diff --git "a/checkpoint-574/trainer_state.json" "b/checkpoint-574/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-574/trainer_state.json"
@@ -0,0 +1,4116 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.999129677980853,
+  "eval_steps": 72,
+  "global_step": 574,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0,
+      "eval_loss": 1.136976718902588,
+      "eval_runtime": 64.1168,
+      "eval_samples_per_second": 73.117,
+      "eval_steps_per_second": 9.14,
+      "step": 0
+    },
+    {
+      "epoch": 0.0017406440382941688,
+      "grad_norm": 1.859375,
+      "learning_rate": 0.0,
+      "loss": 1.1426,
+      "step": 1
+    },
+    {
+      "epoch": 0.0034812880765883376,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.7543859649122808e-07,
+      "loss": 1.1274,
+      "step": 2
+    },
+    {
+      "epoch": 0.005221932114882507,
+      "grad_norm": 1.828125,
+      "learning_rate": 3.5087719298245616e-07,
+      "loss": 1.1205,
+      "step": 3
+    },
+    {
+      "epoch": 0.006962576153176675,
+      "grad_norm": 1.8515625,
+      "learning_rate": 5.263157894736843e-07,
+      "loss": 1.1383,
+      "step": 4
+    },
+    {
+      "epoch": 0.008703220191470844,
+      "grad_norm": 1.796875,
+      "learning_rate": 7.017543859649123e-07,
+      "loss": 1.14,
+      "step": 5
+    },
+    {
+      "epoch": 0.010443864229765013,
+      "grad_norm": 1.734375,
+      "learning_rate": 8.771929824561404e-07,
+      "loss": 1.1031,
+      "step": 6
+    },
+    {
+      "epoch": 0.012184508268059183,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 1.1794,
+      "step": 7
+    },
+    {
+      "epoch": 0.01392515230635335,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.2280701754385965e-06,
+      "loss": 1.1766,
+      "step": 8
+    },
+    {
+      "epoch": 0.015665796344647518,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.4035087719298246e-06,
+      "loss": 1.1251,
+      "step": 9
+    },
+    {
+      "epoch": 0.017406440382941687,
+      "grad_norm": 1.6796875,
+      "learning_rate": 1.5789473684210526e-06,
+      "loss": 1.1622,
+      "step": 10
+    },
+    {
+      "epoch": 0.019147084421235857,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7543859649122807e-06,
+      "loss": 1.0846,
+      "step": 11
+    },
+    {
+      "epoch": 0.020887728459530026,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.929824561403509e-06,
+      "loss": 1.1644,
+      "step": 12
+    },
+    {
+      "epoch": 0.022628372497824196,
+      "grad_norm": 1.7265625,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 1.1361,
+      "step": 13
+    },
+    {
+      "epoch": 0.024369016536118365,
+      "grad_norm": 1.8671875,
+      "learning_rate": 2.280701754385965e-06,
+      "loss": 1.1783,
+      "step": 14
+    },
+    {
+      "epoch": 0.02610966057441253,
+      "grad_norm": 1.828125,
+      "learning_rate": 2.456140350877193e-06,
+      "loss": 1.127,
+      "step": 15
+    },
+    {
+      "epoch": 0.0278503046127067,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.631578947368421e-06,
+      "loss": 1.1587,
+      "step": 16
+    },
+    {
+      "epoch": 0.02959094865100087,
+      "grad_norm": 1.671875,
+      "learning_rate": 2.8070175438596493e-06,
+      "loss": 1.1088,
+      "step": 17
+    },
+    {
+      "epoch": 0.031331592689295036,
+      "grad_norm": 1.59375,
+      "learning_rate": 2.9824561403508774e-06,
+      "loss": 1.174,
+      "step": 18
+    },
+    {
+      "epoch": 0.03307223672758921,
+      "grad_norm": 1.71875,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 1.1453,
+      "step": 19
+    },
+    {
+      "epoch": 0.034812880765883375,
+      "grad_norm": 1.8515625,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.1922,
+      "step": 20
+    },
+    {
+      "epoch": 0.03655352480417755,
+      "grad_norm": 1.5390625,
+      "learning_rate": 3.5087719298245615e-06,
+      "loss": 1.1541,
+      "step": 21
+    },
+    {
+      "epoch": 0.038294168842471714,
+      "grad_norm": 1.5390625,
+      "learning_rate": 3.6842105263157896e-06,
+      "loss": 1.0909,
+      "step": 22
+    },
+    {
+      "epoch": 0.04003481288076589,
+      "grad_norm": 1.515625,
+      "learning_rate": 3.859649122807018e-06,
+      "loss": 1.1498,
+      "step": 23
+    },
+    {
+      "epoch": 0.04177545691906005,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.035087719298246e-06,
+      "loss": 1.1096,
+      "step": 24
+    },
+    {
+      "epoch": 0.04351610095735422,
+      "grad_norm": 1.453125,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 1.1152,
+      "step": 25
+    },
+    {
+      "epoch": 0.04525674499564839,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.385964912280702e-06,
+      "loss": 1.1111,
+      "step": 26
+    },
+    {
+      "epoch": 0.04699738903394256,
+      "grad_norm": 1.296875,
+      "learning_rate": 4.56140350877193e-06,
+      "loss": 1.161,
+      "step": 27
+    },
+    {
+      "epoch": 0.04873803307223673,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.736842105263158e-06,
+      "loss": 1.1557,
+      "step": 28
+    },
+    {
+      "epoch": 0.050478677110530897,
+      "grad_norm": 1.3125,
+      "learning_rate": 4.912280701754386e-06,
+      "loss": 1.1065,
+      "step": 29
+    },
+    {
+      "epoch": 0.05221932114882506,
+      "grad_norm": 1.28125,
+      "learning_rate": 5.087719298245615e-06,
+      "loss": 1.0493,
+      "step": 30
+    },
+    {
+      "epoch": 0.053959965187119235,
+      "grad_norm": 1.1953125,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 1.0652,
+      "step": 31
+    },
+    {
+      "epoch": 0.0557006092254134,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.438596491228071e-06,
+      "loss": 1.0389,
+      "step": 32
+    },
+    {
+      "epoch": 0.057441253263707574,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.6140350877192985e-06,
+      "loss": 1.0349,
+      "step": 33
+    },
+    {
+      "epoch": 0.05918189730200174,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.789473684210527e-06,
+      "loss": 1.1109,
+      "step": 34
+    },
+    {
+      "epoch": 0.060922541340295906,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.964912280701755e-06,
+      "loss": 1.0541,
+      "step": 35
+    },
+    {
+      "epoch": 0.06266318537859007,
+      "grad_norm": 0.9453125,
+      "learning_rate": 6.140350877192983e-06,
+      "loss": 1.1176,
+      "step": 36
+    },
+    {
+      "epoch": 0.06440382941688425,
+      "grad_norm": 0.9140625,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 1.0774,
+      "step": 37
+    },
+    {
+      "epoch": 0.06614447345517842,
+      "grad_norm": 0.9453125,
+      "learning_rate": 6.491228070175439e-06,
+      "loss": 1.031,
+      "step": 38
+    },
+    {
+      "epoch": 0.06788511749347259,
+      "grad_norm": 0.85546875,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.0977,
+      "step": 39
+    },
+    {
+      "epoch": 0.06962576153176675,
+      "grad_norm": 0.8203125,
+      "learning_rate": 6.842105263157896e-06,
+      "loss": 1.0871,
+      "step": 40
+    },
+    {
+      "epoch": 0.07136640557006092,
+      "grad_norm": 0.83984375,
+      "learning_rate": 7.017543859649123e-06,
+      "loss": 1.0856,
+      "step": 41
+    },
+    {
+      "epoch": 0.0731070496083551,
+      "grad_norm": 0.82421875,
+      "learning_rate": 7.192982456140352e-06,
+      "loss": 1.0189,
+      "step": 42
+    },
+    {
+      "epoch": 0.07484769364664925,
+      "grad_norm": 0.76953125,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 1.0633,
+      "step": 43
+    },
+    {
+      "epoch": 0.07658833768494343,
+      "grad_norm": 0.7578125,
+      "learning_rate": 7.5438596491228074e-06,
+      "loss": 1.0582,
+      "step": 44
+    },
+    {
+      "epoch": 0.0783289817232376,
+      "grad_norm": 0.78125,
+      "learning_rate": 7.719298245614036e-06,
+      "loss": 1.0261,
+      "step": 45
+    },
+    {
+      "epoch": 0.08006962576153177,
+      "grad_norm": 0.66796875,
+      "learning_rate": 7.894736842105265e-06,
+      "loss": 0.9962,
+      "step": 46
+    },
+    {
+      "epoch": 0.08181026979982593,
+      "grad_norm": 0.6953125,
+      "learning_rate": 8.070175438596492e-06,
+      "loss": 1.0041,
+      "step": 47
+    },
+    {
+      "epoch": 0.0835509138381201,
+      "grad_norm": 0.6640625,
+      "learning_rate": 8.24561403508772e-06,
+      "loss": 1.0953,
+      "step": 48
+    },
+    {
+      "epoch": 0.08529155787641428,
+      "grad_norm": 0.671875,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 1.0397,
+      "step": 49
+    },
+    {
+      "epoch": 0.08703220191470844,
+      "grad_norm": 0.66015625,
+      "learning_rate": 8.596491228070176e-06,
+      "loss": 1.0366,
+      "step": 50
+    },
+    {
+      "epoch": 0.08877284595300261,
+      "grad_norm": 0.65625,
+      "learning_rate": 8.771929824561405e-06,
+      "loss": 1.005,
+      "step": 51
+    },
+    {
+      "epoch": 0.09051348999129678,
+      "grad_norm": 0.62109375,
+      "learning_rate": 8.947368421052632e-06,
+      "loss": 1.1106,
+      "step": 52
+    },
+    {
+      "epoch": 0.09225413402959094,
+      "grad_norm": 0.59375,
+      "learning_rate": 9.12280701754386e-06,
+      "loss": 1.002,
+      "step": 53
+    },
+    {
+      "epoch": 0.09399477806788512,
+      "grad_norm": 0.5859375,
+      "learning_rate": 9.298245614035088e-06,
+      "loss": 1.0599,
+      "step": 54
+    },
+    {
+      "epoch": 0.09573542210617929,
+      "grad_norm": 0.59765625,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 1.0736,
+      "step": 55
+    },
+    {
+      "epoch": 0.09747606614447346,
+      "grad_norm": 0.56640625,
+      "learning_rate": 9.649122807017545e-06,
+      "loss": 1.0179,
+      "step": 56
+    },
+    {
+      "epoch": 0.09921671018276762,
+      "grad_norm": 0.58203125,
+      "learning_rate": 9.824561403508772e-06,
+      "loss": 1.0308,
+      "step": 57
+    },
+    {
+      "epoch": 0.10095735422106179,
+      "grad_norm": 0.5625,
+      "learning_rate": 1e-05,
+      "loss": 1.0135,
+      "step": 58
+    },
+    {
+      "epoch": 0.10269799825935597,
+      "grad_norm": 0.5859375,
+      "learning_rate": 9.999979270446263e-06,
+      "loss": 0.9598,
+      "step": 59
+    },
+    {
+      "epoch": 0.10443864229765012,
+      "grad_norm": 0.5390625,
+      "learning_rate": 9.999917081956933e-06,
+      "loss": 1.0041,
+      "step": 60
+    },
+    {
+      "epoch": 0.1061792863359443,
+      "grad_norm": 0.53125,
+      "learning_rate": 9.999813435047668e-06,
+      "loss": 1.0726,
+      "step": 61
+    },
+    {
+      "epoch": 0.10791993037423847,
+      "grad_norm": 0.55078125,
+      "learning_rate": 9.99966833057789e-06,
+      "loss": 0.9915,
+      "step": 62
+    },
+    {
+      "epoch": 0.10966057441253264,
+      "grad_norm": 0.55078125,
+      "learning_rate": 9.999481769750779e-06,
+      "loss": 1.0266,
+      "step": 63
+    },
+    {
+      "epoch": 0.1114012184508268,
+      "grad_norm": 0.5234375,
+      "learning_rate": 9.999253754113263e-06,
+      "loss": 1.0524,
+      "step": 64
+    },
+    {
+      "epoch": 0.11314186248912098,
+      "grad_norm": 0.53125,
+      "learning_rate": 9.998984285556008e-06,
+      "loss": 1.0254,
+      "step": 65
+    },
+    {
+      "epoch": 0.11488250652741515,
+      "grad_norm": 0.53515625,
+      "learning_rate": 9.998673366313399e-06,
+      "loss": 1.0132,
+      "step": 66
+    },
+    {
+      "epoch": 0.11662315056570931,
+      "grad_norm": 0.5234375,
+      "learning_rate": 9.998320998963523e-06,
+      "loss": 0.9604,
+      "step": 67
+    },
+    {
+      "epoch": 0.11836379460400348,
+      "grad_norm": 0.54296875,
+      "learning_rate": 9.997927186428145e-06,
+      "loss": 1.0147,
+      "step": 68
+    },
+    {
+      "epoch": 0.12010443864229765,
+      "grad_norm": 0.52734375,
+      "learning_rate": 9.997491931972694e-06,
+      "loss": 0.998,
+      "step": 69
+    },
+    {
+      "epoch": 0.12184508268059181,
+      "grad_norm": 0.51953125,
+      "learning_rate": 9.997015239206216e-06,
+      "loss": 1.034,
+      "step": 70
+    },
+    {
+      "epoch": 0.12358572671888599,
+      "grad_norm": 0.53125,
+      "learning_rate": 9.996497112081365e-06,
+      "loss": 1.0103,
+      "step": 71
+    },
+    {
+      "epoch": 0.12532637075718014,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.99593755489436e-06,
+      "loss": 1.0053,
+      "step": 72
+    },
+    {
+      "epoch": 0.12532637075718014,
+      "eval_loss": 0.9893413186073303,
+      "eval_runtime": 60.3364,
+      "eval_samples_per_second": 77.698,
+      "eval_steps_per_second": 9.712,
+      "step": 72
+    },
+    {
+      "epoch": 0.12706701479547433,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.995336572284945e-06,
+      "loss": 0.9841,
+      "step": 73
+    },
+    {
+      "epoch": 0.1288076588337685,
+      "grad_norm": 0.53125,
+      "learning_rate": 9.994694169236366e-06,
+      "loss": 0.9298,
+      "step": 74
+    },
+    {
+      "epoch": 0.13054830287206268,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.99401035107531e-06,
+      "loss": 1.0449,
+      "step": 75
+    },
+    {
+      "epoch": 0.13228894691035684,
+      "grad_norm": 0.515625,
+      "learning_rate": 9.993285123471878e-06,
+      "loss": 1.0139,
+      "step": 76
+    },
+    {
+      "epoch": 0.134029590948651,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.992518492439526e-06,
+      "loss": 1.0129,
+      "step": 77
+    },
+    {
+      "epoch": 0.13577023498694518,
+      "grad_norm": 0.54296875,
+      "learning_rate": 9.991710464335022e-06,
+      "loss": 0.9696,
+      "step": 78
+    },
+    {
+      "epoch": 0.13751087902523934,
+      "grad_norm": 0.53125,
+      "learning_rate": 9.990861045858392e-06,
+      "loss": 1.0114,
+      "step": 79
+    },
+    {
+      "epoch": 0.1392515230635335,
+      "grad_norm": 0.515625,
+      "learning_rate": 9.989970244052861e-06,
+      "loss": 0.957,
+      "step": 80
+    },
+    {
+      "epoch": 0.1409921671018277,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.9890380663048e-06,
+      "loss": 0.9603,
+      "step": 81
+    },
+    {
+      "epoch": 0.14273281114012185,
+      "grad_norm": 0.52734375,
+      "learning_rate": 9.98806452034366e-06,
+      "loss": 1.0523,
+      "step": 82
+    },
+    {
+      "epoch": 0.144473455178416,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.987049614241907e-06,
+      "loss": 0.9804,
+      "step": 83
+    },
+    {
+      "epoch": 0.1462140992167102,
+      "grad_norm": 0.54296875,
+      "learning_rate": 9.985993356414965e-06,
+      "loss": 1.0307,
+      "step": 84
+    },
+    {
+      "epoch": 0.14795474325500435,
+      "grad_norm": 0.515625,
+      "learning_rate": 9.984895755621136e-06,
+      "loss": 0.9802,
+      "step": 85
+    },
+    {
+      "epoch": 0.1496953872932985,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.983756820961528e-06,
+      "loss": 0.9547,
+      "step": 86
+    },
+    {
+      "epoch": 0.1514360313315927,
+      "grad_norm": 0.5,
+      "learning_rate": 9.982576561879984e-06,
+      "loss": 1.014,
+      "step": 87
+    },
+    {
+      "epoch": 0.15317667536988686,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.981354988163002e-06,
+      "loss": 0.9708,
+      "step": 88
+    },
+    {
+      "epoch": 0.15491731940818101,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.980092109939651e-06,
+      "loss": 0.9717,
+      "step": 89
+    },
+    {
+      "epoch": 0.1566579634464752,
+      "grad_norm": 0.484375,
+      "learning_rate": 9.978787937681496e-06,
+      "loss": 0.9862,
+      "step": 90
+    },
+    {
+      "epoch": 0.15839860748476936,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.977442482202498e-06,
+      "loss": 0.9813,
+      "step": 91
+    },
+    {
+      "epoch": 0.16013925152306355,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.976055754658935e-06,
+      "loss": 0.9551,
+      "step": 92
+    },
+    {
+      "epoch": 0.1618798955613577,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.974627766549301e-06,
+      "loss": 0.9777,
+      "step": 93
+    },
+    {
+      "epoch": 0.16362053959965187,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.973158529714224e-06,
+      "loss": 0.9648,
+      "step": 94
+    },
+    {
+      "epoch": 0.16536118363794605,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.971648056336349e-06,
+      "loss": 0.964,
+      "step": 95
+    },
+    {
+      "epoch": 0.1671018276762402,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.97009635894025e-06,
+      "loss": 0.9635,
+      "step": 96
+    },
+    {
+      "epoch": 0.16884247171453437,
+      "grad_norm": 0.515625,
+      "learning_rate": 9.968503450392332e-06,
+      "loss": 1.0115,
+      "step": 97
+    },
+    {
+      "epoch": 0.17058311575282856,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.966869343900702e-06,
+      "loss": 0.9841,
+      "step": 98
+    },
+    {
+      "epoch": 0.17232375979112272,
+      "grad_norm": 0.484375,
+      "learning_rate": 9.965194053015083e-06,
+      "loss": 1.0241,
+      "step": 99
+    },
+    {
+      "epoch": 0.17406440382941687,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.963477591626687e-06,
+      "loss": 0.968,
+      "step": 100
+    },
+    {
+      "epoch": 0.17580504786771106,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.961719973968102e-06,
+      "loss": 0.946,
+      "step": 101
+    },
+    {
+      "epoch": 0.17754569190600522,
+      "grad_norm": 0.5234375,
+      "learning_rate": 9.959921214613187e-06,
+      "loss": 0.9284,
+      "step": 102
+    },
+    {
+      "epoch": 0.17928633594429938,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.958081328476926e-06,
+      "loss": 0.958,
+      "step": 103
+    },
+    {
+      "epoch": 0.18102697998259357,
+      "grad_norm": 0.486328125,
+      "learning_rate": 9.956200330815329e-06,
+      "loss": 0.9361,
+      "step": 104
+    },
+    {
+      "epoch": 0.18276762402088773,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.954278237225296e-06,
+      "loss": 1.0116,
+      "step": 105
+    },
+    {
+      "epoch": 0.18450826805918188,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.952315063644479e-06,
+      "loss": 1.0041,
+      "step": 106
+    },
+    {
+      "epoch": 0.18624891209747607,
+      "grad_norm": 0.5,
+      "learning_rate": 9.950310826351168e-06,
+      "loss": 0.9725,
+      "step": 107
+    },
+    {
+      "epoch": 0.18798955613577023,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.948265541964136e-06,
+      "loss": 0.9763,
+      "step": 108
+    },
+    {
+      "epoch": 0.18973020017406442,
+      "grad_norm": 0.5,
+      "learning_rate": 9.946179227442521e-06,
+      "loss": 0.9814,
+      "step": 109
+    },
+    {
+      "epoch": 0.19147084421235858,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.944051900085668e-06,
+      "loss": 1.0092,
+      "step": 110
+    },
+    {
+      "epoch": 0.19321148825065274,
+      "grad_norm": 0.515625,
+      "learning_rate": 9.941883577532993e-06,
+      "loss": 0.9836,
+      "step": 111
+    },
+    {
+      "epoch": 0.19495213228894692,
+      "grad_norm": 0.49609375,
+      "learning_rate": 9.939674277763845e-06,
+      "loss": 0.9598,
+      "step": 112
+    },
+    {
+      "epoch": 0.19669277632724108,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.937424019097337e-06,
+      "loss": 0.988,
+      "step": 113
+    },
+    {
+      "epoch": 0.19843342036553524,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.935132820192218e-06,
+      "loss": 0.9512,
+      "step": 114
+    },
+    {
+      "epoch": 0.20017406440382943,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.932800700046697e-06,
+      "loss": 0.9916,
+      "step": 115
+    },
+    {
+      "epoch": 0.20191470844212359,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.9304276779983e-06,
+      "loss": 0.9397,
+      "step": 116
+    },
+    {
+      "epoch": 0.20365535248041775,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.9280137737237e-06,
+      "loss": 0.9791,
+      "step": 117
+    },
+    {
+      "epoch": 0.20539599651871193,
+      "grad_norm": 0.515625,
+      "learning_rate": 9.925559007238564e-06,
+      "loss": 0.9231,
+      "step": 118
+    },
+    {
+      "epoch": 0.2071366405570061,
+      "grad_norm": 0.5,
+      "learning_rate": 9.923063398897372e-06,
+      "loss": 0.9854,
+      "step": 119
+    },
+    {
+      "epoch": 0.20887728459530025,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.920526969393267e-06,
+      "loss": 1.0411,
+      "step": 120
+    },
+    {
+      "epoch": 0.21061792863359444,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.917949739757869e-06,
+      "loss": 1.0254,
+      "step": 121
+    },
+    {
+      "epoch": 0.2123585726718886,
+      "grad_norm": 0.5,
+      "learning_rate": 9.915331731361104e-06,
+      "loss": 0.9208,
+      "step": 122
+    },
+    {
+      "epoch": 0.21409921671018275,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.912672965911034e-06,
+      "loss": 0.9195,
+      "step": 123
+    },
+    {
+      "epoch": 0.21583986074847694,
+      "grad_norm": 0.48046875,
+      "learning_rate": 9.909973465453666e-06,
+      "loss": 0.9938,
+      "step": 124
+    },
+    {
+      "epoch": 0.2175805047867711,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.907233252372775e-06,
+      "loss": 0.904,
+      "step": 125
+    },
+    {
+      "epoch": 0.2193211488250653,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.904452349389717e-06,
+      "loss": 0.9882,
+      "step": 126
+    },
+    {
+      "epoch": 0.22106179286335945,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.901630779563247e-06,
+      "loss": 0.9688,
+      "step": 127
+    },
+    {
+      "epoch": 0.2228024369016536,
+      "grad_norm": 0.51953125,
+      "learning_rate": 9.898768566289316e-06,
+      "loss": 1.0522,
+      "step": 128
+    },
+    {
+      "epoch": 0.2245430809399478,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.895865733300887e-06,
+      "loss": 0.9551,
+      "step": 129
+    },
+    {
+      "epoch": 0.22628372497824195,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.89292230466773e-06,
+      "loss": 0.9623,
+      "step": 130
+    },
+    {
+      "epoch": 0.2280243690165361,
+      "grad_norm": 0.515625,
+      "learning_rate": 9.889938304796236e-06,
+      "loss": 0.9702,
+      "step": 131
+    },
+    {
+      "epoch": 0.2297650130548303,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.886913758429194e-06,
+      "loss": 0.9679,
+      "step": 132
+    },
+    {
+      "epoch": 0.23150565709312446,
+      "grad_norm": 0.466796875,
+      "learning_rate": 9.883848690645601e-06,
+      "loss": 0.9719,
+      "step": 133
+    },
+    {
+      "epoch": 0.23324630113141862,
+      "grad_norm": 0.4921875,
+      "learning_rate": 9.880743126860458e-06,
+      "loss": 0.9717,
+      "step": 134
+    },
+    {
+      "epoch": 0.2349869451697128,
+      "grad_norm": 0.4921875,
+      "learning_rate": 9.87759709282454e-06,
+      "loss": 0.9365,
+      "step": 135
+    },
+    {
+      "epoch": 0.23672758920800696,
+      "grad_norm": 0.478515625,
+      "learning_rate": 9.874410614624202e-06,
+      "loss": 0.9254,
+      "step": 136
+    },
+    {
+      "epoch": 0.23846823324630112,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.871183718681153e-06,
+      "loss": 1.0045,
+      "step": 137
+    },
+    {
+      "epoch": 0.2402088772845953,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.867916431752237e-06,
+      "loss": 0.9693,
+      "step": 138
+    },
+    {
+      "epoch": 0.24194952132288947,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.864608780929218e-06,
+      "loss": 0.9981,
+      "step": 139
+    },
+    {
+      "epoch": 0.24369016536118362,
+      "grad_norm": 0.4921875,
+      "learning_rate": 9.861260793638539e-06,
+      "loss": 0.9569,
+      "step": 140
+    },
+    {
+      "epoch": 0.2454308093994778,
+      "grad_norm": 0.490234375,
+      "learning_rate": 9.857872497641117e-06,
+      "loss": 1.0422,
+      "step": 141
+    },
+    {
+      "epoch": 0.24717145343777197,
+      "grad_norm": 0.49609375,
+      "learning_rate": 9.854443921032098e-06,
+      "loss": 1.0408,
+      "step": 142
+    },
+    {
+      "epoch": 0.24891209747606616,
+      "grad_norm": 0.484375,
+      "learning_rate": 9.850975092240625e-06,
+      "loss": 0.8893,
+      "step": 143
+    },
+    {
+      "epoch": 0.2506527415143603,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.84746604002961e-06,
+      "loss": 0.9679,
+      "step": 144
+    },
+    {
+      "epoch": 0.2506527415143603,
+      "eval_loss": 0.9576423168182373,
+      "eval_runtime": 59.1215,
+      "eval_samples_per_second": 79.294,
+      "eval_steps_per_second": 9.912,
+      "step": 144
+    },
+    {
+      "epoch": 0.2523933855526545,
+      "grad_norm": 0.5,
+      "learning_rate": 9.843916793495487e-06,
+      "loss": 0.9071,
+      "step": 145
+    },
+    {
+      "epoch": 0.25413402959094866,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.840327382067972e-06,
+      "loss": 0.9496,
+      "step": 146
+    },
+    {
+      "epoch": 0.2558746736292428,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.836697835509827e-06,
+      "loss": 0.9864,
+      "step": 147
+    },
+    {
+      "epoch": 0.257615317667537,
+      "grad_norm": 0.5,
+      "learning_rate": 9.833028183916601e-06,
+      "loss": 1.0082,
+      "step": 148
+    },
+    {
+      "epoch": 0.25935596170583114,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.829318457716395e-06,
+      "loss": 0.9591,
+      "step": 149
+    },
+    {
+      "epoch": 0.26109660574412535,
+      "grad_norm": 0.515625,
+      "learning_rate": 9.82556868766959e-06,
+      "loss": 1.0204,
+      "step": 150
+    },
+    {
+      "epoch": 0.2628372497824195,
+      "grad_norm": 0.4921875,
+      "learning_rate": 9.821778904868616e-06,
+      "loss": 0.9805,
+      "step": 151
+    },
+    {
+      "epoch": 0.26457789382071367,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.817949140737672e-06,
+      "loss": 0.9761,
+      "step": 152
+    },
+    {
+      "epoch": 0.26631853785900783,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.81407942703248e-06,
+      "loss": 0.9789,
+      "step": 153
+    },
+    {
+      "epoch": 0.268059181897302,
+      "grad_norm": 0.49609375,
+      "learning_rate": 9.810169795840012e-06,
+      "loss": 0.952,
+      "step": 154
+    },
+    {
+      "epoch": 0.26979982593559615,
+      "grad_norm": 0.484375,
+      "learning_rate": 9.806220279578236e-06,
+      "loss": 0.9431,
+      "step": 155
+    },
+    {
+      "epoch": 0.27154046997389036,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.802230910995833e-06,
+      "loss": 1.0015,
+      "step": 156
+    },
+    {
+      "epoch": 0.2732811140121845,
+      "grad_norm": 0.49609375,
+      "learning_rate": 9.798201723171938e-06,
+      "loss": 0.9513,
+      "step": 157
+    },
+    {
+      "epoch": 0.2750217580504787,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.794132749515854e-06,
+      "loss": 0.9454,
+      "step": 158
+    },
+    {
+      "epoch": 0.27676240208877284,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.790024023766789e-06,
+      "loss": 0.9581,
+      "step": 159
+    },
+    {
+      "epoch": 0.278503046127067,
+      "grad_norm": 0.482421875,
+      "learning_rate": 9.785875579993558e-06,
+      "loss": 0.9874,
+      "step": 160
+    },
+    {
+      "epoch": 0.28024369016536116,
+      "grad_norm": 0.49609375,
+      "learning_rate": 9.781687452594318e-06,
+      "loss": 0.9417,
+      "step": 161
+    },
+    {
+      "epoch": 0.2819843342036554,
+      "grad_norm": 0.478515625,
+      "learning_rate": 9.777459676296276e-06,
+      "loss": 0.9589,
+      "step": 162
+    },
+    {
+      "epoch": 0.28372497824194953,
+      "grad_norm": 0.515625,
+      "learning_rate": 9.773192286155395e-06,
+      "loss": 0.9851,
+      "step": 163
+    },
+    {
+      "epoch": 0.2854656222802437,
+      "grad_norm": 0.49609375,
+      "learning_rate": 9.768885317556116e-06,
+      "loss": 0.98,
+      "step": 164
+    },
+    {
+      "epoch": 0.28720626631853785,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.764538806211052e-06,
+      "loss": 0.9651,
+      "step": 165
+    },
+    {
+      "epoch": 0.288946910356832,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.760152788160697e-06,
+      "loss": 0.9407,
+      "step": 166
+    },
+    {
+      "epoch": 0.2906875543951262,
+      "grad_norm": 0.46875,
+      "learning_rate": 9.755727299773135e-06,
+      "loss": 0.9553,
+      "step": 167
+    },
+    {
+      "epoch": 0.2924281984334204,
+      "grad_norm": 0.470703125,
+      "learning_rate": 9.75126237774372e-06,
+      "loss": 0.9823,
+      "step": 168
+    },
+    {
+      "epoch": 0.29416884247171454,
+      "grad_norm": 0.5,
+      "learning_rate": 9.746758059094791e-06,
+      "loss": 0.9832,
+      "step": 169
+    },
+    {
+      "epoch": 0.2959094865100087,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.742214381175355e-06,
+      "loss": 0.8976,
+      "step": 170
+    },
+    {
+      "epoch": 0.29765013054830286,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.737631381660777e-06,
+      "loss": 0.9331,
+      "step": 171
+    },
+    {
+      "epoch": 0.299390774586597,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.733009098552473e-06,
+      "loss": 0.9666,
+      "step": 172
+    },
+    {
+      "epoch": 0.30113141862489123,
+      "grad_norm": 0.490234375,
+      "learning_rate": 9.728347570177587e-06,
+      "loss": 0.9781,
+      "step": 173
+    },
+    {
+      "epoch": 0.3028720626631854,
+      "grad_norm": 0.484375,
+      "learning_rate": 9.723646835188681e-06,
+      "loss": 0.969,
+      "step": 174
+    },
+    {
+      "epoch": 0.30461270670147955,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.71890693256341e-06,
+      "loss": 0.9481,
+      "step": 175
+    },
+    {
+      "epoch": 0.3063533507397737,
+      "grad_norm": 0.49609375,
+      "learning_rate": 9.7141279016042e-06,
+      "loss": 1.0301,
+      "step": 176
+    },
+    {
+      "epoch": 0.30809399477806787,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.709309781937925e-06,
+      "loss": 1.023,
+      "step": 177
+    },
+    {
+      "epoch": 0.30983463881636203,
+      "grad_norm": 0.490234375,
+      "learning_rate": 9.704452613515571e-06,
+      "loss": 0.9598,
+      "step": 178
+    },
+    {
+      "epoch": 0.31157528285465624,
+      "grad_norm": 0.482421875,
+      "learning_rate": 9.699556436611912e-06,
+      "loss": 0.9699,
+      "step": 179
+    },
+    {
+      "epoch": 0.3133159268929504,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.694621291825174e-06,
+      "loss": 0.903,
+      "step": 180
+    },
+    {
+      "epoch": 0.31505657093124456,
+      "grad_norm": 0.515625,
+      "learning_rate": 9.689647220076696e-06,
+      "loss": 0.9581,
+      "step": 181
+    },
+    {
+      "epoch": 0.3167972149695387,
+      "grad_norm": 0.490234375,
+      "learning_rate": 9.684634262610593e-06,
+      "loss": 0.939,
+      "step": 182
+    },
+    {
+      "epoch": 0.3185378590078329,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.679582460993413e-06,
+      "loss": 0.9363,
+      "step": 183
+    },
+    {
+      "epoch": 0.3202785030461271,
+      "grad_norm": 0.482421875,
+      "learning_rate": 9.674491857113792e-06,
+      "loss": 1.0215,
+      "step": 184
+    },
+    {
+      "epoch": 0.32201914708442125,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.669362493182112e-06,
+      "loss": 0.9464,
+      "step": 185
+    },
+    {
+      "epoch": 0.3237597911227154,
+      "grad_norm": 0.482421875,
+      "learning_rate": 9.66419441173014e-06,
+      "loss": 0.8955,
+      "step": 186
+    },
+    {
+      "epoch": 0.32550043516100957,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.658987655610687e-06,
+      "loss": 0.9503,
+      "step": 187
+    },
+    {
+      "epoch": 0.32724107919930373,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.653742267997245e-06,
+      "loss": 0.9808,
+      "step": 188
+    },
+    {
+      "epoch": 0.3289817232375979,
+      "grad_norm": 0.48046875,
+      "learning_rate": 9.648458292383631e-06,
+      "loss": 1.0155,
+      "step": 189
+    },
+    {
+      "epoch": 0.3307223672758921,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.643135772583627e-06,
+      "loss": 0.929,
+      "step": 190
+    },
+    {
+      "epoch": 0.33246301131418626,
+      "grad_norm": 0.5,
+      "learning_rate": 9.63777475273062e-06,
+      "loss": 0.9214,
+      "step": 191
+    },
+    {
+      "epoch": 0.3342036553524804,
+      "grad_norm": 0.47265625,
+      "learning_rate": 9.632375277277226e-06,
+      "loss": 0.9762,
+      "step": 192
+    },
+    {
+      "epoch": 0.3359442993907746,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.626937390994932e-06,
+      "loss": 0.9734,
+      "step": 193
+    },
+    {
+      "epoch": 0.33768494342906874,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.621461138973725e-06,
+      "loss": 0.9599,
+      "step": 194
+    },
+    {
+      "epoch": 0.3394255874673629,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.615946566621704e-06,
+      "loss": 0.9448,
+      "step": 195
+    },
+    {
+      "epoch": 0.3411662315056571,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.61039371966472e-06,
+      "loss": 0.9721,
+      "step": 196
+    },
+    {
+      "epoch": 0.3429068755439513,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.60480264414599e-06,
+      "loss": 0.97,
+      "step": 197
+    },
+    {
+      "epoch": 0.34464751958224543,
+      "grad_norm": 0.49609375,
+      "learning_rate": 9.599173386425711e-06,
+      "loss": 0.9964,
+      "step": 198
+    },
+    {
+      "epoch": 0.3463881636205396,
+      "grad_norm": 0.486328125,
+      "learning_rate": 9.593505993180687e-06,
+      "loss": 0.9715,
+      "step": 199
+    },
+    {
+      "epoch": 0.34812880765883375,
+      "grad_norm": 0.4921875,
+      "learning_rate": 9.587800511403931e-06,
+      "loss": 0.9655,
+      "step": 200
+    },
+    {
+      "epoch": 0.34986945169712796,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.582056988404276e-06,
+      "loss": 0.9606,
+      "step": 201
+    },
+    {
+      "epoch": 0.3516100957354221,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.576275471805993e-06,
+      "loss": 0.9166,
+      "step": 202
+    },
+    {
+      "epoch": 0.3533507397737163,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.570456009548383e-06,
+      "loss": 0.9063,
+      "step": 203
+    },
+    {
+      "epoch": 0.35509138381201044,
+      "grad_norm": 0.484375,
+      "learning_rate": 9.564598649885391e-06,
+      "loss": 0.9295,
+      "step": 204
+    },
+    {
+      "epoch": 0.3568320278503046,
+      "grad_norm": 0.48046875,
+      "learning_rate": 9.558703441385195e-06,
+      "loss": 0.9933,
+      "step": 205
+    },
+    {
+      "epoch": 0.35857267188859876,
+      "grad_norm": 0.48046875,
+      "learning_rate": 9.552770432929812e-06,
+      "loss": 0.9572,
+      "step": 206
+    },
+    {
+      "epoch": 0.360313315926893,
+      "grad_norm": 0.482421875,
+      "learning_rate": 9.54679967371469e-06,
+      "loss": 0.9484,
+      "step": 207
+    },
+    {
+      "epoch": 0.36205395996518713,
+      "grad_norm": 0.47265625,
+      "learning_rate": 9.540791213248299e-06,
+      "loss": 0.9266,
+      "step": 208
+    },
+    {
+      "epoch": 0.3637946040034813,
+      "grad_norm": 0.5,
+      "learning_rate": 9.534745101351719e-06,
+      "loss": 0.9526,
+      "step": 209
+    },
+    {
+      "epoch": 0.36553524804177545,
+      "grad_norm": 0.5,
+      "learning_rate": 9.528661388158234e-06,
+      "loss": 1.0046,
+      "step": 210
+    },
+    {
+      "epoch": 0.3672758920800696,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.522540124112902e-06,
+      "loss": 0.9757,
+      "step": 211
+    },
+    {
+      "epoch": 0.36901653611836377,
+      "grad_norm": 0.48046875,
+      "learning_rate": 9.516381359972157e-06,
+      "loss": 1.0195,
+      "step": 212
+    },
+    {
+      "epoch": 0.370757180156658,
+      "grad_norm": 0.490234375,
+      "learning_rate": 9.51018514680337e-06,
+      "loss": 0.971,
+      "step": 213
+    },
+    {
+      "epoch": 0.37249782419495214,
+      "grad_norm": 0.5,
+      "learning_rate": 9.503951535984434e-06,
+      "loss": 0.9336,
+      "step": 214
+    },
+    {
+      "epoch": 0.3742384682332463,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.49768057920334e-06,
+      "loss": 0.9887,
+      "step": 215
+    },
+    {
+      "epoch": 0.37597911227154046,
+      "grad_norm": 0.486328125,
+      "learning_rate": 9.491372328457737e-06,
+      "loss": 0.966,
+      "step": 216
+    },
+    {
+      "epoch": 0.37597911227154046,
+      "eval_loss": 0.9439952373504639,
+      "eval_runtime": 59.4647,
+      "eval_samples_per_second": 78.837,
+      "eval_steps_per_second": 9.855,
+      "step": 216
+    },
+    {
+      "epoch": 0.3777197563098346,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.485026836054519e-06,
+      "loss": 0.9771,
+      "step": 217
+    },
+    {
+      "epoch": 0.37946040034812883,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.478644154609372e-06,
+      "loss": 0.9988,
+      "step": 218
+    },
+    {
+      "epoch": 0.381201044386423,
+      "grad_norm": 0.470703125,
+      "learning_rate": 9.472224337046357e-06,
+      "loss": 0.9875,
+      "step": 219
+    },
+    {
+      "epoch": 0.38294168842471715,
+      "grad_norm": 0.482421875,
+      "learning_rate": 9.46576743659745e-06,
+      "loss": 0.9664,
+      "step": 220
+    },
+    {
+      "epoch": 0.3846823324630113,
+      "grad_norm": 0.478515625,
+      "learning_rate": 9.45927350680212e-06,
+      "loss": 0.9693,
+      "step": 221
+    },
+    {
+      "epoch": 0.38642297650130547,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.452742601506873e-06,
+      "loss": 0.9137,
+      "step": 222
+    },
+    {
+      "epoch": 0.38816362053959963,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.446174774864808e-06,
+      "loss": 0.9047,
+      "step": 223
+    },
+    {
+      "epoch": 0.38990426457789384,
+      "grad_norm": 0.486328125,
+      "learning_rate": 9.439570081335173e-06,
+      "loss": 0.9926,
+      "step": 224
+    },
+    {
+      "epoch": 0.391644908616188,
+      "grad_norm": 0.478515625,
+      "learning_rate": 9.432928575682908e-06,
+      "loss": 0.9693,
+      "step": 225
+    },
+    {
+      "epoch": 0.39338555265448216,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.426250312978191e-06,
+      "loss": 0.9106,
+      "step": 226
+    },
+    {
+      "epoch": 0.3951261966927763,
+      "grad_norm": 0.5,
+      "learning_rate": 9.419535348595985e-06,
+      "loss": 0.9639,
+      "step": 227
+    },
+    {
+      "epoch": 0.3968668407310705,
+      "grad_norm": 0.49609375,
+      "learning_rate": 9.412783738215576e-06,
+      "loss": 0.9338,
+      "step": 228
+    },
+    {
+      "epoch": 0.39860748476936464,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.405995537820111e-06,
+      "loss": 1.0216,
+      "step": 229
+    },
+    {
+      "epoch": 0.40034812880765885,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.399170803696139e-06,
+      "loss": 0.942,
+      "step": 230
+    },
+    {
+      "epoch": 0.402088772845953,
+      "grad_norm": 0.486328125,
+      "learning_rate": 9.392309592433134e-06,
+      "loss": 0.9184,
+      "step": 231
+    },
+    {
+      "epoch": 0.40382941688424717,
+      "grad_norm": 0.5234375,
+      "learning_rate": 9.385411960923036e-06,
+      "loss": 0.9178,
+      "step": 232
+    },
+    {
+      "epoch": 0.40557006092254133,
+      "grad_norm": 0.4921875,
+      "learning_rate": 9.378477966359773e-06,
+      "loss": 0.9303,
+      "step": 233
+    },
+    {
+      "epoch": 0.4073107049608355,
+      "grad_norm": 0.5,
+      "learning_rate": 9.371507666238793e-06,
+      "loss": 0.9563,
+      "step": 234
+    },
+    {
+      "epoch": 0.4090513489991297,
+      "grad_norm": 0.47265625,
+      "learning_rate": 9.364501118356579e-06,
+      "loss": 0.92,
+      "step": 235
+    },
+    {
+      "epoch": 0.41079199303742386,
+      "grad_norm": 0.4765625,
+      "learning_rate": 9.357458380810175e-06,
+      "loss": 0.9532,
+      "step": 236
+    },
+    {
+      "epoch": 0.412532637075718,
+      "grad_norm": 0.482421875,
+      "learning_rate": 9.350379511996706e-06,
+      "loss": 0.9604,
+      "step": 237
+    },
+    {
+      "epoch": 0.4142732811140122,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.343264570612883e-06,
+      "loss": 0.9415,
+      "step": 238
+    },
+    {
+      "epoch": 0.41601392515230634,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.336113615654535e-06,
+      "loss": 0.9752,
+      "step": 239
+    },
+    {
+      "epoch": 0.4177545691906005,
+      "grad_norm": 0.484375,
+      "learning_rate": 9.328926706416102e-06,
+      "loss": 0.9517,
+      "step": 240
+    },
+    {
+      "epoch": 0.4194952132288947,
+      "grad_norm": 0.484375,
+      "learning_rate": 9.321703902490152e-06,
+      "loss": 0.9245,
+      "step": 241
+    },
+    {
+      "epoch": 0.4212358572671889,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.314445263766888e-06,
+      "loss": 0.9341,
+      "step": 242
+    },
+    {
+      "epoch": 0.42297650130548303,
+      "grad_norm": 0.47265625,
+      "learning_rate": 9.307150850433643e-06,
+      "loss": 0.9399,
+      "step": 243
+    },
+    {
+      "epoch": 0.4247171453437772,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.299820722974396e-06,
+      "loss": 0.9865,
+      "step": 244
+    },
+    {
+      "epoch": 0.42645778938207135,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.29245494216925e-06,
+      "loss": 0.9538,
+      "step": 245
+    },
+    {
+      "epoch": 0.4281984334203655,
+      "grad_norm": 0.5,
+      "learning_rate": 9.285053569093948e-06,
+      "loss": 1.0095,
+      "step": 246
+    },
+    {
+      "epoch": 0.4299390774586597,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.277616665119352e-06,
+      "loss": 0.9691,
+      "step": 247
+    },
+    {
+      "epoch": 0.4316797214969539,
+      "grad_norm": 0.4921875,
+      "learning_rate": 9.27014429191094e-06,
+      "loss": 0.9854,
+      "step": 248
+    },
+    {
+      "epoch": 0.43342036553524804,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.262636511428304e-06,
+      "loss": 0.9179,
+      "step": 249
+    },
+    {
+      "epoch": 0.4351610095735422,
+      "grad_norm": 0.490234375,
+      "learning_rate": 9.255093385924616e-06,
+      "loss": 0.9388,
+      "step": 250
+    },
+    {
+      "epoch": 0.43690165361183636,
+      "grad_norm": 0.51953125,
+      "learning_rate": 9.247514977946124e-06,
+      "loss": 0.9788,
+      "step": 251
+    },
+    {
+      "epoch": 0.4386422976501306,
+      "grad_norm": 0.515625,
+      "learning_rate": 9.239901350331635e-06,
+      "loss": 0.9301,
+      "step": 252
+    },
+    {
+      "epoch": 0.44038294168842473,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.232252566211993e-06,
+      "loss": 0.9656,
+      "step": 253
+    },
+    {
+      "epoch": 0.4421235857267189,
+      "grad_norm": 0.4921875,
+      "learning_rate": 9.224568689009548e-06,
+      "loss": 1.0119,
+      "step": 254
+    },
+    {
+      "epoch": 0.44386422976501305,
+      "grad_norm": 0.49609375,
+      "learning_rate": 9.216849782437637e-06,
+      "loss": 0.874,
+      "step": 255
+    },
+    {
+      "epoch": 0.4456048738033072,
+      "grad_norm": 0.5,
+      "learning_rate": 9.20909591050006e-06,
+      "loss": 0.9191,
+      "step": 256
+    },
+    {
+      "epoch": 0.44734551784160137,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.201307137490536e-06,
+      "loss": 0.9017,
+      "step": 257
+    },
+    {
+      "epoch": 0.4490861618798956,
+      "grad_norm": 0.4921875,
+      "learning_rate": 9.19348352799218e-06,
+      "loss": 0.9363,
+      "step": 258
+    },
+    {
+      "epoch": 0.45082680591818974,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.185625146876966e-06,
+      "loss": 0.9921,
+      "step": 259
+    },
+    {
+      "epoch": 0.4525674499564839,
+      "grad_norm": 0.482421875,
+      "learning_rate": 9.177732059305187e-06,
+      "loss": 0.9358,
+      "step": 260
+    },
+    {
+      "epoch": 0.45430809399477806,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.169804330724916e-06,
+      "loss": 0.9257,
+      "step": 261
+    },
+    {
+      "epoch": 0.4560487380330722,
+      "grad_norm": 0.49609375,
+      "learning_rate": 9.161842026871465e-06,
+      "loss": 0.9201,
+      "step": 262
+    },
+    {
+      "epoch": 0.4577893820713664,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.153845213766837e-06,
+      "loss": 0.9212,
+      "step": 263
+    },
+    {
+      "epoch": 0.4595300261096606,
+      "grad_norm": 0.5,
+      "learning_rate": 9.145813957719174e-06,
+      "loss": 0.9735,
+      "step": 264
+    },
+    {
+      "epoch": 0.46127067014795475,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.137748325322223e-06,
+      "loss": 0.9585,
+      "step": 265
+    },
+    {
+      "epoch": 0.4630113141862489,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.129648383454764e-06,
+      "loss": 0.9781,
+      "step": 266
+    },
+    {
+      "epoch": 0.46475195822454307,
+      "grad_norm": 0.486328125,
+      "learning_rate": 9.121514199280072e-06,
+      "loss": 0.9759,
+      "step": 267
+    },
+    {
+      "epoch": 0.46649260226283723,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.113345840245348e-06,
+      "loss": 0.9688,
+      "step": 268
+    },
+    {
+      "epoch": 0.46823324630113144,
+      "grad_norm": 0.49609375,
+      "learning_rate": 9.105143374081167e-06,
+      "loss": 0.9092,
+      "step": 269
+    },
+    {
+      "epoch": 0.4699738903394256,
+      "grad_norm": 0.48046875,
+      "learning_rate": 9.096906868800917e-06,
+      "loss": 0.9357,
+      "step": 270
+    },
+    {
+      "epoch": 0.47171453437771976,
+      "grad_norm": 0.4921875,
+      "learning_rate": 9.088636392700227e-06,
+      "loss": 1.0134,
+      "step": 271
+    },
+    {
+      "epoch": 0.4734551784160139,
+      "grad_norm": 0.49609375,
+      "learning_rate": 9.08033201435641e-06,
+      "loss": 0.9494,
+      "step": 272
+    },
+    {
+      "epoch": 0.4751958224543081,
+      "grad_norm": 0.48046875,
+      "learning_rate": 9.071993802627887e-06,
+      "loss": 0.9446,
+      "step": 273
+    },
+    {
+      "epoch": 0.47693646649260224,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.063621826653624e-06,
+      "loss": 0.8926,
+      "step": 274
+    },
+    {
+      "epoch": 0.47867711053089645,
+      "grad_norm": 0.484375,
+      "learning_rate": 9.055216155852548e-06,
+      "loss": 0.9216,
+      "step": 275
+    },
+    {
+      "epoch": 0.4804177545691906,
+      "grad_norm": 0.470703125,
+      "learning_rate": 9.046776859922983e-06,
+      "loss": 0.9442,
+      "step": 276
+    },
+    {
+      "epoch": 0.4821583986074848,
+      "grad_norm": 0.484375,
+      "learning_rate": 9.038304008842064e-06,
+      "loss": 0.9456,
+      "step": 277
+    },
+    {
+      "epoch": 0.48389904264577893,
+      "grad_norm": 0.482421875,
+      "learning_rate": 9.02979767286516e-06,
+      "loss": 0.909,
+      "step": 278
+    },
+    {
+      "epoch": 0.4856396866840731,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.021257922525289e-06,
+      "loss": 0.9597,
+      "step": 279
+    },
+    {
+      "epoch": 0.48738033072236725,
+      "grad_norm": 0.5,
+      "learning_rate": 9.012684828632538e-06,
+      "loss": 0.9646,
+      "step": 280
+    },
+    {
+      "epoch": 0.48912097476066146,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.004078462273471e-06,
+      "loss": 0.9679,
+      "step": 281
+    },
+    {
+      "epoch": 0.4908616187989556,
+      "grad_norm": 0.5078125,
+      "learning_rate": 8.995438894810541e-06,
+      "loss": 0.952,
+      "step": 282
+    },
+    {
+      "epoch": 0.4926022628372498,
+      "grad_norm": 0.484375,
+      "learning_rate": 8.9867661978815e-06,
+      "loss": 0.9792,
+      "step": 283
+    },
+    {
+      "epoch": 0.49434290687554394,
+      "grad_norm": 0.5,
+      "learning_rate": 8.978060443398802e-06,
+      "loss": 0.939,
+      "step": 284
+    },
+    {
+      "epoch": 0.4960835509138381,
+      "grad_norm": 0.486328125,
+      "learning_rate": 8.96932170354901e-06,
+      "loss": 0.9974,
+      "step": 285
+    },
+    {
+      "epoch": 0.4978241949521323,
+      "grad_norm": 0.48828125,
+      "learning_rate": 8.960550050792194e-06,
+      "loss": 0.9947,
+      "step": 286
+    },
+    {
+      "epoch": 0.4995648389904265,
+      "grad_norm": 0.50390625,
+      "learning_rate": 8.951745557861333e-06,
+      "loss": 0.9139,
+      "step": 287
+    },
+    {
+      "epoch": 0.5013054830287206,
+      "grad_norm": 0.49609375,
+      "learning_rate": 8.942908297761712e-06,
+      "loss": 0.9397,
+      "step": 288
+    },
+    {
+      "epoch": 0.5013054830287206,
+      "eval_loss": 0.9357889294624329,
+      "eval_runtime": 60.8471,
+      "eval_samples_per_second": 77.046,
+      "eval_steps_per_second": 9.631,
+      "step": 288
+    },
+    {
+      "epoch": 0.5030461270670148,
+      "grad_norm": 0.482421875,
+      "learning_rate": 8.934038343770312e-06,
+      "loss": 0.9656,
+      "step": 289
+    },
+    {
+      "epoch": 0.504786771105309,
+      "grad_norm": 0.5078125,
+      "learning_rate": 8.925135769435211e-06,
+      "loss": 0.9896,
+      "step": 290
+    },
+    {
+      "epoch": 0.5065274151436031,
+      "grad_norm": 0.5,
+      "learning_rate": 8.916200648574964e-06,
+      "loss": 0.905,
+      "step": 291
+    },
+    {
+      "epoch": 0.5082680591818973,
+      "grad_norm": 0.490234375,
+      "learning_rate": 8.907233055277999e-06,
+      "loss": 0.9309,
+      "step": 292
+    },
+    {
+      "epoch": 0.5100087032201914,
+      "grad_norm": 0.48046875,
+      "learning_rate": 8.898233063902e-06,
+      "loss": 0.9796,
+      "step": 293
+    },
+    {
+      "epoch": 0.5117493472584856,
+      "grad_norm": 0.5234375,
+      "learning_rate": 8.889200749073285e-06,
+      "loss": 0.9335,
+      "step": 294
+    },
+    {
+      "epoch": 0.5134899912967799,
+      "grad_norm": 0.498046875,
+      "learning_rate": 8.880136185686202e-06,
+      "loss": 0.9292,
+      "step": 295
+    },
+    {
+      "epoch": 0.515230635335074,
+      "grad_norm": 0.478515625,
+      "learning_rate": 8.871039448902488e-06,
+      "loss": 0.9116,
+      "step": 296
+    },
+    {
+      "epoch": 0.5169712793733682,
+      "grad_norm": 0.47265625,
+      "learning_rate": 8.861910614150662e-06,
+      "loss": 0.9315,
+      "step": 297
+    },
+    {
+      "epoch": 0.5187119234116623,
+      "grad_norm": 0.478515625,
+      "learning_rate": 8.852749757125392e-06,
+      "loss": 0.9283,
+      "step": 298
+    },
+    {
+      "epoch": 0.5204525674499565,
+      "grad_norm": 0.5078125,
+      "learning_rate": 8.843556953786872e-06,
+      "loss": 0.952,
+      "step": 299
+    },
+    {
+      "epoch": 0.5221932114882507,
+      "grad_norm": 0.48828125,
+      "learning_rate": 8.834332280360181e-06,
+      "loss": 0.9999,
+      "step": 300
+    },
+    {
+      "epoch": 0.5239338555265448,
+      "grad_norm": 0.498046875,
+      "learning_rate": 8.82507581333467e-06,
+      "loss": 0.9453,
+      "step": 301
+    },
+    {
+      "epoch": 0.525674499564839,
+      "grad_norm": 0.482421875,
+      "learning_rate": 8.815787629463306e-06,
+      "loss": 0.8678,
+      "step": 302
+    },
+    {
+      "epoch": 0.5274151436031331,
+      "grad_norm": 0.5,
+      "learning_rate": 8.806467805762056e-06,
+      "loss": 0.9878,
+      "step": 303
+    },
+    {
+      "epoch": 0.5291557876414273,
+      "grad_norm": 0.49609375,
+      "learning_rate": 8.797116419509232e-06,
+      "loss": 0.8964,
+      "step": 304
+    },
+    {
+      "epoch": 0.5308964316797214,
+      "grad_norm": 0.474609375,
+      "learning_rate": 8.78773354824486e-06,
+      "loss": 0.9584,
+      "step": 305
+    },
+    {
+      "epoch": 0.5326370757180157,
+      "grad_norm": 0.482421875,
+      "learning_rate": 8.778319269770033e-06,
+      "loss": 0.9715,
+      "step": 306
+    },
+    {
+      "epoch": 0.5343777197563099,
+      "grad_norm": 0.484375,
+      "learning_rate": 8.768873662146271e-06,
+      "loss": 0.9034,
+      "step": 307
+    },
+    {
+      "epoch": 0.536118363794604,
+      "grad_norm": 0.494140625,
+      "learning_rate": 8.759396803694863e-06,
+      "loss": 0.9189,
+      "step": 308
+    },
+    {
+      "epoch": 0.5378590078328982,
+      "grad_norm": 0.484375,
+      "learning_rate": 8.749888772996226e-06,
+      "loss": 1.0066,
+      "step": 309
+    },
+    {
+      "epoch": 0.5395996518711923,
+      "grad_norm": 0.47265625,
+      "learning_rate": 8.74034964888926e-06,
+      "loss": 1.0147,
+      "step": 310
+    },
+    {
+      "epoch": 0.5413402959094865,
+      "grad_norm": 0.484375,
+      "learning_rate": 8.730779510470672e-06,
+      "loss": 0.9504,
+      "step": 311
+    },
+    {
+      "epoch": 0.5430809399477807,
+      "grad_norm": 0.482421875,
+      "learning_rate": 8.721178437094346e-06,
+      "loss": 0.9239,
+      "step": 312
+    },
+    {
+      "epoch": 0.5448215839860748,
+      "grad_norm": 0.5078125,
+      "learning_rate": 8.711546508370666e-06,
+      "loss": 0.9145,
+      "step": 313
+    },
+    {
+      "epoch": 0.546562228024369,
+      "grad_norm": 0.498046875,
+      "learning_rate": 8.701883804165867e-06,
+      "loss": 0.897,
+      "step": 314
+    },
+    {
+      "epoch": 0.5483028720626631,
+      "grad_norm": 0.490234375,
+      "learning_rate": 8.692190404601368e-06,
+      "loss": 0.8796,
+      "step": 315
+    },
+    {
+      "epoch": 0.5500435161009574,
+      "grad_norm": 0.53515625,
+      "learning_rate": 8.682466390053106e-06,
+      "loss": 1.03,
+      "step": 316
+    },
+    {
+      "epoch": 0.5517841601392516,
+      "grad_norm": 0.482421875,
+      "learning_rate": 8.672711841150877e-06,
+      "loss": 0.9676,
+      "step": 317
+    },
+    {
+      "epoch": 0.5535248041775457,
+      "grad_norm": 0.490234375,
+      "learning_rate": 8.662926838777657e-06,
+      "loss": 1.0098,
+      "step": 318
+    },
+    {
+      "epoch": 0.5552654482158399,
+      "grad_norm": 0.5078125,
+      "learning_rate": 8.653111464068937e-06,
+      "loss": 0.9344,
+      "step": 319
+    },
+    {
+      "epoch": 0.557006092254134,
+      "grad_norm": 0.51171875,
+      "learning_rate": 8.643265798412057e-06,
+      "loss": 0.9555,
+      "step": 320
+    },
+    {
+      "epoch": 0.5587467362924282,
+      "grad_norm": 0.5,
+      "learning_rate": 8.633389923445515e-06,
+      "loss": 0.8845,
+      "step": 321
+    },
+    {
+      "epoch": 0.5604873803307223,
+      "grad_norm": 0.48046875,
+      "learning_rate": 8.623483921058304e-06,
+      "loss": 0.9403,
+      "step": 322
+    },
+    {
+      "epoch": 0.5622280243690165,
+      "grad_norm": 0.482421875,
+      "learning_rate": 8.613547873389228e-06,
+      "loss": 0.9654,
+      "step": 323
+    },
+    {
+      "epoch": 0.5639686684073107,
+      "grad_norm": 0.490234375,
+      "learning_rate": 8.603581862826222e-06,
+      "loss": 1.0108,
+      "step": 324
+    },
+    {
+      "epoch": 0.5657093124456049,
+      "grad_norm": 0.4921875,
+      "learning_rate": 8.593585972005665e-06,
+      "loss": 0.9708,
+      "step": 325
+    },
+    {
+      "epoch": 0.5674499564838991,
+      "grad_norm": 0.484375,
+      "learning_rate": 8.5835602838117e-06,
+      "loss": 0.9333,
+      "step": 326
+    },
+    {
+      "epoch": 0.5691906005221932,
+      "grad_norm": 0.478515625,
+      "learning_rate": 8.573504881375543e-06,
+      "loss": 0.9067,
+      "step": 327
+    },
+    {
+      "epoch": 0.5709312445604874,
+      "grad_norm": 0.486328125,
+      "learning_rate": 8.563419848074798e-06,
+      "loss": 0.9388,
+      "step": 328
+    },
+    {
+      "epoch": 0.5726718885987816,
+      "grad_norm": 0.4921875,
+      "learning_rate": 8.55330526753276e-06,
+      "loss": 0.9246,
+      "step": 329
+    },
+    {
+      "epoch": 0.5744125326370757,
+      "grad_norm": 0.498046875,
+      "learning_rate": 8.543161223617724e-06,
+      "loss": 0.9222,
+      "step": 330
+    },
+    {
+      "epoch": 0.5761531766753699,
+      "grad_norm": 0.490234375,
+      "learning_rate": 8.532987800442292e-06,
+      "loss": 1.0001,
+      "step": 331
+    },
+    {
+      "epoch": 0.577893820713664,
+      "grad_norm": 0.498046875,
+      "learning_rate": 8.522785082362675e-06,
+      "loss": 1.0003,
+      "step": 332
+    },
+    {
+      "epoch": 0.5796344647519582,
+      "grad_norm": 0.498046875,
+      "learning_rate": 8.512553153977988e-06,
+      "loss": 0.9103,
+      "step": 333
+    },
+    {
+      "epoch": 0.5813751087902524,
+      "grad_norm": 0.498046875,
+      "learning_rate": 8.502292100129553e-06,
+      "loss": 0.9895,
+      "step": 334
+    },
+    {
+      "epoch": 0.5831157528285466,
+      "grad_norm": 0.4921875,
+      "learning_rate": 8.492002005900201e-06,
+      "loss": 0.9981,
+      "step": 335
+    },
+    {
+      "epoch": 0.5848563968668408,
+      "grad_norm": 0.4765625,
+      "learning_rate": 8.481682956613555e-06,
+      "loss": 0.8758,
+      "step": 336
+    },
+    {
+      "epoch": 0.5865970409051349,
+      "grad_norm": 0.49609375,
+      "learning_rate": 8.471335037833328e-06,
+      "loss": 0.9845,
+      "step": 337
+    },
+    {
+      "epoch": 0.5883376849434291,
+      "grad_norm": 0.474609375,
+      "learning_rate": 8.460958335362617e-06,
+      "loss": 0.9974,
+      "step": 338
+    },
+    {
+      "epoch": 0.5900783289817232,
+      "grad_norm": 0.4921875,
+      "learning_rate": 8.450552935243186e-06,
+      "loss": 0.9725,
+      "step": 339
+    },
+    {
+      "epoch": 0.5918189730200174,
+      "grad_norm": 0.48046875,
+      "learning_rate": 8.440118923754757e-06,
+      "loss": 0.9362,
+      "step": 340
+    },
+    {
+      "epoch": 0.5935596170583116,
+      "grad_norm": 0.47265625,
+      "learning_rate": 8.429656387414289e-06,
+      "loss": 0.9659,
+      "step": 341
+    },
+    {
+      "epoch": 0.5953002610966057,
+      "grad_norm": 0.5078125,
+      "learning_rate": 8.419165412975265e-06,
+      "loss": 0.9346,
+      "step": 342
+    },
+    {
+      "epoch": 0.5970409051348999,
+      "grad_norm": 0.48828125,
+      "learning_rate": 8.408646087426975e-06,
+      "loss": 0.8759,
+      "step": 343
+    },
+    {
+      "epoch": 0.598781549173194,
+      "grad_norm": 0.5,
+      "learning_rate": 8.398098497993785e-06,
+      "loss": 0.9359,
+      "step": 344
+    },
+    {
+      "epoch": 0.6005221932114883,
+      "grad_norm": 0.5,
+      "learning_rate": 8.387522732134428e-06,
+      "loss": 0.9171,
+      "step": 345
+    },
+    {
+      "epoch": 0.6022628372497825,
+      "grad_norm": 0.49609375,
+      "learning_rate": 8.376918877541263e-06,
+      "loss": 0.919,
+      "step": 346
+    },
+    {
+      "epoch": 0.6040034812880766,
+      "grad_norm": 0.5234375,
+      "learning_rate": 8.36628702213956e-06,
+      "loss": 0.9211,
+      "step": 347
+    },
+    {
+      "epoch": 0.6057441253263708,
+      "grad_norm": 0.494140625,
+      "learning_rate": 8.355627254086771e-06,
+      "loss": 0.9191,
+      "step": 348
+    },
+    {
+      "epoch": 0.6074847693646649,
+      "grad_norm": 0.49609375,
+      "learning_rate": 8.344939661771784e-06,
+      "loss": 0.926,
+      "step": 349
+    },
+    {
+      "epoch": 0.6092254134029591,
+      "grad_norm": 0.5,
+      "learning_rate": 8.334224333814209e-06,
+      "loss": 0.9088,
+      "step": 350
+    },
+    {
+      "epoch": 0.6109660574412533,
+      "grad_norm": 0.50390625,
+      "learning_rate": 8.323481359063631e-06,
+      "loss": 0.9455,
+      "step": 351
+    },
+    {
+      "epoch": 0.6127067014795474,
+      "grad_norm": 0.50390625,
+      "learning_rate": 8.312710826598884e-06,
+      "loss": 0.8873,
+      "step": 352
+    },
+    {
+      "epoch": 0.6144473455178416,
+      "grad_norm": 0.49609375,
+      "learning_rate": 8.301912825727294e-06,
+      "loss": 0.9562,
+      "step": 353
+    },
+    {
+      "epoch": 0.6161879895561357,
+      "grad_norm": 0.48828125,
+      "learning_rate": 8.29108744598396e-06,
+      "loss": 0.9324,
+      "step": 354
+    },
+    {
+      "epoch": 0.61792863359443,
+      "grad_norm": 0.51171875,
+      "learning_rate": 8.280234777131e-06,
+      "loss": 0.9037,
+      "step": 355
+    },
+    {
+      "epoch": 0.6196692776327241,
+      "grad_norm": 0.5078125,
+      "learning_rate": 8.269354909156803e-06,
+      "loss": 0.9127,
+      "step": 356
+    },
+    {
+      "epoch": 0.6214099216710183,
+      "grad_norm": 0.494140625,
+      "learning_rate": 8.258447932275296e-06,
+      "loss": 0.9003,
+      "step": 357
+    },
+    {
+      "epoch": 0.6231505657093125,
+      "grad_norm": 0.494140625,
+      "learning_rate": 8.247513936925182e-06,
+      "loss": 0.9871,
+      "step": 358
+    },
+    {
+      "epoch": 0.6248912097476066,
+      "grad_norm": 0.50390625,
+      "learning_rate": 8.236553013769198e-06,
+      "loss": 0.8732,
+      "step": 359
+    },
+    {
+      "epoch": 0.6266318537859008,
+      "grad_norm": 0.486328125,
+      "learning_rate": 8.225565253693365e-06,
+      "loss": 0.9563,
+      "step": 360
+    },
+    {
+      "epoch": 0.6266318537859008,
+      "eval_loss": 0.9300395846366882,
+      "eval_runtime": 59.3628,
+      "eval_samples_per_second": 78.972,
+      "eval_steps_per_second": 9.871,
+      "step": 360
+    },
+    {
+      "epoch": 0.6283724978241949,
+      "grad_norm": 0.5078125,
+      "learning_rate": 8.214550747806227e-06,
+      "loss": 0.9173,
+      "step": 361
+    },
+    {
+      "epoch": 0.6301131418624891,
+      "grad_norm": 0.494140625,
+      "learning_rate": 8.2035095874381e-06,
+      "loss": 0.9297,
+      "step": 362
+    },
+    {
+      "epoch": 0.6318537859007833,
+      "grad_norm": 0.50390625,
+      "learning_rate": 8.192441864140314e-06,
+      "loss": 0.9582,
+      "step": 363
+    },
+    {
+      "epoch": 0.6335944299390774,
+      "grad_norm": 0.50390625,
+      "learning_rate": 8.181347669684456e-06,
+      "loss": 0.9176,
+      "step": 364
+    },
+    {
+      "epoch": 0.6353350739773717,
+      "grad_norm": 0.494140625,
+      "learning_rate": 8.170227096061607e-06,
+      "loss": 0.9181,
+      "step": 365
+    },
+    {
+      "epoch": 0.6370757180156658,
+      "grad_norm": 0.486328125,
+      "learning_rate": 8.15908023548158e-06,
+      "loss": 0.9708,
+      "step": 366
+    },
+    {
+      "epoch": 0.63881636205396,
+      "grad_norm": 0.48046875,
+      "learning_rate": 8.147907180372147e-06,
+      "loss": 0.9683,
+      "step": 367
+    },
+    {
+      "epoch": 0.6405570060922542,
+      "grad_norm": 0.4765625,
+      "learning_rate": 8.136708023378292e-06,
+      "loss": 0.9728,
+      "step": 368
+    },
+    {
+      "epoch": 0.6422976501305483,
+      "grad_norm": 0.498046875,
+      "learning_rate": 8.125482857361426e-06,
+      "loss": 0.9696,
+      "step": 369
+    },
+    {
+      "epoch": 0.6440382941688425,
+      "grad_norm": 0.49609375,
+      "learning_rate": 8.114231775398618e-06,
+      "loss": 0.9189,
+      "step": 370
+    },
+    {
+      "epoch": 0.6457789382071366,
+      "grad_norm": 0.486328125,
+      "learning_rate": 8.102954870781831e-06,
+      "loss": 0.9361,
+      "step": 371
+    },
+    {
+      "epoch": 0.6475195822454308,
+      "grad_norm": 0.490234375,
+      "learning_rate": 8.091652237017152e-06,
+      "loss": 0.9612,
+      "step": 372
+    },
+    {
+      "epoch": 0.6492602262837249,
+      "grad_norm": 0.498046875,
+      "learning_rate": 8.080323967823993e-06,
+      "loss": 0.9442,
+      "step": 373
+    },
+    {
+      "epoch": 0.6510008703220191,
+      "grad_norm": 0.48828125,
+      "learning_rate": 8.068970157134349e-06,
+      "loss": 0.9346,
+      "step": 374
+    },
+    {
+      "epoch": 0.6527415143603134,
+      "grad_norm": 0.486328125,
+      "learning_rate": 8.057590899091985e-06,
+      "loss": 0.8999,
+      "step": 375
+    },
+    {
+      "epoch": 0.6544821583986075,
+      "grad_norm": 0.498046875,
+      "learning_rate": 8.046186288051681e-06,
+      "loss": 0.9674,
+      "step": 376
+    },
+    {
+      "epoch": 0.6562228024369017,
+      "grad_norm": 0.48046875,
+      "learning_rate": 8.034756418578434e-06,
+      "loss": 0.9364,
+      "step": 377
+    },
+    {
+      "epoch": 0.6579634464751958,
+      "grad_norm": 0.484375,
+      "learning_rate": 8.023301385446682e-06,
+      "loss": 0.9341,
+      "step": 378
+    },
+    {
+      "epoch": 0.65970409051349,
+      "grad_norm": 0.4921875,
+      "learning_rate": 8.011821283639515e-06,
+      "loss": 0.8986,
+      "step": 379
+    },
+    {
+      "epoch": 0.6614447345517842,
+      "grad_norm": 0.515625,
+      "learning_rate": 8.000316208347891e-06,
+      "loss": 0.9266,
+      "step": 380
+    },
+    {
+      "epoch": 0.6631853785900783,
+      "grad_norm": 0.5078125,
+      "learning_rate": 7.988786254969837e-06,
+      "loss": 0.8911,
+      "step": 381
+    },
+    {
+      "epoch": 0.6649260226283725,
+      "grad_norm": 0.5078125,
+      "learning_rate": 7.977231519109665e-06,
+      "loss": 0.9221,
+      "step": 382
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.466796875,
+      "learning_rate": 7.965652096577188e-06,
+      "loss": 0.9635,
+      "step": 383
+    },
+    {
+      "epoch": 0.6684073107049608,
+      "grad_norm": 0.48828125,
+      "learning_rate": 7.954048083386909e-06,
+      "loss": 0.9023,
+      "step": 384
+    },
+    {
+      "epoch": 0.6701479547432551,
+      "grad_norm": 0.48828125,
+      "learning_rate": 7.942419575757235e-06,
+      "loss": 0.9277,
+      "step": 385
+    },
+    {
+      "epoch": 0.6718885987815492,
+      "grad_norm": 0.486328125,
+      "learning_rate": 7.930766670109675e-06,
+      "loss": 0.9826,
+      "step": 386
+    },
+    {
+      "epoch": 0.6736292428198434,
+      "grad_norm": 0.490234375,
+      "learning_rate": 7.919089463068038e-06,
+      "loss": 0.9223,
+      "step": 387
+    },
+    {
+      "epoch": 0.6753698868581375,
+      "grad_norm": 0.51953125,
+      "learning_rate": 7.907388051457647e-06,
+      "loss": 1.0468,
+      "step": 388
+    },
+    {
+      "epoch": 0.6771105308964317,
+      "grad_norm": 0.4921875,
+      "learning_rate": 7.895662532304516e-06,
+      "loss": 0.9233,
+      "step": 389
+    },
+    {
+      "epoch": 0.6788511749347258,
+      "grad_norm": 0.478515625,
+      "learning_rate": 7.883913002834555e-06,
+      "loss": 0.9756,
+      "step": 390
+    },
+    {
+      "epoch": 0.68059181897302,
+      "grad_norm": 0.494140625,
+      "learning_rate": 7.872139560472767e-06,
+      "loss": 0.9,
+      "step": 391
+    },
+    {
+      "epoch": 0.6823324630113142,
+      "grad_norm": 0.51953125,
+      "learning_rate": 7.860342302842434e-06,
+      "loss": 0.894,
+      "step": 392
+    },
+    {
+      "epoch": 0.6840731070496083,
+      "grad_norm": 0.5,
+      "learning_rate": 7.848521327764309e-06,
+      "loss": 0.9369,
+      "step": 393
+    },
+    {
+      "epoch": 0.6858137510879025,
+      "grad_norm": 0.5,
+      "learning_rate": 7.836676733255809e-06,
+      "loss": 0.9416,
+      "step": 394
+    },
+    {
+      "epoch": 0.6875543951261966,
+      "grad_norm": 0.494140625,
+      "learning_rate": 7.824808617530197e-06,
+      "loss": 0.8992,
+      "step": 395
+    },
+    {
+      "epoch": 0.6892950391644909,
+      "grad_norm": 0.486328125,
+      "learning_rate": 7.812917078995769e-06,
+      "loss": 0.9457,
+      "step": 396
+    },
+    {
+      "epoch": 0.6910356832027851,
+      "grad_norm": 0.4765625,
+      "learning_rate": 7.801002216255042e-06,
+      "loss": 0.9484,
+      "step": 397
+    },
+    {
+      "epoch": 0.6927763272410792,
+      "grad_norm": 0.50390625,
+      "learning_rate": 7.78906412810393e-06,
+      "loss": 0.9297,
+      "step": 398
+    },
+    {
+      "epoch": 0.6945169712793734,
+      "grad_norm": 0.498046875,
+      "learning_rate": 7.777102913530927e-06,
+      "loss": 0.951,
+      "step": 399
+    },
+    {
+      "epoch": 0.6962576153176675,
+      "grad_norm": 0.4921875,
+      "learning_rate": 7.76511867171629e-06,
+      "loss": 0.9091,
+      "step": 400
+    },
+    {
+      "epoch": 0.6979982593559617,
+      "grad_norm": 0.49609375,
+      "learning_rate": 7.753111502031214e-06,
+      "loss": 0.9041,
+      "step": 401
+    },
+    {
+      "epoch": 0.6997389033942559,
+      "grad_norm": 0.50390625,
+      "learning_rate": 7.741081504037009e-06,
+      "loss": 0.9463,
+      "step": 402
+    },
+    {
+      "epoch": 0.70147954743255,
+      "grad_norm": 0.486328125,
+      "learning_rate": 7.729028777484266e-06,
+      "loss": 0.9494,
+      "step": 403
+    },
+    {
+      "epoch": 0.7032201914708442,
+      "grad_norm": 0.482421875,
+      "learning_rate": 7.716953422312044e-06,
+      "loss": 0.9266,
+      "step": 404
+    },
+    {
+      "epoch": 0.7049608355091384,
+      "grad_norm": 0.5078125,
+      "learning_rate": 7.704855538647033e-06,
+      "loss": 0.9731,
+      "step": 405
+    },
+    {
+      "epoch": 0.7067014795474326,
+      "grad_norm": 0.490234375,
+      "learning_rate": 7.692735226802729e-06,
+      "loss": 0.961,
+      "step": 406
+    },
+    {
+      "epoch": 0.7084421235857267,
+      "grad_norm": 0.482421875,
+      "learning_rate": 7.680592587278585e-06,
+      "loss": 0.9506,
+      "step": 407
+    },
+    {
+      "epoch": 0.7101827676240209,
+      "grad_norm": 0.5,
+      "learning_rate": 7.668427720759207e-06,
+      "loss": 0.9084,
+      "step": 408
+    },
+    {
+      "epoch": 0.7119234116623151,
+      "grad_norm": 0.5,
+      "learning_rate": 7.656240728113493e-06,
+      "loss": 0.9147,
+      "step": 409
+    },
+    {
+      "epoch": 0.7136640557006092,
+      "grad_norm": 0.4921875,
+      "learning_rate": 7.644031710393815e-06,
+      "loss": 0.9498,
+      "step": 410
+    },
+    {
+      "epoch": 0.7154046997389034,
+      "grad_norm": 0.490234375,
+      "learning_rate": 7.631800768835167e-06,
+      "loss": 0.9145,
+      "step": 411
+    },
+    {
+      "epoch": 0.7171453437771975,
+      "grad_norm": 0.498046875,
+      "learning_rate": 7.619548004854332e-06,
+      "loss": 0.8884,
+      "step": 412
+    },
+    {
+      "epoch": 0.7188859878154917,
+      "grad_norm": 0.48828125,
+      "learning_rate": 7.607273520049041e-06,
+      "loss": 0.9777,
+      "step": 413
+    },
+    {
+      "epoch": 0.720626631853786,
+      "grad_norm": 0.478515625,
+      "learning_rate": 7.594977416197134e-06,
+      "loss": 0.9802,
+      "step": 414
+    },
+    {
+      "epoch": 0.72236727589208,
+      "grad_norm": 0.49609375,
+      "learning_rate": 7.582659795255707e-06,
+      "loss": 0.9484,
+      "step": 415
+    },
+    {
+      "epoch": 0.7241079199303743,
+      "grad_norm": 0.5078125,
+      "learning_rate": 7.570320759360273e-06,
+      "loss": 0.9231,
+      "step": 416
+    },
+    {
+      "epoch": 0.7258485639686684,
+      "grad_norm": 0.498046875,
+      "learning_rate": 7.557960410823917e-06,
+      "loss": 0.9416,
+      "step": 417
+    },
+    {
+      "epoch": 0.7275892080069626,
+      "grad_norm": 0.4921875,
+      "learning_rate": 7.545578852136443e-06,
+      "loss": 1.0058,
+      "step": 418
+    },
+    {
+      "epoch": 0.7293298520452568,
+      "grad_norm": 0.5,
+      "learning_rate": 7.533176185963523e-06,
+      "loss": 0.9426,
+      "step": 419
+    },
+    {
+      "epoch": 0.7310704960835509,
+      "grad_norm": 0.498046875,
+      "learning_rate": 7.520752515145855e-06,
+      "loss": 0.949,
+      "step": 420
+    },
+    {
+      "epoch": 0.7328111401218451,
+      "grad_norm": 0.484375,
+      "learning_rate": 7.508307942698296e-06,
+      "loss": 0.9727,
+      "step": 421
+    },
+    {
+      "epoch": 0.7345517841601392,
+      "grad_norm": 0.515625,
+      "learning_rate": 7.495842571809021e-06,
+      "loss": 0.8763,
+      "step": 422
+    },
+    {
+      "epoch": 0.7362924281984334,
+      "grad_norm": 0.5,
+      "learning_rate": 7.4833565058386595e-06,
+      "loss": 0.8914,
+      "step": 423
+    },
+    {
+      "epoch": 0.7380330722367275,
+      "grad_norm": 0.49609375,
+      "learning_rate": 7.470849848319443e-06,
+      "loss": 0.9157,
+      "step": 424
+    },
+    {
+      "epoch": 0.7397737162750218,
+      "grad_norm": 0.490234375,
+      "learning_rate": 7.458322702954342e-06,
+      "loss": 0.8703,
+      "step": 425
+    },
+    {
+      "epoch": 0.741514360313316,
+      "grad_norm": 0.48828125,
+      "learning_rate": 7.44577517361621e-06,
+      "loss": 0.9106,
+      "step": 426
+    },
+    {
+      "epoch": 0.7432550043516101,
+      "grad_norm": 0.5078125,
+      "learning_rate": 7.4332073643469196e-06,
+      "loss": 0.8928,
+      "step": 427
+    },
+    {
+      "epoch": 0.7449956483899043,
+      "grad_norm": 0.5234375,
+      "learning_rate": 7.420619379356504e-06,
+      "loss": 0.9402,
+      "step": 428
+    },
+    {
+      "epoch": 0.7467362924281984,
+      "grad_norm": 0.490234375,
+      "learning_rate": 7.408011323022286e-06,
+      "loss": 0.8997,
+      "step": 429
+    },
+    {
+      "epoch": 0.7484769364664926,
+      "grad_norm": 0.48828125,
+      "learning_rate": 7.395383299888019e-06,
+      "loss": 0.9831,
+      "step": 430
+    },
+    {
+      "epoch": 0.7502175805047868,
+      "grad_norm": 0.494140625,
+      "learning_rate": 7.382735414663017e-06,
+      "loss": 1.04,
+      "step": 431
+    },
+    {
+      "epoch": 0.7519582245430809,
+      "grad_norm": 0.49609375,
+      "learning_rate": 7.370067772221285e-06,
+      "loss": 0.9034,
+      "step": 432
+    },
+    {
+      "epoch": 0.7519582245430809,
+      "eval_loss": 0.9259106516838074,
+      "eval_runtime": 59.5681,
+      "eval_samples_per_second": 78.7,
+      "eval_steps_per_second": 9.837,
+      "step": 432
+    },
+    {
+      "epoch": 0.7536988685813751,
+      "grad_norm": 0.4921875,
+      "learning_rate": 7.357380477600654e-06,
+      "loss": 0.9201,
+      "step": 433
+    },
+    {
+      "epoch": 0.7554395126196692,
+      "grad_norm": 0.5078125,
+      "learning_rate": 7.3446736360019065e-06,
+      "loss": 0.9291,
+      "step": 434
+    },
+    {
+      "epoch": 0.7571801566579635,
+      "grad_norm": 0.47265625,
+      "learning_rate": 7.331947352787905e-06,
+      "loss": 0.951,
+      "step": 435
+    },
+    {
+      "epoch": 0.7589208006962577,
+      "grad_norm": 0.49609375,
+      "learning_rate": 7.319201733482715e-06,
+      "loss": 0.9208,
+      "step": 436
+    },
+    {
+      "epoch": 0.7606614447345518,
+      "grad_norm": 0.490234375,
+      "learning_rate": 7.3064368837707425e-06,
+      "loss": 0.9116,
+      "step": 437
+    },
+    {
+      "epoch": 0.762402088772846,
+      "grad_norm": 0.50390625,
+      "learning_rate": 7.2936529094958365e-06,
+      "loss": 0.9165,
+      "step": 438
+    },
+    {
+      "epoch": 0.7641427328111401,
+      "grad_norm": 0.494140625,
+      "learning_rate": 7.280849916660434e-06,
+      "loss": 0.9513,
+      "step": 439
+    },
+    {
+      "epoch": 0.7658833768494343,
+      "grad_norm": 0.48828125,
+      "learning_rate": 7.268028011424664e-06,
+      "loss": 0.9725,
+      "step": 440
+    },
+    {
+      "epoch": 0.7676240208877284,
+      "grad_norm": 0.494140625,
+      "learning_rate": 7.255187300105477e-06,
+      "loss": 0.9328,
+      "step": 441
+    },
+    {
+      "epoch": 0.7693646649260226,
+      "grad_norm": 0.50390625,
+      "learning_rate": 7.24232788917576e-06,
+      "loss": 0.8776,
+      "step": 442
+    },
+    {
+      "epoch": 0.7711053089643168,
+      "grad_norm": 0.5078125,
+      "learning_rate": 7.229449885263451e-06,
+      "loss": 0.9432,
+      "step": 443
+    },
+    {
+      "epoch": 0.7728459530026109,
+      "grad_norm": 0.478515625,
+      "learning_rate": 7.21655339515066e-06,
+      "loss": 0.8951,
+      "step": 444
+    },
+    {
+      "epoch": 0.7745865970409052,
+      "grad_norm": 0.4921875,
+      "learning_rate": 7.203638525772783e-06,
+      "loss": 0.961,
+      "step": 445
+    },
+    {
+      "epoch": 0.7763272410791993,
+      "grad_norm": 0.48828125,
+      "learning_rate": 7.1907053842176075e-06,
+      "loss": 0.9498,
+      "step": 446
+    },
+    {
+      "epoch": 0.7780678851174935,
+      "grad_norm": 0.498046875,
+      "learning_rate": 7.17775407772444e-06,
+      "loss": 0.9638,
+      "step": 447
+    },
+    {
+      "epoch": 0.7798085291557877,
+      "grad_norm": 0.490234375,
+      "learning_rate": 7.164784713683197e-06,
+      "loss": 0.9516,
+      "step": 448
+    },
+    {
+      "epoch": 0.7815491731940818,
+      "grad_norm": 0.466796875,
+      "learning_rate": 7.1517973996335335e-06,
+      "loss": 0.9108,
+      "step": 449
+    },
+    {
+      "epoch": 0.783289817232376,
+      "grad_norm": 0.50390625,
+      "learning_rate": 7.138792243263936e-06,
+      "loss": 0.9089,
+      "step": 450
+    },
+    {
+      "epoch": 0.7850304612706701,
+      "grad_norm": 0.50390625,
+      "learning_rate": 7.125769352410845e-06,
+      "loss": 0.9667,
+      "step": 451
+    },
+    {
+      "epoch": 0.7867711053089643,
+      "grad_norm": 0.484375,
+      "learning_rate": 7.112728835057742e-06,
+      "loss": 0.9458,
+      "step": 452
+    },
+    {
+      "epoch": 0.7885117493472585,
+      "grad_norm": 0.478515625,
+      "learning_rate": 7.099670799334269e-06,
+      "loss": 0.899,
+      "step": 453
+    },
+    {
+      "epoch": 0.7902523933855526,
+      "grad_norm": 0.498046875,
+      "learning_rate": 7.08659535351533e-06,
+      "loss": 0.9031,
+      "step": 454
+    },
+    {
+      "epoch": 0.7919930374238469,
+      "grad_norm": 0.49609375,
+      "learning_rate": 7.073502606020187e-06,
+      "loss": 0.9771,
+      "step": 455
+    },
+    {
+      "epoch": 0.793733681462141,
+      "grad_norm": 0.52734375,
+      "learning_rate": 7.060392665411564e-06,
+      "loss": 0.9113,
+      "step": 456
+    },
+    {
+      "epoch": 0.7954743255004352,
+      "grad_norm": 0.498046875,
+      "learning_rate": 7.0472656403947505e-06,
+      "loss": 1.0172,
+      "step": 457
+    },
+    {
+      "epoch": 0.7972149695387293,
+      "grad_norm": 0.4765625,
+      "learning_rate": 7.034121639816691e-06,
+      "loss": 0.9282,
+      "step": 458
+    },
+    {
+      "epoch": 0.7989556135770235,
+      "grad_norm": 0.50390625,
+      "learning_rate": 7.020960772665096e-06,
+      "loss": 0.896,
+      "step": 459
+    },
+    {
+      "epoch": 0.8006962576153177,
+      "grad_norm": 0.490234375,
+      "learning_rate": 7.007783148067524e-06,
+      "loss": 0.881,
+      "step": 460
+    },
+    {
+      "epoch": 0.8024369016536118,
+      "grad_norm": 0.5078125,
+      "learning_rate": 6.994588875290488e-06,
+      "loss": 0.9155,
+      "step": 461
+    },
+    {
+      "epoch": 0.804177545691906,
+      "grad_norm": 0.5078125,
+      "learning_rate": 6.9813780637385385e-06,
+      "loss": 0.9012,
+      "step": 462
+    },
+    {
+      "epoch": 0.8059181897302001,
+      "grad_norm": 0.50390625,
+      "learning_rate": 6.968150822953372e-06,
+      "loss": 0.9085,
+      "step": 463
+    },
+    {
+      "epoch": 0.8076588337684943,
+      "grad_norm": 0.498046875,
+      "learning_rate": 6.954907262612906e-06,
+      "loss": 0.9818,
+      "step": 464
+    },
+    {
+      "epoch": 0.8093994778067886,
+      "grad_norm": 0.484375,
+      "learning_rate": 6.941647492530378e-06,
+      "loss": 0.9717,
+      "step": 465
+    },
+    {
+      "epoch": 0.8111401218450827,
+      "grad_norm": 0.5,
+      "learning_rate": 6.928371622653434e-06,
+      "loss": 0.9369,
+      "step": 466
+    },
+    {
+      "epoch": 0.8128807658833769,
+      "grad_norm": 0.486328125,
+      "learning_rate": 6.91507976306322e-06,
+      "loss": 0.8943,
+      "step": 467
+    },
+    {
+      "epoch": 0.814621409921671,
+      "grad_norm": 0.498046875,
+      "learning_rate": 6.901772023973459e-06,
+      "loss": 0.9113,
+      "step": 468
+    },
+    {
+      "epoch": 0.8163620539599652,
+      "grad_norm": 0.49609375,
+      "learning_rate": 6.888448515729552e-06,
+      "loss": 0.8995,
+      "step": 469
+    },
+    {
+      "epoch": 0.8181026979982594,
+      "grad_norm": 0.486328125,
+      "learning_rate": 6.8751093488076485e-06,
+      "loss": 0.9159,
+      "step": 470
+    },
+    {
+      "epoch": 0.8198433420365535,
+      "grad_norm": 0.498046875,
+      "learning_rate": 6.86175463381374e-06,
+      "loss": 0.9388,
+      "step": 471
+    },
+    {
+      "epoch": 0.8215839860748477,
+      "grad_norm": 0.49609375,
+      "learning_rate": 6.8483844814827405e-06,
+      "loss": 0.9524,
+      "step": 472
+    },
+    {
+      "epoch": 0.8233246301131418,
+      "grad_norm": 0.5078125,
+      "learning_rate": 6.8349990026775656e-06,
+      "loss": 0.9458,
+      "step": 473
+    },
+    {
+      "epoch": 0.825065274151436,
+      "grad_norm": 0.5,
+      "learning_rate": 6.821598308388217e-06,
+      "loss": 0.9258,
+      "step": 474
+    },
+    {
+      "epoch": 0.8268059181897301,
+      "grad_norm": 0.482421875,
+      "learning_rate": 6.8081825097308584e-06,
+      "loss": 0.9159,
+      "step": 475
+    },
+    {
+      "epoch": 0.8285465622280244,
+      "grad_norm": 0.5,
+      "learning_rate": 6.794751717946897e-06,
+      "loss": 0.9029,
+      "step": 476
+    },
+    {
+      "epoch": 0.8302872062663186,
+      "grad_norm": 0.484375,
+      "learning_rate": 6.781306044402064e-06,
+      "loss": 0.8852,
+      "step": 477
+    },
+    {
+      "epoch": 0.8320278503046127,
+      "grad_norm": 0.482421875,
+      "learning_rate": 6.767845600585479e-06,
+      "loss": 0.945,
+      "step": 478
+    },
+    {
+      "epoch": 0.8337684943429069,
+      "grad_norm": 0.474609375,
+      "learning_rate": 6.754370498108747e-06,
+      "loss": 0.9776,
+      "step": 479
+    },
+    {
+      "epoch": 0.835509138381201,
+      "grad_norm": 0.490234375,
+      "learning_rate": 6.740880848705005e-06,
+      "loss": 0.9229,
+      "step": 480
+    },
+    {
+      "epoch": 0.8372497824194952,
+      "grad_norm": 0.50390625,
+      "learning_rate": 6.72737676422802e-06,
+      "loss": 0.9686,
+      "step": 481
+    },
+    {
+      "epoch": 0.8389904264577894,
+      "grad_norm": 0.484375,
+      "learning_rate": 6.713858356651253e-06,
+      "loss": 0.9459,
+      "step": 482
+    },
+    {
+      "epoch": 0.8407310704960835,
+      "grad_norm": 0.486328125,
+      "learning_rate": 6.700325738066923e-06,
+      "loss": 0.9472,
+      "step": 483
+    },
+    {
+      "epoch": 0.8424717145343777,
+      "grad_norm": 0.5,
+      "learning_rate": 6.686779020685089e-06,
+      "loss": 0.9417,
+      "step": 484
+    },
+    {
+      "epoch": 0.8442123585726719,
+      "grad_norm": 0.490234375,
+      "learning_rate": 6.6732183168327146e-06,
+      "loss": 0.9666,
+      "step": 485
+    },
+    {
+      "epoch": 0.8459530026109661,
+      "grad_norm": 0.490234375,
+      "learning_rate": 6.659643738952732e-06,
+      "loss": 0.9368,
+      "step": 486
+    },
+    {
+      "epoch": 0.8476936466492603,
+      "grad_norm": 0.484375,
+      "learning_rate": 6.646055399603122e-06,
+      "loss": 0.9466,
+      "step": 487
+    },
+    {
+      "epoch": 0.8494342906875544,
+      "grad_norm": 0.5,
+      "learning_rate": 6.6324534114559656e-06,
+      "loss": 0.9159,
+      "step": 488
+    },
+    {
+      "epoch": 0.8511749347258486,
+      "grad_norm": 0.5078125,
+      "learning_rate": 6.618837887296523e-06,
+      "loss": 0.9067,
+      "step": 489
+    },
+    {
+      "epoch": 0.8529155787641427,
+      "grad_norm": 0.48046875,
+      "learning_rate": 6.605208940022289e-06,
+      "loss": 0.9608,
+      "step": 490
+    },
+    {
+      "epoch": 0.8546562228024369,
+      "grad_norm": 0.498046875,
+      "learning_rate": 6.591566682642061e-06,
+      "loss": 0.9663,
+      "step": 491
+    },
+    {
+      "epoch": 0.856396866840731,
+      "grad_norm": 0.5234375,
+      "learning_rate": 6.5779112282750035e-06,
+      "loss": 0.8931,
+      "step": 492
+    },
+    {
+      "epoch": 0.8581375108790252,
+      "grad_norm": 0.498046875,
+      "learning_rate": 6.564242690149705e-06,
+      "loss": 0.9383,
+      "step": 493
+    },
+    {
+      "epoch": 0.8598781549173194,
+      "grad_norm": 0.484375,
+      "learning_rate": 6.550561181603244e-06,
+      "loss": 0.9343,
+      "step": 494
+    },
+    {
+      "epoch": 0.8616187989556136,
+      "grad_norm": 0.486328125,
+      "learning_rate": 6.536866816080247e-06,
+      "loss": 0.9999,
+      "step": 495
+    },
+    {
+      "epoch": 0.8633594429939078,
+      "grad_norm": 0.4765625,
+      "learning_rate": 6.523159707131951e-06,
+      "loss": 0.9955,
+      "step": 496
+    },
+    {
+      "epoch": 0.8651000870322019,
+      "grad_norm": 0.50390625,
+      "learning_rate": 6.509439968415252e-06,
+      "loss": 0.8897,
+      "step": 497
+    },
+    {
+      "epoch": 0.8668407310704961,
+      "grad_norm": 0.498046875,
+      "learning_rate": 6.4957077136917776e-06,
+      "loss": 0.9065,
+      "step": 498
+    },
+    {
+      "epoch": 0.8685813751087903,
+      "grad_norm": 0.4921875,
+      "learning_rate": 6.481963056826932e-06,
+      "loss": 0.9224,
+      "step": 499
+    },
+    {
+      "epoch": 0.8703220191470844,
+      "grad_norm": 0.50390625,
+      "learning_rate": 6.468206111788957e-06,
+      "loss": 0.9638,
+      "step": 500
+    },
+    {
+      "epoch": 0.8720626631853786,
+      "grad_norm": 0.5,
+      "learning_rate": 6.454436992647984e-06,
+      "loss": 0.9232,
+      "step": 501
+    },
+    {
+      "epoch": 0.8738033072236727,
+      "grad_norm": 0.51171875,
+      "learning_rate": 6.440655813575093e-06,
+      "loss": 0.9412,
+      "step": 502
+    },
+    {
+      "epoch": 0.8755439512619669,
+      "grad_norm": 0.5078125,
+      "learning_rate": 6.426862688841359e-06,
+      "loss": 0.9051,
+      "step": 503
+    },
+    {
+      "epoch": 0.8772845953002611,
+      "grad_norm": 0.5078125,
+      "learning_rate": 6.413057732816911e-06,
+      "loss": 0.9214,
+      "step": 504
+    },
+    {
+      "epoch": 0.8772845953002611,
+      "eval_loss": 0.9230473637580872,
+      "eval_runtime": 59.2532,
+      "eval_samples_per_second": 79.118,
+      "eval_steps_per_second": 9.89,
+      "step": 504
+    },
+    {
+      "epoch": 0.8790252393385553,
+      "grad_norm": 0.49609375,
+      "learning_rate": 6.3992410599699786e-06,
+      "loss": 0.908,
+      "step": 505
+    },
+    {
+      "epoch": 0.8807658833768495,
+      "grad_norm": 0.48828125,
+      "learning_rate": 6.385412784865948e-06,
+      "loss": 0.9613,
+      "step": 506
+    },
+    {
+      "epoch": 0.8825065274151436,
+      "grad_norm": 0.498046875,
+      "learning_rate": 6.371573022166409e-06,
+      "loss": 0.9061,
+      "step": 507
+    },
+    {
+      "epoch": 0.8842471714534378,
+      "grad_norm": 0.5,
+      "learning_rate": 6.357721886628201e-06,
+      "loss": 0.9117,
+      "step": 508
+    },
+    {
+      "epoch": 0.8859878154917319,
+      "grad_norm": 0.48046875,
+      "learning_rate": 6.34385949310247e-06,
+      "loss": 0.9219,
+      "step": 509
+    },
+    {
+      "epoch": 0.8877284595300261,
+      "grad_norm": 0.5078125,
+      "learning_rate": 6.329985956533708e-06,
+      "loss": 0.976,
+      "step": 510
+    },
+    {
+      "epoch": 0.8894691035683203,
+      "grad_norm": 0.515625,
+      "learning_rate": 6.3161013919588e-06,
+      "loss": 0.9131,
+      "step": 511
+    },
+    {
+      "epoch": 0.8912097476066144,
+      "grad_norm": 0.5,
+      "learning_rate": 6.302205914506083e-06,
+      "loss": 0.908,
+      "step": 512
+    },
+    {
+      "epoch": 0.8929503916449086,
+      "grad_norm": 0.4921875,
+      "learning_rate": 6.2882996393943706e-06,
+      "loss": 0.9786,
+      "step": 513
+    },
+    {
+      "epoch": 0.8946910356832027,
+      "grad_norm": 0.48828125,
+      "learning_rate": 6.274382681932019e-06,
+      "loss": 0.8728,
+      "step": 514
+    },
+    {
+      "epoch": 0.896431679721497,
+      "grad_norm": 0.482421875,
+      "learning_rate": 6.2604551575159476e-06,
+      "loss": 0.9491,
+      "step": 515
+    },
+    {
+      "epoch": 0.8981723237597912,
+      "grad_norm": 0.49609375,
+      "learning_rate": 6.24651718163071e-06,
+      "loss": 0.995,
+      "step": 516
+    },
+    {
+      "epoch": 0.8999129677980853,
+      "grad_norm": 0.482421875,
+      "learning_rate": 6.2325688698475106e-06,
+      "loss": 0.9574,
+      "step": 517
+    },
+    {
+      "epoch": 0.9016536118363795,
+      "grad_norm": 0.486328125,
+      "learning_rate": 6.218610337823262e-06,
+      "loss": 0.9004,
+      "step": 518
+    },
+    {
+      "epoch": 0.9033942558746736,
+      "grad_norm": 0.48046875,
+      "learning_rate": 6.2046417012996195e-06,
+      "loss": 0.9165,
+      "step": 519
+    },
+    {
+      "epoch": 0.9051348999129678,
+      "grad_norm": 0.498046875,
+      "learning_rate": 6.1906630761020245e-06,
+      "loss": 0.9534,
+      "step": 520
+    },
+    {
+      "epoch": 0.906875543951262,
+      "grad_norm": 0.5,
+      "learning_rate": 6.17667457813874e-06,
+      "loss": 0.8883,
+      "step": 521
+    },
+    {
+      "epoch": 0.9086161879895561,
+      "grad_norm": 0.51171875,
+      "learning_rate": 6.162676323399898e-06,
+      "loss": 0.919,
+      "step": 522
+    },
+    {
+      "epoch": 0.9103568320278503,
+      "grad_norm": 0.4921875,
+      "learning_rate": 6.148668427956523e-06,
+      "loss": 0.9187,
+      "step": 523
+    },
+    {
+      "epoch": 0.9120974760661444,
+      "grad_norm": 0.494140625,
+      "learning_rate": 6.134651007959586e-06,
+      "loss": 0.8984,
+      "step": 524
+    },
+    {
+      "epoch": 0.9138381201044387,
+      "grad_norm": 0.4921875,
+      "learning_rate": 6.120624179639032e-06,
+      "loss": 0.9585,
+      "step": 525
+    },
+    {
+      "epoch": 0.9155787641427328,
+      "grad_norm": 0.49609375,
+      "learning_rate": 6.106588059302818e-06,
+      "loss": 0.9848,
+      "step": 526
+    },
+    {
+      "epoch": 0.917319408181027,
+      "grad_norm": 0.4921875,
+      "learning_rate": 6.092542763335947e-06,
+      "loss": 0.9542,
+      "step": 527
+    },
+    {
+      "epoch": 0.9190600522193212,
+      "grad_norm": 0.5,
+      "learning_rate": 6.0784884081995065e-06,
+      "loss": 0.9476,
+      "step": 528
+    },
+    {
+      "epoch": 0.9208006962576153,
+      "grad_norm": 0.498046875,
+      "learning_rate": 6.0644251104296995e-06,
+      "loss": 0.9425,
+      "step": 529
+    },
+    {
+      "epoch": 0.9225413402959095,
+      "grad_norm": 0.5,
+      "learning_rate": 6.0503529866368824e-06,
+      "loss": 0.9532,
+      "step": 530
+    },
+    {
+      "epoch": 0.9242819843342036,
+      "grad_norm": 0.49609375,
+      "learning_rate": 6.036272153504592e-06,
+      "loss": 0.9243,
+      "step": 531
+    },
+    {
+      "epoch": 0.9260226283724978,
+      "grad_norm": 0.490234375,
+      "learning_rate": 6.022182727788586e-06,
+      "loss": 0.9452,
+      "step": 532
+    },
+    {
+      "epoch": 0.927763272410792,
+      "grad_norm": 0.5234375,
+      "learning_rate": 6.008084826315863e-06,
+      "loss": 0.9243,
+      "step": 533
+    },
+    {
+      "epoch": 0.9295039164490861,
+      "grad_norm": 0.498046875,
+      "learning_rate": 5.993978565983709e-06,
+      "loss": 0.8917,
+      "step": 534
+    },
+    {
+      "epoch": 0.9312445604873804,
+      "grad_norm": 0.4921875,
+      "learning_rate": 5.979864063758717e-06,
+      "loss": 0.9562,
+      "step": 535
+    },
+    {
+      "epoch": 0.9329852045256745,
+      "grad_norm": 0.478515625,
+      "learning_rate": 5.965741436675816e-06,
+      "loss": 0.9904,
+      "step": 536
+    },
+    {
+      "epoch": 0.9347258485639687,
+      "grad_norm": 0.5,
+      "learning_rate": 5.9516108018373145e-06,
+      "loss": 0.8953,
+      "step": 537
+    },
+    {
+      "epoch": 0.9364664926022629,
+      "grad_norm": 0.494140625,
+      "learning_rate": 5.937472276411909e-06,
+      "loss": 0.8674,
+      "step": 538
+    },
+    {
+      "epoch": 0.938207136640557,
+      "grad_norm": 0.498046875,
+      "learning_rate": 5.923325977633732e-06,
+      "loss": 1.0281,
+      "step": 539
+    },
+    {
+      "epoch": 0.9399477806788512,
+      "grad_norm": 0.494140625,
+      "learning_rate": 5.909172022801364e-06,
+      "loss": 0.9278,
+      "step": 540
+    },
+    {
+      "epoch": 0.9416884247171453,
+      "grad_norm": 0.486328125,
+      "learning_rate": 5.8950105292768754e-06,
+      "loss": 0.8704,
+      "step": 541
+    },
+    {
+      "epoch": 0.9434290687554395,
+      "grad_norm": 0.515625,
+      "learning_rate": 5.880841614484841e-06,
+      "loss": 0.9561,
+      "step": 542
+    },
+    {
+      "epoch": 0.9451697127937336,
+      "grad_norm": 0.47265625,
+      "learning_rate": 5.866665395911375e-06,
+      "loss": 0.9407,
+      "step": 543
+    },
+    {
+      "epoch": 0.9469103568320278,
+      "grad_norm": 0.484375,
+      "learning_rate": 5.852481991103149e-06,
+      "loss": 0.8833,
+      "step": 544
+    },
+    {
+      "epoch": 0.9486510008703221,
+      "grad_norm": 0.5,
+      "learning_rate": 5.838291517666427e-06,
+      "loss": 0.993,
+      "step": 545
+    },
+    {
+      "epoch": 0.9503916449086162,
+      "grad_norm": 0.494140625,
+      "learning_rate": 5.824094093266077e-06,
+      "loss": 0.9231,
+      "step": 546
+    },
+    {
+      "epoch": 0.9521322889469104,
+      "grad_norm": 0.490234375,
+      "learning_rate": 5.809889835624611e-06,
+      "loss": 0.9512,
+      "step": 547
+    },
+    {
+      "epoch": 0.9538729329852045,
+      "grad_norm": 0.494140625,
+      "learning_rate": 5.795678862521197e-06,
+      "loss": 0.9175,
+      "step": 548
+    },
+    {
+      "epoch": 0.9556135770234987,
+      "grad_norm": 0.486328125,
+      "learning_rate": 5.781461291790687e-06,
+      "loss": 0.8922,
+      "step": 549
+    },
+    {
+      "epoch": 0.9573542210617929,
+      "grad_norm": 0.494140625,
+      "learning_rate": 5.767237241322641e-06,
+      "loss": 0.9701,
+      "step": 550
+    },
+    {
+      "epoch": 0.959094865100087,
+      "grad_norm": 0.49609375,
+      "learning_rate": 5.753006829060343e-06,
+      "loss": 0.9464,
+      "step": 551
+    },
+    {
+      "epoch": 0.9608355091383812,
+      "grad_norm": 0.5,
+      "learning_rate": 5.738770172999835e-06,
+      "loss": 0.9335,
+      "step": 552
+    },
+    {
+      "epoch": 0.9625761531766753,
+      "grad_norm": 0.458984375,
+      "learning_rate": 5.724527391188927e-06,
+      "loss": 0.9689,
+      "step": 553
+    },
+    {
+      "epoch": 0.9643167972149695,
+      "grad_norm": 0.50390625,
+      "learning_rate": 5.710278601726222e-06,
+      "loss": 0.8756,
+      "step": 554
+    },
+    {
+      "epoch": 0.9660574412532638,
+      "grad_norm": 0.484375,
+      "learning_rate": 5.696023922760141e-06,
+      "loss": 0.9459,
+      "step": 555
+    },
+    {
+      "epoch": 0.9677980852915579,
+      "grad_norm": 0.50390625,
+      "learning_rate": 5.681763472487933e-06,
+      "loss": 0.9305,
+      "step": 556
+    },
+    {
+      "epoch": 0.9695387293298521,
+      "grad_norm": 0.50390625,
+      "learning_rate": 5.667497369154712e-06,
+      "loss": 0.8707,
+      "step": 557
+    },
+    {
+      "epoch": 0.9712793733681462,
+      "grad_norm": 0.47265625,
+      "learning_rate": 5.6532257310524565e-06,
+      "loss": 0.8977,
+      "step": 558
+    },
+    {
+      "epoch": 0.9730200174064404,
+      "grad_norm": 0.494140625,
+      "learning_rate": 5.638948676519043e-06,
+      "loss": 0.8735,
+      "step": 559
+    },
+    {
+      "epoch": 0.9747606614447345,
+      "grad_norm": 0.50390625,
+      "learning_rate": 5.624666323937257e-06,
+      "loss": 0.9788,
+      "step": 560
+    },
+    {
+      "epoch": 0.9765013054830287,
+      "grad_norm": 0.49609375,
+      "learning_rate": 5.610378791733821e-06,
+      "loss": 0.9348,
+      "step": 561
+    },
+    {
+      "epoch": 0.9782419495213229,
+      "grad_norm": 0.48046875,
+      "learning_rate": 5.596086198378399e-06,
+      "loss": 0.9258,
+      "step": 562
+    },
+    {
+      "epoch": 0.979982593559617,
+      "grad_norm": 0.490234375,
+      "learning_rate": 5.5817886623826245e-06,
+      "loss": 0.9184,
+      "step": 563
+    },
+    {
+      "epoch": 0.9817232375979112,
+      "grad_norm": 0.490234375,
+      "learning_rate": 5.567486302299112e-06,
+      "loss": 0.9439,
+      "step": 564
+    },
+    {
+      "epoch": 0.9834638816362054,
+      "grad_norm": 0.474609375,
+      "learning_rate": 5.553179236720482e-06,
+      "loss": 0.9887,
+      "step": 565
+    },
+    {
+      "epoch": 0.9852045256744996,
+      "grad_norm": 0.48828125,
+      "learning_rate": 5.5388675842783644e-06,
+      "loss": 0.9439,
+      "step": 566
+    },
+    {
+      "epoch": 0.9869451697127938,
+      "grad_norm": 0.490234375,
+      "learning_rate": 5.524551463642429e-06,
+      "loss": 0.971,
+      "step": 567
+    },
+    {
+      "epoch": 0.9886858137510879,
+      "grad_norm": 0.486328125,
+      "learning_rate": 5.510230993519391e-06,
+      "loss": 0.9583,
+      "step": 568
+    },
+    {
+      "epoch": 0.9904264577893821,
+      "grad_norm": 0.50390625,
+      "learning_rate": 5.495906292652035e-06,
+      "loss": 0.9788,
+      "step": 569
+    },
+    {
+      "epoch": 0.9921671018276762,
+      "grad_norm": 0.4921875,
+      "learning_rate": 5.48157747981822e-06,
+      "loss": 0.9634,
+      "step": 570
+    },
+    {
+      "epoch": 0.9939077458659704,
+      "grad_norm": 0.5078125,
+      "learning_rate": 5.467244673829908e-06,
+      "loss": 0.8834,
+      "step": 571
+    },
+    {
+      "epoch": 0.9956483899042646,
+      "grad_norm": 0.49609375,
+      "learning_rate": 5.452907993532164e-06,
+      "loss": 0.9113,
+      "step": 572
+    },
+    {
+      "epoch": 0.9973890339425587,
+      "grad_norm": 0.50390625,
+      "learning_rate": 5.438567557802186e-06,
+      "loss": 0.8828,
+      "step": 573
+    },
+    {
+      "epoch": 0.999129677980853,
+      "grad_norm": 0.48828125,
+      "learning_rate": 5.424223485548303e-06,
+      "loss": 0.9363,
+      "step": 574
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1148,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 574,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.613017336006574e+18,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}