diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10581 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.4857444561774025,
+  "eval_steps": 60,
+  "global_step": 1480,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 26.0,
+      "learning_rate": 6.666666666666667e-08,
+      "loss": 2.5036,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 2.734731435775757,
+      "eval_runtime": 73.3418,
+      "eval_samples_per_second": 8.849,
+      "eval_steps_per_second": 8.849,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 25.25,
+      "learning_rate": 1.3333333333333334e-07,
+      "loss": 2.8732,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 23.0,
+      "learning_rate": 2e-07,
+      "loss": 2.6451,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 26.375,
+      "learning_rate": 2.6666666666666667e-07,
+      "loss": 2.6047,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 23.0,
+      "learning_rate": 3.333333333333333e-07,
+      "loss": 2.7992,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 24.75,
+      "learning_rate": 4e-07,
+      "loss": 2.7625,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 25.5,
+      "learning_rate": 4.6666666666666666e-07,
+      "loss": 2.5595,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 29.25,
+      "learning_rate": 5.333333333333333e-07,
+      "loss": 2.8246,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 19.75,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 2.4177,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 23.375,
+      "learning_rate": 6.666666666666666e-07,
+      "loss": 2.5124,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 26.375,
+      "learning_rate": 7.333333333333333e-07,
+      "loss": 2.4839,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 23.75,
+      "learning_rate": 8e-07,
+      "loss": 2.7422,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 26.5,
+      "learning_rate": 8.666666666666666e-07,
+      "loss": 2.7973,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 25.0,
+      "learning_rate": 9.333333333333333e-07,
+      "loss": 2.914,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 23.75,
+      "learning_rate": 1e-06,
+      "loss": 2.9031,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 23.125,
+      "learning_rate": 1.0666666666666667e-06,
+      "loss": 2.6065,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 22.75,
+      "learning_rate": 1.1333333333333334e-06,
+      "loss": 2.7575,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 20.375,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 2.8768,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 17.375,
+      "learning_rate": 1.2666666666666667e-06,
+      "loss": 2.7704,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 15.25,
+      "learning_rate": 1.3333333333333332e-06,
+      "loss": 2.5928,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 15.625,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 2.5997,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 18.25,
+      "learning_rate": 1.4666666666666667e-06,
+      "loss": 2.605,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 14.4375,
+      "learning_rate": 1.5333333333333332e-06,
+      "loss": 2.7531,
+      "step": 23
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 14.625,
+      "learning_rate": 1.6e-06,
+      "loss": 2.6625,
+      "step": 24
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 9.1875,
+      "learning_rate": 1.6666666666666669e-06,
+      "loss": 2.5257,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 11.875,
+      "learning_rate": 1.7333333333333332e-06,
+      "loss": 2.5876,
+      "step": 26
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 9.0,
+      "learning_rate": 1.8e-06,
+      "loss": 2.473,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 8.9375,
+      "learning_rate": 1.8666666666666667e-06,
+      "loss": 2.2859,
+      "step": 28
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 8.5625,
+      "learning_rate": 1.9333333333333336e-06,
+      "loss": 2.6034,
+      "step": 29
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 8.6875,
+      "learning_rate": 2e-06,
+      "loss": 2.745,
+      "step": 30
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 8.125,
+      "learning_rate": 2.0666666666666666e-06,
+      "loss": 2.597,
+      "step": 31
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.125,
+      "learning_rate": 2.1333333333333334e-06,
+      "loss": 2.4495,
+      "step": 32
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.75,
+      "learning_rate": 2.1999999999999997e-06,
+      "loss": 2.5442,
+      "step": 33
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 7.09375,
+      "learning_rate": 2.266666666666667e-06,
+      "loss": 2.5841,
+      "step": 34
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.46875,
+      "learning_rate": 2.3333333333333336e-06,
+      "loss": 2.3175,
+      "step": 35
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 5.03125,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 2.4678,
+      "step": 36
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.96875,
+      "learning_rate": 2.4666666666666666e-06,
+      "loss": 2.3823,
+      "step": 37
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.5,
+      "learning_rate": 2.5333333333333334e-06,
+      "loss": 2.2722,
+      "step": 38
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.78125,
+      "learning_rate": 2.6e-06,
+      "loss": 2.6632,
+      "step": 39
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.75,
+      "learning_rate": 2.6666666666666664e-06,
+      "loss": 2.3124,
+      "step": 40
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.3125,
+      "learning_rate": 2.733333333333333e-06,
+      "loss": 2.3315,
+      "step": 41
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 3.796875,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 2.3679,
+      "step": 42
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.8125,
+      "learning_rate": 2.866666666666667e-06,
+      "loss": 2.3584,
+      "step": 43
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.953125,
+      "learning_rate": 2.9333333333333333e-06,
+      "loss": 2.1787,
+      "step": 44
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.1875,
+      "learning_rate": 3e-06,
+      "loss": 2.4291,
+      "step": 45
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.921875,
+      "learning_rate": 3.0666666666666664e-06,
+      "loss": 2.3665,
+      "step": 46
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.28125,
+      "learning_rate": 3.1333333333333335e-06,
+      "loss": 2.4683,
+      "step": 47
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.859375,
+      "learning_rate": 3.2e-06,
+      "loss": 2.327,
+      "step": 48
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 3.40625,
+      "learning_rate": 3.2666666666666666e-06,
+      "loss": 2.3169,
+      "step": 49
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.484375,
+      "learning_rate": 3.3333333333333337e-06,
+      "loss": 2.043,
+      "step": 50
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.0,
+      "learning_rate": 3.4e-06,
+      "loss": 2.4445,
+      "step": 51
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.484375,
+      "learning_rate": 3.4666666666666664e-06,
+      "loss": 2.6345,
+      "step": 52
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.09375,
+      "learning_rate": 3.5333333333333335e-06,
+      "loss": 2.4101,
+      "step": 53
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.28125,
+      "learning_rate": 3.6e-06,
+      "loss": 2.3681,
+      "step": 54
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 2.5,
+      "learning_rate": 3.666666666666667e-06,
+      "loss": 2.4041,
+      "step": 55
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.21875,
+      "learning_rate": 3.7333333333333333e-06,
+      "loss": 2.5763,
+      "step": 56
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.96875,
+      "learning_rate": 3.8e-06,
+      "loss": 2.2547,
+      "step": 57
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.25,
+      "learning_rate": 3.866666666666667e-06,
+      "loss": 2.3994,
+      "step": 58
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9765625,
+      "learning_rate": 3.9333333333333335e-06,
+      "loss": 2.3216,
+      "step": 59
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.25,
+      "learning_rate": 4e-06,
+      "loss": 2.239,
+      "step": 60
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 2.207598924636841,
+      "eval_runtime": 72.558,
+      "eval_samples_per_second": 8.945,
+      "eval_steps_per_second": 8.945,
+      "step": 60
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.0,
+      "learning_rate": 4.066666666666667e-06,
+      "loss": 2.0404,
+      "step": 61
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.133333333333333e-06,
+      "loss": 2.0953,
+      "step": 62
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.078125,
+      "learning_rate": 4.2e-06,
+      "loss": 2.2906,
+      "step": 63
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.609375,
+      "learning_rate": 4.266666666666667e-06,
+      "loss": 2.2262,
+      "step": 64
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.03125,
+      "learning_rate": 4.333333333333333e-06,
+      "loss": 2.2607,
+      "step": 65
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.25,
+      "learning_rate": 4.399999999999999e-06,
+      "loss": 2.2938,
+      "step": 66
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.265625,
+      "learning_rate": 4.4666666666666665e-06,
+      "loss": 2.1453,
+      "step": 67
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.5546875,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 2.0149,
+      "step": 68
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.234375,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 2.4088,
+      "step": 69
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8125,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 2.1434,
+      "step": 70
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.6640625,
+      "learning_rate": 4.7333333333333335e-06,
+      "loss": 2.0821,
+      "step": 71
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 2.1367,
+      "step": 72
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 2.375,
+      "learning_rate": 4.866666666666667e-06,
+      "loss": 2.3554,
+      "step": 73
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.640625,
+      "learning_rate": 4.933333333333333e-06,
+      "loss": 2.1236,
+      "step": 74
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.359375,
+      "learning_rate": 5e-06,
+      "loss": 2.2571,
+      "step": 75
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.03125,
+      "learning_rate": 5.066666666666667e-06,
+      "loss": 2.3625,
+      "step": 76
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.8515625,
+      "learning_rate": 5.133333333333333e-06,
+      "loss": 2.224,
+      "step": 77
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9609375,
+      "learning_rate": 5.2e-06,
+      "loss": 2.2386,
+      "step": 78
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.9296875,
+      "learning_rate": 5.2666666666666665e-06,
+      "loss": 2.2665,
+      "step": 79
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6328125,
+      "learning_rate": 5.333333333333333e-06,
+      "loss": 1.9822,
+      "step": 80
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.71875,
+      "learning_rate": 5.4e-06,
+      "loss": 2.1562,
+      "step": 81
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.9453125,
+      "learning_rate": 5.466666666666666e-06,
+      "loss": 2.1552,
+      "step": 82
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.859375,
+      "learning_rate": 5.5333333333333334e-06,
+      "loss": 2.0242,
+      "step": 83
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.59375,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.9113,
+      "step": 84
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 2.5,
+      "learning_rate": 5.666666666666667e-06,
+      "loss": 2.0728,
+      "step": 85
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.890625,
+      "learning_rate": 5.733333333333334e-06,
+      "loss": 2.1963,
+      "step": 86
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.8828125,
+      "learning_rate": 5.8e-06,
+      "loss": 2.3849,
+      "step": 87
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.734375,
+      "learning_rate": 5.866666666666667e-06,
+      "loss": 2.0595,
+      "step": 88
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.75,
+      "learning_rate": 5.933333333333334e-06,
+      "loss": 2.0931,
+      "step": 89
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.5,
+      "learning_rate": 6e-06,
+      "loss": 1.9384,
+      "step": 90
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.7734375,
+      "learning_rate": 5.999995296023366e-06,
+      "loss": 2.2192,
+      "step": 91
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.999981184109854e-06,
+      "loss": 2.0251,
+      "step": 92
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.5234375,
+      "learning_rate": 5.999957664308636e-06,
+      "loss": 1.9813,
+      "step": 93
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.796875,
+      "learning_rate": 5.999924736701666e-06,
+      "loss": 2.1143,
+      "step": 94
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.625,
+      "learning_rate": 5.999882401403676e-06,
+      "loss": 1.8317,
+      "step": 95
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.25,
+      "learning_rate": 5.999830658562182e-06,
+      "loss": 2.2928,
+      "step": 96
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.999769508357479e-06,
+      "loss": 2.0814,
+      "step": 97
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.6171875,
+      "learning_rate": 5.999698951002636e-06,
+      "loss": 1.9344,
+      "step": 98
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5546875,
+      "learning_rate": 5.99961898674351e-06,
+      "loss": 1.951,
+      "step": 99
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.8203125,
+      "learning_rate": 5.999529615858727e-06,
+      "loss": 2.0772,
+      "step": 100
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.59375,
+      "learning_rate": 5.999430838659697e-06,
+      "loss": 2.1198,
+      "step": 101
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.75,
+      "learning_rate": 5.999322655490598e-06,
+      "loss": 2.0501,
+      "step": 102
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.671875,
+      "learning_rate": 5.999205066728389e-06,
+      "loss": 2.1693,
+      "step": 103
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.703125,
+      "learning_rate": 5.999078072782799e-06,
+      "loss": 2.143,
+      "step": 104
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.625,
+      "learning_rate": 5.9989416740963284e-06,
+      "loss": 2.2024,
+      "step": 105
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 4.96875,
+      "learning_rate": 5.998795871144249e-06,
+      "loss": 2.1681,
+      "step": 106
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.828125,
+      "learning_rate": 5.9986406644346e-06,
+      "loss": 2.0761,
+      "step": 107
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.515625,
+      "learning_rate": 5.998476054508189e-06,
+      "loss": 2.3697,
+      "step": 108
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.8671875,
+      "learning_rate": 5.998302041938586e-06,
+      "loss": 2.2053,
+      "step": 109
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.8671875,
+      "learning_rate": 5.998118627332126e-06,
+      "loss": 2.1624,
+      "step": 110
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.997925811327904e-06,
+      "loss": 2.2316,
+      "step": 111
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.75,
+      "learning_rate": 5.997723594597772e-06,
+      "loss": 2.0206,
+      "step": 112
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.546875,
+      "learning_rate": 5.997511977846342e-06,
+      "loss": 1.9041,
+      "step": 113
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7109375,
+      "learning_rate": 5.997290961810975e-06,
+      "loss": 1.9657,
+      "step": 114
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.7109375,
+      "learning_rate": 5.997060547261785e-06,
+      "loss": 2.0243,
+      "step": 115
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9296875,
+      "learning_rate": 5.996820735001635e-06,
+      "loss": 2.0708,
+      "step": 116
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.4765625,
+      "learning_rate": 5.9965715258661355e-06,
+      "loss": 2.02,
+      "step": 117
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.6171875,
+      "learning_rate": 5.996312920723635e-06,
+      "loss": 2.118,
+      "step": 118
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.578125,
+      "learning_rate": 5.996044920475225e-06,
+      "loss": 1.9616,
+      "step": 119
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.703125,
+      "learning_rate": 5.995767526054733e-06,
+      "loss": 2.1868,
+      "step": 120
+    },
+    {
+      "epoch": 0.2,
+      "eval_loss": 2.064619541168213,
+      "eval_runtime": 72.5673,
+      "eval_samples_per_second": 8.943,
+      "eval_steps_per_second": 8.943,
+      "step": 120
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.578125,
+      "learning_rate": 5.99548073842872e-06,
+      "loss": 2.0924,
+      "step": 121
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.84375,
+      "learning_rate": 5.995184558596476e-06,
+      "loss": 2.2306,
+      "step": 122
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.9948789875900195e-06,
+      "loss": 2.0417,
+      "step": 123
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.994564026474088e-06,
+      "loss": 2.1642,
+      "step": 124
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.994239676346142e-06,
+      "loss": 2.253,
+      "step": 125
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.84375,
+      "learning_rate": 5.9939059383363556e-06,
+      "loss": 1.9898,
+      "step": 126
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.875,
+      "learning_rate": 5.993562813607616e-06,
+      "loss": 2.1938,
+      "step": 127
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.140625,
+      "learning_rate": 5.993210303355513e-06,
+      "loss": 2.1159,
+      "step": 128
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.125,
+      "learning_rate": 5.9928484088083454e-06,
+      "loss": 2.1059,
+      "step": 129
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.609375,
+      "learning_rate": 5.992477131227108e-06,
+      "loss": 2.1826,
+      "step": 130
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 3.40625,
+      "learning_rate": 5.992096471905491e-06,
+      "loss": 1.8935,
+      "step": 131
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.53125,
+      "learning_rate": 5.991706432169873e-06,
+      "loss": 2.1573,
+      "step": 132
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.9453125,
+      "learning_rate": 5.991307013379319e-06,
+      "loss": 2.0751,
+      "step": 133
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.640625,
+      "learning_rate": 5.990898216925575e-06,
+      "loss": 2.1524,
+      "step": 134
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.015625,
+      "learning_rate": 5.990480044233063e-06,
+      "loss": 2.0496,
+      "step": 135
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.5546875,
+      "learning_rate": 5.990052496758875e-06,
+      "loss": 2.0499,
+      "step": 136
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.5546875,
+      "learning_rate": 5.989615575992769e-06,
+      "loss": 1.8557,
+      "step": 137
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.84375,
+      "learning_rate": 5.989169283457163e-06,
+      "loss": 1.9044,
+      "step": 138
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.5859375,
+      "learning_rate": 5.98871362070713e-06,
+      "loss": 1.9648,
+      "step": 139
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.96875,
+      "learning_rate": 5.988248589330397e-06,
+      "loss": 2.1693,
+      "step": 140
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.987774190947329e-06,
+      "loss": 2.1912,
+      "step": 141
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.6328125,
+      "learning_rate": 5.987290427210933e-06,
+      "loss": 2.0069,
+      "step": 142
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.65625,
+      "learning_rate": 5.98679729980685e-06,
+      "loss": 1.9199,
+      "step": 143
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.7734375,
+      "learning_rate": 5.986294810453345e-06,
+      "loss": 2.2007,
+      "step": 144
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.4921875,
+      "learning_rate": 5.985782960901305e-06,
+      "loss": 1.8919,
+      "step": 145
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.3359375,
+      "learning_rate": 5.985261752934234e-06,
+      "loss": 1.9178,
+      "step": 146
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.984731188368243e-06,
+      "loss": 2.0098,
+      "step": 147
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.578125,
+      "learning_rate": 5.984191269052045e-06,
+      "loss": 2.0263,
+      "step": 148
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.5859375,
+      "learning_rate": 5.98364199686695e-06,
+      "loss": 2.1398,
+      "step": 149
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.546875,
+      "learning_rate": 5.983083373726857e-06,
+      "loss": 2.0072,
+      "step": 150
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.9453125,
+      "learning_rate": 5.982515401578246e-06,
+      "loss": 2.1237,
+      "step": 151
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.625,
+      "learning_rate": 5.981938082400176e-06,
+      "loss": 2.2744,
+      "step": 152
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.59375,
+      "learning_rate": 5.9813514182042745e-06,
+      "loss": 1.999,
+      "step": 153
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.359375,
+      "learning_rate": 5.980755411034728e-06,
+      "loss": 1.9153,
+      "step": 154
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.980150062968282e-06,
+      "loss": 1.9111,
+      "step": 155
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.53125,
+      "learning_rate": 5.979535376114227e-06,
+      "loss": 1.9874,
+      "step": 156
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.8828125,
+      "learning_rate": 5.978911352614394e-06,
+      "loss": 2.213,
+      "step": 157
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7109375,
+      "learning_rate": 5.9782779946431475e-06,
+      "loss": 2.0593,
+      "step": 158
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.5,
+      "learning_rate": 5.9776353044073774e-06,
+      "loss": 1.9598,
+      "step": 159
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.875,
+      "learning_rate": 5.9769832841464914e-06,
+      "loss": 2.3477,
+      "step": 160
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.359375,
+      "learning_rate": 5.976321936132406e-06,
+      "loss": 1.8547,
+      "step": 161
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.703125,
+      "learning_rate": 5.975651262669542e-06,
+      "loss": 2.2347,
+      "step": 162
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.703125,
+      "learning_rate": 5.97497126609481e-06,
+      "loss": 2.0735,
+      "step": 163
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.974281948777609e-06,
+      "loss": 1.8234,
+      "step": 164
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.4296875,
+      "learning_rate": 5.973583313119815e-06,
+      "loss": 1.9371,
+      "step": 165
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.59375,
+      "learning_rate": 5.972875361555773e-06,
+      "loss": 2.1144,
+      "step": 166
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.97215809655229e-06,
+      "loss": 2.0255,
+      "step": 167
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.971431520608621e-06,
+      "loss": 2.0186,
+      "step": 168
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.453125,
+      "learning_rate": 5.97069563625647e-06,
+      "loss": 1.933,
+      "step": 169
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.578125,
+      "learning_rate": 5.9699504460599696e-06,
+      "loss": 1.9893,
+      "step": 170
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.453125,
+      "learning_rate": 5.969195952615683e-06,
+      "loss": 1.9635,
+      "step": 171
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.390625,
+      "learning_rate": 5.968432158552586e-06,
+      "loss": 1.7703,
+      "step": 172
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.609375,
+      "learning_rate": 5.967659066532064e-06,
+      "loss": 1.9872,
+      "step": 173
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.9296875,
+      "learning_rate": 5.966876679247899e-06,
+      "loss": 2.2503,
+      "step": 174
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.966084999426265e-06,
+      "loss": 2.0597,
+      "step": 175
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.9652840298257104e-06,
+      "loss": 1.9385,
+      "step": 176
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6328125,
+      "learning_rate": 5.9644737732371555e-06,
+      "loss": 2.0229,
+      "step": 177
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.5703125,
+      "learning_rate": 5.9636542324838804e-06,
+      "loss": 2.0867,
+      "step": 178
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.6171875,
+      "learning_rate": 5.962825410421516e-06,
+      "loss": 2.2203,
+      "step": 179
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.609375,
+      "learning_rate": 5.9619873099380325e-06,
+      "loss": 1.8822,
+      "step": 180
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 2.0232019424438477,
+      "eval_runtime": 72.5156,
+      "eval_samples_per_second": 8.95,
+      "eval_steps_per_second": 8.95,
+      "step": 180
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.2734375,
+      "learning_rate": 5.96113993395373e-06,
+      "loss": 1.8759,
+      "step": 181
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.5078125,
+      "learning_rate": 5.960283285421226e-06,
+      "loss": 2.0248,
+      "step": 182
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.6328125,
+      "learning_rate": 5.959417367325453e-06,
+      "loss": 2.1588,
+      "step": 183
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.40625,
+      "learning_rate": 5.958542182683637e-06,
+      "loss": 2.0872,
+      "step": 184
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.4140625,
+      "learning_rate": 5.957657734545297e-06,
+      "loss": 1.8537,
+      "step": 185
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.578125,
+      "learning_rate": 5.956764025992228e-06,
+      "loss": 2.0764,
+      "step": 186
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.53125,
+      "learning_rate": 5.9558610601384905e-06,
+      "loss": 2.0138,
+      "step": 187
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.8046875,
+      "learning_rate": 5.9549488401304055e-06,
+      "loss": 2.0897,
+      "step": 188
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.5,
+      "learning_rate": 5.954027369146535e-06,
+      "loss": 2.0864,
+      "step": 189
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.6640625,
+      "learning_rate": 5.9530966503976796e-06,
+      "loss": 2.128,
+      "step": 190
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.734375,
+      "learning_rate": 5.952156687126859e-06,
+      "loss": 2.1293,
+      "step": 191
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 5.951207482609308e-06,
+      "loss": 2.0601,
+      "step": 192
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.546875,
+      "learning_rate": 5.95024904015246e-06,
+      "loss": 2.0303,
+      "step": 193
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.8125,
+      "learning_rate": 5.949281363095939e-06,
+      "loss": 2.0027,
+      "step": 194
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.5546875,
+      "learning_rate": 5.9483044548115445e-06,
+      "loss": 2.0815,
+      "step": 195
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.53125,
+      "learning_rate": 5.947318318703242e-06,
+      "loss": 2.111,
+      "step": 196
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.734375,
+      "learning_rate": 5.94632295820715e-06,
+      "loss": 2.0938,
+      "step": 197
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.34375,
+      "learning_rate": 5.945318376791531e-06,
+      "loss": 1.8695,
+      "step": 198
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.375,
+      "learning_rate": 5.944304577956776e-06,
+      "loss": 1.947,
+      "step": 199
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.546875,
+      "learning_rate": 5.9432815652353924e-06,
+      "loss": 1.9312,
+      "step": 200
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.6953125,
+      "learning_rate": 5.942249342191993e-06,
+      "loss": 2.0169,
+      "step": 201
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.5859375,
+      "learning_rate": 5.941207912423285e-06,
+      "loss": 1.9519,
+      "step": 202
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.71875,
+      "learning_rate": 5.940157279558055e-06,
+      "loss": 2.0954,
+      "step": 203
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.046875,
+      "learning_rate": 5.9390974472571535e-06,
+      "loss": 2.1871,
+      "step": 204
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.4296875,
+      "learning_rate": 5.9380284192134925e-06,
+      "loss": 1.9921,
+      "step": 205
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.703125,
+      "learning_rate": 5.936950199152021e-06,
+      "loss": 2.0208,
+      "step": 206
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.935862790829718e-06,
+      "loss": 1.9512,
+      "step": 207
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.934766198035577e-06,
+      "loss": 1.9911,
+      "step": 208
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.3671875,
+      "learning_rate": 5.933660424590599e-06,
+      "loss": 1.9075,
+      "step": 209
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.765625,
+      "learning_rate": 5.932545474347769e-06,
+      "loss": 2.1189,
+      "step": 210
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.931421351192052e-06,
+      "loss": 2.0672,
+      "step": 211
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.8359375,
+      "learning_rate": 5.9302880590403685e-06,
+      "loss": 2.1281,
+      "step": 212
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.929145601841597e-06,
+      "loss": 1.7286,
+      "step": 213
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.7421875,
+      "learning_rate": 5.927993983576545e-06,
+      "loss": 2.0305,
+      "step": 214
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.926833208257942e-06,
+      "loss": 2.1612,
+      "step": 215
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.453125,
+      "learning_rate": 5.925663279930424e-06,
+      "loss": 2.0265,
+      "step": 216
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.453125,
+      "learning_rate": 5.924484202670524e-06,
+      "loss": 1.7555,
+      "step": 217
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.923295980586648e-06,
+      "loss": 2.053,
+      "step": 218
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.734375,
+      "learning_rate": 5.922098617819071e-06,
+      "loss": 2.162,
+      "step": 219
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.7109375,
+      "learning_rate": 5.920892118539916e-06,
+      "loss": 2.1272,
+      "step": 220
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 1.375,
+      "learning_rate": 5.919676486953141e-06,
+      "loss": 1.8841,
+      "step": 221
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.40625,
+      "learning_rate": 5.918451727294526e-06,
+      "loss": 1.8209,
+      "step": 222
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.4765625,
+      "learning_rate": 5.917217843831657e-06,
+      "loss": 1.965,
+      "step": 223
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.3828125,
+      "learning_rate": 5.915974840863911e-06,
+      "loss": 2.0254,
+      "step": 224
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.4296875,
+      "learning_rate": 5.9147227227224415e-06,
+      "loss": 2.0056,
+      "step": 225
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.609375,
+      "learning_rate": 5.913461493770162e-06,
+      "loss": 2.0089,
+      "step": 226
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.703125,
+      "learning_rate": 5.912191158401734e-06,
+      "loss": 1.9855,
+      "step": 227
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.625,
+      "learning_rate": 5.910911721043549e-06,
+      "loss": 2.051,
+      "step": 228
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.578125,
+      "learning_rate": 5.90962318615371e-06,
+      "loss": 1.9122,
+      "step": 229
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.71875,
+      "learning_rate": 5.908325558222027e-06,
+      "loss": 2.1777,
+      "step": 230
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4296875,
+      "learning_rate": 5.907018841769988e-06,
+      "loss": 1.9074,
+      "step": 231
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4375,
+      "learning_rate": 5.905703041350752e-06,
+      "loss": 1.9642,
+      "step": 232
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.359375,
+      "learning_rate": 5.904378161549129e-06,
+      "loss": 1.8773,
+      "step": 233
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.671875,
+      "learning_rate": 5.903044206981566e-06,
+      "loss": 2.2381,
+      "step": 234
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.901701182296131e-06,
+      "loss": 1.9551,
+      "step": 235
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.900349092172495e-06,
+      "loss": 1.9436,
+      "step": 236
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.328125,
+      "learning_rate": 5.898987941321919e-06,
+      "loss": 1.8453,
+      "step": 237
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.5703125,
+      "learning_rate": 5.897617734487233e-06,
+      "loss": 1.9952,
+      "step": 238
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.59375,
+      "learning_rate": 5.896238476442822e-06,
+      "loss": 2.1063,
+      "step": 239
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5625,
+      "learning_rate": 5.894850171994611e-06,
+      "loss": 1.9618,
+      "step": 240
+    },
+    {
+      "epoch": 0.41,
+      "eval_loss": 2.005347967147827,
+      "eval_runtime": 72.6485,
+      "eval_samples_per_second": 8.933,
+      "eval_steps_per_second": 8.933,
+      "step": 240
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.3203125,
+      "learning_rate": 5.893452825980047e-06,
+      "loss": 1.947,
+      "step": 241
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.578125,
+      "learning_rate": 5.892046443268078e-06,
+      "loss": 2.2132,
+      "step": 242
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.890631028759143e-06,
+      "loss": 2.0863,
+      "step": 243
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.5859375,
+      "learning_rate": 5.8892065873851515e-06,
+      "loss": 2.0921,
+      "step": 244
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.234375,
+      "learning_rate": 5.887773124109465e-06,
+      "loss": 1.7609,
+      "step": 245
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.59375,
+      "learning_rate": 5.886330643926881e-06,
+      "loss": 2.0622,
+      "step": 246
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.5234375,
+      "learning_rate": 5.884879151863618e-06,
+      "loss": 2.05,
+      "step": 247
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.3125,
+      "learning_rate": 5.8834186529772945e-06,
+      "loss": 1.9644,
+      "step": 248
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.3671875,
+      "learning_rate": 5.881949152356911e-06,
+      "loss": 1.912,
+      "step": 249
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.880470655122837e-06,
+      "loss": 1.9836,
+      "step": 250
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.625,
+      "learning_rate": 5.878983166426787e-06,
+      "loss": 2.1238,
+      "step": 251
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.3046875,
+      "learning_rate": 5.8774866914518075e-06,
+      "loss": 1.9278,
+      "step": 252
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.875981235412256e-06,
+      "loss": 1.9561,
+      "step": 253
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.874466803553785e-06,
+      "loss": 1.9926,
+      "step": 254
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.5703125,
+      "learning_rate": 5.8729434011533235e-06,
+      "loss": 1.9142,
+      "step": 255
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.7890625,
+      "learning_rate": 5.871411033519052e-06,
+      "loss": 1.964,
+      "step": 256
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.625,
+      "learning_rate": 5.8698697059903975e-06,
+      "loss": 1.8374,
+      "step": 257
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.4765625,
+      "learning_rate": 5.868319423938003e-06,
+      "loss": 2.068,
+      "step": 258
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.4921875,
+      "learning_rate": 5.866760192763712e-06,
+      "loss": 2.1991,
+      "step": 259
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.5546875,
+      "learning_rate": 5.865192017900552e-06,
+      "loss": 2.0453,
+      "step": 260
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.4296875,
+      "learning_rate": 5.863614904812716e-06,
+      "loss": 1.8102,
+      "step": 261
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.59375,
+      "learning_rate": 5.862028858995538e-06,
+      "loss": 1.9502,
+      "step": 262
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.7109375,
+      "learning_rate": 5.8604338859754785e-06,
+      "loss": 2.2038,
+      "step": 263
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.858829991310106e-06,
+      "loss": 2.0411,
+      "step": 264
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.5234375,
+      "learning_rate": 5.857217180588075e-06,
+      "loss": 2.1519,
+      "step": 265
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.4921875,
+      "learning_rate": 5.855595459429105e-06,
+      "loss": 2.0955,
+      "step": 266
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.4296875,
+      "learning_rate": 5.853964833483967e-06,
+      "loss": 1.9125,
+      "step": 267
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.852325308434457e-06,
+      "loss": 2.0982,
+      "step": 268
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.4921875,
+      "learning_rate": 5.850676889993383e-06,
+      "loss": 2.0996,
+      "step": 269
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.6640625,
+      "learning_rate": 5.849019583904536e-06,
+      "loss": 2.049,
+      "step": 270
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.3828125,
+      "learning_rate": 5.84735339594268e-06,
+      "loss": 1.9399,
+      "step": 271
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.3125,
+      "learning_rate": 5.845678331913525e-06,
+      "loss": 2.0352,
+      "step": 272
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.375,
+      "learning_rate": 5.843994397653709e-06,
+      "loss": 1.9942,
+      "step": 273
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.484375,
+      "learning_rate": 5.842301599030779e-06,
+      "loss": 2.0872,
+      "step": 274
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.4375,
+      "learning_rate": 5.840599941943169e-06,
+      "loss": 1.9365,
+      "step": 275
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.4765625,
+      "learning_rate": 5.838889432320178e-06,
+      "loss": 1.9919,
+      "step": 276
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.4140625,
+      "learning_rate": 5.8371700761219525e-06,
+      "loss": 2.0302,
+      "step": 277
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.5703125,
+      "learning_rate": 5.835441879339465e-06,
+      "loss": 2.0213,
+      "step": 278
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.59375,
+      "learning_rate": 5.833704847994489e-06,
+      "loss": 1.9393,
+      "step": 279
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.375,
+      "learning_rate": 5.831958988139585e-06,
+      "loss": 1.8929,
+      "step": 280
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.6953125,
+      "learning_rate": 5.830204305858075e-06,
+      "loss": 2.0949,
+      "step": 281
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.4296875,
+      "learning_rate": 5.828440807264021e-06,
+      "loss": 1.953,
+      "step": 282
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.546875,
+      "learning_rate": 5.826668498502205e-06,
+      "loss": 1.8952,
+      "step": 283
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.4140625,
+      "learning_rate": 5.824887385748107e-06,
+      "loss": 2.1153,
+      "step": 284
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.921875,
+      "learning_rate": 5.823097475207888e-06,
+      "loss": 2.1172,
+      "step": 285
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.53125,
+      "learning_rate": 5.821298773118357e-06,
+      "loss": 1.8706,
+      "step": 286
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.6015625,
+      "learning_rate": 5.8194912857469615e-06,
+      "loss": 2.0176,
+      "step": 287
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.53125,
+      "learning_rate": 5.81767501939176e-06,
+      "loss": 2.0283,
+      "step": 288
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.359375,
+      "learning_rate": 5.815849980381401e-06,
+      "loss": 1.9713,
+      "step": 289
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.640625,
+      "learning_rate": 5.814016175075099e-06,
+      "loss": 2.022,
+      "step": 290
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.8121736098626144e-06,
+      "loss": 2.1311,
+      "step": 291
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.828125,
+      "learning_rate": 5.8103222911642335e-06,
+      "loss": 2.0839,
+      "step": 292
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.7578125,
+      "learning_rate": 5.808462225430739e-06,
+      "loss": 1.9339,
+      "step": 293
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.8065934191433955e-06,
+      "loss": 1.9463,
+      "step": 294
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.804715878813923e-06,
+      "loss": 1.9973,
+      "step": 295
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.5234375,
+      "learning_rate": 5.8028296109844736e-06,
+      "loss": 1.9792,
+      "step": 296
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.484375,
+      "learning_rate": 5.800934622227608e-06,
+      "loss": 2.005,
+      "step": 297
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.453125,
+      "learning_rate": 5.79903091914628e-06,
+      "loss": 1.9834,
+      "step": 298
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.7890625,
+      "learning_rate": 5.797118508373802e-06,
+      "loss": 2.2975,
+      "step": 299
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.3515625,
+      "learning_rate": 5.79519739657383e-06,
+      "loss": 2.0384,
+      "step": 300
+    },
+    {
+      "epoch": 0.51,
+      "eval_loss": 1.993146538734436,
+      "eval_runtime": 73.1349,
+      "eval_samples_per_second": 8.874,
+      "eval_steps_per_second": 8.874,
+      "step": 300
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.793267590440339e-06,
+      "loss": 2.041,
+      "step": 301
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.4140625,
+      "learning_rate": 5.791329096697598e-06,
+      "loss": 1.7858,
+      "step": 302
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.71875,
+      "learning_rate": 5.789381922100146e-06,
+      "loss": 2.1009,
+      "step": 303
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.453125,
+      "learning_rate": 5.787426073432772e-06,
+      "loss": 2.0599,
+      "step": 304
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.265625,
+      "learning_rate": 5.785461557510487e-06,
+      "loss": 1.8,
+      "step": 305
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.53125,
+      "learning_rate": 5.783488381178505e-06,
+      "loss": 2.0379,
+      "step": 306
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.3515625,
+      "learning_rate": 5.781506551312215e-06,
+      "loss": 1.9764,
+      "step": 307
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.546875,
+      "learning_rate": 5.779516074817158e-06,
+      "loss": 2.0071,
+      "step": 308
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.3359375,
+      "learning_rate": 5.777516958629004e-06,
+      "loss": 1.9035,
+      "step": 309
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.5703125,
+      "learning_rate": 5.77550920971353e-06,
+      "loss": 1.8642,
+      "step": 310
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.5,
+      "learning_rate": 5.773492835066588e-06,
+      "loss": 2.0698,
+      "step": 311
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.6171875,
+      "learning_rate": 5.77146784171409e-06,
+      "loss": 2.0487,
+      "step": 312
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.4140625,
+      "learning_rate": 5.769434236711978e-06,
+      "loss": 2.0178,
+      "step": 313
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.421875,
+      "learning_rate": 5.7673920271462e-06,
+      "loss": 2.0158,
+      "step": 314
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.3515625,
+      "learning_rate": 5.765341220132685e-06,
+      "loss": 1.8776,
+      "step": 315
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.390625,
+      "learning_rate": 5.763281822817322e-06,
+      "loss": 1.945,
+      "step": 316
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.5234375,
+      "learning_rate": 5.761213842375933e-06,
+      "loss": 1.9996,
+      "step": 317
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.759137286014241e-06,
+      "loss": 2.0427,
+      "step": 318
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.757052160967858e-06,
+      "loss": 1.9874,
+      "step": 319
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.2265625,
+      "learning_rate": 5.75495847450225e-06,
+      "loss": 1.7006,
+      "step": 320
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.752856233912714e-06,
+      "loss": 2.0536,
+      "step": 321
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 1.8359375,
+      "learning_rate": 5.750745446524355e-06,
+      "loss": 2.08,
+      "step": 322
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.5078125,
+      "learning_rate": 5.748626119692058e-06,
+      "loss": 1.9435,
+      "step": 323
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.671875,
+      "learning_rate": 5.746498260800462e-06,
+      "loss": 2.0348,
+      "step": 324
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7578125,
+      "learning_rate": 5.744361877263938e-06,
+      "loss": 2.2069,
+      "step": 325
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.742216976526557e-06,
+      "loss": 2.002,
+      "step": 326
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.6015625,
+      "learning_rate": 5.7400635660620685e-06,
+      "loss": 2.0927,
+      "step": 327
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.359375,
+      "learning_rate": 5.737901653373878e-06,
+      "loss": 1.6721,
+      "step": 328
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.4921875,
+      "learning_rate": 5.7357312459950085e-06,
+      "loss": 1.8958,
+      "step": 329
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.375,
+      "learning_rate": 5.733552351488089e-06,
+      "loss": 1.8287,
+      "step": 330
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.4296875,
+      "learning_rate": 5.7313649774453185e-06,
+      "loss": 1.9405,
+      "step": 331
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.4140625,
+      "learning_rate": 5.729169131488443e-06,
+      "loss": 1.8981,
+      "step": 332
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.734375,
+      "learning_rate": 5.726964821268725e-06,
+      "loss": 1.8419,
+      "step": 333
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.671875,
+      "learning_rate": 5.724752054466926e-06,
+      "loss": 1.9149,
+      "step": 334
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.722530838793268e-06,
+      "loss": 2.0117,
+      "step": 335
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.720301181987417e-06,
+      "loss": 2.0528,
+      "step": 336
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.4453125,
+      "learning_rate": 5.7180630918184485e-06,
+      "loss": 2.0948,
+      "step": 337
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.3828125,
+      "learning_rate": 5.7158165760848265e-06,
+      "loss": 1.9506,
+      "step": 338
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.6640625,
+      "learning_rate": 5.713561642614369e-06,
+      "loss": 1.8097,
+      "step": 339
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.453125,
+      "learning_rate": 5.71129829926423e-06,
+      "loss": 1.9875,
+      "step": 340
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7890625,
+      "learning_rate": 5.709026553920862e-06,
+      "loss": 2.1816,
+      "step": 341
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.706746414499997e-06,
+      "loss": 2.0024,
+      "step": 342
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.5,
+      "learning_rate": 5.704457888946614e-06,
+      "loss": 1.8431,
+      "step": 343
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.7734375,
+      "learning_rate": 5.702160985234916e-06,
+      "loss": 2.3138,
+      "step": 344
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.4140625,
+      "learning_rate": 5.6998557113682935e-06,
+      "loss": 2.0798,
+      "step": 345
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.4765625,
+      "learning_rate": 5.697542075379308e-06,
+      "loss": 1.8796,
+      "step": 346
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.328125,
+      "learning_rate": 5.695220085329653e-06,
+      "loss": 2.0439,
+      "step": 347
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.5625,
+      "learning_rate": 5.692889749310135e-06,
+      "loss": 2.0048,
+      "step": 348
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.53125,
+      "learning_rate": 5.690551075440642e-06,
+      "loss": 1.9745,
+      "step": 349
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.546875,
+      "learning_rate": 5.688204071870108e-06,
+      "loss": 2.0887,
+      "step": 350
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.685848746776501e-06,
+      "loss": 2.0145,
+      "step": 351
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.6171875,
+      "learning_rate": 5.683485108366776e-06,
+      "loss": 1.9202,
+      "step": 352
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.5625,
+      "learning_rate": 5.681113164876861e-06,
+      "loss": 2.1363,
+      "step": 353
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.678732924571618e-06,
+      "loss": 1.8549,
+      "step": 354
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.3125,
+      "learning_rate": 5.6763443957448236e-06,
+      "loss": 1.8595,
+      "step": 355
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.67394758671913e-06,
+      "loss": 1.9278,
+      "step": 356
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.3515625,
+      "learning_rate": 5.671542505846046e-06,
+      "loss": 1.973,
+      "step": 357
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.3828125,
+      "learning_rate": 5.669129161505899e-06,
+      "loss": 1.9534,
+      "step": 358
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.3828125,
+      "learning_rate": 5.666707562107812e-06,
+      "loss": 1.9176,
+      "step": 359
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.8359375,
+      "learning_rate": 5.6642777160896705e-06,
+      "loss": 1.978,
+      "step": 360
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 1.9843242168426514,
+      "eval_runtime": 72.526,
+      "eval_samples_per_second": 8.949,
+      "eval_steps_per_second": 8.949,
+      "step": 360
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.3046875,
+      "learning_rate": 5.661839631918097e-06,
+      "loss": 1.7639,
+      "step": 361
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.75,
+      "learning_rate": 5.659393318088419e-06,
+      "loss": 2.1257,
+      "step": 362
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.656938783124639e-06,
+      "loss": 2.0958,
+      "step": 363
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.4921875,
+      "learning_rate": 5.654476035579402e-06,
+      "loss": 1.9213,
+      "step": 364
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.5703125,
+      "learning_rate": 5.652005084033976e-06,
+      "loss": 2.1154,
+      "step": 365
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.421875,
+      "learning_rate": 5.64952593709821e-06,
+      "loss": 2.1574,
+      "step": 366
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.4765625,
+      "learning_rate": 5.64703860341051e-06,
+      "loss": 1.964,
+      "step": 367
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.4765625,
+      "learning_rate": 5.64454309163781e-06,
+      "loss": 1.9656,
+      "step": 368
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.046875,
+      "learning_rate": 5.64203941047554e-06,
+      "loss": 2.008,
+      "step": 369
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.3671875,
+      "learning_rate": 5.639527568647593e-06,
+      "loss": 1.9648,
+      "step": 370
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.3984375,
+      "learning_rate": 5.6370075749063e-06,
+      "loss": 1.9933,
+      "step": 371
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.5625,
+      "learning_rate": 5.634479438032394e-06,
+      "loss": 2.0599,
+      "step": 372
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.4140625,
+      "learning_rate": 5.631943166834985e-06,
+      "loss": 1.9217,
+      "step": 373
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.8203125,
+      "learning_rate": 5.629398770151526e-06,
+      "loss": 2.035,
+      "step": 374
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.40625,
+      "learning_rate": 5.626846256847779e-06,
+      "loss": 1.7794,
+      "step": 375
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.609375,
+      "learning_rate": 5.624285635817793e-06,
+      "loss": 2.104,
+      "step": 376
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.3984375,
+      "learning_rate": 5.621716915983865e-06,
+      "loss": 1.859,
+      "step": 377
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.5625,
+      "learning_rate": 5.619140106296511e-06,
+      "loss": 1.8899,
+      "step": 378
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.616555215734438e-06,
+      "loss": 1.9897,
+      "step": 379
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.6139622533045085e-06,
+      "loss": 2.2321,
+      "step": 380
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.390625,
+      "learning_rate": 5.61136122804171e-06,
+      "loss": 1.9103,
+      "step": 381
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.53125,
+      "learning_rate": 5.608752149009127e-06,
+      "loss": 1.9372,
+      "step": 382
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.125,
+      "learning_rate": 5.606135025297905e-06,
+      "loss": 2.1246,
+      "step": 383
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.4921875,
+      "learning_rate": 5.60350986602722e-06,
+      "loss": 1.9047,
+      "step": 384
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.4921875,
+      "learning_rate": 5.600876680344249e-06,
+      "loss": 1.755,
+      "step": 385
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 2.234375,
+      "learning_rate": 5.5982354774241355e-06,
+      "loss": 1.925,
+      "step": 386
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.4453125,
+      "learning_rate": 5.595586266469958e-06,
+      "loss": 1.9832,
+      "step": 387
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.5859375,
+      "learning_rate": 5.5929290567127e-06,
+      "loss": 2.0026,
+      "step": 388
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.375,
+      "learning_rate": 5.590263857411215e-06,
+      "loss": 2.0023,
+      "step": 389
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.4453125,
+      "learning_rate": 5.587590677852195e-06,
+      "loss": 1.801,
+      "step": 390
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.453125,
+      "learning_rate": 5.584909527350141e-06,
+      "loss": 1.8022,
+      "step": 391
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.582220415247325e-06,
+      "loss": 2.1502,
+      "step": 392
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.6015625,
+      "learning_rate": 5.579523350913763e-06,
+      "loss": 1.9927,
+      "step": 393
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.6015625,
+      "learning_rate": 5.576818343747179e-06,
+      "loss": 2.1792,
+      "step": 394
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.28125,
+      "learning_rate": 5.574105403172973e-06,
+      "loss": 1.832,
+      "step": 395
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.8125,
+      "learning_rate": 5.571384538644189e-06,
+      "loss": 1.9887,
+      "step": 396
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.375,
+      "learning_rate": 5.568655759641482e-06,
+      "loss": 2.0233,
+      "step": 397
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.3515625,
+      "learning_rate": 5.565919075673082e-06,
+      "loss": 1.8999,
+      "step": 398
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.3671875,
+      "learning_rate": 5.563174496274767e-06,
+      "loss": 2.0715,
+      "step": 399
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.5859375,
+      "learning_rate": 5.560422031009821e-06,
+      "loss": 2.0547,
+      "step": 400
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 2.171875,
+      "learning_rate": 5.557661689469011e-06,
+      "loss": 2.1603,
+      "step": 401
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.5,
+      "learning_rate": 5.554893481270546e-06,
+      "loss": 1.9174,
+      "step": 402
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.3359375,
+      "learning_rate": 5.552117416060045e-06,
+      "loss": 1.8991,
+      "step": 403
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.3515625,
+      "learning_rate": 5.549333503510504e-06,
+      "loss": 1.8692,
+      "step": 404
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.640625,
+      "learning_rate": 5.546541753322264e-06,
+      "loss": 1.9568,
+      "step": 405
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.578125,
+      "learning_rate": 5.543742175222977e-06,
+      "loss": 2.1146,
+      "step": 406
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.4453125,
+      "learning_rate": 5.540934778967567e-06,
+      "loss": 1.9401,
+      "step": 407
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.3828125,
+      "learning_rate": 5.538119574338203e-06,
+      "loss": 1.878,
+      "step": 408
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.609375,
+      "learning_rate": 5.535296571144259e-06,
+      "loss": 2.0552,
+      "step": 409
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.532465779222283e-06,
+      "loss": 2.1755,
+      "step": 410
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.3671875,
+      "learning_rate": 5.529627208435967e-06,
+      "loss": 1.9412,
+      "step": 411
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.4453125,
+      "learning_rate": 5.5267808686761e-06,
+      "loss": 1.9386,
+      "step": 412
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.421875,
+      "learning_rate": 5.523926769860549e-06,
+      "loss": 1.9339,
+      "step": 413
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.52106492193421e-06,
+      "loss": 1.8639,
+      "step": 414
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.5181953348689865e-06,
+      "loss": 2.0624,
+      "step": 415
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.296875,
+      "learning_rate": 5.515318018663744e-06,
+      "loss": 1.8617,
+      "step": 416
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.5078125,
+      "learning_rate": 5.512432983344281e-06,
+      "loss": 2.2311,
+      "step": 417
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.578125,
+      "learning_rate": 5.509540238963293e-06,
+      "loss": 1.9526,
+      "step": 418
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.421875,
+      "learning_rate": 5.506639795600338e-06,
+      "loss": 2.0589,
+      "step": 419
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.390625,
+      "learning_rate": 5.5037316633618e-06,
+      "loss": 1.8341,
+      "step": 420
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 1.9771074056625366,
+      "eval_runtime": 72.6606,
+      "eval_samples_per_second": 8.932,
+      "eval_steps_per_second": 8.932,
+      "step": 420
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.50081585238085e-06,
+      "loss": 1.8993,
+      "step": 421
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.4375,
+      "learning_rate": 5.497892372817423e-06,
+      "loss": 1.9337,
+      "step": 422
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.484375,
+      "learning_rate": 5.494961234858169e-06,
+      "loss": 2.0068,
+      "step": 423
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.4296875,
+      "learning_rate": 5.492022448716425e-06,
+      "loss": 2.0354,
+      "step": 424
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.484375,
+      "learning_rate": 5.489076024632178e-06,
+      "loss": 1.8277,
+      "step": 425
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 2.265625,
+      "learning_rate": 5.486121972872026e-06,
+      "loss": 1.8635,
+      "step": 426
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.5234375,
+      "learning_rate": 5.483160303729148e-06,
+      "loss": 2.0839,
+      "step": 427
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.480191027523265e-06,
+      "loss": 2.0322,
+      "step": 428
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.6015625,
+      "learning_rate": 5.477214154600602e-06,
+      "loss": 1.9341,
+      "step": 429
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.474229695333859e-06,
+      "loss": 2.0939,
+      "step": 430
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.6015625,
+      "learning_rate": 5.4712376601221625e-06,
+      "loss": 1.9471,
+      "step": 431
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.5859375,
+      "learning_rate": 5.468238059391045e-06,
+      "loss": 2.1356,
+      "step": 432
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.671875,
+      "learning_rate": 5.465230903592394e-06,
+      "loss": 2.104,
+      "step": 433
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.462216203204427e-06,
+      "loss": 1.8952,
+      "step": 434
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.3671875,
+      "learning_rate": 5.459193968731645e-06,
+      "loss": 1.9731,
+      "step": 435
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.4296875,
+      "learning_rate": 5.456164210704808e-06,
+      "loss": 1.9873,
+      "step": 436
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.484375,
+      "learning_rate": 5.453126939680882e-06,
+      "loss": 1.9,
+      "step": 437
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.3984375,
+      "learning_rate": 5.45008216624302e-06,
+      "loss": 1.8434,
+      "step": 438
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.5234375,
+      "learning_rate": 5.447029901000513e-06,
+      "loss": 1.7828,
+      "step": 439
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.6328125,
+      "learning_rate": 5.443970154588756e-06,
+      "loss": 2.045,
+      "step": 440
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.6171875,
+      "learning_rate": 5.440902937669213e-06,
+      "loss": 2.0628,
+      "step": 441
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.8046875,
+      "learning_rate": 5.4378282609293775e-06,
+      "loss": 2.138,
+      "step": 442
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.434746135082736e-06,
+      "loss": 1.8824,
+      "step": 443
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.431656570868731e-06,
+      "loss": 2.0143,
+      "step": 444
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.5,
+      "learning_rate": 5.428559579052726e-06,
+      "loss": 2.0449,
+      "step": 445
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.296875,
+      "learning_rate": 5.425455170425963e-06,
+      "loss": 1.734,
+      "step": 446
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.2578125,
+      "learning_rate": 5.422343355805526e-06,
+      "loss": 1.7411,
+      "step": 447
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.419224146034307e-06,
+      "loss": 1.9042,
+      "step": 448
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.546875,
+      "learning_rate": 5.4160975519809665e-06,
+      "loss": 2.1657,
+      "step": 449
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.59375,
+      "learning_rate": 5.412963584539892e-06,
+      "loss": 2.1677,
+      "step": 450
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.359375,
+      "learning_rate": 5.409822254631165e-06,
+      "loss": 1.6379,
+      "step": 451
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.453125,
+      "learning_rate": 5.40667357320052e-06,
+      "loss": 1.9644,
+      "step": 452
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.5625,
+      "learning_rate": 5.403517551219309e-06,
+      "loss": 1.9929,
+      "step": 453
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.640625,
+      "learning_rate": 5.400354199684463e-06,
+      "loss": 2.0832,
+      "step": 454
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.6328125,
+      "learning_rate": 5.3971835296184455e-06,
+      "loss": 1.9306,
+      "step": 455
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.671875,
+      "learning_rate": 5.394005552069228e-06,
+      "loss": 2.2007,
+      "step": 456
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.40625,
+      "learning_rate": 5.39082027811024e-06,
+      "loss": 1.881,
+      "step": 457
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.625,
+      "learning_rate": 5.387627718840339e-06,
+      "loss": 2.0157,
+      "step": 458
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.359375,
+      "learning_rate": 5.3844278853837626e-06,
+      "loss": 1.9403,
+      "step": 459
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.453125,
+      "learning_rate": 5.3812207888900985e-06,
+      "loss": 2.1445,
+      "step": 460
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.796875,
+      "learning_rate": 5.378006440534241e-06,
+      "loss": 1.9988,
+      "step": 461
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.734375,
+      "learning_rate": 5.374784851516352e-06,
+      "loss": 2.1576,
+      "step": 462
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.4765625,
+      "learning_rate": 5.3715560330618245e-06,
+      "loss": 1.9325,
+      "step": 463
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.3515625,
+      "learning_rate": 5.3683199964212405e-06,
+      "loss": 1.9047,
+      "step": 464
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.484375,
+      "learning_rate": 5.365076752870335e-06,
+      "loss": 2.0143,
+      "step": 465
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.361826313709952e-06,
+      "loss": 1.8602,
+      "step": 466
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.358568690266011e-06,
+      "loss": 2.1187,
+      "step": 467
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.5234375,
+      "learning_rate": 5.355303893889463e-06,
+      "loss": 1.9371,
+      "step": 468
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.453125,
+      "learning_rate": 5.3520319359562545e-06,
+      "loss": 1.8244,
+      "step": 469
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.3046875,
+      "learning_rate": 5.348752827867283e-06,
+      "loss": 1.8302,
+      "step": 470
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.4921875,
+      "learning_rate": 5.345466581048362e-06,
+      "loss": 1.8959,
+      "step": 471
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.484375,
+      "learning_rate": 5.34217320695018e-06,
+      "loss": 1.8984,
+      "step": 472
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.338872717048259e-06,
+      "loss": 1.9635,
+      "step": 473
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.6171875,
+      "learning_rate": 5.3355651228429135e-06,
+      "loss": 1.8611,
+      "step": 474
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.59375,
+      "learning_rate": 5.332250435859216e-06,
+      "loss": 2.0737,
+      "step": 475
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.3289286676469534e-06,
+      "loss": 1.8734,
+      "step": 476
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.5078125,
+      "learning_rate": 5.325599829780581e-06,
+      "loss": 1.9806,
+      "step": 477
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.453125,
+      "learning_rate": 5.322263933859196e-06,
+      "loss": 1.9084,
+      "step": 478
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.609375,
+      "learning_rate": 5.318920991506481e-06,
+      "loss": 2.115,
+      "step": 479
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.315571014370679e-06,
+      "loss": 1.7521,
+      "step": 480
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 1.9717546701431274,
+      "eval_runtime": 73.075,
+      "eval_samples_per_second": 8.881,
+      "eval_steps_per_second": 8.881,
+      "step": 480
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.6328125,
+      "learning_rate": 5.312214014124538e-06,
+      "loss": 2.1913,
+      "step": 481
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.5859375,
+      "learning_rate": 5.308850002465282e-06,
+      "loss": 2.0011,
+      "step": 482
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.305478991114565e-06,
+      "loss": 2.1085,
+      "step": 483
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.3984375,
+      "learning_rate": 5.302100991818428e-06,
+      "loss": 1.83,
+      "step": 484
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.59375,
+      "learning_rate": 5.298716016347267e-06,
+      "loss": 2.2321,
+      "step": 485
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.4921875,
+      "learning_rate": 5.29532407649578e-06,
+      "loss": 1.9749,
+      "step": 486
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.3984375,
+      "learning_rate": 5.291925184082935e-06,
+      "loss": 1.8921,
+      "step": 487
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.40625,
+      "learning_rate": 5.288519350951925e-06,
+      "loss": 1.9127,
+      "step": 488
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.4375,
+      "learning_rate": 5.2851065889701265e-06,
+      "loss": 1.8932,
+      "step": 489
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.6953125,
+      "learning_rate": 5.281686910029061e-06,
+      "loss": 2.2113,
+      "step": 490
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.27826032604435e-06,
+      "loss": 1.9799,
+      "step": 491
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.5,
+      "learning_rate": 5.274826848955677e-06,
+      "loss": 1.9245,
+      "step": 492
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.578125,
+      "learning_rate": 5.271386490726741e-06,
+      "loss": 1.8607,
+      "step": 493
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.296875,
+      "learning_rate": 5.26793926334522e-06,
+      "loss": 1.8167,
+      "step": 494
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.0625,
+      "learning_rate": 5.264485178822728e-06,
+      "loss": 2.165,
+      "step": 495
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 2.125,
+      "learning_rate": 5.261024249194769e-06,
+      "loss": 1.8488,
+      "step": 496
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.2575564865207e-06,
+      "loss": 1.8926,
+      "step": 497
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.3984375,
+      "learning_rate": 5.254081902883689e-06,
+      "loss": 1.9392,
+      "step": 498
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.250600510390669e-06,
+      "loss": 2.1148,
+      "step": 499
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.247112321172299e-06,
+      "loss": 2.0925,
+      "step": 500
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.243617347382919e-06,
+      "loss": 1.898,
+      "step": 501
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.546875,
+      "learning_rate": 5.240115601200511e-06,
+      "loss": 2.0985,
+      "step": 502
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.9609375,
+      "learning_rate": 5.236607094826653e-06,
+      "loss": 1.9642,
+      "step": 503
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.2330918404864815e-06,
+      "loss": 1.8557,
+      "step": 504
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.546875,
+      "learning_rate": 5.229569850428646e-06,
+      "loss": 1.8097,
+      "step": 505
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.22604113692526e-06,
+      "loss": 1.9262,
+      "step": 506
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.5859375,
+      "learning_rate": 5.22250571227187e-06,
+      "loss": 2.0331,
+      "step": 507
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.421875,
+      "learning_rate": 5.218963588787407e-06,
+      "loss": 1.7557,
+      "step": 508
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.3984375,
+      "learning_rate": 5.215414778814143e-06,
+      "loss": 2.0086,
+      "step": 509
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.4375,
+      "learning_rate": 5.211859294717644e-06,
+      "loss": 1.9163,
+      "step": 510
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.2082971488867385e-06,
+      "loss": 2.1011,
+      "step": 511
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.453125,
+      "learning_rate": 5.204728353733462e-06,
+      "loss": 1.9178,
+      "step": 512
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.375,
+      "learning_rate": 5.201152921693024e-06,
+      "loss": 1.9985,
+      "step": 513
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.375,
+      "learning_rate": 5.197570865223754e-06,
+      "loss": 1.8964,
+      "step": 514
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.40625,
+      "learning_rate": 5.193982196807067e-06,
+      "loss": 1.9792,
+      "step": 515
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.2890625,
+      "learning_rate": 5.190386928947419e-06,
+      "loss": 1.7589,
+      "step": 516
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.578125,
+      "learning_rate": 5.186785074172257e-06,
+      "loss": 2.0091,
+      "step": 517
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.183176645031982e-06,
+      "loss": 2.1835,
+      "step": 518
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.3046875,
+      "learning_rate": 5.179561654099903e-06,
+      "loss": 1.8255,
+      "step": 519
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.234375,
+      "learning_rate": 5.1759401139721905e-06,
+      "loss": 1.8562,
+      "step": 520
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.5859375,
+      "learning_rate": 5.1723120372678385e-06,
+      "loss": 1.9574,
+      "step": 521
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.3671875,
+      "learning_rate": 5.168677436628616e-06,
+      "loss": 1.8051,
+      "step": 522
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.421875,
+      "learning_rate": 5.165036324719025e-06,
+      "loss": 1.9382,
+      "step": 523
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.4453125,
+      "learning_rate": 5.161388714226253e-06,
+      "loss": 1.7465,
+      "step": 524
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.5234375,
+      "learning_rate": 5.157734617860134e-06,
+      "loss": 1.8892,
+      "step": 525
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.5703125,
+      "learning_rate": 5.1540740483531e-06,
+      "loss": 1.994,
+      "step": 526
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.150407018460139e-06,
+      "loss": 1.8736,
+      "step": 527
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.4375,
+      "learning_rate": 5.146733540958751e-06,
+      "loss": 1.876,
+      "step": 528
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.421875,
+      "learning_rate": 5.143053628648897e-06,
+      "loss": 1.9668,
+      "step": 529
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.3671875,
+      "learning_rate": 5.139367294352966e-06,
+      "loss": 2.0875,
+      "step": 530
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.3984375,
+      "learning_rate": 5.135674550915721e-06,
+      "loss": 1.8277,
+      "step": 531
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.5078125,
+      "learning_rate": 5.131975411204257e-06,
+      "loss": 1.9919,
+      "step": 532
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.3203125,
+      "learning_rate": 5.1282698881079574e-06,
+      "loss": 1.8748,
+      "step": 533
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.3671875,
+      "learning_rate": 5.124557994538446e-06,
+      "loss": 1.6749,
+      "step": 534
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.5625,
+      "learning_rate": 5.1208397434295465e-06,
+      "loss": 2.0019,
+      "step": 535
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.4765625,
+      "learning_rate": 5.117115147737234e-06,
+      "loss": 2.0223,
+      "step": 536
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.2890625,
+      "learning_rate": 5.1133842204395914e-06,
+      "loss": 1.9228,
+      "step": 537
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.390625,
+      "learning_rate": 5.109646974536761e-06,
+      "loss": 2.0,
+      "step": 538
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.359375,
+      "learning_rate": 5.105903423050907e-06,
+      "loss": 1.9856,
+      "step": 539
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.53125,
+      "learning_rate": 5.102153579026159e-06,
+      "loss": 1.9863,
+      "step": 540
+    },
+    {
+      "epoch": 0.91,
+      "eval_loss": 1.9678696393966675,
+      "eval_runtime": 72.6404,
+      "eval_samples_per_second": 8.934,
+      "eval_steps_per_second": 8.934,
+      "step": 540
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.484375,
+      "learning_rate": 5.098397455528576e-06,
+      "loss": 1.9114,
+      "step": 541
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.094635065646096e-06,
+      "loss": 2.1427,
+      "step": 542
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.6015625,
+      "learning_rate": 5.090866422488495e-06,
+      "loss": 2.0377,
+      "step": 543
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.9140625,
+      "learning_rate": 5.087091539187333e-06,
+      "loss": 2.0124,
+      "step": 544
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.3515625,
+      "learning_rate": 5.083310428895916e-06,
+      "loss": 1.878,
+      "step": 545
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.28125,
+      "learning_rate": 5.079523104789248e-06,
+      "loss": 1.6719,
+      "step": 546
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.40625,
+      "learning_rate": 5.0757295800639846e-06,
+      "loss": 1.8464,
+      "step": 547
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.484375,
+      "learning_rate": 5.071929867938385e-06,
+      "loss": 2.02,
+      "step": 548
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.4765625,
+      "learning_rate": 5.06812398165227e-06,
+      "loss": 2.1547,
+      "step": 549
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.3671875,
+      "learning_rate": 5.064311934466972e-06,
+      "loss": 1.8403,
+      "step": 550
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.8984375,
+      "learning_rate": 5.060493739665293e-06,
+      "loss": 1.9878,
+      "step": 551
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.4609375,
+      "learning_rate": 5.056669410551454e-06,
+      "loss": 1.9279,
+      "step": 552
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.05283896045105e-06,
+      "loss": 2.2803,
+      "step": 553
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.049002402711008e-06,
+      "loss": 1.9853,
+      "step": 554
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.4296875,
+      "learning_rate": 5.045159750699532e-06,
+      "loss": 2.0884,
+      "step": 555
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.3984375,
+      "learning_rate": 5.041311017806063e-06,
+      "loss": 1.9208,
+      "step": 556
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.7421875,
+      "learning_rate": 5.037456217441231e-06,
+      "loss": 2.0662,
+      "step": 557
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.3984375,
+      "learning_rate": 5.033595363036805e-06,
+      "loss": 1.9996,
+      "step": 558
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.029728468045654e-06,
+      "loss": 1.8858,
+      "step": 559
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.4375,
+      "learning_rate": 5.0258555459416885e-06,
+      "loss": 1.8735,
+      "step": 560
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.4296875,
+      "learning_rate": 5.021976610219827e-06,
+      "loss": 1.8138,
+      "step": 561
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.6015625,
+      "learning_rate": 5.018091674395937e-06,
+      "loss": 2.1505,
+      "step": 562
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.9609375,
+      "learning_rate": 5.014200752006792e-06,
+      "loss": 1.8376,
+      "step": 563
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.53125,
+      "learning_rate": 5.0103038566100295e-06,
+      "loss": 2.1275,
+      "step": 564
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.5078125,
+      "learning_rate": 5.006401001784099e-06,
+      "loss": 1.9623,
+      "step": 565
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.4765625,
+      "learning_rate": 5.002492201128211e-06,
+      "loss": 1.7547,
+      "step": 566
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.5,
+      "learning_rate": 4.998577468262297e-06,
+      "loss": 2.0941,
+      "step": 567
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.3828125,
+      "learning_rate": 4.9946568168269615e-06,
+      "loss": 1.9377,
+      "step": 568
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.990730260483425e-06,
+      "loss": 2.0837,
+      "step": 569
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.5,
+      "learning_rate": 4.98679781291349e-06,
+      "loss": 1.8037,
+      "step": 570
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.484375,
+      "learning_rate": 4.98285948781948e-06,
+      "loss": 2.0301,
+      "step": 571
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.3515625,
+      "learning_rate": 4.9789152989242045e-06,
+      "loss": 1.999,
+      "step": 572
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.3984375,
+      "learning_rate": 4.974965259970902e-06,
+      "loss": 1.9322,
+      "step": 573
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.9710093847231955e-06,
+      "loss": 1.8256,
+      "step": 574
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.967047686965043e-06,
+      "loss": 2.0594,
+      "step": 575
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.4765625,
+      "learning_rate": 4.963080180500692e-06,
+      "loss": 2.1973,
+      "step": 576
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.4921875,
+      "learning_rate": 4.95910687915463e-06,
+      "loss": 2.0154,
+      "step": 577
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.28125,
+      "learning_rate": 4.9551277967715376e-06,
+      "loss": 1.9276,
+      "step": 578
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.25,
+      "learning_rate": 4.951142947216235e-06,
+      "loss": 1.8117,
+      "step": 579
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.515625,
+      "learning_rate": 4.9471523443736395e-06,
+      "loss": 1.9427,
+      "step": 580
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.453125,
+      "learning_rate": 4.943156002148719e-06,
+      "loss": 1.9366,
+      "step": 581
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.5,
+      "learning_rate": 4.939153934466435e-06,
+      "loss": 1.9254,
+      "step": 582
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.4375,
+      "learning_rate": 4.9351461552717004e-06,
+      "loss": 1.9774,
+      "step": 583
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.4140625,
+      "learning_rate": 4.9311326785293315e-06,
+      "loss": 1.8385,
+      "step": 584
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.3515625,
+      "learning_rate": 4.927113518223993e-06,
+      "loss": 1.9024,
+      "step": 585
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.2578125,
+      "learning_rate": 4.923088688360158e-06,
+      "loss": 1.8878,
+      "step": 586
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.919058202962052e-06,
+      "loss": 1.8002,
+      "step": 587
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.3515625,
+      "learning_rate": 4.915022076073608e-06,
+      "loss": 1.9392,
+      "step": 588
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.5,
+      "learning_rate": 4.910980321758417e-06,
+      "loss": 2.0364,
+      "step": 589
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.328125,
+      "learning_rate": 4.906932954099677e-06,
+      "loss": 1.8918,
+      "step": 590
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.625,
+      "learning_rate": 4.902879987200145e-06,
+      "loss": 2.2092,
+      "step": 591
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.5,
+      "learning_rate": 4.898821435182088e-06,
+      "loss": 1.985,
+      "step": 592
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.40625,
+      "learning_rate": 4.894757312187237e-06,
+      "loss": 1.9997,
+      "step": 593
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.5546875,
+      "learning_rate": 4.890687632376731e-06,
+      "loss": 1.9681,
+      "step": 594
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.886612409931073e-06,
+      "loss": 1.9677,
+      "step": 595
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.882531659050079e-06,
+      "loss": 1.9783,
+      "step": 596
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.4140625,
+      "learning_rate": 4.8784453939528295e-06,
+      "loss": 2.1565,
+      "step": 597
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.8743536288776144e-06,
+      "loss": 2.0744,
+      "step": 598
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.328125,
+      "learning_rate": 4.8702563780818926e-06,
+      "loss": 1.9511,
+      "step": 599
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.515625,
+      "learning_rate": 4.8661536558422356e-06,
+      "loss": 1.7847,
+      "step": 600
+    },
+    {
+      "epoch": 1.01,
+      "eval_loss": 1.9644558429718018,
+      "eval_runtime": 72.8255,
+      "eval_samples_per_second": 8.912,
+      "eval_steps_per_second": 8.912,
+      "step": 600
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.546875,
+      "learning_rate": 4.86204547645428e-06,
+      "loss": 2.1285,
+      "step": 601
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.5078125,
+      "learning_rate": 4.857931854232679e-06,
+      "loss": 1.7975,
+      "step": 602
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.853812803511047e-06,
+      "loss": 1.8594,
+      "step": 603
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.28125,
+      "learning_rate": 4.849688338641919e-06,
+      "loss": 1.6401,
+      "step": 604
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.84555847399669e-06,
+      "loss": 1.8446,
+      "step": 605
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.546875,
+      "learning_rate": 4.8414232239655735e-06,
+      "loss": 1.8468,
+      "step": 606
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.5859375,
+      "learning_rate": 4.837282602957547e-06,
+      "loss": 1.922,
+      "step": 607
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.421875,
+      "learning_rate": 4.833136625400302e-06,
+      "loss": 2.0706,
+      "step": 608
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.828985305740195e-06,
+      "loss": 1.9775,
+      "step": 609
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.28125,
+      "learning_rate": 4.824828658442199e-06,
+      "loss": 1.9182,
+      "step": 610
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.515625,
+      "learning_rate": 4.820666697989845e-06,
+      "loss": 2.1004,
+      "step": 611
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.4375,
+      "learning_rate": 4.816499438885184e-06,
+      "loss": 1.8518,
+      "step": 612
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.421875,
+      "learning_rate": 4.812326895648727e-06,
+      "loss": 2.0677,
+      "step": 613
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.808149082819394e-06,
+      "loss": 1.9105,
+      "step": 614
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.803966014954471e-06,
+      "loss": 1.9669,
+      "step": 615
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.46875,
+      "learning_rate": 4.799777706629555e-06,
+      "loss": 1.9302,
+      "step": 616
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.3046875,
+      "learning_rate": 4.795584172438498e-06,
+      "loss": 1.7689,
+      "step": 617
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.4140625,
+      "learning_rate": 4.791385426993366e-06,
+      "loss": 1.8315,
+      "step": 618
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.787181484924381e-06,
+      "loss": 1.8672,
+      "step": 619
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.7829723608798736e-06,
+      "loss": 2.0861,
+      "step": 620
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.3828125,
+      "learning_rate": 4.7787580695262295e-06,
+      "loss": 1.9532,
+      "step": 621
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.4375,
+      "learning_rate": 4.7745386255478385e-06,
+      "loss": 1.7258,
+      "step": 622
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.546875,
+      "learning_rate": 4.7703140436470486e-06,
+      "loss": 1.7769,
+      "step": 623
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.296875,
+      "learning_rate": 4.766084338544105e-06,
+      "loss": 2.0005,
+      "step": 624
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.761849524977107e-06,
+      "loss": 1.9948,
+      "step": 625
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.4296875,
+      "learning_rate": 4.757609617701957e-06,
+      "loss": 1.9489,
+      "step": 626
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.375,
+      "learning_rate": 4.753364631492299e-06,
+      "loss": 1.863,
+      "step": 627
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.74911458113948e-06,
+      "loss": 1.947,
+      "step": 628
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.744859481452491e-06,
+      "loss": 1.9949,
+      "step": 629
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.3828125,
+      "learning_rate": 4.740599347257917e-06,
+      "loss": 1.9907,
+      "step": 630
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.40625,
+      "learning_rate": 4.736334193399884e-06,
+      "loss": 2.0075,
+      "step": 631
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.625,
+      "learning_rate": 4.73206403474001e-06,
+      "loss": 1.7911,
+      "step": 632
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.3203125,
+      "learning_rate": 4.7277888861573546e-06,
+      "loss": 1.8767,
+      "step": 633
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.484375,
+      "learning_rate": 4.723508762548356e-06,
+      "loss": 1.8936,
+      "step": 634
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.28125,
+      "learning_rate": 4.719223678826798e-06,
+      "loss": 2.052,
+      "step": 635
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.714933649923741e-06,
+      "loss": 2.0797,
+      "step": 636
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.3515625,
+      "learning_rate": 4.710638690787478e-06,
+      "loss": 1.9925,
+      "step": 637
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.484375,
+      "learning_rate": 4.70633881638348e-06,
+      "loss": 1.9232,
+      "step": 638
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.2109375,
+      "learning_rate": 4.702034041694348e-06,
+      "loss": 1.8591,
+      "step": 639
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.6977243817197546e-06,
+      "loss": 1.723,
+      "step": 640
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.5546875,
+      "learning_rate": 4.693409851476397e-06,
+      "loss": 2.0273,
+      "step": 641
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.6890904659979405e-06,
+      "loss": 1.9799,
+      "step": 642
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.515625,
+      "learning_rate": 4.6847662403349685e-06,
+      "loss": 1.9064,
+      "step": 643
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.5859375,
+      "learning_rate": 4.680437189554931e-06,
+      "loss": 1.8878,
+      "step": 644
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.546875,
+      "learning_rate": 4.676103328742091e-06,
+      "loss": 2.0737,
+      "step": 645
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 2.0,
+      "learning_rate": 4.671764672997469e-06,
+      "loss": 1.7712,
+      "step": 646
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.5390625,
+      "learning_rate": 4.6674212374387955e-06,
+      "loss": 1.7269,
+      "step": 647
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.53125,
+      "learning_rate": 4.663073037200457e-06,
+      "loss": 2.0273,
+      "step": 648
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.546875,
+      "learning_rate": 4.658720087433437e-06,
+      "loss": 2.0479,
+      "step": 649
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.654362403305274e-06,
+      "loss": 1.9415,
+      "step": 650
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.546875,
+      "learning_rate": 4.65e-06,
+      "loss": 1.9311,
+      "step": 651
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.5859375,
+      "learning_rate": 4.645632892718092e-06,
+      "loss": 1.8846,
+      "step": 652
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.3984375,
+      "learning_rate": 4.641261096676417e-06,
+      "loss": 1.9105,
+      "step": 653
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.636884627108181e-06,
+      "loss": 1.7741,
+      "step": 654
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.375,
+      "learning_rate": 4.63250349926287e-06,
+      "loss": 2.0834,
+      "step": 655
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.628117728406209e-06,
+      "loss": 1.9291,
+      "step": 656
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.46875,
+      "learning_rate": 4.623727329820092e-06,
+      "loss": 1.9338,
+      "step": 657
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.619332318802547e-06,
+      "loss": 1.9905,
+      "step": 658
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.5859375,
+      "learning_rate": 4.6149327106676686e-06,
+      "loss": 2.0903,
+      "step": 659
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.61052852074557e-06,
+      "loss": 1.9864,
+      "step": 660
+    },
+    {
+      "epoch": 1.11,
+      "eval_loss": 1.9622352123260498,
+      "eval_runtime": 72.6996,
+      "eval_samples_per_second": 8.927,
+      "eval_steps_per_second": 8.927,
+      "step": 660
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.53125,
+      "learning_rate": 4.606119764382332e-06,
+      "loss": 2.0866,
+      "step": 661
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.5,
+      "learning_rate": 4.601706456939942e-06,
+      "loss": 1.8834,
+      "step": 662
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.5078125,
+      "learning_rate": 4.597288613796251e-06,
+      "loss": 2.0631,
+      "step": 663
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.592866250344909e-06,
+      "loss": 1.9722,
+      "step": 664
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.588439381995321e-06,
+      "loss": 1.9687,
+      "step": 665
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.53125,
+      "learning_rate": 4.584008024172588e-06,
+      "loss": 1.9895,
+      "step": 666
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.57957219231745e-06,
+      "loss": 1.8916,
+      "step": 667
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.4375,
+      "learning_rate": 4.575131901886244e-06,
+      "loss": 1.9296,
+      "step": 668
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.5078125,
+      "learning_rate": 4.570687168350835e-06,
+      "loss": 1.9725,
+      "step": 669
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.7265625,
+      "learning_rate": 4.566238007198575e-06,
+      "loss": 1.9202,
+      "step": 670
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.5390625,
+      "learning_rate": 4.561784433932243e-06,
+      "loss": 1.9359,
+      "step": 671
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.557326464069988e-06,
+      "loss": 2.1014,
+      "step": 672
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.552864113145281e-06,
+      "loss": 1.9649,
+      "step": 673
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.548397396706861e-06,
+      "loss": 1.7436,
+      "step": 674
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.203125,
+      "learning_rate": 4.543926330318675e-06,
+      "loss": 1.7933,
+      "step": 675
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.4765625,
+      "learning_rate": 4.5394509295598275e-06,
+      "loss": 1.8842,
+      "step": 676
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.515625,
+      "learning_rate": 4.534971210024528e-06,
+      "loss": 1.9636,
+      "step": 677
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.530487187322031e-06,
+      "loss": 1.9325,
+      "step": 678
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.525998877076587e-06,
+      "loss": 2.1198,
+      "step": 679
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.521506294927387e-06,
+      "loss": 1.8449,
+      "step": 680
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.3125,
+      "learning_rate": 4.517009456528504e-06,
+      "loss": 1.7361,
+      "step": 681
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.512508377548847e-06,
+      "loss": 1.9924,
+      "step": 682
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.5080030736720956e-06,
+      "loss": 1.888,
+      "step": 683
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.503493560596654e-06,
+      "loss": 1.809,
+      "step": 684
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.421875,
+      "learning_rate": 4.49897985403559e-06,
+      "loss": 1.8108,
+      "step": 685
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.7109375,
+      "learning_rate": 4.494461969716588e-06,
+      "loss": 2.2171,
+      "step": 686
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.2578125,
+      "learning_rate": 4.489939923381884e-06,
+      "loss": 1.7997,
+      "step": 687
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.46875,
+      "learning_rate": 4.485413730788222e-06,
+      "loss": 1.9431,
+      "step": 688
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.4808834077067865e-06,
+      "loss": 1.7227,
+      "step": 689
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.476348969923162e-06,
+      "loss": 1.96,
+      "step": 690
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.484375,
+      "learning_rate": 4.471810433237263e-06,
+      "loss": 1.9317,
+      "step": 691
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.4296875,
+      "learning_rate": 4.467267813463294e-06,
+      "loss": 1.933,
+      "step": 692
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.462721126429678e-06,
+      "loss": 1.9302,
+      "step": 693
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.625,
+      "learning_rate": 4.458170387979017e-06,
+      "loss": 2.0257,
+      "step": 694
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.4140625,
+      "learning_rate": 4.453615613968028e-06,
+      "loss": 2.0193,
+      "step": 695
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.3203125,
+      "learning_rate": 4.44905682026749e-06,
+      "loss": 1.9601,
+      "step": 696
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.444494022762184e-06,
+      "loss": 2.0479,
+      "step": 697
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.6953125,
+      "learning_rate": 4.43992723735085e-06,
+      "loss": 1.9051,
+      "step": 698
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.5,
+      "learning_rate": 4.435356479946116e-06,
+      "loss": 1.8617,
+      "step": 699
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.40625,
+      "learning_rate": 4.430781766474452e-06,
+      "loss": 1.8091,
+      "step": 700
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.3046875,
+      "learning_rate": 4.426203112876117e-06,
+      "loss": 2.005,
+      "step": 701
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.5234375,
+      "learning_rate": 4.421620535105094e-06,
+      "loss": 1.9742,
+      "step": 702
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.40625,
+      "learning_rate": 4.417034049129042e-06,
+      "loss": 1.856,
+      "step": 703
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.5234375,
+      "learning_rate": 4.4124436709292355e-06,
+      "loss": 2.0985,
+      "step": 704
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.407849416500515e-06,
+      "loss": 1.9179,
+      "step": 705
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.5859375,
+      "learning_rate": 4.403251301851225e-06,
+      "loss": 1.7981,
+      "step": 706
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.398649343003162e-06,
+      "loss": 1.9737,
+      "step": 707
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.7734375,
+      "learning_rate": 4.394043555991512e-06,
+      "loss": 1.782,
+      "step": 708
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.3894339568648086e-06,
+      "loss": 1.8265,
+      "step": 709
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.384820561684862e-06,
+      "loss": 1.9757,
+      "step": 710
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.5,
+      "learning_rate": 4.3802033865267116e-06,
+      "loss": 1.8808,
+      "step": 711
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.6171875,
+      "learning_rate": 4.375582447478569e-06,
+      "loss": 1.908,
+      "step": 712
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.370957760641759e-06,
+      "loss": 1.837,
+      "step": 713
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.3515625,
+      "learning_rate": 4.366329342130667e-06,
+      "loss": 1.8169,
+      "step": 714
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.4140625,
+      "learning_rate": 4.361697208072677e-06,
+      "loss": 1.8081,
+      "step": 715
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.357061374608125e-06,
+      "loss": 1.9752,
+      "step": 716
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.46875,
+      "learning_rate": 4.352421857890234e-06,
+      "loss": 1.8072,
+      "step": 717
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.347778674085065e-06,
+      "loss": 1.896,
+      "step": 718
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.5234375,
+      "learning_rate": 4.343131839371447e-06,
+      "loss": 1.752,
+      "step": 719
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.338481369940943e-06,
+      "loss": 1.8197,
+      "step": 720
+    },
+    {
+      "epoch": 1.21,
+      "eval_loss": 1.9600226879119873,
+      "eval_runtime": 72.5961,
+      "eval_samples_per_second": 8.94,
+      "eval_steps_per_second": 8.94,
+      "step": 720
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.515625,
+      "learning_rate": 4.3338272819977736e-06,
+      "loss": 1.9461,
+      "step": 721
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.5546875,
+      "learning_rate": 4.329169591758768e-06,
+      "loss": 1.9262,
+      "step": 722
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.324508315453309e-06,
+      "loss": 1.8615,
+      "step": 723
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.625,
+      "learning_rate": 4.319843469323273e-06,
+      "loss": 1.7861,
+      "step": 724
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.421875,
+      "learning_rate": 4.315175069622978e-06,
+      "loss": 1.8771,
+      "step": 725
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.515625,
+      "learning_rate": 4.310503132619122e-06,
+      "loss": 2.0245,
+      "step": 726
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.4296875,
+      "learning_rate": 4.305827674590731e-06,
+      "loss": 1.8933,
+      "step": 727
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.301148711829094e-06,
+      "loss": 2.0775,
+      "step": 728
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.6640625,
+      "learning_rate": 4.2964662606377195e-06,
+      "loss": 2.0916,
+      "step": 729
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.5078125,
+      "learning_rate": 4.291780337332265e-06,
+      "loss": 1.9606,
+      "step": 730
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.6484375,
+      "learning_rate": 4.28709095824049e-06,
+      "loss": 1.9152,
+      "step": 731
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.515625,
+      "learning_rate": 4.282398139702194e-06,
+      "loss": 1.8195,
+      "step": 732
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.6953125,
+      "learning_rate": 4.277701898069161e-06,
+      "loss": 1.7624,
+      "step": 733
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.4765625,
+      "learning_rate": 4.273002249705102e-06,
+      "loss": 1.8801,
+      "step": 734
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.5390625,
+      "learning_rate": 4.268299210985601e-06,
+      "loss": 2.0596,
+      "step": 735
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.453125,
+      "learning_rate": 4.2635927982980544e-06,
+      "loss": 2.053,
+      "step": 736
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.4296875,
+      "learning_rate": 4.258883028041611e-06,
+      "loss": 1.902,
+      "step": 737
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.328125,
+      "learning_rate": 4.254169916627126e-06,
+      "loss": 1.735,
+      "step": 738
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.249453480477089e-06,
+      "loss": 1.8353,
+      "step": 739
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.6640625,
+      "learning_rate": 4.244733736025581e-06,
+      "loss": 1.9626,
+      "step": 740
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.484375,
+      "learning_rate": 4.240010699718206e-06,
+      "loss": 1.821,
+      "step": 741
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.5,
+      "learning_rate": 4.23528438801204e-06,
+      "loss": 1.7279,
+      "step": 742
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.4375,
+      "learning_rate": 4.230554817375572e-06,
+      "loss": 1.9597,
+      "step": 743
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.6484375,
+      "learning_rate": 4.225822004288648e-06,
+      "loss": 2.2468,
+      "step": 744
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.221085965242408e-06,
+      "loss": 1.9827,
+      "step": 745
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.3515625,
+      "learning_rate": 4.216346716739235e-06,
+      "loss": 1.8237,
+      "step": 746
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.53125,
+      "learning_rate": 4.211604275292696e-06,
+      "loss": 1.9025,
+      "step": 747
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.4375,
+      "learning_rate": 4.206858657427482e-06,
+      "loss": 2.0194,
+      "step": 748
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.202109879679353e-06,
+      "loss": 1.9849,
+      "step": 749
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.4921875,
+      "learning_rate": 4.1973579585950805e-06,
+      "loss": 2.0062,
+      "step": 750
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.7578125,
+      "learning_rate": 4.192602910732385e-06,
+      "loss": 2.0088,
+      "step": 751
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.5546875,
+      "learning_rate": 4.1878447526598875e-06,
+      "loss": 1.8793,
+      "step": 752
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.4140625,
+      "learning_rate": 4.183083500957039e-06,
+      "loss": 1.8007,
+      "step": 753
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.178319172214075e-06,
+      "loss": 1.6391,
+      "step": 754
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.1735517830319554e-06,
+      "loss": 2.0541,
+      "step": 755
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.453125,
+      "learning_rate": 4.168781350022296e-06,
+      "loss": 1.8274,
+      "step": 756
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.164007889807325e-06,
+      "loss": 1.9349,
+      "step": 757
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.5,
+      "learning_rate": 4.159231419019818e-06,
+      "loss": 2.2892,
+      "step": 758
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.375,
+      "learning_rate": 4.154451954303038e-06,
+      "loss": 2.0556,
+      "step": 759
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.5546875,
+      "learning_rate": 4.1496695123106805e-06,
+      "loss": 1.9248,
+      "step": 760
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.4609375,
+      "learning_rate": 4.1448841097068166e-06,
+      "loss": 2.0287,
+      "step": 761
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.3828125,
+      "learning_rate": 4.140095763165836e-06,
+      "loss": 1.9871,
+      "step": 762
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.135304489372379e-06,
+      "loss": 1.8919,
+      "step": 763
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.6640625,
+      "learning_rate": 4.130510305021293e-06,
+      "loss": 2.1305,
+      "step": 764
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.484375,
+      "learning_rate": 4.125713226817563e-06,
+      "loss": 1.8925,
+      "step": 765
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.3203125,
+      "learning_rate": 4.120913271476259e-06,
+      "loss": 1.8463,
+      "step": 766
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.3984375,
+      "learning_rate": 4.116110455722474e-06,
+      "loss": 1.5727,
+      "step": 767
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.111304796291272e-06,
+      "loss": 1.9468,
+      "step": 768
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.4296875,
+      "learning_rate": 4.10649630992762e-06,
+      "loss": 1.9911,
+      "step": 769
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.6796875,
+      "learning_rate": 4.1016850133863425e-06,
+      "loss": 2.0594,
+      "step": 770
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.40625,
+      "learning_rate": 4.096870923432048e-06,
+      "loss": 1.838,
+      "step": 771
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.3125,
+      "learning_rate": 4.092054056839085e-06,
+      "loss": 1.8663,
+      "step": 772
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.087234430391472e-06,
+      "loss": 2.1704,
+      "step": 773
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.46875,
+      "learning_rate": 4.0824120608828466e-06,
+      "loss": 1.8005,
+      "step": 774
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.077586965116406e-06,
+      "loss": 1.9346,
+      "step": 775
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.6484375,
+      "learning_rate": 4.072759159904841e-06,
+      "loss": 2.179,
+      "step": 776
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.3984375,
+      "learning_rate": 4.067928662070291e-06,
+      "loss": 2.0209,
+      "step": 777
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.6640625,
+      "learning_rate": 4.063095488444271e-06,
+      "loss": 1.9372,
+      "step": 778
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.5,
+      "learning_rate": 4.058259655867623e-06,
+      "loss": 1.9265,
+      "step": 779
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.46875,
+      "learning_rate": 4.053421181190454e-06,
+      "loss": 1.9848,
+      "step": 780
+    },
+    {
+      "epoch": 1.31,
+      "eval_loss": 1.9584399461746216,
+      "eval_runtime": 72.5746,
+      "eval_samples_per_second": 8.943,
+      "eval_steps_per_second": 8.943,
+      "step": 780
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.5859375,
+      "learning_rate": 4.048580081272078e-06,
+      "loss": 2.0108,
+      "step": 781
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.421875,
+      "learning_rate": 4.0437363729809515e-06,
+      "loss": 1.8503,
+      "step": 782
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.3984375,
+      "learning_rate": 4.038890073194625e-06,
+      "loss": 1.7675,
+      "step": 783
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.034041198799675e-06,
+      "loss": 2.0385,
+      "step": 784
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.390625,
+      "learning_rate": 4.029189766691654e-06,
+      "loss": 1.8441,
+      "step": 785
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.453125,
+      "learning_rate": 4.024335793775022e-06,
+      "loss": 1.8735,
+      "step": 786
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.53125,
+      "learning_rate": 4.019479296963094e-06,
+      "loss": 2.027,
+      "step": 787
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.53125,
+      "learning_rate": 4.0146202931779806e-06,
+      "loss": 1.9666,
+      "step": 788
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.6640625,
+      "learning_rate": 4.009758799350525e-06,
+      "loss": 1.8787,
+      "step": 789
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.3984375,
+      "learning_rate": 4.004894832420249e-06,
+      "loss": 1.8996,
+      "step": 790
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.2890625,
+      "learning_rate": 4.0000284093352885e-06,
+      "loss": 1.8736,
+      "step": 791
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.6796875,
+      "learning_rate": 3.995159547052345e-06,
+      "loss": 2.0839,
+      "step": 792
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.453125,
+      "learning_rate": 3.9902882625366095e-06,
+      "loss": 2.0342,
+      "step": 793
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.40625,
+      "learning_rate": 3.985414572761721e-06,
+      "loss": 1.9169,
+      "step": 794
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.5625,
+      "learning_rate": 3.980538494709696e-06,
+      "loss": 1.8123,
+      "step": 795
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.4609375,
+      "learning_rate": 3.975660045370871e-06,
+      "loss": 1.7921,
+      "step": 796
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.970779241743851e-06,
+      "loss": 1.7977,
+      "step": 797
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.578125,
+      "learning_rate": 3.965896100835437e-06,
+      "loss": 1.826,
+      "step": 798
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.40625,
+      "learning_rate": 3.96101063966058e-06,
+      "loss": 1.9129,
+      "step": 799
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.6796875,
+      "learning_rate": 3.956122875242313e-06,
+      "loss": 2.0396,
+      "step": 800
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.5546875,
+      "learning_rate": 3.9512328246116925e-06,
+      "loss": 2.0096,
+      "step": 801
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.40625,
+      "learning_rate": 3.946340504807748e-06,
+      "loss": 1.9644,
+      "step": 802
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.6328125,
+      "learning_rate": 3.9414459328774095e-06,
+      "loss": 2.0962,
+      "step": 803
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.53125,
+      "learning_rate": 3.936549125875454e-06,
+      "loss": 2.032,
+      "step": 804
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.6875,
+      "learning_rate": 3.931650100864452e-06,
+      "loss": 2.0593,
+      "step": 805
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.296875,
+      "learning_rate": 3.9267488749146966e-06,
+      "loss": 1.9617,
+      "step": 806
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.3359375,
+      "learning_rate": 3.921845465104155e-06,
+      "loss": 1.9175,
+      "step": 807
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.5859375,
+      "learning_rate": 3.916939888518399e-06,
+      "loss": 1.8773,
+      "step": 808
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.40625,
+      "learning_rate": 3.9120321622505544e-06,
+      "loss": 2.002,
+      "step": 809
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.421875,
+      "learning_rate": 3.907122303401236e-06,
+      "loss": 1.9718,
+      "step": 810
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.5859375,
+      "learning_rate": 3.902210329078487e-06,
+      "loss": 2.0348,
+      "step": 811
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.53125,
+      "learning_rate": 3.897296256397727e-06,
+      "loss": 1.8621,
+      "step": 812
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.8923801024816825e-06,
+      "loss": 2.0355,
+      "step": 813
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.4296875,
+      "learning_rate": 3.887461884460332e-06,
+      "loss": 1.7751,
+      "step": 814
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.5625,
+      "learning_rate": 3.882541619470851e-06,
+      "loss": 1.876,
+      "step": 815
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.328125,
+      "learning_rate": 3.8776193246575425e-06,
+      "loss": 1.9031,
+      "step": 816
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.71875,
+      "learning_rate": 3.872695017171786e-06,
+      "loss": 1.8683,
+      "step": 817
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.4921875,
+      "learning_rate": 3.867768714171971e-06,
+      "loss": 2.0642,
+      "step": 818
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.862840432823439e-06,
+      "loss": 1.9005,
+      "step": 819
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.421875,
+      "learning_rate": 3.857910190298431e-06,
+      "loss": 1.9538,
+      "step": 820
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.5625,
+      "learning_rate": 3.852978003776015e-06,
+      "loss": 1.7715,
+      "step": 821
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.4140625,
+      "learning_rate": 3.848043890442038e-06,
+      "loss": 1.8265,
+      "step": 822
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.5546875,
+      "learning_rate": 3.843107867489056e-06,
+      "loss": 1.9873,
+      "step": 823
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.4609375,
+      "learning_rate": 3.8381699521162834e-06,
+      "loss": 2.0191,
+      "step": 824
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.359375,
+      "learning_rate": 3.8332301615295265e-06,
+      "loss": 1.8491,
+      "step": 825
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.5859375,
+      "learning_rate": 3.828288512941122e-06,
+      "loss": 1.9185,
+      "step": 826
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.703125,
+      "learning_rate": 3.823345023569889e-06,
+      "loss": 2.0706,
+      "step": 827
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.818399710641055e-06,
+      "loss": 2.1711,
+      "step": 828
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.3515625,
+      "learning_rate": 3.8134525913861997e-06,
+      "loss": 1.8938,
+      "step": 829
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.8085036830432033e-06,
+      "loss": 1.8994,
+      "step": 830
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.2890625,
+      "learning_rate": 3.8035530028561736e-06,
+      "loss": 1.7224,
+      "step": 831
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.5546875,
+      "learning_rate": 3.798600568075397e-06,
+      "loss": 2.2053,
+      "step": 832
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.5234375,
+      "learning_rate": 3.7936463959572707e-06,
+      "loss": 1.8359,
+      "step": 833
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.5859375,
+      "learning_rate": 3.7886905037642454e-06,
+      "loss": 1.8931,
+      "step": 834
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.5,
+      "learning_rate": 3.783732908764769e-06,
+      "loss": 1.7162,
+      "step": 835
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.4140625,
+      "learning_rate": 3.778773628233217e-06,
+      "loss": 1.9625,
+      "step": 836
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.40625,
+      "learning_rate": 3.7738126794498437e-06,
+      "loss": 1.847,
+      "step": 837
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.484375,
+      "learning_rate": 3.7688500797007126e-06,
+      "loss": 1.6776,
+      "step": 838
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.515625,
+      "learning_rate": 3.7638858462776402e-06,
+      "loss": 1.8384,
+      "step": 839
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.5625,
+      "learning_rate": 3.7589199964781372e-06,
+      "loss": 1.9262,
+      "step": 840
+    },
+    {
+      "epoch": 1.41,
+      "eval_loss": 1.9571281671524048,
+      "eval_runtime": 72.4769,
+      "eval_samples_per_second": 8.955,
+      "eval_steps_per_second": 8.955,
+      "step": 840
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.753952547605344e-06,
+      "loss": 1.9915,
+      "step": 841
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.5234375,
+      "learning_rate": 3.748983516967977e-06,
+      "loss": 2.0314,
+      "step": 842
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.40625,
+      "learning_rate": 3.7440129218802585e-06,
+      "loss": 1.9247,
+      "step": 843
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.3828125,
+      "learning_rate": 3.739040779661867e-06,
+      "loss": 1.9936,
+      "step": 844
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.3203125,
+      "learning_rate": 3.7340671076378677e-06,
+      "loss": 1.6597,
+      "step": 845
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.515625,
+      "learning_rate": 3.72909192313866e-06,
+      "loss": 1.6903,
+      "step": 846
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.484375,
+      "learning_rate": 3.724115243499912e-06,
+      "loss": 1.7956,
+      "step": 847
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.4765625,
+      "learning_rate": 3.7191370860624998e-06,
+      "loss": 1.8839,
+      "step": 848
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.5234375,
+      "learning_rate": 3.7141574681724506e-06,
+      "loss": 1.8973,
+      "step": 849
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.5703125,
+      "learning_rate": 3.7091764071808814e-06,
+      "loss": 2.0264,
+      "step": 850
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.53125,
+      "learning_rate": 3.7041939204439346e-06,
+      "loss": 1.7518,
+      "step": 851
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.390625,
+      "learning_rate": 3.6992100253227237e-06,
+      "loss": 1.9336,
+      "step": 852
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.4609375,
+      "learning_rate": 3.6942247391832676e-06,
+      "loss": 1.7735,
+      "step": 853
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.5859375,
+      "learning_rate": 3.689238079396432e-06,
+      "loss": 1.891,
+      "step": 854
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.4296875,
+      "learning_rate": 3.6842500633378707e-06,
+      "loss": 2.0223,
+      "step": 855
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.4453125,
+      "learning_rate": 3.6792607083879604e-06,
+      "loss": 2.0054,
+      "step": 856
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.375,
+      "learning_rate": 3.6742700319317468e-06,
+      "loss": 1.81,
+      "step": 857
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.669278051358877e-06,
+      "loss": 1.9736,
+      "step": 858
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.3828125,
+      "learning_rate": 3.664284784063544e-06,
+      "loss": 1.9764,
+      "step": 859
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.6484375,
+      "learning_rate": 3.6592902474444248e-06,
+      "loss": 2.1843,
+      "step": 860
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.28125,
+      "learning_rate": 3.6542944589046175e-06,
+      "loss": 1.8826,
+      "step": 861
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.40625,
+      "learning_rate": 3.6492974358515837e-06,
+      "loss": 1.8483,
+      "step": 862
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.3828125,
+      "learning_rate": 3.644299195697085e-06,
+      "loss": 1.6964,
+      "step": 863
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.625,
+      "learning_rate": 3.639299755857125e-06,
+      "loss": 2.0857,
+      "step": 864
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.6342991337518898e-06,
+      "loss": 1.8794,
+      "step": 865
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.421875,
+      "learning_rate": 3.62929734680568e-06,
+      "loss": 1.8154,
+      "step": 866
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.40625,
+      "learning_rate": 3.6242944124468605e-06,
+      "loss": 1.852,
+      "step": 867
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.4296875,
+      "learning_rate": 3.6192903481077883e-06,
+      "loss": 2.0246,
+      "step": 868
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.5625,
+      "learning_rate": 3.614285171224765e-06,
+      "loss": 1.953,
+      "step": 869
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.46875,
+      "learning_rate": 3.6092788992379605e-06,
+      "loss": 1.7793,
+      "step": 870
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.3828125,
+      "learning_rate": 3.604271549591367e-06,
+      "loss": 1.846,
+      "step": 871
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.5992631397327303e-06,
+      "loss": 1.84,
+      "step": 872
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.4140625,
+      "learning_rate": 3.5942536871134875e-06,
+      "loss": 1.9115,
+      "step": 873
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.59375,
+      "learning_rate": 3.589243209188713e-06,
+      "loss": 1.7918,
+      "step": 874
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.53125,
+      "learning_rate": 3.584231723417051e-06,
+      "loss": 2.1052,
+      "step": 875
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.4296875,
+      "learning_rate": 3.5792192472606596e-06,
+      "loss": 2.0174,
+      "step": 876
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.4765625,
+      "learning_rate": 3.5742057981851454e-06,
+      "loss": 1.9988,
+      "step": 877
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.359375,
+      "learning_rate": 3.5691913936595063e-06,
+      "loss": 1.9073,
+      "step": 878
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.453125,
+      "learning_rate": 3.564176051156072e-06,
+      "loss": 1.9155,
+      "step": 879
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.4140625,
+      "learning_rate": 3.559159788150435e-06,
+      "loss": 1.8487,
+      "step": 880
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.4609375,
+      "learning_rate": 3.554142622121399e-06,
+      "loss": 2.015,
+      "step": 881
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.5703125,
+      "learning_rate": 3.5491245705509153e-06,
+      "loss": 2.0941,
+      "step": 882
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.3828125,
+      "learning_rate": 3.5441056509240177e-06,
+      "loss": 1.9948,
+      "step": 883
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.515625,
+      "learning_rate": 3.5390858807287653e-06,
+      "loss": 2.1411,
+      "step": 884
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.534065277456182e-06,
+      "loss": 2.1164,
+      "step": 885
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.328125,
+      "learning_rate": 3.529043858600194e-06,
+      "loss": 1.8789,
+      "step": 886
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.5546875,
+      "learning_rate": 3.5240216416575704e-06,
+      "loss": 1.9746,
+      "step": 887
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.3828125,
+      "learning_rate": 3.518998644127859e-06,
+      "loss": 1.7627,
+      "step": 888
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.46875,
+      "learning_rate": 3.513974883513329e-06,
+      "loss": 1.938,
+      "step": 889
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.4453125,
+      "learning_rate": 3.5089503773189084e-06,
+      "loss": 1.9516,
+      "step": 890
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.3359375,
+      "learning_rate": 3.5039251430521237e-06,
+      "loss": 1.8827,
+      "step": 891
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.4988991982230364e-06,
+      "loss": 1.8046,
+      "step": 892
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.46875,
+      "learning_rate": 3.4938725603441848e-06,
+      "loss": 1.9488,
+      "step": 893
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.4888452469305233e-06,
+      "loss": 1.9693,
+      "step": 894
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.4609375,
+      "learning_rate": 3.4838172754993586e-06,
+      "loss": 1.9955,
+      "step": 895
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.478788663570291e-06,
+      "loss": 1.947,
+      "step": 896
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.5234375,
+      "learning_rate": 3.473759428665152e-06,
+      "loss": 1.9457,
+      "step": 897
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.4765625,
+      "learning_rate": 3.4687295883079457e-06,
+      "loss": 1.9994,
+      "step": 898
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.463699160024783e-06,
+      "loss": 1.7545,
+      "step": 899
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.4586681613438247e-06,
+      "loss": 2.0359,
+      "step": 900
+    },
+    {
+      "epoch": 1.51,
+      "eval_loss": 1.9557842016220093,
+      "eval_runtime": 72.8827,
+      "eval_samples_per_second": 8.905,
+      "eval_steps_per_second": 8.905,
+      "step": 900
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.5546875,
+      "learning_rate": 3.4536366097952182e-06,
+      "loss": 1.8549,
+      "step": 901
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.4296875,
+      "learning_rate": 3.448604522911041e-06,
+      "loss": 1.8572,
+      "step": 902
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.46875,
+      "learning_rate": 3.4435719182252302e-06,
+      "loss": 1.8748,
+      "step": 903
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.484375,
+      "learning_rate": 3.438538813273532e-06,
+      "loss": 1.9317,
+      "step": 904
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.484375,
+      "learning_rate": 3.4335052255934334e-06,
+      "loss": 1.928,
+      "step": 905
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.484375,
+      "learning_rate": 3.428471172724105e-06,
+      "loss": 1.933,
+      "step": 906
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.3828125,
+      "learning_rate": 3.423436672206335e-06,
+      "loss": 1.9946,
+      "step": 907
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.4921875,
+      "learning_rate": 3.418401741582474e-06,
+      "loss": 1.9355,
+      "step": 908
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.413366398396373e-06,
+      "loss": 1.9021,
+      "step": 909
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.4921875,
+      "learning_rate": 3.4083306601933165e-06,
+      "loss": 2.0507,
+      "step": 910
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.403294544519968e-06,
+      "loss": 1.9046,
+      "step": 911
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.34375,
+      "learning_rate": 3.3982580689243053e-06,
+      "loss": 1.8577,
+      "step": 912
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.265625,
+      "learning_rate": 3.3932212509555613e-06,
+      "loss": 1.9135,
+      "step": 913
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.4296875,
+      "learning_rate": 3.388184108164161e-06,
+      "loss": 2.0145,
+      "step": 914
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.46875,
+      "learning_rate": 3.383146658101661e-06,
+      "loss": 2.0523,
+      "step": 915
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.3984375,
+      "learning_rate": 3.378108918320688e-06,
+      "loss": 1.9216,
+      "step": 916
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.4140625,
+      "learning_rate": 3.373070906374882e-06,
+      "loss": 1.78,
+      "step": 917
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.2578125,
+      "learning_rate": 3.368032639818825e-06,
+      "loss": 1.8309,
+      "step": 918
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.3046875,
+      "learning_rate": 3.3629941362079933e-06,
+      "loss": 1.861,
+      "step": 919
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.375,
+      "learning_rate": 3.3579554130986842e-06,
+      "loss": 2.0568,
+      "step": 920
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.5,
+      "learning_rate": 3.35291648804796e-06,
+      "loss": 1.8546,
+      "step": 921
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.3515625,
+      "learning_rate": 3.34787737861359e-06,
+      "loss": 1.9779,
+      "step": 922
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.5625,
+      "learning_rate": 3.3428381023539818e-06,
+      "loss": 1.876,
+      "step": 923
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.4609375,
+      "learning_rate": 3.3377986768281275e-06,
+      "loss": 1.8134,
+      "step": 924
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.40625,
+      "learning_rate": 3.3327591195955384e-06,
+      "loss": 1.999,
+      "step": 925
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.5390625,
+      "learning_rate": 3.327719448216183e-06,
+      "loss": 2.1062,
+      "step": 926
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.53125,
+      "learning_rate": 3.3226796802504303e-06,
+      "loss": 2.1174,
+      "step": 927
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.4296875,
+      "learning_rate": 3.317639833258985e-06,
+      "loss": 1.7734,
+      "step": 928
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.3671875,
+      "learning_rate": 3.312599924802826e-06,
+      "loss": 1.8079,
+      "step": 929
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.4296875,
+      "learning_rate": 3.3075599724431464e-06,
+      "loss": 1.7901,
+      "step": 930
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.484375,
+      "learning_rate": 3.302519993741294e-06,
+      "loss": 2.0501,
+      "step": 931
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.40625,
+      "learning_rate": 3.2974800062587066e-06,
+      "loss": 1.8301,
+      "step": 932
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.2924400275568548e-06,
+      "loss": 1.8754,
+      "step": 933
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.3515625,
+      "learning_rate": 3.287400075197175e-06,
+      "loss": 1.767,
+      "step": 934
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.703125,
+      "learning_rate": 3.2823601667410156e-06,
+      "loss": 2.1026,
+      "step": 935
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.5625,
+      "learning_rate": 3.2773203197495704e-06,
+      "loss": 1.929,
+      "step": 936
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.4765625,
+      "learning_rate": 3.272280551783818e-06,
+      "loss": 1.9192,
+      "step": 937
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.359375,
+      "learning_rate": 3.2672408804044636e-06,
+      "loss": 1.8981,
+      "step": 938
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.4296875,
+      "learning_rate": 3.2622013231718724e-06,
+      "loss": 1.9101,
+      "step": 939
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.453125,
+      "learning_rate": 3.2571618976460186e-06,
+      "loss": 2.024,
+      "step": 940
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.2521226213864105e-06,
+      "loss": 1.8604,
+      "step": 941
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.24708351195204e-06,
+      "loss": 1.8735,
+      "step": 942
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.5625,
+      "learning_rate": 3.2420445869013165e-06,
+      "loss": 1.8987,
+      "step": 943
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.5234375,
+      "learning_rate": 3.2370058637920066e-06,
+      "loss": 1.9249,
+      "step": 944
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.390625,
+      "learning_rate": 3.2319673601811747e-06,
+      "loss": 1.82,
+      "step": 945
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.578125,
+      "learning_rate": 3.2269290936251187e-06,
+      "loss": 1.9913,
+      "step": 946
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.5703125,
+      "learning_rate": 3.2218910816793125e-06,
+      "loss": 1.9907,
+      "step": 947
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.75,
+      "learning_rate": 3.2168533418983404e-06,
+      "loss": 2.0234,
+      "step": 948
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.3671875,
+      "learning_rate": 3.21181589183584e-06,
+      "loss": 1.7599,
+      "step": 949
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.328125,
+      "learning_rate": 3.2067787490444394e-06,
+      "loss": 1.7259,
+      "step": 950
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.390625,
+      "learning_rate": 3.2017419310756955e-06,
+      "loss": 1.9728,
+      "step": 951
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.5,
+      "learning_rate": 3.196705455480033e-06,
+      "loss": 1.83,
+      "step": 952
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.3203125,
+      "learning_rate": 3.1916693398066847e-06,
+      "loss": 1.662,
+      "step": 953
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.578125,
+      "learning_rate": 3.186633601603627e-06,
+      "loss": 1.8832,
+      "step": 954
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.4140625,
+      "learning_rate": 3.181598258417526e-06,
+      "loss": 2.0136,
+      "step": 955
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.4453125,
+      "learning_rate": 3.1765633277936655e-06,
+      "loss": 1.8746,
+      "step": 956
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.5625,
+      "learning_rate": 3.171528827275896e-06,
+      "loss": 1.8762,
+      "step": 957
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.4609375,
+      "learning_rate": 3.1664947744065665e-06,
+      "loss": 1.9944,
+      "step": 958
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.46875,
+      "learning_rate": 3.1614611867264677e-06,
+      "loss": 1.8449,
+      "step": 959
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.5,
+      "learning_rate": 3.1564280817747697e-06,
+      "loss": 1.8945,
+      "step": 960
+    },
+    {
+      "epoch": 1.61,
+      "eval_loss": 1.9549872875213623,
+      "eval_runtime": 72.7206,
+      "eval_samples_per_second": 8.925,
+      "eval_steps_per_second": 8.925,
+      "step": 960
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.2734375,
+      "learning_rate": 3.1513954770889597e-06,
+      "loss": 1.8645,
+      "step": 961
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.4765625,
+      "learning_rate": 3.146363390204782e-06,
+      "loss": 1.8943,
+      "step": 962
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.515625,
+      "learning_rate": 3.1413318386561765e-06,
+      "loss": 2.1025,
+      "step": 963
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.136300839975218e-06,
+      "loss": 1.9225,
+      "step": 964
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.453125,
+      "learning_rate": 3.1312704116920546e-06,
+      "loss": 2.0847,
+      "step": 965
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.3046875,
+      "learning_rate": 3.126240571334848e-06,
+      "loss": 1.7204,
+      "step": 966
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.12121133642971e-06,
+      "loss": 1.8552,
+      "step": 967
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.116182724500642e-06,
+      "loss": 1.8653,
+      "step": 968
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.34375,
+      "learning_rate": 3.1111547530694767e-06,
+      "loss": 1.9546,
+      "step": 969
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.53125,
+      "learning_rate": 3.106127439655816e-06,
+      "loss": 2.0012,
+      "step": 970
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.484375,
+      "learning_rate": 3.1011008017769644e-06,
+      "loss": 1.9694,
+      "step": 971
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.3515625,
+      "learning_rate": 3.096074856947877e-06,
+      "loss": 2.0119,
+      "step": 972
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.234375,
+      "learning_rate": 3.0910496226810924e-06,
+      "loss": 1.7685,
+      "step": 973
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.086025116486671e-06,
+      "loss": 1.9522,
+      "step": 974
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.5,
+      "learning_rate": 3.081001355872141e-06,
+      "loss": 2.0921,
+      "step": 975
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.5390625,
+      "learning_rate": 3.0759783583424303e-06,
+      "loss": 1.9473,
+      "step": 976
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.359375,
+      "learning_rate": 3.0709561413998067e-06,
+      "loss": 1.868,
+      "step": 977
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.515625,
+      "learning_rate": 3.065934722543819e-06,
+      "loss": 2.0399,
+      "step": 978
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.4375,
+      "learning_rate": 3.0609141192712355e-06,
+      "loss": 2.0686,
+      "step": 979
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.5,
+      "learning_rate": 3.055894349075983e-06,
+      "loss": 1.7349,
+      "step": 980
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.3125,
+      "learning_rate": 3.0508754294490854e-06,
+      "loss": 1.8013,
+      "step": 981
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.3203125,
+      "learning_rate": 3.045857377878601e-06,
+      "loss": 1.7967,
+      "step": 982
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.3984375,
+      "learning_rate": 3.0408402118495656e-06,
+      "loss": 1.8798,
+      "step": 983
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.4453125,
+      "learning_rate": 3.035823948843928e-06,
+      "loss": 1.9516,
+      "step": 984
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.0308086063404936e-06,
+      "loss": 1.9727,
+      "step": 985
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.3828125,
+      "learning_rate": 3.0257942018148554e-06,
+      "loss": 1.7369,
+      "step": 986
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.359375,
+      "learning_rate": 3.020780752739341e-06,
+      "loss": 1.9381,
+      "step": 987
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.515625,
+      "learning_rate": 3.0157682765829498e-06,
+      "loss": 1.8097,
+      "step": 988
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.3125,
+      "learning_rate": 3.010756790811287e-06,
+      "loss": 1.8961,
+      "step": 989
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.421875,
+      "learning_rate": 3.0057463128865125e-06,
+      "loss": 1.9253,
+      "step": 990
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.0007368602672705e-06,
+      "loss": 1.861,
+      "step": 991
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.9957284504086337e-06,
+      "loss": 2.093,
+      "step": 992
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.9907211007620407e-06,
+      "loss": 1.873,
+      "step": 993
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.4765625,
+      "learning_rate": 2.985714828775237e-06,
+      "loss": 1.9115,
+      "step": 994
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.3203125,
+      "learning_rate": 2.9807096518922116e-06,
+      "loss": 1.8313,
+      "step": 995
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.515625,
+      "learning_rate": 2.975705587553141e-06,
+      "loss": 1.8919,
+      "step": 996
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.3828125,
+      "learning_rate": 2.9707026531943204e-06,
+      "loss": 1.8009,
+      "step": 997
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.3828125,
+      "learning_rate": 2.965700866248111e-06,
+      "loss": 2.006,
+      "step": 998
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.5078125,
+      "learning_rate": 2.9607002441428745e-06,
+      "loss": 1.9849,
+      "step": 999
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.375,
+      "learning_rate": 2.9557008043029154e-06,
+      "loss": 1.8528,
+      "step": 1000
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.950702564148417e-06,
+      "loss": 2.0375,
+      "step": 1001
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.945705541095383e-06,
+      "loss": 1.9146,
+      "step": 1002
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.28125,
+      "learning_rate": 2.9407097525555764e-06,
+      "loss": 1.8071,
+      "step": 1003
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.53125,
+      "learning_rate": 2.9357152159364554e-06,
+      "loss": 1.8935,
+      "step": 1004
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.515625,
+      "learning_rate": 2.9307219486411228e-06,
+      "loss": 1.8324,
+      "step": 1005
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.4375,
+      "learning_rate": 2.925729968068254e-06,
+      "loss": 1.9883,
+      "step": 1006
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.92073929161204e-06,
+      "loss": 1.9158,
+      "step": 1007
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.7578125,
+      "learning_rate": 2.915749936662131e-06,
+      "loss": 1.9927,
+      "step": 1008
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.9107619206035687e-06,
+      "loss": 1.9653,
+      "step": 1009
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.4765625,
+      "learning_rate": 2.905775260816733e-06,
+      "loss": 1.9164,
+      "step": 1010
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.34375,
+      "learning_rate": 2.900789974677276e-06,
+      "loss": 1.8986,
+      "step": 1011
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.578125,
+      "learning_rate": 2.895806079556066e-06,
+      "loss": 2.0182,
+      "step": 1012
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.3125,
+      "learning_rate": 2.8908235928191198e-06,
+      "loss": 1.7159,
+      "step": 1013
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.5078125,
+      "learning_rate": 2.8858425318275493e-06,
+      "loss": 2.0026,
+      "step": 1014
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.880862913937501e-06,
+      "loss": 2.0248,
+      "step": 1015
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.53125,
+      "learning_rate": 2.8758847565000887e-06,
+      "loss": 1.8751,
+      "step": 1016
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.4453125,
+      "learning_rate": 2.8709080768613406e-06,
+      "loss": 1.7352,
+      "step": 1017
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.421875,
+      "learning_rate": 2.8659328923621335e-06,
+      "loss": 1.9528,
+      "step": 1018
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.4140625,
+      "learning_rate": 2.860959220338134e-06,
+      "loss": 1.8967,
+      "step": 1019
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.4375,
+      "learning_rate": 2.8559870781197414e-06,
+      "loss": 1.9923,
+      "step": 1020
+    },
+    {
+      "epoch": 1.71,
+      "eval_loss": 1.953945279121399,
+      "eval_runtime": 72.8216,
+      "eval_samples_per_second": 8.912,
+      "eval_steps_per_second": 8.912,
+      "step": 1020
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.3828125,
+      "learning_rate": 2.851016483032024e-06,
+      "loss": 1.8657,
+      "step": 1021
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.846047452394656e-06,
+      "loss": 1.8847,
+      "step": 1022
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.328125,
+      "learning_rate": 2.8410800035218635e-06,
+      "loss": 2.02,
+      "step": 1023
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.390625,
+      "learning_rate": 2.83611415372236e-06,
+      "loss": 1.8547,
+      "step": 1024
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.4453125,
+      "learning_rate": 2.831149920299288e-06,
+      "loss": 1.7854,
+      "step": 1025
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.826187320550157e-06,
+      "loss": 1.7301,
+      "step": 1026
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.8212263717667836e-06,
+      "loss": 1.8653,
+      "step": 1027
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.4140625,
+      "learning_rate": 2.8162670912352323e-06,
+      "loss": 1.7948,
+      "step": 1028
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.811309496235754e-06,
+      "loss": 1.9075,
+      "step": 1029
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.4375,
+      "learning_rate": 2.80635360404273e-06,
+      "loss": 1.8065,
+      "step": 1030
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.515625,
+      "learning_rate": 2.801399431924603e-06,
+      "loss": 1.9205,
+      "step": 1031
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.5703125,
+      "learning_rate": 2.796446997143827e-06,
+      "loss": 1.9284,
+      "step": 1032
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.4296875,
+      "learning_rate": 2.791496316956798e-06,
+      "loss": 1.9412,
+      "step": 1033
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.5078125,
+      "learning_rate": 2.7865474086138003e-06,
+      "loss": 1.985,
+      "step": 1034
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.515625,
+      "learning_rate": 2.781600289358946e-06,
+      "loss": 1.8522,
+      "step": 1035
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.3984375,
+      "learning_rate": 2.776654976430111e-06,
+      "loss": 1.8031,
+      "step": 1036
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.3515625,
+      "learning_rate": 2.771711487058878e-06,
+      "loss": 1.8725,
+      "step": 1037
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.766769838470476e-06,
+      "loss": 2.0065,
+      "step": 1038
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.761830047883717e-06,
+      "loss": 2.0794,
+      "step": 1039
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.515625,
+      "learning_rate": 2.7568921325109447e-06,
+      "loss": 1.7627,
+      "step": 1040
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.421875,
+      "learning_rate": 2.7519561095579635e-06,
+      "loss": 1.8511,
+      "step": 1041
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.7470219962239855e-06,
+      "loss": 1.9931,
+      "step": 1042
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.4453125,
+      "learning_rate": 2.742089809701571e-06,
+      "loss": 1.8826,
+      "step": 1043
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.737159567176561e-06,
+      "loss": 1.9399,
+      "step": 1044
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.7322312858280303e-06,
+      "loss": 1.7728,
+      "step": 1045
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.5,
+      "learning_rate": 2.727304982828215e-06,
+      "loss": 2.062,
+      "step": 1046
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.4140625,
+      "learning_rate": 2.722380675342458e-06,
+      "loss": 1.8854,
+      "step": 1047
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.71745838052915e-06,
+      "loss": 2.0153,
+      "step": 1048
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.3203125,
+      "learning_rate": 2.7125381155396677e-06,
+      "loss": 1.7867,
+      "step": 1049
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.7890625,
+      "learning_rate": 2.7076198975183187e-06,
+      "loss": 1.6952,
+      "step": 1050
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.421875,
+      "learning_rate": 2.7027037436022736e-06,
+      "loss": 1.8918,
+      "step": 1051
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.6977896709215137e-06,
+      "loss": 1.8403,
+      "step": 1052
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.3828125,
+      "learning_rate": 2.6928776965987655e-06,
+      "loss": 1.8227,
+      "step": 1053
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.625,
+      "learning_rate": 2.6879678377494463e-06,
+      "loss": 1.8661,
+      "step": 1054
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.5546875,
+      "learning_rate": 2.6830601114816014e-06,
+      "loss": 1.905,
+      "step": 1055
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.375,
+      "learning_rate": 2.6781545348958467e-06,
+      "loss": 1.6843,
+      "step": 1056
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.6732511250853046e-06,
+      "loss": 1.8012,
+      "step": 1057
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.375,
+      "learning_rate": 2.6683498991355496e-06,
+      "loss": 1.9774,
+      "step": 1058
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.59375,
+      "learning_rate": 2.6634508741245463e-06,
+      "loss": 2.1853,
+      "step": 1059
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.3828125,
+      "learning_rate": 2.658554067122592e-06,
+      "loss": 1.9452,
+      "step": 1060
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.515625,
+      "learning_rate": 2.653659495192253e-06,
+      "loss": 1.9929,
+      "step": 1061
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.4375,
+      "learning_rate": 2.6487671753883074e-06,
+      "loss": 2.026,
+      "step": 1062
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.375,
+      "learning_rate": 2.643877124757689e-06,
+      "loss": 1.9421,
+      "step": 1063
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.63898936033942e-06,
+      "loss": 2.0657,
+      "step": 1064
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.5625,
+      "learning_rate": 2.634103899164563e-06,
+      "loss": 1.964,
+      "step": 1065
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.3984375,
+      "learning_rate": 2.6292207582561494e-06,
+      "loss": 1.938,
+      "step": 1066
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.546875,
+      "learning_rate": 2.6243399546291285e-06,
+      "loss": 2.0169,
+      "step": 1067
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.5234375,
+      "learning_rate": 2.619461505290305e-06,
+      "loss": 1.913,
+      "step": 1068
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.4765625,
+      "learning_rate": 2.6145854272382794e-06,
+      "loss": 1.8626,
+      "step": 1069
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.3828125,
+      "learning_rate": 2.6097117374633904e-06,
+      "loss": 1.8187,
+      "step": 1070
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.4765625,
+      "learning_rate": 2.604840452947656e-06,
+      "loss": 2.019,
+      "step": 1071
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.4375,
+      "learning_rate": 2.5999715906647114e-06,
+      "loss": 1.9029,
+      "step": 1072
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.515625,
+      "learning_rate": 2.5951051675797527e-06,
+      "loss": 1.8707,
+      "step": 1073
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.5,
+      "learning_rate": 2.5902412006494745e-06,
+      "loss": 1.9438,
+      "step": 1074
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.515625,
+      "learning_rate": 2.5853797068220198e-06,
+      "loss": 2.0,
+      "step": 1075
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.484375,
+      "learning_rate": 2.580520703036906e-06,
+      "loss": 1.8732,
+      "step": 1076
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.575664206224979e-06,
+      "loss": 1.9073,
+      "step": 1077
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.5708102333083473e-06,
+      "loss": 1.8259,
+      "step": 1078
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.4140625,
+      "learning_rate": 2.5659588012003246e-06,
+      "loss": 1.9296,
+      "step": 1079
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.3984375,
+      "learning_rate": 2.561109926805377e-06,
+      "loss": 1.9241,
+      "step": 1080
+    },
+    {
+      "epoch": 1.82,
+      "eval_loss": 1.953582763671875,
+      "eval_runtime": 72.6019,
+      "eval_samples_per_second": 8.939,
+      "eval_steps_per_second": 8.939,
+      "step": 1080
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.515625,
+      "learning_rate": 2.5562636270190496e-06,
+      "loss": 1.9805,
+      "step": 1081
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.4296875,
+      "learning_rate": 2.551419918727923e-06,
+      "loss": 1.9033,
+      "step": 1082
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.359375,
+      "learning_rate": 2.5465788188095457e-06,
+      "loss": 1.8066,
+      "step": 1083
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.5,
+      "learning_rate": 2.541740344132377e-06,
+      "loss": 1.8673,
+      "step": 1084
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.4765625,
+      "learning_rate": 2.5369045115557296e-06,
+      "loss": 1.8727,
+      "step": 1085
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.3828125,
+      "learning_rate": 2.5320713379297097e-06,
+      "loss": 1.8473,
+      "step": 1086
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.5703125,
+      "learning_rate": 2.52724084009516e-06,
+      "loss": 1.802,
+      "step": 1087
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.5078125,
+      "learning_rate": 2.522413034883596e-06,
+      "loss": 2.0202,
+      "step": 1088
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.3984375,
+      "learning_rate": 2.5175879391171534e-06,
+      "loss": 1.8701,
+      "step": 1089
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.4296875,
+      "learning_rate": 2.512765569608529e-06,
+      "loss": 2.0256,
+      "step": 1090
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.507945943160916e-06,
+      "loss": 1.8134,
+      "step": 1091
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.3984375,
+      "learning_rate": 2.5031290765679528e-06,
+      "loss": 1.8631,
+      "step": 1092
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.3203125,
+      "learning_rate": 2.4983149866136587e-06,
+      "loss": 1.8127,
+      "step": 1093
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.4935036900723794e-06,
+      "loss": 1.9904,
+      "step": 1094
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.5078125,
+      "learning_rate": 2.488695203708729e-06,
+      "loss": 1.7608,
+      "step": 1095
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.4140625,
+      "learning_rate": 2.4838895442775264e-06,
+      "loss": 1.8589,
+      "step": 1096
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.5625,
+      "learning_rate": 2.4790867285237423e-06,
+      "loss": 2.0151,
+      "step": 1097
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.3359375,
+      "learning_rate": 2.474286773182438e-06,
+      "loss": 1.8685,
+      "step": 1098
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.578125,
+      "learning_rate": 2.4694896949787075e-06,
+      "loss": 1.683,
+      "step": 1099
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.5,
+      "learning_rate": 2.464695510627622e-06,
+      "loss": 1.9681,
+      "step": 1100
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.3828125,
+      "learning_rate": 2.4599042368341653e-06,
+      "loss": 1.794,
+      "step": 1101
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.3828125,
+      "learning_rate": 2.455115890293184e-06,
+      "loss": 1.7639,
+      "step": 1102
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.7890625,
+      "learning_rate": 2.450330487689321e-06,
+      "loss": 1.8602,
+      "step": 1103
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.4140625,
+      "learning_rate": 2.445548045696964e-06,
+      "loss": 1.7026,
+      "step": 1104
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.375,
+      "learning_rate": 2.440768580980183e-06,
+      "loss": 1.8607,
+      "step": 1105
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.4359921101926755e-06,
+      "loss": 1.8726,
+      "step": 1106
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.431218649977705e-06,
+      "loss": 2.0619,
+      "step": 1107
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.34375,
+      "learning_rate": 2.426448216968046e-06,
+      "loss": 1.7456,
+      "step": 1108
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.4216808277859242e-06,
+      "loss": 1.8947,
+      "step": 1109
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.53125,
+      "learning_rate": 2.4169164990429614e-06,
+      "loss": 1.8927,
+      "step": 1110
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.75,
+      "learning_rate": 2.4121552473401137e-06,
+      "loss": 1.8026,
+      "step": 1111
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.407397089267614e-06,
+      "loss": 1.9866,
+      "step": 1112
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.4921875,
+      "learning_rate": 2.4026420414049195e-06,
+      "loss": 1.8591,
+      "step": 1113
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.546875,
+      "learning_rate": 2.3978901203206457e-06,
+      "loss": 2.0265,
+      "step": 1114
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.3984375,
+      "learning_rate": 2.393141342572518e-06,
+      "loss": 1.7797,
+      "step": 1115
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.859375,
+      "learning_rate": 2.388395724707305e-06,
+      "loss": 2.0544,
+      "step": 1116
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.4140625,
+      "learning_rate": 2.383653283260766e-06,
+      "loss": 1.8944,
+      "step": 1117
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.378914034757593e-06,
+      "loss": 2.0089,
+      "step": 1118
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.3515625,
+      "learning_rate": 2.3741779957113526e-06,
+      "loss": 1.7302,
+      "step": 1119
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.40625,
+      "learning_rate": 2.369445182624428e-06,
+      "loss": 1.9784,
+      "step": 1120
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.546875,
+      "learning_rate": 2.3647156119879606e-06,
+      "loss": 2.0007,
+      "step": 1121
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.5078125,
+      "learning_rate": 2.359989300281795e-06,
+      "loss": 2.016,
+      "step": 1122
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.3552662639744206e-06,
+      "loss": 1.9814,
+      "step": 1123
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.3359375,
+      "learning_rate": 2.3505465195229108e-06,
+      "loss": 1.7757,
+      "step": 1124
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.4765625,
+      "learning_rate": 2.345830083372875e-06,
+      "loss": 1.9739,
+      "step": 1125
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.734375,
+      "learning_rate": 2.3411169719583887e-06,
+      "loss": 1.9702,
+      "step": 1126
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.3364072017019467e-06,
+      "loss": 1.7358,
+      "step": 1127
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.328125,
+      "learning_rate": 2.331700789014399e-06,
+      "loss": 1.8713,
+      "step": 1128
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.421875,
+      "learning_rate": 2.326997750294898e-06,
+      "loss": 1.9347,
+      "step": 1129
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.3359375,
+      "learning_rate": 2.32229810193084e-06,
+      "loss": 1.6961,
+      "step": 1130
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.34375,
+      "learning_rate": 2.317601860297807e-06,
+      "loss": 1.7551,
+      "step": 1131
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.4296875,
+      "learning_rate": 2.3129090417595107e-06,
+      "loss": 1.998,
+      "step": 1132
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.3359375,
+      "learning_rate": 2.308219662667736e-06,
+      "loss": 1.8426,
+      "step": 1133
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.3515625,
+      "learning_rate": 2.3035337393622816e-06,
+      "loss": 1.9068,
+      "step": 1134
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.2988512881709063e-06,
+      "loss": 1.9824,
+      "step": 1135
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.29417232540927e-06,
+      "loss": 1.8839,
+      "step": 1136
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.3984375,
+      "learning_rate": 2.289496867380878e-06,
+      "loss": 1.7661,
+      "step": 1137
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.578125,
+      "learning_rate": 2.284824930377023e-06,
+      "loss": 1.9512,
+      "step": 1138
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.3671875,
+      "learning_rate": 2.280156530676727e-06,
+      "loss": 1.9426,
+      "step": 1139
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.8203125,
+      "learning_rate": 2.275491684546692e-06,
+      "loss": 2.1988,
+      "step": 1140
+    },
+    {
+      "epoch": 1.92,
+      "eval_loss": 1.9526869058609009,
+      "eval_runtime": 72.6995,
+      "eval_samples_per_second": 8.927,
+      "eval_steps_per_second": 8.927,
+      "step": 1140
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.421875,
+      "learning_rate": 2.270830408241233e-06,
+      "loss": 1.7793,
+      "step": 1141
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.40625,
+      "learning_rate": 2.266172718002227e-06,
+      "loss": 1.838,
+      "step": 1142
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.484375,
+      "learning_rate": 2.2615186300590567e-06,
+      "loss": 1.6998,
+      "step": 1143
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.234375,
+      "learning_rate": 2.2568681606285527e-06,
+      "loss": 1.8002,
+      "step": 1144
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.252221325914937e-06,
+      "loss": 2.073,
+      "step": 1145
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.2475781421097657e-06,
+      "loss": 1.887,
+      "step": 1146
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.3984375,
+      "learning_rate": 2.242938625391876e-06,
+      "loss": 1.7672,
+      "step": 1147
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.4453125,
+      "learning_rate": 2.238302791927324e-06,
+      "loss": 1.8141,
+      "step": 1148
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.4296875,
+      "learning_rate": 2.2336706578693345e-06,
+      "loss": 1.8366,
+      "step": 1149
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.2290422393582413e-06,
+      "loss": 1.9289,
+      "step": 1150
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.484375,
+      "learning_rate": 2.2244175525214313e-06,
+      "loss": 2.059,
+      "step": 1151
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.421875,
+      "learning_rate": 2.219796613473289e-06,
+      "loss": 1.8753,
+      "step": 1152
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.4609375,
+      "learning_rate": 2.215179438315139e-06,
+      "loss": 1.774,
+      "step": 1153
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.210566043135192e-06,
+      "loss": 1.8905,
+      "step": 1154
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.3671875,
+      "learning_rate": 2.2059564440084882e-06,
+      "loss": 1.6761,
+      "step": 1155
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.3828125,
+      "learning_rate": 2.201350656996839e-06,
+      "loss": 1.8904,
+      "step": 1156
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.4453125,
+      "learning_rate": 2.1967486981487753e-06,
+      "loss": 1.8736,
+      "step": 1157
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.4296875,
+      "learning_rate": 2.1921505834994855e-06,
+      "loss": 1.8248,
+      "step": 1158
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.875,
+      "learning_rate": 2.187556329070765e-06,
+      "loss": 1.9285,
+      "step": 1159
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.546875,
+      "learning_rate": 2.182965950870959e-06,
+      "loss": 1.9256,
+      "step": 1160
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.5234375,
+      "learning_rate": 2.178379464894907e-06,
+      "loss": 1.9867,
+      "step": 1161
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.28125,
+      "learning_rate": 2.173796887123885e-06,
+      "loss": 1.7247,
+      "step": 1162
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.3984375,
+      "learning_rate": 2.1692182335255495e-06,
+      "loss": 1.9383,
+      "step": 1163
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.4921875,
+      "learning_rate": 2.1646435200538856e-06,
+      "loss": 1.7344,
+      "step": 1164
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.375,
+      "learning_rate": 2.160072762649151e-06,
+      "loss": 2.015,
+      "step": 1165
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.671875,
+      "learning_rate": 2.155505977237816e-06,
+      "loss": 2.136,
+      "step": 1166
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.3046875,
+      "learning_rate": 2.150943179732511e-06,
+      "loss": 1.8755,
+      "step": 1167
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.1463843860319723e-06,
+      "loss": 1.8817,
+      "step": 1168
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.28125,
+      "learning_rate": 2.1418296120209825e-06,
+      "loss": 1.8505,
+      "step": 1169
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.3984375,
+      "learning_rate": 2.1372788735703227e-06,
+      "loss": 1.814,
+      "step": 1170
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.132732186536708e-06,
+      "loss": 1.9639,
+      "step": 1171
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.375,
+      "learning_rate": 2.1281895667627374e-06,
+      "loss": 1.7337,
+      "step": 1172
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.5546875,
+      "learning_rate": 2.1236510300768383e-06,
+      "loss": 1.9272,
+      "step": 1173
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.1191165922932134e-06,
+      "loss": 1.9235,
+      "step": 1174
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.375,
+      "learning_rate": 2.11458626921178e-06,
+      "loss": 2.0302,
+      "step": 1175
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.5,
+      "learning_rate": 2.110060076618117e-06,
+      "loss": 2.0152,
+      "step": 1176
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.484375,
+      "learning_rate": 2.105538030283414e-06,
+      "loss": 1.9395,
+      "step": 1177
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.1010201459644097e-06,
+      "loss": 1.9026,
+      "step": 1178
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.096506439403347e-06,
+      "loss": 2.0946,
+      "step": 1179
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.671875,
+      "learning_rate": 2.091996926327905e-06,
+      "loss": 1.8123,
+      "step": 1180
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.546875,
+      "learning_rate": 2.087491622451154e-06,
+      "loss": 2.0841,
+      "step": 1181
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.5859375,
+      "learning_rate": 2.082990543471497e-06,
+      "loss": 1.8813,
+      "step": 1182
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.5078125,
+      "learning_rate": 2.0784937050726136e-06,
+      "loss": 1.8978,
+      "step": 1183
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.2890625,
+      "learning_rate": 2.0740011229234137e-06,
+      "loss": 1.7163,
+      "step": 1184
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.0695128126779696e-06,
+      "loss": 1.9937,
+      "step": 1185
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.0650287899754726e-06,
+      "loss": 2.1517,
+      "step": 1186
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.3671875,
+      "learning_rate": 2.060549070440172e-06,
+      "loss": 1.7763,
+      "step": 1187
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.5859375,
+      "learning_rate": 2.056073669681324e-06,
+      "loss": 1.9543,
+      "step": 1188
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.78125,
+      "learning_rate": 2.051602603293139e-06,
+      "loss": 1.8608,
+      "step": 1189
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.0471358868547196e-06,
+      "loss": 1.9044,
+      "step": 1190
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.359375,
+      "learning_rate": 2.0426735359300136e-06,
+      "loss": 1.861,
+      "step": 1191
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.390625,
+      "learning_rate": 2.038215566067759e-06,
+      "loss": 1.9506,
+      "step": 1192
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.3828125,
+      "learning_rate": 2.033761992801425e-06,
+      "loss": 1.9469,
+      "step": 1193
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.029312831649165e-06,
+      "loss": 1.8447,
+      "step": 1194
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.4765625,
+      "learning_rate": 2.0248680981137574e-06,
+      "loss": 1.9012,
+      "step": 1195
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.0204278076825505e-06,
+      "loss": 1.9715,
+      "step": 1196
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.53125,
+      "learning_rate": 2.0159919758274136e-06,
+      "loss": 2.2625,
+      "step": 1197
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.34375,
+      "learning_rate": 2.011560618004679e-06,
+      "loss": 1.9621,
+      "step": 1198
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.5234375,
+      "learning_rate": 2.0071337496550908e-06,
+      "loss": 2.0286,
+      "step": 1199
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.4765625,
+      "learning_rate": 2.00271138620375e-06,
+      "loss": 1.7899,
+      "step": 1200
+    },
+    {
+      "epoch": 2.01,
+      "eval_loss": 1.952858805656433,
+      "eval_runtime": 72.6356,
+      "eval_samples_per_second": 8.935,
+      "eval_steps_per_second": 8.935,
+      "step": 1200
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.998293543060058e-06,
+      "loss": 1.9355,
+      "step": 1201
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9938802356176688e-06,
+      "loss": 1.8697,
+      "step": 1202
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.989471479254429e-06,
+      "loss": 1.7158,
+      "step": 1203
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.985067289332332e-06,
+      "loss": 1.8872,
+      "step": 1204
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.9806676811974535e-06,
+      "loss": 2.0431,
+      "step": 1205
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.976272670179909e-06,
+      "loss": 1.8193,
+      "step": 1206
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9718822715937937e-06,
+      "loss": 1.9256,
+      "step": 1207
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.9674965007371302e-06,
+      "loss": 1.8832,
+      "step": 1208
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.96311537289182e-06,
+      "loss": 1.8657,
+      "step": 1209
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.958738903323583e-06,
+      "loss": 1.9369,
+      "step": 1210
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.954367107281908e-06,
+      "loss": 2.0463,
+      "step": 1211
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.950000000000001e-06,
+      "loss": 1.9903,
+      "step": 1212
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9456375966947263e-06,
+      "loss": 1.8925,
+      "step": 1213
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.9412799125665633e-06,
+      "loss": 1.7835,
+      "step": 1214
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.936926962799544e-06,
+      "loss": 1.9661,
+      "step": 1215
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9325787625612044e-06,
+      "loss": 1.5986,
+      "step": 1216
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.9282353270025313e-06,
+      "loss": 1.8952,
+      "step": 1217
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.92389667125791e-06,
+      "loss": 1.9931,
+      "step": 1218
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.9195628104450694e-06,
+      "loss": 1.9302,
+      "step": 1219
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.9152337596650327e-06,
+      "loss": 1.7964,
+      "step": 1220
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.910909534002061e-06,
+      "loss": 1.8921,
+      "step": 1221
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9065901485236046e-06,
+      "loss": 1.8691,
+      "step": 1222
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.9022756182802456e-06,
+      "loss": 1.7877,
+      "step": 1223
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8979659583056525e-06,
+      "loss": 1.9212,
+      "step": 1224
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.8936611836165204e-06,
+      "loss": 1.7434,
+      "step": 1225
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8893613092125233e-06,
+      "loss": 1.6815,
+      "step": 1226
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.8850663500762597e-06,
+      "loss": 1.7606,
+      "step": 1227
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8807763211732013e-06,
+      "loss": 1.9047,
+      "step": 1228
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.8764912374516437e-06,
+      "loss": 1.8633,
+      "step": 1229
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8722111138426466e-06,
+      "loss": 1.9066,
+      "step": 1230
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.8679359652599896e-06,
+      "loss": 1.8429,
+      "step": 1231
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.8636658066001164e-06,
+      "loss": 1.9299,
+      "step": 1232
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.8594006527420838e-06,
+      "loss": 1.8668,
+      "step": 1233
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.8551405185475094e-06,
+      "loss": 1.778,
+      "step": 1234
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8508854188605209e-06,
+      "loss": 1.8144,
+      "step": 1235
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.846635368507702e-06,
+      "loss": 1.7307,
+      "step": 1236
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.8423903822980448e-06,
+      "loss": 1.9232,
+      "step": 1237
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.8381504750228928e-06,
+      "loss": 1.993,
+      "step": 1238
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.8339156614558957e-06,
+      "loss": 1.8617,
+      "step": 1239
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.8296859563529524e-06,
+      "loss": 1.7955,
+      "step": 1240
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.825461374452162e-06,
+      "loss": 1.8889,
+      "step": 1241
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8212419304737717e-06,
+      "loss": 1.8679,
+      "step": 1242
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.8170276391201263e-06,
+      "loss": 2.0214,
+      "step": 1243
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.8128185150756188e-06,
+      "loss": 2.0213,
+      "step": 1244
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8086145730066346e-06,
+      "loss": 1.7442,
+      "step": 1245
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.8044158275615025e-06,
+      "loss": 1.9483,
+      "step": 1246
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.800222293370446e-06,
+      "loss": 1.8626,
+      "step": 1247
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.7960339850455288e-06,
+      "loss": 1.8822,
+      "step": 1248
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.7918509171806064e-06,
+      "loss": 1.6832,
+      "step": 1249
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.7876731043512742e-06,
+      "loss": 1.9366,
+      "step": 1250
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.7835005611148163e-06,
+      "loss": 1.7482,
+      "step": 1251
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.7793333020101554e-06,
+      "loss": 1.926,
+      "step": 1252
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.7751713415578018e-06,
+      "loss": 1.6752,
+      "step": 1253
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.7710146942598051e-06,
+      "loss": 1.7899,
+      "step": 1254
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.7668633745996988e-06,
+      "loss": 1.8611,
+      "step": 1255
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.7627173970424541e-06,
+      "loss": 1.9212,
+      "step": 1256
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.7585767760344273e-06,
+      "loss": 1.803,
+      "step": 1257
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.75444152600331e-06,
+      "loss": 1.9312,
+      "step": 1258
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.750311661358081e-06,
+      "loss": 1.8898,
+      "step": 1259
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.5,
+      "learning_rate": 1.7461871964889528e-06,
+      "loss": 1.8785,
+      "step": 1260
+    },
+    {
+      "epoch": 2.11,
+      "eval_loss": 1.9536255598068237,
+      "eval_runtime": 72.8027,
+      "eval_samples_per_second": 8.915,
+      "eval_steps_per_second": 8.915,
+      "step": 1260
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.7420681457673218e-06,
+      "loss": 1.8501,
+      "step": 1261
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.73795452354572e-06,
+      "loss": 2.0064,
+      "step": 1262
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.7338463441577654e-06,
+      "loss": 1.9383,
+      "step": 1263
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.7297436219181084e-06,
+      "loss": 1.8458,
+      "step": 1264
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.7256463711223872e-06,
+      "loss": 1.935,
+      "step": 1265
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.7215546060471724e-06,
+      "loss": 1.755,
+      "step": 1266
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.7174683409499212e-06,
+      "loss": 1.8595,
+      "step": 1267
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.7133875900689265e-06,
+      "loss": 1.8653,
+      "step": 1268
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.7093123676232692e-06,
+      "loss": 1.7989,
+      "step": 1269
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.375,
+      "learning_rate": 1.7052426878127634e-06,
+      "loss": 1.8884,
+      "step": 1270
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.5,
+      "learning_rate": 1.7011785648179122e-06,
+      "loss": 1.8548,
+      "step": 1271
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.6971200127998564e-06,
+      "loss": 1.8745,
+      "step": 1272
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.34375,
+      "learning_rate": 1.693067045900323e-06,
+      "loss": 1.7291,
+      "step": 1273
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.6890196782415828e-06,
+      "loss": 1.9105,
+      "step": 1274
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.684977923926392e-06,
+      "loss": 1.9538,
+      "step": 1275
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.6809417970379485e-06,
+      "loss": 1.8723,
+      "step": 1276
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.34375,
+      "learning_rate": 1.676911311639843e-06,
+      "loss": 1.9465,
+      "step": 1277
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.672886481776008e-06,
+      "loss": 1.8716,
+      "step": 1278
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.6688673214706703e-06,
+      "loss": 1.8797,
+      "step": 1279
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.6648538447283005e-06,
+      "loss": 1.8554,
+      "step": 1280
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.6608460655335666e-06,
+      "loss": 2.0352,
+      "step": 1281
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.6568439978512823e-06,
+      "loss": 1.9972,
+      "step": 1282
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.6528476556263606e-06,
+      "loss": 1.8229,
+      "step": 1283
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.6488570527837662e-06,
+      "loss": 1.7553,
+      "step": 1284
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.375,
+      "learning_rate": 1.6448722032284636e-06,
+      "loss": 1.9446,
+      "step": 1285
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.6408931208453699e-06,
+      "loss": 1.9821,
+      "step": 1286
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.6369198194993087e-06,
+      "loss": 1.8894,
+      "step": 1287
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.6329523130349573e-06,
+      "loss": 1.66,
+      "step": 1288
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.6289906152768046e-06,
+      "loss": 1.8564,
+      "step": 1289
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.6250347400290976e-06,
+      "loss": 1.9308,
+      "step": 1290
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.6210847010757952e-06,
+      "loss": 1.9478,
+      "step": 1291
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.61714051218052e-06,
+      "loss": 1.9453,
+      "step": 1292
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.6132021870865111e-06,
+      "loss": 1.977,
+      "step": 1293
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.609269739516575e-06,
+      "loss": 1.7592,
+      "step": 1294
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.6053431831730393e-06,
+      "loss": 1.9182,
+      "step": 1295
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.6014225317377031e-06,
+      "loss": 1.8977,
+      "step": 1296
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.5975077988717903e-06,
+      "loss": 1.6754,
+      "step": 1297
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.5935989982159013e-06,
+      "loss": 2.006,
+      "step": 1298
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.5896961433899705e-06,
+      "loss": 2.0322,
+      "step": 1299
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.5857992479932086e-06,
+      "loss": 1.8969,
+      "step": 1300
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5819083256040646e-06,
+      "loss": 1.7744,
+      "step": 1301
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.5780233897801735e-06,
+      "loss": 1.7507,
+      "step": 1302
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.5741444540583106e-06,
+      "loss": 1.7593,
+      "step": 1303
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.6796875,
+      "learning_rate": 1.5702715319543467e-06,
+      "loss": 2.0022,
+      "step": 1304
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.566404636963195e-06,
+      "loss": 2.1224,
+      "step": 1305
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.5625437825587703e-06,
+      "loss": 1.9616,
+      "step": 1306
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.375,
+      "learning_rate": 1.5586889821939378e-06,
+      "loss": 1.9013,
+      "step": 1307
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5548402493004688e-06,
+      "loss": 1.7741,
+      "step": 1308
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.625,
+      "learning_rate": 1.5509975972889925e-06,
+      "loss": 1.8791,
+      "step": 1309
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.5471610395489502e-06,
+      "loss": 1.8647,
+      "step": 1310
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.5433305894485472e-06,
+      "loss": 1.8453,
+      "step": 1311
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.539506260334708e-06,
+      "loss": 1.8188,
+      "step": 1312
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.6953125,
+      "learning_rate": 1.535688065533028e-06,
+      "loss": 1.9197,
+      "step": 1313
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.5318760183477305e-06,
+      "loss": 1.9553,
+      "step": 1314
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.528070132061615e-06,
+      "loss": 1.9533,
+      "step": 1315
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.375,
+      "learning_rate": 1.5242704199360158e-06,
+      "loss": 1.7673,
+      "step": 1316
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.520476895210752e-06,
+      "loss": 2.0074,
+      "step": 1317
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.5166895711040836e-06,
+      "loss": 1.9309,
+      "step": 1318
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.512908460812667e-06,
+      "loss": 1.8433,
+      "step": 1319
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.296875,
+      "learning_rate": 1.5091335775115052e-06,
+      "loss": 1.7208,
+      "step": 1320
+    },
+    {
+      "epoch": 2.22,
+      "eval_loss": 1.9534618854522705,
+      "eval_runtime": 72.4867,
+      "eval_samples_per_second": 8.953,
+      "eval_steps_per_second": 8.953,
+      "step": 1320
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.5053649343539036e-06,
+      "loss": 1.9372,
+      "step": 1321
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5016025444714247e-06,
+      "loss": 1.948,
+      "step": 1322
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.497846420973842e-06,
+      "loss": 1.9946,
+      "step": 1323
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.4940965769490942e-06,
+      "loss": 2.0408,
+      "step": 1324
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.4903530254632393e-06,
+      "loss": 1.8153,
+      "step": 1325
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.625,
+      "learning_rate": 1.4866157795604095e-06,
+      "loss": 1.8674,
+      "step": 1326
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.4828848522627667e-06,
+      "loss": 1.8996,
+      "step": 1327
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.4791602565704536e-06,
+      "loss": 1.9489,
+      "step": 1328
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.4754420054615543e-06,
+      "loss": 2.0395,
+      "step": 1329
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.4717301118920437e-06,
+      "loss": 1.8873,
+      "step": 1330
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4680245887957436e-06,
+      "loss": 1.7733,
+      "step": 1331
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.4643254490842798e-06,
+      "loss": 1.826,
+      "step": 1332
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.4606327056470336e-06,
+      "loss": 2.0092,
+      "step": 1333
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.456946371351103e-06,
+      "loss": 1.9064,
+      "step": 1334
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.4532664590412498e-06,
+      "loss": 2.0053,
+      "step": 1335
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.4495929815398604e-06,
+      "loss": 1.8985,
+      "step": 1336
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4459259516468998e-06,
+      "loss": 1.666,
+      "step": 1337
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.442265382139866e-06,
+      "loss": 1.8388,
+      "step": 1338
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.4386112857737473e-06,
+      "loss": 1.9407,
+      "step": 1339
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.4349636752809757e-06,
+      "loss": 1.8795,
+      "step": 1340
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.2265625,
+      "learning_rate": 1.4313225633713843e-06,
+      "loss": 1.9593,
+      "step": 1341
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4276879627321624e-06,
+      "loss": 1.7267,
+      "step": 1342
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.42405988602781e-06,
+      "loss": 1.9268,
+      "step": 1343
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.4204383459000984e-06,
+      "loss": 1.8601,
+      "step": 1344
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.4168233549680186e-06,
+      "loss": 1.9448,
+      "step": 1345
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4132149258277436e-06,
+      "loss": 1.9202,
+      "step": 1346
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4096130710525814e-06,
+      "loss": 1.9888,
+      "step": 1347
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.4060178031929324e-06,
+      "loss": 1.9818,
+      "step": 1348
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.4024291347762465e-06,
+      "loss": 1.7319,
+      "step": 1349
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.3988470783069768e-06,
+      "loss": 1.8546,
+      "step": 1350
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.3952716462665378e-06,
+      "loss": 1.7922,
+      "step": 1351
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.3917028511132623e-06,
+      "loss": 1.7291,
+      "step": 1352
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.3881407052823564e-06,
+      "loss": 1.894,
+      "step": 1353
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.3845852211858582e-06,
+      "loss": 1.9447,
+      "step": 1354
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.381036411212593e-06,
+      "loss": 1.8265,
+      "step": 1355
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.3774942877281303e-06,
+      "loss": 1.837,
+      "step": 1356
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.3739588630747414e-06,
+      "loss": 1.8976,
+      "step": 1357
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.3704301495713553e-06,
+      "loss": 1.9525,
+      "step": 1358
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.3669081595135182e-06,
+      "loss": 1.8708,
+      "step": 1359
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.3633929051733473e-06,
+      "loss": 1.9955,
+      "step": 1360
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.3598843987994905e-06,
+      "loss": 1.8324,
+      "step": 1361
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.3563826526170821e-06,
+      "loss": 1.9427,
+      "step": 1362
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.352887678827701e-06,
+      "loss": 2.0053,
+      "step": 1363
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.3493994896093306e-06,
+      "loss": 1.9065,
+      "step": 1364
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.3459180971163105e-06,
+      "loss": 1.9167,
+      "step": 1365
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.3424435134793002e-06,
+      "loss": 1.772,
+      "step": 1366
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.7109375,
+      "learning_rate": 1.338975750805232e-06,
+      "loss": 2.0155,
+      "step": 1367
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.3355148211772732e-06,
+      "loss": 1.9183,
+      "step": 1368
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.3320607366547808e-06,
+      "loss": 1.8341,
+      "step": 1369
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.3286135092732599e-06,
+      "loss": 1.9931,
+      "step": 1370
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.3251731510443243e-06,
+      "loss": 2.027,
+      "step": 1371
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.3217396739556507e-06,
+      "loss": 1.8582,
+      "step": 1372
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.3183130899709393e-06,
+      "loss": 1.8663,
+      "step": 1373
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.3148934110298738e-06,
+      "loss": 1.8957,
+      "step": 1374
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.311480649048076e-06,
+      "loss": 1.7826,
+      "step": 1375
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.308074815917066e-06,
+      "loss": 1.7203,
+      "step": 1376
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.5,
+      "learning_rate": 1.304675923504221e-06,
+      "loss": 1.9513,
+      "step": 1377
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.301283983652733e-06,
+      "loss": 2.0152,
+      "step": 1378
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.2978990081815716e-06,
+      "loss": 1.9463,
+      "step": 1379
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.294521008885436e-06,
+      "loss": 1.9643,
+      "step": 1380
+    },
+    {
+      "epoch": 2.32,
+      "eval_loss": 1.953567385673523,
+      "eval_runtime": 72.5011,
+      "eval_samples_per_second": 8.952,
+      "eval_steps_per_second": 8.952,
+      "step": 1380
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.2911499975347184e-06,
+      "loss": 2.1475,
+      "step": 1381
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.2877859858754626e-06,
+      "loss": 1.8881,
+      "step": 1382
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.284428985629322e-06,
+      "loss": 1.7487,
+      "step": 1383
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.281079008493519e-06,
+      "loss": 1.891,
+      "step": 1384
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.2777360661408053e-06,
+      "loss": 1.8939,
+      "step": 1385
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.2744001702194199e-06,
+      "loss": 2.0425,
+      "step": 1386
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.625,
+      "learning_rate": 1.2710713323530486e-06,
+      "loss": 2.2004,
+      "step": 1387
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.2677495641407838e-06,
+      "loss": 1.6771,
+      "step": 1388
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.264434877157087e-06,
+      "loss": 1.9232,
+      "step": 1389
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.2611272829517423e-06,
+      "loss": 2.0209,
+      "step": 1390
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2578267930498205e-06,
+      "loss": 2.027,
+      "step": 1391
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.254533418951638e-06,
+      "loss": 1.9216,
+      "step": 1392
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2512471721327165e-06,
+      "loss": 1.8335,
+      "step": 1393
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.2479680640437454e-06,
+      "loss": 1.9975,
+      "step": 1394
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.2446961061105367e-06,
+      "loss": 1.7855,
+      "step": 1395
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.2414313097339893e-06,
+      "loss": 1.8574,
+      "step": 1396
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.2381736862900485e-06,
+      "loss": 2.0234,
+      "step": 1397
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.2349232471296659e-06,
+      "loss": 1.8838,
+      "step": 1398
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.2316800035787598e-06,
+      "loss": 1.9116,
+      "step": 1399
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.2284439669381758e-06,
+      "loss": 2.0191,
+      "step": 1400
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.2252151484836484e-06,
+      "loss": 1.9649,
+      "step": 1401
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.22199355946576e-06,
+      "loss": 1.8811,
+      "step": 1402
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.2187792111099016e-06,
+      "loss": 1.9588,
+      "step": 1403
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.2155721146162376e-06,
+      "loss": 1.8881,
+      "step": 1404
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.212372281159662e-06,
+      "loss": 1.7173,
+      "step": 1405
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.20917972188976e-06,
+      "loss": 1.8664,
+      "step": 1406
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.205994447930773e-06,
+      "loss": 1.9208,
+      "step": 1407
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.2028164703815544e-06,
+      "loss": 1.9237,
+      "step": 1408
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.1996458003155372e-06,
+      "loss": 1.8487,
+      "step": 1409
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.625,
+      "learning_rate": 1.19648244878069e-06,
+      "loss": 1.9926,
+      "step": 1410
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.34375,
+      "learning_rate": 1.1933264267994795e-06,
+      "loss": 1.8457,
+      "step": 1411
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.1901777453688357e-06,
+      "loss": 1.9027,
+      "step": 1412
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.5,
+      "learning_rate": 1.1870364154601086e-06,
+      "loss": 2.1226,
+      "step": 1413
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.183902448019034e-06,
+      "loss": 1.9132,
+      "step": 1414
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.375,
+      "learning_rate": 1.180775853965693e-06,
+      "loss": 1.7712,
+      "step": 1415
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.1776566441944746e-06,
+      "loss": 1.8856,
+      "step": 1416
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.1745448295740368e-06,
+      "loss": 2.0439,
+      "step": 1417
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1714404209472736e-06,
+      "loss": 1.8875,
+      "step": 1418
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.375,
+      "learning_rate": 1.1683434291312685e-06,
+      "loss": 1.7842,
+      "step": 1419
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.165253864917265e-06,
+      "loss": 1.8657,
+      "step": 1420
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.1621717390706236e-06,
+      "loss": 2.0253,
+      "step": 1421
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.159097062330787e-06,
+      "loss": 1.8576,
+      "step": 1422
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.1560298454112436e-06,
+      "loss": 1.9776,
+      "step": 1423
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.1529700989994868e-06,
+      "loss": 1.9832,
+      "step": 1424
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.1499178337569794e-06,
+      "loss": 2.0509,
+      "step": 1425
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.146873060319118e-06,
+      "loss": 1.9491,
+      "step": 1426
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.1438357892951932e-06,
+      "loss": 1.9559,
+      "step": 1427
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.140806031268355e-06,
+      "loss": 1.8652,
+      "step": 1428
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.137783796795574e-06,
+      "loss": 1.7705,
+      "step": 1429
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.1347690964076065e-06,
+      "loss": 1.8483,
+      "step": 1430
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.131761940608956e-06,
+      "loss": 1.9094,
+      "step": 1431
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.1287623398778372e-06,
+      "loss": 1.8403,
+      "step": 1432
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.1257703046661418e-06,
+      "loss": 1.778,
+      "step": 1433
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.1227858453993977e-06,
+      "loss": 2.0312,
+      "step": 1434
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.1198089724767356e-06,
+      "loss": 2.0088,
+      "step": 1435
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.1168396962708524e-06,
+      "loss": 1.8605,
+      "step": 1436
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.113878027127974e-06,
+      "loss": 1.9716,
+      "step": 1437
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.1109239753678222e-06,
+      "loss": 1.9367,
+      "step": 1438
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.5,
+      "learning_rate": 1.1079775512835748e-06,
+      "loss": 1.9521,
+      "step": 1439
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.1050387651418307e-06,
+      "loss": 1.9099,
+      "step": 1440
+    },
+    {
+      "epoch": 2.42,
+      "eval_loss": 1.953331470489502,
+      "eval_runtime": 72.686,
+      "eval_samples_per_second": 8.929,
+      "eval_steps_per_second": 8.929,
+      "step": 1440
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.1021076271825771e-06,
+      "loss": 1.7965,
+      "step": 1441
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.0991841476191504e-06,
+      "loss": 1.8824,
+      "step": 1442
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0962683366382017e-06,
+      "loss": 1.8532,
+      "step": 1443
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.0933602043996624e-06,
+      "loss": 1.9575,
+      "step": 1444
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.090459761036707e-06,
+      "loss": 1.717,
+      "step": 1445
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.08756701665572e-06,
+      "loss": 1.8202,
+      "step": 1446
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.0846819813362563e-06,
+      "loss": 2.0111,
+      "step": 1447
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.081804665131014e-06,
+      "loss": 1.9468,
+      "step": 1448
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0789350780657898e-06,
+      "loss": 2.1138,
+      "step": 1449
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.0760732301394517e-06,
+      "loss": 1.903,
+      "step": 1450
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.0732191313239e-06,
+      "loss": 1.7584,
+      "step": 1451
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.375,
+      "learning_rate": 1.0703727915640331e-06,
+      "loss": 1.8592,
+      "step": 1452
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.0675342207777166e-06,
+      "loss": 1.8527,
+      "step": 1453
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.0647034288557416e-06,
+      "loss": 1.8394,
+      "step": 1454
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.0618804256617976e-06,
+      "loss": 2.0342,
+      "step": 1455
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.059065221032433e-06,
+      "loss": 1.8342,
+      "step": 1456
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0562578247770232e-06,
+      "loss": 1.7947,
+      "step": 1457
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.0534582466777362e-06,
+      "loss": 1.8489,
+      "step": 1458
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.0506664964894971e-06,
+      "loss": 1.9293,
+      "step": 1459
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.0478825839399567e-06,
+      "loss": 1.7394,
+      "step": 1460
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.0451065187294553e-06,
+      "loss": 1.6778,
+      "step": 1461
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.042338310530989e-06,
+      "loss": 1.7995,
+      "step": 1462
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.0395779689901793e-06,
+      "loss": 1.9903,
+      "step": 1463
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.036825503725234e-06,
+      "loss": 1.7324,
+      "step": 1464
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.0340809243269183e-06,
+      "loss": 1.8404,
+      "step": 1465
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.0313442403585187e-06,
+      "loss": 1.8433,
+      "step": 1466
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.0286154613558106e-06,
+      "loss": 1.8111,
+      "step": 1467
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.025894596827027e-06,
+      "loss": 1.9719,
+      "step": 1468
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.023181656252821e-06,
+      "loss": 1.8663,
+      "step": 1469
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.020476649086237e-06,
+      "loss": 1.6536,
+      "step": 1470
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.0177795847526752e-06,
+      "loss": 1.7711,
+      "step": 1471
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.0150904726498593e-06,
+      "loss": 2.0681,
+      "step": 1472
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.0124093221478047e-06,
+      "loss": 1.9075,
+      "step": 1473
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.0097361425887852e-06,
+      "loss": 1.9701,
+      "step": 1474
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.3046875,
+      "learning_rate": 1.0070709432873005e-06,
+      "loss": 1.7582,
+      "step": 1475
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.0044137335300421e-06,
+      "loss": 2.0247,
+      "step": 1476
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.5,
+      "learning_rate": 1.0017645225758648e-06,
+      "loss": 1.9363,
+      "step": 1477
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.99123319655751e-07,
+      "loss": 2.0171,
+      "step": 1478
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.9765625,
+      "learning_rate": 9.9649013397278e-07,
+      "loss": 1.8441,
+      "step": 1479
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 1.4140625,
+      "learning_rate": 9.938649747020956e-07,
+      "loss": 1.8559,
+      "step": 1480
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1773,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 148,
+  "total_flos": 8.73512712108245e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}