rate": 1.49129510522803e-07,
+      "loss": 0.8576,
+      "step": 1724
+    },
+    {
+      "epoch": 2.79030303030303,
+      "grad_norm": 3.03125,
+      "learning_rate": 1.4685528919403446e-07,
+      "loss": 0.9184,
+      "step": 1725
+    },
+    {
+      "epoch": 2.7919191919191917,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.445982838062676e-07,
+      "loss": 0.75,
+      "step": 1726
+    },
+    {
+      "epoch": 2.7935353535353533,
+      "grad_norm": 3.59375,
+      "learning_rate": 1.423585023659546e-07,
+      "loss": 0.8554,
+      "step": 1727
+    },
+    {
+      "epoch": 2.795151515151515,
+      "grad_norm": 3.140625,
+      "learning_rate": 1.4013595281844872e-07,
+      "loss": 0.4573,
+      "step": 1728
+    },
+    {
+      "epoch": 2.7967676767676766,
+      "grad_norm": 3.8125,
+      "learning_rate": 1.3793064304797266e-07,
+      "loss": 0.589,
+      "step": 1729
+    },
+    {
+      "epoch": 2.798383838383838,
+      "grad_norm": 3.15625,
+      "learning_rate": 1.357425808775964e-07,
+      "loss": 0.5394,
+      "step": 1730
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 3.1875,
+      "learning_rate": 1.335717740692033e-07,
+      "loss": 0.7066,
+      "step": 1731
+    },
+    {
+      "epoch": 2.8016161616161614,
+      "grad_norm": 3.203125,
+      "learning_rate": 1.3141823032346736e-07,
+      "loss": 0.6943,
+      "step": 1732
+    },
+    {
+      "epoch": 2.803232323232323,
+      "grad_norm": 3.296875,
+      "learning_rate": 1.2928195727982317e-07,
+      "loss": 0.6387,
+      "step": 1733
+    },
+    {
+      "epoch": 2.8048484848484847,
+      "grad_norm": 3.421875,
+      "learning_rate": 1.2716296251644e-07,
+      "loss": 1.024,
+      "step": 1734
+    },
+    {
+      "epoch": 2.8064646464646463,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.2506125355019327e-07,
+      "loss": 0.6218,
+      "step": 1735
+    },
+    {
+      "epoch": 2.808080808080808,
+      "grad_norm": 3.625,
+      "learning_rate": 1.2297683783664138e-07,
+      "loss": 0.7948,
+      "step": 1736
+    },
+    {
+      "epoch": 2.8096969696969696,
+      "grad_norm": 3.3125,
+      "learning_rate": 1.2090972276999513e-07,
+      "loss": 0.6258,
+      "step": 1737
+    },
+    {
+      "epoch": 2.811313131313131,
+      "grad_norm": 3.515625,
+      "learning_rate": 1.1885991568309385e-07,
+      "loss": 0.7143,
+      "step": 1738
+    },
+    {
+      "epoch": 2.812929292929293,
+      "grad_norm": 3.375,
+      "learning_rate": 1.1682742384737878e-07,
+      "loss": 0.7272,
+      "step": 1739
+    },
+    {
+      "epoch": 2.8145454545454545,
+      "grad_norm": 3.265625,
+      "learning_rate": 1.1481225447286803e-07,
+      "loss": 0.7612,
+      "step": 1740
+    },
+    {
+      "epoch": 2.816161616161616,
+      "grad_norm": 3.1875,
+      "learning_rate": 1.128144147081306e-07,
+      "loss": 0.5836,
+      "step": 1741
+    },
+    {
+      "epoch": 2.8177777777777777,
+      "grad_norm": 3.453125,
+      "learning_rate": 1.1083391164025903e-07,
+      "loss": 0.7418,
+      "step": 1742
+    },
+    {
+      "epoch": 2.8193939393939393,
+      "grad_norm": 3.25,
+      "learning_rate": 1.0887075229484789e-07,
+      "loss": 0.9544,
+      "step": 1743
+    },
+    {
+      "epoch": 2.821010101010101,
+      "grad_norm": 3.375,
+      "learning_rate": 1.069249436359665e-07,
+      "loss": 0.9414,
+      "step": 1744
+    },
+    {
+      "epoch": 2.8226262626262626,
+      "grad_norm": 3.171875,
+      "learning_rate": 1.0499649256613453e-07,
+      "loss": 1.1887,
+      "step": 1745
+    },
+    {
+      "epoch": 2.824242424242424,
+      "grad_norm": 3.546875,
+      "learning_rate": 1.0308540592629756e-07,
+      "loss": 0.8466,
+      "step": 1746
+    },
+    {
+      "epoch": 2.825858585858586,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.011916904958038e-07,
+      "loss": 1.0053,
+      "step": 1747
+    },
+    {
+      "epoch": 2.8274747474747475,
+      "grad_norm": 3.296875,
+      "learning_rate": 9.931535299237737e-08,
+      "loss": 0.8013,
+      "step": 1748
+    },
+    {
+      "epoch": 2.829090909090909,
+      "grad_norm": 3.28125,
+      "learning_rate": 9.745640007209844e-08,
+      "loss": 0.8139,
+      "step": 1749
+    },
+    {
+      "epoch": 2.8307070707070707,
+      "grad_norm": 3.125,
+      "learning_rate": 9.561483832937535e-08,
+      "loss": 0.6891,
+      "step": 1750
+    },
+    {
+      "epoch": 2.8323232323232324,
+      "grad_norm": 3.28125,
+      "learning_rate": 9.379067429692635e-08,
+      "loss": 1.0838,
+      "step": 1751
+    },
+    {
+      "epoch": 2.833939393939394,
+      "grad_norm": 3.28125,
+      "learning_rate": 9.198391444575072e-08,
+      "loss": 0.8658,
+      "step": 1752
+    },
+    {
+      "epoch": 2.8355555555555556,
+      "grad_norm": 3.578125,
+      "learning_rate": 9.019456518510938e-08,
+      "loss": 0.8982,
+      "step": 1753
+    },
+    {
+      "epoch": 2.8371717171717172,
+      "grad_norm": 2.984375,
+      "learning_rate": 8.842263286250208e-08,
+      "loss": 0.7117,
+      "step": 1754
+    },
+    {
+      "epoch": 2.838787878787879,
+      "grad_norm": 3.21875,
+      "learning_rate": 8.666812376364187e-08,
+      "loss": 0.6688,
+      "step": 1755
+    },
+    {
+      "epoch": 2.8404040404040405,
+      "grad_norm": 3.21875,
+      "learning_rate": 8.493104411243791e-08,
+      "loss": 0.7352,
+      "step": 1756
+    },
+    {
+      "epoch": 2.842020202020202,
+      "grad_norm": 3.34375,
+      "learning_rate": 8.321140007096828e-08,
+      "loss": 0.6682,
+      "step": 1757
+    },
+    {
+      "epoch": 2.8436363636363637,
+      "grad_norm": 3.6875,
+      "learning_rate": 8.150919773946165e-08,
+      "loss": 0.7482,
+      "step": 1758
+    },
+    {
+      "epoch": 2.8452525252525254,
+      "grad_norm": 3.234375,
+      "learning_rate": 7.98244431562739e-08,
+      "loss": 0.8074,
+      "step": 1759
+    },
+    {
+      "epoch": 2.846868686868687,
+      "grad_norm": 3.171875,
+      "learning_rate": 7.81571422978672e-08,
+      "loss": 0.9279,
+      "step": 1760
+    },
+    {
+      "epoch": 2.8484848484848486,
+      "grad_norm": 3.0,
+      "learning_rate": 7.650730107878812e-08,
+      "loss": 0.7131,
+      "step": 1761
+    },
+    {
+      "epoch": 2.8501010101010102,
+      "grad_norm": 3.453125,
+      "learning_rate": 7.487492535164842e-08,
+      "loss": 0.7527,
+      "step": 1762
+    },
+    {
+      "epoch": 2.851717171717172,
+      "grad_norm": 3.375,
+      "learning_rate": 7.326002090710327e-08,
+      "loss": 0.9338,
+      "step": 1763
+    },
+    {
+      "epoch": 2.8533333333333335,
+      "grad_norm": 3.34375,
+      "learning_rate": 7.166259347382854e-08,
+      "loss": 0.9222,
+      "step": 1764
+    },
+    {
+      "epoch": 2.854949494949495,
+      "grad_norm": 3.203125,
+      "learning_rate": 7.008264871850411e-08,
+      "loss": 0.8662,
+      "step": 1765
+    },
+    {
+      "epoch": 2.8565656565656568,
+      "grad_norm": 3.234375,
+      "learning_rate": 6.852019224579287e-08,
+      "loss": 0.71,
+      "step": 1766
+    },
+    {
+      "epoch": 2.8581818181818184,
+      "grad_norm": 3.296875,
+      "learning_rate": 6.697522959831837e-08,
+      "loss": 0.5704,
+      "step": 1767
+    },
+    {
+      "epoch": 2.85979797979798,
+      "grad_norm": 3.5625,
+      "learning_rate": 6.544776625664829e-08,
+      "loss": 0.6104,
+      "step": 1768
+    },
+    {
+      "epoch": 2.8614141414141416,
+      "grad_norm": 3.34375,
+      "learning_rate": 6.393780763927271e-08,
+      "loss": 0.8029,
+      "step": 1769
+    },
+    {
+      "epoch": 2.8630303030303033,
+      "grad_norm": 3.03125,
+      "learning_rate": 6.244535910258697e-08,
+      "loss": 0.9581,
+      "step": 1770
+    },
+    {
+      "epoch": 2.864646464646465,
+      "grad_norm": 3.3125,
+      "learning_rate": 6.097042594087054e-08,
+      "loss": 0.5292,
+      "step": 1771
+    },
+    {
+      "epoch": 2.8662626262626265,
+      "grad_norm": 3.546875,
+      "learning_rate": 5.95130133862698e-08,
+      "loss": 0.8581,
+      "step": 1772
+    },
+    {
+      "epoch": 2.867878787878788,
+      "grad_norm": 3.09375,
+      "learning_rate": 5.8073126608778064e-08,
+      "loss": 0.4921,
+      "step": 1773
+    },
+    {
+      "epoch": 2.8694949494949498,
+      "grad_norm": 3.734375,
+      "learning_rate": 5.665077071621894e-08,
+      "loss": 0.8834,
+      "step": 1774
+    },
+    {
+      "epoch": 2.871111111111111,
+      "grad_norm": 3.265625,
+      "learning_rate": 5.524595075422745e-08,
+      "loss": 1.0098,
+      "step": 1775
+    },
+    {
+      "epoch": 2.8727272727272726,
+      "grad_norm": 3.484375,
+      "learning_rate": 5.3858671706230605e-08,
+      "loss": 0.8919,
+      "step": 1776
+    },
+    {
+      "epoch": 2.874343434343434,
+      "grad_norm": 3.453125,
+      "learning_rate": 5.2488938493433506e-08,
+      "loss": 0.7971,
+      "step": 1777
+    },
+    {
+      "epoch": 2.875959595959596,
+      "grad_norm": 3.515625,
+      "learning_rate": 5.1136755974797724e-08,
+      "loss": 0.7461,
+      "step": 1778
+    },
+    {
+      "epoch": 2.8775757575757575,
+      "grad_norm": 3.359375,
+      "learning_rate": 4.98021289470263e-08,
+      "loss": 0.7591,
+      "step": 1779
+    },
+    {
+      "epoch": 2.879191919191919,
+      "grad_norm": 3.609375,
+      "learning_rate": 4.848506214454651e-08,
+      "loss": 1.0085,
+      "step": 1780
+    },
+    {
+      "epoch": 2.8808080808080807,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.7185560239492724e-08,
+      "loss": 0.6251,
+      "step": 1781
+    },
+    {
+      "epoch": 2.8824242424242423,
+      "grad_norm": 3.28125,
+      "learning_rate": 4.590362784169022e-08,
+      "loss": 0.7066,
+      "step": 1782
+    },
+    {
+      "epoch": 2.884040404040404,
+      "grad_norm": 3.328125,
+      "learning_rate": 4.4639269498637506e-08,
+      "loss": 0.5631,
+      "step": 1783
+    },
+    {
+      "epoch": 2.8856565656565656,
+      "grad_norm": 2.890625,
+      "learning_rate": 4.3392489695493475e-08,
+      "loss": 0.6411,
+      "step": 1784
+    },
+    {
+      "epoch": 2.887272727272727,
+      "grad_norm": 3.25,
+      "learning_rate": 4.2163292855056936e-08,
+      "loss": 0.9655,
+      "step": 1785
+    },
+    {
+      "epoch": 2.888888888888889,
+      "grad_norm": 3.640625,
+      "learning_rate": 4.0951683337754345e-08,
+      "loss": 0.8352,
+      "step": 1786
+    },
+    {
+      "epoch": 2.8905050505050505,
+      "grad_norm": 3.359375,
+      "learning_rate": 3.9757665441623735e-08,
+      "loss": 0.7442,
+      "step": 1787
+    },
+    {
+      "epoch": 2.892121212121212,
+      "grad_norm": 3.015625,
+      "learning_rate": 3.858124340229863e-08,
+      "loss": 1.1369,
+      "step": 1788
+    },
+    {
+      "epoch": 2.8937373737373737,
+      "grad_norm": 3.140625,
+      "learning_rate": 3.742242139299246e-08,
+      "loss": 0.7454,
+      "step": 1789
+    },
+    {
+      "epoch": 2.8953535353535353,
+      "grad_norm": 3.375,
+      "learning_rate": 3.628120352448583e-08,
+      "loss": 0.5551,
+      "step": 1790
+    },
+    {
+      "epoch": 2.896969696969697,
+      "grad_norm": 3.5,
+      "learning_rate": 3.515759384510986e-08,
+      "loss": 1.0619,
+      "step": 1791
+    },
+    {
+      "epoch": 2.8985858585858586,
+      "grad_norm": 3.484375,
+      "learning_rate": 3.405159634073452e-08,
+      "loss": 1.0028,
+      "step": 1792
+    },
+    {
+      "epoch": 2.9002020202020202,
+      "grad_norm": 3.34375,
+      "learning_rate": 3.29632149347503e-08,
+      "loss": 0.7127,
+      "step": 1793
+    },
+    {
+      "epoch": 2.901818181818182,
+      "grad_norm": 2.90625,
+      "learning_rate": 3.1892453488058803e-08,
+      "loss": 0.7959,
+      "step": 1794
+    },
+    {
+      "epoch": 2.9034343434343435,
+      "grad_norm": 3.25,
+      "learning_rate": 3.0839315799055526e-08,
+      "loss": 0.6159,
+      "step": 1795
+    },
+    {
+      "epoch": 2.905050505050505,
+      "grad_norm": 3.1875,
+      "learning_rate": 2.9803805603619283e-08,
+      "loss": 1.1245,
+      "step": 1796
+    },
+    {
+      "epoch": 2.9066666666666667,
+      "grad_norm": 5.4375,
+      "learning_rate": 2.87859265750956e-08,
+      "loss": 0.7249,
+      "step": 1797
+    },
+    {
+      "epoch": 2.9082828282828284,
+      "grad_norm": 3.25,
+      "learning_rate": 2.77856823242878e-08,
+      "loss": 0.6923,
+      "step": 1798
+    },
+    {
+      "epoch": 2.90989898989899,
+      "grad_norm": 3.109375,
+      "learning_rate": 2.6803076399439244e-08,
+      "loss": 0.7782,
+      "step": 1799
+    },
+    {
+      "epoch": 2.9115151515151516,
+      "grad_norm": 3.40625,
+      "learning_rate": 2.5838112286226123e-08,
+      "loss": 0.6833,
+      "step": 1800
+    },
+    {
+      "epoch": 2.9131313131313132,
+      "grad_norm": 3.265625,
+      "learning_rate": 2.489079340774081e-08,
+      "loss": 0.7404,
+      "step": 1801
+    },
+    {
+      "epoch": 2.914747474747475,
+      "grad_norm": 3.453125,
+      "learning_rate": 2.39611231244824e-08,
+      "loss": 0.8805,
+      "step": 1802
+    },
+    {
+      "epoch": 2.9163636363636365,
+      "grad_norm": 3.328125,
+      "learning_rate": 2.304910473434341e-08,
+      "loss": 0.7132,
+      "step": 1803
+    },
+    {
+      "epoch": 2.917979797979798,
+      "grad_norm": 3.15625,
+      "learning_rate": 2.2154741472596996e-08,
+      "loss": 0.8298,
+      "step": 1804
+    },
+    {
+      "epoch": 2.9195959595959597,
+      "grad_norm": 3.1875,
+      "learning_rate": 2.1278036511889756e-08,
+      "loss": 1.1185,
+      "step": 1805
+    },
+    {
+      "epoch": 2.9212121212121214,
+      "grad_norm": 3.75,
+      "learning_rate": 2.0418992962224495e-08,
+      "loss": 0.8491,
+      "step": 1806
+    },
+    {
+      "epoch": 2.9228282828282826,
+      "grad_norm": 3.78125,
+      "learning_rate": 1.9577613870954694e-08,
+      "loss": 0.6341,
+      "step": 1807
+    },
+    {
+      "epoch": 2.924444444444444,
+      "grad_norm": 3.3125,
+      "learning_rate": 1.8753902222770627e-08,
+      "loss": 0.865,
+      "step": 1808
+    },
+    {
+      "epoch": 2.926060606060606,
+      "grad_norm": 3.3125,
+      "learning_rate": 1.7947860939688255e-08,
+      "loss": 0.6369,
+      "step": 1809
+    },
+    {
+      "epoch": 2.9276767676767674,
+      "grad_norm": 3.03125,
+      "learning_rate": 1.7159492881041462e-08,
+      "loss": 0.697,
+      "step": 1810
+    },
+    {
+      "epoch": 2.929292929292929,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.6388800843470387e-08,
+      "loss": 0.5284,
+      "step": 1811
+    },
+    {
+      "epoch": 2.9309090909090907,
+      "grad_norm": 3.53125,
+      "learning_rate": 1.563578756091144e-08,
+      "loss": 0.767,
+      "step": 1812
+    },
+    {
+      "epoch": 2.9325252525252523,
+      "grad_norm": 3.3125,
+      "learning_rate": 1.4900455704587869e-08,
+      "loss": 1.037,
+      "step": 1813
+    },
+    {
+      "epoch": 2.934141414141414,
+      "grad_norm": 3.40625,
+      "learning_rate": 1.4182807882999194e-08,
+      "loss": 1.0857,
+      "step": 1814
+    },
+    {
+      "epoch": 2.9357575757575756,
+      "grad_norm": 3.59375,
+      "learning_rate": 1.3482846641914572e-08,
+      "loss": 0.763,
+      "step": 1815
+    },
+    {
+      "epoch": 2.937373737373737,
+      "grad_norm": 2.921875,
+      "learning_rate": 1.2800574464361115e-08,
+      "loss": 0.7171,
+      "step": 1816
+    },
+    {
+      "epoch": 2.938989898989899,
+      "grad_norm": 3.109375,
+      "learning_rate": 1.2135993770616693e-08,
+      "loss": 0.531,
+      "step": 1817
+    },
+    {
+      "epoch": 2.9406060606060604,
+      "grad_norm": 3.65625,
+      "learning_rate": 1.1489106918200487e-08,
+      "loss": 0.9201,
+      "step": 1818
+    },
+    {
+      "epoch": 2.942222222222222,
+      "grad_norm": 3.171875,
+      "learning_rate": 1.0859916201864663e-08,
+      "loss": 1.016,
+      "step": 1819
+    },
+    {
+      "epoch": 2.9438383838383837,
+      "grad_norm": 3.0625,
+      "learning_rate": 1.0248423853587154e-08,
+      "loss": 0.9085,
+      "step": 1820
+    },
+    {
+      "epoch": 2.9454545454545453,
+      "grad_norm": 3.453125,
+      "learning_rate": 9.654632042562229e-09,
+      "loss": 0.4909,
+      "step": 1821
+    },
+    {
+      "epoch": 2.947070707070707,
+      "grad_norm": 3.5625,
+      "learning_rate": 9.07854287519494e-09,
+      "loss": 0.6964,
+      "step": 1822
+    },
+    {
+      "epoch": 2.9486868686868686,
+      "grad_norm": 3.65625,
+      "learning_rate": 8.520158395091127e-09,
+      "loss": 0.555,
+      "step": 1823
+    },
+    {
+      "epoch": 2.95030303030303,
+      "grad_norm": 3.109375,
+      "learning_rate": 7.979480583052423e-09,
+      "loss": 0.6408,
+      "step": 1824
+    },
+    {
+      "epoch": 2.951919191919192,
+      "grad_norm": 3.265625,
+      "learning_rate": 7.45651135706682e-09,
+      "loss": 0.7698,
+      "step": 1825
+    },
+    {
+      "epoch": 2.9535353535353535,
+      "grad_norm": 3.671875,
+      "learning_rate": 6.951252572304224e-09,
+      "loss": 0.8088,
+      "step": 1826
+    },
+    {
+      "epoch": 2.955151515151515,
+      "grad_norm": 3.546875,
+      "learning_rate": 6.4637060211092395e-09,
+      "loss": 0.9291,
+      "step": 1827
+    },
+    {
+      "epoch": 2.9567676767676767,
+      "grad_norm": 3.046875,
+      "learning_rate": 5.993873432993957e-09,
+      "loss": 1.0298,
+      "step": 1828
+    },
+    {
+      "epoch": 2.9583838383838383,
+      "grad_norm": 3.34375,
+      "learning_rate": 5.541756474631843e-09,
+      "loss": 0.8321,
+      "step": 1829
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 3.125,
+      "learning_rate": 5.107356749853298e-09,
+      "loss": 0.7262,
+      "step": 1830
+    },
+    {
+      "epoch": 2.9616161616161616,
+      "grad_norm": 3.375,
+      "learning_rate": 4.690675799638445e-09,
+      "loss": 0.9449,
+      "step": 1831
+    },
+    {
+      "epoch": 2.963232323232323,
+      "grad_norm": 3.25,
+      "learning_rate": 4.291715102112126e-09,
+      "loss": 0.9254,
+      "step": 1832
+    },
+    {
+      "epoch": 2.964848484848485,
+      "grad_norm": 3.578125,
+      "learning_rate": 3.910476072539471e-09,
+      "loss": 0.7683,
+      "step": 1833
+    },
+    {
+      "epoch": 2.9664646464646465,
+      "grad_norm": 3.734375,
+      "learning_rate": 3.546960063319227e-09,
+      "loss": 0.7185,
+      "step": 1834
+    },
+    {
+      "epoch": 2.968080808080808,
+      "grad_norm": 3.34375,
+      "learning_rate": 3.2011683639798783e-09,
+      "loss": 0.6909,
+      "step": 1835
+    },
+    {
+      "epoch": 2.9696969696969697,
+      "grad_norm": 3.203125,
+      "learning_rate": 2.8731022011757593e-09,
+      "loss": 0.968,
+      "step": 1836
+    },
+    {
+      "epoch": 2.9713131313131314,
+      "grad_norm": 3.125,
+      "learning_rate": 2.562762738682056e-09,
+      "loss": 0.6775,
+      "step": 1837
+    },
+    {
+      "epoch": 2.972929292929293,
+      "grad_norm": 3.46875,
+      "learning_rate": 2.27015107739037e-09,
+      "loss": 0.9623,
+      "step": 1838
+    },
+    {
+      "epoch": 2.9745454545454546,
+      "grad_norm": 3.390625,
+      "learning_rate": 1.9952682553042722e-09,
+      "loss": 0.8031,
+      "step": 1839
+    },
+    {
+      "epoch": 2.9761616161616162,
+      "grad_norm": 2.953125,
+      "learning_rate": 1.7381152475376416e-09,
+      "loss": 0.5292,
+      "step": 1840
+    },
+    {
+      "epoch": 2.977777777777778,
+      "grad_norm": 3.046875,
+      "learning_rate": 1.4986929663091122e-09,
+      "loss": 0.6783,
+      "step": 1841
+    },
+    {
+      "epoch": 2.9793939393939395,
+      "grad_norm": 3.578125,
+      "learning_rate": 1.2770022609409628e-09,
+      "loss": 0.7034,
+      "step": 1842
+    },
+    {
+      "epoch": 2.981010101010101,
+      "grad_norm": 3.28125,
+      "learning_rate": 1.0730439178530117e-09,
+      "loss": 0.8043,
+      "step": 1843
+    },
+    {
+      "epoch": 2.9826262626262627,
+      "grad_norm": 3.375,
+      "learning_rate": 8.868186605631712e-10,
+      "loss": 0.722,
+      "step": 1844
+    },
+    {
+      "epoch": 2.9842424242424244,
+      "grad_norm": 3.171875,
+      "learning_rate": 7.18327149683562e-10,
+      "loss": 1.0399,
+      "step": 1845
+    },
+    {
+      "epoch": 2.9842424242424244,
+      "eval_loss": 0.8116065859794617,
+      "eval_runtime": 34.1224,
+      "eval_samples_per_second": 32.237,
+      "eval_steps_per_second": 4.044,
+      "step": 1845
+    },
+    {
+      "epoch": 2.985858585858586,
+      "grad_norm": 3.375,
+      "learning_rate": 5.675699829160719e-10,
+      "loss": 0.8431,
+      "step": 1846
+    },
+    {
+      "epoch": 2.9874747474747476,
+      "grad_norm": 3.5,
+      "learning_rate": 4.345476950540217e-10,
+      "loss": 0.7127,
+      "step": 1847
+    },
+    {
+      "epoch": 2.9890909090909092,
+      "grad_norm": 3.109375,
+      "learning_rate": 3.1926075797827914e-10,
+      "loss": 1.0254,
+      "step": 1848
+    },
+    {
+      "epoch": 2.990707070707071,
+      "grad_norm": 3.453125,
+      "learning_rate": 2.2170958065448334e-10,
+      "loss": 0.8006,
+      "step": 1849
+    },
+    {
+      "epoch": 2.9923232323232325,
+      "grad_norm": 3.953125,
+      "learning_rate": 1.4189450913415505e-10,
+      "loss": 1.2,
+      "step": 1850
+    },
+    {
+      "epoch": 2.993939393939394,
+      "grad_norm": 3.203125,
+      "learning_rate": 7.9815826551366e-11,
+      "loss": 0.7911,
+      "step": 1851
+    },
+    {
+      "epoch": 2.9955555555555557,
+      "grad_norm": 3.25,
+      "learning_rate": 3.547375312218382e-11,
+      "loss": 0.8875,
+      "step": 1852
+    },
+    {
+      "epoch": 2.9971717171717174,
+      "grad_norm": 3.296875,
+      "learning_rate": 8.868446145227083e-12,
+      "loss": 0.7878,
+      "step": 1853
+    },
+    {
+      "epoch": 2.998787878787879,
+      "grad_norm": 3.546875,
+      "learning_rate": 0.0,
+      "loss": 0.8058,
+      "step": 1854
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1854,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.9570062090438246e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}