diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,23329 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.3854739652870494,
+  "eval_steps": 71,
+  "global_step": 3276,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.078738059471418,
+      "learning_rate": 1e-08,
+      "loss": 0.9787,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 0.9907150864601135,
+      "eval_runtime": 6914.8687,
+      "eval_samples_per_second": 41.995,
+      "eval_steps_per_second": 2.1,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.611292476924348,
+      "learning_rate": 2e-08,
+      "loss": 1.0125,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.680888573298947,
+      "learning_rate": 3.0000000000000004e-08,
+      "loss": 0.9921,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.223600490850279,
+      "learning_rate": 4e-08,
+      "loss": 0.9905,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 13.16229268115578,
+      "learning_rate": 5.0000000000000004e-08,
+      "loss": 1.0453,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.866725221480236,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 1.0109,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.384069184647979,
+      "learning_rate": 7e-08,
+      "loss": 0.9907,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.950365392573932,
+      "learning_rate": 8e-08,
+      "loss": 1.0039,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.269911995803245,
+      "learning_rate": 9e-08,
+      "loss": 0.992,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.784110104038653,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 0.9909,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.530150291016858,
+      "learning_rate": 1.1e-07,
+      "loss": 0.9717,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.46072146393034,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 0.9813,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 10.983926810290162,
+      "learning_rate": 1.3e-07,
+      "loss": 1.0025,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.258765973186346,
+      "learning_rate": 1.4e-07,
+      "loss": 0.9927,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.213060806605753,
+      "learning_rate": 1.5000000000000002e-07,
+      "loss": 0.9729,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.271084778147063,
+      "learning_rate": 1.6e-07,
+      "loss": 0.9861,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.40785228773617,
+      "learning_rate": 1.7000000000000001e-07,
+      "loss": 1.004,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.071262500540902,
+      "learning_rate": 1.8e-07,
+      "loss": 0.999,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 10.939577936304328,
+      "learning_rate": 1.9e-07,
+      "loss": 0.9951,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 10.543379116586797,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 0.9573,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 10.648729634073065,
+      "learning_rate": 2.1000000000000003e-07,
+      "loss": 0.95,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.442791775540536,
+      "learning_rate": 2.2e-07,
+      "loss": 0.9315,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.458344300581993,
+      "learning_rate": 2.3000000000000002e-07,
+      "loss": 0.942,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.249373081558684,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 0.9065,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.104568818785538,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 0.9251,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.776787693093286,
+      "learning_rate": 2.6e-07,
+      "loss": 0.9381,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.809034639899487,
+      "learning_rate": 2.7e-07,
+      "loss": 0.8958,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 8.588186330852137,
+      "learning_rate": 2.8e-07,
+      "loss": 0.9345,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.054903161127684,
+      "learning_rate": 2.9000000000000003e-07,
+      "loss": 0.926,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.409541396614469,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 0.8518,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.079371789402381,
+      "learning_rate": 3.1000000000000005e-07,
+      "loss": 0.8247,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.285936528164377,
+      "learning_rate": 3.2e-07,
+      "loss": 0.8839,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.0378580616482385,
+      "learning_rate": 3.3e-07,
+      "loss": 0.8482,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.8906753179918745,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 0.8848,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.673664611548944,
+      "learning_rate": 3.5000000000000004e-07,
+      "loss": 0.8514,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.672079486287008,
+      "learning_rate": 3.6e-07,
+      "loss": 0.8531,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.534392734183333,
+      "learning_rate": 3.7e-07,
+      "loss": 0.8408,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.426822133821268,
+      "learning_rate": 3.8e-07,
+      "loss": 0.8421,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.050598149897309,
+      "learning_rate": 3.9e-07,
+      "loss": 0.8143,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.3681521726267727,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 0.7697,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.168148018142974,
+      "learning_rate": 4.1000000000000004e-07,
+      "loss": 0.7522,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.988199405686652,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 0.7721,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.7711442613950577,
+      "learning_rate": 4.3e-07,
+      "loss": 0.7777,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.3134967643199253,
+      "learning_rate": 4.4e-07,
+      "loss": 0.7478,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.102821646941925,
+      "learning_rate": 4.5000000000000003e-07,
+      "loss": 0.7614,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.806705011993839,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 0.7344,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.07648392275885,
+      "learning_rate": 4.7000000000000005e-07,
+      "loss": 0.7469,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.9203144322745775,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 0.7611,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.8850095224554737,
+      "learning_rate": 4.900000000000001e-07,
+      "loss": 0.7892,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.8950995227228287,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 0.7249,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.220629645536677,
+      "learning_rate": 5.1e-07,
+      "loss": 0.7286,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6217121963418404,
+      "learning_rate": 5.2e-07,
+      "loss": 0.72,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.7536100967521506,
+      "learning_rate": 5.3e-07,
+      "loss": 0.7303,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.8632645115840507,
+      "learning_rate": 5.4e-07,
+      "loss": 0.7041,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.6438907360560255,
+      "learning_rate": 5.5e-07,
+      "loss": 0.6778,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4220708453129016,
+      "learning_rate": 5.6e-07,
+      "loss": 0.6928,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.3065263284110715,
+      "learning_rate": 5.7e-07,
+      "loss": 0.6705,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.160277550171688,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.714,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0619267911744315,
+      "learning_rate": 5.900000000000001e-07,
+      "loss": 0.6578,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0978405000752092,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 0.6827,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0203948283683908,
+      "learning_rate": 6.100000000000001e-07,
+      "loss": 0.6945,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0376432923241783,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 0.6785,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.1199898784937596,
+      "learning_rate": 6.3e-07,
+      "loss": 0.6879,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9323677710694938,
+      "learning_rate": 6.4e-07,
+      "loss": 0.6958,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9242075211877638,
+      "learning_rate": 6.5e-07,
+      "loss": 0.6632,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9436047323924038,
+      "learning_rate": 6.6e-07,
+      "loss": 0.679,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8822704045541839,
+      "learning_rate": 6.7e-07,
+      "loss": 0.6936,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8869148114540986,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 0.6598,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9122382330757471,
+      "learning_rate": 6.900000000000001e-07,
+      "loss": 0.6733,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9213292233226152,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 0.6868,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7942384246431622,
+      "learning_rate": 7.1e-07,
+      "loss": 0.6438,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 0.6540535092353821,
+      "eval_runtime": 6906.1355,
+      "eval_samples_per_second": 42.048,
+      "eval_steps_per_second": 2.102,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8205470446647939,
+      "learning_rate": 7.2e-07,
+      "loss": 0.6462,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7946802028186414,
+      "learning_rate": 7.3e-07,
+      "loss": 0.6806,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8828845496654094,
+      "learning_rate": 7.4e-07,
+      "loss": 0.6664,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.9383931456659222,
+      "learning_rate": 7.5e-07,
+      "loss": 0.6137,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8776340800086451,
+      "learning_rate": 7.6e-07,
+      "loss": 0.6453,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7928098345984148,
+      "learning_rate": 7.7e-07,
+      "loss": 0.6245,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8548051569752384,
+      "learning_rate": 7.8e-07,
+      "loss": 0.6588,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7989762163409693,
+      "learning_rate": 7.900000000000001e-07,
+      "loss": 0.6225,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7487196972831345,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.6306,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7503963380748643,
+      "learning_rate": 8.100000000000001e-07,
+      "loss": 0.642,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7715020340302331,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 0.637,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7461624365856538,
+      "learning_rate": 8.300000000000001e-07,
+      "loss": 0.6456,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7873038726512404,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 0.6301,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8371533025271144,
+      "learning_rate": 8.500000000000001e-07,
+      "loss": 0.6294,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7722902390890481,
+      "learning_rate": 8.6e-07,
+      "loss": 0.6285,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7349010275266745,
+      "learning_rate": 8.7e-07,
+      "loss": 0.656,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8710625748918476,
+      "learning_rate": 8.8e-07,
+      "loss": 0.6156,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7833176365407506,
+      "learning_rate": 8.900000000000001e-07,
+      "loss": 0.6329,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7956237199904019,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 0.604,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7663333987942103,
+      "learning_rate": 9.100000000000001e-07,
+      "loss": 0.6132,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7350037692506446,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 0.5959,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7095481904743137,
+      "learning_rate": 9.300000000000001e-07,
+      "loss": 0.581,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7669373034765433,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 0.6059,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7042131948418571,
+      "learning_rate": 9.500000000000001e-07,
+      "loss": 0.6159,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7415353323840956,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 0.6046,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7038043696743754,
+      "learning_rate": 9.7e-07,
+      "loss": 0.6313,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7144659980129581,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.6091,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7198651744206456,
+      "learning_rate": 9.9e-07,
+      "loss": 0.6296,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7518090561897931,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.6172,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7176227042870604,
+      "learning_rate": 1.01e-06,
+      "loss": 0.5913,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7227768155981023,
+      "learning_rate": 1.02e-06,
+      "loss": 0.592,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.7111683414742458,
+      "learning_rate": 1.03e-06,
+      "loss": 0.625,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6992806259847806,
+      "learning_rate": 1.04e-06,
+      "loss": 0.5955,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.673513907571938,
+      "learning_rate": 1.0500000000000001e-06,
+      "loss": 0.6244,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6626454057249019,
+      "learning_rate": 1.06e-06,
+      "loss": 0.6246,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6807547559648992,
+      "learning_rate": 1.0700000000000001e-06,
+      "loss": 0.6302,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7118070037431555,
+      "learning_rate": 1.08e-06,
+      "loss": 0.612,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7101111808553712,
+      "learning_rate": 1.0900000000000002e-06,
+      "loss": 0.6299,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7361716975541328,
+      "learning_rate": 1.1e-06,
+      "loss": 0.5889,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7507623271913958,
+      "learning_rate": 1.1100000000000002e-06,
+      "loss": 0.5939,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.69232482698427,
+      "learning_rate": 1.12e-06,
+      "loss": 0.5857,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7560268686291741,
+      "learning_rate": 1.1300000000000002e-06,
+      "loss": 0.5926,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6909950381406327,
+      "learning_rate": 1.14e-06,
+      "loss": 0.6259,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.705457142068379,
+      "learning_rate": 1.1500000000000002e-06,
+      "loss": 0.594,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7479733597296523,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 0.5941,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6703707169078938,
+      "learning_rate": 1.1700000000000002e-06,
+      "loss": 0.6002,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7142454896237415,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.5721,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6410921583173411,
+      "learning_rate": 1.19e-06,
+      "loss": 0.5549,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7371839254744956,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 0.6076,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6642983393421674,
+      "learning_rate": 1.21e-06,
+      "loss": 0.6164,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6622892175821214,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 0.5916,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6836278867814578,
+      "learning_rate": 1.23e-06,
+      "loss": 0.6055,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7449640043860738,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 0.5706,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7030914815199217,
+      "learning_rate": 1.25e-06,
+      "loss": 0.5962,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6636464764371871,
+      "learning_rate": 1.26e-06,
+      "loss": 0.6004,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7292101825675744,
+      "learning_rate": 1.2700000000000001e-06,
+      "loss": 0.6002,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7002155035767813,
+      "learning_rate": 1.28e-06,
+      "loss": 0.597,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.697947240938124,
+      "learning_rate": 1.2900000000000001e-06,
+      "loss": 0.5788,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7106171095112853,
+      "learning_rate": 1.3e-06,
+      "loss": 0.5878,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7712154049962048,
+      "learning_rate": 1.3100000000000002e-06,
+      "loss": 0.6198,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9336735438619129,
+      "learning_rate": 1.32e-06,
+      "loss": 0.5883,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6689366967241089,
+      "learning_rate": 1.3300000000000002e-06,
+      "loss": 0.5881,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6553033178045365,
+      "learning_rate": 1.34e-06,
+      "loss": 0.564,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7780421140633135,
+      "learning_rate": 1.3500000000000002e-06,
+      "loss": 0.5939,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7512017485981449,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 0.5967,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7273946733225628,
+      "learning_rate": 1.3700000000000002e-06,
+      "loss": 0.593,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7070418593498196,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.5848,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7716412837183378,
+      "learning_rate": 1.3900000000000002e-06,
+      "loss": 0.6087,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7211339598744784,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 0.5711,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6962486858771757,
+      "learning_rate": 1.41e-06,
+      "loss": 0.5791,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7001118725454738,
+      "learning_rate": 1.42e-06,
+      "loss": 0.5772,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "eval_loss": 0.5818271040916443,
+      "eval_runtime": 6911.9695,
+      "eval_samples_per_second": 42.012,
+      "eval_steps_per_second": 2.101,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.764854682764742,
+      "learning_rate": 1.43e-06,
+      "loss": 0.6009,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.730261637210628,
+      "learning_rate": 1.44e-06,
+      "loss": 0.5785,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.8038474557194998,
+      "learning_rate": 1.45e-06,
+      "loss": 0.607,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6835086511414872,
+      "learning_rate": 1.46e-06,
+      "loss": 0.5825,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7505101234005623,
+      "learning_rate": 1.4700000000000001e-06,
+      "loss": 0.6049,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7840302962113346,
+      "learning_rate": 1.48e-06,
+      "loss": 0.5569,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6936446529890948,
+      "learning_rate": 1.4900000000000001e-06,
+      "loss": 0.5713,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7331878475223359,
+      "learning_rate": 1.5e-06,
+      "loss": 0.588,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6745518982361585,
+      "learning_rate": 1.5100000000000002e-06,
+      "loss": 0.5576,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.6950585712176994,
+      "learning_rate": 1.52e-06,
+      "loss": 0.5853,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7339007955624004,
+      "learning_rate": 1.5300000000000002e-06,
+      "loss": 0.5887,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7086482597013368,
+      "learning_rate": 1.54e-06,
+      "loss": 0.5857,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6929345752180732,
+      "learning_rate": 1.5500000000000002e-06,
+      "loss": 0.5937,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.716359352444948,
+      "learning_rate": 1.56e-06,
+      "loss": 0.5459,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7432089483211889,
+      "learning_rate": 1.5700000000000002e-06,
+      "loss": 0.5766,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6876861316239075,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.5729,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7652117324379928,
+      "learning_rate": 1.5900000000000002e-06,
+      "loss": 0.5714,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6874557208270947,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.5786,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6873913339602188,
+      "learning_rate": 1.6100000000000003e-06,
+      "loss": 0.588,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.683965861475918,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 0.5772,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.774874549937846,
+      "learning_rate": 1.6300000000000003e-06,
+      "loss": 0.6,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7021871067570723,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 0.5689,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7209423777996005,
+      "learning_rate": 1.6500000000000003e-06,
+      "loss": 0.5607,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6803679044895155,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 0.609,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7712655814872096,
+      "learning_rate": 1.6700000000000003e-06,
+      "loss": 0.5764,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7619624904596949,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 0.5599,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.669785223369481,
+      "learning_rate": 1.6900000000000003e-06,
+      "loss": 0.5518,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7392059624798044,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 0.595,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6847168836492608,
+      "learning_rate": 1.7100000000000004e-06,
+      "loss": 0.5554,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7793064159857329,
+      "learning_rate": 1.72e-06,
+      "loss": 0.5905,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6760936921129691,
+      "learning_rate": 1.73e-06,
+      "loss": 0.5278,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.7120881185318579,
+      "learning_rate": 1.74e-06,
+      "loss": 0.562,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.6777641027946281,
+      "learning_rate": 1.75e-06,
+      "loss": 0.5967,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6644124127387283,
+      "learning_rate": 1.76e-06,
+      "loss": 0.5685,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6633049014407767,
+      "learning_rate": 1.77e-06,
+      "loss": 0.5694,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6598336805084065,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.5556,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7006058904587646,
+      "learning_rate": 1.79e-06,
+      "loss": 0.5769,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7269153745138233,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 0.5725,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7282646554724388,
+      "learning_rate": 1.81e-06,
+      "loss": 0.5593,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.857099360195062,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 0.5669,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6950980962926347,
+      "learning_rate": 1.83e-06,
+      "loss": 0.5777,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6604073281679678,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 0.5918,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6690212574223983,
+      "learning_rate": 1.85e-06,
+      "loss": 0.561,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6959851809345552,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 0.5523,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.375144332117638,
+      "learning_rate": 1.87e-06,
+      "loss": 0.5612,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7710335154558103,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 0.5515,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7387587574522239,
+      "learning_rate": 1.8900000000000001e-06,
+      "loss": 0.548,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6895693730643745,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 0.5525,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6809790348929371,
+      "learning_rate": 1.9100000000000003e-06,
+      "loss": 0.5539,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6975774380503847,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 0.5576,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.65978668039197,
+      "learning_rate": 1.93e-06,
+      "loss": 0.5546,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7381481487351611,
+      "learning_rate": 1.94e-06,
+      "loss": 0.5362,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7590514250142023,
+      "learning_rate": 1.9500000000000004e-06,
+      "loss": 0.5496,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6814269181772578,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 0.595,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7290336017009336,
+      "learning_rate": 1.97e-06,
+      "loss": 0.5845,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7087771732688267,
+      "learning_rate": 1.98e-06,
+      "loss": 0.5425,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6740991887260622,
+      "learning_rate": 1.9900000000000004e-06,
+      "loss": 0.5697,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7692213485546533,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.5832,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7310442342385465,
+      "learning_rate": 2.0100000000000002e-06,
+      "loss": 0.5484,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6807932721845763,
+      "learning_rate": 2.02e-06,
+      "loss": 0.5476,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.703371626783706,
+      "learning_rate": 2.0300000000000005e-06,
+      "loss": 0.6031,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6802894338699406,
+      "learning_rate": 2.04e-06,
+      "loss": 0.5828,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6652596394872644,
+      "learning_rate": 2.05e-06,
+      "loss": 0.5614,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.737938303801843,
+      "learning_rate": 2.06e-06,
+      "loss": 0.5649,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6851616685774734,
+      "learning_rate": 2.07e-06,
+      "loss": 0.5923,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7193070222820541,
+      "learning_rate": 2.08e-06,
+      "loss": 0.5344,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6982256320037129,
+      "learning_rate": 2.09e-06,
+      "loss": 0.524,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6731186167503178,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 0.524,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.1798176224305597,
+      "learning_rate": 2.11e-06,
+      "loss": 0.5603,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7003894417531635,
+      "learning_rate": 2.12e-06,
+      "loss": 0.5547,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6678276891220589,
+      "learning_rate": 2.13e-06,
+      "loss": 0.543,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 0.5565809607505798,
+      "eval_runtime": 6927.1816,
+      "eval_samples_per_second": 41.92,
+      "eval_steps_per_second": 2.096,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6757438454080638,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 0.5468,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.770869092645212,
+      "learning_rate": 2.15e-06,
+      "loss": 0.56,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7073954416641812,
+      "learning_rate": 2.16e-06,
+      "loss": 0.5481,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7155585643449607,
+      "learning_rate": 2.17e-06,
+      "loss": 0.5616,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7290481543044689,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.5418,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.8118091409501089,
+      "learning_rate": 2.19e-06,
+      "loss": 0.5438,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.8302800073178411,
+      "learning_rate": 2.2e-06,
+      "loss": 0.5487,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.7364426707486922,
+      "learning_rate": 2.21e-06,
+      "loss": 0.5855,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.456340142180152,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 0.5836,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7385740541544212,
+      "learning_rate": 2.2300000000000002e-06,
+      "loss": 0.5697,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.84040439389078,
+      "learning_rate": 2.24e-06,
+      "loss": 0.5688,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6791205410588949,
+      "learning_rate": 2.25e-06,
+      "loss": 0.5504,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7249144826404104,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 0.5573,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7512914278711404,
+      "learning_rate": 2.2700000000000003e-06,
+      "loss": 0.5443,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7283685135498201,
+      "learning_rate": 2.28e-06,
+      "loss": 0.5423,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.745183565987288,
+      "learning_rate": 2.29e-06,
+      "loss": 0.5304,
+      "step": 229
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.738561325322298,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 0.5406,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8285144394816474,
+      "learning_rate": 2.3100000000000003e-06,
+      "loss": 0.5299,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6778837717707735,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 0.5195,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7719966350372699,
+      "learning_rate": 2.33e-06,
+      "loss": 0.5571,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.76143480039083,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 0.5392,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7799555692877767,
+      "learning_rate": 2.35e-06,
+      "loss": 0.5675,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.172177763095597,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 0.5751,
+      "step": 236
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7382398234329495,
+      "learning_rate": 2.37e-06,
+      "loss": 0.541,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7869817068341356,
+      "learning_rate": 2.38e-06,
+      "loss": 0.5616,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.747149871498409,
+      "learning_rate": 2.39e-06,
+      "loss": 0.5368,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.709533915967845,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.5401,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8009167743191223,
+      "learning_rate": 2.4100000000000002e-06,
+      "loss": 0.5553,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7824397361977652,
+      "learning_rate": 2.42e-06,
+      "loss": 0.5582,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7713371690459996,
+      "learning_rate": 2.43e-06,
+      "loss": 0.5567,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6841472190627085,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 0.5255,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.7395314286170461,
+      "learning_rate": 2.4500000000000003e-06,
+      "loss": 0.5474,
+      "step": 245
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9037560995126637,
+      "learning_rate": 2.46e-06,
+      "loss": 0.5443,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7394243896953805,
+      "learning_rate": 2.47e-06,
+      "loss": 0.5654,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7368266522326133,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 0.5588,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7216141071484306,
+      "learning_rate": 2.4900000000000003e-06,
+      "loss": 0.546,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7381090709238742,
+      "learning_rate": 2.5e-06,
+      "loss": 0.5359,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.81508667979967,
+      "learning_rate": 2.51e-06,
+      "loss": 0.5831,
+      "step": 251
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7507679622871583,
+      "learning_rate": 2.52e-06,
+      "loss": 0.5536,
+      "step": 252
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.725818735593307,
+      "learning_rate": 2.5300000000000003e-06,
+      "loss": 0.5301,
+      "step": 253
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7171172879542945,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 0.562,
+      "step": 254
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8845290493858748,
+      "learning_rate": 2.55e-06,
+      "loss": 0.5641,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7748353959313085,
+      "learning_rate": 2.56e-06,
+      "loss": 0.5617,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.4792826780398232,
+      "learning_rate": 2.5700000000000004e-06,
+      "loss": 0.5218,
+      "step": 257
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7755584686345505,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.5549,
+      "step": 258
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7144098990925435,
+      "learning_rate": 2.59e-06,
+      "loss": 0.5452,
+      "step": 259
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7016327562812078,
+      "learning_rate": 2.6e-06,
+      "loss": 0.5359,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7674707640501007,
+      "learning_rate": 2.6100000000000004e-06,
+      "loss": 0.5492,
+      "step": 261
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6991460195923854,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 0.5739,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7481462435947455,
+      "learning_rate": 2.6300000000000002e-06,
+      "loss": 0.5382,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6903104793011363,
+      "learning_rate": 2.64e-06,
+      "loss": 0.5774,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7276434522232335,
+      "learning_rate": 2.6500000000000005e-06,
+      "loss": 0.5395,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7049348082872935,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 0.5077,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6860853746928447,
+      "learning_rate": 2.6700000000000003e-06,
+      "loss": 0.5365,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7019598422590565,
+      "learning_rate": 2.68e-06,
+      "loss": 0.5405,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7452321432305177,
+      "learning_rate": 2.6900000000000005e-06,
+      "loss": 0.5301,
+      "step": 269
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6869176763831081,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 0.5294,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.923474966878648,
+      "learning_rate": 2.7100000000000003e-06,
+      "loss": 0.5341,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6685065451328585,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 0.5364,
+      "step": 272
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7024497350510225,
+      "learning_rate": 2.7300000000000005e-06,
+      "loss": 0.5411,
+      "step": 273
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6954294556378668,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 0.5599,
+      "step": 274
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7307559316445182,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 0.5546,
+      "step": 275
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6817035611374646,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 0.5252,
+      "step": 276
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6749522949619997,
+      "learning_rate": 2.7700000000000006e-06,
+      "loss": 0.5329,
+      "step": 277
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6885720885016431,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.5437,
+      "step": 278
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6847069171457737,
+      "learning_rate": 2.7900000000000004e-06,
+      "loss": 0.5298,
+      "step": 279
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6540194496966271,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 0.5129,
+      "step": 280
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7232645465800646,
+      "learning_rate": 2.8100000000000006e-06,
+      "loss": 0.5727,
+      "step": 281
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6686647596073149,
+      "learning_rate": 2.82e-06,
+      "loss": 0.5507,
+      "step": 282
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6600278121030401,
+      "learning_rate": 2.83e-06,
+      "loss": 0.5413,
+      "step": 283
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7295788704850122,
+      "learning_rate": 2.84e-06,
+      "loss": 0.5644,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 0.5427850484848022,
+      "eval_runtime": 6910.0669,
+      "eval_samples_per_second": 42.024,
+      "eval_steps_per_second": 2.101,
+      "step": 284
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7172749652762687,
+      "learning_rate": 2.85e-06,
+      "loss": 0.5586,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7146040704416872,
+      "learning_rate": 2.86e-06,
+      "loss": 0.5468,
+      "step": 286
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6773322028612333,
+      "learning_rate": 2.87e-06,
+      "loss": 0.5439,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6807935236750091,
+      "learning_rate": 2.88e-06,
+      "loss": 0.541,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7137692746953846,
+      "learning_rate": 2.89e-06,
+      "loss": 0.5299,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7079425743756034,
+      "learning_rate": 2.9e-06,
+      "loss": 0.5591,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.6971956190090158,
+      "learning_rate": 2.91e-06,
+      "loss": 0.5515,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.7203057497367688,
+      "learning_rate": 2.92e-06,
+      "loss": 0.5335,
+      "step": 292
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6508573931612972,
+      "learning_rate": 2.93e-06,
+      "loss": 0.5242,
+      "step": 293
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.682319005640009,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 0.544,
+      "step": 294
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6467072142132468,
+      "learning_rate": 2.95e-06,
+      "loss": 0.5236,
+      "step": 295
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6776154655530738,
+      "learning_rate": 2.96e-06,
+      "loss": 0.5271,
+      "step": 296
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6938107980975571,
+      "learning_rate": 2.97e-06,
+      "loss": 0.5382,
+      "step": 297
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.721981231896848,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.5911,
+      "step": 298
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.712390399983667,
+      "learning_rate": 2.99e-06,
+      "loss": 0.5546,
+      "step": 299
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7104303392210323,
+      "learning_rate": 3e-06,
+      "loss": 0.5519,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6771404308712767,
+      "learning_rate": 3.01e-06,
+      "loss": 0.535,
+      "step": 301
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6911636460013854,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 0.5293,
+      "step": 302
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6692540043016396,
+      "learning_rate": 3.0300000000000002e-06,
+      "loss": 0.542,
+      "step": 303
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6744768013007634,
+      "learning_rate": 3.04e-06,
+      "loss": 0.515,
+      "step": 304
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.670911158940749,
+      "learning_rate": 3.05e-06,
+      "loss": 0.5163,
+      "step": 305
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.663474423719772,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 0.5305,
+      "step": 306
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6925074950155179,
+      "learning_rate": 3.0700000000000003e-06,
+      "loss": 0.5415,
+      "step": 307
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6934197367843478,
+      "learning_rate": 3.08e-06,
+      "loss": 0.5525,
+      "step": 308
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6661951817270549,
+      "learning_rate": 3.09e-06,
+      "loss": 0.5608,
+      "step": 309
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6502761715692771,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 0.5597,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.644832829029986,
+      "learning_rate": 3.1100000000000003e-06,
+      "loss": 0.5438,
+      "step": 311
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6264165348789397,
+      "learning_rate": 3.12e-06,
+      "loss": 0.5292,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7028440661397689,
+      "learning_rate": 3.13e-06,
+      "loss": 0.5475,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6530445345568231,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 0.5567,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6365729580283129,
+      "learning_rate": 3.1500000000000003e-06,
+      "loss": 0.5375,
+      "step": 315
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7033548602718471,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 0.5691,
+      "step": 316
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6757049797519596,
+      "learning_rate": 3.17e-06,
+      "loss": 0.5206,
+      "step": 317
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6685908127465012,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.5304,
+      "step": 318
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6793556517907352,
+      "learning_rate": 3.1900000000000004e-06,
+      "loss": 0.5217,
+      "step": 319
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6523324979691255,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.501,
+      "step": 320
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6443915969872007,
+      "learning_rate": 3.21e-06,
+      "loss": 0.5315,
+      "step": 321
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6681792930023069,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 0.5411,
+      "step": 322
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6976917063523176,
+      "learning_rate": 3.2300000000000004e-06,
+      "loss": 0.5356,
+      "step": 323
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6455932268651956,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 0.5085,
+      "step": 324
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7324457281506692,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 0.5804,
+      "step": 325
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6887957564513499,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 0.5315,
+      "step": 326
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6933538113778166,
+      "learning_rate": 3.2700000000000005e-06,
+      "loss": 0.529,
+      "step": 327
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6968756135404809,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 0.509,
+      "step": 328
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6547487769661117,
+      "learning_rate": 3.2900000000000003e-06,
+      "loss": 0.5319,
+      "step": 329
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6918958083567368,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 0.5369,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7537331163722618,
+      "learning_rate": 3.3100000000000005e-06,
+      "loss": 0.5559,
+      "step": 331
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6622519534641422,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 0.5399,
+      "step": 332
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7164771428347083,
+      "learning_rate": 3.3300000000000003e-06,
+      "loss": 0.5565,
+      "step": 333
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7082376525584928,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 0.5212,
+      "step": 334
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6726486174520817,
+      "learning_rate": 3.3500000000000005e-06,
+      "loss": 0.5214,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6698599823643026,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 0.5392,
+      "step": 336
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7412977141620687,
+      "learning_rate": 3.3700000000000003e-06,
+      "loss": 0.5786,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7232716108771607,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.5315,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.6392581387125058,
+      "learning_rate": 3.3900000000000006e-06,
+      "loss": 0.5443,
+      "step": 339
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.709924249993068,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 0.5518,
+      "step": 340
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6805700381054698,
+      "learning_rate": 3.4100000000000004e-06,
+      "loss": 0.51,
+      "step": 341
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6404690210430515,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 0.5482,
+      "step": 342
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7350376016254183,
+      "learning_rate": 3.4300000000000006e-06,
+      "loss": 0.4998,
+      "step": 343
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.692941753107107,
+      "learning_rate": 3.44e-06,
+      "loss": 0.5174,
+      "step": 344
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6917522453335027,
+      "learning_rate": 3.45e-06,
+      "loss": 0.5252,
+      "step": 345
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7698041778535101,
+      "learning_rate": 3.46e-06,
+      "loss": 0.5116,
+      "step": 346
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6534415447218267,
+      "learning_rate": 3.4700000000000002e-06,
+      "loss": 0.5324,
+      "step": 347
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6553279473118478,
+      "learning_rate": 3.48e-06,
+      "loss": 0.4979,
+      "step": 348
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7260157286864319,
+      "learning_rate": 3.49e-06,
+      "loss": 0.5883,
+      "step": 349
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6509209097177096,
+      "learning_rate": 3.5e-06,
+      "loss": 0.5532,
+      "step": 350
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7431590152540184,
+      "learning_rate": 3.5100000000000003e-06,
+      "loss": 0.5651,
+      "step": 351
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7154996648263839,
+      "learning_rate": 3.52e-06,
+      "loss": 0.5522,
+      "step": 352
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6949365648624505,
+      "learning_rate": 3.53e-06,
+      "loss": 0.5648,
+      "step": 353
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6507409922513757,
+      "learning_rate": 3.54e-06,
+      "loss": 0.5278,
+      "step": 354
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7005428814595557,
+      "learning_rate": 3.5500000000000003e-06,
+      "loss": 0.5555,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "eval_loss": 0.5336291193962097,
+      "eval_runtime": 6911.9871,
+      "eval_samples_per_second": 42.012,
+      "eval_steps_per_second": 2.101,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6932495434274139,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 0.515,
+      "step": 356
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6592137259570553,
+      "learning_rate": 3.57e-06,
+      "loss": 0.5593,
+      "step": 357
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6621976521925507,
+      "learning_rate": 3.58e-06,
+      "loss": 0.5314,
+      "step": 358
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6743898928954292,
+      "learning_rate": 3.5900000000000004e-06,
+      "loss": 0.5458,
+      "step": 359
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6270336100237296,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 0.5281,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.718834892269395,
+      "learning_rate": 3.61e-06,
+      "loss": 0.54,
+      "step": 361
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.6096717034209563,
+      "learning_rate": 3.62e-06,
+      "loss": 0.5023,
+      "step": 362
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6952055757879536,
+      "learning_rate": 3.6300000000000004e-06,
+      "loss": 0.5268,
+      "step": 363
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6571216773223735,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 0.5314,
+      "step": 364
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.644276415756237,
+      "learning_rate": 3.65e-06,
+      "loss": 0.5194,
+      "step": 365
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.298671233861733,
+      "learning_rate": 3.66e-06,
+      "loss": 0.5117,
+      "step": 366
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7103634954229385,
+      "learning_rate": 3.6700000000000004e-06,
+      "loss": 0.5385,
+      "step": 367
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6104474479946076,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 0.4998,
+      "step": 368
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6896768428677847,
+      "learning_rate": 3.6900000000000002e-06,
+      "loss": 0.533,
+      "step": 369
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6413918813463226,
+      "learning_rate": 3.7e-06,
+      "loss": 0.5312,
+      "step": 370
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7573960000478444,
+      "learning_rate": 3.7100000000000005e-06,
+      "loss": 0.5118,
+      "step": 371
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6754673328160921,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 0.531,
+      "step": 372
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7380227752448969,
+      "learning_rate": 3.7300000000000003e-06,
+      "loss": 0.553,
+      "step": 373
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7101835866153544,
+      "learning_rate": 3.74e-06,
+      "loss": 0.5356,
+      "step": 374
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6854482179866157,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.5241,
+      "step": 375
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6626980381085176,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 0.51,
+      "step": 376
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6510834607877704,
+      "learning_rate": 3.7700000000000003e-06,
+      "loss": 0.5231,
+      "step": 377
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6659115075290825,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.5164,
+      "step": 378
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6977709143095813,
+      "learning_rate": 3.79e-06,
+      "loss": 0.5529,
+      "step": 379
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6429659300311611,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 0.5177,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6780609817770847,
+      "learning_rate": 3.8100000000000004e-06,
+      "loss": 0.5017,
+      "step": 381
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6312430732177394,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 0.5034,
+      "step": 382
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6557950564147815,
+      "learning_rate": 3.830000000000001e-06,
+      "loss": 0.5297,
+      "step": 383
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7189637600222241,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 0.5396,
+      "step": 384
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7535141909531238,
+      "learning_rate": 3.85e-06,
+      "loss": 0.5572,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6848030312310904,
+      "learning_rate": 3.86e-06,
+      "loss": 0.5135,
+      "step": 386
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6605297036299905,
+      "learning_rate": 3.87e-06,
+      "loss": 0.5576,
+      "step": 387
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6988172021818264,
+      "learning_rate": 3.88e-06,
+      "loss": 0.5281,
+      "step": 388
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6511359929973235,
+      "learning_rate": 3.89e-06,
+      "loss": 0.5148,
+      "step": 389
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.74475476203548,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 0.5313,
+      "step": 390
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6768085651878416,
+      "learning_rate": 3.910000000000001e-06,
+      "loss": 0.5289,
+      "step": 391
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6590289960576611,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 0.526,
+      "step": 392
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6294447240216576,
+      "learning_rate": 3.9300000000000005e-06,
+      "loss": 0.5066,
+      "step": 393
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6795503573063048,
+      "learning_rate": 3.94e-06,
+      "loss": 0.5265,
+      "step": 394
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6327421219484214,
+      "learning_rate": 3.95e-06,
+      "loss": 0.5262,
+      "step": 395
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6702321778478743,
+      "learning_rate": 3.96e-06,
+      "loss": 0.5521,
+      "step": 396
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6601601082950667,
+      "learning_rate": 3.97e-06,
+      "loss": 0.5511,
+      "step": 397
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.672312807049255,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.581,
+      "step": 398
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.70887535443515,
+      "learning_rate": 3.990000000000001e-06,
+      "loss": 0.5693,
+      "step": 399
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.674197659542202,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.5304,
+      "step": 400
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6367460409014842,
+      "learning_rate": 4.0100000000000006e-06,
+      "loss": 0.5081,
+      "step": 401
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6857813275083907,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 0.4973,
+      "step": 402
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6556472016511119,
+      "learning_rate": 4.03e-06,
+      "loss": 0.5389,
+      "step": 403
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6985422078815098,
+      "learning_rate": 4.04e-06,
+      "loss": 0.5289,
+      "step": 404
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6649519288148354,
+      "learning_rate": 4.05e-06,
+      "loss": 0.5379,
+      "step": 405
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6263707838794045,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 0.5421,
+      "step": 406
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.7664906236471626,
+      "learning_rate": 4.07e-06,
+      "loss": 0.5535,
+      "step": 407
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6553639230258015,
+      "learning_rate": 4.08e-06,
+      "loss": 0.5349,
+      "step": 408
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.6750627011135212,
+      "learning_rate": 4.09e-06,
+      "loss": 0.5453,
+      "step": 409
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6577154298564163,
+      "learning_rate": 4.1e-06,
+      "loss": 0.5082,
+      "step": 410
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6630263227345423,
+      "learning_rate": 4.1100000000000005e-06,
+      "loss": 0.5856,
+      "step": 411
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6574426137463745,
+      "learning_rate": 4.12e-06,
+      "loss": 0.5384,
+      "step": 412
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6988289040575354,
+      "learning_rate": 4.13e-06,
+      "loss": 0.5399,
+      "step": 413
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6306882029774344,
+      "learning_rate": 4.14e-06,
+      "loss": 0.5336,
+      "step": 414
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7179639205026864,
+      "learning_rate": 4.15e-06,
+      "loss": 0.518,
+      "step": 415
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6516513551525797,
+      "learning_rate": 4.16e-06,
+      "loss": 0.5257,
+      "step": 416
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6601218324564917,
+      "learning_rate": 4.17e-06,
+      "loss": 0.5289,
+      "step": 417
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7211814488287548,
+      "learning_rate": 4.18e-06,
+      "loss": 0.5276,
+      "step": 418
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6699099078689619,
+      "learning_rate": 4.1900000000000005e-06,
+      "loss": 0.5244,
+      "step": 419
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6916838990628401,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 0.5181,
+      "step": 420
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6545058846876347,
+      "learning_rate": 4.21e-06,
+      "loss": 0.5475,
+      "step": 421
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7236179142807477,
+      "learning_rate": 4.22e-06,
+      "loss": 0.5049,
+      "step": 422
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6610815551752731,
+      "learning_rate": 4.23e-06,
+      "loss": 0.5463,
+      "step": 423
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6821422605987203,
+      "learning_rate": 4.24e-06,
+      "loss": 0.5307,
+      "step": 424
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6431222169441803,
+      "learning_rate": 4.25e-06,
+      "loss": 0.521,
+      "step": 425
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6952795202176767,
+      "learning_rate": 4.26e-06,
+      "loss": 0.514,
+      "step": 426
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 0.5265827775001526,
+      "eval_runtime": 6909.3691,
+      "eval_samples_per_second": 42.028,
+      "eval_steps_per_second": 2.101,
+      "step": 426
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6134347482653176,
+      "learning_rate": 4.270000000000001e-06,
+      "loss": 0.5251,
+      "step": 427
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7147159318510563,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 0.5309,
+      "step": 428
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7026933031500548,
+      "learning_rate": 4.2900000000000004e-06,
+      "loss": 0.5336,
+      "step": 429
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7005957676093472,
+      "learning_rate": 4.3e-06,
+      "loss": 0.5518,
+      "step": 430
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.697347284426043,
+      "learning_rate": 4.31e-06,
+      "loss": 0.5222,
+      "step": 431
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6618459969675489,
+      "learning_rate": 4.32e-06,
+      "loss": 0.5077,
+      "step": 432
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.6436931259589193,
+      "learning_rate": 4.33e-06,
+      "loss": 0.5155,
+      "step": 433
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6589831621019552,
+      "learning_rate": 4.34e-06,
+      "loss": 0.5293,
+      "step": 434
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6487932842171147,
+      "learning_rate": 4.350000000000001e-06,
+      "loss": 0.5234,
+      "step": 435
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6388950573406754,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 0.5247,
+      "step": 436
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6484480185597394,
+      "learning_rate": 4.3700000000000005e-06,
+      "loss": 0.5493,
+      "step": 437
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6656552397893157,
+      "learning_rate": 4.38e-06,
+      "loss": 0.5122,
+      "step": 438
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6100914831240443,
+      "learning_rate": 4.39e-06,
+      "loss": 0.5223,
+      "step": 439
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6806948048082133,
+      "learning_rate": 4.4e-06,
+      "loss": 0.5366,
+      "step": 440
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6792634934940536,
+      "learning_rate": 4.41e-06,
+      "loss": 0.5371,
+      "step": 441
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6163101784317894,
+      "learning_rate": 4.42e-06,
+      "loss": 0.5001,
+      "step": 442
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6711982512846896,
+      "learning_rate": 4.430000000000001e-06,
+      "loss": 0.5431,
+      "step": 443
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6656044261856343,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 0.4961,
+      "step": 444
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6090638329534794,
+      "learning_rate": 4.450000000000001e-06,
+      "loss": 0.5124,
+      "step": 445
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6642264166031999,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 0.5401,
+      "step": 446
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6146557964679851,
+      "learning_rate": 4.47e-06,
+      "loss": 0.521,
+      "step": 447
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6540634116911361,
+      "learning_rate": 4.48e-06,
+      "loss": 0.5472,
+      "step": 448
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6394159843807653,
+      "learning_rate": 4.49e-06,
+      "loss": 0.4984,
+      "step": 449
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6320717472264333,
+      "learning_rate": 4.5e-06,
+      "loss": 0.539,
+      "step": 450
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6335986050650263,
+      "learning_rate": 4.510000000000001e-06,
+      "loss": 0.5279,
+      "step": 451
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6238922538054539,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 0.5632,
+      "step": 452
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6628688965370182,
+      "learning_rate": 4.530000000000001e-06,
+      "loss": 0.5417,
+      "step": 453
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6559295875868244,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 0.5127,
+      "step": 454
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6429887683654244,
+      "learning_rate": 4.5500000000000005e-06,
+      "loss": 0.5116,
+      "step": 455
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6925428894756206,
+      "learning_rate": 4.56e-06,
+      "loss": 0.5264,
+      "step": 456
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6604588640436396,
+      "learning_rate": 4.57e-06,
+      "loss": 0.5284,
+      "step": 457
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6392985738009899,
+      "learning_rate": 4.58e-06,
+      "loss": 0.5188,
+      "step": 458
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6765577462503485,
+      "learning_rate": 4.590000000000001e-06,
+      "loss": 0.514,
+      "step": 459
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8495240761386383,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 0.5268,
+      "step": 460
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6836958527754253,
+      "learning_rate": 4.610000000000001e-06,
+      "loss": 0.5305,
+      "step": 461
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6502080642686838,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 0.5178,
+      "step": 462
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6342249164223178,
+      "learning_rate": 4.6300000000000006e-06,
+      "loss": 0.5217,
+      "step": 463
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6777938665629211,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 0.534,
+      "step": 464
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7062865741492855,
+      "learning_rate": 4.65e-06,
+      "loss": 0.5077,
+      "step": 465
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6383144907337746,
+      "learning_rate": 4.66e-06,
+      "loss": 0.5134,
+      "step": 466
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6730064655141996,
+      "learning_rate": 4.670000000000001e-06,
+      "loss": 0.552,
+      "step": 467
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6394012481326041,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 0.4908,
+      "step": 468
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7061191203371422,
+      "learning_rate": 4.69e-06,
+      "loss": 0.5354,
+      "step": 469
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6289046687203929,
+      "learning_rate": 4.7e-06,
+      "loss": 0.524,
+      "step": 470
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6880276584214321,
+      "learning_rate": 4.71e-06,
+      "loss": 0.5317,
+      "step": 471
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6626736935804997,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 0.5293,
+      "step": 472
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6704211606772497,
+      "learning_rate": 4.7300000000000005e-06,
+      "loss": 0.5135,
+      "step": 473
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6671310077095398,
+      "learning_rate": 4.74e-06,
+      "loss": 0.5256,
+      "step": 474
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.637683155549069,
+      "learning_rate": 4.75e-06,
+      "loss": 0.5113,
+      "step": 475
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7243606567809928,
+      "learning_rate": 4.76e-06,
+      "loss": 0.5222,
+      "step": 476
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6898226976279818,
+      "learning_rate": 4.77e-06,
+      "loss": 0.5434,
+      "step": 477
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7102119430718769,
+      "learning_rate": 4.78e-06,
+      "loss": 0.5584,
+      "step": 478
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.6357099843026002,
+      "learning_rate": 4.79e-06,
+      "loss": 0.5377,
+      "step": 479
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7078871919110719,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.51,
+      "step": 480
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7004243288260213,
+      "learning_rate": 4.8100000000000005e-06,
+      "loss": 0.5098,
+      "step": 481
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7218886636410997,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 0.5473,
+      "step": 482
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6908387236569687,
+      "learning_rate": 4.83e-06,
+      "loss": 0.5155,
+      "step": 483
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7042724565499465,
+      "learning_rate": 4.84e-06,
+      "loss": 0.5152,
+      "step": 484
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7088947897764216,
+      "learning_rate": 4.85e-06,
+      "loss": 0.5417,
+      "step": 485
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6279155418925114,
+      "learning_rate": 4.86e-06,
+      "loss": 0.5349,
+      "step": 486
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7207841117522383,
+      "learning_rate": 4.87e-06,
+      "loss": 0.5261,
+      "step": 487
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6652618430741107,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 0.5355,
+      "step": 488
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6296330144736813,
+      "learning_rate": 4.890000000000001e-06,
+      "loss": 0.5302,
+      "step": 489
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6197754222224622,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 0.4987,
+      "step": 490
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.642766778231823,
+      "learning_rate": 4.9100000000000004e-06,
+      "loss": 0.5247,
+      "step": 491
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.643792980820932,
+      "learning_rate": 4.92e-06,
+      "loss": 0.5347,
+      "step": 492
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.625452787303321,
+      "learning_rate": 4.93e-06,
+      "loss": 0.499,
+      "step": 493
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.5964088928926613,
+      "learning_rate": 4.94e-06,
+      "loss": 0.5407,
+      "step": 494
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6500032443873983,
+      "learning_rate": 4.95e-06,
+      "loss": 0.5464,
+      "step": 495
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6460792843444659,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 0.4949,
+      "step": 496
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7395445647719444,
+      "learning_rate": 4.970000000000001e-06,
+      "loss": 0.5291,
+      "step": 497
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 0.5214883685112,
+      "eval_runtime": 6920.669,
+      "eval_samples_per_second": 41.96,
+      "eval_steps_per_second": 2.098,
+      "step": 497
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6537595312676089,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.5176,
+      "step": 498
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6463304730768601,
+      "learning_rate": 4.9900000000000005e-06,
+      "loss": 0.5206,
+      "step": 499
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6973321215012371,
+      "learning_rate": 5e-06,
+      "loss": 0.5404,
+      "step": 500
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6779339517089825,
+      "learning_rate": 4.999999293914693e-06,
+      "loss": 0.5073,
+      "step": 501
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6916291409636109,
+      "learning_rate": 4.99999717565917e-06,
+      "loss": 0.5277,
+      "step": 502
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6019781233779008,
+      "learning_rate": 4.999993645234629e-06,
+      "loss": 0.488,
+      "step": 503
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6767553900584968,
+      "learning_rate": 4.999988702643063e-06,
+      "loss": 0.5517,
+      "step": 504
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6399171255962077,
+      "learning_rate": 4.999982347887264e-06,
+      "loss": 0.5363,
+      "step": 505
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6957517244632938,
+      "learning_rate": 4.999974580970822e-06,
+      "loss": 0.5425,
+      "step": 506
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6303242233074258,
+      "learning_rate": 4.999965401898124e-06,
+      "loss": 0.5359,
+      "step": 507
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6857079786662916,
+      "learning_rate": 4.999954810674355e-06,
+      "loss": 0.4941,
+      "step": 508
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6875532162149117,
+      "learning_rate": 4.999942807305497e-06,
+      "loss": 0.5345,
+      "step": 509
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6275865235654622,
+      "learning_rate": 4.9999293917983325e-06,
+      "loss": 0.5099,
+      "step": 510
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7718521188057177,
+      "learning_rate": 4.999914564160437e-06,
+      "loss": 0.5181,
+      "step": 511
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6534743581262111,
+      "learning_rate": 4.999898324400187e-06,
+      "loss": 0.5126,
+      "step": 512
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.674911757416486,
+      "learning_rate": 4.999880672526757e-06,
+      "loss": 0.5368,
+      "step": 513
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.764101502643344,
+      "learning_rate": 4.999861608550116e-06,
+      "loss": 0.5105,
+      "step": 514
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6607294039970933,
+      "learning_rate": 4.999841132481035e-06,
+      "loss": 0.5231,
+      "step": 515
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7724967744040852,
+      "learning_rate": 4.999819244331078e-06,
+      "loss": 0.5489,
+      "step": 516
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6730774325770473,
+      "learning_rate": 4.99979594411261e-06,
+      "loss": 0.5326,
+      "step": 517
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7136063574644014,
+      "learning_rate": 4.999771231838792e-06,
+      "loss": 0.5144,
+      "step": 518
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6477743762407223,
+      "learning_rate": 4.999745107523583e-06,
+      "loss": 0.5111,
+      "step": 519
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6916139988086453,
+      "learning_rate": 4.999717571181742e-06,
+      "loss": 0.529,
+      "step": 520
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6564109960699416,
+      "learning_rate": 4.999688622828821e-06,
+      "loss": 0.5101,
+      "step": 521
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6610425029900775,
+      "learning_rate": 4.999658262481173e-06,
+      "loss": 0.5095,
+      "step": 522
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6062090231312307,
+      "learning_rate": 4.999626490155947e-06,
+      "loss": 0.5072,
+      "step": 523
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.629980273760487,
+      "learning_rate": 4.999593305871091e-06,
+      "loss": 0.5382,
+      "step": 524
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6597246057010635,
+      "learning_rate": 4.999558709645349e-06,
+      "loss": 0.5307,
+      "step": 525
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.636745442438904,
+      "learning_rate": 4.999522701498263e-06,
+      "loss": 0.5302,
+      "step": 526
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6324462143762214,
+      "learning_rate": 4.999485281450174e-06,
+      "loss": 0.4996,
+      "step": 527
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6069558558098991,
+      "learning_rate": 4.99944644952222e-06,
+      "loss": 0.5058,
+      "step": 528
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6006305147544736,
+      "learning_rate": 4.999406205736334e-06,
+      "loss": 0.5218,
+      "step": 529
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6354602393667919,
+      "learning_rate": 4.9993645501152485e-06,
+      "loss": 0.498,
+      "step": 530
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6299647790639523,
+      "learning_rate": 4.999321482682495e-06,
+      "loss": 0.5435,
+      "step": 531
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5702190944475204,
+      "learning_rate": 4.9992770034624e-06,
+      "loss": 0.5259,
+      "step": 532
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6629005033715545,
+      "learning_rate": 4.999231112480088e-06,
+      "loss": 0.5247,
+      "step": 533
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6161756194165398,
+      "learning_rate": 4.999183809761481e-06,
+      "loss": 0.5233,
+      "step": 534
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6143755401649461,
+      "learning_rate": 4.999135095333301e-06,
+      "loss": 0.4983,
+      "step": 535
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6374604849318725,
+      "learning_rate": 4.999084969223064e-06,
+      "loss": 0.5407,
+      "step": 536
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6489680492749198,
+      "learning_rate": 4.999033431459084e-06,
+      "loss": 0.5216,
+      "step": 537
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6128330475540421,
+      "learning_rate": 4.998980482070473e-06,
+      "loss": 0.4956,
+      "step": 538
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6084501316124987,
+      "learning_rate": 4.998926121087142e-06,
+      "loss": 0.5105,
+      "step": 539
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6718312705515297,
+      "learning_rate": 4.998870348539797e-06,
+      "loss": 0.4978,
+      "step": 540
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.5917495579968527,
+      "learning_rate": 4.998813164459942e-06,
+      "loss": 0.5005,
+      "step": 541
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7002888012401992,
+      "learning_rate": 4.9987545688798765e-06,
+      "loss": 0.5158,
+      "step": 542
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6215562316702827,
+      "learning_rate": 4.998694561832703e-06,
+      "loss": 0.5063,
+      "step": 543
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6463810547164883,
+      "learning_rate": 4.998633143352315e-06,
+      "loss": 0.5216,
+      "step": 544
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6607126999162819,
+      "learning_rate": 4.998570313473408e-06,
+      "loss": 0.5176,
+      "step": 545
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6896680501018091,
+      "learning_rate": 4.998506072231469e-06,
+      "loss": 0.5142,
+      "step": 546
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.64482281681983,
+      "learning_rate": 4.99844041966279e-06,
+      "loss": 0.5223,
+      "step": 547
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6804316388777557,
+      "learning_rate": 4.998373355804454e-06,
+      "loss": 0.512,
+      "step": 548
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6799252023969709,
+      "learning_rate": 4.998304880694342e-06,
+      "loss": 0.4962,
+      "step": 549
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.637365085524163,
+      "learning_rate": 4.998234994371135e-06,
+      "loss": 0.5234,
+      "step": 550
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7036436945946638,
+      "learning_rate": 4.99816369687431e-06,
+      "loss": 0.5141,
+      "step": 551
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6240264386156906,
+      "learning_rate": 4.99809098824414e-06,
+      "loss": 0.5365,
+      "step": 552
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7403699319338876,
+      "learning_rate": 4.998016868521695e-06,
+      "loss": 0.5326,
+      "step": 553
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6543848290779426,
+      "learning_rate": 4.997941337748845e-06,
+      "loss": 0.5146,
+      "step": 554
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7042287590734285,
+      "learning_rate": 4.997864395968252e-06,
+      "loss": 0.5259,
+      "step": 555
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6574049565188761,
+      "learning_rate": 4.997786043223381e-06,
+      "loss": 0.5137,
+      "step": 556
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6165676577046679,
+      "learning_rate": 4.9977062795584895e-06,
+      "loss": 0.5473,
+      "step": 557
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7044566948417562,
+      "learning_rate": 4.997625105018634e-06,
+      "loss": 0.5067,
+      "step": 558
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6447891576595195,
+      "learning_rate": 4.9975425196496656e-06,
+      "loss": 0.5117,
+      "step": 559
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6083915401749171,
+      "learning_rate": 4.997458523498236e-06,
+      "loss": 0.4752,
+      "step": 560
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6990442708117952,
+      "learning_rate": 4.997373116611792e-06,
+      "loss": 0.5145,
+      "step": 561
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6792107396650091,
+      "learning_rate": 4.997286299038576e-06,
+      "loss": 0.5143,
+      "step": 562
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6274185410270217,
+      "learning_rate": 4.997198070827629e-06,
+      "loss": 0.5079,
+      "step": 563
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7317503385718371,
+      "learning_rate": 4.99710843202879e-06,
+      "loss": 0.5058,
+      "step": 564
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6602507519356526,
+      "learning_rate": 4.99701738269269e-06,
+      "loss": 0.5269,
+      "step": 565
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7158785502037818,
+      "learning_rate": 4.9969249228707625e-06,
+      "loss": 0.5449,
+      "step": 566
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6678036728294128,
+      "learning_rate": 4.996831052615234e-06,
+      "loss": 0.4996,
+      "step": 567
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6635048447906537,
+      "learning_rate": 4.996735771979129e-06,
+      "loss": 0.5117,
+      "step": 568
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 0.517206609249115,
+      "eval_runtime": 6917.186,
+      "eval_samples_per_second": 41.981,
+      "eval_steps_per_second": 2.099,
+      "step": 568
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6631113641261142,
+      "learning_rate": 4.996639081016268e-06,
+      "loss": 0.5103,
+      "step": 569
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6279444558937123,
+      "learning_rate": 4.996540979781269e-06,
+      "loss": 0.5397,
+      "step": 570
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6876493931054276,
+      "learning_rate": 4.996441468329547e-06,
+      "loss": 0.521,
+      "step": 571
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6664423383196236,
+      "learning_rate": 4.996340546717312e-06,
+      "loss": 0.5449,
+      "step": 572
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6101029466977869,
+      "learning_rate": 4.996238215001571e-06,
+      "loss": 0.5443,
+      "step": 573
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6622002597747987,
+      "learning_rate": 4.99613447324013e-06,
+      "loss": 0.5245,
+      "step": 574
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.60547382309911,
+      "learning_rate": 4.996029321491587e-06,
+      "loss": 0.5101,
+      "step": 575
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6685376758007866,
+      "learning_rate": 4.9959227598153395e-06,
+      "loss": 0.5226,
+      "step": 576
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6325127114162676,
+      "learning_rate": 4.995814788271582e-06,
+      "loss": 0.525,
+      "step": 577
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6462009207619525,
+      "learning_rate": 4.995705406921303e-06,
+      "loss": 0.5181,
+      "step": 578
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6514032258227322,
+      "learning_rate": 4.995594615826289e-06,
+      "loss": 0.5225,
+      "step": 579
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6013786935845273,
+      "learning_rate": 4.995482415049123e-06,
+      "loss": 0.5105,
+      "step": 580
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6722327524821109,
+      "learning_rate": 4.995368804653182e-06,
+      "loss": 0.5088,
+      "step": 581
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6486288344146888,
+      "learning_rate": 4.995253784702643e-06,
+      "loss": 0.4924,
+      "step": 582
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.609052561499619,
+      "learning_rate": 4.995137355262475e-06,
+      "loss": 0.5091,
+      "step": 583
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6149869407753081,
+      "learning_rate": 4.995019516398447e-06,
+      "loss": 0.522,
+      "step": 584
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.649711066334699,
+      "learning_rate": 4.994900268177121e-06,
+      "loss": 0.5067,
+      "step": 585
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6492055338072015,
+      "learning_rate": 4.994779610665858e-06,
+      "loss": 0.5531,
+      "step": 586
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6055259963194052,
+      "learning_rate": 4.9946575439328124e-06,
+      "loss": 0.4979,
+      "step": 587
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5966649511641912,
+      "learning_rate": 4.994534068046936e-06,
+      "loss": 0.511,
+      "step": 588
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6406292315887588,
+      "learning_rate": 4.994409183077979e-06,
+      "loss": 0.5423,
+      "step": 589
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6273737605531128,
+      "learning_rate": 4.99428288909648e-06,
+      "loss": 0.5,
+      "step": 590
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6106470129106254,
+      "learning_rate": 4.994155186173782e-06,
+      "loss": 0.5107,
+      "step": 591
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6047596492090179,
+      "learning_rate": 4.994026074382019e-06,
+      "loss": 0.5267,
+      "step": 592
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5986927472135537,
+      "learning_rate": 4.993895553794123e-06,
+      "loss": 0.5025,
+      "step": 593
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6429583180993461,
+      "learning_rate": 4.993763624483821e-06,
+      "loss": 0.5158,
+      "step": 594
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.5881062972343085,
+      "learning_rate": 4.993630286525634e-06,
+      "loss": 0.5061,
+      "step": 595
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6368950879373987,
+      "learning_rate": 4.993495539994882e-06,
+      "loss": 0.5203,
+      "step": 596
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6716881530561206,
+      "learning_rate": 4.99335938496768e-06,
+      "loss": 0.5228,
+      "step": 597
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6023179095017582,
+      "learning_rate": 4.993221821520935e-06,
+      "loss": 0.5349,
+      "step": 598
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6118755989606954,
+      "learning_rate": 4.993082849732353e-06,
+      "loss": 0.5071,
+      "step": 599
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6160351740855943,
+      "learning_rate": 4.992942469680437e-06,
+      "loss": 0.4796,
+      "step": 600
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6291124034547457,
+      "learning_rate": 4.99280068144448e-06,
+      "loss": 0.5545,
+      "step": 601
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6534616170585308,
+      "learning_rate": 4.992657485104575e-06,
+      "loss": 0.5358,
+      "step": 602
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6264541141447443,
+      "learning_rate": 4.99251288074161e-06,
+      "loss": 0.5292,
+      "step": 603
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6329545583821921,
+      "learning_rate": 4.992366868437266e-06,
+      "loss": 0.5032,
+      "step": 604
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6262380789197445,
+      "learning_rate": 4.992219448274022e-06,
+      "loss": 0.5178,
+      "step": 605
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6003543172945967,
+      "learning_rate": 4.9920706203351495e-06,
+      "loss": 0.5037,
+      "step": 606
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6614794995436638,
+      "learning_rate": 4.9919203847047185e-06,
+      "loss": 0.5391,
+      "step": 607
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6276911443763706,
+      "learning_rate": 4.99176874146759e-06,
+      "loss": 0.4837,
+      "step": 608
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6140831194461105,
+      "learning_rate": 4.9916156907094246e-06,
+      "loss": 0.4944,
+      "step": 609
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6520299553825838,
+      "learning_rate": 4.991461232516675e-06,
+      "loss": 0.4984,
+      "step": 610
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6785930087303712,
+      "learning_rate": 4.99130536697659e-06,
+      "loss": 0.5289,
+      "step": 611
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6751246335138484,
+      "learning_rate": 4.991148094177212e-06,
+      "loss": 0.517,
+      "step": 612
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.7063631237307217,
+      "learning_rate": 4.990989414207381e-06,
+      "loss": 0.4747,
+      "step": 613
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.633915838850146,
+      "learning_rate": 4.990829327156729e-06,
+      "loss": 0.5182,
+      "step": 614
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6346239919506149,
+      "learning_rate": 4.990667833115684e-06,
+      "loss": 0.4991,
+      "step": 615
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6922170062238953,
+      "learning_rate": 4.990504932175471e-06,
+      "loss": 0.4997,
+      "step": 616
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6497250780582481,
+      "learning_rate": 4.990340624428105e-06,
+      "loss": 0.5025,
+      "step": 617
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6480587404624075,
+      "learning_rate": 4.990174909966399e-06,
+      "loss": 0.5164,
+      "step": 618
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.5647887869895122,
+      "learning_rate": 4.9900077888839606e-06,
+      "loss": 0.4974,
+      "step": 619
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6157405333654503,
+      "learning_rate": 4.989839261275191e-06,
+      "loss": 0.5119,
+      "step": 620
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6350389596733836,
+      "learning_rate": 4.989669327235285e-06,
+      "loss": 0.5252,
+      "step": 621
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.584442178782563,
+      "learning_rate": 4.989497986860234e-06,
+      "loss": 0.5043,
+      "step": 622
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6196796070950197,
+      "learning_rate": 4.989325240246823e-06,
+      "loss": 0.5176,
+      "step": 623
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6015855990105409,
+      "learning_rate": 4.98915108749263e-06,
+      "loss": 0.5065,
+      "step": 624
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5944351660408316,
+      "learning_rate": 4.988975528696028e-06,
+      "loss": 0.5027,
+      "step": 625
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5799955435183154,
+      "learning_rate": 4.988798563956186e-06,
+      "loss": 0.4897,
+      "step": 626
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6252189457639663,
+      "learning_rate": 4.988620193373066e-06,
+      "loss": 0.5185,
+      "step": 627
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6091789072393431,
+      "learning_rate": 4.988440417047424e-06,
+      "loss": 0.5364,
+      "step": 628
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5991774516054336,
+      "learning_rate": 4.988259235080807e-06,
+      "loss": 0.5218,
+      "step": 629
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6004079133450607,
+      "learning_rate": 4.988076647575562e-06,
+      "loss": 0.5346,
+      "step": 630
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6055525535102732,
+      "learning_rate": 4.987892654634825e-06,
+      "loss": 0.4794,
+      "step": 631
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6231744592672184,
+      "learning_rate": 4.987707256362529e-06,
+      "loss": 0.5332,
+      "step": 632
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5943003024735096,
+      "learning_rate": 4.9875204528633995e-06,
+      "loss": 0.4958,
+      "step": 633
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6429217088606017,
+      "learning_rate": 4.987332244242955e-06,
+      "loss": 0.5049,
+      "step": 634
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6474023001802436,
+      "learning_rate": 4.98714263060751e-06,
+      "loss": 0.4769,
+      "step": 635
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6404945389649939,
+      "learning_rate": 4.9869516120641705e-06,
+      "loss": 0.4947,
+      "step": 636
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6640610934148168,
+      "learning_rate": 4.986759188720836e-06,
+      "loss": 0.5258,
+      "step": 637
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6232010288109409,
+      "learning_rate": 4.986565360686201e-06,
+      "loss": 0.5189,
+      "step": 638
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6143458681147979,
+      "learning_rate": 4.9863701280697535e-06,
+      "loss": 0.4986,
+      "step": 639
+    },
+    {
+      "epoch": 0.27,
+      "eval_loss": 0.5126649141311646,
+      "eval_runtime": 6910.7711,
+      "eval_samples_per_second": 42.02,
+      "eval_steps_per_second": 2.101,
+      "step": 639
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6231312288980349,
+      "learning_rate": 4.986173490981773e-06,
+      "loss": 0.5218,
+      "step": 640
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6521330675888304,
+      "learning_rate": 4.985975449533335e-06,
+      "loss": 0.4984,
+      "step": 641
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6192181654765082,
+      "learning_rate": 4.9857760038363045e-06,
+      "loss": 0.5257,
+      "step": 642
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.6486948358491045,
+      "learning_rate": 4.9855751540033446e-06,
+      "loss": 0.5147,
+      "step": 643
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5969342371638264,
+      "learning_rate": 4.985372900147907e-06,
+      "loss": 0.5185,
+      "step": 644
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6320704370838177,
+      "learning_rate": 4.9851692423842406e-06,
+      "loss": 0.5383,
+      "step": 645
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5966997098167783,
+      "learning_rate": 4.984964180827383e-06,
+      "loss": 0.5044,
+      "step": 646
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6109011858617819,
+      "learning_rate": 4.984757715593168e-06,
+      "loss": 0.5091,
+      "step": 647
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.653805713393856,
+      "learning_rate": 4.984549846798221e-06,
+      "loss": 0.5161,
+      "step": 648
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6199215842335395,
+      "learning_rate": 4.984340574559961e-06,
+      "loss": 0.5152,
+      "step": 649
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6068098393695217,
+      "learning_rate": 4.984129898996599e-06,
+      "loss": 0.5209,
+      "step": 650
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5796411094349677,
+      "learning_rate": 4.9839178202271375e-06,
+      "loss": 0.502,
+      "step": 651
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6053468178000233,
+      "learning_rate": 4.983704338371375e-06,
+      "loss": 0.5283,
+      "step": 652
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6156634669849913,
+      "learning_rate": 4.983489453549901e-06,
+      "loss": 0.4947,
+      "step": 653
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6241545798810763,
+      "learning_rate": 4.983273165884096e-06,
+      "loss": 0.5176,
+      "step": 654
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5895664409761231,
+      "learning_rate": 4.983055475496134e-06,
+      "loss": 0.5105,
+      "step": 655
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5935426249245771,
+      "learning_rate": 4.982836382508981e-06,
+      "loss": 0.5,
+      "step": 656
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6233645788446041,
+      "learning_rate": 4.9826158870463955e-06,
+      "loss": 0.5099,
+      "step": 657
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5991332439052809,
+      "learning_rate": 4.982393989232931e-06,
+      "loss": 0.4924,
+      "step": 658
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5880708311480832,
+      "learning_rate": 4.982170689193927e-06,
+      "loss": 0.4978,
+      "step": 659
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.627672707017739,
+      "learning_rate": 4.981945987055521e-06,
+      "loss": 0.5077,
+      "step": 660
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6040524425886902,
+      "learning_rate": 4.981719882944639e-06,
+      "loss": 0.493,
+      "step": 661
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5875699138537736,
+      "learning_rate": 4.981492376989001e-06,
+      "loss": 0.506,
+      "step": 662
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6071888999132754,
+      "learning_rate": 4.981263469317116e-06,
+      "loss": 0.4954,
+      "step": 663
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6343036259474204,
+      "learning_rate": 4.981033160058289e-06,
+      "loss": 0.5302,
+      "step": 664
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6270477353053345,
+      "learning_rate": 4.9808014493426124e-06,
+      "loss": 0.4676,
+      "step": 665
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5957359905347063,
+      "learning_rate": 4.9805683373009746e-06,
+      "loss": 0.5108,
+      "step": 666
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6267707786737462,
+      "learning_rate": 4.980333824065051e-06,
+      "loss": 0.5293,
+      "step": 667
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.577921156213165,
+      "learning_rate": 4.980097909767311e-06,
+      "loss": 0.5115,
+      "step": 668
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6306888663442499,
+      "learning_rate": 4.9798605945410156e-06,
+      "loss": 0.5123,
+      "step": 669
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5963346695333823,
+      "learning_rate": 4.979621878520217e-06,
+      "loss": 0.5026,
+      "step": 670
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.590690163604036,
+      "learning_rate": 4.979381761839757e-06,
+      "loss": 0.503,
+      "step": 671
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6410334373270873,
+      "learning_rate": 4.979140244635271e-06,
+      "loss": 0.5116,
+      "step": 672
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6105251339526676,
+      "learning_rate": 4.978897327043185e-06,
+      "loss": 0.503,
+      "step": 673
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.611118980970183,
+      "learning_rate": 4.978653009200713e-06,
+      "loss": 0.5005,
+      "step": 674
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.61089260561556,
+      "learning_rate": 4.978407291245866e-06,
+      "loss": 0.4745,
+      "step": 675
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6296501736035641,
+      "learning_rate": 4.978160173317439e-06,
+      "loss": 0.5413,
+      "step": 676
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5754436830327142,
+      "learning_rate": 4.977911655555022e-06,
+      "loss": 0.5258,
+      "step": 677
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6080486185142544,
+      "learning_rate": 4.977661738098996e-06,
+      "loss": 0.5022,
+      "step": 678
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6095469549060126,
+      "learning_rate": 4.97741042109053e-06,
+      "loss": 0.5236,
+      "step": 679
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6246514201425122,
+      "learning_rate": 4.977157704671585e-06,
+      "loss": 0.4948,
+      "step": 680
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6043169735120094,
+      "learning_rate": 4.976903588984913e-06,
+      "loss": 0.5174,
+      "step": 681
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5782841931005864,
+      "learning_rate": 4.976648074174056e-06,
+      "loss": 0.4935,
+      "step": 682
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6172869080905491,
+      "learning_rate": 4.976391160383347e-06,
+      "loss": 0.5068,
+      "step": 683
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5986800192103562,
+      "learning_rate": 4.976132847757906e-06,
+      "loss": 0.5293,
+      "step": 684
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5997059239072083,
+      "learning_rate": 4.975873136443649e-06,
+      "loss": 0.5311,
+      "step": 685
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.586987259487267,
+      "learning_rate": 4.9756120265872755e-06,
+      "loss": 0.513,
+      "step": 686
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.6278820584072181,
+      "learning_rate": 4.97534951833628e-06,
+      "loss": 0.5278,
+      "step": 687
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5792346089546118,
+      "learning_rate": 4.975085611838944e-06,
+      "loss": 0.5048,
+      "step": 688
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5926608888246347,
+      "learning_rate": 4.974820307244341e-06,
+      "loss": 0.4979,
+      "step": 689
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5905921659974432,
+      "learning_rate": 4.974553604702332e-06,
+      "loss": 0.4977,
+      "step": 690
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6604840701536976,
+      "learning_rate": 4.974285504363569e-06,
+      "loss": 0.504,
+      "step": 691
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6243772158616288,
+      "learning_rate": 4.974016006379495e-06,
+      "loss": 0.5288,
+      "step": 692
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6240076780093031,
+      "learning_rate": 4.973745110902339e-06,
+      "loss": 0.5189,
+      "step": 693
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6107829102505847,
+      "learning_rate": 4.973472818085122e-06,
+      "loss": 0.5359,
+      "step": 694
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6345275208943483,
+      "learning_rate": 4.9731991280816534e-06,
+      "loss": 0.5096,
+      "step": 695
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6136833545759933,
+      "learning_rate": 4.9729240410465315e-06,
+      "loss": 0.5048,
+      "step": 696
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6405628475705748,
+      "learning_rate": 4.972647557135146e-06,
+      "loss": 0.5234,
+      "step": 697
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5807229777227344,
+      "learning_rate": 4.972369676503672e-06,
+      "loss": 0.516,
+      "step": 698
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6808588571301922,
+      "learning_rate": 4.972090399309075e-06,
+      "loss": 0.4935,
+      "step": 699
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.598882478841374,
+      "learning_rate": 4.971809725709112e-06,
+      "loss": 0.5167,
+      "step": 700
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6597647873322787,
+      "learning_rate": 4.971527655862325e-06,
+      "loss": 0.4986,
+      "step": 701
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6205471007307612,
+      "learning_rate": 4.9712441899280475e-06,
+      "loss": 0.481,
+      "step": 702
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6501750674785319,
+      "learning_rate": 4.970959328066399e-06,
+      "loss": 0.5141,
+      "step": 703
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6389770637630756,
+      "learning_rate": 4.97067307043829e-06,
+      "loss": 0.5045,
+      "step": 704
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6071448493335176,
+      "learning_rate": 4.970385417205418e-06,
+      "loss": 0.5231,
+      "step": 705
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6307400799413926,
+      "learning_rate": 4.9700963685302685e-06,
+      "loss": 0.5132,
+      "step": 706
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6345555266447565,
+      "learning_rate": 4.969805924576116e-06,
+      "loss": 0.5165,
+      "step": 707
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.5992064057317342,
+      "learning_rate": 4.969514085507025e-06,
+      "loss": 0.4933,
+      "step": 708
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6327334322592149,
+      "learning_rate": 4.9692208514878445e-06,
+      "loss": 0.4959,
+      "step": 709
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6250597389392478,
+      "learning_rate": 4.968926222684213e-06,
+      "loss": 0.5097,
+      "step": 710
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 0.5087815523147583,
+      "eval_runtime": 6910.3283,
+      "eval_samples_per_second": 42.022,
+      "eval_steps_per_second": 2.101,
+      "step": 710
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6069692793366778,
+      "learning_rate": 4.9686301992625575e-06,
+      "loss": 0.4892,
+      "step": 711
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6516138876780239,
+      "learning_rate": 4.968332781390092e-06,
+      "loss": 0.5303,
+      "step": 712
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6143846915255134,
+      "learning_rate": 4.968033969234818e-06,
+      "loss": 0.4919,
+      "step": 713
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5981365954582023,
+      "learning_rate": 4.967733762965526e-06,
+      "loss": 0.5053,
+      "step": 714
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6237699240689051,
+      "learning_rate": 4.967432162751792e-06,
+      "loss": 0.4727,
+      "step": 715
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.666141553803713,
+      "learning_rate": 4.967129168763981e-06,
+      "loss": 0.4991,
+      "step": 716
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.646564036075848,
+      "learning_rate": 4.966824781173245e-06,
+      "loss": 0.5251,
+      "step": 717
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6401865563361007,
+      "learning_rate": 4.966519000151522e-06,
+      "loss": 0.4882,
+      "step": 718
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6532940949381087,
+      "learning_rate": 4.966211825871538e-06,
+      "loss": 0.4859,
+      "step": 719
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6240667658033411,
+      "learning_rate": 4.965903258506806e-06,
+      "loss": 0.5165,
+      "step": 720
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6050182788187826,
+      "learning_rate": 4.965593298231627e-06,
+      "loss": 0.4972,
+      "step": 721
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6394796781675777,
+      "learning_rate": 4.965281945221086e-06,
+      "loss": 0.4863,
+      "step": 722
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5879647113680488,
+      "learning_rate": 4.964969199651059e-06,
+      "loss": 0.5268,
+      "step": 723
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6719126547265977,
+      "learning_rate": 4.964655061698204e-06,
+      "loss": 0.5094,
+      "step": 724
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6247081850307215,
+      "learning_rate": 4.964339531539967e-06,
+      "loss": 0.5042,
+      "step": 725
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6448631581908335,
+      "learning_rate": 4.964022609354583e-06,
+      "loss": 0.548,
+      "step": 726
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6155726808462035,
+      "learning_rate": 4.963704295321069e-06,
+      "loss": 0.5079,
+      "step": 727
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5856540757636703,
+      "learning_rate": 4.963384589619233e-06,
+      "loss": 0.5344,
+      "step": 728
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6160053187700611,
+      "learning_rate": 4.963063492429665e-06,
+      "loss": 0.5174,
+      "step": 729
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6092787845104115,
+      "learning_rate": 4.9627410039337426e-06,
+      "loss": 0.5069,
+      "step": 730
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.572582103354353,
+      "learning_rate": 4.96241712431363e-06,
+      "loss": 0.4989,
+      "step": 731
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5711072343879144,
+      "learning_rate": 4.962091853752276e-06,
+      "loss": 0.5148,
+      "step": 732
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6061761244724242,
+      "learning_rate": 4.961765192433415e-06,
+      "loss": 0.5287,
+      "step": 733
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6033994961802631,
+      "learning_rate": 4.961437140541569e-06,
+      "loss": 0.4927,
+      "step": 734
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6209778806623933,
+      "learning_rate": 4.9611076982620445e-06,
+      "loss": 0.5358,
+      "step": 735
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6501849583238293,
+      "learning_rate": 4.960776865780931e-06,
+      "loss": 0.5183,
+      "step": 736
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.5740429036937207,
+      "learning_rate": 4.9604446432851064e-06,
+      "loss": 0.481,
+      "step": 737
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5986038003667088,
+      "learning_rate": 4.960111030962232e-06,
+      "loss": 0.5035,
+      "step": 738
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6269593371981165,
+      "learning_rate": 4.959776029000756e-06,
+      "loss": 0.5019,
+      "step": 739
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.586820399261423,
+      "learning_rate": 4.959439637589909e-06,
+      "loss": 0.4907,
+      "step": 740
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.591243683238131,
+      "learning_rate": 4.959101856919709e-06,
+      "loss": 0.5128,
+      "step": 741
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6043813840833816,
+      "learning_rate": 4.9587626871809564e-06,
+      "loss": 0.5254,
+      "step": 742
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6088008476429794,
+      "learning_rate": 4.958422128565238e-06,
+      "loss": 0.5099,
+      "step": 743
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5919159624960704,
+      "learning_rate": 4.958080181264926e-06,
+      "loss": 0.4813,
+      "step": 744
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5767332938769787,
+      "learning_rate": 4.957736845473173e-06,
+      "loss": 0.5118,
+      "step": 745
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6115434997604602,
+      "learning_rate": 4.957392121383919e-06,
+      "loss": 0.5178,
+      "step": 746
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.606547936666955,
+      "learning_rate": 4.957046009191889e-06,
+      "loss": 0.5125,
+      "step": 747
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6281682365690769,
+      "learning_rate": 4.956698509092591e-06,
+      "loss": 0.5302,
+      "step": 748
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6041160141331913,
+      "learning_rate": 4.956349621282315e-06,
+      "loss": 0.4981,
+      "step": 749
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6082725831189176,
+      "learning_rate": 4.9559993459581375e-06,
+      "loss": 0.5022,
+      "step": 750
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6308489518539907,
+      "learning_rate": 4.9556476833179185e-06,
+      "loss": 0.5145,
+      "step": 751
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6122107491320039,
+      "learning_rate": 4.9552946335603006e-06,
+      "loss": 0.4803,
+      "step": 752
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6247807256313168,
+      "learning_rate": 4.95494019688471e-06,
+      "loss": 0.5033,
+      "step": 753
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6133850387446359,
+      "learning_rate": 4.954584373491357e-06,
+      "loss": 0.5094,
+      "step": 754
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6610062403122814,
+      "learning_rate": 4.954227163581234e-06,
+      "loss": 0.5246,
+      "step": 755
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.5719697098479033,
+      "learning_rate": 4.953868567356121e-06,
+      "loss": 0.4986,
+      "step": 756
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.611277655866564,
+      "learning_rate": 4.953508585018573e-06,
+      "loss": 0.5084,
+      "step": 757
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.570613285492476,
+      "learning_rate": 4.953147216771935e-06,
+      "loss": 0.5154,
+      "step": 758
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6081820557545713,
+      "learning_rate": 4.952784462820333e-06,
+      "loss": 0.5217,
+      "step": 759
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.6233401854792252,
+      "learning_rate": 4.952420323368673e-06,
+      "loss": 0.4905,
+      "step": 760
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5903715996857434,
+      "learning_rate": 4.952054798622649e-06,
+      "loss": 0.4612,
+      "step": 761
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5976026591907524,
+      "learning_rate": 4.951687888788731e-06,
+      "loss": 0.5114,
+      "step": 762
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6424120319126979,
+      "learning_rate": 4.9513195940741764e-06,
+      "loss": 0.4839,
+      "step": 763
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6168679097191889,
+      "learning_rate": 4.950949914687024e-06,
+      "loss": 0.5134,
+      "step": 764
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6400860199848029,
+      "learning_rate": 4.950578850836092e-06,
+      "loss": 0.5087,
+      "step": 765
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6011473106721757,
+      "learning_rate": 4.950206402730984e-06,
+      "loss": 0.526,
+      "step": 766
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6076752750689335,
+      "learning_rate": 4.949832570582083e-06,
+      "loss": 0.5243,
+      "step": 767
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.611589146611987,
+      "learning_rate": 4.949457354600556e-06,
+      "loss": 0.5602,
+      "step": 768
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6379045005420435,
+      "learning_rate": 4.94908075499835e-06,
+      "loss": 0.485,
+      "step": 769
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6111544996406524,
+      "learning_rate": 4.948702771988195e-06,
+      "loss": 0.4961,
+      "step": 770
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6440163967432894,
+      "learning_rate": 4.9483234057836e-06,
+      "loss": 0.5247,
+      "step": 771
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6424136239524529,
+      "learning_rate": 4.9479426565988585e-06,
+      "loss": 0.5095,
+      "step": 772
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5845387618126129,
+      "learning_rate": 4.947560524649043e-06,
+      "loss": 0.4667,
+      "step": 773
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5581508613420673,
+      "learning_rate": 4.947177010150007e-06,
+      "loss": 0.4915,
+      "step": 774
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6002089992364806,
+      "learning_rate": 4.9467921133183864e-06,
+      "loss": 0.4957,
+      "step": 775
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.571455007692429,
+      "learning_rate": 4.946405834371598e-06,
+      "loss": 0.4998,
+      "step": 776
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5996615965227683,
+      "learning_rate": 4.9460181735278365e-06,
+      "loss": 0.5058,
+      "step": 777
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5933857632393196,
+      "learning_rate": 4.945629131006081e-06,
+      "loss": 0.5165,
+      "step": 778
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5922936609778912,
+      "learning_rate": 4.945238707026087e-06,
+      "loss": 0.497,
+      "step": 779
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6468383832092212,
+      "learning_rate": 4.944846901808397e-06,
+      "loss": 0.4988,
+      "step": 780
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5963282275866615,
+      "learning_rate": 4.9444537155743245e-06,
+      "loss": 0.5146,
+      "step": 781
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 0.5058240294456482,
+      "eval_runtime": 6911.9149,
+      "eval_samples_per_second": 42.013,
+      "eval_steps_per_second": 2.101,
+      "step": 781
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.616517236291001,
+      "learning_rate": 4.944059148545971e-06,
+      "loss": 0.5123,
+      "step": 782
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6002931009806305,
+      "learning_rate": 4.943663200946213e-06,
+      "loss": 0.501,
+      "step": 783
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.5875304361967715,
+      "learning_rate": 4.94326587299871e-06,
+      "loss": 0.4847,
+      "step": 784
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5836573087771422,
+      "learning_rate": 4.942867164927899e-06,
+      "loss": 0.4998,
+      "step": 785
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5783319062666641,
+      "learning_rate": 4.942467076958999e-06,
+      "loss": 0.4989,
+      "step": 786
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6295939909126551,
+      "learning_rate": 4.9420656093180056e-06,
+      "loss": 0.4886,
+      "step": 787
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5808772327348201,
+      "learning_rate": 4.941662762231695e-06,
+      "loss": 0.5036,
+      "step": 788
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5757656711915483,
+      "learning_rate": 4.9412585359276235e-06,
+      "loss": 0.4892,
+      "step": 789
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6156971766935416,
+      "learning_rate": 4.940852930634126e-06,
+      "loss": 0.5189,
+      "step": 790
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5964903100014775,
+      "learning_rate": 4.940445946580315e-06,
+      "loss": 0.4893,
+      "step": 791
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5930502537178118,
+      "learning_rate": 4.9400375839960826e-06,
+      "loss": 0.4854,
+      "step": 792
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6216632090251537,
+      "learning_rate": 4.939627843112102e-06,
+      "loss": 0.52,
+      "step": 793
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5981307705816444,
+      "learning_rate": 4.939216724159821e-06,
+      "loss": 0.4924,
+      "step": 794
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6171104301125759,
+      "learning_rate": 4.938804227371467e-06,
+      "loss": 0.4949,
+      "step": 795
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6232653150523578,
+      "learning_rate": 4.938390352980049e-06,
+      "loss": 0.5276,
+      "step": 796
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6319806108460523,
+      "learning_rate": 4.93797510121935e-06,
+      "loss": 0.4867,
+      "step": 797
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6373229220396098,
+      "learning_rate": 4.937558472323932e-06,
+      "loss": 0.5175,
+      "step": 798
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6381263394133697,
+      "learning_rate": 4.937140466529135e-06,
+      "loss": 0.5212,
+      "step": 799
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6889087852800232,
+      "learning_rate": 4.936721084071079e-06,
+      "loss": 0.5068,
+      "step": 800
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6217485975923667,
+      "learning_rate": 4.936300325186659e-06,
+      "loss": 0.4926,
+      "step": 801
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.571769849631971,
+      "learning_rate": 4.9358781901135485e-06,
+      "loss": 0.509,
+      "step": 802
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6780444728173729,
+      "learning_rate": 4.935454679090197e-06,
+      "loss": 0.5165,
+      "step": 803
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6076278839686122,
+      "learning_rate": 4.935029792355834e-06,
+      "loss": 0.5039,
+      "step": 804
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6538852692558306,
+      "learning_rate": 4.9346035301504644e-06,
+      "loss": 0.4962,
+      "step": 805
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6171376333190426,
+      "learning_rate": 4.934175892714869e-06,
+      "loss": 0.5112,
+      "step": 806
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6558521386937952,
+      "learning_rate": 4.933746880290607e-06,
+      "loss": 0.4807,
+      "step": 807
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5755989068902357,
+      "learning_rate": 4.933316493120015e-06,
+      "loss": 0.4821,
+      "step": 808
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6612931100425695,
+      "learning_rate": 4.932884731446204e-06,
+      "loss": 0.517,
+      "step": 809
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6216763697335129,
+      "learning_rate": 4.932451595513063e-06,
+      "loss": 0.5269,
+      "step": 810
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6136117850372235,
+      "learning_rate": 4.932017085565256e-06,
+      "loss": 0.5197,
+      "step": 811
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5783344700139442,
+      "learning_rate": 4.931581201848224e-06,
+      "loss": 0.5167,
+      "step": 812
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5691685860925245,
+      "learning_rate": 4.931143944608184e-06,
+      "loss": 0.4957,
+      "step": 813
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6136146632090349,
+      "learning_rate": 4.93070531409213e-06,
+      "loss": 0.4923,
+      "step": 814
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6146071241365089,
+      "learning_rate": 4.930265310547829e-06,
+      "loss": 0.498,
+      "step": 815
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6002352821905501,
+      "learning_rate": 4.9298239342238255e-06,
+      "loss": 0.4994,
+      "step": 816
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5889451517506399,
+      "learning_rate": 4.929381185369438e-06,
+      "loss": 0.5001,
+      "step": 817
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6082258257150579,
+      "learning_rate": 4.928937064234764e-06,
+      "loss": 0.5521,
+      "step": 818
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6330482315113839,
+      "learning_rate": 4.928491571070669e-06,
+      "loss": 0.5219,
+      "step": 819
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6004600333139476,
+      "learning_rate": 4.928044706128803e-06,
+      "loss": 0.5005,
+      "step": 820
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6189100444908509,
+      "learning_rate": 4.927596469661582e-06,
+      "loss": 0.5176,
+      "step": 821
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.606015342163562,
+      "learning_rate": 4.9271468619222015e-06,
+      "loss": 0.496,
+      "step": 822
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.8556526555027408,
+      "learning_rate": 4.926695883164632e-06,
+      "loss": 0.5257,
+      "step": 823
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6258732371599479,
+      "learning_rate": 4.926243533643615e-06,
+      "loss": 0.487,
+      "step": 824
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5994220294037795,
+      "learning_rate": 4.92578981361467e-06,
+      "loss": 0.5072,
+      "step": 825
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5716588082101919,
+      "learning_rate": 4.925334723334088e-06,
+      "loss": 0.4994,
+      "step": 826
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.585486108238998,
+      "learning_rate": 4.924878263058937e-06,
+      "loss": 0.5226,
+      "step": 827
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5747575013418114,
+      "learning_rate": 4.924420433047055e-06,
+      "loss": 0.4965,
+      "step": 828
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5919850393540613,
+      "learning_rate": 4.9239612335570555e-06,
+      "loss": 0.4948,
+      "step": 829
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.5830251031451256,
+      "learning_rate": 4.923500664848327e-06,
+      "loss": 0.487,
+      "step": 830
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6178725623700653,
+      "learning_rate": 4.923038727181028e-06,
+      "loss": 0.4862,
+      "step": 831
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5814875278732987,
+      "learning_rate": 4.922575420816095e-06,
+      "loss": 0.5166,
+      "step": 832
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5986635615376367,
+      "learning_rate": 4.922110746015234e-06,
+      "loss": 0.489,
+      "step": 833
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5859519526703544,
+      "learning_rate": 4.921644703040925e-06,
+      "loss": 0.4906,
+      "step": 834
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5743263620222753,
+      "learning_rate": 4.9211772921564205e-06,
+      "loss": 0.4843,
+      "step": 835
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8492032385838142,
+      "learning_rate": 4.920708513625746e-06,
+      "loss": 0.5099,
+      "step": 836
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6140314776814894,
+      "learning_rate": 4.9202383677137005e-06,
+      "loss": 0.5006,
+      "step": 837
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6527049634117856,
+      "learning_rate": 4.919766854685854e-06,
+      "loss": 0.5264,
+      "step": 838
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6114157924940975,
+      "learning_rate": 4.91929397480855e-06,
+      "loss": 0.4989,
+      "step": 839
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.660553349533273,
+      "learning_rate": 4.918819728348901e-06,
+      "loss": 0.4894,
+      "step": 840
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6011284115605714,
+      "learning_rate": 4.918344115574797e-06,
+      "loss": 0.5038,
+      "step": 841
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5690628939882818,
+      "learning_rate": 4.917867136754894e-06,
+      "loss": 0.5419,
+      "step": 842
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5962796608244493,
+      "learning_rate": 4.917388792158623e-06,
+      "loss": 0.4726,
+      "step": 843
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6420232226244765,
+      "learning_rate": 4.9169090820561845e-06,
+      "loss": 0.5121,
+      "step": 844
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5977037225796433,
+      "learning_rate": 4.916428006718555e-06,
+      "loss": 0.5125,
+      "step": 845
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6028773887667327,
+      "learning_rate": 4.9159455664174756e-06,
+      "loss": 0.4987,
+      "step": 846
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6366562430011422,
+      "learning_rate": 4.9154617614254616e-06,
+      "loss": 0.482,
+      "step": 847
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6591094862661601,
+      "learning_rate": 4.914976592015801e-06,
+      "loss": 0.5366,
+      "step": 848
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5958159768787876,
+      "learning_rate": 4.914490058462549e-06,
+      "loss": 0.4975,
+      "step": 849
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.60494927183394,
+      "learning_rate": 4.9140021610405335e-06,
+      "loss": 0.5224,
+      "step": 850
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6105732826310788,
+      "learning_rate": 4.913512900025351e-06,
+      "loss": 0.52,
+      "step": 851
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5967159340815495,
+      "learning_rate": 4.913022275693372e-06,
+      "loss": 0.4667,
+      "step": 852
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 0.5032065510749817,
+      "eval_runtime": 6906.5208,
+      "eval_samples_per_second": 42.046,
+      "eval_steps_per_second": 2.102,
+      "step": 852
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.6009525199808483,
+      "learning_rate": 4.912530288321733e-06,
+      "loss": 0.4868,
+      "step": 853
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5830517125297875,
+      "learning_rate": 4.912036938188342e-06,
+      "loss": 0.5266,
+      "step": 854
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5889940979301209,
+      "learning_rate": 4.911542225571877e-06,
+      "loss": 0.5029,
+      "step": 855
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6026973021517633,
+      "learning_rate": 4.911046150751786e-06,
+      "loss": 0.517,
+      "step": 856
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5785861902321248,
+      "learning_rate": 4.910548714008285e-06,
+      "loss": 0.4926,
+      "step": 857
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6270047301502678,
+      "learning_rate": 4.910049915622361e-06,
+      "loss": 0.5154,
+      "step": 858
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5882468386280117,
+      "learning_rate": 4.90954975587577e-06,
+      "loss": 0.5096,
+      "step": 859
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5663550702566633,
+      "learning_rate": 4.909048235051033e-06,
+      "loss": 0.4908,
+      "step": 860
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6017236125989268,
+      "learning_rate": 4.9085453534314474e-06,
+      "loss": 0.5193,
+      "step": 861
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6222382484211336,
+      "learning_rate": 4.908041111301074e-06,
+      "loss": 0.5167,
+      "step": 862
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5963402145379366,
+      "learning_rate": 4.90753550894474e-06,
+      "loss": 0.4786,
+      "step": 863
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5654270988232517,
+      "learning_rate": 4.907028546648049e-06,
+      "loss": 0.509,
+      "step": 864
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6108797196445513,
+      "learning_rate": 4.906520224697364e-06,
+      "loss": 0.5055,
+      "step": 865
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6008153268576729,
+      "learning_rate": 4.906010543379821e-06,
+      "loss": 0.5043,
+      "step": 866
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6448956256200653,
+      "learning_rate": 4.905499502983325e-06,
+      "loss": 0.5257,
+      "step": 867
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5595461401241696,
+      "learning_rate": 4.904987103796544e-06,
+      "loss": 0.5017,
+      "step": 868
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5941241562315258,
+      "learning_rate": 4.904473346108916e-06,
+      "loss": 0.5052,
+      "step": 869
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6018542340595553,
+      "learning_rate": 4.903958230210647e-06,
+      "loss": 0.4875,
+      "step": 870
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5681619089604921,
+      "learning_rate": 4.9034417563927105e-06,
+      "loss": 0.4876,
+      "step": 871
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.580332576906727,
+      "learning_rate": 4.902923924946845e-06,
+      "loss": 0.503,
+      "step": 872
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5991899633835461,
+      "learning_rate": 4.902404736165557e-06,
+      "loss": 0.4792,
+      "step": 873
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5788544217689144,
+      "learning_rate": 4.901884190342121e-06,
+      "loss": 0.534,
+      "step": 874
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5873037059704003,
+      "learning_rate": 4.901362287770576e-06,
+      "loss": 0.5138,
+      "step": 875
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5673844198994867,
+      "learning_rate": 4.900839028745727e-06,
+      "loss": 0.5094,
+      "step": 876
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.5992496148744582,
+      "learning_rate": 4.900314413563149e-06,
+      "loss": 0.5296,
+      "step": 877
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.619714807218819,
+      "learning_rate": 4.899788442519178e-06,
+      "loss": 0.5174,
+      "step": 878
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5833898092793183,
+      "learning_rate": 4.899261115910919e-06,
+      "loss": 0.4845,
+      "step": 879
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5731340631423973,
+      "learning_rate": 4.8987324340362445e-06,
+      "loss": 0.494,
+      "step": 880
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6335983797182833,
+      "learning_rate": 4.898202397193787e-06,
+      "loss": 0.4902,
+      "step": 881
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6021252115369131,
+      "learning_rate": 4.897671005682948e-06,
+      "loss": 0.5039,
+      "step": 882
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6136802203792989,
+      "learning_rate": 4.8971382598038945e-06,
+      "loss": 0.5108,
+      "step": 883
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6131873733015205,
+      "learning_rate": 4.896604159857557e-06,
+      "loss": 0.4997,
+      "step": 884
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6159506568258571,
+      "learning_rate": 4.896068706145632e-06,
+      "loss": 0.5292,
+      "step": 885
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6041398474984774,
+      "learning_rate": 4.8955318989705814e-06,
+      "loss": 0.51,
+      "step": 886
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6148781569224463,
+      "learning_rate": 4.8949937386356284e-06,
+      "loss": 0.5094,
+      "step": 887
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5976753831281014,
+      "learning_rate": 4.894454225444764e-06,
+      "loss": 0.5026,
+      "step": 888
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.608033075624346,
+      "learning_rate": 4.893913359702742e-06,
+      "loss": 0.5011,
+      "step": 889
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5610693452613091,
+      "learning_rate": 4.89337114171508e-06,
+      "loss": 0.4903,
+      "step": 890
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5740287354727569,
+      "learning_rate": 4.89282757178806e-06,
+      "loss": 0.5355,
+      "step": 891
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5808789786027553,
+      "learning_rate": 4.892282650228728e-06,
+      "loss": 0.4798,
+      "step": 892
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5636801724275967,
+      "learning_rate": 4.891736377344891e-06,
+      "loss": 0.4929,
+      "step": 893
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6092124636402675,
+      "learning_rate": 4.891188753445122e-06,
+      "loss": 0.4976,
+      "step": 894
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5840172519663077,
+      "learning_rate": 4.890639778838757e-06,
+      "loss": 0.4841,
+      "step": 895
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5819312884051389,
+      "learning_rate": 4.890089453835894e-06,
+      "loss": 0.4869,
+      "step": 896
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6120829375007303,
+      "learning_rate": 4.889537778747396e-06,
+      "loss": 0.5265,
+      "step": 897
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5726939025833054,
+      "learning_rate": 4.888984753884882e-06,
+      "loss": 0.4708,
+      "step": 898
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6080885452408311,
+      "learning_rate": 4.8884303795607424e-06,
+      "loss": 0.501,
+      "step": 899
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.5909973175245595,
+      "learning_rate": 4.887874656088124e-06,
+      "loss": 0.5017,
+      "step": 900
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.591651916865507,
+      "learning_rate": 4.887317583780937e-06,
+      "loss": 0.483,
+      "step": 901
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5745658546447854,
+      "learning_rate": 4.886759162953856e-06,
+      "loss": 0.4942,
+      "step": 902
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6021387624200228,
+      "learning_rate": 4.886199393922313e-06,
+      "loss": 0.4906,
+      "step": 903
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5993298907461448,
+      "learning_rate": 4.885638277002503e-06,
+      "loss": 0.4987,
+      "step": 904
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6004161472848657,
+      "learning_rate": 4.885075812511386e-06,
+      "loss": 0.5152,
+      "step": 905
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5799301402705993,
+      "learning_rate": 4.884512000766679e-06,
+      "loss": 0.5045,
+      "step": 906
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6121599376461038,
+      "learning_rate": 4.883946842086861e-06,
+      "loss": 0.4915,
+      "step": 907
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5663312512818208,
+      "learning_rate": 4.883380336791172e-06,
+      "loss": 0.4983,
+      "step": 908
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6095573669970138,
+      "learning_rate": 4.882812485199614e-06,
+      "loss": 0.5285,
+      "step": 909
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5921811411224858,
+      "learning_rate": 4.882243287632947e-06,
+      "loss": 0.5194,
+      "step": 910
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6006510756243623,
+      "learning_rate": 4.8816727444126935e-06,
+      "loss": 0.492,
+      "step": 911
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5649111234747042,
+      "learning_rate": 4.881100855861134e-06,
+      "loss": 0.4916,
+      "step": 912
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5762188689086207,
+      "learning_rate": 4.880527622301312e-06,
+      "loss": 0.4887,
+      "step": 913
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6441465256010206,
+      "learning_rate": 4.879953044057028e-06,
+      "loss": 0.511,
+      "step": 914
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5684710047872453,
+      "learning_rate": 4.879377121452844e-06,
+      "loss": 0.5001,
+      "step": 915
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5686046518640278,
+      "learning_rate": 4.8787998548140794e-06,
+      "loss": 0.5014,
+      "step": 916
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5905740541193955,
+      "learning_rate": 4.878221244466813e-06,
+      "loss": 0.486,
+      "step": 917
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5952876200881407,
+      "learning_rate": 4.8776412907378845e-06,
+      "loss": 0.481,
+      "step": 918
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.5872355311893122,
+      "learning_rate": 4.877059993954891e-06,
+      "loss": 0.5039,
+      "step": 919
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6074745120236803,
+      "learning_rate": 4.8764773544461895e-06,
+      "loss": 0.5108,
+      "step": 920
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6130696940827612,
+      "learning_rate": 4.875893372540893e-06,
+      "loss": 0.5048,
+      "step": 921
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6004208834442807,
+      "learning_rate": 4.875308048568875e-06,
+      "loss": 0.4988,
+      "step": 922
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6892465313158495,
+      "learning_rate": 4.8747213828607675e-06,
+      "loss": 0.483,
+      "step": 923
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 0.5001593232154846,
+      "eval_runtime": 6909.7667,
+      "eval_samples_per_second": 42.026,
+      "eval_steps_per_second": 2.101,
+      "step": 923
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.6134636983656643,
+      "learning_rate": 4.874133375747957e-06,
+      "loss": 0.5137,
+      "step": 924
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6440085903145949,
+      "learning_rate": 4.873544027562593e-06,
+      "loss": 0.5176,
+      "step": 925
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6899079462308539,
+      "learning_rate": 4.8729533386375775e-06,
+      "loss": 0.5162,
+      "step": 926
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6146416139724011,
+      "learning_rate": 4.872361309306572e-06,
+      "loss": 0.5113,
+      "step": 927
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.585989400430957,
+      "learning_rate": 4.8717679399039954e-06,
+      "loss": 0.4987,
+      "step": 928
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6462954923557843,
+      "learning_rate": 4.871173230765024e-06,
+      "loss": 0.501,
+      "step": 929
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6369012674472616,
+      "learning_rate": 4.8705771822255895e-06,
+      "loss": 0.4859,
+      "step": 930
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6299056450791893,
+      "learning_rate": 4.8699797946223805e-06,
+      "loss": 0.5061,
+      "step": 931
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6098008991534455,
+      "learning_rate": 4.869381068292842e-06,
+      "loss": 0.4871,
+      "step": 932
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.63534813466544,
+      "learning_rate": 4.868781003575176e-06,
+      "loss": 0.4946,
+      "step": 933
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.607178762605803,
+      "learning_rate": 4.86817960080834e-06,
+      "loss": 0.4986,
+      "step": 934
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5983782529586613,
+      "learning_rate": 4.867576860332048e-06,
+      "loss": 0.4795,
+      "step": 935
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6738363380499427,
+      "learning_rate": 4.8669727824867686e-06,
+      "loss": 0.5166,
+      "step": 936
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.598996652424602,
+      "learning_rate": 4.866367367613725e-06,
+      "loss": 0.5197,
+      "step": 937
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5815833611063322,
+      "learning_rate": 4.865760616054899e-06,
+      "loss": 0.4844,
+      "step": 938
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.603444821349114,
+      "learning_rate": 4.865152528153022e-06,
+      "loss": 0.5207,
+      "step": 939
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5867776906541347,
+      "learning_rate": 4.864543104251587e-06,
+      "loss": 0.529,
+      "step": 940
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5650043009819257,
+      "learning_rate": 4.863932344694837e-06,
+      "loss": 0.4959,
+      "step": 941
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.6917490151688115,
+      "learning_rate": 4.8633202498277695e-06,
+      "loss": 0.5137,
+      "step": 942
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.549493622060356,
+      "learning_rate": 4.862706819996139e-06,
+      "loss": 0.4845,
+      "step": 943
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5858437778240971,
+      "learning_rate": 4.8620920555464515e-06,
+      "loss": 0.5006,
+      "step": 944
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5993940522874021,
+      "learning_rate": 4.8614759568259685e-06,
+      "loss": 0.506,
+      "step": 945
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5948965330805686,
+      "learning_rate": 4.860858524182704e-06,
+      "loss": 0.5192,
+      "step": 946
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.5944946909841916,
+      "learning_rate": 4.860239757965428e-06,
+      "loss": 0.5071,
+      "step": 947
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5915006708411217,
+      "learning_rate": 4.8596196585236595e-06,
+      "loss": 0.4853,
+      "step": 948
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5630831967485228,
+      "learning_rate": 4.858998226207674e-06,
+      "loss": 0.484,
+      "step": 949
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5666674173283087,
+      "learning_rate": 4.858375461368499e-06,
+      "loss": 0.512,
+      "step": 950
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5836639113649253,
+      "learning_rate": 4.857751364357913e-06,
+      "loss": 0.514,
+      "step": 951
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5744731439956018,
+      "learning_rate": 4.857125935528451e-06,
+      "loss": 0.5177,
+      "step": 952
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5926955465496379,
+      "learning_rate": 4.8564991752333975e-06,
+      "loss": 0.4955,
+      "step": 953
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5628507262484403,
+      "learning_rate": 4.855871083826789e-06,
+      "loss": 0.4995,
+      "step": 954
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5866653009845907,
+      "learning_rate": 4.855241661663413e-06,
+      "loss": 0.5002,
+      "step": 955
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5599328850779348,
+      "learning_rate": 4.854610909098813e-06,
+      "loss": 0.4593,
+      "step": 956
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5973389278804867,
+      "learning_rate": 4.853978826489277e-06,
+      "loss": 0.4913,
+      "step": 957
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.643022571476318,
+      "learning_rate": 4.8533454141918525e-06,
+      "loss": 0.5036,
+      "step": 958
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5688007484429585,
+      "learning_rate": 4.852710672564332e-06,
+      "loss": 0.5036,
+      "step": 959
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5969281444464002,
+      "learning_rate": 4.852074601965261e-06,
+      "loss": 0.5344,
+      "step": 960
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5730247632836935,
+      "learning_rate": 4.851437202753936e-06,
+      "loss": 0.4809,
+      "step": 961
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5799456612673418,
+      "learning_rate": 4.850798475290403e-06,
+      "loss": 0.4536,
+      "step": 962
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6205297443588728,
+      "learning_rate": 4.8501584199354604e-06,
+      "loss": 0.5099,
+      "step": 963
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5681608576563654,
+      "learning_rate": 4.849517037050653e-06,
+      "loss": 0.4903,
+      "step": 964
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6025610912235543,
+      "learning_rate": 4.848874326998279e-06,
+      "loss": 0.4836,
+      "step": 965
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5901338922642949,
+      "learning_rate": 4.848230290141383e-06,
+      "loss": 0.4662,
+      "step": 966
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5688713468398793,
+      "learning_rate": 4.847584926843765e-06,
+      "loss": 0.4884,
+      "step": 967
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5778466913445185,
+      "learning_rate": 4.846938237469966e-06,
+      "loss": 0.5096,
+      "step": 968
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.603606926382752,
+      "learning_rate": 4.846290222385282e-06,
+      "loss": 0.5415,
+      "step": 969
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.594103598829822,
+      "learning_rate": 4.845640881955757e-06,
+      "loss": 0.5007,
+      "step": 970
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5983275671847197,
+      "learning_rate": 4.844990216548181e-06,
+      "loss": 0.501,
+      "step": 971
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6107199353721016,
+      "learning_rate": 4.844338226530095e-06,
+      "loss": 0.5303,
+      "step": 972
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6032407557209453,
+      "learning_rate": 4.843684912269789e-06,
+      "loss": 0.4946,
+      "step": 973
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5413313146483812,
+      "learning_rate": 4.843030274136297e-06,
+      "loss": 0.4956,
+      "step": 974
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6106091128154526,
+      "learning_rate": 4.842374312499405e-06,
+      "loss": 0.5105,
+      "step": 975
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.574289734465813,
+      "learning_rate": 4.841717027729643e-06,
+      "loss": 0.5215,
+      "step": 976
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6083050534381844,
+      "learning_rate": 4.8410584201982934e-06,
+      "loss": 0.4862,
+      "step": 977
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5766091053101502,
+      "learning_rate": 4.84039849027738e-06,
+      "loss": 0.5117,
+      "step": 978
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5711574672193085,
+      "learning_rate": 4.8397372383396765e-06,
+      "loss": 0.4722,
+      "step": 979
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5774240645659215,
+      "learning_rate": 4.839074664758705e-06,
+      "loss": 0.522,
+      "step": 980
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5572922524103036,
+      "learning_rate": 4.8384107699087305e-06,
+      "loss": 0.4989,
+      "step": 981
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6070310913343694,
+      "learning_rate": 4.837745554164766e-06,
+      "loss": 0.4857,
+      "step": 982
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.60261744740847,
+      "learning_rate": 4.8370790179025715e-06,
+      "loss": 0.5182,
+      "step": 983
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5659965233077204,
+      "learning_rate": 4.836411161498653e-06,
+      "loss": 0.4828,
+      "step": 984
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5957726040740746,
+      "learning_rate": 4.835741985330259e-06,
+      "loss": 0.4724,
+      "step": 985
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6024381722019819,
+      "learning_rate": 4.835071489775388e-06,
+      "loss": 0.5038,
+      "step": 986
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5889077069135094,
+      "learning_rate": 4.834399675212781e-06,
+      "loss": 0.4837,
+      "step": 987
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6240890961484814,
+      "learning_rate": 4.8337265420219245e-06,
+      "loss": 0.5006,
+      "step": 988
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5986554935817346,
+      "learning_rate": 4.833052090583052e-06,
+      "loss": 0.4603,
+      "step": 989
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5632943529640556,
+      "learning_rate": 4.832376321277136e-06,
+      "loss": 0.478,
+      "step": 990
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5811104630507077,
+      "learning_rate": 4.831699234485899e-06,
+      "loss": 0.503,
+      "step": 991
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.6286701964893617,
+      "learning_rate": 4.831020830591806e-06,
+      "loss": 0.5186,
+      "step": 992
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5656632345758932,
+      "learning_rate": 4.8303411099780665e-06,
+      "loss": 0.4863,
+      "step": 993
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.5914309687581247,
+      "learning_rate": 4.829660073028631e-06,
+      "loss": 0.5056,
+      "step": 994
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 0.4973524808883667,
+      "eval_runtime": 6914.8778,
+      "eval_samples_per_second": 41.995,
+      "eval_steps_per_second": 2.1,
+      "step": 994
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6155119630859821,
+      "learning_rate": 4.828977720128198e-06,
+      "loss": 0.494,
+      "step": 995
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5711544453505207,
+      "learning_rate": 4.828294051662206e-06,
+      "loss": 0.4958,
+      "step": 996
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5630820865144905,
+      "learning_rate": 4.827609068016836e-06,
+      "loss": 0.5164,
+      "step": 997
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5954033434874677,
+      "learning_rate": 4.826922769579017e-06,
+      "loss": 0.5199,
+      "step": 998
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5740037081231429,
+      "learning_rate": 4.826235156736414e-06,
+      "loss": 0.4798,
+      "step": 999
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5812019395259428,
+      "learning_rate": 4.825546229877439e-06,
+      "loss": 0.4865,
+      "step": 1000
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5893943585416656,
+      "learning_rate": 4.824855989391245e-06,
+      "loss": 0.4912,
+      "step": 1001
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6065006741354068,
+      "learning_rate": 4.824164435667727e-06,
+      "loss": 0.5072,
+      "step": 1002
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6019073096492751,
+      "learning_rate": 4.823471569097521e-06,
+      "loss": 0.4971,
+      "step": 1003
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5883223828151468,
+      "learning_rate": 4.822777390072006e-06,
+      "loss": 0.514,
+      "step": 1004
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6604784109834196,
+      "learning_rate": 4.822081898983302e-06,
+      "loss": 0.5065,
+      "step": 1005
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5937829112220406,
+      "learning_rate": 4.821385096224268e-06,
+      "loss": 0.4714,
+      "step": 1006
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5748353294494913,
+      "learning_rate": 4.820686982188508e-06,
+      "loss": 0.5059,
+      "step": 1007
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6069950364399875,
+      "learning_rate": 4.819987557270364e-06,
+      "loss": 0.5204,
+      "step": 1008
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6358630633744264,
+      "learning_rate": 4.819286821864917e-06,
+      "loss": 0.4903,
+      "step": 1009
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6204642750728111,
+      "learning_rate": 4.818584776367992e-06,
+      "loss": 0.5128,
+      "step": 1010
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6829577526239162,
+      "learning_rate": 4.817881421176153e-06,
+      "loss": 0.5197,
+      "step": 1011
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5922818255526525,
+      "learning_rate": 4.817176756686701e-06,
+      "loss": 0.4918,
+      "step": 1012
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5740860408805906,
+      "learning_rate": 4.816470783297679e-06,
+      "loss": 0.4982,
+      "step": 1013
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5910300282185704,
+      "learning_rate": 4.815763501407869e-06,
+      "loss": 0.485,
+      "step": 1014
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5715165207150573,
+      "learning_rate": 4.815054911416795e-06,
+      "loss": 0.4849,
+      "step": 1015
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6024018069444244,
+      "learning_rate": 4.8143450137247116e-06,
+      "loss": 0.4994,
+      "step": 1016
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.6266053738868206,
+      "learning_rate": 4.8136338087326214e-06,
+      "loss": 0.5286,
+      "step": 1017
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.5953155175370475,
+      "learning_rate": 4.812921296842261e-06,
+      "loss": 0.4776,
+      "step": 1018
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6031211740166059,
+      "learning_rate": 4.812207478456105e-06,
+      "loss": 0.5148,
+      "step": 1019
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6273922038641672,
+      "learning_rate": 4.811492353977366e-06,
+      "loss": 0.5029,
+      "step": 1020
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5826143241371425,
+      "learning_rate": 4.810775923809996e-06,
+      "loss": 0.5056,
+      "step": 1021
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.582077622885555,
+      "learning_rate": 4.810058188358685e-06,
+      "loss": 0.487,
+      "step": 1022
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5767038876088614,
+      "learning_rate": 4.809339148028857e-06,
+      "loss": 0.4805,
+      "step": 1023
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5850526215638874,
+      "learning_rate": 4.808618803226675e-06,
+      "loss": 0.4686,
+      "step": 1024
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6178237737641012,
+      "learning_rate": 4.80789715435904e-06,
+      "loss": 0.5005,
+      "step": 1025
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5579649875995263,
+      "learning_rate": 4.807174201833589e-06,
+      "loss": 0.4848,
+      "step": 1026
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.576234598032971,
+      "learning_rate": 4.8064499460586926e-06,
+      "loss": 0.4824,
+      "step": 1027
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5826322787338029,
+      "learning_rate": 4.8057243874434625e-06,
+      "loss": 0.4997,
+      "step": 1028
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5580813074509025,
+      "learning_rate": 4.8049975263977416e-06,
+      "loss": 0.4693,
+      "step": 1029
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5695474003960879,
+      "learning_rate": 4.804269363332112e-06,
+      "loss": 0.492,
+      "step": 1030
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5905987109265735,
+      "learning_rate": 4.80353989865789e-06,
+      "loss": 0.4966,
+      "step": 1031
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5685057462956427,
+      "learning_rate": 4.802809132787125e-06,
+      "loss": 0.5016,
+      "step": 1032
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5886266001836196,
+      "learning_rate": 4.802077066132607e-06,
+      "loss": 0.4946,
+      "step": 1033
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6035821042805889,
+      "learning_rate": 4.801343699107854e-06,
+      "loss": 0.4959,
+      "step": 1034
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.566265021004039,
+      "learning_rate": 4.800609032127123e-06,
+      "loss": 0.4972,
+      "step": 1035
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5614736780544856,
+      "learning_rate": 4.799873065605404e-06,
+      "loss": 0.4807,
+      "step": 1036
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.600684352457,
+      "learning_rate": 4.799135799958421e-06,
+      "loss": 0.496,
+      "step": 1037
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.591873988888275,
+      "learning_rate": 4.798397235602632e-06,
+      "loss": 0.4698,
+      "step": 1038
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6143917690140963,
+      "learning_rate": 4.797657372955228e-06,
+      "loss": 0.4906,
+      "step": 1039
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5837819971908614,
+      "learning_rate": 4.7969162124341354e-06,
+      "loss": 0.4711,
+      "step": 1040
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6109135143710297,
+      "learning_rate": 4.79617375445801e-06,
+      "loss": 0.5127,
+      "step": 1041
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5763531175283814,
+      "learning_rate": 4.795429999446246e-06,
+      "loss": 0.4889,
+      "step": 1042
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5818174333398002,
+      "learning_rate": 4.794684947818964e-06,
+      "loss": 0.4956,
+      "step": 1043
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6195177609111279,
+      "learning_rate": 4.793938599997021e-06,
+      "loss": 0.4926,
+      "step": 1044
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5686953952488751,
+      "learning_rate": 4.793190956402005e-06,
+      "loss": 0.4815,
+      "step": 1045
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5648932079137525,
+      "learning_rate": 4.792442017456237e-06,
+      "loss": 0.5092,
+      "step": 1046
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5851943135122919,
+      "learning_rate": 4.791691783582768e-06,
+      "loss": 0.4765,
+      "step": 1047
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.591274407551804,
+      "learning_rate": 4.790940255205381e-06,
+      "loss": 0.5056,
+      "step": 1048
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6379099025664374,
+      "learning_rate": 4.790187432748591e-06,
+      "loss": 0.5158,
+      "step": 1049
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6181747896706304,
+      "learning_rate": 4.789433316637644e-06,
+      "loss": 0.5146,
+      "step": 1050
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5531345989453061,
+      "learning_rate": 4.788677907298516e-06,
+      "loss": 0.4878,
+      "step": 1051
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5876103896392866,
+      "learning_rate": 4.7879212051579124e-06,
+      "loss": 0.5136,
+      "step": 1052
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 3.875613200014087,
+      "learning_rate": 4.787163210643272e-06,
+      "loss": 0.4833,
+      "step": 1053
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6036074704422942,
+      "learning_rate": 4.786403924182761e-06,
+      "loss": 0.5099,
+      "step": 1054
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.621921611720999,
+      "learning_rate": 4.785643346205277e-06,
+      "loss": 0.5052,
+      "step": 1055
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6125351583146175,
+      "learning_rate": 4.784881477140445e-06,
+      "loss": 0.5053,
+      "step": 1056
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5999869964811588,
+      "learning_rate": 4.784118317418621e-06,
+      "loss": 0.5077,
+      "step": 1057
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5923633626129878,
+      "learning_rate": 4.7833538674708905e-06,
+      "loss": 0.4784,
+      "step": 1058
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5564286456120893,
+      "learning_rate": 4.782588127729066e-06,
+      "loss": 0.4985,
+      "step": 1059
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5651692637493795,
+      "learning_rate": 4.781821098625691e-06,
+      "loss": 0.4796,
+      "step": 1060
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6611434126482291,
+      "learning_rate": 4.7810527805940344e-06,
+      "loss": 0.5125,
+      "step": 1061
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.6079384633788307,
+      "learning_rate": 4.7802831740680955e-06,
+      "loss": 0.4909,
+      "step": 1062
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5887501058130822,
+      "learning_rate": 4.7795122794826e-06,
+      "loss": 0.4729,
+      "step": 1063
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8409421704273247,
+      "learning_rate": 4.778740097273003e-06,
+      "loss": 0.5162,
+      "step": 1064
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6099196443182712,
+      "learning_rate": 4.777966627875484e-06,
+      "loss": 0.5268,
+      "step": 1065
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 0.49549439549446106,
+      "eval_runtime": 6918.4924,
+      "eval_samples_per_second": 41.973,
+      "eval_steps_per_second": 2.099,
+      "step": 1065
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.63487506837019,
+      "learning_rate": 4.777191871726951e-06,
+      "loss": 0.4854,
+      "step": 1066
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.810054670245462,
+      "learning_rate": 4.776415829265043e-06,
+      "loss": 0.5234,
+      "step": 1067
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5924971114450747,
+      "learning_rate": 4.775638500928117e-06,
+      "loss": 0.5024,
+      "step": 1068
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5596644374385136,
+      "learning_rate": 4.774859887155263e-06,
+      "loss": 0.4743,
+      "step": 1069
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6022292614035627,
+      "learning_rate": 4.7740799883862966e-06,
+      "loss": 0.4998,
+      "step": 1070
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5685916266892248,
+      "learning_rate": 4.773298805061756e-06,
+      "loss": 0.4877,
+      "step": 1071
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5962356632555038,
+      "learning_rate": 4.772516337622907e-06,
+      "loss": 0.4977,
+      "step": 1072
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6113056469088662,
+      "learning_rate": 4.771732586511741e-06,
+      "loss": 0.5129,
+      "step": 1073
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5729626434260354,
+      "learning_rate": 4.7709475521709745e-06,
+      "loss": 0.5059,
+      "step": 1074
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5952807746932272,
+      "learning_rate": 4.770161235044047e-06,
+      "loss": 0.4947,
+      "step": 1075
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.583828215474725,
+      "learning_rate": 4.769373635575127e-06,
+      "loss": 0.4919,
+      "step": 1076
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6213165462594205,
+      "learning_rate": 4.768584754209101e-06,
+      "loss": 0.5161,
+      "step": 1077
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5695247509557161,
+      "learning_rate": 4.767794591391585e-06,
+      "loss": 0.4995,
+      "step": 1078
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5741341553365366,
+      "learning_rate": 4.767003147568917e-06,
+      "loss": 0.475,
+      "step": 1079
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5768424395874503,
+      "learning_rate": 4.766210423188158e-06,
+      "loss": 0.4994,
+      "step": 1080
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5626374906483355,
+      "learning_rate": 4.765416418697092e-06,
+      "loss": 0.503,
+      "step": 1081
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5683372940335709,
+      "learning_rate": 4.764621134544229e-06,
+      "loss": 0.5083,
+      "step": 1082
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.9131547521845851,
+      "learning_rate": 4.763824571178798e-06,
+      "loss": 0.4672,
+      "step": 1083
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5723059333795633,
+      "learning_rate": 4.763026729050752e-06,
+      "loss": 0.483,
+      "step": 1084
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5906452138705621,
+      "learning_rate": 4.7622276086107685e-06,
+      "loss": 0.5082,
+      "step": 1085
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5774338186179943,
+      "learning_rate": 4.761427210310244e-06,
+      "loss": 0.4797,
+      "step": 1086
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5795580232694232,
+      "learning_rate": 4.760625534601299e-06,
+      "loss": 0.4718,
+      "step": 1087
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.5917263527753107,
+      "learning_rate": 4.759822581936773e-06,
+      "loss": 0.4773,
+      "step": 1088
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5606273835003459,
+      "learning_rate": 4.759018352770229e-06,
+      "loss": 0.4876,
+      "step": 1089
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5763982202540356,
+      "learning_rate": 4.758212847555953e-06,
+      "loss": 0.476,
+      "step": 1090
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5761656331674372,
+      "learning_rate": 4.757406066748947e-06,
+      "loss": 0.4526,
+      "step": 1091
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6337475882937359,
+      "learning_rate": 4.756598010804935e-06,
+      "loss": 0.5034,
+      "step": 1092
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.598332500270868,
+      "learning_rate": 4.755788680180363e-06,
+      "loss": 0.5204,
+      "step": 1093
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5983336967750388,
+      "learning_rate": 4.754978075332398e-06,
+      "loss": 0.4992,
+      "step": 1094
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5932701762917508,
+      "learning_rate": 4.7541661967189225e-06,
+      "loss": 0.4924,
+      "step": 1095
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.571498332890339,
+      "learning_rate": 4.7533530447985424e-06,
+      "loss": 0.507,
+      "step": 1096
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6003239552991582,
+      "learning_rate": 4.752538620030581e-06,
+      "loss": 0.4797,
+      "step": 1097
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5906168083325332,
+      "learning_rate": 4.7517229228750804e-06,
+      "loss": 0.5193,
+      "step": 1098
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5931501865913923,
+      "learning_rate": 4.750905953792803e-06,
+      "loss": 0.4812,
+      "step": 1099
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6290969275442341,
+      "learning_rate": 4.750087713245227e-06,
+      "loss": 0.5057,
+      "step": 1100
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5604034491261822,
+      "learning_rate": 4.749268201694553e-06,
+      "loss": 0.4678,
+      "step": 1101
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6014101011761356,
+      "learning_rate": 4.748447419603696e-06,
+      "loss": 0.4999,
+      "step": 1102
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5926144054570872,
+      "learning_rate": 4.747625367436288e-06,
+      "loss": 0.4803,
+      "step": 1103
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5820804611550561,
+      "learning_rate": 4.746802045656683e-06,
+      "loss": 0.5047,
+      "step": 1104
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6087772177843124,
+      "learning_rate": 4.745977454729947e-06,
+      "loss": 0.5042,
+      "step": 1105
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.594424789965164,
+      "learning_rate": 4.7451515951218675e-06,
+      "loss": 0.5102,
+      "step": 1106
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5731428331636981,
+      "learning_rate": 4.744324467298944e-06,
+      "loss": 0.5191,
+      "step": 1107
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5657909010178551,
+      "learning_rate": 4.743496071728396e-06,
+      "loss": 0.4829,
+      "step": 1108
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5670594056702332,
+      "learning_rate": 4.7426664088781585e-06,
+      "loss": 0.4896,
+      "step": 1109
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6028975438711562,
+      "learning_rate": 4.74183547921688e-06,
+      "loss": 0.4996,
+      "step": 1110
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6072539145327235,
+      "learning_rate": 4.741003283213928e-06,
+      "loss": 0.5041,
+      "step": 1111
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5867589851114696,
+      "learning_rate": 4.740169821339381e-06,
+      "loss": 0.4992,
+      "step": 1112
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.586826402405483,
+      "learning_rate": 4.739335094064038e-06,
+      "loss": 0.4616,
+      "step": 1113
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5943697459318228,
+      "learning_rate": 4.738499101859409e-06,
+      "loss": 0.4942,
+      "step": 1114
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6118584069615158,
+      "learning_rate": 4.7376618451977195e-06,
+      "loss": 0.4826,
+      "step": 1115
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5877105797297991,
+      "learning_rate": 4.736823324551909e-06,
+      "loss": 0.4845,
+      "step": 1116
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6099949628606918,
+      "learning_rate": 4.735983540395631e-06,
+      "loss": 0.4982,
+      "step": 1117
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5834134958716418,
+      "learning_rate": 4.735142493203253e-06,
+      "loss": 0.4944,
+      "step": 1118
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5566576275979142,
+      "learning_rate": 4.734300183449856e-06,
+      "loss": 0.5016,
+      "step": 1119
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6126927082777508,
+      "learning_rate": 4.733456611611233e-06,
+      "loss": 0.5104,
+      "step": 1120
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5844938513436758,
+      "learning_rate": 4.732611778163894e-06,
+      "loss": 0.5373,
+      "step": 1121
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5486863613853675,
+      "learning_rate": 4.7317656835850544e-06,
+      "loss": 0.5098,
+      "step": 1122
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.6075051942047663,
+      "learning_rate": 4.73091832835265e-06,
+      "loss": 0.4901,
+      "step": 1123
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5805346057610045,
+      "learning_rate": 4.730069712945322e-06,
+      "loss": 0.511,
+      "step": 1124
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5775795364832905,
+      "learning_rate": 4.729219837842427e-06,
+      "loss": 0.4916,
+      "step": 1125
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5931365026299273,
+      "learning_rate": 4.728368703524034e-06,
+      "loss": 0.4897,
+      "step": 1126
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5538844072984839,
+      "learning_rate": 4.72751631047092e-06,
+      "loss": 0.4937,
+      "step": 1127
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.585745992804038,
+      "learning_rate": 4.726662659164576e-06,
+      "loss": 0.4791,
+      "step": 1128
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.58228568438699,
+      "learning_rate": 4.725807750087201e-06,
+      "loss": 0.4897,
+      "step": 1129
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.623272001177232,
+      "learning_rate": 4.7249515837217075e-06,
+      "loss": 0.4713,
+      "step": 1130
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5882528036227618,
+      "learning_rate": 4.724094160551716e-06,
+      "loss": 0.4888,
+      "step": 1131
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5665571281823603,
+      "learning_rate": 4.7232354810615575e-06,
+      "loss": 0.4634,
+      "step": 1132
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5547983753857056,
+      "learning_rate": 4.722375545736273e-06,
+      "loss": 0.4908,
+      "step": 1133
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5637582498617196,
+      "learning_rate": 4.7215143550616124e-06,
+      "loss": 0.4912,
+      "step": 1134
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.5691678446202798,
+      "learning_rate": 4.720651909524037e-06,
+      "loss": 0.5088,
+      "step": 1135
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6037047316013417,
+      "learning_rate": 4.719788209610711e-06,
+      "loss": 0.5063,
+      "step": 1136
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 0.49330827593803406,
+      "eval_runtime": 6918.0742,
+      "eval_samples_per_second": 41.975,
+      "eval_steps_per_second": 2.099,
+      "step": 1136
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7053964822007601,
+      "learning_rate": 4.718923255809514e-06,
+      "loss": 0.4887,
+      "step": 1137
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6112035128658351,
+      "learning_rate": 4.71805704860903e-06,
+      "loss": 0.515,
+      "step": 1138
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6144098315190231,
+      "learning_rate": 4.717189588498552e-06,
+      "loss": 0.4933,
+      "step": 1139
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5899433076087655,
+      "learning_rate": 4.716320875968081e-06,
+      "loss": 0.5082,
+      "step": 1140
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5827664203359575,
+      "learning_rate": 4.715450911508324e-06,
+      "loss": 0.4962,
+      "step": 1141
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.559390267287684,
+      "learning_rate": 4.714579695610698e-06,
+      "loss": 0.496,
+      "step": 1142
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5970166193989745,
+      "learning_rate": 4.7137072287673244e-06,
+      "loss": 0.4963,
+      "step": 1143
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5862371291827102,
+      "learning_rate": 4.712833511471032e-06,
+      "loss": 0.5107,
+      "step": 1144
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5621562211030774,
+      "learning_rate": 4.711958544215355e-06,
+      "loss": 0.4848,
+      "step": 1145
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5957285834779159,
+      "learning_rate": 4.711082327494536e-06,
+      "loss": 0.5068,
+      "step": 1146
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6068577769642044,
+      "learning_rate": 4.710204861803522e-06,
+      "loss": 0.4806,
+      "step": 1147
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5871604299043601,
+      "learning_rate": 4.709326147637965e-06,
+      "loss": 0.5263,
+      "step": 1148
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6470536987852363,
+      "learning_rate": 4.708446185494222e-06,
+      "loss": 0.4911,
+      "step": 1149
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.581159959235341,
+      "learning_rate": 4.707564975869357e-06,
+      "loss": 0.5169,
+      "step": 1150
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.5891669395141927,
+      "learning_rate": 4.706682519261137e-06,
+      "loss": 0.4872,
+      "step": 1151
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6080392901510567,
+      "learning_rate": 4.7057988161680325e-06,
+      "loss": 0.5072,
+      "step": 1152
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6173464560574663,
+      "learning_rate": 4.704913867089221e-06,
+      "loss": 0.4857,
+      "step": 1153
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6353962573655269,
+      "learning_rate": 4.704027672524582e-06,
+      "loss": 0.5113,
+      "step": 1154
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6161269437993665,
+      "learning_rate": 4.703140232974697e-06,
+      "loss": 0.4904,
+      "step": 1155
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6096752081924692,
+      "learning_rate": 4.7022515489408536e-06,
+      "loss": 0.4925,
+      "step": 1156
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.669481066717579,
+      "learning_rate": 4.701361620925041e-06,
+      "loss": 0.4996,
+      "step": 1157
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6192786083709042,
+      "learning_rate": 4.700470449429952e-06,
+      "loss": 0.5126,
+      "step": 1158
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.643144004386803,
+      "learning_rate": 4.699578034958981e-06,
+      "loss": 0.4969,
+      "step": 1159
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5846720216828944,
+      "learning_rate": 4.698684378016223e-06,
+      "loss": 0.4776,
+      "step": 1160
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5741587597255912,
+      "learning_rate": 4.697789479106479e-06,
+      "loss": 0.4923,
+      "step": 1161
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5636128681731576,
+      "learning_rate": 4.696893338735246e-06,
+      "loss": 0.4946,
+      "step": 1162
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5616937977408087,
+      "learning_rate": 4.6959959574087265e-06,
+      "loss": 0.4956,
+      "step": 1163
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5752353921016059,
+      "learning_rate": 4.695097335633823e-06,
+      "loss": 0.4951,
+      "step": 1164
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5975316140588034,
+      "learning_rate": 4.694197473918139e-06,
+      "loss": 0.4702,
+      "step": 1165
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5712072339658532,
+      "learning_rate": 4.693296372769978e-06,
+      "loss": 0.5235,
+      "step": 1166
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5634240355738215,
+      "learning_rate": 4.692394032698341e-06,
+      "loss": 0.4926,
+      "step": 1167
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5804932429546033,
+      "learning_rate": 4.691490454212933e-06,
+      "loss": 0.4799,
+      "step": 1168
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.553753907689284,
+      "learning_rate": 4.690585637824158e-06,
+      "loss": 0.4689,
+      "step": 1169
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5690679915211406,
+      "learning_rate": 4.6896795840431155e-06,
+      "loss": 0.4839,
+      "step": 1170
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6053719083807559,
+      "learning_rate": 4.688772293381608e-06,
+      "loss": 0.5041,
+      "step": 1171
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5622187791991081,
+      "learning_rate": 4.687863766352134e-06,
+      "loss": 0.5178,
+      "step": 1172
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5706828038584673,
+      "learning_rate": 4.686954003467894e-06,
+      "loss": 0.5148,
+      "step": 1173
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5814193933038267,
+      "learning_rate": 4.686043005242781e-06,
+      "loss": 0.4925,
+      "step": 1174
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5841581837583776,
+      "learning_rate": 4.685130772191392e-06,
+      "loss": 0.5043,
+      "step": 1175
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5432146572491381,
+      "learning_rate": 4.684217304829017e-06,
+      "loss": 0.497,
+      "step": 1176
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6238281748493371,
+      "learning_rate": 4.683302603671644e-06,
+      "loss": 0.5117,
+      "step": 1177
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6120786023963148,
+      "learning_rate": 4.682386669235959e-06,
+      "loss": 0.5043,
+      "step": 1178
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6097201971727148,
+      "learning_rate": 4.681469502039345e-06,
+      "loss": 0.4781,
+      "step": 1179
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6099227512690751,
+      "learning_rate": 4.680551102599881e-06,
+      "loss": 0.484,
+      "step": 1180
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6118633618285224,
+      "learning_rate": 4.67963147143634e-06,
+      "loss": 0.4973,
+      "step": 1181
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5963607635102951,
+      "learning_rate": 4.678710609068193e-06,
+      "loss": 0.5189,
+      "step": 1182
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6095853439513462,
+      "learning_rate": 4.677788516015608e-06,
+      "loss": 0.4954,
+      "step": 1183
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5621028815997745,
+      "learning_rate": 4.676865192799443e-06,
+      "loss": 0.4872,
+      "step": 1184
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5945331480044012,
+      "learning_rate": 4.675940639941256e-06,
+      "loss": 0.5155,
+      "step": 1185
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5753731126662963,
+      "learning_rate": 4.675014857963297e-06,
+      "loss": 0.4699,
+      "step": 1186
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5747798420504858,
+      "learning_rate": 4.674087847388511e-06,
+      "loss": 0.4952,
+      "step": 1187
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5749508507302784,
+      "learning_rate": 4.673159608740536e-06,
+      "loss": 0.4896,
+      "step": 1188
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5809285237560987,
+      "learning_rate": 4.6722301425437056e-06,
+      "loss": 0.4659,
+      "step": 1189
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6013752084991827,
+      "learning_rate": 4.671299449323045e-06,
+      "loss": 0.5326,
+      "step": 1190
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6050011396855485,
+      "learning_rate": 4.670367529604274e-06,
+      "loss": 0.5158,
+      "step": 1191
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5740809770574845,
+      "learning_rate": 4.669434383913803e-06,
+      "loss": 0.4945,
+      "step": 1192
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6493339789648531,
+      "learning_rate": 4.668500012778738e-06,
+      "loss": 0.5303,
+      "step": 1193
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5707017426293574,
+      "learning_rate": 4.667564416726875e-06,
+      "loss": 0.5118,
+      "step": 1194
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5849224346254286,
+      "learning_rate": 4.666627596286702e-06,
+      "loss": 0.489,
+      "step": 1195
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5747420247370756,
+      "learning_rate": 4.6656895519874e-06,
+      "loss": 0.4815,
+      "step": 1196
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5840838885906293,
+      "learning_rate": 4.664750284358841e-06,
+      "loss": 0.5038,
+      "step": 1197
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5865365275558818,
+      "learning_rate": 4.663809793931585e-06,
+      "loss": 0.481,
+      "step": 1198
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6430958410357724,
+      "learning_rate": 4.662868081236887e-06,
+      "loss": 0.4874,
+      "step": 1199
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5694523106768161,
+      "learning_rate": 4.66192514680669e-06,
+      "loss": 0.4681,
+      "step": 1200
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6361578693239596,
+      "learning_rate": 4.660980991173628e-06,
+      "loss": 0.5068,
+      "step": 1201
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6082855757365931,
+      "learning_rate": 4.660035614871024e-06,
+      "loss": 0.4994,
+      "step": 1202
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5711249982493317,
+      "learning_rate": 4.659089018432893e-06,
+      "loss": 0.4821,
+      "step": 1203
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.6147166083946165,
+      "learning_rate": 4.658141202393935e-06,
+      "loss": 0.5056,
+      "step": 1204
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5988711912518265,
+      "learning_rate": 4.657192167289542e-06,
+      "loss": 0.4711,
+      "step": 1205
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5741231990207563,
+      "learning_rate": 4.6562419136557935e-06,
+      "loss": 0.5082,
+      "step": 1206
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6046056425511495,
+      "learning_rate": 4.655290442029459e-06,
+      "loss": 0.4681,
+      "step": 1207
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 0.4913150668144226,
+      "eval_runtime": 6919.579,
+      "eval_samples_per_second": 41.966,
+      "eval_steps_per_second": 2.098,
+      "step": 1207
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5744167724479108,
+      "learning_rate": 4.654337752947992e-06,
+      "loss": 0.4737,
+      "step": 1208
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5907262180093524,
+      "learning_rate": 4.653383846949539e-06,
+      "loss": 0.4771,
+      "step": 1209
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6398591796202622,
+      "learning_rate": 4.652428724572929e-06,
+      "loss": 0.5307,
+      "step": 1210
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6121432006011495,
+      "learning_rate": 4.6514723863576815e-06,
+      "loss": 0.4952,
+      "step": 1211
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6039842346032075,
+      "learning_rate": 4.650514832844002e-06,
+      "loss": 0.5106,
+      "step": 1212
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5945723569093062,
+      "learning_rate": 4.649556064572781e-06,
+      "loss": 0.5187,
+      "step": 1213
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6127059471922567,
+      "learning_rate": 4.648596082085597e-06,
+      "loss": 0.5023,
+      "step": 1214
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6148342317689859,
+      "learning_rate": 4.647634885924713e-06,
+      "loss": 0.5041,
+      "step": 1215
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6171661212182703,
+      "learning_rate": 4.64667247663308e-06,
+      "loss": 0.4656,
+      "step": 1216
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5811162956002015,
+      "learning_rate": 4.645708854754329e-06,
+      "loss": 0.4793,
+      "step": 1217
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5897461617987049,
+      "learning_rate": 4.644744020832782e-06,
+      "loss": 0.511,
+      "step": 1218
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6360033916365583,
+      "learning_rate": 4.6437779754134424e-06,
+      "loss": 0.4959,
+      "step": 1219
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5920841965277126,
+      "learning_rate": 4.642810719041999e-06,
+      "loss": 0.5078,
+      "step": 1220
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5764590932419221,
+      "learning_rate": 4.641842252264824e-06,
+      "loss": 0.4816,
+      "step": 1221
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.606347475674508,
+      "learning_rate": 4.640872575628973e-06,
+      "loss": 0.5039,
+      "step": 1222
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5601452063877889,
+      "learning_rate": 4.639901689682186e-06,
+      "loss": 0.483,
+      "step": 1223
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.60107394353492,
+      "learning_rate": 4.638929594972885e-06,
+      "loss": 0.5017,
+      "step": 1224
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.5832229161651298,
+      "learning_rate": 4.637956292050176e-06,
+      "loss": 0.4881,
+      "step": 1225
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6230227342266171,
+      "learning_rate": 4.636981781463848e-06,
+      "loss": 0.4771,
+      "step": 1226
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6267479062825622,
+      "learning_rate": 4.636006063764369e-06,
+      "loss": 0.4898,
+      "step": 1227
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.6267200657967705,
+      "learning_rate": 4.635029139502892e-06,
+      "loss": 0.4901,
+      "step": 1228
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6484932594616627,
+      "learning_rate": 4.634051009231251e-06,
+      "loss": 0.5206,
+      "step": 1229
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5931663500547755,
+      "learning_rate": 4.63307167350196e-06,
+      "loss": 0.4702,
+      "step": 1230
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5860181057294145,
+      "learning_rate": 4.632091132868214e-06,
+      "loss": 0.506,
+      "step": 1231
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5931231715432055,
+      "learning_rate": 4.631109387883891e-06,
+      "loss": 0.5044,
+      "step": 1232
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5821409581725687,
+      "learning_rate": 4.630126439103546e-06,
+      "loss": 0.4844,
+      "step": 1233
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.549630228044515,
+      "learning_rate": 4.629142287082416e-06,
+      "loss": 0.4804,
+      "step": 1234
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5406774405919889,
+      "learning_rate": 4.628156932376419e-06,
+      "loss": 0.4937,
+      "step": 1235
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5446742036382618,
+      "learning_rate": 4.627170375542147e-06,
+      "loss": 0.4896,
+      "step": 1236
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5801553090659803,
+      "learning_rate": 4.626182617136877e-06,
+      "loss": 0.5097,
+      "step": 1237
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.547761487753955,
+      "learning_rate": 4.625193657718563e-06,
+      "loss": 0.486,
+      "step": 1238
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5718577243267876,
+      "learning_rate": 4.624203497845835e-06,
+      "loss": 0.4951,
+      "step": 1239
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5960930128816431,
+      "learning_rate": 4.623212138078004e-06,
+      "loss": 0.5076,
+      "step": 1240
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5721738998469788,
+      "learning_rate": 4.622219578975057e-06,
+      "loss": 0.4883,
+      "step": 1241
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5957161415624224,
+      "learning_rate": 4.62122582109766e-06,
+      "loss": 0.4832,
+      "step": 1242
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5704351658199357,
+      "learning_rate": 4.620230865007154e-06,
+      "loss": 0.5335,
+      "step": 1243
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5613480503706471,
+      "learning_rate": 4.619234711265558e-06,
+      "loss": 0.4999,
+      "step": 1244
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5990237352315096,
+      "learning_rate": 4.61823736043557e-06,
+      "loss": 0.5194,
+      "step": 1245
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5646085384828234,
+      "learning_rate": 4.617238813080559e-06,
+      "loss": 0.5143,
+      "step": 1246
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5868684924014452,
+      "learning_rate": 4.616239069764574e-06,
+      "loss": 0.5031,
+      "step": 1247
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6029024571519792,
+      "learning_rate": 4.615238131052339e-06,
+      "loss": 0.4964,
+      "step": 1248
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6047661298443577,
+      "learning_rate": 4.614235997509251e-06,
+      "loss": 0.515,
+      "step": 1249
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5787855934766027,
+      "learning_rate": 4.613232669701384e-06,
+      "loss": 0.5045,
+      "step": 1250
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5628826075983923,
+      "learning_rate": 4.612228148195486e-06,
+      "loss": 0.4866,
+      "step": 1251
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.5482732141245912,
+      "learning_rate": 4.61122243355898e-06,
+      "loss": 0.4854,
+      "step": 1252
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5570796537586561,
+      "learning_rate": 4.610215526359961e-06,
+      "loss": 0.4892,
+      "step": 1253
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5726673029878695,
+      "learning_rate": 4.609207427167201e-06,
+      "loss": 0.4801,
+      "step": 1254
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6214832894338556,
+      "learning_rate": 4.60819813655014e-06,
+      "loss": 0.5322,
+      "step": 1255
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5714531549826166,
+      "learning_rate": 4.607187655078896e-06,
+      "loss": 0.4983,
+      "step": 1256
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5854107865928243,
+      "learning_rate": 4.6061759833242585e-06,
+      "loss": 0.5097,
+      "step": 1257
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5877571773366539,
+      "learning_rate": 4.605163121857688e-06,
+      "loss": 0.4764,
+      "step": 1258
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6161387817508228,
+      "learning_rate": 4.604149071251318e-06,
+      "loss": 0.4921,
+      "step": 1259
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5863778207848662,
+      "learning_rate": 4.603133832077953e-06,
+      "loss": 0.4652,
+      "step": 1260
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5889518886848664,
+      "learning_rate": 4.602117404911071e-06,
+      "loss": 0.4978,
+      "step": 1261
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.569076904297205,
+      "learning_rate": 4.601099790324817e-06,
+      "loss": 0.5452,
+      "step": 1262
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5823464063147468,
+      "learning_rate": 4.6000809888940105e-06,
+      "loss": 0.516,
+      "step": 1263
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5762572387418361,
+      "learning_rate": 4.59906100119414e-06,
+      "loss": 0.4883,
+      "step": 1264
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5710353785095638,
+      "learning_rate": 4.598039827801364e-06,
+      "loss": 0.5114,
+      "step": 1265
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5484965886436899,
+      "learning_rate": 4.597017469292511e-06,
+      "loss": 0.4806,
+      "step": 1266
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5809463207101558,
+      "learning_rate": 4.5959939262450796e-06,
+      "loss": 0.4614,
+      "step": 1267
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5845778185226627,
+      "learning_rate": 4.594969199237235e-06,
+      "loss": 0.5241,
+      "step": 1268
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5908229060244614,
+      "learning_rate": 4.593943288847814e-06,
+      "loss": 0.5007,
+      "step": 1269
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5823679847675488,
+      "learning_rate": 4.592916195656322e-06,
+      "loss": 0.4949,
+      "step": 1270
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6773394663731939,
+      "learning_rate": 4.591887920242929e-06,
+      "loss": 0.4771,
+      "step": 1271
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5600458296952812,
+      "learning_rate": 4.590858463188477e-06,
+      "loss": 0.4636,
+      "step": 1272
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.5983807787136752,
+      "learning_rate": 4.589827825074472e-06,
+      "loss": 0.4664,
+      "step": 1273
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6098172385279027,
+      "learning_rate": 4.58879600648309e-06,
+      "loss": 0.4971,
+      "step": 1274
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6237066298991583,
+      "learning_rate": 4.587763007997173e-06,
+      "loss": 0.4906,
+      "step": 1275
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5912938391288759,
+      "learning_rate": 4.586728830200227e-06,
+      "loss": 0.5064,
+      "step": 1276
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6381599023642128,
+      "learning_rate": 4.585693473676428e-06,
+      "loss": 0.4548,
+      "step": 1277
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5844695919916449,
+      "learning_rate": 4.584656939010615e-06,
+      "loss": 0.5001,
+      "step": 1278
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 0.48919913172721863,
+      "eval_runtime": 6918.8774,
+      "eval_samples_per_second": 41.971,
+      "eval_steps_per_second": 2.099,
+      "step": 1278
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.586366005962443,
+      "learning_rate": 4.583619226788294e-06,
+      "loss": 0.5084,
+      "step": 1279
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5845148658515918,
+      "learning_rate": 4.582580337595636e-06,
+      "loss": 0.4912,
+      "step": 1280
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6242513167522349,
+      "learning_rate": 4.581540272019476e-06,
+      "loss": 0.519,
+      "step": 1281
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5978675772328417,
+      "learning_rate": 4.580499030647314e-06,
+      "loss": 0.505,
+      "step": 1282
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5716562163085582,
+      "learning_rate": 4.579456614067315e-06,
+      "loss": 0.4805,
+      "step": 1283
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5887843179010744,
+      "learning_rate": 4.578413022868305e-06,
+      "loss": 0.4805,
+      "step": 1284
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5448511038289022,
+      "learning_rate": 4.577368257639778e-06,
+      "loss": 0.5194,
+      "step": 1285
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5583269034513156,
+      "learning_rate": 4.576322318971888e-06,
+      "loss": 0.4665,
+      "step": 1286
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6175638858305248,
+      "learning_rate": 4.575275207455451e-06,
+      "loss": 0.4953,
+      "step": 1287
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5893120538439389,
+      "learning_rate": 4.5742269236819485e-06,
+      "loss": 0.5078,
+      "step": 1288
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5759225882638491,
+      "learning_rate": 4.5731774682435225e-06,
+      "loss": 0.4967,
+      "step": 1289
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6027419371051689,
+      "learning_rate": 4.572126841732977e-06,
+      "loss": 0.4927,
+      "step": 1290
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5713265952976949,
+      "learning_rate": 4.571075044743778e-06,
+      "loss": 0.4818,
+      "step": 1291
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6136093959980795,
+      "learning_rate": 4.570022077870051e-06,
+      "loss": 0.4928,
+      "step": 1292
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5812274200588445,
+      "learning_rate": 4.568967941706584e-06,
+      "loss": 0.461,
+      "step": 1293
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5705816273671851,
+      "learning_rate": 4.567912636848826e-06,
+      "loss": 0.4902,
+      "step": 1294
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5683273724782274,
+      "learning_rate": 4.566856163892884e-06,
+      "loss": 0.4751,
+      "step": 1295
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5783916742928776,
+      "learning_rate": 4.565798523435528e-06,
+      "loss": 0.5014,
+      "step": 1296
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6008921391069879,
+      "learning_rate": 4.564739716074182e-06,
+      "loss": 0.4751,
+      "step": 1297
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5977176200664817,
+      "learning_rate": 4.563679742406935e-06,
+      "loss": 0.4861,
+      "step": 1298
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5819179757150398,
+      "learning_rate": 4.562618603032533e-06,
+      "loss": 0.5001,
+      "step": 1299
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.609985142660628,
+      "learning_rate": 4.561556298550379e-06,
+      "loss": 0.5011,
+      "step": 1300
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.628383300337832,
+      "learning_rate": 4.560492829560535e-06,
+      "loss": 0.4917,
+      "step": 1301
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.556934869126385,
+      "learning_rate": 4.559428196663721e-06,
+      "loss": 0.4939,
+      "step": 1302
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5753689242601436,
+      "learning_rate": 4.558362400461315e-06,
+      "loss": 0.4798,
+      "step": 1303
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.608958621920625,
+      "learning_rate": 4.55729544155535e-06,
+      "loss": 0.5147,
+      "step": 1304
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5559662431460121,
+      "learning_rate": 4.556227320548519e-06,
+      "loss": 0.4939,
+      "step": 1305
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5704357834306742,
+      "learning_rate": 4.555158038044167e-06,
+      "loss": 0.4716,
+      "step": 1306
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6333282168527011,
+      "learning_rate": 4.5540875946463e-06,
+      "loss": 0.465,
+      "step": 1307
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6080793831354635,
+      "learning_rate": 4.553015990959577e-06,
+      "loss": 0.4881,
+      "step": 1308
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5706855311890715,
+      "learning_rate": 4.551943227589314e-06,
+      "loss": 0.4647,
+      "step": 1309
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5821734570269693,
+      "learning_rate": 4.550869305141478e-06,
+      "loss": 0.5317,
+      "step": 1310
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5827695023416871,
+      "learning_rate": 4.549794224222697e-06,
+      "loss": 0.5124,
+      "step": 1311
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5511547366384472,
+      "learning_rate": 4.548717985440247e-06,
+      "loss": 0.4722,
+      "step": 1312
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6041209279498603,
+      "learning_rate": 4.547640589402063e-06,
+      "loss": 0.5015,
+      "step": 1313
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5861854523134222,
+      "learning_rate": 4.546562036716732e-06,
+      "loss": 0.4808,
+      "step": 1314
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5860300360146689,
+      "learning_rate": 4.5454823279934924e-06,
+      "loss": 0.4905,
+      "step": 1315
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5653121567715198,
+      "learning_rate": 4.5444014638422396e-06,
+      "loss": 0.4877,
+      "step": 1316
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5828331265940039,
+      "learning_rate": 4.543319444873517e-06,
+      "loss": 0.4966,
+      "step": 1317
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5524931399027349,
+      "learning_rate": 4.5422362716985255e-06,
+      "loss": 0.4934,
+      "step": 1318
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5848515428912962,
+      "learning_rate": 4.541151944929114e-06,
+      "loss": 0.4874,
+      "step": 1319
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5981291916343784,
+      "learning_rate": 4.5400664651777835e-06,
+      "loss": 0.4998,
+      "step": 1320
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.5655813478901666,
+      "learning_rate": 4.538979833057688e-06,
+      "loss": 0.4519,
+      "step": 1321
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6562430108814516,
+      "learning_rate": 4.537892049182631e-06,
+      "loss": 0.5272,
+      "step": 1322
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6017818414342826,
+      "learning_rate": 4.536803114167067e-06,
+      "loss": 0.4821,
+      "step": 1323
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.558603956646909,
+      "learning_rate": 4.535713028626101e-06,
+      "loss": 0.4879,
+      "step": 1324
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5722300913660474,
+      "learning_rate": 4.534621793175488e-06,
+      "loss": 0.4873,
+      "step": 1325
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5997686300961845,
+      "learning_rate": 4.533529408431632e-06,
+      "loss": 0.5033,
+      "step": 1326
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5829364793352837,
+      "learning_rate": 4.532435875011586e-06,
+      "loss": 0.5,
+      "step": 1327
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.570165895880547,
+      "learning_rate": 4.531341193533053e-06,
+      "loss": 0.4978,
+      "step": 1328
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.579729697723111,
+      "learning_rate": 4.530245364614384e-06,
+      "loss": 0.4837,
+      "step": 1329
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5809138960078671,
+      "learning_rate": 4.529148388874577e-06,
+      "loss": 0.5126,
+      "step": 1330
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5916301742756499,
+      "learning_rate": 4.528050266933279e-06,
+      "loss": 0.4984,
+      "step": 1331
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5820120794112843,
+      "learning_rate": 4.526950999410785e-06,
+      "loss": 0.4617,
+      "step": 1332
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5602665039817353,
+      "learning_rate": 4.525850586928036e-06,
+      "loss": 0.4761,
+      "step": 1333
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.569022812704669,
+      "learning_rate": 4.52474903010662e-06,
+      "loss": 0.4734,
+      "step": 1334
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5942689466703626,
+      "learning_rate": 4.523646329568771e-06,
+      "loss": 0.5025,
+      "step": 1335
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5652312432297659,
+      "learning_rate": 4.522542485937369e-06,
+      "loss": 0.4848,
+      "step": 1336
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.6230515838830604,
+      "learning_rate": 4.521437499835942e-06,
+      "loss": 0.5154,
+      "step": 1337
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5789207150950728,
+      "learning_rate": 4.52033137188866e-06,
+      "loss": 0.492,
+      "step": 1338
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5723737075235371,
+      "learning_rate": 4.519224102720341e-06,
+      "loss": 0.4898,
+      "step": 1339
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5921182647904031,
+      "learning_rate": 4.518115692956445e-06,
+      "loss": 0.4786,
+      "step": 1340
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5661075262133396,
+      "learning_rate": 4.517006143223078e-06,
+      "loss": 0.4819,
+      "step": 1341
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5639183149476357,
+      "learning_rate": 4.515895454146989e-06,
+      "loss": 0.4899,
+      "step": 1342
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5844775926241916,
+      "learning_rate": 4.514783626355571e-06,
+      "loss": 0.4749,
+      "step": 1343
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5554143190741964,
+      "learning_rate": 4.513670660476861e-06,
+      "loss": 0.4904,
+      "step": 1344
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.5730633115430818,
+      "learning_rate": 4.512556557139538e-06,
+      "loss": 0.4745,
+      "step": 1345
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5651701903378715,
+      "learning_rate": 4.5114413169729224e-06,
+      "loss": 0.4944,
+      "step": 1346
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5926181506414837,
+      "learning_rate": 4.51032494060698e-06,
+      "loss": 0.5002,
+      "step": 1347
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5603459457356793,
+      "learning_rate": 4.509207428672313e-06,
+      "loss": 0.4831,
+      "step": 1348
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5864508189913115,
+      "learning_rate": 4.508088781800172e-06,
+      "loss": 0.4759,
+      "step": 1349
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 0.4873812198638916,
+      "eval_runtime": 6917.3872,
+      "eval_samples_per_second": 41.98,
+      "eval_steps_per_second": 2.099,
+      "step": 1349
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5924744693021977,
+      "learning_rate": 4.506969000622443e-06,
+      "loss": 0.4916,
+      "step": 1350
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7833760299298924,
+      "learning_rate": 4.5058480857716554e-06,
+      "loss": 0.4883,
+      "step": 1351
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.6032529206566636,
+      "learning_rate": 4.504726037880978e-06,
+      "loss": 0.4588,
+      "step": 1352
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5651861924712768,
+      "learning_rate": 4.5036028575842215e-06,
+      "loss": 0.4935,
+      "step": 1353
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5811213046043772,
+      "learning_rate": 4.502478545515833e-06,
+      "loss": 0.4927,
+      "step": 1354
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5602131651168747,
+      "learning_rate": 4.501353102310901e-06,
+      "loss": 0.5045,
+      "step": 1355
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5609065838079279,
+      "learning_rate": 4.500226528605154e-06,
+      "loss": 0.5024,
+      "step": 1356
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5926162131260848,
+      "learning_rate": 4.499098825034956e-06,
+      "loss": 0.5022,
+      "step": 1357
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.581609311231316,
+      "learning_rate": 4.497969992237312e-06,
+      "loss": 0.5069,
+      "step": 1358
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5734162550066819,
+      "learning_rate": 4.496840030849864e-06,
+      "loss": 0.4718,
+      "step": 1359
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5832809240820751,
+      "learning_rate": 4.49570894151089e-06,
+      "loss": 0.4884,
+      "step": 1360
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5845131849772652,
+      "learning_rate": 4.494576724859307e-06,
+      "loss": 0.5017,
+      "step": 1361
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5917877188606556,
+      "learning_rate": 4.49344338153467e-06,
+      "loss": 0.5087,
+      "step": 1362
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5336678295406534,
+      "learning_rate": 4.492308912177166e-06,
+      "loss": 0.5018,
+      "step": 1363
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5351847391697913,
+      "learning_rate": 4.491173317427622e-06,
+      "loss": 0.4747,
+      "step": 1364
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.598617114856617,
+      "learning_rate": 4.490036597927499e-06,
+      "loss": 0.4894,
+      "step": 1365
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5814612225515823,
+      "learning_rate": 4.488898754318894e-06,
+      "loss": 0.5018,
+      "step": 1366
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5887096610888417,
+      "learning_rate": 4.48775978724454e-06,
+      "loss": 0.4913,
+      "step": 1367
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5831429888356454,
+      "learning_rate": 4.4866196973478e-06,
+      "loss": 0.4789,
+      "step": 1368
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.5920011899513401,
+      "learning_rate": 4.485478485272678e-06,
+      "loss": 0.4933,
+      "step": 1369
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5646646883743509,
+      "learning_rate": 4.484336151663807e-06,
+      "loss": 0.4657,
+      "step": 1370
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5498567442056715,
+      "learning_rate": 4.483192697166455e-06,
+      "loss": 0.4762,
+      "step": 1371
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5802015082280455,
+      "learning_rate": 4.482048122426523e-06,
+      "loss": 0.498,
+      "step": 1372
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5960641848798551,
+      "learning_rate": 4.480902428090546e-06,
+      "loss": 0.4868,
+      "step": 1373
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5951553770712328,
+      "learning_rate": 4.4797556148056884e-06,
+      "loss": 0.4796,
+      "step": 1374
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5704287014934469,
+      "learning_rate": 4.47860768321975e-06,
+      "loss": 0.519,
+      "step": 1375
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5911943453713712,
+      "learning_rate": 4.477458633981161e-06,
+      "loss": 0.4969,
+      "step": 1376
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5706191668874873,
+      "learning_rate": 4.476308467738982e-06,
+      "loss": 0.4754,
+      "step": 1377
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5941395696773826,
+      "learning_rate": 4.4751571851429054e-06,
+      "loss": 0.4993,
+      "step": 1378
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5947043585408122,
+      "learning_rate": 4.474004786843256e-06,
+      "loss": 0.5004,
+      "step": 1379
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5573964245620489,
+      "learning_rate": 4.472851273490985e-06,
+      "loss": 0.4914,
+      "step": 1380
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5600880890049603,
+      "learning_rate": 4.471696645737675e-06,
+      "loss": 0.4713,
+      "step": 1381
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6082765235407795,
+      "learning_rate": 4.470540904235541e-06,
+      "loss": 0.4619,
+      "step": 1382
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.579769492076148,
+      "learning_rate": 4.469384049637423e-06,
+      "loss": 0.5104,
+      "step": 1383
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5594418549752033,
+      "learning_rate": 4.468226082596792e-06,
+      "loss": 0.4688,
+      "step": 1384
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6322690165548234,
+      "learning_rate": 4.467067003767745e-06,
+      "loss": 0.4867,
+      "step": 1385
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5935187981482064,
+      "learning_rate": 4.465906813805012e-06,
+      "loss": 0.5181,
+      "step": 1386
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5944767421902712,
+      "learning_rate": 4.464745513363945e-06,
+      "loss": 0.4891,
+      "step": 1387
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5765100142428179,
+      "learning_rate": 4.463583103100527e-06,
+      "loss": 0.4829,
+      "step": 1388
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6169183106704456,
+      "learning_rate": 4.462419583671366e-06,
+      "loss": 0.4845,
+      "step": 1389
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5803513203340209,
+      "learning_rate": 4.4612549557336975e-06,
+      "loss": 0.4658,
+      "step": 1390
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.562253781436203,
+      "learning_rate": 4.460089219945383e-06,
+      "loss": 0.4771,
+      "step": 1391
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.5584896620934483,
+      "learning_rate": 4.458922376964909e-06,
+      "loss": 0.452,
+      "step": 1392
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5894231534862258,
+      "learning_rate": 4.457754427451389e-06,
+      "loss": 0.4713,
+      "step": 1393
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5847332266508826,
+      "learning_rate": 4.456585372064559e-06,
+      "loss": 0.4912,
+      "step": 1394
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5590333023061157,
+      "learning_rate": 4.455415211464783e-06,
+      "loss": 0.4849,
+      "step": 1395
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5390426978695757,
+      "learning_rate": 4.454243946313047e-06,
+      "loss": 0.5039,
+      "step": 1396
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5719225582437959,
+      "learning_rate": 4.453071577270961e-06,
+      "loss": 0.4779,
+      "step": 1397
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5850243105245156,
+      "learning_rate": 4.451898105000759e-06,
+      "loss": 0.4898,
+      "step": 1398
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5324711572140098,
+      "learning_rate": 4.450723530165299e-06,
+      "loss": 0.4757,
+      "step": 1399
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5560225385184383,
+      "learning_rate": 4.449547853428061e-06,
+      "loss": 0.4883,
+      "step": 1400
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5485424264260553,
+      "learning_rate": 4.448371075453147e-06,
+      "loss": 0.4701,
+      "step": 1401
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5667730723100881,
+      "learning_rate": 4.4471931969052816e-06,
+      "loss": 0.492,
+      "step": 1402
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6246169641223871,
+      "learning_rate": 4.446014218449811e-06,
+      "loss": 0.4884,
+      "step": 1403
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5799291856487485,
+      "learning_rate": 4.444834140752702e-06,
+      "loss": 0.4933,
+      "step": 1404
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5851580198155035,
+      "learning_rate": 4.443652964480544e-06,
+      "loss": 0.4821,
+      "step": 1405
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6056180941381275,
+      "learning_rate": 4.442470690300546e-06,
+      "loss": 0.4947,
+      "step": 1406
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5758464475466127,
+      "learning_rate": 4.441287318880537e-06,
+      "loss": 0.5095,
+      "step": 1407
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6109440245331582,
+      "learning_rate": 4.4401028508889645e-06,
+      "loss": 0.4702,
+      "step": 1408
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5763086784688934,
+      "learning_rate": 4.4389172869949e-06,
+      "loss": 0.4869,
+      "step": 1409
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6069906437238171,
+      "learning_rate": 4.437730627868028e-06,
+      "loss": 0.4947,
+      "step": 1410
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5927086729110405,
+      "learning_rate": 4.4365428741786554e-06,
+      "loss": 0.4606,
+      "step": 1411
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6090925352724758,
+      "learning_rate": 4.435354026597707e-06,
+      "loss": 0.4914,
+      "step": 1412
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5746470686497894,
+      "learning_rate": 4.434164085796724e-06,
+      "loss": 0.4891,
+      "step": 1413
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6111669276029154,
+      "learning_rate": 4.432973052447868e-06,
+      "loss": 0.4752,
+      "step": 1414
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5736428596336413,
+      "learning_rate": 4.4317809272239145e-06,
+      "loss": 0.4849,
+      "step": 1415
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5279491616141355,
+      "learning_rate": 4.430587710798257e-06,
+      "loss": 0.4732,
+      "step": 1416
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5799544850676969,
+      "learning_rate": 4.429393403844906e-06,
+      "loss": 0.5072,
+      "step": 1417
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5345098644918358,
+      "learning_rate": 4.428198007038489e-06,
+      "loss": 0.4731,
+      "step": 1418
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5979310441080966,
+      "learning_rate": 4.427001521054245e-06,
+      "loss": 0.4888,
+      "step": 1419
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.580732593585383,
+      "learning_rate": 4.425803946568033e-06,
+      "loss": 0.4849,
+      "step": 1420
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 0.4855799674987793,
+      "eval_runtime": 6914.623,
+      "eval_samples_per_second": 41.996,
+      "eval_steps_per_second": 2.1,
+      "step": 1420
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5404083046391809,
+      "learning_rate": 4.424605284256323e-06,
+      "loss": 0.4898,
+      "step": 1421
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5484706120901545,
+      "learning_rate": 4.423405534796204e-06,
+      "loss": 0.4973,
+      "step": 1422
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5788605549686905,
+      "learning_rate": 4.422204698865374e-06,
+      "loss": 0.4837,
+      "step": 1423
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6106952748012678,
+      "learning_rate": 4.421002777142148e-06,
+      "loss": 0.5014,
+      "step": 1424
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5448876773549647,
+      "learning_rate": 4.419799770305453e-06,
+      "loss": 0.4775,
+      "step": 1425
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5628932799070395,
+      "learning_rate": 4.41859567903483e-06,
+      "loss": 0.4877,
+      "step": 1426
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.58401703660086,
+      "learning_rate": 4.417390504010432e-06,
+      "loss": 0.5241,
+      "step": 1427
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5916298435437614,
+      "learning_rate": 4.416184245913022e-06,
+      "loss": 0.4664,
+      "step": 1428
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5713018784420761,
+      "learning_rate": 4.41497690542398e-06,
+      "loss": 0.5127,
+      "step": 1429
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5997368182029638,
+      "learning_rate": 4.413768483225292e-06,
+      "loss": 0.4928,
+      "step": 1430
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5793090886150154,
+      "learning_rate": 4.4125589799995585e-06,
+      "loss": 0.4658,
+      "step": 1431
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.563370967957538,
+      "learning_rate": 4.411348396429989e-06,
+      "loss": 0.4933,
+      "step": 1432
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5541515539221555,
+      "learning_rate": 4.410136733200404e-06,
+      "loss": 0.4635,
+      "step": 1433
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6348388776122461,
+      "learning_rate": 4.4089239909952335e-06,
+      "loss": 0.5072,
+      "step": 1434
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5613188088037165,
+      "learning_rate": 4.407710170499517e-06,
+      "loss": 0.4722,
+      "step": 1435
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5702782150061315,
+      "learning_rate": 4.406495272398903e-06,
+      "loss": 0.4601,
+      "step": 1436
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5706119901840532,
+      "learning_rate": 4.405279297379648e-06,
+      "loss": 0.4873,
+      "step": 1437
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.6397976458766784,
+      "learning_rate": 4.404062246128621e-06,
+      "loss": 0.4943,
+      "step": 1438
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.5973771040772421,
+      "learning_rate": 4.4028441193332914e-06,
+      "loss": 0.5085,
+      "step": 1439
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5914091770327425,
+      "learning_rate": 4.401624917681743e-06,
+      "loss": 0.4922,
+      "step": 1440
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6055784851510099,
+      "learning_rate": 4.400404641862664e-06,
+      "loss": 0.4745,
+      "step": 1441
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5764544930402707,
+      "learning_rate": 4.399183292565347e-06,
+      "loss": 0.4857,
+      "step": 1442
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5996098894523001,
+      "learning_rate": 4.397960870479696e-06,
+      "loss": 0.4978,
+      "step": 1443
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5521355170405737,
+      "learning_rate": 4.396737376296218e-06,
+      "loss": 0.4587,
+      "step": 1444
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.59045938437032,
+      "learning_rate": 4.395512810706026e-06,
+      "loss": 0.4908,
+      "step": 1445
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.583988820333397,
+      "learning_rate": 4.394287174400838e-06,
+      "loss": 0.4869,
+      "step": 1446
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6065684858025009,
+      "learning_rate": 4.393060468072976e-06,
+      "loss": 0.4892,
+      "step": 1447
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5703024949579703,
+      "learning_rate": 4.3918326924153685e-06,
+      "loss": 0.4759,
+      "step": 1448
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5745917135551377,
+      "learning_rate": 4.390603848121546e-06,
+      "loss": 0.4869,
+      "step": 1449
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5720705945602651,
+      "learning_rate": 4.3893739358856465e-06,
+      "loss": 0.4663,
+      "step": 1450
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5748475515933826,
+      "learning_rate": 4.388142956402405e-06,
+      "loss": 0.4981,
+      "step": 1451
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.1756158657989677,
+      "learning_rate": 4.3869109103671645e-06,
+      "loss": 0.5071,
+      "step": 1452
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5696921062936988,
+      "learning_rate": 4.385677798475868e-06,
+      "loss": 0.4764,
+      "step": 1453
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.557690101161529,
+      "learning_rate": 4.384443621425062e-06,
+      "loss": 0.5103,
+      "step": 1454
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5542852620725858,
+      "learning_rate": 4.383208379911893e-06,
+      "loss": 0.4835,
+      "step": 1455
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5402197457616243,
+      "learning_rate": 4.38197207463411e-06,
+      "loss": 0.4963,
+      "step": 1456
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.584415569246153,
+      "learning_rate": 4.380734706290063e-06,
+      "loss": 0.4835,
+      "step": 1457
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5483678859464396,
+      "learning_rate": 4.379496275578701e-06,
+      "loss": 0.489,
+      "step": 1458
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6061260112046732,
+      "learning_rate": 4.378256783199575e-06,
+      "loss": 0.507,
+      "step": 1459
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5542441873465928,
+      "learning_rate": 4.377016229852836e-06,
+      "loss": 0.4915,
+      "step": 1460
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.5545139083535812,
+      "learning_rate": 4.375774616239231e-06,
+      "loss": 0.5185,
+      "step": 1461
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.552239627065928,
+      "learning_rate": 4.374531943060109e-06,
+      "loss": 0.4783,
+      "step": 1462
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5335369265994653,
+      "learning_rate": 4.373288211017418e-06,
+      "loss": 0.454,
+      "step": 1463
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5454066478470697,
+      "learning_rate": 4.3720434208137015e-06,
+      "loss": 0.4824,
+      "step": 1464
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6037061519362923,
+      "learning_rate": 4.370797573152101e-06,
+      "loss": 0.4977,
+      "step": 1465
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5805121340778033,
+      "learning_rate": 4.369550668736358e-06,
+      "loss": 0.4894,
+      "step": 1466
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.551429386005362,
+      "learning_rate": 4.3683027082708085e-06,
+      "loss": 0.4612,
+      "step": 1467
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5772151081375165,
+      "learning_rate": 4.3670536924603855e-06,
+      "loss": 0.4893,
+      "step": 1468
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5600194793378279,
+      "learning_rate": 4.365803622010618e-06,
+      "loss": 0.4956,
+      "step": 1469
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5530119828515296,
+      "learning_rate": 4.364552497627632e-06,
+      "loss": 0.4612,
+      "step": 1470
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5805548135765888,
+      "learning_rate": 4.363300320018148e-06,
+      "loss": 0.4659,
+      "step": 1471
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5501610854660056,
+      "learning_rate": 4.36204708988948e-06,
+      "loss": 0.4842,
+      "step": 1472
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5997289615402112,
+      "learning_rate": 4.36079280794954e-06,
+      "loss": 0.52,
+      "step": 1473
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5805877442370376,
+      "learning_rate": 4.359537474906831e-06,
+      "loss": 0.487,
+      "step": 1474
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.558823842875777,
+      "learning_rate": 4.35828109147045e-06,
+      "loss": 0.4664,
+      "step": 1475
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5889535016987915,
+      "learning_rate": 4.357023658350089e-06,
+      "loss": 0.4989,
+      "step": 1476
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.6277448104948404,
+      "learning_rate": 4.3557651762560316e-06,
+      "loss": 0.5118,
+      "step": 1477
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5679808700739567,
+      "learning_rate": 4.3545056458991556e-06,
+      "loss": 0.463,
+      "step": 1478
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5962236218865731,
+      "learning_rate": 4.353245067990928e-06,
+      "loss": 0.4798,
+      "step": 1479
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5808142134901263,
+      "learning_rate": 4.3519834432434095e-06,
+      "loss": 0.514,
+      "step": 1480
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5889258035229223,
+      "learning_rate": 4.350720772369252e-06,
+      "loss": 0.5016,
+      "step": 1481
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5618362131992736,
+      "learning_rate": 4.3494570560817e-06,
+      "loss": 0.4861,
+      "step": 1482
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5693422718547901,
+      "learning_rate": 4.348192295094585e-06,
+      "loss": 0.4737,
+      "step": 1483
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5554250264653801,
+      "learning_rate": 4.346926490122329e-06,
+      "loss": 0.4959,
+      "step": 1484
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5541704411153275,
+      "learning_rate": 4.345659641879948e-06,
+      "loss": 0.489,
+      "step": 1485
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.5749556736212126,
+      "learning_rate": 4.344391751083043e-06,
+      "loss": 0.4919,
+      "step": 1486
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5815536068592329,
+      "learning_rate": 4.343122818447804e-06,
+      "loss": 0.478,
+      "step": 1487
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5807918654560094,
+      "learning_rate": 4.341852844691012e-06,
+      "loss": 0.4893,
+      "step": 1488
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5772854343547457,
+      "learning_rate": 4.340581830530036e-06,
+      "loss": 0.4883,
+      "step": 1489
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6083000277946388,
+      "learning_rate": 4.33930977668283e-06,
+      "loss": 0.5033,
+      "step": 1490
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5740926343714129,
+      "learning_rate": 4.338036683867936e-06,
+      "loss": 0.4943,
+      "step": 1491
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 0.4840275049209595,
+      "eval_runtime": 6927.0119,
+      "eval_samples_per_second": 41.921,
+      "eval_steps_per_second": 2.096,
+      "step": 1491
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5684026018424353,
+      "learning_rate": 4.336762552804485e-06,
+      "loss": 0.4718,
+      "step": 1492
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.562055744831237,
+      "learning_rate": 4.335487384212194e-06,
+      "loss": 0.5053,
+      "step": 1493
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5816786952791797,
+      "learning_rate": 4.334211178811364e-06,
+      "loss": 0.4864,
+      "step": 1494
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.543579418092536,
+      "learning_rate": 4.332933937322883e-06,
+      "loss": 0.4937,
+      "step": 1495
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.576270682270156,
+      "learning_rate": 4.331655660468224e-06,
+      "loss": 0.4587,
+      "step": 1496
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5770731512199584,
+      "learning_rate": 4.330376348969445e-06,
+      "loss": 0.4996,
+      "step": 1497
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5523724826628789,
+      "learning_rate": 4.329096003549189e-06,
+      "loss": 0.4748,
+      "step": 1498
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.603518080716257,
+      "learning_rate": 4.3278146249306825e-06,
+      "loss": 0.4974,
+      "step": 1499
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5503858002021573,
+      "learning_rate": 4.326532213837735e-06,
+      "loss": 0.4927,
+      "step": 1500
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5464005046510432,
+      "learning_rate": 4.325248770994741e-06,
+      "loss": 0.4834,
+      "step": 1501
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.591865173361713,
+      "learning_rate": 4.323964297126675e-06,
+      "loss": 0.4651,
+      "step": 1502
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5839558098791746,
+      "learning_rate": 4.3226787929590965e-06,
+      "loss": 0.4726,
+      "step": 1503
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5870418970742052,
+      "learning_rate": 4.3213922592181455e-06,
+      "loss": 0.4996,
+      "step": 1504
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5949338157726647,
+      "learning_rate": 4.320104696630544e-06,
+      "loss": 0.4882,
+      "step": 1505
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5822457164207244,
+      "learning_rate": 4.318816105923596e-06,
+      "loss": 0.4999,
+      "step": 1506
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.6042290491386848,
+      "learning_rate": 4.317526487825185e-06,
+      "loss": 0.4743,
+      "step": 1507
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.576454722720242,
+      "learning_rate": 4.316235843063775e-06,
+      "loss": 0.5038,
+      "step": 1508
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5715912329774304,
+      "learning_rate": 4.314944172368411e-06,
+      "loss": 0.5029,
+      "step": 1509
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5530105706856633,
+      "learning_rate": 4.3136514764687155e-06,
+      "loss": 0.4861,
+      "step": 1510
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5500767838459567,
+      "learning_rate": 4.312357756094892e-06,
+      "loss": 0.4858,
+      "step": 1511
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.54693416081482,
+      "learning_rate": 4.311063011977723e-06,
+      "loss": 0.5017,
+      "step": 1512
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5419499341904167,
+      "learning_rate": 4.309767244848567e-06,
+      "loss": 0.4891,
+      "step": 1513
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5614950534369257,
+      "learning_rate": 4.308470455439362e-06,
+      "loss": 0.4971,
+      "step": 1514
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5349156359990223,
+      "learning_rate": 4.3071726444826244e-06,
+      "loss": 0.4777,
+      "step": 1515
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6003648657933892,
+      "learning_rate": 4.305873812711445e-06,
+      "loss": 0.4883,
+      "step": 1516
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5561939266607488,
+      "learning_rate": 4.304573960859493e-06,
+      "loss": 0.477,
+      "step": 1517
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5772579264539893,
+      "learning_rate": 4.303273089661013e-06,
+      "loss": 0.4636,
+      "step": 1518
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5555633091494279,
+      "learning_rate": 4.301971199850826e-06,
+      "loss": 0.4598,
+      "step": 1519
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5909063796287866,
+      "learning_rate": 4.300668292164329e-06,
+      "loss": 0.4704,
+      "step": 1520
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5321519349592307,
+      "learning_rate": 4.299364367337493e-06,
+      "loss": 0.4726,
+      "step": 1521
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5948749563634228,
+      "learning_rate": 4.298059426106864e-06,
+      "loss": 0.5042,
+      "step": 1522
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5590529380799898,
+      "learning_rate": 4.29675346920956e-06,
+      "loss": 0.4771,
+      "step": 1523
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.573778432963487,
+      "learning_rate": 4.2954464973832765e-06,
+      "loss": 0.4841,
+      "step": 1524
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5531613083984294,
+      "learning_rate": 4.29413851136628e-06,
+      "loss": 0.4695,
+      "step": 1525
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5612844436545967,
+      "learning_rate": 4.292829511897409e-06,
+      "loss": 0.4804,
+      "step": 1526
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.7125770424238018,
+      "learning_rate": 4.2915194997160774e-06,
+      "loss": 0.4788,
+      "step": 1527
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.6254458428657401,
+      "learning_rate": 4.2902084755622685e-06,
+      "loss": 0.4875,
+      "step": 1528
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5699982607183243,
+      "learning_rate": 4.288896440176539e-06,
+      "loss": 0.5127,
+      "step": 1529
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5359705267411138,
+      "learning_rate": 4.287583394300016e-06,
+      "loss": 0.4592,
+      "step": 1530
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.573098209766848,
+      "learning_rate": 4.286269338674396e-06,
+      "loss": 0.4872,
+      "step": 1531
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.582145535892815,
+      "learning_rate": 4.284954274041949e-06,
+      "loss": 0.4932,
+      "step": 1532
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.569146953738728,
+      "learning_rate": 4.283638201145512e-06,
+      "loss": 0.4842,
+      "step": 1533
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6060270899238028,
+      "learning_rate": 4.282321120728493e-06,
+      "loss": 0.5012,
+      "step": 1534
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5636460053991561,
+      "learning_rate": 4.28100303353487e-06,
+      "loss": 0.4912,
+      "step": 1535
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5594354604572797,
+      "learning_rate": 4.279683940309187e-06,
+      "loss": 0.4742,
+      "step": 1536
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6148223579588533,
+      "learning_rate": 4.278363841796559e-06,
+      "loss": 0.4705,
+      "step": 1537
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5510025520117999,
+      "learning_rate": 4.277042738742668e-06,
+      "loss": 0.4813,
+      "step": 1538
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6189865762055806,
+      "learning_rate": 4.2757206318937625e-06,
+      "loss": 0.5076,
+      "step": 1539
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5549678033446015,
+      "learning_rate": 4.274397521996658e-06,
+      "loss": 0.5005,
+      "step": 1540
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5444972036743465,
+      "learning_rate": 4.27307340979874e-06,
+      "loss": 0.4738,
+      "step": 1541
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5863310201010252,
+      "learning_rate": 4.271748296047953e-06,
+      "loss": 0.5036,
+      "step": 1542
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5834850598785314,
+      "learning_rate": 4.270422181492815e-06,
+      "loss": 0.4789,
+      "step": 1543
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5617538920507555,
+      "learning_rate": 4.269095066882406e-06,
+      "loss": 0.4632,
+      "step": 1544
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5907455073189339,
+      "learning_rate": 4.267766952966369e-06,
+      "loss": 0.4953,
+      "step": 1545
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5983165451822997,
+      "learning_rate": 4.266437840494915e-06,
+      "loss": 0.4796,
+      "step": 1546
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.569812624676123,
+      "learning_rate": 4.265107730218817e-06,
+      "loss": 0.4514,
+      "step": 1547
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5847228353394706,
+      "learning_rate": 4.2637766228894115e-06,
+      "loss": 0.4656,
+      "step": 1548
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5925087756170236,
+      "learning_rate": 4.2624445192585994e-06,
+      "loss": 0.5031,
+      "step": 1549
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5703001753744149,
+      "learning_rate": 4.261111420078844e-06,
+      "loss": 0.4648,
+      "step": 1550
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5777371677357935,
+      "learning_rate": 4.259777326103169e-06,
+      "loss": 0.5158,
+      "step": 1551
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6186779369597184,
+      "learning_rate": 4.258442238085164e-06,
+      "loss": 0.4643,
+      "step": 1552
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.64721013399992,
+      "learning_rate": 4.2571061567789766e-06,
+      "loss": 0.4947,
+      "step": 1553
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.536482893670931,
+      "learning_rate": 4.255769082939316e-06,
+      "loss": 0.4661,
+      "step": 1554
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.5970665352812153,
+      "learning_rate": 4.2544310173214546e-06,
+      "loss": 0.4676,
+      "step": 1555
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6564985474844536,
+      "learning_rate": 4.253091960681222e-06,
+      "loss": 0.4875,
+      "step": 1556
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5742936073076695,
+      "learning_rate": 4.251751913775009e-06,
+      "loss": 0.4685,
+      "step": 1557
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5791371405621296,
+      "learning_rate": 4.250410877359765e-06,
+      "loss": 0.4816,
+      "step": 1558
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5879477020083694,
+      "learning_rate": 4.2490688521930005e-06,
+      "loss": 0.4648,
+      "step": 1559
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5644425455654086,
+      "learning_rate": 4.247725839032781e-06,
+      "loss": 0.4979,
+      "step": 1560
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5816219717795591,
+      "learning_rate": 4.246381838637733e-06,
+      "loss": 0.4767,
+      "step": 1561
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6181448897628907,
+      "learning_rate": 4.245036851767039e-06,
+      "loss": 0.5092,
+      "step": 1562
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 0.4825836420059204,
+      "eval_runtime": 6927.1849,
+      "eval_samples_per_second": 41.92,
+      "eval_steps_per_second": 2.096,
+      "step": 1562
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6319339756032352,
+      "learning_rate": 4.243690879180441e-06,
+      "loss": 0.4999,
+      "step": 1563
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5766409130292584,
+      "learning_rate": 4.242343921638235e-06,
+      "loss": 0.4912,
+      "step": 1564
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6169305231670678,
+      "learning_rate": 4.240995979901273e-06,
+      "loss": 0.5037,
+      "step": 1565
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6002589034049418,
+      "learning_rate": 4.239647054730966e-06,
+      "loss": 0.506,
+      "step": 1566
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6023521389889079,
+      "learning_rate": 4.238297146889281e-06,
+      "loss": 0.5257,
+      "step": 1567
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5158122792594769,
+      "learning_rate": 4.236946257138734e-06,
+      "loss": 0.441,
+      "step": 1568
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5495762183794795,
+      "learning_rate": 4.235594386242403e-06,
+      "loss": 0.4743,
+      "step": 1569
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6628519895823336,
+      "learning_rate": 4.234241534963916e-06,
+      "loss": 0.5272,
+      "step": 1570
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5959806073232053,
+      "learning_rate": 4.232887704067455e-06,
+      "loss": 0.4951,
+      "step": 1571
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5572508686653875,
+      "learning_rate": 4.231532894317757e-06,
+      "loss": 0.4923,
+      "step": 1572
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5899131939103285,
+      "learning_rate": 4.23017710648011e-06,
+      "loss": 0.4837,
+      "step": 1573
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.6064850421779246,
+      "learning_rate": 4.228820341320356e-06,
+      "loss": 0.5005,
+      "step": 1574
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5780537164257747,
+      "learning_rate": 4.227462599604889e-06,
+      "loss": 0.4785,
+      "step": 1575
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5810376085859065,
+      "learning_rate": 4.226103882100654e-06,
+      "loss": 0.4714,
+      "step": 1576
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5909147691169829,
+      "learning_rate": 4.224744189575148e-06,
+      "loss": 0.5021,
+      "step": 1577
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5942835635277367,
+      "learning_rate": 4.2233835227964145e-06,
+      "loss": 0.4924,
+      "step": 1578
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.5618720908082389,
+      "learning_rate": 4.222021882533056e-06,
+      "loss": 0.4724,
+      "step": 1579
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5522683879852052,
+      "learning_rate": 4.220659269554217e-06,
+      "loss": 0.476,
+      "step": 1580
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5626316433669417,
+      "learning_rate": 4.219295684629595e-06,
+      "loss": 0.5038,
+      "step": 1581
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5832875815941423,
+      "learning_rate": 4.217931128529436e-06,
+      "loss": 0.4942,
+      "step": 1582
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5515392841114919,
+      "learning_rate": 4.216565602024533e-06,
+      "loss": 0.4832,
+      "step": 1583
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6821325914674553,
+      "learning_rate": 4.2151991058862314e-06,
+      "loss": 0.4696,
+      "step": 1584
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5918911213299775,
+      "learning_rate": 4.21383164088642e-06,
+      "loss": 0.4981,
+      "step": 1585
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.567039525371673,
+      "learning_rate": 4.212463207797535e-06,
+      "loss": 0.474,
+      "step": 1586
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5743234762161183,
+      "learning_rate": 4.211093807392562e-06,
+      "loss": 0.4675,
+      "step": 1587
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5855603678522909,
+      "learning_rate": 4.209723440445032e-06,
+      "loss": 0.4796,
+      "step": 1588
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5500461666318762,
+      "learning_rate": 4.208352107729021e-06,
+      "loss": 0.4844,
+      "step": 1589
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5791356596651714,
+      "learning_rate": 4.206979810019153e-06,
+      "loss": 0.4769,
+      "step": 1590
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5943166707550008,
+      "learning_rate": 4.205606548090593e-06,
+      "loss": 0.4842,
+      "step": 1591
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5765468748074813,
+      "learning_rate": 4.204232322719055e-06,
+      "loss": 0.4877,
+      "step": 1592
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5705538264405167,
+      "learning_rate": 4.202857134680795e-06,
+      "loss": 0.4848,
+      "step": 1593
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.550378639192882,
+      "learning_rate": 4.201480984752612e-06,
+      "loss": 0.4709,
+      "step": 1594
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5843384273029046,
+      "learning_rate": 4.20010387371185e-06,
+      "loss": 0.5116,
+      "step": 1595
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5430286998603538,
+      "learning_rate": 4.198725802336395e-06,
+      "loss": 0.4755,
+      "step": 1596
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5613235833509066,
+      "learning_rate": 4.197346771404677e-06,
+      "loss": 0.483,
+      "step": 1597
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5431464537859271,
+      "learning_rate": 4.1959667816956654e-06,
+      "loss": 0.4954,
+      "step": 1598
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5816367873246644,
+      "learning_rate": 4.194585833988873e-06,
+      "loss": 0.5123,
+      "step": 1599
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5842017694622618,
+      "learning_rate": 4.1932039290643534e-06,
+      "loss": 0.4751,
+      "step": 1600
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5983057704757485,
+      "learning_rate": 4.191821067702701e-06,
+      "loss": 0.476,
+      "step": 1601
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.5785359587175982,
+      "learning_rate": 4.190437250685049e-06,
+      "loss": 0.4882,
+      "step": 1602
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.6048435561330398,
+      "learning_rate": 4.189052478793074e-06,
+      "loss": 0.5045,
+      "step": 1603
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5886694967565423,
+      "learning_rate": 4.187666752808987e-06,
+      "loss": 0.4809,
+      "step": 1604
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5662837567043483,
+      "learning_rate": 4.186280073515543e-06,
+      "loss": 0.508,
+      "step": 1605
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5507517273321046,
+      "learning_rate": 4.184892441696031e-06,
+      "loss": 0.4574,
+      "step": 1606
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5506339868366799,
+      "learning_rate": 4.183503858134283e-06,
+      "loss": 0.4983,
+      "step": 1607
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.551124570714273,
+      "learning_rate": 4.182114323614662e-06,
+      "loss": 0.4964,
+      "step": 1608
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5632058519052752,
+      "learning_rate": 4.180723838922076e-06,
+      "loss": 0.5091,
+      "step": 1609
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5523436692520042,
+      "learning_rate": 4.179332404841963e-06,
+      "loss": 0.484,
+      "step": 1610
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6137884481921679,
+      "learning_rate": 4.177940022160299e-06,
+      "loss": 0.4796,
+      "step": 1611
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6400564535674431,
+      "learning_rate": 4.1765466916636e-06,
+      "loss": 0.4813,
+      "step": 1612
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6137552397160586,
+      "learning_rate": 4.175152414138911e-06,
+      "loss": 0.5128,
+      "step": 1613
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5878395349717804,
+      "learning_rate": 4.173757190373817e-06,
+      "loss": 0.4921,
+      "step": 1614
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.638177925386608,
+      "learning_rate": 4.172361021156436e-06,
+      "loss": 0.4697,
+      "step": 1615
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.55678238023355,
+      "learning_rate": 4.170963907275418e-06,
+      "loss": 0.456,
+      "step": 1616
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5856822237716187,
+      "learning_rate": 4.169565849519949e-06,
+      "loss": 0.4762,
+      "step": 1617
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6465303799069492,
+      "learning_rate": 4.1681668486797475e-06,
+      "loss": 0.4964,
+      "step": 1618
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.6003634481531572,
+      "learning_rate": 4.166766905545064e-06,
+      "loss": 0.4864,
+      "step": 1619
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5916487532205914,
+      "learning_rate": 4.1653660209066835e-06,
+      "loss": 0.5002,
+      "step": 1620
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5898166480233019,
+      "learning_rate": 4.1639641955559205e-06,
+      "loss": 0.4827,
+      "step": 1621
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5866832054250863,
+      "learning_rate": 4.162561430284621e-06,
+      "loss": 0.5073,
+      "step": 1622
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5553096089138998,
+      "learning_rate": 4.161157725885163e-06,
+      "loss": 0.4507,
+      "step": 1623
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5797616225893526,
+      "learning_rate": 4.159753083150455e-06,
+      "loss": 0.4781,
+      "step": 1624
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5782876181988972,
+      "learning_rate": 4.158347502873933e-06,
+      "loss": 0.486,
+      "step": 1625
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5593458949862828,
+      "learning_rate": 4.156940985849568e-06,
+      "loss": 0.4722,
+      "step": 1626
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5566363926907538,
+      "learning_rate": 4.155533532871855e-06,
+      "loss": 0.471,
+      "step": 1627
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5687649356444076,
+      "learning_rate": 4.154125144735819e-06,
+      "loss": 0.4593,
+      "step": 1628
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5805813995919267,
+      "learning_rate": 4.1527158222370134e-06,
+      "loss": 0.4825,
+      "step": 1629
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5698382648767448,
+      "learning_rate": 4.151305566171521e-06,
+      "loss": 0.4744,
+      "step": 1630
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5478296421305997,
+      "learning_rate": 4.149894377335951e-06,
+      "loss": 0.4708,
+      "step": 1631
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5579153962795045,
+      "learning_rate": 4.148482256527438e-06,
+      "loss": 0.4838,
+      "step": 1632
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5487271513915988,
+      "learning_rate": 4.147069204543645e-06,
+      "loss": 0.4538,
+      "step": 1633
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 0.4806346595287323,
+      "eval_runtime": 6927.3711,
+      "eval_samples_per_second": 41.919,
+      "eval_steps_per_second": 2.096,
+      "step": 1633
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5571550450326609,
+      "learning_rate": 4.14565522218276e-06,
+      "loss": 0.4696,
+      "step": 1634
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.573294746763072,
+      "learning_rate": 4.144240310243496e-06,
+      "loss": 0.476,
+      "step": 1635
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5571754212391226,
+      "learning_rate": 4.142824469525093e-06,
+      "loss": 0.4823,
+      "step": 1636
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5921760399908439,
+      "learning_rate": 4.1414077008273134e-06,
+      "loss": 0.4787,
+      "step": 1637
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5663213930738107,
+      "learning_rate": 4.139990004950446e-06,
+      "loss": 0.4837,
+      "step": 1638
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6090434172084536,
+      "learning_rate": 4.138571382695301e-06,
+      "loss": 0.472,
+      "step": 1639
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6277842926973998,
+      "learning_rate": 4.137151834863213e-06,
+      "loss": 0.4627,
+      "step": 1640
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5518929513736244,
+      "learning_rate": 4.13573136225604e-06,
+      "loss": 0.4815,
+      "step": 1641
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5750076801420694,
+      "learning_rate": 4.1343099656761635e-06,
+      "loss": 0.4713,
+      "step": 1642
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5636682836359242,
+      "learning_rate": 4.132887645926482e-06,
+      "loss": 0.4739,
+      "step": 1643
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5582878941115875,
+      "learning_rate": 4.1314644038104215e-06,
+      "loss": 0.4846,
+      "step": 1644
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5355410249521172,
+      "learning_rate": 4.130040240131925e-06,
+      "loss": 0.487,
+      "step": 1645
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.6050538286426506,
+      "learning_rate": 4.128615155695458e-06,
+      "loss": 0.4791,
+      "step": 1646
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.551251347571686,
+      "learning_rate": 4.127189151306004e-06,
+      "loss": 0.4636,
+      "step": 1647
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5586211992866543,
+      "learning_rate": 4.12576222776907e-06,
+      "loss": 0.5068,
+      "step": 1648
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.5541371767356188,
+      "learning_rate": 4.124334385890678e-06,
+      "loss": 0.4666,
+      "step": 1649
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.597101794217058,
+      "learning_rate": 4.122905626477371e-06,
+      "loss": 0.4905,
+      "step": 1650
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5675341113402599,
+      "learning_rate": 4.121475950336209e-06,
+      "loss": 0.5052,
+      "step": 1651
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5863054446966842,
+      "learning_rate": 4.120045358274772e-06,
+      "loss": 0.4974,
+      "step": 1652
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.553734529174061,
+      "learning_rate": 4.118613851101156e-06,
+      "loss": 0.4674,
+      "step": 1653
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5949725607990067,
+      "learning_rate": 4.117181429623973e-06,
+      "loss": 0.4659,
+      "step": 1654
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5834567071239104,
+      "learning_rate": 4.115748094652352e-06,
+      "loss": 0.4743,
+      "step": 1655
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5535876065946561,
+      "learning_rate": 4.114313846995941e-06,
+      "loss": 0.5171,
+      "step": 1656
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5838621642562685,
+      "learning_rate": 4.112878687464898e-06,
+      "loss": 0.5032,
+      "step": 1657
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5823106011343094,
+      "learning_rate": 4.111442616869901e-06,
+      "loss": 0.4829,
+      "step": 1658
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5559063219280944,
+      "learning_rate": 4.110005636022138e-06,
+      "loss": 0.4636,
+      "step": 1659
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5736354627247767,
+      "learning_rate": 4.108567745733318e-06,
+      "loss": 0.4933,
+      "step": 1660
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5690533891379516,
+      "learning_rate": 4.107128946815657e-06,
+      "loss": 0.4771,
+      "step": 1661
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5816163405766323,
+      "learning_rate": 4.105689240081886e-06,
+      "loss": 0.4599,
+      "step": 1662
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5686008337838981,
+      "learning_rate": 4.104248626345252e-06,
+      "loss": 0.4588,
+      "step": 1663
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5713712745582588,
+      "learning_rate": 4.102807106419511e-06,
+      "loss": 0.4707,
+      "step": 1664
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5568768914986018,
+      "learning_rate": 4.10136468111893e-06,
+      "loss": 0.491,
+      "step": 1665
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5680603398114878,
+      "learning_rate": 4.099921351258292e-06,
+      "loss": 0.4887,
+      "step": 1666
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.555747852519542,
+      "learning_rate": 4.098477117652887e-06,
+      "loss": 0.4678,
+      "step": 1667
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5732419584091223,
+      "learning_rate": 4.097031981118516e-06,
+      "loss": 0.5059,
+      "step": 1668
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5978764739542107,
+      "learning_rate": 4.095585942471492e-06,
+      "loss": 0.4862,
+      "step": 1669
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5641825612908864,
+      "learning_rate": 4.094139002528635e-06,
+      "loss": 0.4892,
+      "step": 1670
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.59291044888431,
+      "learning_rate": 4.092691162107277e-06,
+      "loss": 0.4689,
+      "step": 1671
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5633228261402596,
+      "learning_rate": 4.091242422025256e-06,
+      "loss": 0.4867,
+      "step": 1672
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.5446421814689791,
+      "learning_rate": 4.08979278310092e-06,
+      "loss": 0.4963,
+      "step": 1673
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5900103216285727,
+      "learning_rate": 4.088342246153123e-06,
+      "loss": 0.4784,
+      "step": 1674
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5662846183673187,
+      "learning_rate": 4.086890812001228e-06,
+      "loss": 0.4728,
+      "step": 1675
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5481824376820944,
+      "learning_rate": 4.085438481465104e-06,
+      "loss": 0.4925,
+      "step": 1676
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.603893688943784,
+      "learning_rate": 4.083985255365127e-06,
+      "loss": 0.4775,
+      "step": 1677
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5616169558946099,
+      "learning_rate": 4.082531134522176e-06,
+      "loss": 0.4836,
+      "step": 1678
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5771671536691606,
+      "learning_rate": 4.0810761197576405e-06,
+      "loss": 0.4729,
+      "step": 1679
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5356849896576423,
+      "learning_rate": 4.07962021189341e-06,
+      "loss": 0.4831,
+      "step": 1680
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5705452601718124,
+      "learning_rate": 4.078163411751882e-06,
+      "loss": 0.5036,
+      "step": 1681
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5853521037301916,
+      "learning_rate": 4.076705720155956e-06,
+      "loss": 0.4663,
+      "step": 1682
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5767304367108156,
+      "learning_rate": 4.075247137929036e-06,
+      "loss": 0.4696,
+      "step": 1683
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5469609445931426,
+      "learning_rate": 4.073787665895029e-06,
+      "loss": 0.4861,
+      "step": 1684
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5647896369896342,
+      "learning_rate": 4.0723273048783426e-06,
+      "loss": 0.4751,
+      "step": 1685
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5471961307604523,
+      "learning_rate": 4.070866055703892e-06,
+      "loss": 0.4755,
+      "step": 1686
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5409469094446197,
+      "learning_rate": 4.069403919197087e-06,
+      "loss": 0.4485,
+      "step": 1687
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5493683437426159,
+      "learning_rate": 4.067940896183843e-06,
+      "loss": 0.466,
+      "step": 1688
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6184085336074597,
+      "learning_rate": 4.0664769874905765e-06,
+      "loss": 0.4941,
+      "step": 1689
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5897131879671903,
+      "learning_rate": 4.065012193944201e-06,
+      "loss": 0.4949,
+      "step": 1690
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5476514705576049,
+      "learning_rate": 4.063546516372134e-06,
+      "loss": 0.4627,
+      "step": 1691
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5639046127555836,
+      "learning_rate": 4.0620799556022886e-06,
+      "loss": 0.507,
+      "step": 1692
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.569493917934072,
+      "learning_rate": 4.060612512463079e-06,
+      "loss": 0.4689,
+      "step": 1693
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5576467520547685,
+      "learning_rate": 4.059144187783417e-06,
+      "loss": 0.4533,
+      "step": 1694
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5785819029326567,
+      "learning_rate": 4.057674982392713e-06,
+      "loss": 0.484,
+      "step": 1695
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.5800649358100449,
+      "learning_rate": 4.056204897120875e-06,
+      "loss": 0.475,
+      "step": 1696
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5547737396991846,
+      "learning_rate": 4.054733932798306e-06,
+      "loss": 0.4617,
+      "step": 1697
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.55033684788177,
+      "learning_rate": 4.053262090255908e-06,
+      "loss": 0.4748,
+      "step": 1698
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5624353739448029,
+      "learning_rate": 4.051789370325079e-06,
+      "loss": 0.4869,
+      "step": 1699
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5452044799501949,
+      "learning_rate": 4.050315773837708e-06,
+      "loss": 0.4463,
+      "step": 1700
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5802412977647277,
+      "learning_rate": 4.048841301626188e-06,
+      "loss": 0.4821,
+      "step": 1701
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5909786502202287,
+      "learning_rate": 4.047365954523398e-06,
+      "loss": 0.4666,
+      "step": 1702
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5825689334745631,
+      "learning_rate": 4.045889733362717e-06,
+      "loss": 0.4833,
+      "step": 1703
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5429276122164189,
+      "learning_rate": 4.044412638978012e-06,
+      "loss": 0.4792,
+      "step": 1704
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 0.4791695773601532,
+      "eval_runtime": 6928.6024,
+      "eval_samples_per_second": 41.912,
+      "eval_steps_per_second": 2.096,
+      "step": 1704
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5506114214338211,
+      "learning_rate": 4.042934672203651e-06,
+      "loss": 0.4495,
+      "step": 1705
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5578625634328879,
+      "learning_rate": 4.041455833874488e-06,
+      "loss": 0.4522,
+      "step": 1706
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5795772583330789,
+      "learning_rate": 4.039976124825872e-06,
+      "loss": 0.5001,
+      "step": 1707
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5536178873684257,
+      "learning_rate": 4.038495545893643e-06,
+      "loss": 0.4803,
+      "step": 1708
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5392842977706797,
+      "learning_rate": 4.037014097914135e-06,
+      "loss": 0.4682,
+      "step": 1709
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5708019402077392,
+      "learning_rate": 4.0355317817241705e-06,
+      "loss": 0.4785,
+      "step": 1710
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6113657504825251,
+      "learning_rate": 4.034048598161061e-06,
+      "loss": 0.4797,
+      "step": 1711
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5620870459786824,
+      "learning_rate": 4.032564548062612e-06,
+      "loss": 0.472,
+      "step": 1712
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5660234184012428,
+      "learning_rate": 4.0310796322671144e-06,
+      "loss": 0.4867,
+      "step": 1713
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5428002908374318,
+      "learning_rate": 4.029593851613351e-06,
+      "loss": 0.4772,
+      "step": 1714
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5576743567400213,
+      "learning_rate": 4.028107206940592e-06,
+      "loss": 0.484,
+      "step": 1715
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5472128158032424,
+      "learning_rate": 4.0266196990885955e-06,
+      "loss": 0.5003,
+      "step": 1716
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5194449059148619,
+      "learning_rate": 4.025131328897608e-06,
+      "loss": 0.4718,
+      "step": 1717
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.5611014049931291,
+      "learning_rate": 4.023642097208362e-06,
+      "loss": 0.4812,
+      "step": 1718
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.6211118572373506,
+      "learning_rate": 4.022152004862079e-06,
+      "loss": 0.4877,
+      "step": 1719
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.541074320013028,
+      "learning_rate": 4.020661052700462e-06,
+      "loss": 0.4996,
+      "step": 1720
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5552304848575638,
+      "learning_rate": 4.019169241565704e-06,
+      "loss": 0.5153,
+      "step": 1721
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.553005883252194,
+      "learning_rate": 4.0176765723004805e-06,
+      "loss": 0.4583,
+      "step": 1722
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5609121544313215,
+      "learning_rate": 4.0161830457479555e-06,
+      "loss": 0.4524,
+      "step": 1723
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5641653969492663,
+      "learning_rate": 4.014688662751773e-06,
+      "loss": 0.4543,
+      "step": 1724
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5328244993549478,
+      "learning_rate": 4.013193424156062e-06,
+      "loss": 0.4417,
+      "step": 1725
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5561692361326591,
+      "learning_rate": 4.011697330805436e-06,
+      "loss": 0.488,
+      "step": 1726
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5700577799667438,
+      "learning_rate": 4.010200383544992e-06,
+      "loss": 0.4964,
+      "step": 1727
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5451521766339095,
+      "learning_rate": 4.0087025832203065e-06,
+      "loss": 0.4556,
+      "step": 1728
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5817456153792706,
+      "learning_rate": 4.007203930677438e-06,
+      "loss": 0.4983,
+      "step": 1729
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6380188720129754,
+      "learning_rate": 4.00570442676293e-06,
+      "loss": 0.4805,
+      "step": 1730
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5205063698001692,
+      "learning_rate": 4.0042040723238055e-06,
+      "loss": 0.4659,
+      "step": 1731
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5768671155441834,
+      "learning_rate": 4.002702868207563e-06,
+      "loss": 0.4873,
+      "step": 1732
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5494137373354541,
+      "learning_rate": 4.001200815262188e-06,
+      "loss": 0.4723,
+      "step": 1733
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5585419846591574,
+      "learning_rate": 3.999697914336143e-06,
+      "loss": 0.4873,
+      "step": 1734
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.606023064302735,
+      "learning_rate": 3.9981941662783675e-06,
+      "loss": 0.5029,
+      "step": 1735
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5623066425478451,
+      "learning_rate": 3.996689571938282e-06,
+      "loss": 0.463,
+      "step": 1736
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5971959876411371,
+      "learning_rate": 3.995184132165783e-06,
+      "loss": 0.4753,
+      "step": 1737
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.54900424681578,
+      "learning_rate": 3.993677847811247e-06,
+      "loss": 0.4872,
+      "step": 1738
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.574046829299625,
+      "learning_rate": 3.992170719725524e-06,
+      "loss": 0.4602,
+      "step": 1739
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.588055937602866,
+      "learning_rate": 3.990662748759946e-06,
+      "loss": 0.4915,
+      "step": 1740
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5684082620173647,
+      "learning_rate": 3.989153935766314e-06,
+      "loss": 0.4765,
+      "step": 1741
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.5755931683742708,
+      "learning_rate": 3.987644281596913e-06,
+      "loss": 0.4909,
+      "step": 1742
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.553216450579215,
+      "learning_rate": 3.986133787104496e-06,
+      "loss": 0.4723,
+      "step": 1743
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5366461565908377,
+      "learning_rate": 3.984622453142293e-06,
+      "loss": 0.4624,
+      "step": 1744
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5610507340686801,
+      "learning_rate": 3.983110280564009e-06,
+      "loss": 0.5002,
+      "step": 1745
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5676547948280201,
+      "learning_rate": 3.981597270223822e-06,
+      "loss": 0.4897,
+      "step": 1746
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5702435461535129,
+      "learning_rate": 3.980083422976386e-06,
+      "loss": 0.4706,
+      "step": 1747
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5682513989307584,
+      "learning_rate": 3.978568739676822e-06,
+      "loss": 0.4631,
+      "step": 1748
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5668309362429137,
+      "learning_rate": 3.977053221180729e-06,
+      "loss": 0.4878,
+      "step": 1749
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5630149523518507,
+      "learning_rate": 3.975536868344174e-06,
+      "loss": 0.4971,
+      "step": 1750
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5525662343852639,
+      "learning_rate": 3.974019682023695e-06,
+      "loss": 0.4765,
+      "step": 1751
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5456185795288315,
+      "learning_rate": 3.972501663076306e-06,
+      "loss": 0.4695,
+      "step": 1752
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.543284167845247,
+      "learning_rate": 3.9709828123594855e-06,
+      "loss": 0.4546,
+      "step": 1753
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5783138022249001,
+      "learning_rate": 3.969463130731183e-06,
+      "loss": 0.5026,
+      "step": 1754
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5821103906157934,
+      "learning_rate": 3.96794261904982e-06,
+      "loss": 0.4685,
+      "step": 1755
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5776445776955914,
+      "learning_rate": 3.9664212781742865e-06,
+      "loss": 0.4696,
+      "step": 1756
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5444444652574508,
+      "learning_rate": 3.964899108963937e-06,
+      "loss": 0.4715,
+      "step": 1757
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5791101056295087,
+      "learning_rate": 3.963376112278597e-06,
+      "loss": 0.4892,
+      "step": 1758
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5409942798323824,
+      "learning_rate": 3.96185228897856e-06,
+      "loss": 0.4679,
+      "step": 1759
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5313906519312775,
+      "learning_rate": 3.9603276399245864e-06,
+      "loss": 0.4364,
+      "step": 1760
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5590372661816775,
+      "learning_rate": 3.9588021659779e-06,
+      "loss": 0.4478,
+      "step": 1761
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5535499771886758,
+      "learning_rate": 3.957275868000192e-06,
+      "loss": 0.4771,
+      "step": 1762
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5343930689867736,
+      "learning_rate": 3.9557487468536225e-06,
+      "loss": 0.4735,
+      "step": 1763
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5610690419155818,
+      "learning_rate": 3.954220803400811e-06,
+      "loss": 0.4762,
+      "step": 1764
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5305051159735634,
+      "learning_rate": 3.9526920385048465e-06,
+      "loss": 0.4619,
+      "step": 1765
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5436426867711756,
+      "learning_rate": 3.951162453029278e-06,
+      "loss": 0.4793,
+      "step": 1766
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.5326612510661454,
+      "learning_rate": 3.94963204783812e-06,
+      "loss": 0.4849,
+      "step": 1767
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5515320255957604,
+      "learning_rate": 3.948100823795851e-06,
+      "loss": 0.503,
+      "step": 1768
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5694215728858405,
+      "learning_rate": 3.946568781767409e-06,
+      "loss": 0.4945,
+      "step": 1769
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5641781422463842,
+      "learning_rate": 3.945035922618198e-06,
+      "loss": 0.478,
+      "step": 1770
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5588846997248246,
+      "learning_rate": 3.94350224721408e-06,
+      "loss": 0.4798,
+      "step": 1771
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5569535296640705,
+      "learning_rate": 3.9419677564213795e-06,
+      "loss": 0.4866,
+      "step": 1772
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6058189007310192,
+      "learning_rate": 3.9404324511068825e-06,
+      "loss": 0.4777,
+      "step": 1773
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5488550968313441,
+      "learning_rate": 3.938896332137834e-06,
+      "loss": 0.4885,
+      "step": 1774
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5717371621896641,
+      "learning_rate": 3.937359400381938e-06,
+      "loss": 0.481,
+      "step": 1775
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 0.47748449444770813,
+      "eval_runtime": 6921.6137,
+      "eval_samples_per_second": 41.954,
+      "eval_steps_per_second": 2.098,
+      "step": 1775
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5658570817407027,
+      "learning_rate": 3.935821656707359e-06,
+      "loss": 0.4739,
+      "step": 1776
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6083719336101632,
+      "learning_rate": 3.93428310198272e-06,
+      "loss": 0.5087,
+      "step": 1777
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5358828780708986,
+      "learning_rate": 3.932743737077101e-06,
+      "loss": 0.4849,
+      "step": 1778
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.550854913111367,
+      "learning_rate": 3.931203562860042e-06,
+      "loss": 0.5071,
+      "step": 1779
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5347707663990211,
+      "learning_rate": 3.929662580201536e-06,
+      "loss": 0.4794,
+      "step": 1780
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5551468944100317,
+      "learning_rate": 3.928120789972036e-06,
+      "loss": 0.4689,
+      "step": 1781
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5727517444264358,
+      "learning_rate": 3.926578193042451e-06,
+      "loss": 0.4952,
+      "step": 1782
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5623029835503726,
+      "learning_rate": 3.9250347902841456e-06,
+      "loss": 0.4723,
+      "step": 1783
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5452388086485114,
+      "learning_rate": 3.923490582568937e-06,
+      "loss": 0.4746,
+      "step": 1784
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.525242855196112,
+      "learning_rate": 3.9219455707691004e-06,
+      "loss": 0.4802,
+      "step": 1785
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.6002892698931297,
+      "learning_rate": 3.920399755757365e-06,
+      "loss": 0.4951,
+      "step": 1786
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5602495779757628,
+      "learning_rate": 3.9188531384069095e-06,
+      "loss": 0.4791,
+      "step": 1787
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5483232695317782,
+      "learning_rate": 3.917305719591372e-06,
+      "loss": 0.4699,
+      "step": 1788
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5742707937304822,
+      "learning_rate": 3.915757500184838e-06,
+      "loss": 0.4729,
+      "step": 1789
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.5424718840301662,
+      "learning_rate": 3.91420848106185e-06,
+      "loss": 0.4983,
+      "step": 1790
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5297073714321522,
+      "learning_rate": 3.912658663097396e-06,
+      "loss": 0.4725,
+      "step": 1791
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5391456048094847,
+      "learning_rate": 3.911108047166924e-06,
+      "loss": 0.4588,
+      "step": 1792
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5422617623881943,
+      "learning_rate": 3.909556634146323e-06,
+      "loss": 0.461,
+      "step": 1793
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5559883587716514,
+      "learning_rate": 3.908004424911939e-06,
+      "loss": 0.4981,
+      "step": 1794
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5355343530496602,
+      "learning_rate": 3.906451420340566e-06,
+      "loss": 0.4845,
+      "step": 1795
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6731733861479123,
+      "learning_rate": 3.904897621309446e-06,
+      "loss": 0.4936,
+      "step": 1796
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5495562245224338,
+      "learning_rate": 3.9033430286962714e-06,
+      "loss": 0.4577,
+      "step": 1797
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5574105971150268,
+      "learning_rate": 3.901787643379183e-06,
+      "loss": 0.4693,
+      "step": 1798
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5755389775504788,
+      "learning_rate": 3.900231466236766e-06,
+      "loss": 0.4844,
+      "step": 1799
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6078323318223101,
+      "learning_rate": 3.898674498148058e-06,
+      "loss": 0.4728,
+      "step": 1800
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5907606344024237,
+      "learning_rate": 3.897116739992539e-06,
+      "loss": 0.485,
+      "step": 1801
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5509448118458101,
+      "learning_rate": 3.89555819265014e-06,
+      "loss": 0.4868,
+      "step": 1802
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5247691225819013,
+      "learning_rate": 3.893998857001231e-06,
+      "loss": 0.4919,
+      "step": 1803
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5708932504567875,
+      "learning_rate": 3.892438733926634e-06,
+      "loss": 0.4607,
+      "step": 1804
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8363516955156233,
+      "learning_rate": 3.890877824307611e-06,
+      "loss": 0.4721,
+      "step": 1805
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5591473348688127,
+      "learning_rate": 3.889316129025873e-06,
+      "loss": 0.4905,
+      "step": 1806
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5918209400494773,
+      "learning_rate": 3.887753648963569e-06,
+      "loss": 0.473,
+      "step": 1807
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5897196038146727,
+      "learning_rate": 3.886190385003297e-06,
+      "loss": 0.4528,
+      "step": 1808
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6070139742560858,
+      "learning_rate": 3.884626338028093e-06,
+      "loss": 0.4734,
+      "step": 1809
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6106551950471045,
+      "learning_rate": 3.883061508921439e-06,
+      "loss": 0.5053,
+      "step": 1810
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5919165573743996,
+      "learning_rate": 3.8814958985672564e-06,
+      "loss": 0.4792,
+      "step": 1811
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5683775432095641,
+      "learning_rate": 3.87992950784991e-06,
+      "loss": 0.4869,
+      "step": 1812
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.5693276502530639,
+      "learning_rate": 3.878362337654203e-06,
+      "loss": 0.4919,
+      "step": 1813
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6456814342581492,
+      "learning_rate": 3.87679438886538e-06,
+      "loss": 0.5012,
+      "step": 1814
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6047567254595965,
+      "learning_rate": 3.875225662369125e-06,
+      "loss": 0.489,
+      "step": 1815
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6019100134305163,
+      "learning_rate": 3.8736561590515646e-06,
+      "loss": 0.4998,
+      "step": 1816
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5455946748258141,
+      "learning_rate": 3.872085879799258e-06,
+      "loss": 0.4617,
+      "step": 1817
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.593910191752196,
+      "learning_rate": 3.870514825499208e-06,
+      "loss": 0.4898,
+      "step": 1818
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6125446125342663,
+      "learning_rate": 3.868942997038853e-06,
+      "loss": 0.4781,
+      "step": 1819
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5571000922345232,
+      "learning_rate": 3.8673703953060685e-06,
+      "loss": 0.5001,
+      "step": 1820
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6338578928227893,
+      "learning_rate": 3.865797021189167e-06,
+      "loss": 0.5032,
+      "step": 1821
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5693059830561665,
+      "learning_rate": 3.864222875576898e-06,
+      "loss": 0.4717,
+      "step": 1822
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5713524469999574,
+      "learning_rate": 3.862647959358447e-06,
+      "loss": 0.469,
+      "step": 1823
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5469501972304562,
+      "learning_rate": 3.861072273423434e-06,
+      "loss": 0.5001,
+      "step": 1824
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5643553209363702,
+      "learning_rate": 3.859495818661914e-06,
+      "loss": 0.4752,
+      "step": 1825
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6079497862523476,
+      "learning_rate": 3.857918595964375e-06,
+      "loss": 0.4657,
+      "step": 1826
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5940401601609584,
+      "learning_rate": 3.8563406062217405e-06,
+      "loss": 0.4794,
+      "step": 1827
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5224134371483142,
+      "learning_rate": 3.8547618503253685e-06,
+      "loss": 0.4596,
+      "step": 1828
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5691746499773465,
+      "learning_rate": 3.8531823291670455e-06,
+      "loss": 0.4705,
+      "step": 1829
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5709823691070093,
+      "learning_rate": 3.8516020436389945e-06,
+      "loss": 0.4802,
+      "step": 1830
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5674686057693449,
+      "learning_rate": 3.850020994633869e-06,
+      "loss": 0.5046,
+      "step": 1831
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5942932229729947,
+      "learning_rate": 3.848439183044751e-06,
+      "loss": 0.465,
+      "step": 1832
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.6007440298375468,
+      "learning_rate": 3.846856609765158e-06,
+      "loss": 0.485,
+      "step": 1833
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5905695400663254,
+      "learning_rate": 3.845273275689035e-06,
+      "loss": 0.4898,
+      "step": 1834
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5526340123188758,
+      "learning_rate": 3.843689181710756e-06,
+      "loss": 0.4895,
+      "step": 1835
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.585057913171662,
+      "learning_rate": 3.842104328725127e-06,
+      "loss": 0.457,
+      "step": 1836
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5917091841007728,
+      "learning_rate": 3.8405187176273794e-06,
+      "loss": 0.4745,
+      "step": 1837
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5831734022479653,
+      "learning_rate": 3.838932349313176e-06,
+      "loss": 0.5003,
+      "step": 1838
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5475814395504766,
+      "learning_rate": 3.837345224678605e-06,
+      "loss": 0.4875,
+      "step": 1839
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5752248721709146,
+      "learning_rate": 3.835757344620183e-06,
+      "loss": 0.4636,
+      "step": 1840
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5901567903691399,
+      "learning_rate": 3.8341687100348536e-06,
+      "loss": 0.4924,
+      "step": 1841
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5434020555712842,
+      "learning_rate": 3.832579321819985e-06,
+      "loss": 0.4634,
+      "step": 1842
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5492021117737101,
+      "learning_rate": 3.830989180873373e-06,
+      "loss": 0.4777,
+      "step": 1843
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5306994578054051,
+      "learning_rate": 3.829398288093237e-06,
+      "loss": 0.472,
+      "step": 1844
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5921460420396056,
+      "learning_rate": 3.827806644378221e-06,
+      "loss": 0.4951,
+      "step": 1845
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5211955272058552,
+      "learning_rate": 3.826214250627397e-06,
+      "loss": 0.4513,
+      "step": 1846
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 0.47622737288475037,
+      "eval_runtime": 6919.5773,
+      "eval_samples_per_second": 41.966,
+      "eval_steps_per_second": 2.098,
+      "step": 1846
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.572627752021322,
+      "learning_rate": 3.824621107740255e-06,
+      "loss": 0.5055,
+      "step": 1847
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5855256130142836,
+      "learning_rate": 3.823027216616711e-06,
+      "loss": 0.4712,
+      "step": 1848
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5330067035136337,
+      "learning_rate": 3.821432578157105e-06,
+      "loss": 0.461,
+      "step": 1849
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5379073734644638,
+      "learning_rate": 3.819837193262197e-06,
+      "loss": 0.4646,
+      "step": 1850
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5550409399367109,
+      "learning_rate": 3.818241062833168e-06,
+      "loss": 0.5092,
+      "step": 1851
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5475883585070609,
+      "learning_rate": 3.816644187771624e-06,
+      "loss": 0.4845,
+      "step": 1852
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5304094349485199,
+      "learning_rate": 3.815046568979585e-06,
+      "loss": 0.5045,
+      "step": 1853
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5902094724693763,
+      "learning_rate": 3.8134482073594997e-06,
+      "loss": 0.4703,
+      "step": 1854
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5373657261604207,
+      "learning_rate": 3.811849103814229e-06,
+      "loss": 0.4438,
+      "step": 1855
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5500441237475201,
+      "learning_rate": 3.8102492592470562e-06,
+      "loss": 0.4808,
+      "step": 1856
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.587067338502479,
+      "learning_rate": 3.808648674561683e-06,
+      "loss": 0.484,
+      "step": 1857
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5762212544281782,
+      "learning_rate": 3.8070473506622283e-06,
+      "loss": 0.4765,
+      "step": 1858
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5750217699914456,
+      "learning_rate": 3.80544528845323e-06,
+      "loss": 0.4695,
+      "step": 1859
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5928903124224453,
+      "learning_rate": 3.803842488839642e-06,
+      "loss": 0.4931,
+      "step": 1860
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5211947228388228,
+      "learning_rate": 3.8022389527268344e-06,
+      "loss": 0.4671,
+      "step": 1861
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5624138113341162,
+      "learning_rate": 3.8006346810205935e-06,
+      "loss": 0.462,
+      "step": 1862
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6055264031899258,
+      "learning_rate": 3.7990296746271227e-06,
+      "loss": 0.4951,
+      "step": 1863
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.575871000906671,
+      "learning_rate": 3.797423934453038e-06,
+      "loss": 0.4613,
+      "step": 1864
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5436757648363332,
+      "learning_rate": 3.795817461405372e-06,
+      "loss": 0.495,
+      "step": 1865
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5489759393271353,
+      "learning_rate": 3.7942102563915693e-06,
+      "loss": 0.4761,
+      "step": 1866
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5799706124416135,
+      "learning_rate": 3.79260232031949e-06,
+      "loss": 0.4992,
+      "step": 1867
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6021574593638155,
+      "learning_rate": 3.7909936540974052e-06,
+      "loss": 0.4573,
+      "step": 1868
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5528129676449581,
+      "learning_rate": 3.7893842586340003e-06,
+      "loss": 0.515,
+      "step": 1869
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5695311954254395,
+      "learning_rate": 3.7877741348383703e-06,
+      "loss": 0.4728,
+      "step": 1870
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5592177344188098,
+      "learning_rate": 3.7861632836200245e-06,
+      "loss": 0.4673,
+      "step": 1871
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5611705084883242,
+      "learning_rate": 3.784551705888881e-06,
+      "loss": 0.4584,
+      "step": 1872
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5539834069231695,
+      "learning_rate": 3.7829394025552684e-06,
+      "loss": 0.4644,
+      "step": 1873
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.537446301860318,
+      "learning_rate": 3.7813263745299257e-06,
+      "loss": 0.4927,
+      "step": 1874
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5596544860638828,
+      "learning_rate": 3.779712622724003e-06,
+      "loss": 0.4647,
+      "step": 1875
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.544652883431319,
+      "learning_rate": 3.7780981480490554e-06,
+      "loss": 0.4955,
+      "step": 1876
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5512313969653201,
+      "learning_rate": 3.776482951417049e-06,
+      "loss": 0.4644,
+      "step": 1877
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5497506195632861,
+      "learning_rate": 3.774867033740357e-06,
+      "loss": 0.4789,
+      "step": 1878
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5742672691362793,
+      "learning_rate": 3.77325039593176e-06,
+      "loss": 0.4729,
+      "step": 1879
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5629948393220319,
+      "learning_rate": 3.7716330389044463e-06,
+      "loss": 0.4723,
+      "step": 1880
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5888605808332452,
+      "learning_rate": 3.7700149635720086e-06,
+      "loss": 0.4716,
+      "step": 1881
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5578428602453922,
+      "learning_rate": 3.768396170848445e-06,
+      "loss": 0.4958,
+      "step": 1882
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5466856701572237,
+      "learning_rate": 3.766776661648163e-06,
+      "loss": 0.4712,
+      "step": 1883
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5650833519106695,
+      "learning_rate": 3.76515643688597e-06,
+      "loss": 0.4812,
+      "step": 1884
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5662580501911987,
+      "learning_rate": 3.76353549747708e-06,
+      "loss": 0.477,
+      "step": 1885
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5515090242427748,
+      "learning_rate": 3.76191384433711e-06,
+      "loss": 0.4784,
+      "step": 1886
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5651933942612933,
+      "learning_rate": 3.76029147838208e-06,
+      "loss": 0.4493,
+      "step": 1887
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5534626182885516,
+      "learning_rate": 3.7586684005284146e-06,
+      "loss": 0.4823,
+      "step": 1888
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5308600240405152,
+      "learning_rate": 3.7570446116929372e-06,
+      "loss": 0.448,
+      "step": 1889
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5654195108385544,
+      "learning_rate": 3.7554201127928747e-06,
+      "loss": 0.4679,
+      "step": 1890
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5815086921921774,
+      "learning_rate": 3.7537949047458567e-06,
+      "loss": 0.4637,
+      "step": 1891
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5603546362158033,
+      "learning_rate": 3.7521689884699093e-06,
+      "loss": 0.4805,
+      "step": 1892
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5835928980888774,
+      "learning_rate": 3.750542364883462e-06,
+      "loss": 0.4665,
+      "step": 1893
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5706169202769774,
+      "learning_rate": 3.748915034905344e-06,
+      "loss": 0.4889,
+      "step": 1894
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5269442750245277,
+      "learning_rate": 3.74728699945478e-06,
+      "loss": 0.465,
+      "step": 1895
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5547808175518558,
+      "learning_rate": 3.745658259451397e-06,
+      "loss": 0.4639,
+      "step": 1896
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6267511707684223,
+      "learning_rate": 3.744028815815219e-06,
+      "loss": 0.4865,
+      "step": 1897
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5643243359785219,
+      "learning_rate": 3.742398669466665e-06,
+      "loss": 0.4897,
+      "step": 1898
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5600481323676141,
+      "learning_rate": 3.740767821326555e-06,
+      "loss": 0.4708,
+      "step": 1899
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5421854980261044,
+      "learning_rate": 3.739136272316102e-06,
+      "loss": 0.4797,
+      "step": 1900
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5504823302075539,
+      "learning_rate": 3.737504023356916e-06,
+      "loss": 0.4477,
+      "step": 1901
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5596996727298843,
+      "learning_rate": 3.735871075371004e-06,
+      "loss": 0.4751,
+      "step": 1902
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5490200023377907,
+      "learning_rate": 3.734237429280766e-06,
+      "loss": 0.4667,
+      "step": 1903
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.6262743935742706,
+      "learning_rate": 3.7326030860089955e-06,
+      "loss": 0.4838,
+      "step": 1904
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5824379568686556,
+      "learning_rate": 3.7309680464788835e-06,
+      "loss": 0.4905,
+      "step": 1905
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.53774534973763,
+      "learning_rate": 3.72933231161401e-06,
+      "loss": 0.4701,
+      "step": 1906
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5754065009216981,
+      "learning_rate": 3.72769588233835e-06,
+      "loss": 0.5086,
+      "step": 1907
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5438398948047919,
+      "learning_rate": 3.726058759576271e-06,
+      "loss": 0.5028,
+      "step": 1908
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5545263081214099,
+      "learning_rate": 3.724420944252531e-06,
+      "loss": 0.4742,
+      "step": 1909
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5598750002548621,
+      "learning_rate": 3.72278243729228e-06,
+      "loss": 0.4568,
+      "step": 1910
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5657656081137544,
+      "learning_rate": 3.7211432396210595e-06,
+      "loss": 0.4478,
+      "step": 1911
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5621097014773546,
+      "learning_rate": 3.7195033521647987e-06,
+      "loss": 0.4779,
+      "step": 1912
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5412943721977179,
+      "learning_rate": 3.7178627758498194e-06,
+      "loss": 0.4553,
+      "step": 1913
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.577522286854261,
+      "learning_rate": 3.71622151160283e-06,
+      "loss": 0.4829,
+      "step": 1914
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.543384848094331,
+      "learning_rate": 3.7145795603509282e-06,
+      "loss": 0.4582,
+      "step": 1915
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5778457930279592,
+      "learning_rate": 3.712936923021602e-06,
+      "loss": 0.4713,
+      "step": 1916
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5766320078694117,
+      "learning_rate": 3.7112936005427237e-06,
+      "loss": 0.4683,
+      "step": 1917
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 0.4745747447013855,
+      "eval_runtime": 6921.2455,
+      "eval_samples_per_second": 41.956,
+      "eval_steps_per_second": 2.098,
+      "step": 1917
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5432751258589548,
+      "learning_rate": 3.7096495938425537e-06,
+      "loss": 0.4674,
+      "step": 1918
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.569125410720449,
+      "learning_rate": 3.7080049038497405e-06,
+      "loss": 0.4793,
+      "step": 1919
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5681975559829856,
+      "learning_rate": 3.706359531493316e-06,
+      "loss": 0.4942,
+      "step": 1920
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5598180203564832,
+      "learning_rate": 3.704713477702699e-06,
+      "loss": 0.4869,
+      "step": 1921
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5358569958499817,
+      "learning_rate": 3.703066743407694e-06,
+      "loss": 0.493,
+      "step": 1922
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5694771092869512,
+      "learning_rate": 3.701419329538487e-06,
+      "loss": 0.4798,
+      "step": 1923
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5515301941280839,
+      "learning_rate": 3.699771237025652e-06,
+      "loss": 0.4732,
+      "step": 1924
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5836180114241711,
+      "learning_rate": 3.6981224668001427e-06,
+      "loss": 0.4868,
+      "step": 1925
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5724166590716755,
+      "learning_rate": 3.696473019793297e-06,
+      "loss": 0.4631,
+      "step": 1926
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5794404466148781,
+      "learning_rate": 3.694822896936836e-06,
+      "loss": 0.5093,
+      "step": 1927
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5788548411305892,
+      "learning_rate": 3.6931720991628613e-06,
+      "loss": 0.4849,
+      "step": 1928
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5388008666947813,
+      "learning_rate": 3.691520627403856e-06,
+      "loss": 0.4653,
+      "step": 1929
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.548773537353004,
+      "learning_rate": 3.6898684825926845e-06,
+      "loss": 0.4823,
+      "step": 1930
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5588102780031893,
+      "learning_rate": 3.68821566566259e-06,
+      "loss": 0.4644,
+      "step": 1931
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5614220676467329,
+      "learning_rate": 3.686562177547197e-06,
+      "loss": 0.4392,
+      "step": 1932
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.57266859860218,
+      "learning_rate": 3.6849080191805087e-06,
+      "loss": 0.4784,
+      "step": 1933
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5752230473014237,
+      "learning_rate": 3.683253191496906e-06,
+      "loss": 0.4782,
+      "step": 1934
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5692017163618623,
+      "learning_rate": 3.681597695431149e-06,
+      "loss": 0.4845,
+      "step": 1935
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5999017797504378,
+      "learning_rate": 3.6799415319183753e-06,
+      "loss": 0.5121,
+      "step": 1936
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5542057486345733,
+      "learning_rate": 3.678284701894097e-06,
+      "loss": 0.4545,
+      "step": 1937
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5744953689894631,
+      "learning_rate": 3.6766272062942066e-06,
+      "loss": 0.4673,
+      "step": 1938
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5519474373087149,
+      "learning_rate": 3.6749690460549704e-06,
+      "loss": 0.4768,
+      "step": 1939
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5308709424185206,
+      "learning_rate": 3.6733102221130303e-06,
+      "loss": 0.4675,
+      "step": 1940
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5543162922631799,
+      "learning_rate": 3.6716507354054044e-06,
+      "loss": 0.4884,
+      "step": 1941
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5525721347222098,
+      "learning_rate": 3.669990586869482e-06,
+      "loss": 0.4633,
+      "step": 1942
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5676565038664707,
+      "learning_rate": 3.6683297774430287e-06,
+      "loss": 0.4915,
+      "step": 1943
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5354820116105071,
+      "learning_rate": 3.6666683080641846e-06,
+      "loss": 0.471,
+      "step": 1944
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5521553904563508,
+      "learning_rate": 3.6650061796714597e-06,
+      "loss": 0.4683,
+      "step": 1945
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5853163676085599,
+      "learning_rate": 3.6633433932037376e-06,
+      "loss": 0.5075,
+      "step": 1946
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5751517974852843,
+      "learning_rate": 3.661679949600275e-06,
+      "loss": 0.4778,
+      "step": 1947
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5440950912602847,
+      "learning_rate": 3.6600158498006955e-06,
+      "loss": 0.4706,
+      "step": 1948
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5476897010825572,
+      "learning_rate": 3.6583510947449983e-06,
+      "loss": 0.4612,
+      "step": 1949
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5562096216167524,
+      "learning_rate": 3.656685685373552e-06,
+      "loss": 0.4692,
+      "step": 1950
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.551268497700442,
+      "learning_rate": 3.6550196226270894e-06,
+      "loss": 0.4598,
+      "step": 1951
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5391818232642948,
+      "learning_rate": 3.65335290744672e-06,
+      "loss": 0.4695,
+      "step": 1952
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5338996664057097,
+      "learning_rate": 3.6516855407739164e-06,
+      "loss": 0.4665,
+      "step": 1953
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.5579648908867901,
+      "learning_rate": 3.6500175235505226e-06,
+      "loss": 0.4883,
+      "step": 1954
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5331291112438312,
+      "learning_rate": 3.6483488567187473e-06,
+      "loss": 0.4836,
+      "step": 1955
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.556976603909284,
+      "learning_rate": 3.646679541221168e-06,
+      "loss": 0.4893,
+      "step": 1956
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5623081423655151,
+      "learning_rate": 3.6450095780007277e-06,
+      "loss": 0.4966,
+      "step": 1957
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5490180827544557,
+      "learning_rate": 3.643338968000736e-06,
+      "loss": 0.4592,
+      "step": 1958
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5673115477683868,
+      "learning_rate": 3.641667712164867e-06,
+      "loss": 0.488,
+      "step": 1959
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5550807543029536,
+      "learning_rate": 3.6399958114371597e-06,
+      "loss": 0.477,
+      "step": 1960
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5857922488282928,
+      "learning_rate": 3.6383232667620195e-06,
+      "loss": 0.4715,
+      "step": 1961
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.31408407563173,
+      "learning_rate": 3.6366500790842113e-06,
+      "loss": 0.4798,
+      "step": 1962
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5616626137169414,
+      "learning_rate": 3.634976249348867e-06,
+      "loss": 0.4779,
+      "step": 1963
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5426687056510846,
+      "learning_rate": 3.633301778501481e-06,
+      "loss": 0.478,
+      "step": 1964
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5819270419939396,
+      "learning_rate": 3.631626667487906e-06,
+      "loss": 0.4819,
+      "step": 1965
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5530881380000352,
+      "learning_rate": 3.6299509172543616e-06,
+      "loss": 0.4939,
+      "step": 1966
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5705862965767243,
+      "learning_rate": 3.628274528747424e-06,
+      "loss": 0.4805,
+      "step": 1967
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6039400830717704,
+      "learning_rate": 3.6265975029140334e-06,
+      "loss": 0.4728,
+      "step": 1968
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5576346901828154,
+      "learning_rate": 3.624919840701488e-06,
+      "loss": 0.4876,
+      "step": 1969
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.58645769749298,
+      "learning_rate": 3.623241543057445e-06,
+      "loss": 0.4852,
+      "step": 1970
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5694595348460465,
+      "learning_rate": 3.6215626109299218e-06,
+      "loss": 0.4943,
+      "step": 1971
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5505430973110726,
+      "learning_rate": 3.6198830452672944e-06,
+      "loss": 0.4823,
+      "step": 1972
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5666099153148945,
+      "learning_rate": 3.618202847018296e-06,
+      "loss": 0.4794,
+      "step": 1973
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5649501462357687,
+      "learning_rate": 3.616522017132017e-06,
+      "loss": 0.4729,
+      "step": 1974
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6007742923118407,
+      "learning_rate": 3.614840556557905e-06,
+      "loss": 0.4757,
+      "step": 1975
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5766917548962726,
+      "learning_rate": 3.613158466245763e-06,
+      "loss": 0.4645,
+      "step": 1976
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.5540384171742403,
+      "learning_rate": 3.6114757471457514e-06,
+      "loss": 0.4525,
+      "step": 1977
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5488107248965611,
+      "learning_rate": 3.6097924002083838e-06,
+      "loss": 0.4632,
+      "step": 1978
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.556985216041662,
+      "learning_rate": 3.60810842638453e-06,
+      "loss": 0.4871,
+      "step": 1979
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5302372510269003,
+      "learning_rate": 3.606423826625414e-06,
+      "loss": 0.4819,
+      "step": 1980
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5650443172227675,
+      "learning_rate": 3.604738601882612e-06,
+      "loss": 0.4668,
+      "step": 1981
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5687893746146396,
+      "learning_rate": 3.6030527531080533e-06,
+      "loss": 0.4668,
+      "step": 1982
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5483149007131946,
+      "learning_rate": 3.6013662812540217e-06,
+      "loss": 0.4633,
+      "step": 1983
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5666223715356314,
+      "learning_rate": 3.5996791872731508e-06,
+      "loss": 0.4996,
+      "step": 1984
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5610496348612013,
+      "learning_rate": 3.5979914721184263e-06,
+      "loss": 0.4638,
+      "step": 1985
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.556813398196998,
+      "learning_rate": 3.5963031367431856e-06,
+      "loss": 0.489,
+      "step": 1986
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.555985238033488,
+      "learning_rate": 3.594614182101115e-06,
+      "loss": 0.4644,
+      "step": 1987
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.2268257169954715,
+      "learning_rate": 3.592924609146251e-06,
+      "loss": 0.4715,
+      "step": 1988
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 0.4732716977596283,
+      "eval_runtime": 6918.9919,
+      "eval_samples_per_second": 41.97,
+      "eval_steps_per_second": 2.099,
+      "step": 1988
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5830835777523167,
+      "learning_rate": 3.5912344188329812e-06,
+      "loss": 0.4622,
+      "step": 1989
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5532809255407818,
+      "learning_rate": 3.5895436121160388e-06,
+      "loss": 0.5057,
+      "step": 1990
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5548636273482329,
+      "learning_rate": 3.5878521899505083e-06,
+      "loss": 0.4717,
+      "step": 1991
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5756081970951604,
+      "learning_rate": 3.5861601532918188e-06,
+      "loss": 0.4828,
+      "step": 1992
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.561307751747662,
+      "learning_rate": 3.58446750309575e-06,
+      "loss": 0.471,
+      "step": 1993
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5716452195722197,
+      "learning_rate": 3.5827742403184246e-06,
+      "loss": 0.4799,
+      "step": 1994
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6374115187137489,
+      "learning_rate": 3.5810803659163136e-06,
+      "loss": 0.4685,
+      "step": 1995
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5584668047661877,
+      "learning_rate": 3.579385880846232e-06,
+      "loss": 0.4733,
+      "step": 1996
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5582439256078539,
+      "learning_rate": 3.577690786065343e-06,
+      "loss": 0.4657,
+      "step": 1997
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5551491606524519,
+      "learning_rate": 3.5759950825311497e-06,
+      "loss": 0.499,
+      "step": 1998
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.6015108198668244,
+      "learning_rate": 3.5742987712015016e-06,
+      "loss": 0.4971,
+      "step": 1999
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5552048556458112,
+      "learning_rate": 3.5726018530345913e-06,
+      "loss": 0.4661,
+      "step": 2000
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.5346633218936818,
+      "learning_rate": 3.5709043289889538e-06,
+      "loss": 0.4608,
+      "step": 2001
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5373071208933639,
+      "learning_rate": 3.5692062000234663e-06,
+      "loss": 0.4647,
+      "step": 2002
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5416519357021035,
+      "learning_rate": 3.5675074670973485e-06,
+      "loss": 0.4738,
+      "step": 2003
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.569652388343622,
+      "learning_rate": 3.565808131170161e-06,
+      "loss": 0.4749,
+      "step": 2004
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.540670625527253,
+      "learning_rate": 3.564108193201804e-06,
+      "loss": 0.4843,
+      "step": 2005
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5637489546305884,
+      "learning_rate": 3.562407654152518e-06,
+      "loss": 0.4718,
+      "step": 2006
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5563549147978012,
+      "learning_rate": 3.5607065149828845e-06,
+      "loss": 0.4862,
+      "step": 2007
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5371061058438552,
+      "learning_rate": 3.559004776653823e-06,
+      "loss": 0.4974,
+      "step": 2008
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.544463201871563,
+      "learning_rate": 3.557302440126591e-06,
+      "loss": 0.4674,
+      "step": 2009
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5563948529476817,
+      "learning_rate": 3.5555995063627842e-06,
+      "loss": 0.4703,
+      "step": 2010
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5433738422335685,
+      "learning_rate": 3.5538959763243363e-06,
+      "loss": 0.4801,
+      "step": 2011
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5616100198019316,
+      "learning_rate": 3.552191850973517e-06,
+      "loss": 0.496,
+      "step": 2012
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5891584297064337,
+      "learning_rate": 3.550487131272933e-06,
+      "loss": 0.4766,
+      "step": 2013
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5684784781174675,
+      "learning_rate": 3.5487818181855253e-06,
+      "loss": 0.4633,
+      "step": 2014
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.544328038939811,
+      "learning_rate": 3.5470759126745726e-06,
+      "loss": 0.4694,
+      "step": 2015
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5505644001026635,
+      "learning_rate": 3.545369415703685e-06,
+      "loss": 0.4834,
+      "step": 2016
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.605202877335372,
+      "learning_rate": 3.54366232823681e-06,
+      "loss": 0.4879,
+      "step": 2017
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5543782398842882,
+      "learning_rate": 3.5419546512382264e-06,
+      "loss": 0.461,
+      "step": 2018
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.564246989631432,
+      "learning_rate": 3.540246385672547e-06,
+      "loss": 0.4655,
+      "step": 2019
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5286400658391471,
+      "learning_rate": 3.5385375325047167e-06,
+      "loss": 0.4352,
+      "step": 2020
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5577835291298394,
+      "learning_rate": 3.536828092700012e-06,
+      "loss": 0.4591,
+      "step": 2021
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5649431678973527,
+      "learning_rate": 3.5351180672240413e-06,
+      "loss": 0.4604,
+      "step": 2022
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5672528493306068,
+      "learning_rate": 3.5334074570427444e-06,
+      "loss": 0.5038,
+      "step": 2023
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.5562216889012358,
+      "learning_rate": 3.5316962631223896e-06,
+      "loss": 0.4651,
+      "step": 2024
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5639239221337747,
+      "learning_rate": 3.5299844864295773e-06,
+      "loss": 0.438,
+      "step": 2025
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5501230990120523,
+      "learning_rate": 3.5282721279312343e-06,
+      "loss": 0.4518,
+      "step": 2026
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.565754118751582,
+      "learning_rate": 3.5265591885946184e-06,
+      "loss": 0.5105,
+      "step": 2027
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5488876736412479,
+      "learning_rate": 3.5248456693873152e-06,
+      "loss": 0.4705,
+      "step": 2028
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.567410215517672,
+      "learning_rate": 3.523131571277235e-06,
+      "loss": 0.4649,
+      "step": 2029
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.56236312948077,
+      "learning_rate": 3.5214168952326205e-06,
+      "loss": 0.4625,
+      "step": 2030
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5713132266773925,
+      "learning_rate": 3.519701642222036e-06,
+      "loss": 0.4613,
+      "step": 2031
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5586906230581218,
+      "learning_rate": 3.5179858132143727e-06,
+      "loss": 0.4749,
+      "step": 2032
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5452698201212837,
+      "learning_rate": 3.5162694091788506e-06,
+      "loss": 0.4704,
+      "step": 2033
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5472112238723226,
+      "learning_rate": 3.5145524310850088e-06,
+      "loss": 0.4723,
+      "step": 2034
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5239386822886098,
+      "learning_rate": 3.5128348799027157e-06,
+      "loss": 0.4832,
+      "step": 2035
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5582354151966005,
+      "learning_rate": 3.5111167566021607e-06,
+      "loss": 0.4825,
+      "step": 2036
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5661871633973496,
+      "learning_rate": 3.509398062153857e-06,
+      "loss": 0.4808,
+      "step": 2037
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5450889485025441,
+      "learning_rate": 3.507678797528641e-06,
+      "loss": 0.4908,
+      "step": 2038
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5299540248896896,
+      "learning_rate": 3.5059589636976704e-06,
+      "loss": 0.4392,
+      "step": 2039
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5347817988173922,
+      "learning_rate": 3.5042385616324243e-06,
+      "loss": 0.4673,
+      "step": 2040
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.545451063247587,
+      "learning_rate": 3.5025175923047034e-06,
+      "loss": 0.4531,
+      "step": 2041
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.657050095176287,
+      "learning_rate": 3.5007960566866296e-06,
+      "loss": 0.4599,
+      "step": 2042
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5948751733409405,
+      "learning_rate": 3.499073955750642e-06,
+      "loss": 0.4796,
+      "step": 2043
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5467310616780172,
+      "learning_rate": 3.497351290469503e-06,
+      "loss": 0.5012,
+      "step": 2044
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.538398730069935,
+      "learning_rate": 3.4956280618162887e-06,
+      "loss": 0.4799,
+      "step": 2045
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5563044950019395,
+      "learning_rate": 3.4939042707643983e-06,
+      "loss": 0.4456,
+      "step": 2046
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5552862407440945,
+      "learning_rate": 3.492179918287547e-06,
+      "loss": 0.4715,
+      "step": 2047
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5443648119446477,
+      "learning_rate": 3.4904550053597646e-06,
+      "loss": 0.4737,
+      "step": 2048
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.533397172596125,
+      "learning_rate": 3.488729532955401e-06,
+      "loss": 0.4327,
+      "step": 2049
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5290432865271623,
+      "learning_rate": 3.4870035020491216e-06,
+      "loss": 0.4568,
+      "step": 2050
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5653745834920318,
+      "learning_rate": 3.4852769136159047e-06,
+      "loss": 0.4719,
+      "step": 2051
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 3.260354916091318,
+      "learning_rate": 3.4835497686310458e-06,
+      "loss": 0.4502,
+      "step": 2052
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5407916301520294,
+      "learning_rate": 3.4818220680701554e-06,
+      "loss": 0.493,
+      "step": 2053
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5530804605769145,
+      "learning_rate": 3.480093812909155e-06,
+      "loss": 0.4591,
+      "step": 2054
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5560882191153973,
+      "learning_rate": 3.4783650041242823e-06,
+      "loss": 0.4814,
+      "step": 2055
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5491339314657214,
+      "learning_rate": 3.4766356426920854e-06,
+      "loss": 0.459,
+      "step": 2056
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5446665246830417,
+      "learning_rate": 3.474905729589427e-06,
+      "loss": 0.4721,
+      "step": 2057
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5597955365837155,
+      "learning_rate": 3.4731752657934793e-06,
+      "loss": 0.4657,
+      "step": 2058
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5257446537011154,
+      "learning_rate": 3.471444252281726e-06,
+      "loss": 0.4791,
+      "step": 2059
+    },
+    {
+      "epoch": 0.88,
+      "eval_loss": 0.47191861271858215,
+      "eval_runtime": 6926.5022,
+      "eval_samples_per_second": 41.924,
+      "eval_steps_per_second": 2.096,
+      "step": 2059
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5550797881194796,
+      "learning_rate": 3.469712690031962e-06,
+      "loss": 0.4586,
+      "step": 2060
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5435199223357358,
+      "learning_rate": 3.467980580022293e-06,
+      "loss": 0.4572,
+      "step": 2061
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5445376220038961,
+      "learning_rate": 3.466247923231131e-06,
+      "loss": 0.4535,
+      "step": 2062
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5438364103684316,
+      "learning_rate": 3.4645147206371997e-06,
+      "loss": 0.469,
+      "step": 2063
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5622544652606651,
+      "learning_rate": 3.4627809732195306e-06,
+      "loss": 0.5053,
+      "step": 2064
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5787220304673627,
+      "learning_rate": 3.4610466819574617e-06,
+      "loss": 0.4916,
+      "step": 2065
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5665331755731509,
+      "learning_rate": 3.45931184783064e-06,
+      "loss": 0.461,
+      "step": 2066
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5465798109205211,
+      "learning_rate": 3.4575764718190174e-06,
+      "loss": 0.4681,
+      "step": 2067
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5519620649686449,
+      "learning_rate": 3.455840554902853e-06,
+      "loss": 0.4842,
+      "step": 2068
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5679253676438992,
+      "learning_rate": 3.4541040980627117e-06,
+      "loss": 0.4714,
+      "step": 2069
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5537751246266448,
+      "learning_rate": 3.4523671022794612e-06,
+      "loss": 0.4746,
+      "step": 2070
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.5895813898958205,
+      "learning_rate": 3.450629568534277e-06,
+      "loss": 0.483,
+      "step": 2071
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5820282697400692,
+      "learning_rate": 3.448891497808636e-06,
+      "loss": 0.4948,
+      "step": 2072
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5849936796434134,
+      "learning_rate": 3.4471528910843193e-06,
+      "loss": 0.4604,
+      "step": 2073
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5994615032661528,
+      "learning_rate": 3.4454137493434107e-06,
+      "loss": 0.4893,
+      "step": 2074
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5648890671471385,
+      "learning_rate": 3.443674073568296e-06,
+      "loss": 0.4761,
+      "step": 2075
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5800587534232966,
+      "learning_rate": 3.441933864741663e-06,
+      "loss": 0.4747,
+      "step": 2076
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5515407287133164,
+      "learning_rate": 3.4401931238464996e-06,
+      "loss": 0.4339,
+      "step": 2077
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5814192658954972,
+      "learning_rate": 3.438451851866097e-06,
+      "loss": 0.4769,
+      "step": 2078
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5393565470119203,
+      "learning_rate": 3.4367100497840416e-06,
+      "loss": 0.4843,
+      "step": 2079
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5789119658212182,
+      "learning_rate": 3.4349677185842246e-06,
+      "loss": 0.4767,
+      "step": 2080
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5605931723414512,
+      "learning_rate": 3.433224859250832e-06,
+      "loss": 0.4735,
+      "step": 2081
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6040063691004713,
+      "learning_rate": 3.4314814727683506e-06,
+      "loss": 0.4968,
+      "step": 2082
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5148532844857273,
+      "learning_rate": 3.429737560121564e-06,
+      "loss": 0.4516,
+      "step": 2083
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5936934662433498,
+      "learning_rate": 3.427993122295552e-06,
+      "loss": 0.4827,
+      "step": 2084
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5666076166587783,
+      "learning_rate": 3.4262481602756937e-06,
+      "loss": 0.4535,
+      "step": 2085
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.542895247264907,
+      "learning_rate": 3.4245026750476618e-06,
+      "loss": 0.4531,
+      "step": 2086
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.541516478103584,
+      "learning_rate": 3.4227566675974256e-06,
+      "loss": 0.4892,
+      "step": 2087
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5605571680518421,
+      "learning_rate": 3.421010138911249e-06,
+      "loss": 0.4508,
+      "step": 2088
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5849102411377991,
+      "learning_rate": 3.4192630899756924e-06,
+      "loss": 0.4528,
+      "step": 2089
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5336283093255381,
+      "learning_rate": 3.4175155217776057e-06,
+      "loss": 0.4516,
+      "step": 2090
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5558368518531489,
+      "learning_rate": 3.4157674353041358e-06,
+      "loss": 0.4703,
+      "step": 2091
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5552349223460866,
+      "learning_rate": 3.4140188315427216e-06,
+      "loss": 0.497,
+      "step": 2092
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5511369795033969,
+      "learning_rate": 3.4122697114810934e-06,
+      "loss": 0.5119,
+      "step": 2093
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.5583288164174555,
+      "learning_rate": 3.410520076107273e-06,
+      "loss": 0.4827,
+      "step": 2094
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.567817131338669,
+      "learning_rate": 3.4087699264095746e-06,
+      "loss": 0.4634,
+      "step": 2095
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5815719817752771,
+      "learning_rate": 3.4070192633766025e-06,
+      "loss": 0.4886,
+      "step": 2096
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5479218943057999,
+      "learning_rate": 3.405268087997251e-06,
+      "loss": 0.4886,
+      "step": 2097
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5538188239242408,
+      "learning_rate": 3.4035164012607013e-06,
+      "loss": 0.4459,
+      "step": 2098
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.54601055265423,
+      "learning_rate": 3.401764204156428e-06,
+      "loss": 0.4682,
+      "step": 2099
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5752200802837831,
+      "learning_rate": 3.4000114976741905e-06,
+      "loss": 0.4858,
+      "step": 2100
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5329434779873036,
+      "learning_rate": 3.3982582828040373e-06,
+      "loss": 0.4508,
+      "step": 2101
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5626679379467617,
+      "learning_rate": 3.3965045605363036e-06,
+      "loss": 0.4974,
+      "step": 2102
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5227078861625193,
+      "learning_rate": 3.3947503318616117e-06,
+      "loss": 0.447,
+      "step": 2103
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5318507506941723,
+      "learning_rate": 3.3929955977708686e-06,
+      "loss": 0.4611,
+      "step": 2104
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5652337604907314,
+      "learning_rate": 3.391240359255269e-06,
+      "loss": 0.4825,
+      "step": 2105
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5661117477966576,
+      "learning_rate": 3.3894846173062917e-06,
+      "loss": 0.4569,
+      "step": 2106
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5225946246169721,
+      "learning_rate": 3.3877283729156983e-06,
+      "loss": 0.4646,
+      "step": 2107
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.522801955720487,
+      "learning_rate": 3.385971627075537e-06,
+      "loss": 0.4846,
+      "step": 2108
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5654497625396676,
+      "learning_rate": 3.3842143807781363e-06,
+      "loss": 0.4641,
+      "step": 2109
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.553968034176698,
+      "learning_rate": 3.38245663501611e-06,
+      "loss": 0.4771,
+      "step": 2110
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5512083756758241,
+      "learning_rate": 3.3806983907823526e-06,
+      "loss": 0.4711,
+      "step": 2111
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5612387241279982,
+      "learning_rate": 3.378939649070039e-06,
+      "loss": 0.4929,
+      "step": 2112
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5634378274470825,
+      "learning_rate": 3.3771804108726294e-06,
+      "loss": 0.4668,
+      "step": 2113
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5691963273905253,
+      "learning_rate": 3.375420677183859e-06,
+      "loss": 0.4857,
+      "step": 2114
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.556084686207028,
+      "learning_rate": 3.3736604489977465e-06,
+      "loss": 0.4756,
+      "step": 2115
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5453441170478283,
+      "learning_rate": 3.3718997273085883e-06,
+      "loss": 0.4633,
+      "step": 2116
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5336519493865814,
+      "learning_rate": 3.3701385131109617e-06,
+      "loss": 0.4587,
+      "step": 2117
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.54183517888334,
+      "learning_rate": 3.368376807399719e-06,
+      "loss": 0.4897,
+      "step": 2118
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.537940362868582,
+      "learning_rate": 3.3666146111699926e-06,
+      "loss": 0.4613,
+      "step": 2119
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5229756265690545,
+      "learning_rate": 3.3648519254171906e-06,
+      "loss": 0.4689,
+      "step": 2120
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6066336811051063,
+      "learning_rate": 3.363088751136999e-06,
+      "loss": 0.5001,
+      "step": 2121
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5518196727992455,
+      "learning_rate": 3.3613250893253794e-06,
+      "loss": 0.5078,
+      "step": 2122
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5602072507870511,
+      "learning_rate": 3.3595609409785668e-06,
+      "loss": 0.479,
+      "step": 2123
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5609819914853433,
+      "learning_rate": 3.357796307093074e-06,
+      "loss": 0.4798,
+      "step": 2124
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.540124268427881,
+      "learning_rate": 3.3560311886656855e-06,
+      "loss": 0.4707,
+      "step": 2125
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5273549257492,
+      "learning_rate": 3.3542655866934613e-06,
+      "loss": 0.451,
+      "step": 2126
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5257257097173317,
+      "learning_rate": 3.352499502173734e-06,
+      "loss": 0.4422,
+      "step": 2127
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5242099348991289,
+      "learning_rate": 3.350732936104108e-06,
+      "loss": 0.4553,
+      "step": 2128
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5871633066769245,
+      "learning_rate": 3.3489658894824614e-06,
+      "loss": 0.4706,
+      "step": 2129
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5558589637343108,
+      "learning_rate": 3.3471983633069414e-06,
+      "loss": 0.4506,
+      "step": 2130
+    },
+    {
+      "epoch": 0.91,
+      "eval_loss": 0.4703803062438965,
+      "eval_runtime": 6929.62,
+      "eval_samples_per_second": 41.905,
+      "eval_steps_per_second": 2.095,
+      "step": 2130
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.538163796065571,
+      "learning_rate": 3.3454303585759684e-06,
+      "loss": 0.4616,
+      "step": 2131
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.6059394097926508,
+      "learning_rate": 3.3436618762882322e-06,
+      "loss": 0.4803,
+      "step": 2132
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.0796137145313163,
+      "learning_rate": 3.3418929174426918e-06,
+      "loss": 0.4908,
+      "step": 2133
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5577675000716914,
+      "learning_rate": 3.3401234830385753e-06,
+      "loss": 0.457,
+      "step": 2134
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5830147544071842,
+      "learning_rate": 3.3383535740753813e-06,
+      "loss": 0.4715,
+      "step": 2135
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5860738051813335,
+      "learning_rate": 3.336583191552876e-06,
+      "loss": 0.4749,
+      "step": 2136
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5718370942498533,
+      "learning_rate": 3.334812336471089e-06,
+      "loss": 0.462,
+      "step": 2137
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5677796507381937,
+      "learning_rate": 3.3330410098303224e-06,
+      "loss": 0.4658,
+      "step": 2138
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5660363318045506,
+      "learning_rate": 3.3312692126311424e-06,
+      "loss": 0.4654,
+      "step": 2139
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5529225478270423,
+      "learning_rate": 3.32949694587438e-06,
+      "loss": 0.4853,
+      "step": 2140
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.558224467702825,
+      "learning_rate": 3.3277242105611334e-06,
+      "loss": 0.4635,
+      "step": 2141
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5895844984799891,
+      "learning_rate": 3.3259510076927644e-06,
+      "loss": 0.4831,
+      "step": 2142
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5474505234053266,
+      "learning_rate": 3.324177338270898e-06,
+      "loss": 0.4723,
+      "step": 2143
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5565011263756835,
+      "learning_rate": 3.322403203297424e-06,
+      "loss": 0.4524,
+      "step": 2144
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.653742264632734,
+      "learning_rate": 3.320628603774496e-06,
+      "loss": 0.4539,
+      "step": 2145
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5762670304100043,
+      "learning_rate": 3.3188535407045274e-06,
+      "loss": 0.4779,
+      "step": 2146
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.549175111157888,
+      "learning_rate": 3.317078015090197e-06,
+      "loss": 0.5013,
+      "step": 2147
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5736471852199507,
+      "learning_rate": 3.315302027934441e-06,
+      "loss": 0.4816,
+      "step": 2148
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.557613887379806,
+      "learning_rate": 3.313525580240459e-06,
+      "loss": 0.4756,
+      "step": 2149
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5468948120560473,
+      "learning_rate": 3.3117486730117092e-06,
+      "loss": 0.4753,
+      "step": 2150
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5267251356037557,
+      "learning_rate": 3.309971307251911e-06,
+      "loss": 0.4672,
+      "step": 2151
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5661351158225999,
+      "learning_rate": 3.3081934839650404e-06,
+      "loss": 0.4607,
+      "step": 2152
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5777629397975975,
+      "learning_rate": 3.3064152041553356e-06,
+      "loss": 0.4451,
+      "step": 2153
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5700643158197273,
+      "learning_rate": 3.304636468827288e-06,
+      "loss": 0.4695,
+      "step": 2154
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5615418059663319,
+      "learning_rate": 3.3028572789856507e-06,
+      "loss": 0.469,
+      "step": 2155
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5871509589578386,
+      "learning_rate": 3.30107763563543e-06,
+      "loss": 0.4827,
+      "step": 2156
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5558568799307586,
+      "learning_rate": 3.299297539781891e-06,
+      "loss": 0.4716,
+      "step": 2157
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5545006329981151,
+      "learning_rate": 3.2975169924305524e-06,
+      "loss": 0.4463,
+      "step": 2158
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 3.548575566071066,
+      "learning_rate": 3.29573599458719e-06,
+      "loss": 0.4889,
+      "step": 2159
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5213212611415827,
+      "learning_rate": 3.2939545472578314e-06,
+      "loss": 0.4867,
+      "step": 2160
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5843877680577335,
+      "learning_rate": 3.292172651448761e-06,
+      "loss": 0.4658,
+      "step": 2161
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5710485476550279,
+      "learning_rate": 3.290390308166515e-06,
+      "loss": 0.4722,
+      "step": 2162
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.5539562494433544,
+      "learning_rate": 3.2886075184178817e-06,
+      "loss": 0.4885,
+      "step": 2163
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6485074098343331,
+      "learning_rate": 3.2868242832099034e-06,
+      "loss": 0.471,
+      "step": 2164
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.6332549066175103,
+      "learning_rate": 3.285040603549872e-06,
+      "loss": 0.4821,
+      "step": 2165
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5894833989878483,
+      "learning_rate": 3.2832564804453327e-06,
+      "loss": 0.4571,
+      "step": 2166
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5558289841136125,
+      "learning_rate": 3.281471914904079e-06,
+      "loss": 0.4744,
+      "step": 2167
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5809032359736926,
+      "learning_rate": 3.2796869079341555e-06,
+      "loss": 0.4868,
+      "step": 2168
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5635833248240615,
+      "learning_rate": 3.2779014605438563e-06,
+      "loss": 0.4736,
+      "step": 2169
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5668615337342484,
+      "learning_rate": 3.276115573741724e-06,
+      "loss": 0.4737,
+      "step": 2170
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5702039104950186,
+      "learning_rate": 3.274329248536548e-06,
+      "loss": 0.459,
+      "step": 2171
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5220961360234252,
+      "learning_rate": 3.272542485937369e-06,
+      "loss": 0.4656,
+      "step": 2172
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5594033494520262,
+      "learning_rate": 3.270755286953471e-06,
+      "loss": 0.4814,
+      "step": 2173
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5741756929068141,
+      "learning_rate": 3.2689676525943854e-06,
+      "loss": 0.4815,
+      "step": 2174
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5464966733679589,
+      "learning_rate": 3.267179583869892e-06,
+      "loss": 0.4593,
+      "step": 2175
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5721831411264321,
+      "learning_rate": 3.265391081790012e-06,
+      "loss": 0.4761,
+      "step": 2176
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5456090495324294,
+      "learning_rate": 3.2636021473650143e-06,
+      "loss": 0.4875,
+      "step": 2177
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5463724345795404,
+      "learning_rate": 3.2618127816054117e-06,
+      "loss": 0.4726,
+      "step": 2178
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5367573084821134,
+      "learning_rate": 3.2600229855219595e-06,
+      "loss": 0.473,
+      "step": 2179
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5855822548291622,
+      "learning_rate": 3.2582327601256567e-06,
+      "loss": 0.4782,
+      "step": 2180
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5565384228650471,
+      "learning_rate": 3.256442106427745e-06,
+      "loss": 0.457,
+      "step": 2181
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5386204130175131,
+      "learning_rate": 3.254651025439707e-06,
+      "loss": 0.4684,
+      "step": 2182
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5334182218422864,
+      "learning_rate": 3.252859518173269e-06,
+      "loss": 0.4645,
+      "step": 2183
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5715830265673014,
+      "learning_rate": 3.251067585640395e-06,
+      "loss": 0.4772,
+      "step": 2184
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5741015269134933,
+      "learning_rate": 3.249275228853292e-06,
+      "loss": 0.4627,
+      "step": 2185
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5672462313761777,
+      "learning_rate": 3.247482448824405e-06,
+      "loss": 0.4903,
+      "step": 2186
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5373081330698016,
+      "learning_rate": 3.245689246566418e-06,
+      "loss": 0.4816,
+      "step": 2187
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.5476617186268243,
+      "learning_rate": 3.243895623092254e-06,
+      "loss": 0.4886,
+      "step": 2188
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5356939267254818,
+      "learning_rate": 3.2421015794150755e-06,
+      "loss": 0.4765,
+      "step": 2189
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5494604506543436,
+      "learning_rate": 3.240307116548279e-06,
+      "loss": 0.5035,
+      "step": 2190
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5930523473140368,
+      "learning_rate": 3.2385122355055004e-06,
+      "loss": 0.4962,
+      "step": 2191
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.602008763063395,
+      "learning_rate": 3.2367169373006114e-06,
+      "loss": 0.4892,
+      "step": 2192
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5651243722247239,
+      "learning_rate": 3.234921222947718e-06,
+      "loss": 0.4829,
+      "step": 2193
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5385560583261362,
+      "learning_rate": 3.2331250934611623e-06,
+      "loss": 0.482,
+      "step": 2194
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5547490494472022,
+      "learning_rate": 3.231328549855522e-06,
+      "loss": 0.4612,
+      "step": 2195
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5470853885746786,
+      "learning_rate": 3.2295315931456057e-06,
+      "loss": 0.4593,
+      "step": 2196
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5786280065830184,
+      "learning_rate": 3.227734224346458e-06,
+      "loss": 0.4769,
+      "step": 2197
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5344368582425154,
+      "learning_rate": 3.2259364444733567e-06,
+      "loss": 0.466,
+      "step": 2198
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5555074280873257,
+      "learning_rate": 3.2241382545418087e-06,
+      "loss": 0.4648,
+      "step": 2199
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5482269284810969,
+      "learning_rate": 3.222339655567556e-06,
+      "loss": 0.4769,
+      "step": 2200
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5667532379426644,
+      "learning_rate": 3.2205406485665693e-06,
+      "loss": 0.4695,
+      "step": 2201
+    },
+    {
+      "epoch": 0.94,
+      "eval_loss": 0.46910035610198975,
+      "eval_runtime": 6929.3666,
+      "eval_samples_per_second": 41.907,
+      "eval_steps_per_second": 2.095,
+      "step": 2201
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5516775609671802,
+      "learning_rate": 3.2187412345550493e-06,
+      "loss": 0.4626,
+      "step": 2202
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5274995673676252,
+      "learning_rate": 3.2169414145494306e-06,
+      "loss": 0.4542,
+      "step": 2203
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5594394034476171,
+      "learning_rate": 3.2151411895663713e-06,
+      "loss": 0.5077,
+      "step": 2204
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5592366706538346,
+      "learning_rate": 3.2133405606227636e-06,
+      "loss": 0.4836,
+      "step": 2205
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5598764473379652,
+      "learning_rate": 3.2115395287357247e-06,
+      "loss": 0.4808,
+      "step": 2206
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5593396084233587,
+      "learning_rate": 3.2097380949226004e-06,
+      "loss": 0.4692,
+      "step": 2207
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5338314791656622,
+      "learning_rate": 3.2079362602009633e-06,
+      "loss": 0.4823,
+      "step": 2208
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.530750479189892,
+      "learning_rate": 3.2061340255886135e-06,
+      "loss": 0.4787,
+      "step": 2209
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5710624896170967,
+      "learning_rate": 3.2043313921035747e-06,
+      "loss": 0.4536,
+      "step": 2210
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5731148940480776,
+      "learning_rate": 3.2025283607640985e-06,
+      "loss": 0.4813,
+      "step": 2211
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5946357271072944,
+      "learning_rate": 3.200724932588659e-06,
+      "loss": 0.4819,
+      "step": 2212
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5728672004460359,
+      "learning_rate": 3.1989211085959558e-06,
+      "loss": 0.4919,
+      "step": 2213
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5568909639263977,
+      "learning_rate": 3.197116889804913e-06,
+      "loss": 0.4864,
+      "step": 2214
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5378921515155873,
+      "learning_rate": 3.1953122772346757e-06,
+      "loss": 0.4708,
+      "step": 2215
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5540326517638553,
+      "learning_rate": 3.193507271904612e-06,
+      "loss": 0.4961,
+      "step": 2216
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5264231160255753,
+      "learning_rate": 3.191701874834312e-06,
+      "loss": 0.4335,
+      "step": 2217
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5506768071233391,
+      "learning_rate": 3.1898960870435875e-06,
+      "loss": 0.4793,
+      "step": 2218
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5399042085543564,
+      "learning_rate": 3.1880899095524698e-06,
+      "loss": 0.4538,
+      "step": 2219
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5470743327622264,
+      "learning_rate": 3.1862833433812137e-06,
+      "loss": 0.4624,
+      "step": 2220
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5366708239822617,
+      "learning_rate": 3.1844763895502876e-06,
+      "loss": 0.4735,
+      "step": 2221
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5593381531413127,
+      "learning_rate": 3.1826690490803846e-06,
+      "loss": 0.4577,
+      "step": 2222
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5380013310652985,
+      "learning_rate": 3.180861322992414e-06,
+      "loss": 0.4526,
+      "step": 2223
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5415307669206325,
+      "learning_rate": 3.179053212307502e-06,
+      "loss": 0.484,
+      "step": 2224
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5610218584434514,
+      "learning_rate": 3.1772447180469934e-06,
+      "loss": 0.4998,
+      "step": 2225
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5407354208309232,
+      "learning_rate": 3.1754358412324483e-06,
+      "loss": 0.4603,
+      "step": 2226
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.557457231879274,
+      "learning_rate": 3.173626582885645e-06,
+      "loss": 0.4657,
+      "step": 2227
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5712441602450918,
+      "learning_rate": 3.1718169440285763e-06,
+      "loss": 0.4893,
+      "step": 2228
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5433914547191268,
+      "learning_rate": 3.1700069256834478e-06,
+      "loss": 0.503,
+      "step": 2229
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.529494115478655,
+      "learning_rate": 3.1681965288726825e-06,
+      "loss": 0.4474,
+      "step": 2230
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5572186853202381,
+      "learning_rate": 3.166385754618917e-06,
+      "loss": 0.4572,
+      "step": 2231
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5258044718918907,
+      "learning_rate": 3.1645746039449987e-06,
+      "loss": 0.4461,
+      "step": 2232
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5420554964039433,
+      "learning_rate": 3.16276307787399e-06,
+      "loss": 0.4553,
+      "step": 2233
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5405232645917257,
+      "learning_rate": 3.1609511774291646e-06,
+      "loss": 0.4557,
+      "step": 2234
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.5847047282231195,
+      "learning_rate": 3.1591389036340064e-06,
+      "loss": 0.4527,
+      "step": 2235
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5423932667911106,
+      "learning_rate": 3.157326257512212e-06,
+      "loss": 0.457,
+      "step": 2236
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5692656034737968,
+      "learning_rate": 3.1555132400876877e-06,
+      "loss": 0.4943,
+      "step": 2237
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5413038156860533,
+      "learning_rate": 3.15369985238455e-06,
+      "loss": 0.4516,
+      "step": 2238
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5381862150816308,
+      "learning_rate": 3.151886095427123e-06,
+      "loss": 0.4753,
+      "step": 2239
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5727293376987612,
+      "learning_rate": 3.1500719702399406e-06,
+      "loss": 0.5059,
+      "step": 2240
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5805006268410494,
+      "learning_rate": 3.1482574778477447e-06,
+      "loss": 0.4524,
+      "step": 2241
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.528343362595312,
+      "learning_rate": 3.146442619275486e-06,
+      "loss": 0.471,
+      "step": 2242
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5677192912627839,
+      "learning_rate": 3.1446273955483173e-06,
+      "loss": 0.4462,
+      "step": 2243
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.569474282068865,
+      "learning_rate": 3.142811807691603e-06,
+      "loss": 0.4663,
+      "step": 2244
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5669828732246334,
+      "learning_rate": 3.1409958567309114e-06,
+      "loss": 0.4694,
+      "step": 2245
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.529668789550494,
+      "learning_rate": 3.1391795436920136e-06,
+      "loss": 0.4679,
+      "step": 2246
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5397155007947799,
+      "learning_rate": 3.1373628696008883e-06,
+      "loss": 0.4856,
+      "step": 2247
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5495612328617239,
+      "learning_rate": 3.1355458354837183e-06,
+      "loss": 0.4909,
+      "step": 2248
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5196306467160406,
+      "learning_rate": 3.133728442366885e-06,
+      "loss": 0.4525,
+      "step": 2249
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5278171503733031,
+      "learning_rate": 3.1319106912769797e-06,
+      "loss": 0.4458,
+      "step": 2250
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5621344379861554,
+      "learning_rate": 3.13009258324079e-06,
+      "loss": 0.4288,
+      "step": 2251
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5340342660151414,
+      "learning_rate": 3.128274119285309e-06,
+      "loss": 0.449,
+      "step": 2252
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5771753795948189,
+      "learning_rate": 3.1264553004377285e-06,
+      "loss": 0.493,
+      "step": 2253
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.6109163801950218,
+      "learning_rate": 3.1246361277254405e-06,
+      "loss": 0.5092,
+      "step": 2254
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5464493547347894,
+      "learning_rate": 3.122816602176039e-06,
+      "loss": 0.4831,
+      "step": 2255
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5388061503869606,
+      "learning_rate": 3.1209967248173167e-06,
+      "loss": 0.474,
+      "step": 2256
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5671598890046339,
+      "learning_rate": 3.119176496677263e-06,
+      "loss": 0.4911,
+      "step": 2257
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.5456085685153534,
+      "learning_rate": 3.1173559187840683e-06,
+      "loss": 0.4614,
+      "step": 2258
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.54513522335301,
+      "learning_rate": 3.115534992166119e-06,
+      "loss": 0.4692,
+      "step": 2259
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5262225869939646,
+      "learning_rate": 3.1137137178519983e-06,
+      "loss": 0.4568,
+      "step": 2260
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5473252038967218,
+      "learning_rate": 3.111892096870487e-06,
+      "loss": 0.4611,
+      "step": 2261
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5599766804442933,
+      "learning_rate": 3.1100701302505586e-06,
+      "loss": 0.4753,
+      "step": 2262
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5756313356809942,
+      "learning_rate": 3.1082478190213872e-06,
+      "loss": 0.4845,
+      "step": 2263
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5480923324133978,
+      "learning_rate": 3.106425164212338e-06,
+      "loss": 0.487,
+      "step": 2264
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5315541819290027,
+      "learning_rate": 3.1046021668529684e-06,
+      "loss": 0.4702,
+      "step": 2265
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5476930314697923,
+      "learning_rate": 3.1027788279730343e-06,
+      "loss": 0.4817,
+      "step": 2266
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.544515736247213,
+      "learning_rate": 3.1009551486024814e-06,
+      "loss": 0.4587,
+      "step": 2267
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5703546973497986,
+      "learning_rate": 3.099131129771448e-06,
+      "loss": 0.4617,
+      "step": 2268
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5424534272905377,
+      "learning_rate": 3.0973067725102636e-06,
+      "loss": 0.4484,
+      "step": 2269
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5733254545058311,
+      "learning_rate": 3.0954820778494516e-06,
+      "loss": 0.49,
+      "step": 2270
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5844361084620427,
+      "learning_rate": 3.093657046819722e-06,
+      "loss": 0.4734,
+      "step": 2271
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5574279308944012,
+      "learning_rate": 3.0918316804519784e-06,
+      "loss": 0.4574,
+      "step": 2272
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 0.467759370803833,
+      "eval_runtime": 6932.3606,
+      "eval_samples_per_second": 41.889,
+      "eval_steps_per_second": 2.095,
+      "step": 2272
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.546522412221111,
+      "learning_rate": 3.0900059797773114e-06,
+      "loss": 0.4886,
+      "step": 2273
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.553571296978123,
+      "learning_rate": 3.0881799458270005e-06,
+      "loss": 0.4543,
+      "step": 2274
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5237411938748142,
+      "learning_rate": 3.0863535796325173e-06,
+      "loss": 0.4556,
+      "step": 2275
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.544211874410188,
+      "learning_rate": 3.0845268822255155e-06,
+      "loss": 0.4775,
+      "step": 2276
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5661742324216515,
+      "learning_rate": 3.0826998546378385e-06,
+      "loss": 0.4602,
+      "step": 2277
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5241341056853029,
+      "learning_rate": 3.080872497901518e-06,
+      "loss": 0.4673,
+      "step": 2278
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5483140951807146,
+      "learning_rate": 3.079044813048768e-06,
+      "loss": 0.4986,
+      "step": 2279
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5515056093607098,
+      "learning_rate": 3.0772168011119894e-06,
+      "loss": 0.4472,
+      "step": 2280
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5594300479805302,
+      "learning_rate": 3.0753884631237706e-06,
+      "loss": 0.4762,
+      "step": 2281
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.5326582325370033,
+      "learning_rate": 3.073559800116879e-06,
+      "loss": 0.4652,
+      "step": 2282
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5354171305776954,
+      "learning_rate": 3.0717308131242695e-06,
+      "loss": 0.4585,
+      "step": 2283
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5414502214146208,
+      "learning_rate": 3.069901503179079e-06,
+      "loss": 0.4721,
+      "step": 2284
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5586551837722931,
+      "learning_rate": 3.068071871314626e-06,
+      "loss": 0.4283,
+      "step": 2285
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5627424065729489,
+      "learning_rate": 3.0662419185644117e-06,
+      "loss": 0.4811,
+      "step": 2286
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5553189229312547,
+      "learning_rate": 3.0644116459621177e-06,
+      "loss": 0.4749,
+      "step": 2287
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5758016314417642,
+      "learning_rate": 3.0625810545416066e-06,
+      "loss": 0.4696,
+      "step": 2288
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5625491392976549,
+      "learning_rate": 3.060750145336924e-06,
+      "loss": 0.4641,
+      "step": 2289
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5325354662459477,
+      "learning_rate": 3.0589189193822894e-06,
+      "loss": 0.4501,
+      "step": 2290
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5498616147688943,
+      "learning_rate": 3.057087377712106e-06,
+      "loss": 0.4697,
+      "step": 2291
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5402471923474919,
+      "learning_rate": 3.0552555213609526e-06,
+      "loss": 0.4746,
+      "step": 2292
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5742344531078118,
+      "learning_rate": 3.0534233513635863e-06,
+      "loss": 0.4597,
+      "step": 2293
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5660060095582498,
+      "learning_rate": 3.0515908687549427e-06,
+      "loss": 0.4712,
+      "step": 2294
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5449862197515469,
+      "learning_rate": 3.0497580745701334e-06,
+      "loss": 0.4658,
+      "step": 2295
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5637723063866732,
+      "learning_rate": 3.047924969844444e-06,
+      "loss": 0.4863,
+      "step": 2296
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5495824769869212,
+      "learning_rate": 3.046091555613339e-06,
+      "loss": 0.4703,
+      "step": 2297
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5382416875131586,
+      "learning_rate": 3.0442578329124545e-06,
+      "loss": 0.4584,
+      "step": 2298
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5463125942933166,
+      "learning_rate": 3.042423802777602e-06,
+      "loss": 0.4772,
+      "step": 2299
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5597328954873347,
+      "learning_rate": 3.0405894662447682e-06,
+      "loss": 0.4639,
+      "step": 2300
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5559561549207039,
+      "learning_rate": 3.038754824350111e-06,
+      "loss": 0.4847,
+      "step": 2301
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.548445944370762,
+      "learning_rate": 3.0369198781299615e-06,
+      "loss": 0.47,
+      "step": 2302
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5292092956017875,
+      "learning_rate": 3.0350846286208223e-06,
+      "loss": 0.4515,
+      "step": 2303
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5674894268853128,
+      "learning_rate": 3.0332490768593676e-06,
+      "loss": 0.4685,
+      "step": 2304
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5336634810588987,
+      "learning_rate": 3.0314132238824416e-06,
+      "loss": 0.4581,
+      "step": 2305
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5533286565854473,
+      "learning_rate": 3.029577070727061e-06,
+      "loss": 0.4477,
+      "step": 2306
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5680828232808299,
+      "learning_rate": 3.027740618430409e-06,
+      "loss": 0.4584,
+      "step": 2307
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5469367387429259,
+      "learning_rate": 3.0259038680298403e-06,
+      "loss": 0.4566,
+      "step": 2308
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.575161918224578,
+      "learning_rate": 3.0240668205628757e-06,
+      "loss": 0.4918,
+      "step": 2309
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5576124636437852,
+      "learning_rate": 3.0222294770672054e-06,
+      "loss": 0.4358,
+      "step": 2310
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5802410529155669,
+      "learning_rate": 3.0203918385806874e-06,
+      "loss": 0.4712,
+      "step": 2311
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5305957990981474,
+      "learning_rate": 3.018553906141343e-06,
+      "loss": 0.4656,
+      "step": 2312
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5503984759313564,
+      "learning_rate": 3.0167156807873637e-06,
+      "loss": 0.4711,
+      "step": 2313
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5629590646395772,
+      "learning_rate": 3.014877163557105e-06,
+      "loss": 0.4604,
+      "step": 2314
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5731070115431295,
+      "learning_rate": 3.013038355489086e-06,
+      "loss": 0.4543,
+      "step": 2315
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5625771430390166,
+      "learning_rate": 3.0111992576219905e-06,
+      "loss": 0.4564,
+      "step": 2316
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5811573131231536,
+      "learning_rate": 3.009359870994668e-06,
+      "loss": 0.4792,
+      "step": 2317
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5768867352155639,
+      "learning_rate": 3.0075201966461286e-06,
+      "loss": 0.4802,
+      "step": 2318
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5833346195049011,
+      "learning_rate": 3.0056802356155455e-06,
+      "loss": 0.4669,
+      "step": 2319
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5506865485168906,
+      "learning_rate": 3.0038399889422553e-06,
+      "loss": 0.4483,
+      "step": 2320
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.549893343788274,
+      "learning_rate": 3.001999457665754e-06,
+      "loss": 0.4608,
+      "step": 2321
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5742059606229357,
+      "learning_rate": 3.0001586428257006e-06,
+      "loss": 0.4632,
+      "step": 2322
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5710353252614344,
+      "learning_rate": 2.9983175454619114e-06,
+      "loss": 0.4906,
+      "step": 2323
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5661140686401654,
+      "learning_rate": 2.9964761666143638e-06,
+      "loss": 0.4691,
+      "step": 2324
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5832412124730281,
+      "learning_rate": 2.9946345073231964e-06,
+      "loss": 0.4715,
+      "step": 2325
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5322313638153089,
+      "learning_rate": 2.9927925686287006e-06,
+      "loss": 0.4397,
+      "step": 2326
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5532684507212673,
+      "learning_rate": 2.9909503515713324e-06,
+      "loss": 0.4526,
+      "step": 2327
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.5380218677936013,
+      "learning_rate": 2.9891078571917004e-06,
+      "loss": 0.4627,
+      "step": 2328
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.575331323008352,
+      "learning_rate": 2.987265086530571e-06,
+      "loss": 0.4582,
+      "step": 2329
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5720361448465936,
+      "learning_rate": 2.985422040628867e-06,
+      "loss": 0.4663,
+      "step": 2330
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.54998684465839,
+      "learning_rate": 2.983578720527667e-06,
+      "loss": 0.455,
+      "step": 2331
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5552738272333081,
+      "learning_rate": 2.981735127268202e-06,
+      "loss": 0.4705,
+      "step": 2332
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.552744557893692,
+      "learning_rate": 2.9798912618918617e-06,
+      "loss": 0.4633,
+      "step": 2333
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5445751179840383,
+      "learning_rate": 2.9780471254401868e-06,
+      "loss": 0.4736,
+      "step": 2334
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5333392734486967,
+      "learning_rate": 2.976202718954869e-06,
+      "loss": 0.4777,
+      "step": 2335
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.572472547878964,
+      "learning_rate": 2.9743580434777586e-06,
+      "loss": 0.487,
+      "step": 2336
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5344393742783472,
+      "learning_rate": 2.972513100050851e-06,
+      "loss": 0.4707,
+      "step": 2337
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5630355400902859,
+      "learning_rate": 2.970667889716298e-06,
+      "loss": 0.4366,
+      "step": 2338
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5643967515499763,
+      "learning_rate": 2.9688224135164e-06,
+      "loss": 0.466,
+      "step": 2339
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5624350508625989,
+      "learning_rate": 2.9669766724936074e-06,
+      "loss": 0.4352,
+      "step": 2340
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5863321023005329,
+      "learning_rate": 2.9651306676905213e-06,
+      "loss": 0.4963,
+      "step": 2341
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5263776129880875,
+      "learning_rate": 2.9632844001498908e-06,
+      "loss": 0.4565,
+      "step": 2342
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5610974245603717,
+      "learning_rate": 2.9614378709146136e-06,
+      "loss": 0.4661,
+      "step": 2343
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.4664944112300873,
+      "eval_runtime": 6923.3541,
+      "eval_samples_per_second": 41.943,
+      "eval_steps_per_second": 2.097,
+      "step": 2343
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5460765503187325,
+      "learning_rate": 2.9595910810277367e-06,
+      "loss": 0.4629,
+      "step": 2344
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5468673482394827,
+      "learning_rate": 2.957744031532451e-06,
+      "loss": 0.4583,
+      "step": 2345
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5470321763441149,
+      "learning_rate": 2.9558967234720976e-06,
+      "loss": 0.5088,
+      "step": 2346
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5599348363396319,
+      "learning_rate": 2.9540491578901625e-06,
+      "loss": 0.4596,
+      "step": 2347
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.554159234404833,
+      "learning_rate": 2.9522013358302754e-06,
+      "loss": 0.4577,
+      "step": 2348
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6005976463431216,
+      "learning_rate": 2.9503532583362126e-06,
+      "loss": 0.468,
+      "step": 2349
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5407554418974608,
+      "learning_rate": 2.948504926451896e-06,
+      "loss": 0.4715,
+      "step": 2350
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5744854947285752,
+      "learning_rate": 2.9466563412213873e-06,
+      "loss": 0.4941,
+      "step": 2351
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5637411990253361,
+      "learning_rate": 2.9448075036888944e-06,
+      "loss": 0.4573,
+      "step": 2352
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5656040322788783,
+      "learning_rate": 2.942958414898768e-06,
+      "loss": 0.4586,
+      "step": 2353
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5735314101382244,
+      "learning_rate": 2.941109075895499e-06,
+      "loss": 0.4493,
+      "step": 2354
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5467877193091774,
+      "learning_rate": 2.9392594877237194e-06,
+      "loss": 0.5006,
+      "step": 2355
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5362633073872797,
+      "learning_rate": 2.937409651428205e-06,
+      "loss": 0.4874,
+      "step": 2356
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5325968068801722,
+      "learning_rate": 2.935559568053867e-06,
+      "loss": 0.4914,
+      "step": 2357
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5378549381823456,
+      "learning_rate": 2.93370923864576e-06,
+      "loss": 0.4607,
+      "step": 2358
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5271154873814673,
+      "learning_rate": 2.9318586642490766e-06,
+      "loss": 0.44,
+      "step": 2359
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5924972443494668,
+      "learning_rate": 2.930007845909146e-06,
+      "loss": 0.5007,
+      "step": 2360
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5759884987421106,
+      "learning_rate": 2.9281567846714393e-06,
+      "loss": 0.4847,
+      "step": 2361
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5452440313358291,
+      "learning_rate": 2.92630548158156e-06,
+      "loss": 0.4728,
+      "step": 2362
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5491210743756382,
+      "learning_rate": 2.924453937685251e-06,
+      "loss": 0.4814,
+      "step": 2363
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5497417413168237,
+      "learning_rate": 2.9226021540283914e-06,
+      "loss": 0.479,
+      "step": 2364
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5440302176322239,
+      "learning_rate": 2.9207501316569936e-06,
+      "loss": 0.469,
+      "step": 2365
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5307306470354942,
+      "learning_rate": 2.918897871617207e-06,
+      "loss": 0.4494,
+      "step": 2366
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5533694018060741,
+      "learning_rate": 2.9170453749553158e-06,
+      "loss": 0.4528,
+      "step": 2367
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5821093073073446,
+      "learning_rate": 2.9151926427177345e-06,
+      "loss": 0.4847,
+      "step": 2368
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5810561155735557,
+      "learning_rate": 2.913339675951014e-06,
+      "loss": 0.4577,
+      "step": 2369
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5637656659230893,
+      "learning_rate": 2.911486475701835e-06,
+      "loss": 0.4705,
+      "step": 2370
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5646507829525883,
+      "learning_rate": 2.909633043017013e-06,
+      "loss": 0.4655,
+      "step": 2371
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5560085235238604,
+      "learning_rate": 2.9077793789434925e-06,
+      "loss": 0.4645,
+      "step": 2372
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5804591657223265,
+      "learning_rate": 2.905925484528349e-06,
+      "loss": 0.4473,
+      "step": 2373
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5183741960616122,
+      "learning_rate": 2.9040713608187896e-06,
+      "loss": 0.4561,
+      "step": 2374
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6496654700785708,
+      "learning_rate": 2.9022170088621497e-06,
+      "loss": 0.4481,
+      "step": 2375
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6258547954302262,
+      "learning_rate": 2.900362429705893e-06,
+      "loss": 0.4322,
+      "step": 2376
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7675149067928225,
+      "learning_rate": 2.8985076243976133e-06,
+      "loss": 0.3944,
+      "step": 2377
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6600899912780784,
+      "learning_rate": 2.896652593985031e-06,
+      "loss": 0.4245,
+      "step": 2378
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7306824483448988,
+      "learning_rate": 2.8947973395159934e-06,
+      "loss": 0.4426,
+      "step": 2379
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6516484002884269,
+      "learning_rate": 2.892941862038475e-06,
+      "loss": 0.4297,
+      "step": 2380
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5912395370707995,
+      "learning_rate": 2.8910861626005774e-06,
+      "loss": 0.4273,
+      "step": 2381
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6132721088402558,
+      "learning_rate": 2.889230242250525e-06,
+      "loss": 0.3981,
+      "step": 2382
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6491777150140556,
+      "learning_rate": 2.887374102036668e-06,
+      "loss": 0.4167,
+      "step": 2383
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6143918069720361,
+      "learning_rate": 2.8855177430074817e-06,
+      "loss": 0.4215,
+      "step": 2384
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5796986965059269,
+      "learning_rate": 2.883661166211564e-06,
+      "loss": 0.4328,
+      "step": 2385
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.6385596001339221,
+      "learning_rate": 2.881804372697637e-06,
+      "loss": 0.4189,
+      "step": 2386
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.6774158174574826,
+      "learning_rate": 2.879947363514543e-06,
+      "loss": 0.4257,
+      "step": 2387
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.6439324568387629,
+      "learning_rate": 2.878090139711249e-06,
+      "loss": 0.4203,
+      "step": 2388
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5928966275845924,
+      "learning_rate": 2.8762327023368408e-06,
+      "loss": 0.4118,
+      "step": 2389
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.6017304771309222,
+      "learning_rate": 2.8743750524405254e-06,
+      "loss": 0.4698,
+      "step": 2390
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5709073058681067,
+      "learning_rate": 2.872517191071631e-06,
+      "loss": 0.4357,
+      "step": 2391
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5680125069824045,
+      "learning_rate": 2.870659119279605e-06,
+      "loss": 0.4331,
+      "step": 2392
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5694888697631014,
+      "learning_rate": 2.8688008381140126e-06,
+      "loss": 0.4225,
+      "step": 2393
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.533441546679153,
+      "learning_rate": 2.866942348624538e-06,
+      "loss": 0.4106,
+      "step": 2394
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.6115603135836928,
+      "learning_rate": 2.8650836518609814e-06,
+      "loss": 0.4167,
+      "step": 2395
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.6206156844939555,
+      "learning_rate": 2.863224748873264e-06,
+      "loss": 0.4153,
+      "step": 2396
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5748859460500882,
+      "learning_rate": 2.8613656407114197e-06,
+      "loss": 0.4134,
+      "step": 2397
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5420087579510725,
+      "learning_rate": 2.8595063284255997e-06,
+      "loss": 0.42,
+      "step": 2398
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.6393458020851205,
+      "learning_rate": 2.85764681306607e-06,
+      "loss": 0.4448,
+      "step": 2399
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5570280986330635,
+      "learning_rate": 2.8557870956832135e-06,
+      "loss": 0.4192,
+      "step": 2400
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5742816886506813,
+      "learning_rate": 2.853927177327524e-06,
+      "loss": 0.4063,
+      "step": 2401
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5631462441989407,
+      "learning_rate": 2.85206705904961e-06,
+      "loss": 0.4187,
+      "step": 2402
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5759070659432811,
+      "learning_rate": 2.850206741900195e-06,
+      "loss": 0.4294,
+      "step": 2403
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5954714993206741,
+      "learning_rate": 2.8483462269301117e-06,
+      "loss": 0.4414,
+      "step": 2404
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5847923954668218,
+      "learning_rate": 2.8464855151903065e-06,
+      "loss": 0.4153,
+      "step": 2405
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.6128854329193607,
+      "learning_rate": 2.844624607731836e-06,
+      "loss": 0.4528,
+      "step": 2406
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5703811176973029,
+      "learning_rate": 2.842763505605867e-06,
+      "loss": 0.4222,
+      "step": 2407
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.5552066352368414,
+      "learning_rate": 2.8409022098636797e-06,
+      "loss": 0.436,
+      "step": 2408
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.6157208158352602,
+      "learning_rate": 2.839040721556658e-06,
+      "loss": 0.4302,
+      "step": 2409
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5647862801761142,
+      "learning_rate": 2.837179041736299e-06,
+      "loss": 0.4083,
+      "step": 2410
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5547355674329157,
+      "learning_rate": 2.835317171454206e-06,
+      "loss": 0.4162,
+      "step": 2411
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5766014610774604,
+      "learning_rate": 2.8334551117620908e-06,
+      "loss": 0.4226,
+      "step": 2412
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.6015597806488204,
+      "learning_rate": 2.8315928637117713e-06,
+      "loss": 0.4291,
+      "step": 2413
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5788471843088463,
+      "learning_rate": 2.829730428355173e-06,
+      "loss": 0.4251,
+      "step": 2414
+    },
+    {
+      "epoch": 1.02,
+      "eval_loss": 0.4677947163581848,
+      "eval_runtime": 6939.1205,
+      "eval_samples_per_second": 41.848,
+      "eval_steps_per_second": 2.092,
+      "step": 2414
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.586414114042501,
+      "learning_rate": 2.8278678067443255e-06,
+      "loss": 0.4145,
+      "step": 2415
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5691685054792884,
+      "learning_rate": 2.826004999931365e-06,
+      "loss": 0.4376,
+      "step": 2416
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5972775147112437,
+      "learning_rate": 2.8241420089685327e-06,
+      "loss": 0.4306,
+      "step": 2417
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5620380464963803,
+      "learning_rate": 2.8222788349081724e-06,
+      "loss": 0.4096,
+      "step": 2418
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5784135735472834,
+      "learning_rate": 2.820415478802733e-06,
+      "loss": 0.4322,
+      "step": 2419
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5885392171954246,
+      "learning_rate": 2.8185519417047624e-06,
+      "loss": 0.4039,
+      "step": 2420
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5536074618340685,
+      "learning_rate": 2.8166882246669158e-06,
+      "loss": 0.4291,
+      "step": 2421
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.581909431594722,
+      "learning_rate": 2.814824328741948e-06,
+      "loss": 0.4144,
+      "step": 2422
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5675861206607421,
+      "learning_rate": 2.8129602549827133e-06,
+      "loss": 0.419,
+      "step": 2423
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5632432133699884,
+      "learning_rate": 2.811096004442168e-06,
+      "loss": 0.4252,
+      "step": 2424
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5650569070218456,
+      "learning_rate": 2.80923157817337e-06,
+      "loss": 0.4136,
+      "step": 2425
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.572161063598354,
+      "learning_rate": 2.8073669772294714e-06,
+      "loss": 0.4208,
+      "step": 2426
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5719901179312831,
+      "learning_rate": 2.805502202663728e-06,
+      "loss": 0.4375,
+      "step": 2427
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5612496619770833,
+      "learning_rate": 2.8036372555294916e-06,
+      "loss": 0.4047,
+      "step": 2428
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5900526811179282,
+      "learning_rate": 2.8017721368802105e-06,
+      "loss": 0.4295,
+      "step": 2429
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5789988982356299,
+      "learning_rate": 2.799906847769433e-06,
+      "loss": 0.4138,
+      "step": 2430
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5438179046584847,
+      "learning_rate": 2.7980413892507995e-06,
+      "loss": 0.408,
+      "step": 2431
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.5857502120912527,
+      "learning_rate": 2.79617576237805e-06,
+      "loss": 0.406,
+      "step": 2432
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5798751246080152,
+      "learning_rate": 2.7943099682050174e-06,
+      "loss": 0.4173,
+      "step": 2433
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5830921819980437,
+      "learning_rate": 2.7924440077856284e-06,
+      "loss": 0.4353,
+      "step": 2434
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5550145268591252,
+      "learning_rate": 2.790577882173906e-06,
+      "loss": 0.4033,
+      "step": 2435
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5665020917284002,
+      "learning_rate": 2.788711592423966e-06,
+      "loss": 0.4282,
+      "step": 2436
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5619717205524801,
+      "learning_rate": 2.786845139590014e-06,
+      "loss": 0.4161,
+      "step": 2437
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5647215964187846,
+      "learning_rate": 2.7849785247263515e-06,
+      "loss": 0.419,
+      "step": 2438
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.6047710310723298,
+      "learning_rate": 2.7831117488873703e-06,
+      "loss": 0.4489,
+      "step": 2439
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5729169302378637,
+      "learning_rate": 2.781244813127552e-06,
+      "loss": 0.4258,
+      "step": 2440
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5726895367017322,
+      "learning_rate": 2.779377718501469e-06,
+      "loss": 0.4163,
+      "step": 2441
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5898478621812457,
+      "learning_rate": 2.7775104660637847e-06,
+      "loss": 0.4383,
+      "step": 2442
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5776286383338888,
+      "learning_rate": 2.77564305686925e-06,
+      "loss": 0.4172,
+      "step": 2443
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5708128845728648,
+      "learning_rate": 2.7737754919727057e-06,
+      "loss": 0.433,
+      "step": 2444
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5838647193488192,
+      "learning_rate": 2.7719077724290793e-06,
+      "loss": 0.4319,
+      "step": 2445
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5839908538746356,
+      "learning_rate": 2.7700398992933865e-06,
+      "loss": 0.3981,
+      "step": 2446
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5793472385138957,
+      "learning_rate": 2.76817187362073e-06,
+      "loss": 0.4075,
+      "step": 2447
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5727590490167658,
+      "learning_rate": 2.7663036964662967e-06,
+      "loss": 0.4312,
+      "step": 2448
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5920736043473027,
+      "learning_rate": 2.764435368885362e-06,
+      "loss": 0.4138,
+      "step": 2449
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5723166701159088,
+      "learning_rate": 2.762566891933285e-06,
+      "loss": 0.4073,
+      "step": 2450
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5624668781788753,
+      "learning_rate": 2.7606982666655074e-06,
+      "loss": 0.429,
+      "step": 2451
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5704524501724009,
+      "learning_rate": 2.758829494137557e-06,
+      "loss": 0.4196,
+      "step": 2452
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5700116449229535,
+      "learning_rate": 2.7569605754050455e-06,
+      "loss": 0.4317,
+      "step": 2453
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.599236712281217,
+      "learning_rate": 2.7550915115236636e-06,
+      "loss": 0.4087,
+      "step": 2454
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5731708478648513,
+      "learning_rate": 2.7532223035491877e-06,
+      "loss": 0.4149,
+      "step": 2455
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5748213937213151,
+      "learning_rate": 2.7513529525374725e-06,
+      "loss": 0.4041,
+      "step": 2456
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5736837321137316,
+      "learning_rate": 2.749483459544457e-06,
+      "loss": 0.4282,
+      "step": 2457
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5637414256487769,
+      "learning_rate": 2.7476138256261575e-06,
+      "loss": 0.3905,
+      "step": 2458
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.6013990347615658,
+      "learning_rate": 2.74574405183867e-06,
+      "loss": 0.4183,
+      "step": 2459
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5581978111233568,
+      "learning_rate": 2.743874139238171e-06,
+      "loss": 0.4325,
+      "step": 2460
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5658199630731812,
+      "learning_rate": 2.7420040888809153e-06,
+      "loss": 0.4035,
+      "step": 2461
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5691729761752431,
+      "learning_rate": 2.740133901823234e-06,
+      "loss": 0.4171,
+      "step": 2462
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5571943454995789,
+      "learning_rate": 2.7382635791215368e-06,
+      "loss": 0.4092,
+      "step": 2463
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5584570649456075,
+      "learning_rate": 2.7363931218323103e-06,
+      "loss": 0.4192,
+      "step": 2464
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.579128238271553,
+      "learning_rate": 2.7345225310121155e-06,
+      "loss": 0.4404,
+      "step": 2465
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5929460640663197,
+      "learning_rate": 2.7326518077175897e-06,
+      "loss": 0.4197,
+      "step": 2466
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5689782317669538,
+      "learning_rate": 2.7307809530054456e-06,
+      "loss": 0.4316,
+      "step": 2467
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5854607172774103,
+      "learning_rate": 2.7289099679324686e-06,
+      "loss": 0.4431,
+      "step": 2468
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5694064345867531,
+      "learning_rate": 2.7270388535555207e-06,
+      "loss": 0.4229,
+      "step": 2469
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5694897278258253,
+      "learning_rate": 2.725167610931534e-06,
+      "loss": 0.4122,
+      "step": 2470
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5799667414945328,
+      "learning_rate": 2.7232962411175128e-06,
+      "loss": 0.4227,
+      "step": 2471
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5880421037268969,
+      "learning_rate": 2.721424745170537e-06,
+      "loss": 0.4239,
+      "step": 2472
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5645195680029945,
+      "learning_rate": 2.719553124147753e-06,
+      "loss": 0.4131,
+      "step": 2473
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5549310632780857,
+      "learning_rate": 2.717681379106381e-06,
+      "loss": 0.4221,
+      "step": 2474
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5724363209257333,
+      "learning_rate": 2.715809511103711e-06,
+      "loss": 0.4331,
+      "step": 2475
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5646087777604074,
+      "learning_rate": 2.7139375211971e-06,
+      "loss": 0.4285,
+      "step": 2476
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5589589680565078,
+      "learning_rate": 2.712065410443977e-06,
+      "loss": 0.4225,
+      "step": 2477
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5975042188527351,
+      "learning_rate": 2.710193179901838e-06,
+      "loss": 0.4475,
+      "step": 2478
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5632223523751451,
+      "learning_rate": 2.7083208306282455e-06,
+      "loss": 0.4195,
+      "step": 2479
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.6271794519728584,
+      "learning_rate": 2.7064483636808314e-06,
+      "loss": 0.4383,
+      "step": 2480
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.6193844438411981,
+      "learning_rate": 2.7045757801172918e-06,
+      "loss": 0.4405,
+      "step": 2481
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5762357913780729,
+      "learning_rate": 2.70270308099539e-06,
+      "loss": 0.4256,
+      "step": 2482
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5686615359961465,
+      "learning_rate": 2.7008302673729556e-06,
+      "loss": 0.4228,
+      "step": 2483
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5846035613411025,
+      "learning_rate": 2.6989573403078793e-06,
+      "loss": 0.4424,
+      "step": 2484
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.57106132811842,
+      "learning_rate": 2.69708430085812e-06,
+      "loss": 0.4249,
+      "step": 2485
+    },
+    {
+      "epoch": 1.05,
+      "eval_loss": 0.4666215181350708,
+      "eval_runtime": 6941.4114,
+      "eval_samples_per_second": 41.834,
+      "eval_steps_per_second": 2.092,
+      "step": 2485
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5599313713923679,
+      "learning_rate": 2.6952111500816972e-06,
+      "loss": 0.4247,
+      "step": 2486
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5633818405700421,
+      "learning_rate": 2.6933378890366945e-06,
+      "loss": 0.4292,
+      "step": 2487
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5687785418378354,
+      "learning_rate": 2.6914645187812573e-06,
+      "loss": 0.4238,
+      "step": 2488
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5882193651407125,
+      "learning_rate": 2.6895910403735938e-06,
+      "loss": 0.4105,
+      "step": 2489
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5745904096533959,
+      "learning_rate": 2.687717454871971e-06,
+      "loss": 0.4307,
+      "step": 2490
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5600088359405079,
+      "learning_rate": 2.6858437633347197e-06,
+      "loss": 0.4149,
+      "step": 2491
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5900737321936005,
+      "learning_rate": 2.6839699668202275e-06,
+      "loss": 0.4426,
+      "step": 2492
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5993093534443086,
+      "learning_rate": 2.682096066386943e-06,
+      "loss": 0.4079,
+      "step": 2493
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.573184519185299,
+      "learning_rate": 2.680222063093372e-06,
+      "loss": 0.4052,
+      "step": 2494
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5618782823660212,
+      "learning_rate": 2.678347957998081e-06,
+      "loss": 0.4364,
+      "step": 2495
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.584859311676964,
+      "learning_rate": 2.6764737521596917e-06,
+      "loss": 0.4052,
+      "step": 2496
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5477234902368038,
+      "learning_rate": 2.6745994466368846e-06,
+      "loss": 0.4302,
+      "step": 2497
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.6063996852367219,
+      "learning_rate": 2.672725042488393e-06,
+      "loss": 0.4181,
+      "step": 2498
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5760512448701784,
+      "learning_rate": 2.6708505407730106e-06,
+      "loss": 0.4134,
+      "step": 2499
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5519722703829206,
+      "learning_rate": 2.6689759425495833e-06,
+      "loss": 0.4222,
+      "step": 2500
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5683672052752755,
+      "learning_rate": 2.6671012488770104e-06,
+      "loss": 0.4268,
+      "step": 2501
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.5877420100279226,
+      "learning_rate": 2.6652264608142487e-06,
+      "loss": 0.4178,
+      "step": 2502
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.607133857685082,
+      "learning_rate": 2.663351579420307e-06,
+      "loss": 0.417,
+      "step": 2503
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5772639249289018,
+      "learning_rate": 2.661476605754244e-06,
+      "loss": 0.4377,
+      "step": 2504
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.6163851599426933,
+      "learning_rate": 2.659601540875174e-06,
+      "loss": 0.4059,
+      "step": 2505
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.6150547368429744,
+      "learning_rate": 2.6577263858422623e-06,
+      "loss": 0.4473,
+      "step": 2506
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.6277681834624624,
+      "learning_rate": 2.6558511417147225e-06,
+      "loss": 0.4312,
+      "step": 2507
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5576110114818135,
+      "learning_rate": 2.653975809551823e-06,
+      "loss": 0.415,
+      "step": 2508
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5603219178351719,
+      "learning_rate": 2.6521003904128772e-06,
+      "loss": 0.4021,
+      "step": 2509
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5842308669890593,
+      "learning_rate": 2.650224885357251e-06,
+      "loss": 0.4377,
+      "step": 2510
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.6027928815424282,
+      "learning_rate": 2.648349295444358e-06,
+      "loss": 0.4326,
+      "step": 2511
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.6056130362871082,
+      "learning_rate": 2.646473621733658e-06,
+      "loss": 0.4253,
+      "step": 2512
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.6243077090025687,
+      "learning_rate": 2.6445978652846605e-06,
+      "loss": 0.4478,
+      "step": 2513
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.6116645505267284,
+      "learning_rate": 2.6427220271569206e-06,
+      "loss": 0.4576,
+      "step": 2514
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5491346480936342,
+      "learning_rate": 2.640846108410039e-06,
+      "loss": 0.3972,
+      "step": 2515
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5681342786794281,
+      "learning_rate": 2.6389701101036635e-06,
+      "loss": 0.4351,
+      "step": 2516
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5866245904284294,
+      "learning_rate": 2.6370940332974864e-06,
+      "loss": 0.4201,
+      "step": 2517
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5501785457503201,
+      "learning_rate": 2.6352178790512425e-06,
+      "loss": 0.3985,
+      "step": 2518
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5783704626903385,
+      "learning_rate": 2.6333416484247126e-06,
+      "loss": 0.4582,
+      "step": 2519
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5601581473403204,
+      "learning_rate": 2.6314653424777194e-06,
+      "loss": 0.4502,
+      "step": 2520
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5776661184654673,
+      "learning_rate": 2.6295889622701287e-06,
+      "loss": 0.3946,
+      "step": 2521
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.6070561840129438,
+      "learning_rate": 2.6277125088618496e-06,
+      "loss": 0.4253,
+      "step": 2522
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5860572342161754,
+      "learning_rate": 2.6258359833128284e-06,
+      "loss": 0.4134,
+      "step": 2523
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5814356096505094,
+      "learning_rate": 2.623959386683056e-06,
+      "loss": 0.4144,
+      "step": 2524
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.5645239452943359,
+      "learning_rate": 2.6220827200325628e-06,
+      "loss": 0.4072,
+      "step": 2525
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5555072889719695,
+      "learning_rate": 2.620205984421418e-06,
+      "loss": 0.4115,
+      "step": 2526
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5857812167489317,
+      "learning_rate": 2.618329180909728e-06,
+      "loss": 0.3889,
+      "step": 2527
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5540712238808715,
+      "learning_rate": 2.6164523105576436e-06,
+      "loss": 0.4143,
+      "step": 2528
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5726575598093935,
+      "learning_rate": 2.614575374425345e-06,
+      "loss": 0.4062,
+      "step": 2529
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.580473478944349,
+      "learning_rate": 2.612698373573056e-06,
+      "loss": 0.4155,
+      "step": 2530
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5921217133427389,
+      "learning_rate": 2.6108213090610352e-06,
+      "loss": 0.4462,
+      "step": 2531
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5801087342197929,
+      "learning_rate": 2.608944181949575e-06,
+      "loss": 0.4099,
+      "step": 2532
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5749901636296175,
+      "learning_rate": 2.607066993299007e-06,
+      "loss": 0.4093,
+      "step": 2533
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.6164831721085331,
+      "learning_rate": 2.6051897441696926e-06,
+      "loss": 0.4283,
+      "step": 2534
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5874464536341129,
+      "learning_rate": 2.603312435622033e-06,
+      "loss": 0.4446,
+      "step": 2535
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5662147896859149,
+      "learning_rate": 2.6014350687164598e-06,
+      "loss": 0.4194,
+      "step": 2536
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5495971494931703,
+      "learning_rate": 2.5995576445134364e-06,
+      "loss": 0.4057,
+      "step": 2537
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5722070713750189,
+      "learning_rate": 2.5976801640734604e-06,
+      "loss": 0.4355,
+      "step": 2538
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5594258522910831,
+      "learning_rate": 2.595802628457063e-06,
+      "loss": 0.4238,
+      "step": 2539
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5700766032916892,
+      "learning_rate": 2.593925038724802e-06,
+      "loss": 0.4286,
+      "step": 2540
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5778449462446359,
+      "learning_rate": 2.5920473959372695e-06,
+      "loss": 0.4337,
+      "step": 2541
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.584060965567766,
+      "learning_rate": 2.5901697011550857e-06,
+      "loss": 0.4259,
+      "step": 2542
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.563752954121171,
+      "learning_rate": 2.5882919554389007e-06,
+      "loss": 0.4309,
+      "step": 2543
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5899104452933682,
+      "learning_rate": 2.586414159849394e-06,
+      "loss": 0.4461,
+      "step": 2544
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5820330641642761,
+      "learning_rate": 2.5845363154472725e-06,
+      "loss": 0.4301,
+      "step": 2545
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5446163156937063,
+      "learning_rate": 2.5826584232932707e-06,
+      "loss": 0.3957,
+      "step": 2546
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5945376021090512,
+      "learning_rate": 2.5807804844481506e-06,
+      "loss": 0.4091,
+      "step": 2547
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5638002168999431,
+      "learning_rate": 2.5789024999727e-06,
+      "loss": 0.414,
+      "step": 2548
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5951090126233547,
+      "learning_rate": 2.577024470927732e-06,
+      "loss": 0.4237,
+      "step": 2549
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5684255566290534,
+      "learning_rate": 2.575146398374087e-06,
+      "loss": 0.4431,
+      "step": 2550
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5595853495309726,
+      "learning_rate": 2.5732682833726274e-06,
+      "loss": 0.4221,
+      "step": 2551
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5742831043384634,
+      "learning_rate": 2.5713901269842405e-06,
+      "loss": 0.4285,
+      "step": 2552
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5640089992024232,
+      "learning_rate": 2.569511930269839e-06,
+      "loss": 0.4196,
+      "step": 2553
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5572383872005084,
+      "learning_rate": 2.5676336942903547e-06,
+      "loss": 0.4015,
+      "step": 2554
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.6180725350224047,
+      "learning_rate": 2.565755420106744e-06,
+      "loss": 0.4346,
+      "step": 2555
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.586931012696472,
+      "learning_rate": 2.5638771087799857e-06,
+      "loss": 0.4377,
+      "step": 2556
+    },
+    {
+      "epoch": 1.08,
+      "eval_loss": 0.46661555767059326,
+      "eval_runtime": 6940.8186,
+      "eval_samples_per_second": 41.838,
+      "eval_steps_per_second": 2.092,
+      "step": 2556
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.60890234512916,
+      "learning_rate": 2.5619987613710757e-06,
+      "loss": 0.4203,
+      "step": 2557
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.6130809794076856,
+      "learning_rate": 2.5601203789410344e-06,
+      "loss": 0.3926,
+      "step": 2558
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5758560877978278,
+      "learning_rate": 2.5582419625509004e-06,
+      "loss": 0.4193,
+      "step": 2559
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5611835821629346,
+      "learning_rate": 2.5563635132617305e-06,
+      "loss": 0.4106,
+      "step": 2560
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5500032992402379,
+      "learning_rate": 2.5544850321346026e-06,
+      "loss": 0.4103,
+      "step": 2561
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5991303836828948,
+      "learning_rate": 2.55260652023061e-06,
+      "loss": 0.4176,
+      "step": 2562
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5495267980558447,
+      "learning_rate": 2.550727978610864e-06,
+      "loss": 0.412,
+      "step": 2563
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5903181299884372,
+      "learning_rate": 2.5488494083364946e-06,
+      "loss": 0.4186,
+      "step": 2564
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5775725024291222,
+      "learning_rate": 2.5469708104686452e-06,
+      "loss": 0.4017,
+      "step": 2565
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5794156104604055,
+      "learning_rate": 2.5450921860684765e-06,
+      "loss": 0.3933,
+      "step": 2566
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5929127352748987,
+      "learning_rate": 2.543213536197164e-06,
+      "loss": 0.4474,
+      "step": 2567
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.6307527970701059,
+      "learning_rate": 2.5413348619158966e-06,
+      "loss": 0.4367,
+      "step": 2568
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5792027225417186,
+      "learning_rate": 2.5394561642858785e-06,
+      "loss": 0.4371,
+      "step": 2569
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5781330553234431,
+      "learning_rate": 2.5375774443683263e-06,
+      "loss": 0.416,
+      "step": 2570
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.574771297321989,
+      "learning_rate": 2.5356987032244686e-06,
+      "loss": 0.4075,
+      "step": 2571
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.5558908214262773,
+      "learning_rate": 2.5338199419155473e-06,
+      "loss": 0.41,
+      "step": 2572
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.570022965299663,
+      "learning_rate": 2.5319411615028144e-06,
+      "loss": 0.4197,
+      "step": 2573
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5953309626927854,
+      "learning_rate": 2.530062363047534e-06,
+      "loss": 0.4489,
+      "step": 2574
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5678767865025237,
+      "learning_rate": 2.5281835476109796e-06,
+      "loss": 0.4328,
+      "step": 2575
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5830901884956002,
+      "learning_rate": 2.5263047162544335e-06,
+      "loss": 0.4117,
+      "step": 2576
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.565926311345139,
+      "learning_rate": 2.5244258700391888e-06,
+      "loss": 0.4232,
+      "step": 2577
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5813850996625575,
+      "learning_rate": 2.522547010026546e-06,
+      "loss": 0.4351,
+      "step": 2578
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5766636305766427,
+      "learning_rate": 2.5206681372778126e-06,
+      "loss": 0.4281,
+      "step": 2579
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.547136463994455,
+      "learning_rate": 2.518789252854305e-06,
+      "loss": 0.399,
+      "step": 2580
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5673733087045142,
+      "learning_rate": 2.5169103578173455e-06,
+      "loss": 0.4429,
+      "step": 2581
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.577885776445999,
+      "learning_rate": 2.5150314532282615e-06,
+      "loss": 0.4373,
+      "step": 2582
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5681163890434137,
+      "learning_rate": 2.5131525401483863e-06,
+      "loss": 0.4314,
+      "step": 2583
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5914855836747088,
+      "learning_rate": 2.51127361963906e-06,
+      "loss": 0.4156,
+      "step": 2584
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5695378199879212,
+      "learning_rate": 2.5093946927616227e-06,
+      "loss": 0.4137,
+      "step": 2585
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.615520565415579,
+      "learning_rate": 2.507515760577423e-06,
+      "loss": 0.4202,
+      "step": 2586
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5962868782142557,
+      "learning_rate": 2.505636824147808e-06,
+      "loss": 0.4433,
+      "step": 2587
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5642173652784211,
+      "learning_rate": 2.50375788453413e-06,
+      "loss": 0.4296,
+      "step": 2588
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5643198205173939,
+      "learning_rate": 2.501878942797743e-06,
+      "loss": 0.4163,
+      "step": 2589
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5730693457148456,
+      "learning_rate": 2.5e-06,
+      "loss": 0.4184,
+      "step": 2590
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5648030559058376,
+      "learning_rate": 2.498121057202258e-06,
+      "loss": 0.4286,
+      "step": 2591
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.7310458154947486,
+      "learning_rate": 2.4962421154658706e-06,
+      "loss": 0.4552,
+      "step": 2592
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.8642590895577826,
+      "learning_rate": 2.4943631758521924e-06,
+      "loss": 0.43,
+      "step": 2593
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5957297069878567,
+      "learning_rate": 2.492484239422578e-06,
+      "loss": 0.4174,
+      "step": 2594
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5288913445348521,
+      "learning_rate": 2.4906053072383773e-06,
+      "loss": 0.4197,
+      "step": 2595
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5669908964180154,
+      "learning_rate": 2.4887263803609415e-06,
+      "loss": 0.4228,
+      "step": 2596
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5619854120471871,
+      "learning_rate": 2.486847459851614e-06,
+      "loss": 0.4204,
+      "step": 2597
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.573591559908782,
+      "learning_rate": 2.4849685467717397e-06,
+      "loss": 0.4043,
+      "step": 2598
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5660399643338767,
+      "learning_rate": 2.4830896421826554e-06,
+      "loss": 0.4151,
+      "step": 2599
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.6061575332078436,
+      "learning_rate": 2.4812107471456958e-06,
+      "loss": 0.4261,
+      "step": 2600
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.558833085401998,
+      "learning_rate": 2.479331862722188e-06,
+      "loss": 0.4253,
+      "step": 2601
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5661726057886762,
+      "learning_rate": 2.477452989973455e-06,
+      "loss": 0.4282,
+      "step": 2602
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5765250370061781,
+      "learning_rate": 2.475574129960812e-06,
+      "loss": 0.3895,
+      "step": 2603
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.6094913200037948,
+      "learning_rate": 2.4736952837455665e-06,
+      "loss": 0.4438,
+      "step": 2604
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5958403528694752,
+      "learning_rate": 2.4718164523890212e-06,
+      "loss": 0.4276,
+      "step": 2605
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5711855859912395,
+      "learning_rate": 2.4699376369524665e-06,
+      "loss": 0.422,
+      "step": 2606
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5827746728273364,
+      "learning_rate": 2.4680588384971864e-06,
+      "loss": 0.4094,
+      "step": 2607
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5652093041093819,
+      "learning_rate": 2.4661800580844535e-06,
+      "loss": 0.4404,
+      "step": 2608
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5815716319807065,
+      "learning_rate": 2.4643012967755327e-06,
+      "loss": 0.4188,
+      "step": 2609
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5667592036448388,
+      "learning_rate": 2.4624225556316745e-06,
+      "loss": 0.4103,
+      "step": 2610
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.6003054597720718,
+      "learning_rate": 2.4605438357141223e-06,
+      "loss": 0.4422,
+      "step": 2611
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5718504015079162,
+      "learning_rate": 2.458665138084104e-06,
+      "loss": 0.4057,
+      "step": 2612
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5894624642862655,
+      "learning_rate": 2.4567864638028374e-06,
+      "loss": 0.4164,
+      "step": 2613
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.6181729940652014,
+      "learning_rate": 2.4549078139315243e-06,
+      "loss": 0.4354,
+      "step": 2614
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5831842560896243,
+      "learning_rate": 2.453029189531356e-06,
+      "loss": 0.4059,
+      "step": 2615
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5711907705253331,
+      "learning_rate": 2.451150591663506e-06,
+      "loss": 0.4161,
+      "step": 2616
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5840971955664861,
+      "learning_rate": 2.449272021389136e-06,
+      "loss": 0.4266,
+      "step": 2617
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.6151549174106447,
+      "learning_rate": 2.4473934797693908e-06,
+      "loss": 0.4451,
+      "step": 2618
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5912926857180598,
+      "learning_rate": 2.4455149678653982e-06,
+      "loss": 0.424,
+      "step": 2619
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5690720352177238,
+      "learning_rate": 2.44363648673827e-06,
+      "loss": 0.4094,
+      "step": 2620
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5880654368148057,
+      "learning_rate": 2.4417580374491e-06,
+      "loss": 0.4374,
+      "step": 2621
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.592818300351086,
+      "learning_rate": 2.439879621058966e-06,
+      "loss": 0.4392,
+      "step": 2622
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5904613946521675,
+      "learning_rate": 2.438001238628925e-06,
+      "loss": 0.421,
+      "step": 2623
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5780768348236618,
+      "learning_rate": 2.436122891220016e-06,
+      "loss": 0.412,
+      "step": 2624
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.6165111418892802,
+      "learning_rate": 2.4342445798932563e-06,
+      "loss": 0.4225,
+      "step": 2625
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5864395429245397,
+      "learning_rate": 2.4323663057096466e-06,
+      "loss": 0.4158,
+      "step": 2626
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5790868452034496,
+      "learning_rate": 2.4304880697301615e-06,
+      "loss": 0.4018,
+      "step": 2627
+    },
+    {
+      "epoch": 1.11,
+      "eval_loss": 0.46582120656967163,
+      "eval_runtime": 6934.7187,
+      "eval_samples_per_second": 41.875,
+      "eval_steps_per_second": 2.094,
+      "step": 2627
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5854060000762511,
+      "learning_rate": 2.42860987301576e-06,
+      "loss": 0.429,
+      "step": 2628
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5793144094485164,
+      "learning_rate": 2.4267317166273734e-06,
+      "loss": 0.4008,
+      "step": 2629
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.557725397610062,
+      "learning_rate": 2.4248536016259137e-06,
+      "loss": 0.4339,
+      "step": 2630
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5707416944794899,
+      "learning_rate": 2.422975529072269e-06,
+      "loss": 0.4144,
+      "step": 2631
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5781786267589856,
+      "learning_rate": 2.4210975000273005e-06,
+      "loss": 0.4475,
+      "step": 2632
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5688305465034205,
+      "learning_rate": 2.41921951555185e-06,
+      "loss": 0.4145,
+      "step": 2633
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.583143118932043,
+      "learning_rate": 2.4173415767067297e-06,
+      "loss": 0.4286,
+      "step": 2634
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5507249017882659,
+      "learning_rate": 2.4154636845527284e-06,
+      "loss": 0.4137,
+      "step": 2635
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.6128066967756481,
+      "learning_rate": 2.4135858401506066e-06,
+      "loss": 0.423,
+      "step": 2636
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5984198722869004,
+      "learning_rate": 2.4117080445611e-06,
+      "loss": 0.4315,
+      "step": 2637
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.566032787321914,
+      "learning_rate": 2.4098302988449147e-06,
+      "loss": 0.4614,
+      "step": 2638
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5507111424965526,
+      "learning_rate": 2.4079526040627318e-06,
+      "loss": 0.4359,
+      "step": 2639
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5688047320342015,
+      "learning_rate": 2.4060749612751987e-06,
+      "loss": 0.4466,
+      "step": 2640
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.580032240572495,
+      "learning_rate": 2.404197371542938e-06,
+      "loss": 0.4032,
+      "step": 2641
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.5820043004603282,
+      "learning_rate": 2.40231983592654e-06,
+      "loss": 0.42,
+      "step": 2642
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5788132559135092,
+      "learning_rate": 2.400442355486564e-06,
+      "loss": 0.4183,
+      "step": 2643
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5859683700448465,
+      "learning_rate": 2.398564931283541e-06,
+      "loss": 0.4123,
+      "step": 2644
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5862017007386782,
+      "learning_rate": 2.396687564377967e-06,
+      "loss": 0.4285,
+      "step": 2645
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5925607615927038,
+      "learning_rate": 2.394810255830308e-06,
+      "loss": 0.4283,
+      "step": 2646
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5703213873210229,
+      "learning_rate": 2.3929330067009944e-06,
+      "loss": 0.4331,
+      "step": 2647
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5429662869911331,
+      "learning_rate": 2.391055818050426e-06,
+      "loss": 0.4009,
+      "step": 2648
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5805429721405676,
+      "learning_rate": 2.389178690938965e-06,
+      "loss": 0.4191,
+      "step": 2649
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5669011883051497,
+      "learning_rate": 2.3873016264269446e-06,
+      "loss": 0.4112,
+      "step": 2650
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.567460249220951,
+      "learning_rate": 2.3854246255746555e-06,
+      "loss": 0.4315,
+      "step": 2651
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5623656342926924,
+      "learning_rate": 2.3835476894423577e-06,
+      "loss": 0.4059,
+      "step": 2652
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5887115656599653,
+      "learning_rate": 2.3816708190902722e-06,
+      "loss": 0.4292,
+      "step": 2653
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5801851574567665,
+      "learning_rate": 2.3797940155785837e-06,
+      "loss": 0.4377,
+      "step": 2654
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5957884506649995,
+      "learning_rate": 2.3779172799674377e-06,
+      "loss": 0.4291,
+      "step": 2655
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5922308503207093,
+      "learning_rate": 2.376040613316944e-06,
+      "loss": 0.4219,
+      "step": 2656
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5657227192474519,
+      "learning_rate": 2.374164016687173e-06,
+      "loss": 0.4338,
+      "step": 2657
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.6033671454063749,
+      "learning_rate": 2.3722874911381517e-06,
+      "loss": 0.4554,
+      "step": 2658
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5691097763031018,
+      "learning_rate": 2.3704110377298717e-06,
+      "loss": 0.4105,
+      "step": 2659
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5617846566458164,
+      "learning_rate": 2.368534657522281e-06,
+      "loss": 0.4439,
+      "step": 2660
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5551937150400084,
+      "learning_rate": 2.366658351575288e-06,
+      "loss": 0.4266,
+      "step": 2661
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5726467432147245,
+      "learning_rate": 2.364782120948758e-06,
+      "loss": 0.417,
+      "step": 2662
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.6092370186148129,
+      "learning_rate": 2.362905966702515e-06,
+      "loss": 0.4201,
+      "step": 2663
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5647441929287327,
+      "learning_rate": 2.3610298898963373e-06,
+      "loss": 0.4024,
+      "step": 2664
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5612482266868575,
+      "learning_rate": 2.359153891589962e-06,
+      "loss": 0.4272,
+      "step": 2665
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5404969509411939,
+      "learning_rate": 2.35727797284308e-06,
+      "loss": 0.4097,
+      "step": 2666
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5847499166914444,
+      "learning_rate": 2.3554021347153403e-06,
+      "loss": 0.4522,
+      "step": 2667
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5671574895969574,
+      "learning_rate": 2.3535263782663425e-06,
+      "loss": 0.4028,
+      "step": 2668
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.552780149490217,
+      "learning_rate": 2.351650704555643e-06,
+      "loss": 0.4086,
+      "step": 2669
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5803614826769898,
+      "learning_rate": 2.3497751146427494e-06,
+      "loss": 0.4155,
+      "step": 2670
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5652376067469718,
+      "learning_rate": 2.3478996095871228e-06,
+      "loss": 0.4084,
+      "step": 2671
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5845008843251347,
+      "learning_rate": 2.3460241904481778e-06,
+      "loss": 0.4027,
+      "step": 2672
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5799385879129909,
+      "learning_rate": 2.3441488582852774e-06,
+      "loss": 0.4268,
+      "step": 2673
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5738670567740277,
+      "learning_rate": 2.342273614157739e-06,
+      "loss": 0.4464,
+      "step": 2674
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.544552930903809,
+      "learning_rate": 2.3403984591248265e-06,
+      "loss": 0.3964,
+      "step": 2675
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.6124679370289692,
+      "learning_rate": 2.3385233942457574e-06,
+      "loss": 0.4202,
+      "step": 2676
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5610056639464951,
+      "learning_rate": 2.336648420579694e-06,
+      "loss": 0.4094,
+      "step": 2677
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5615215146670192,
+      "learning_rate": 2.3347735391857517e-06,
+      "loss": 0.3789,
+      "step": 2678
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5472440549240616,
+      "learning_rate": 2.33289875112299e-06,
+      "loss": 0.3896,
+      "step": 2679
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5874793154794317,
+      "learning_rate": 2.3310240574504184e-06,
+      "loss": 0.4205,
+      "step": 2680
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5738763492928816,
+      "learning_rate": 2.3291494592269902e-06,
+      "loss": 0.4152,
+      "step": 2681
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5752705880264377,
+      "learning_rate": 2.327274957511607e-06,
+      "loss": 0.4032,
+      "step": 2682
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.6094503057397006,
+      "learning_rate": 2.3254005533631162e-06,
+      "loss": 0.4271,
+      "step": 2683
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.6066369045019371,
+      "learning_rate": 2.3235262478403082e-06,
+      "loss": 0.4241,
+      "step": 2684
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5939355951481937,
+      "learning_rate": 2.3216520420019194e-06,
+      "loss": 0.3942,
+      "step": 2685
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5616286358779028,
+      "learning_rate": 2.3197779369066287e-06,
+      "loss": 0.4052,
+      "step": 2686
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5714073547567622,
+      "learning_rate": 2.3179039336130588e-06,
+      "loss": 0.4156,
+      "step": 2687
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5870213755012056,
+      "learning_rate": 2.3160300331797734e-06,
+      "loss": 0.419,
+      "step": 2688
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5872084599723584,
+      "learning_rate": 2.3141562366652816e-06,
+      "loss": 0.4094,
+      "step": 2689
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5712310144109413,
+      "learning_rate": 2.3122825451280294e-06,
+      "loss": 0.4185,
+      "step": 2690
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5974668557328553,
+      "learning_rate": 2.3104089596264075e-06,
+      "loss": 0.4271,
+      "step": 2691
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5925710616458487,
+      "learning_rate": 2.3085354812187436e-06,
+      "loss": 0.4445,
+      "step": 2692
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.574277995467648,
+      "learning_rate": 2.306662110963307e-06,
+      "loss": 0.4042,
+      "step": 2693
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5763082849196323,
+      "learning_rate": 2.3047888499183036e-06,
+      "loss": 0.417,
+      "step": 2694
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5593667983969396,
+      "learning_rate": 2.30291569914188e-06,
+      "loss": 0.412,
+      "step": 2695
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5768991359362865,
+      "learning_rate": 2.301042659692121e-06,
+      "loss": 0.4041,
+      "step": 2696
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.592377033641057,
+      "learning_rate": 2.299169732627045e-06,
+      "loss": 0.3905,
+      "step": 2697
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5978581423236226,
+      "learning_rate": 2.2972969190046104e-06,
+      "loss": 0.3998,
+      "step": 2698
+    },
+    {
+      "epoch": 1.14,
+      "eval_loss": 0.4642565846443176,
+      "eval_runtime": 6929.6676,
+      "eval_samples_per_second": 41.905,
+      "eval_steps_per_second": 2.095,
+      "step": 2698
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5739162693130145,
+      "learning_rate": 2.2954242198827082e-06,
+      "loss": 0.4196,
+      "step": 2699
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5844044219912439,
+      "learning_rate": 2.2935516363191695e-06,
+      "loss": 0.433,
+      "step": 2700
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5576878153425602,
+      "learning_rate": 2.2916791693717553e-06,
+      "loss": 0.4116,
+      "step": 2701
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5975344250607371,
+      "learning_rate": 2.2898068200981633e-06,
+      "loss": 0.4239,
+      "step": 2702
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5486064916195111,
+      "learning_rate": 2.287934589556024e-06,
+      "loss": 0.3899,
+      "step": 2703
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5954206100798385,
+      "learning_rate": 2.2860624788029013e-06,
+      "loss": 0.3948,
+      "step": 2704
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.582033730964023,
+      "learning_rate": 2.2841904888962903e-06,
+      "loss": 0.4024,
+      "step": 2705
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.61366156641062,
+      "learning_rate": 2.2823186208936205e-06,
+      "loss": 0.4492,
+      "step": 2706
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5837374648931264,
+      "learning_rate": 2.280446875852248e-06,
+      "loss": 0.4394,
+      "step": 2707
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.6159725968971995,
+      "learning_rate": 2.2785752548294637e-06,
+      "loss": 0.4345,
+      "step": 2708
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.561818864484124,
+      "learning_rate": 2.2767037588824877e-06,
+      "loss": 0.4053,
+      "step": 2709
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5693511275506383,
+      "learning_rate": 2.2748323890684664e-06,
+      "loss": 0.4167,
+      "step": 2710
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5676875307416003,
+      "learning_rate": 2.2729611464444797e-06,
+      "loss": 0.4087,
+      "step": 2711
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5865113412123945,
+      "learning_rate": 2.2710900320675314e-06,
+      "loss": 0.4044,
+      "step": 2712
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5877911286647893,
+      "learning_rate": 2.2692190469945557e-06,
+      "loss": 0.4416,
+      "step": 2713
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5405283913293546,
+      "learning_rate": 2.267348192282411e-06,
+      "loss": 0.3994,
+      "step": 2714
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5721166898641502,
+      "learning_rate": 2.2654774689878862e-06,
+      "loss": 0.4269,
+      "step": 2715
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5664221347108425,
+      "learning_rate": 2.2636068781676905e-06,
+      "loss": 0.4172,
+      "step": 2716
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5609241860387546,
+      "learning_rate": 2.261736420878464e-06,
+      "loss": 0.4323,
+      "step": 2717
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5846699602141291,
+      "learning_rate": 2.2598660981767667e-06,
+      "loss": 0.4173,
+      "step": 2718
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5487147727790713,
+      "learning_rate": 2.257995911119086e-06,
+      "loss": 0.4091,
+      "step": 2719
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5642527354564277,
+      "learning_rate": 2.2561258607618296e-06,
+      "loss": 0.4208,
+      "step": 2720
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5596940402216087,
+      "learning_rate": 2.25425594816133e-06,
+      "loss": 0.4177,
+      "step": 2721
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5737740293350555,
+      "learning_rate": 2.2523861743738433e-06,
+      "loss": 0.4079,
+      "step": 2722
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5542429347678349,
+      "learning_rate": 2.2505165404555434e-06,
+      "loss": 0.4177,
+      "step": 2723
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.58808254423625,
+      "learning_rate": 2.248647047462528e-06,
+      "loss": 0.4558,
+      "step": 2724
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.6083740436546075,
+      "learning_rate": 2.246777696450813e-06,
+      "loss": 0.4367,
+      "step": 2725
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.577052191765705,
+      "learning_rate": 2.244908488476337e-06,
+      "loss": 0.4133,
+      "step": 2726
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.59097397240932,
+      "learning_rate": 2.2430394245949553e-06,
+      "loss": 0.4327,
+      "step": 2727
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5524670352667417,
+      "learning_rate": 2.2411705058624437e-06,
+      "loss": 0.4364,
+      "step": 2728
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5889869749199854,
+      "learning_rate": 2.2393017333344935e-06,
+      "loss": 0.4389,
+      "step": 2729
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5781616268517972,
+      "learning_rate": 2.2374331080667168e-06,
+      "loss": 0.4146,
+      "step": 2730
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5516338673454813,
+      "learning_rate": 2.235564631114639e-06,
+      "loss": 0.42,
+      "step": 2731
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5737980509227198,
+      "learning_rate": 2.2336963035337037e-06,
+      "loss": 0.4049,
+      "step": 2732
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.6099792863000669,
+      "learning_rate": 2.2318281263792714e-06,
+      "loss": 0.4329,
+      "step": 2733
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.6215436378061963,
+      "learning_rate": 2.229960100706614e-06,
+      "loss": 0.4064,
+      "step": 2734
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.6070304940895107,
+      "learning_rate": 2.2280922275709216e-06,
+      "loss": 0.4172,
+      "step": 2735
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.6137903976399841,
+      "learning_rate": 2.2262245080272947e-06,
+      "loss": 0.4191,
+      "step": 2736
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.589617423215821,
+      "learning_rate": 2.2243569431307506e-06,
+      "loss": 0.4236,
+      "step": 2737
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5714692395337472,
+      "learning_rate": 2.2224895339362153e-06,
+      "loss": 0.4174,
+      "step": 2738
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.566616099477279,
+      "learning_rate": 2.2206222814985316e-06,
+      "loss": 0.4172,
+      "step": 2739
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5918370633556324,
+      "learning_rate": 2.2187551868724487e-06,
+      "loss": 0.4022,
+      "step": 2740
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5621764463135852,
+      "learning_rate": 2.2168882511126306e-06,
+      "loss": 0.4033,
+      "step": 2741
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5602317292982957,
+      "learning_rate": 2.215021475273649e-06,
+      "loss": 0.3948,
+      "step": 2742
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5821498763067597,
+      "learning_rate": 2.213154860409987e-06,
+      "loss": 0.4342,
+      "step": 2743
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.56289957110064,
+      "learning_rate": 2.211288407576035e-06,
+      "loss": 0.4224,
+      "step": 2744
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5891254592215949,
+      "learning_rate": 2.209422117826094e-06,
+      "loss": 0.4217,
+      "step": 2745
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5729813741965974,
+      "learning_rate": 2.207555992214372e-06,
+      "loss": 0.4369,
+      "step": 2746
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5673765067747608,
+      "learning_rate": 2.2056900317949835e-06,
+      "loss": 0.4242,
+      "step": 2747
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5690722337898687,
+      "learning_rate": 2.203824237621951e-06,
+      "loss": 0.4284,
+      "step": 2748
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5690375147524728,
+      "learning_rate": 2.2019586107492005e-06,
+      "loss": 0.4238,
+      "step": 2749
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5507462792620564,
+      "learning_rate": 2.200093152230568e-06,
+      "loss": 0.4135,
+      "step": 2750
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5719332339424014,
+      "learning_rate": 2.1982278631197895e-06,
+      "loss": 0.4174,
+      "step": 2751
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5874278219304234,
+      "learning_rate": 2.1963627444705097e-06,
+      "loss": 0.4053,
+      "step": 2752
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.568889174310937,
+      "learning_rate": 2.1944977973362728e-06,
+      "loss": 0.4123,
+      "step": 2753
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.6038767993949933,
+      "learning_rate": 2.19263302277053e-06,
+      "loss": 0.4349,
+      "step": 2754
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5872897799798638,
+      "learning_rate": 2.190768421826631e-06,
+      "loss": 0.4083,
+      "step": 2755
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5905095099111151,
+      "learning_rate": 2.1889039955578327e-06,
+      "loss": 0.4146,
+      "step": 2756
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5527103292721002,
+      "learning_rate": 2.1870397450172876e-06,
+      "loss": 0.4191,
+      "step": 2757
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5818092591622248,
+      "learning_rate": 2.1851756712580526e-06,
+      "loss": 0.3905,
+      "step": 2758
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5560986512545047,
+      "learning_rate": 2.1833117753330847e-06,
+      "loss": 0.4212,
+      "step": 2759
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.583827443059487,
+      "learning_rate": 2.1814480582952376e-06,
+      "loss": 0.439,
+      "step": 2760
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.6047213755513012,
+      "learning_rate": 2.1795845211972684e-06,
+      "loss": 0.4098,
+      "step": 2761
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5719149813855363,
+      "learning_rate": 2.1777211650918276e-06,
+      "loss": 0.4436,
+      "step": 2762
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5636987195875225,
+      "learning_rate": 2.1758579910314677e-06,
+      "loss": 0.4212,
+      "step": 2763
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5821057683244233,
+      "learning_rate": 2.1739950000686354e-06,
+      "loss": 0.4157,
+      "step": 2764
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5763014032864897,
+      "learning_rate": 2.1721321932556753e-06,
+      "loss": 0.4167,
+      "step": 2765
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5741607793695896,
+      "learning_rate": 2.1702695716448276e-06,
+      "loss": 0.4118,
+      "step": 2766
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.583977923198581,
+      "learning_rate": 2.16840713628823e-06,
+      "loss": 0.4362,
+      "step": 2767
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.567095712657335,
+      "learning_rate": 2.16654488823791e-06,
+      "loss": 0.3994,
+      "step": 2768
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5882945712372565,
+      "learning_rate": 2.164682828545795e-06,
+      "loss": 0.415,
+      "step": 2769
+    },
+    {
+      "epoch": 1.17,
+      "eval_loss": 0.4636525809764862,
+      "eval_runtime": 6931.4457,
+      "eval_samples_per_second": 41.894,
+      "eval_steps_per_second": 2.095,
+      "step": 2769
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5800810184022894,
+      "learning_rate": 2.1628209582637024e-06,
+      "loss": 0.423,
+      "step": 2770
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5877561785586146,
+      "learning_rate": 2.160959278443342e-06,
+      "loss": 0.4191,
+      "step": 2771
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5450021019526348,
+      "learning_rate": 2.1590977901363215e-06,
+      "loss": 0.4121,
+      "step": 2772
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.586036868971513,
+      "learning_rate": 2.157236494394133e-06,
+      "loss": 0.4231,
+      "step": 2773
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.6082884207803223,
+      "learning_rate": 2.155375392268165e-06,
+      "loss": 0.4347,
+      "step": 2774
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5818956219426625,
+      "learning_rate": 2.1535144848096943e-06,
+      "loss": 0.4267,
+      "step": 2775
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5882181261304475,
+      "learning_rate": 2.1516537730698895e-06,
+      "loss": 0.4122,
+      "step": 2776
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5855723087856304,
+      "learning_rate": 2.1497932580998055e-06,
+      "loss": 0.4196,
+      "step": 2777
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5621462555991732,
+      "learning_rate": 2.147932940950391e-06,
+      "loss": 0.4313,
+      "step": 2778
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5619630139053288,
+      "learning_rate": 2.1460728226724768e-06,
+      "loss": 0.4171,
+      "step": 2779
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5803428278169429,
+      "learning_rate": 2.1442129043167877e-06,
+      "loss": 0.4137,
+      "step": 2780
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.6003743965203625,
+      "learning_rate": 2.1423531869339307e-06,
+      "loss": 0.4094,
+      "step": 2781
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5838530366206433,
+      "learning_rate": 2.140493671574402e-06,
+      "loss": 0.4256,
+      "step": 2782
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5750052865571543,
+      "learning_rate": 2.138634359288581e-06,
+      "loss": 0.4182,
+      "step": 2783
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.555105523968193,
+      "learning_rate": 2.1367752511267366e-06,
+      "loss": 0.4174,
+      "step": 2784
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.569042426766486,
+      "learning_rate": 2.134916348139019e-06,
+      "loss": 0.4137,
+      "step": 2785
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5907654743125312,
+      "learning_rate": 2.133057651375463e-06,
+      "loss": 0.4148,
+      "step": 2786
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5717943994371465,
+      "learning_rate": 2.1311991618859883e-06,
+      "loss": 0.4307,
+      "step": 2787
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5562054796755491,
+      "learning_rate": 2.129340880720395e-06,
+      "loss": 0.412,
+      "step": 2788
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5745208858282651,
+      "learning_rate": 2.1274828089283696e-06,
+      "loss": 0.4315,
+      "step": 2789
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.6261475696552913,
+      "learning_rate": 2.125624947559475e-06,
+      "loss": 0.4141,
+      "step": 2790
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.587239695175737,
+      "learning_rate": 2.123767297663161e-06,
+      "loss": 0.4384,
+      "step": 2791
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5583403908079667,
+      "learning_rate": 2.1219098602887524e-06,
+      "loss": 0.4215,
+      "step": 2792
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5697359070123751,
+      "learning_rate": 2.1200526364854583e-06,
+      "loss": 0.4195,
+      "step": 2793
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5746434077589057,
+      "learning_rate": 2.118195627302364e-06,
+      "loss": 0.4397,
+      "step": 2794
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5770878978662568,
+      "learning_rate": 2.116338833788437e-06,
+      "loss": 0.4142,
+      "step": 2795
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.615557087883441,
+      "learning_rate": 2.114482256992519e-06,
+      "loss": 0.4124,
+      "step": 2796
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5665043322437245,
+      "learning_rate": 2.112625897963333e-06,
+      "loss": 0.423,
+      "step": 2797
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5598727728785423,
+      "learning_rate": 2.1107697577494764e-06,
+      "loss": 0.4014,
+      "step": 2798
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.567119767667523,
+      "learning_rate": 2.1089138373994226e-06,
+      "loss": 0.4072,
+      "step": 2799
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5672415415140308,
+      "learning_rate": 2.1070581379615253e-06,
+      "loss": 0.4336,
+      "step": 2800
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5503253498723211,
+      "learning_rate": 2.1052026604840066e-06,
+      "loss": 0.4181,
+      "step": 2801
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5837370512357433,
+      "learning_rate": 2.10334740601497e-06,
+      "loss": 0.4014,
+      "step": 2802
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5485492339607526,
+      "learning_rate": 2.101492375602387e-06,
+      "loss": 0.3878,
+      "step": 2803
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5901125098812506,
+      "learning_rate": 2.099637570294108e-06,
+      "loss": 0.4315,
+      "step": 2804
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5790878839442004,
+      "learning_rate": 2.0977829911378507e-06,
+      "loss": 0.4082,
+      "step": 2805
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5486030886397184,
+      "learning_rate": 2.0959286391812116e-06,
+      "loss": 0.4131,
+      "step": 2806
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5731467050022245,
+      "learning_rate": 2.0940745154716516e-06,
+      "loss": 0.4213,
+      "step": 2807
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.563249940690925,
+      "learning_rate": 2.0922206210565088e-06,
+      "loss": 0.4275,
+      "step": 2808
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5588267746939021,
+      "learning_rate": 2.090366956982988e-06,
+      "loss": 0.4053,
+      "step": 2809
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.6276548597898437,
+      "learning_rate": 2.088513524298165e-06,
+      "loss": 0.4044,
+      "step": 2810
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5654177414506385,
+      "learning_rate": 2.086660324048987e-06,
+      "loss": 0.3936,
+      "step": 2811
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5912299181232848,
+      "learning_rate": 2.084807357282266e-06,
+      "loss": 0.4449,
+      "step": 2812
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5934179796428065,
+      "learning_rate": 2.0829546250446846e-06,
+      "loss": 0.4316,
+      "step": 2813
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.660477044543163,
+      "learning_rate": 2.0811021283827928e-06,
+      "loss": 0.4134,
+      "step": 2814
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5785577080236974,
+      "learning_rate": 2.0792498683430072e-06,
+      "loss": 0.4289,
+      "step": 2815
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5359940863049542,
+      "learning_rate": 2.077397845971609e-06,
+      "loss": 0.3887,
+      "step": 2816
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5579305558402582,
+      "learning_rate": 2.07554606231475e-06,
+      "loss": 0.4046,
+      "step": 2817
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5768033221733195,
+      "learning_rate": 2.0736945184184406e-06,
+      "loss": 0.4306,
+      "step": 2818
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5574558414654748,
+      "learning_rate": 2.0718432153285615e-06,
+      "loss": 0.4266,
+      "step": 2819
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5523510875719263,
+      "learning_rate": 2.0699921540908542e-06,
+      "loss": 0.4209,
+      "step": 2820
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.586454598962645,
+      "learning_rate": 2.068141335750925e-06,
+      "loss": 0.4038,
+      "step": 2821
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5767355936333549,
+      "learning_rate": 2.0662907613542405e-06,
+      "loss": 0.4219,
+      "step": 2822
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5748539680981339,
+      "learning_rate": 2.064440431946133e-06,
+      "loss": 0.382,
+      "step": 2823
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5642219467594125,
+      "learning_rate": 2.062590348571796e-06,
+      "loss": 0.4051,
+      "step": 2824
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5825964555184913,
+      "learning_rate": 2.0607405122762806e-06,
+      "loss": 0.4096,
+      "step": 2825
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5629400713760981,
+      "learning_rate": 2.058890924104502e-06,
+      "loss": 0.4,
+      "step": 2826
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5463214353524036,
+      "learning_rate": 2.057041585101232e-06,
+      "loss": 0.3901,
+      "step": 2827
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5728170268667269,
+      "learning_rate": 2.0551924963111064e-06,
+      "loss": 0.4252,
+      "step": 2828
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.5543782845949665,
+      "learning_rate": 2.053343658778613e-06,
+      "loss": 0.4214,
+      "step": 2829
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.58082955878978,
+      "learning_rate": 2.0514950735481053e-06,
+      "loss": 0.4137,
+      "step": 2830
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5531229406229621,
+      "learning_rate": 2.049646741663788e-06,
+      "loss": 0.4131,
+      "step": 2831
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5541337155553405,
+      "learning_rate": 2.0477986641697263e-06,
+      "loss": 0.4068,
+      "step": 2832
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5512293318719139,
+      "learning_rate": 2.0459508421098383e-06,
+      "loss": 0.4193,
+      "step": 2833
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.6481077588847556,
+      "learning_rate": 2.0441032765279036e-06,
+      "loss": 0.4195,
+      "step": 2834
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5553462347610719,
+      "learning_rate": 2.0422559684675498e-06,
+      "loss": 0.4131,
+      "step": 2835
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5854214572008192,
+      "learning_rate": 2.040408918972264e-06,
+      "loss": 0.4223,
+      "step": 2836
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5524192472655216,
+      "learning_rate": 2.038562129085387e-06,
+      "loss": 0.3845,
+      "step": 2837
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.56829946589284,
+      "learning_rate": 2.0367155998501092e-06,
+      "loss": 0.4013,
+      "step": 2838
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5878475779332522,
+      "learning_rate": 2.03486933230948e-06,
+      "loss": 0.4206,
+      "step": 2839
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5637907040986664,
+      "learning_rate": 2.033023327506393e-06,
+      "loss": 0.4052,
+      "step": 2840
+    },
+    {
+      "epoch": 1.2,
+      "eval_loss": 0.4627860486507416,
+      "eval_runtime": 6936.5234,
+      "eval_samples_per_second": 41.864,
+      "eval_steps_per_second": 2.093,
+      "step": 2840
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5782227933222694,
+      "learning_rate": 2.0311775864836007e-06,
+      "loss": 0.4184,
+      "step": 2841
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5726678150106956,
+      "learning_rate": 2.0293321102837023e-06,
+      "loss": 0.4379,
+      "step": 2842
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5653275322857352,
+      "learning_rate": 2.0274868999491496e-06,
+      "loss": 0.411,
+      "step": 2843
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.6143518059743222,
+      "learning_rate": 2.0256419565222423e-06,
+      "loss": 0.4202,
+      "step": 2844
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5533020572133284,
+      "learning_rate": 2.023797281045132e-06,
+      "loss": 0.4015,
+      "step": 2845
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5551626428007986,
+      "learning_rate": 2.0219528745598145e-06,
+      "loss": 0.4148,
+      "step": 2846
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5889239778826716,
+      "learning_rate": 2.020108738108139e-06,
+      "loss": 0.4052,
+      "step": 2847
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.6024562268514936,
+      "learning_rate": 2.0182648727317986e-06,
+      "loss": 0.4287,
+      "step": 2848
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5746780507322024,
+      "learning_rate": 2.0164212794723336e-06,
+      "loss": 0.4165,
+      "step": 2849
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5908540121904199,
+      "learning_rate": 2.014577959371134e-06,
+      "loss": 0.4391,
+      "step": 2850
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5820354589271028,
+      "learning_rate": 2.012734913469429e-06,
+      "loss": 0.4151,
+      "step": 2851
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5473344454525257,
+      "learning_rate": 2.0108921428083e-06,
+      "loss": 0.4279,
+      "step": 2852
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5832936596852468,
+      "learning_rate": 2.009049648428668e-06,
+      "loss": 0.4288,
+      "step": 2853
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.6084388529390307,
+      "learning_rate": 2.0072074313713e-06,
+      "loss": 0.4461,
+      "step": 2854
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5740745870244656,
+      "learning_rate": 2.0053654926768044e-06,
+      "loss": 0.4021,
+      "step": 2855
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5515506740065009,
+      "learning_rate": 2.003523833385637e-06,
+      "loss": 0.3928,
+      "step": 2856
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5624075725007869,
+      "learning_rate": 2.0016824545380895e-06,
+      "loss": 0.4191,
+      "step": 2857
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5863935111984939,
+      "learning_rate": 1.9998413571743006e-06,
+      "loss": 0.4224,
+      "step": 2858
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5721820837476251,
+      "learning_rate": 1.9980005423342462e-06,
+      "loss": 0.4364,
+      "step": 2859
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5729299550286993,
+      "learning_rate": 1.996160011057746e-06,
+      "loss": 0.4112,
+      "step": 2860
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5933829400685822,
+      "learning_rate": 1.9943197643844554e-06,
+      "loss": 0.399,
+      "step": 2861
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.601305825995989,
+      "learning_rate": 1.992479803353872e-06,
+      "loss": 0.4265,
+      "step": 2862
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5692294662214427,
+      "learning_rate": 1.9906401290053323e-06,
+      "loss": 0.4071,
+      "step": 2863
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.570531202757177,
+      "learning_rate": 1.9888007423780095e-06,
+      "loss": 0.4161,
+      "step": 2864
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5765127200524025,
+      "learning_rate": 1.9869616445109146e-06,
+      "loss": 0.42,
+      "step": 2865
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5652493468336516,
+      "learning_rate": 1.985122836442895e-06,
+      "loss": 0.4447,
+      "step": 2866
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5345921455849048,
+      "learning_rate": 1.9832843192126367e-06,
+      "loss": 0.4114,
+      "step": 2867
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.553833707765255,
+      "learning_rate": 1.9814460938586572e-06,
+      "loss": 0.4192,
+      "step": 2868
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5659115457713205,
+      "learning_rate": 1.9796081614193143e-06,
+      "loss": 0.4195,
+      "step": 2869
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5667385460753466,
+      "learning_rate": 1.9777705229327954e-06,
+      "loss": 0.4177,
+      "step": 2870
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5807688119962365,
+      "learning_rate": 1.9759331794371255e-06,
+      "loss": 0.4015,
+      "step": 2871
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5637812729284788,
+      "learning_rate": 1.97409613197016e-06,
+      "loss": 0.4288,
+      "step": 2872
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5695007837691306,
+      "learning_rate": 1.972259381569592e-06,
+      "loss": 0.4029,
+      "step": 2873
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5977467234875765,
+      "learning_rate": 1.9704229292729393e-06,
+      "loss": 0.4189,
+      "step": 2874
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.569978022020393,
+      "learning_rate": 1.9685867761175584e-06,
+      "loss": 0.3981,
+      "step": 2875
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5622730363649805,
+      "learning_rate": 1.9667509231406332e-06,
+      "loss": 0.4072,
+      "step": 2876
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5772397599288341,
+      "learning_rate": 1.964915371379178e-06,
+      "loss": 0.4224,
+      "step": 2877
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5550624772060395,
+      "learning_rate": 1.9630801218700397e-06,
+      "loss": 0.4168,
+      "step": 2878
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.593384835711817,
+      "learning_rate": 1.961245175649889e-06,
+      "loss": 0.4082,
+      "step": 2879
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5648138827206016,
+      "learning_rate": 1.959410533755232e-06,
+      "loss": 0.4386,
+      "step": 2880
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.590345590782358,
+      "learning_rate": 1.9575761972223983e-06,
+      "loss": 0.4318,
+      "step": 2881
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5814485952827325,
+      "learning_rate": 1.955742167087547e-06,
+      "loss": 0.4288,
+      "step": 2882
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5833916546222742,
+      "learning_rate": 1.953908444386662e-06,
+      "loss": 0.4093,
+      "step": 2883
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.599723766340312,
+      "learning_rate": 1.9520750301555574e-06,
+      "loss": 0.4177,
+      "step": 2884
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.6655917345901515,
+      "learning_rate": 1.9502419254298674e-06,
+      "loss": 0.4105,
+      "step": 2885
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5717968178433537,
+      "learning_rate": 1.9484091312450577e-06,
+      "loss": 0.4214,
+      "step": 2886
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5844960834468691,
+      "learning_rate": 1.9465766486364145e-06,
+      "loss": 0.4249,
+      "step": 2887
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5832680992228242,
+      "learning_rate": 1.944744478639048e-06,
+      "loss": 0.4095,
+      "step": 2888
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5814549183663302,
+      "learning_rate": 1.9429126222878954e-06,
+      "loss": 0.4303,
+      "step": 2889
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5589845040306947,
+      "learning_rate": 1.9410810806177105e-06,
+      "loss": 0.4008,
+      "step": 2890
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5823353970607823,
+      "learning_rate": 1.9392498546630767e-06,
+      "loss": 0.4206,
+      "step": 2891
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5773798607020071,
+      "learning_rate": 1.937418945458393e-06,
+      "loss": 0.4107,
+      "step": 2892
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5758173233706595,
+      "learning_rate": 1.935588354037883e-06,
+      "loss": 0.4299,
+      "step": 2893
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5823417001041828,
+      "learning_rate": 1.9337580814355887e-06,
+      "loss": 0.4109,
+      "step": 2894
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5969616099880778,
+      "learning_rate": 1.931928128685375e-06,
+      "loss": 0.4005,
+      "step": 2895
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5732522814857013,
+      "learning_rate": 1.9300984968209215e-06,
+      "loss": 0.4251,
+      "step": 2896
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.6361313903565526,
+      "learning_rate": 1.928269186875731e-06,
+      "loss": 0.4294,
+      "step": 2897
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5936983346727575,
+      "learning_rate": 1.9264401998831213e-06,
+      "loss": 0.4503,
+      "step": 2898
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.566199487669491,
+      "learning_rate": 1.9246115368762307e-06,
+      "loss": 0.4267,
+      "step": 2899
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5658880403316346,
+      "learning_rate": 1.922783198888011e-06,
+      "loss": 0.4211,
+      "step": 2900
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5861694408434711,
+      "learning_rate": 1.9209551869512326e-06,
+      "loss": 0.4142,
+      "step": 2901
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5861893981468904,
+      "learning_rate": 1.919127502098483e-06,
+      "loss": 0.4285,
+      "step": 2902
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.6057636807953072,
+      "learning_rate": 1.9173001453621615e-06,
+      "loss": 0.4334,
+      "step": 2903
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5825991271880557,
+      "learning_rate": 1.9154731177744858e-06,
+      "loss": 0.4135,
+      "step": 2904
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5685728541574715,
+      "learning_rate": 1.913646420367483e-06,
+      "loss": 0.4239,
+      "step": 2905
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5435808297310615,
+      "learning_rate": 1.911820054173e-06,
+      "loss": 0.4092,
+      "step": 2906
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5768278588163496,
+      "learning_rate": 1.9099940202226895e-06,
+      "loss": 0.4192,
+      "step": 2907
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5528574586286196,
+      "learning_rate": 1.908168319548023e-06,
+      "loss": 0.4395,
+      "step": 2908
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5460290788495442,
+      "learning_rate": 1.9063429531802788e-06,
+      "loss": 0.4089,
+      "step": 2909
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5836425311067438,
+      "learning_rate": 1.9045179221505497e-06,
+      "loss": 0.4096,
+      "step": 2910
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5368200883277836,
+      "learning_rate": 1.902693227489737e-06,
+      "loss": 0.4098,
+      "step": 2911
+    },
+    {
+      "epoch": 1.23,
+      "eval_loss": 0.46161049604415894,
+      "eval_runtime": 6937.863,
+      "eval_samples_per_second": 41.856,
+      "eval_steps_per_second": 2.093,
+      "step": 2911
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5737532658966875,
+      "learning_rate": 1.9008688702285532e-06,
+      "loss": 0.4076,
+      "step": 2912
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5929933734472119,
+      "learning_rate": 1.899044851397519e-06,
+      "loss": 0.386,
+      "step": 2913
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5904060193816585,
+      "learning_rate": 1.8972211720269657e-06,
+      "loss": 0.4006,
+      "step": 2914
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5744303339968841,
+      "learning_rate": 1.8953978331470322e-06,
+      "loss": 0.3955,
+      "step": 2915
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5805107950321803,
+      "learning_rate": 1.8935748357876626e-06,
+      "loss": 0.3978,
+      "step": 2916
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.6035250400166989,
+      "learning_rate": 1.8917521809786136e-06,
+      "loss": 0.4243,
+      "step": 2917
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5889359436095863,
+      "learning_rate": 1.8899298697494413e-06,
+      "loss": 0.4332,
+      "step": 2918
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.638150416673843,
+      "learning_rate": 1.8881079031295147e-06,
+      "loss": 0.4289,
+      "step": 2919
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5711856241860922,
+      "learning_rate": 1.8862862821480023e-06,
+      "loss": 0.4198,
+      "step": 2920
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5695997739109594,
+      "learning_rate": 1.8844650078338818e-06,
+      "loss": 0.4008,
+      "step": 2921
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5593680674464713,
+      "learning_rate": 1.8826440812159321e-06,
+      "loss": 0.3921,
+      "step": 2922
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5684210443299574,
+      "learning_rate": 1.8808235033227378e-06,
+      "loss": 0.4109,
+      "step": 2923
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.791352852330744,
+      "learning_rate": 1.8790032751826839e-06,
+      "loss": 0.4426,
+      "step": 2924
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.590981142715526,
+      "learning_rate": 1.8771833978239615e-06,
+      "loss": 0.4044,
+      "step": 2925
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5929104028257183,
+      "learning_rate": 1.8753638722745601e-06,
+      "loss": 0.4251,
+      "step": 2926
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.6088394356034341,
+      "learning_rate": 1.8735446995622719e-06,
+      "loss": 0.422,
+      "step": 2927
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5812034504925151,
+      "learning_rate": 1.8717258807146918e-06,
+      "loss": 0.4417,
+      "step": 2928
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5634954353609875,
+      "learning_rate": 1.8699074167592097e-06,
+      "loss": 0.4221,
+      "step": 2929
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5805870682157763,
+      "learning_rate": 1.8680893087230207e-06,
+      "loss": 0.4148,
+      "step": 2930
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5978992039621629,
+      "learning_rate": 1.866271557633115e-06,
+      "loss": 0.427,
+      "step": 2931
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5433676698516975,
+      "learning_rate": 1.8644541645162834e-06,
+      "loss": 0.4378,
+      "step": 2932
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5654081584633399,
+      "learning_rate": 1.862637130399112e-06,
+      "loss": 0.4255,
+      "step": 2933
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5745394540754084,
+      "learning_rate": 1.8608204563079874e-06,
+      "loss": 0.4138,
+      "step": 2934
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5651328725524883,
+      "learning_rate": 1.8590041432690895e-06,
+      "loss": 0.4289,
+      "step": 2935
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5543604900455075,
+      "learning_rate": 1.8571881923083976e-06,
+      "loss": 0.4098,
+      "step": 2936
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5541075458786751,
+      "learning_rate": 1.8553726044516835e-06,
+      "loss": 0.405,
+      "step": 2937
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5662395052454143,
+      "learning_rate": 1.8535573807245155e-06,
+      "loss": 0.398,
+      "step": 2938
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5415458339050451,
+      "learning_rate": 1.8517425221522555e-06,
+      "loss": 0.3944,
+      "step": 2939
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5975828808599232,
+      "learning_rate": 1.8499280297600594e-06,
+      "loss": 0.4102,
+      "step": 2940
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5586543614587621,
+      "learning_rate": 1.848113904572878e-06,
+      "loss": 0.4071,
+      "step": 2941
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5462110270977989,
+      "learning_rate": 1.8463001476154508e-06,
+      "loss": 0.4151,
+      "step": 2942
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5750571044180495,
+      "learning_rate": 1.844486759912313e-06,
+      "loss": 0.4111,
+      "step": 2943
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5670618407119518,
+      "learning_rate": 1.8426737424877883e-06,
+      "loss": 0.4188,
+      "step": 2944
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5916977571948979,
+      "learning_rate": 1.840861096365995e-06,
+      "loss": 0.4323,
+      "step": 2945
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5723973722291331,
+      "learning_rate": 1.8390488225708364e-06,
+      "loss": 0.4216,
+      "step": 2946
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5788397104566307,
+      "learning_rate": 1.8372369221260106e-06,
+      "loss": 0.4185,
+      "step": 2947
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.605618630484289,
+      "learning_rate": 1.8354253960550017e-06,
+      "loss": 0.4073,
+      "step": 2948
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5986751285211731,
+      "learning_rate": 1.833614245381084e-06,
+      "loss": 0.4254,
+      "step": 2949
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5555679218850047,
+      "learning_rate": 1.8318034711273181e-06,
+      "loss": 0.4029,
+      "step": 2950
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5541685467504466,
+      "learning_rate": 1.8299930743165537e-06,
+      "loss": 0.3933,
+      "step": 2951
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5797747508458493,
+      "learning_rate": 1.8281830559714248e-06,
+      "loss": 0.4167,
+      "step": 2952
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5724155369928727,
+      "learning_rate": 1.8263734171143552e-06,
+      "loss": 0.4065,
+      "step": 2953
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5842672183423349,
+      "learning_rate": 1.8245641587675523e-06,
+      "loss": 0.4193,
+      "step": 2954
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5670445430376003,
+      "learning_rate": 1.822755281953007e-06,
+      "loss": 0.4362,
+      "step": 2955
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5785669095113156,
+      "learning_rate": 1.8209467876924992e-06,
+      "loss": 0.4115,
+      "step": 2956
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5942980368738244,
+      "learning_rate": 1.8191386770075863e-06,
+      "loss": 0.4273,
+      "step": 2957
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5928392121690472,
+      "learning_rate": 1.8173309509196158e-06,
+      "loss": 0.4224,
+      "step": 2958
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5922420607132926,
+      "learning_rate": 1.8155236104497128e-06,
+      "loss": 0.4394,
+      "step": 2959
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.6223806379306809,
+      "learning_rate": 1.813716656618788e-06,
+      "loss": 0.434,
+      "step": 2960
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5845357998232491,
+      "learning_rate": 1.8119100904475306e-06,
+      "loss": 0.441,
+      "step": 2961
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5576654968046494,
+      "learning_rate": 1.8101039129564142e-06,
+      "loss": 0.4018,
+      "step": 2962
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5734931825622845,
+      "learning_rate": 1.8082981251656887e-06,
+      "loss": 0.4122,
+      "step": 2963
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5992264311549484,
+      "learning_rate": 1.8064927280953893e-06,
+      "loss": 0.4426,
+      "step": 2964
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5768107642343978,
+      "learning_rate": 1.8046877227653248e-06,
+      "loss": 0.388,
+      "step": 2965
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5821336521989418,
+      "learning_rate": 1.8028831101950866e-06,
+      "loss": 0.4245,
+      "step": 2966
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5390570291656819,
+      "learning_rate": 1.8010788914040444e-06,
+      "loss": 0.3867,
+      "step": 2967
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5704847089539564,
+      "learning_rate": 1.7992750674113414e-06,
+      "loss": 0.4278,
+      "step": 2968
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.6111690436052009,
+      "learning_rate": 1.7974716392359026e-06,
+      "loss": 0.4479,
+      "step": 2969
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.5802046152902072,
+      "learning_rate": 1.7956686078964257e-06,
+      "loss": 0.4298,
+      "step": 2970
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5455052953532722,
+      "learning_rate": 1.793865974411388e-06,
+      "loss": 0.4181,
+      "step": 2971
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5686949830786789,
+      "learning_rate": 1.7920637397990373e-06,
+      "loss": 0.4203,
+      "step": 2972
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5957176939693254,
+      "learning_rate": 1.7902619050774006e-06,
+      "loss": 0.4026,
+      "step": 2973
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.6022161318571446,
+      "learning_rate": 1.788460471264276e-06,
+      "loss": 0.3995,
+      "step": 2974
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5831579158285504,
+      "learning_rate": 1.7866594393772375e-06,
+      "loss": 0.4035,
+      "step": 2975
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5801852393556498,
+      "learning_rate": 1.7848588104336293e-06,
+      "loss": 0.4123,
+      "step": 2976
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5961588266575963,
+      "learning_rate": 1.783058585450571e-06,
+      "loss": 0.4317,
+      "step": 2977
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.573944779319003,
+      "learning_rate": 1.781258765444951e-06,
+      "loss": 0.4344,
+      "step": 2978
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5879109677820142,
+      "learning_rate": 1.7794593514334313e-06,
+      "loss": 0.4182,
+      "step": 2979
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5888697412742766,
+      "learning_rate": 1.7776603444324445e-06,
+      "loss": 0.4119,
+      "step": 2980
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.570679582164104,
+      "learning_rate": 1.775861745458191e-06,
+      "loss": 0.4411,
+      "step": 2981
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5775235217718448,
+      "learning_rate": 1.774063555526644e-06,
+      "loss": 0.4003,
+      "step": 2982
+    },
+    {
+      "epoch": 1.26,
+      "eval_loss": 0.4607752859592438,
+      "eval_runtime": 6938.4518,
+      "eval_samples_per_second": 41.852,
+      "eval_steps_per_second": 2.093,
+      "step": 2982
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5772173427303933,
+      "learning_rate": 1.7722657756535422e-06,
+      "loss": 0.4292,
+      "step": 2983
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5620226835290322,
+      "learning_rate": 1.7704684068543953e-06,
+      "loss": 0.396,
+      "step": 2984
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.6017908973225675,
+      "learning_rate": 1.7686714501444791e-06,
+      "loss": 0.4228,
+      "step": 2985
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5842773295597439,
+      "learning_rate": 1.7668749065388385e-06,
+      "loss": 0.4235,
+      "step": 2986
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5881893778638159,
+      "learning_rate": 1.7650787770522831e-06,
+      "loss": 0.3885,
+      "step": 2987
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5921874088208304,
+      "learning_rate": 1.76328306269939e-06,
+      "loss": 0.4118,
+      "step": 2988
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5860478876629458,
+      "learning_rate": 1.7614877644945002e-06,
+      "loss": 0.4259,
+      "step": 2989
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5566234951598134,
+      "learning_rate": 1.759692883451721e-06,
+      "loss": 0.4062,
+      "step": 2990
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5886058120181612,
+      "learning_rate": 1.757898420584925e-06,
+      "loss": 0.4157,
+      "step": 2991
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5835204891594994,
+      "learning_rate": 1.756104376907746e-06,
+      "loss": 0.4287,
+      "step": 2992
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5820795493170282,
+      "learning_rate": 1.7543107534335828e-06,
+      "loss": 0.4224,
+      "step": 2993
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.6013433453081255,
+      "learning_rate": 1.752517551175596e-06,
+      "loss": 0.4354,
+      "step": 2994
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.6033893634449149,
+      "learning_rate": 1.750724771146709e-06,
+      "loss": 0.4286,
+      "step": 2995
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.54509644376269,
+      "learning_rate": 1.748932414359605e-06,
+      "loss": 0.4082,
+      "step": 2996
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5826151129215978,
+      "learning_rate": 1.7471404818267319e-06,
+      "loss": 0.4051,
+      "step": 2997
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5880061063869555,
+      "learning_rate": 1.745348974560293e-06,
+      "loss": 0.4386,
+      "step": 2998
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5656468304808548,
+      "learning_rate": 1.743557893572256e-06,
+      "loss": 0.4146,
+      "step": 2999
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5689192538618583,
+      "learning_rate": 1.741767239874344e-06,
+      "loss": 0.4103,
+      "step": 3000
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5840118133437147,
+      "learning_rate": 1.7399770144780414e-06,
+      "loss": 0.4199,
+      "step": 3001
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5550874833815063,
+      "learning_rate": 1.7381872183945885e-06,
+      "loss": 0.3935,
+      "step": 3002
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5731500465902329,
+      "learning_rate": 1.7363978526349857e-06,
+      "loss": 0.4307,
+      "step": 3003
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5579055837980196,
+      "learning_rate": 1.734608918209989e-06,
+      "loss": 0.4152,
+      "step": 3004
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5619015288206511,
+      "learning_rate": 1.7328204161301084e-06,
+      "loss": 0.4257,
+      "step": 3005
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5685518518480854,
+      "learning_rate": 1.7310323474056154e-06,
+      "loss": 0.4292,
+      "step": 3006
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.572223039710208,
+      "learning_rate": 1.7292447130465296e-06,
+      "loss": 0.4027,
+      "step": 3007
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5923422187875147,
+      "learning_rate": 1.7274575140626318e-06,
+      "loss": 0.4185,
+      "step": 3008
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.578248418530715,
+      "learning_rate": 1.7256707514634521e-06,
+      "loss": 0.4225,
+      "step": 3009
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5738266276609935,
+      "learning_rate": 1.723884426258277e-06,
+      "loss": 0.4073,
+      "step": 3010
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5804211498201226,
+      "learning_rate": 1.7220985394561445e-06,
+      "loss": 0.4253,
+      "step": 3011
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5703132294176713,
+      "learning_rate": 1.7203130920658457e-06,
+      "loss": 0.4127,
+      "step": 3012
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.580926877910347,
+      "learning_rate": 1.7185280850959215e-06,
+      "loss": 0.4152,
+      "step": 3013
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5807574727618313,
+      "learning_rate": 1.7167435195546683e-06,
+      "loss": 0.4133,
+      "step": 3014
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5834413664173587,
+      "learning_rate": 1.7149593964501285e-06,
+      "loss": 0.4132,
+      "step": 3015
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.592960565681535,
+      "learning_rate": 1.7131757167900966e-06,
+      "loss": 0.4159,
+      "step": 3016
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5734356639268737,
+      "learning_rate": 1.711392481582119e-06,
+      "loss": 0.4037,
+      "step": 3017
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.608289566001257,
+      "learning_rate": 1.7096096918334853e-06,
+      "loss": 0.4225,
+      "step": 3018
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5553937161005909,
+      "learning_rate": 1.7078273485512392e-06,
+      "loss": 0.428,
+      "step": 3019
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5800080754367256,
+      "learning_rate": 1.7060454527421688e-06,
+      "loss": 0.4497,
+      "step": 3020
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5956520471145508,
+      "learning_rate": 1.7042640054128112e-06,
+      "loss": 0.4023,
+      "step": 3021
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5513802042755411,
+      "learning_rate": 1.7024830075694483e-06,
+      "loss": 0.4057,
+      "step": 3022
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5663784002610377,
+      "learning_rate": 1.7007024602181105e-06,
+      "loss": 0.4072,
+      "step": 3023
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5728768465657827,
+      "learning_rate": 1.6989223643645706e-06,
+      "loss": 0.4027,
+      "step": 3024
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5540154660402091,
+      "learning_rate": 1.6971427210143503e-06,
+      "loss": 0.417,
+      "step": 3025
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.580580249925815,
+      "learning_rate": 1.6953635311727126e-06,
+      "loss": 0.4203,
+      "step": 3026
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.6204069176031851,
+      "learning_rate": 1.6935847958446657e-06,
+      "loss": 0.438,
+      "step": 3027
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5536941783725637,
+      "learning_rate": 1.6918065160349604e-06,
+      "loss": 0.4255,
+      "step": 3028
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.569556045523807,
+      "learning_rate": 1.6900286927480898e-06,
+      "loss": 0.4122,
+      "step": 3029
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5546113410095393,
+      "learning_rate": 1.6882513269882916e-06,
+      "loss": 0.4194,
+      "step": 3030
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5749490985614878,
+      "learning_rate": 1.6864744197595418e-06,
+      "loss": 0.4213,
+      "step": 3031
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5849700553588103,
+      "learning_rate": 1.68469797206556e-06,
+      "loss": 0.414,
+      "step": 3032
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5642184593848872,
+      "learning_rate": 1.6829219849098035e-06,
+      "loss": 0.4217,
+      "step": 3033
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5928018316436391,
+      "learning_rate": 1.681146459295473e-06,
+      "loss": 0.4244,
+      "step": 3034
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.591820965440071,
+      "learning_rate": 1.6793713962255043e-06,
+      "loss": 0.4609,
+      "step": 3035
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5886651969708854,
+      "learning_rate": 1.6775967967025764e-06,
+      "loss": 0.418,
+      "step": 3036
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.563924203947021,
+      "learning_rate": 1.675822661729103e-06,
+      "loss": 0.3846,
+      "step": 3037
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5882580889291531,
+      "learning_rate": 1.674048992307237e-06,
+      "loss": 0.4225,
+      "step": 3038
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5609824212530277,
+      "learning_rate": 1.6722757894388675e-06,
+      "loss": 0.3968,
+      "step": 3039
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5783821889365726,
+      "learning_rate": 1.6705030541256211e-06,
+      "loss": 0.4044,
+      "step": 3040
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5904279554578172,
+      "learning_rate": 1.6687307873688583e-06,
+      "loss": 0.4334,
+      "step": 3041
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.6379074710732456,
+      "learning_rate": 1.6669589901696778e-06,
+      "loss": 0.4264,
+      "step": 3042
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5522942159204164,
+      "learning_rate": 1.665187663528912e-06,
+      "loss": 0.4256,
+      "step": 3043
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5665275799582181,
+      "learning_rate": 1.6634168084471252e-06,
+      "loss": 0.4099,
+      "step": 3044
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5800540818159535,
+      "learning_rate": 1.661646425924619e-06,
+      "loss": 0.4272,
+      "step": 3045
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.59834302679002,
+      "learning_rate": 1.6598765169614245e-06,
+      "loss": 0.417,
+      "step": 3046
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5425819631570858,
+      "learning_rate": 1.6581070825573093e-06,
+      "loss": 0.4123,
+      "step": 3047
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.549947011538962,
+      "learning_rate": 1.6563381237117688e-06,
+      "loss": 0.4141,
+      "step": 3048
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5708925242028148,
+      "learning_rate": 1.6545696414240326e-06,
+      "loss": 0.4306,
+      "step": 3049
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5742846219284804,
+      "learning_rate": 1.6528016366930594e-06,
+      "loss": 0.4147,
+      "step": 3050
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.6145663624908198,
+      "learning_rate": 1.6510341105175401e-06,
+      "loss": 0.4262,
+      "step": 3051
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.6072916721819851,
+      "learning_rate": 1.6492670638958924e-06,
+      "loss": 0.441,
+      "step": 3052
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5834453619650581,
+      "learning_rate": 1.647500497826267e-06,
+      "loss": 0.4159,
+      "step": 3053
+    },
+    {
+      "epoch": 1.29,
+      "eval_loss": 0.4601620137691498,
+      "eval_runtime": 6944.585,
+      "eval_samples_per_second": 41.815,
+      "eval_steps_per_second": 2.091,
+      "step": 3053
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.558687925184084,
+      "learning_rate": 1.6457344133065395e-06,
+      "loss": 0.4053,
+      "step": 3054
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5674610161351168,
+      "learning_rate": 1.643968811334315e-06,
+      "loss": 0.417,
+      "step": 3055
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5640253416506891,
+      "learning_rate": 1.642203692906927e-06,
+      "loss": 0.3994,
+      "step": 3056
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5508590652466895,
+      "learning_rate": 1.640439059021433e-06,
+      "loss": 0.4221,
+      "step": 3057
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5938928064517502,
+      "learning_rate": 1.6386749106746214e-06,
+      "loss": 0.4267,
+      "step": 3058
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5570296609089503,
+      "learning_rate": 1.6369112488630009e-06,
+      "loss": 0.4032,
+      "step": 3059
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5854299661276051,
+      "learning_rate": 1.6351480745828098e-06,
+      "loss": 0.4095,
+      "step": 3060
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.559667720654985,
+      "learning_rate": 1.6333853888300083e-06,
+      "loss": 0.4221,
+      "step": 3061
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5735241871511867,
+      "learning_rate": 1.6316231926002823e-06,
+      "loss": 0.395,
+      "step": 3062
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5416537755969776,
+      "learning_rate": 1.629861486889039e-06,
+      "loss": 0.3963,
+      "step": 3063
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5673587035921341,
+      "learning_rate": 1.6281002726914125e-06,
+      "loss": 0.4109,
+      "step": 3064
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5433392636524627,
+      "learning_rate": 1.6263395510022546e-06,
+      "loss": 0.4001,
+      "step": 3065
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5550782192229988,
+      "learning_rate": 1.6245793228161421e-06,
+      "loss": 0.4045,
+      "step": 3066
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5861247355677058,
+      "learning_rate": 1.622819589127372e-06,
+      "loss": 0.4346,
+      "step": 3067
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.607146634130001,
+      "learning_rate": 1.6210603509299604e-06,
+      "loss": 0.4115,
+      "step": 3068
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5562854235248,
+      "learning_rate": 1.6193016092176484e-06,
+      "loss": 0.4203,
+      "step": 3069
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5627927077991777,
+      "learning_rate": 1.6175433649838901e-06,
+      "loss": 0.4262,
+      "step": 3070
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5750081673543436,
+      "learning_rate": 1.615785619221864e-06,
+      "loss": 0.3833,
+      "step": 3071
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5936159776999882,
+      "learning_rate": 1.6140283729244638e-06,
+      "loss": 0.4203,
+      "step": 3072
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5718250335162808,
+      "learning_rate": 1.6122716270843025e-06,
+      "loss": 0.4065,
+      "step": 3073
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5771598792199492,
+      "learning_rate": 1.6105153826937087e-06,
+      "loss": 0.4185,
+      "step": 3074
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5759622276853105,
+      "learning_rate": 1.6087596407447314e-06,
+      "loss": 0.42,
+      "step": 3075
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.611717304812912,
+      "learning_rate": 1.607004402229132e-06,
+      "loss": 0.4072,
+      "step": 3076
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5872035309615434,
+      "learning_rate": 1.60524966813839e-06,
+      "loss": 0.4354,
+      "step": 3077
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5811950216335751,
+      "learning_rate": 1.6034954394636977e-06,
+      "loss": 0.4073,
+      "step": 3078
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5706512241287943,
+      "learning_rate": 1.6017417171959643e-06,
+      "loss": 0.4203,
+      "step": 3079
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5826675112936385,
+      "learning_rate": 1.5999885023258099e-06,
+      "loss": 0.3995,
+      "step": 3080
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5682279192683956,
+      "learning_rate": 1.5982357958435723e-06,
+      "loss": 0.3878,
+      "step": 3081
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5713141319558672,
+      "learning_rate": 1.5964835987392991e-06,
+      "loss": 0.4014,
+      "step": 3082
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5747223757155583,
+      "learning_rate": 1.59473191200275e-06,
+      "loss": 0.4146,
+      "step": 3083
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5809784351004627,
+      "learning_rate": 1.5929807366233979e-06,
+      "loss": 0.4445,
+      "step": 3084
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5632117792726165,
+      "learning_rate": 1.5912300735904252e-06,
+      "loss": 0.3946,
+      "step": 3085
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.581279032610451,
+      "learning_rate": 1.5894799238927277e-06,
+      "loss": 0.395,
+      "step": 3086
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5877431541038257,
+      "learning_rate": 1.5877302885189077e-06,
+      "loss": 0.4212,
+      "step": 3087
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.6178627095748951,
+      "learning_rate": 1.5859811684572796e-06,
+      "loss": 0.4237,
+      "step": 3088
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5791179237596596,
+      "learning_rate": 1.584232564695865e-06,
+      "loss": 0.4243,
+      "step": 3089
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5657862133711392,
+      "learning_rate": 1.5824844782223956e-06,
+      "loss": 0.4234,
+      "step": 3090
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5767524507801428,
+      "learning_rate": 1.5807369100243084e-06,
+      "loss": 0.4211,
+      "step": 3091
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5697331723482661,
+      "learning_rate": 1.578989861088751e-06,
+      "loss": 0.414,
+      "step": 3092
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5476753293523711,
+      "learning_rate": 1.5772433324025748e-06,
+      "loss": 0.406,
+      "step": 3093
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5561716385297626,
+      "learning_rate": 1.5754973249523387e-06,
+      "loss": 0.4155,
+      "step": 3094
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.572512956688457,
+      "learning_rate": 1.5737518397243074e-06,
+      "loss": 0.4286,
+      "step": 3095
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5939672288238211,
+      "learning_rate": 1.5720068777044479e-06,
+      "loss": 0.4276,
+      "step": 3096
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5966500948903306,
+      "learning_rate": 1.570262439878437e-06,
+      "loss": 0.4211,
+      "step": 3097
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5952090362230588,
+      "learning_rate": 1.56851852723165e-06,
+      "loss": 0.4447,
+      "step": 3098
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5673871145350723,
+      "learning_rate": 1.5667751407491689e-06,
+      "loss": 0.4408,
+      "step": 3099
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5663044725908228,
+      "learning_rate": 1.5650322814157764e-06,
+      "loss": 0.4219,
+      "step": 3100
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5902068842643253,
+      "learning_rate": 1.5632899502159594e-06,
+      "loss": 0.4104,
+      "step": 3101
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5961011802361124,
+      "learning_rate": 1.561548148133904e-06,
+      "loss": 0.3999,
+      "step": 3102
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5676333875668861,
+      "learning_rate": 1.559806876153501e-06,
+      "loss": 0.42,
+      "step": 3103
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5717204537217901,
+      "learning_rate": 1.5580661352583377e-06,
+      "loss": 0.3927,
+      "step": 3104
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5932365555296837,
+      "learning_rate": 1.5563259264317048e-06,
+      "loss": 0.4069,
+      "step": 3105
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.60609083336309,
+      "learning_rate": 1.55458625065659e-06,
+      "loss": 0.4232,
+      "step": 3106
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5568592972383142,
+      "learning_rate": 1.5528471089156805e-06,
+      "loss": 0.4035,
+      "step": 3107
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5811068115782658,
+      "learning_rate": 1.5511085021913644e-06,
+      "loss": 0.4381,
+      "step": 3108
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5978207369401186,
+      "learning_rate": 1.5493704314657232e-06,
+      "loss": 0.438,
+      "step": 3109
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5720438095205538,
+      "learning_rate": 1.5476328977205396e-06,
+      "loss": 0.4279,
+      "step": 3110
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5893474852498958,
+      "learning_rate": 1.5458959019372893e-06,
+      "loss": 0.431,
+      "step": 3111
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5490466454197072,
+      "learning_rate": 1.544159445097148e-06,
+      "loss": 0.4092,
+      "step": 3112
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5384899530575747,
+      "learning_rate": 1.542423528180983e-06,
+      "loss": 0.3936,
+      "step": 3113
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5850096033351098,
+      "learning_rate": 1.5406881521693606e-06,
+      "loss": 0.4019,
+      "step": 3114
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.587195704302166,
+      "learning_rate": 1.5389533180425387e-06,
+      "loss": 0.4144,
+      "step": 3115
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5635366537239104,
+      "learning_rate": 1.5372190267804704e-06,
+      "loss": 0.4129,
+      "step": 3116
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5718044990589992,
+      "learning_rate": 1.5354852793628007e-06,
+      "loss": 0.4009,
+      "step": 3117
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5926450332610386,
+      "learning_rate": 1.53375207676887e-06,
+      "loss": 0.4348,
+      "step": 3118
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.564012302965777,
+      "learning_rate": 1.5320194199777078e-06,
+      "loss": 0.4008,
+      "step": 3119
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5881564721462378,
+      "learning_rate": 1.5302873099680378e-06,
+      "loss": 0.4116,
+      "step": 3120
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5805570882385258,
+      "learning_rate": 1.5285557477182744e-06,
+      "loss": 0.3955,
+      "step": 3121
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.601026663572398,
+      "learning_rate": 1.5268247342065215e-06,
+      "loss": 0.4222,
+      "step": 3122
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5688916955974609,
+      "learning_rate": 1.525094270410574e-06,
+      "loss": 0.4077,
+      "step": 3123
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.6376814875861784,
+      "learning_rate": 1.5233643573079148e-06,
+      "loss": 0.4198,
+      "step": 3124
+    },
+    {
+      "epoch": 1.32,
+      "eval_loss": 0.45952433347702026,
+      "eval_runtime": 6938.4565,
+      "eval_samples_per_second": 41.852,
+      "eval_steps_per_second": 2.093,
+      "step": 3124
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5675589935533134,
+      "learning_rate": 1.5216349958757187e-06,
+      "loss": 0.4276,
+      "step": 3125
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5819253780607699,
+      "learning_rate": 1.5199061870908457e-06,
+      "loss": 0.4135,
+      "step": 3126
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5473928976130037,
+      "learning_rate": 1.518177931929846e-06,
+      "loss": 0.39,
+      "step": 3127
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5668895033063771,
+      "learning_rate": 1.516450231368955e-06,
+      "loss": 0.4145,
+      "step": 3128
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5566976729596672,
+      "learning_rate": 1.5147230863840968e-06,
+      "loss": 0.417,
+      "step": 3129
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5599445867608915,
+      "learning_rate": 1.5129964979508792e-06,
+      "loss": 0.4049,
+      "step": 3130
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5689525691882638,
+      "learning_rate": 1.5112704670445994e-06,
+      "loss": 0.414,
+      "step": 3131
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5745693019904166,
+      "learning_rate": 1.509544994640236e-06,
+      "loss": 0.415,
+      "step": 3132
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5703134113139972,
+      "learning_rate": 1.507820081712454e-06,
+      "loss": 0.4202,
+      "step": 3133
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5798680593678225,
+      "learning_rate": 1.5060957292356021e-06,
+      "loss": 0.4224,
+      "step": 3134
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5634109776233843,
+      "learning_rate": 1.5043719381837113e-06,
+      "loss": 0.4165,
+      "step": 3135
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.6007001621304139,
+      "learning_rate": 1.5026487095304982e-06,
+      "loss": 0.4143,
+      "step": 3136
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5816452933961378,
+      "learning_rate": 1.5009260442493582e-06,
+      "loss": 0.3987,
+      "step": 3137
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5520751522048931,
+      "learning_rate": 1.4992039433133715e-06,
+      "loss": 0.4151,
+      "step": 3138
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.6053997190844257,
+      "learning_rate": 1.497482407695297e-06,
+      "loss": 0.4311,
+      "step": 3139
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5574155007809499,
+      "learning_rate": 1.495761438367577e-06,
+      "loss": 0.3934,
+      "step": 3140
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5612828220380264,
+      "learning_rate": 1.4940410363023306e-06,
+      "loss": 0.3898,
+      "step": 3141
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5900288585719455,
+      "learning_rate": 1.4923212024713602e-06,
+      "loss": 0.4369,
+      "step": 3142
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5818357215539803,
+      "learning_rate": 1.4906019378461437e-06,
+      "loss": 0.4448,
+      "step": 3143
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.553139692347851,
+      "learning_rate": 1.4888832433978403e-06,
+      "loss": 0.426,
+      "step": 3144
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5952446958459735,
+      "learning_rate": 1.4871651200972854e-06,
+      "loss": 0.4246,
+      "step": 3145
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.6187484191420308,
+      "learning_rate": 1.485447568914991e-06,
+      "loss": 0.4145,
+      "step": 3146
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.6002358417607456,
+      "learning_rate": 1.4837305908211502e-06,
+      "loss": 0.407,
+      "step": 3147
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.55885003361605,
+      "learning_rate": 1.4820141867856268e-06,
+      "loss": 0.3977,
+      "step": 3148
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5703620407770618,
+      "learning_rate": 1.4802983577779651e-06,
+      "loss": 0.4402,
+      "step": 3149
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5752264024437074,
+      "learning_rate": 1.47858310476738e-06,
+      "loss": 0.4081,
+      "step": 3150
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.587484294352918,
+      "learning_rate": 1.4768684287227652e-06,
+      "loss": 0.4258,
+      "step": 3151
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.577277316397856,
+      "learning_rate": 1.4751543306126856e-06,
+      "loss": 0.4076,
+      "step": 3152
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5585293413380346,
+      "learning_rate": 1.4734408114053822e-06,
+      "loss": 0.4066,
+      "step": 3153
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5996038294832453,
+      "learning_rate": 1.471727872068766e-06,
+      "loss": 0.413,
+      "step": 3154
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5908728837224524,
+      "learning_rate": 1.470015513570424e-06,
+      "loss": 0.4135,
+      "step": 3155
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5973053416179713,
+      "learning_rate": 1.468303736877611e-06,
+      "loss": 0.4216,
+      "step": 3156
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5783789948908563,
+      "learning_rate": 1.466592542957257e-06,
+      "loss": 0.4292,
+      "step": 3157
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.549412909563158,
+      "learning_rate": 1.4648819327759589e-06,
+      "loss": 0.4151,
+      "step": 3158
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5638177225917886,
+      "learning_rate": 1.4631719072999884e-06,
+      "loss": 0.43,
+      "step": 3159
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5811841444119168,
+      "learning_rate": 1.4614624674952843e-06,
+      "loss": 0.4199,
+      "step": 3160
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5745757615240445,
+      "learning_rate": 1.4597536143274537e-06,
+      "loss": 0.4223,
+      "step": 3161
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5695416398603609,
+      "learning_rate": 1.4580453487617747e-06,
+      "loss": 0.403,
+      "step": 3162
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5643072461632591,
+      "learning_rate": 1.4563376717631906e-06,
+      "loss": 0.4024,
+      "step": 3163
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5855374862753647,
+      "learning_rate": 1.4546305842963156e-06,
+      "loss": 0.4169,
+      "step": 3164
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5640742939680888,
+      "learning_rate": 1.452924087325428e-06,
+      "loss": 0.3986,
+      "step": 3165
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5854422692893917,
+      "learning_rate": 1.4512181818144763e-06,
+      "loss": 0.4092,
+      "step": 3166
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6013198747012948,
+      "learning_rate": 1.4495128687270682e-06,
+      "loss": 0.4147,
+      "step": 3167
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6349112878398844,
+      "learning_rate": 1.4478081490264841e-06,
+      "loss": 0.4013,
+      "step": 3168
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6018833225607165,
+      "learning_rate": 1.4461040236756643e-06,
+      "loss": 0.4034,
+      "step": 3169
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.572757916476602,
+      "learning_rate": 1.4444004936372166e-06,
+      "loss": 0.4139,
+      "step": 3170
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5964839570478256,
+      "learning_rate": 1.4426975598734103e-06,
+      "loss": 0.4403,
+      "step": 3171
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6532887857894302,
+      "learning_rate": 1.4409952233461777e-06,
+      "loss": 0.4089,
+      "step": 3172
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.599992570422661,
+      "learning_rate": 1.4392934850171161e-06,
+      "loss": 0.4328,
+      "step": 3173
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5924831441960442,
+      "learning_rate": 1.4375923458474822e-06,
+      "loss": 0.4397,
+      "step": 3174
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5694039028586787,
+      "learning_rate": 1.4358918067981969e-06,
+      "loss": 0.4052,
+      "step": 3175
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5638688789990264,
+      "learning_rate": 1.434191868829839e-06,
+      "loss": 0.4306,
+      "step": 3176
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6070863428317758,
+      "learning_rate": 1.4324925329026526e-06,
+      "loss": 0.4069,
+      "step": 3177
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5970977280572529,
+      "learning_rate": 1.4307937999765343e-06,
+      "loss": 0.4264,
+      "step": 3178
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5400844435082709,
+      "learning_rate": 1.4290956710110477e-06,
+      "loss": 0.4114,
+      "step": 3179
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6189005664696162,
+      "learning_rate": 1.4273981469654093e-06,
+      "loss": 0.4393,
+      "step": 3180
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.5551865309592269,
+      "learning_rate": 1.4257012287984994e-06,
+      "loss": 0.3815,
+      "step": 3181
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5788766743190364,
+      "learning_rate": 1.4240049174688514e-06,
+      "loss": 0.4129,
+      "step": 3182
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.6145740654828784,
+      "learning_rate": 1.4223092139346583e-06,
+      "loss": 0.4465,
+      "step": 3183
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.56123366419593,
+      "learning_rate": 1.4206141191537681e-06,
+      "loss": 0.4226,
+      "step": 3184
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5702082132713671,
+      "learning_rate": 1.4189196340836866e-06,
+      "loss": 0.4407,
+      "step": 3185
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.568086319678547,
+      "learning_rate": 1.4172257596815762e-06,
+      "loss": 0.4235,
+      "step": 3186
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.570650833706996,
+      "learning_rate": 1.41553249690425e-06,
+      "loss": 0.397,
+      "step": 3187
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5758906759310528,
+      "learning_rate": 1.413839846708182e-06,
+      "loss": 0.4008,
+      "step": 3188
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5867057883601531,
+      "learning_rate": 1.4121478100494926e-06,
+      "loss": 0.3931,
+      "step": 3189
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.6216370173036398,
+      "learning_rate": 1.4104563878839623e-06,
+      "loss": 0.432,
+      "step": 3190
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5834896894589426,
+      "learning_rate": 1.4087655811670196e-06,
+      "loss": 0.3892,
+      "step": 3191
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.55572187008008,
+      "learning_rate": 1.4070753908537498e-06,
+      "loss": 0.3945,
+      "step": 3192
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.6195431150659669,
+      "learning_rate": 1.4053858178988866e-06,
+      "loss": 0.4252,
+      "step": 3193
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5855307932366061,
+      "learning_rate": 1.4036968632568163e-06,
+      "loss": 0.4023,
+      "step": 3194
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5785248405460688,
+      "learning_rate": 1.4020085278815745e-06,
+      "loss": 0.4187,
+      "step": 3195
+    },
+    {
+      "epoch": 1.35,
+      "eval_loss": 0.4585270285606384,
+      "eval_runtime": 6935.0493,
+      "eval_samples_per_second": 41.873,
+      "eval_steps_per_second": 2.094,
+      "step": 3195
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5793269431905544,
+      "learning_rate": 1.4003208127268503e-06,
+      "loss": 0.4362,
+      "step": 3196
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5635960927267711,
+      "learning_rate": 1.3986337187459787e-06,
+      "loss": 0.396,
+      "step": 3197
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5675588328279935,
+      "learning_rate": 1.3969472468919462e-06,
+      "loss": 0.4357,
+      "step": 3198
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5813371441566617,
+      "learning_rate": 1.3952613981173894e-06,
+      "loss": 0.4202,
+      "step": 3199
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5567802235455163,
+      "learning_rate": 1.3935761733745865e-06,
+      "loss": 0.4131,
+      "step": 3200
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5821477938771911,
+      "learning_rate": 1.3918915736154704e-06,
+      "loss": 0.4214,
+      "step": 3201
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5964536001348693,
+      "learning_rate": 1.3902075997916164e-06,
+      "loss": 0.4336,
+      "step": 3202
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.6086865744806292,
+      "learning_rate": 1.3885242528542497e-06,
+      "loss": 0.4551,
+      "step": 3203
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.575959160272487,
+      "learning_rate": 1.3868415337542382e-06,
+      "loss": 0.4224,
+      "step": 3204
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.6026326885124447,
+      "learning_rate": 1.3851594434420968e-06,
+      "loss": 0.4202,
+      "step": 3205
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5493404687568753,
+      "learning_rate": 1.383477982867984e-06,
+      "loss": 0.3954,
+      "step": 3206
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5781348712939294,
+      "learning_rate": 1.3817971529817054e-06,
+      "loss": 0.3956,
+      "step": 3207
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5848758929707726,
+      "learning_rate": 1.380116954732706e-06,
+      "loss": 0.4298,
+      "step": 3208
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5839753363394727,
+      "learning_rate": 1.3784373890700789e-06,
+      "loss": 0.4159,
+      "step": 3209
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5473659827496421,
+      "learning_rate": 1.3767584569425562e-06,
+      "loss": 0.3982,
+      "step": 3210
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5854856789628843,
+      "learning_rate": 1.375080159298513e-06,
+      "loss": 0.3985,
+      "step": 3211
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.6223285089698908,
+      "learning_rate": 1.3734024970859672e-06,
+      "loss": 0.3924,
+      "step": 3212
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5818260019212478,
+      "learning_rate": 1.3717254712525758e-06,
+      "loss": 0.4091,
+      "step": 3213
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.588671059523623,
+      "learning_rate": 1.3700490827456393e-06,
+      "loss": 0.4278,
+      "step": 3214
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.542860404907633,
+      "learning_rate": 1.3683733325120934e-06,
+      "loss": 0.4131,
+      "step": 3215
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5562613227316864,
+      "learning_rate": 1.3666982214985208e-06,
+      "loss": 0.4245,
+      "step": 3216
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 4.123028508699682,
+      "learning_rate": 1.3650237506511333e-06,
+      "loss": 0.4637,
+      "step": 3217
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.590425927959101,
+      "learning_rate": 1.3633499209157898e-06,
+      "loss": 0.4267,
+      "step": 3218
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5832844350075718,
+      "learning_rate": 1.3616767332379815e-06,
+      "loss": 0.4182,
+      "step": 3219
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5960509584308913,
+      "learning_rate": 1.360004188562841e-06,
+      "loss": 0.4169,
+      "step": 3220
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.592810575458578,
+      "learning_rate": 1.3583322878351346e-06,
+      "loss": 0.427,
+      "step": 3221
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5842647152175641,
+      "learning_rate": 1.3566610319992658e-06,
+      "loss": 0.4101,
+      "step": 3222
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5687572730019927,
+      "learning_rate": 1.3549904219992732e-06,
+      "loss": 0.4027,
+      "step": 3223
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.563025461962013,
+      "learning_rate": 1.3533204587788323e-06,
+      "loss": 0.4068,
+      "step": 3224
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5694816221991685,
+      "learning_rate": 1.351651143281253e-06,
+      "loss": 0.4101,
+      "step": 3225
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5642547386099277,
+      "learning_rate": 1.3499824764494773e-06,
+      "loss": 0.4127,
+      "step": 3226
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5480375629168616,
+      "learning_rate": 1.3483144592260844e-06,
+      "loss": 0.397,
+      "step": 3227
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5653603918404666,
+      "learning_rate": 1.346647092553281e-06,
+      "loss": 0.4109,
+      "step": 3228
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5806409790053582,
+      "learning_rate": 1.3449803773729115e-06,
+      "loss": 0.3967,
+      "step": 3229
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5605529434144851,
+      "learning_rate": 1.3433143146264494e-06,
+      "loss": 0.4287,
+      "step": 3230
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5722146410788806,
+      "learning_rate": 1.3416489052550019e-06,
+      "loss": 0.4006,
+      "step": 3231
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.575347179976502,
+      "learning_rate": 1.3399841501993056e-06,
+      "loss": 0.4116,
+      "step": 3232
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5949791657298646,
+      "learning_rate": 1.338320050399727e-06,
+      "loss": 0.4303,
+      "step": 3233
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5794211854463649,
+      "learning_rate": 1.3366566067962628e-06,
+      "loss": 0.4124,
+      "step": 3234
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5793797643264423,
+      "learning_rate": 1.3349938203285412e-06,
+      "loss": 0.4275,
+      "step": 3235
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5842493222488487,
+      "learning_rate": 1.3333316919358159e-06,
+      "loss": 0.4209,
+      "step": 3236
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5663991497625657,
+      "learning_rate": 1.3316702225569708e-06,
+      "loss": 0.3918,
+      "step": 3237
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5486915223283709,
+      "learning_rate": 1.3300094131305196e-06,
+      "loss": 0.4073,
+      "step": 3238
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5350475982264186,
+      "learning_rate": 1.3283492645945966e-06,
+      "loss": 0.3985,
+      "step": 3239
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5886624292798438,
+      "learning_rate": 1.3266897778869704e-06,
+      "loss": 0.393,
+      "step": 3240
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5755756067045215,
+      "learning_rate": 1.3250309539450298e-06,
+      "loss": 0.4184,
+      "step": 3241
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5677276610035034,
+      "learning_rate": 1.3233727937057938e-06,
+      "loss": 0.4299,
+      "step": 3242
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5690350952201221,
+      "learning_rate": 1.3217152981059043e-06,
+      "loss": 0.4292,
+      "step": 3243
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5739165525454375,
+      "learning_rate": 1.320058468081627e-06,
+      "loss": 0.4076,
+      "step": 3244
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5613813103721713,
+      "learning_rate": 1.3184023045688515e-06,
+      "loss": 0.4214,
+      "step": 3245
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5954222168994903,
+      "learning_rate": 1.3167468085030948e-06,
+      "loss": 0.4513,
+      "step": 3246
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5502927352970999,
+      "learning_rate": 1.3150919808194917e-06,
+      "loss": 0.4156,
+      "step": 3247
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5608089205143509,
+      "learning_rate": 1.3134378224528026e-06,
+      "loss": 0.416,
+      "step": 3248
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5536001064152207,
+      "learning_rate": 1.311784334337411e-06,
+      "loss": 0.4126,
+      "step": 3249
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5766677004057259,
+      "learning_rate": 1.3101315174073162e-06,
+      "loss": 0.4304,
+      "step": 3250
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5776009437096029,
+      "learning_rate": 1.3084793725961447e-06,
+      "loss": 0.3962,
+      "step": 3251
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5479613505888217,
+      "learning_rate": 1.3068279008371387e-06,
+      "loss": 0.3994,
+      "step": 3252
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5690887730011218,
+      "learning_rate": 1.3051771030631644e-06,
+      "loss": 0.4088,
+      "step": 3253
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5799412277505956,
+      "learning_rate": 1.303526980206704e-06,
+      "loss": 0.422,
+      "step": 3254
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5570959933884828,
+      "learning_rate": 1.301877533199859e-06,
+      "loss": 0.4104,
+      "step": 3255
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5643212278599219,
+      "learning_rate": 1.3002287629743488e-06,
+      "loss": 0.4161,
+      "step": 3256
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5618675409375589,
+      "learning_rate": 1.2985806704615139e-06,
+      "loss": 0.4352,
+      "step": 3257
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5795363521047754,
+      "learning_rate": 1.2969332565923068e-06,
+      "loss": 0.4149,
+      "step": 3258
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5607134789936746,
+      "learning_rate": 1.2952865222973015e-06,
+      "loss": 0.3938,
+      "step": 3259
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5774563064680311,
+      "learning_rate": 1.2936404685066852e-06,
+      "loss": 0.4031,
+      "step": 3260
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5784824116412987,
+      "learning_rate": 1.2919950961502603e-06,
+      "loss": 0.4175,
+      "step": 3261
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5685744391843301,
+      "learning_rate": 1.2903504061574467e-06,
+      "loss": 0.4102,
+      "step": 3262
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.6133989126312784,
+      "learning_rate": 1.2887063994572765e-06,
+      "loss": 0.4397,
+      "step": 3263
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5576015312128622,
+      "learning_rate": 1.2870630769783985e-06,
+      "loss": 0.4254,
+      "step": 3264
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5854336579154686,
+      "learning_rate": 1.2854204396490722e-06,
+      "loss": 0.425,
+      "step": 3265
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5851979998505886,
+      "learning_rate": 1.2837784883971716e-06,
+      "loss": 0.4158,
+      "step": 3266
+    },
+    {
+      "epoch": 1.38,
+      "eval_loss": 0.4578353464603424,
+      "eval_runtime": 6934.1495,
+      "eval_samples_per_second": 41.878,
+      "eval_steps_per_second": 2.094,
+      "step": 3266
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5605246475286948,
+      "learning_rate": 1.2821372241501814e-06,
+      "loss": 0.4197,
+      "step": 3267
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5907220198204197,
+      "learning_rate": 1.280496647835202e-06,
+      "loss": 0.4239,
+      "step": 3268
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5543362298552139,
+      "learning_rate": 1.278856760378941e-06,
+      "loss": 0.4027,
+      "step": 3269
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5824324760965199,
+      "learning_rate": 1.2772175627077204e-06,
+      "loss": 0.3936,
+      "step": 3270
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5596093376012918,
+      "learning_rate": 1.27557905574747e-06,
+      "loss": 0.3892,
+      "step": 3271
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5847030750956056,
+      "learning_rate": 1.2739412404237306e-06,
+      "loss": 0.4517,
+      "step": 3272
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5532504862482353,
+      "learning_rate": 1.2723041176616508e-06,
+      "loss": 0.4178,
+      "step": 3273
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5842377189415942,
+      "learning_rate": 1.2706676883859902e-06,
+      "loss": 0.4135,
+      "step": 3274
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.5442392432462477,
+      "learning_rate": 1.2690319535211171e-06,
+      "loss": 0.3921,
+      "step": 3275
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.5655162202220977,
+      "learning_rate": 1.2673969139910047e-06,
+      "loss": 0.4281,
+      "step": 3276
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 4680,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 468,
+  "total_flos": 3429507897753600.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}