diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7033 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.0853342435652647,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 8.53342435652647e-05,
+      "grad_norm": 0.8203125,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.8,
+      "step": 1
+    },
+    {
+      "epoch": 0.0001706684871305294,
+      "grad_norm": 0.9140625,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.8859,
+      "step": 2
+    },
+    {
+      "epoch": 0.0002560027306957941,
+      "grad_norm": 0.90625,
+      "learning_rate": 3e-06,
+      "loss": 1.7006,
+      "step": 3
+    },
+    {
+      "epoch": 0.0003413369742610588,
+      "grad_norm": 0.76171875,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.6897,
+      "step": 4
+    },
+    {
+      "epoch": 0.0004266712178263235,
+      "grad_norm": 0.77734375,
+      "learning_rate": 5e-06,
+      "loss": 1.6493,
+      "step": 5
+    },
+    {
+      "epoch": 0.0005120054613915882,
+      "grad_norm": 0.80859375,
+      "learning_rate": 6e-06,
+      "loss": 1.7298,
+      "step": 6
+    },
+    {
+      "epoch": 0.0005973397049568529,
+      "grad_norm": 0.82421875,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 1.7985,
+      "step": 7
+    },
+    {
+      "epoch": 0.0006826739485221176,
+      "grad_norm": 0.875,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.6986,
+      "step": 8
+    },
+    {
+      "epoch": 0.0007680081920873823,
+      "grad_norm": 0.8671875,
+      "learning_rate": 9e-06,
+      "loss": 1.9523,
+      "step": 9
+    },
+    {
+      "epoch": 0.000853342435652647,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1e-05,
+      "loss": 1.8498,
+      "step": 10
+    },
+    {
+      "epoch": 0.0009386766792179117,
+      "grad_norm": 0.921875,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.6866,
+      "step": 11
+    },
+    {
+      "epoch": 0.0010240109227831764,
+      "grad_norm": 0.90625,
+      "learning_rate": 1.2e-05,
+      "loss": 2.0208,
+      "step": 12
+    },
+    {
+      "epoch": 0.001109345166348441,
+      "grad_norm": 0.8046875,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.7539,
+      "step": 13
+    },
+    {
+      "epoch": 0.0011946794099137058,
+      "grad_norm": 0.921875,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 1.8917,
+      "step": 14
+    },
+    {
+      "epoch": 0.0012800136534789704,
+      "grad_norm": 0.890625,
+      "learning_rate": 1.5e-05,
+      "loss": 1.7578,
+      "step": 15
+    },
+    {
+      "epoch": 0.0013653478970442352,
+      "grad_norm": 0.84375,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.6821,
+      "step": 16
+    },
+    {
+      "epoch": 0.0014506821406094998,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 1.9614,
+      "step": 17
+    },
+    {
+      "epoch": 0.0015360163841747646,
+      "grad_norm": 0.86328125,
+      "learning_rate": 1.8e-05,
+      "loss": 1.693,
+      "step": 18
+    },
+    {
+      "epoch": 0.0016213506277400291,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.9e-05,
+      "loss": 1.5387,
+      "step": 19
+    },
+    {
+      "epoch": 0.001706684871305294,
+      "grad_norm": 0.8515625,
+      "learning_rate": 2e-05,
+      "loss": 1.8614,
+      "step": 20
+    },
+    {
+      "epoch": 0.0017920191148705585,
+      "grad_norm": 0.83984375,
+      "learning_rate": 2.1e-05,
+      "loss": 1.713,
+      "step": 21
+    },
+    {
+      "epoch": 0.0018773533584358233,
+      "grad_norm": 0.80859375,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.8531,
+      "step": 22
+    },
+    {
+      "epoch": 0.001962687602001088,
+      "grad_norm": 0.80078125,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 2.006,
+      "step": 23
+    },
+    {
+      "epoch": 0.0020480218455663527,
+      "grad_norm": 0.79296875,
+      "learning_rate": 2.4e-05,
+      "loss": 1.6356,
+      "step": 24
+    },
+    {
+      "epoch": 0.0021333560891316173,
+      "grad_norm": 0.984375,
+      "learning_rate": 2.5e-05,
+      "loss": 1.7681,
+      "step": 25
+    },
+    {
+      "epoch": 0.002218690332696882,
+      "grad_norm": 0.81640625,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 2.0056,
+      "step": 26
+    },
+    {
+      "epoch": 0.002304024576262147,
+      "grad_norm": 0.9453125,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 1.9787,
+      "step": 27
+    },
+    {
+      "epoch": 0.0023893588198274115,
+      "grad_norm": 0.85546875,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 1.7908,
+      "step": 28
+    },
+    {
+      "epoch": 0.002474693063392676,
+      "grad_norm": 0.765625,
+      "learning_rate": 2.9e-05,
+      "loss": 1.8873,
+      "step": 29
+    },
+    {
+      "epoch": 0.0025600273069579407,
+      "grad_norm": 0.82421875,
+      "learning_rate": 3e-05,
+      "loss": 1.7073,
+      "step": 30
+    },
+    {
+      "epoch": 0.0026453615505232057,
+      "grad_norm": 0.796875,
+      "learning_rate": 3.1e-05,
+      "loss": 1.9116,
+      "step": 31
+    },
+    {
+      "epoch": 0.0027306957940884703,
+      "grad_norm": 0.73828125,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 1.6336,
+      "step": 32
+    },
+    {
+      "epoch": 0.002816030037653735,
+      "grad_norm": 0.80078125,
+      "learning_rate": 3.3e-05,
+      "loss": 1.8687,
+      "step": 33
+    },
+    {
+      "epoch": 0.0029013642812189995,
+      "grad_norm": 0.7265625,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 1.7556,
+      "step": 34
+    },
+    {
+      "epoch": 0.0029866985247842645,
+      "grad_norm": 0.71875,
+      "learning_rate": 3.5e-05,
+      "loss": 1.646,
+      "step": 35
+    },
+    {
+      "epoch": 0.003072032768349529,
+      "grad_norm": 0.89453125,
+      "learning_rate": 3.6e-05,
+      "loss": 1.8949,
+      "step": 36
+    },
+    {
+      "epoch": 0.0031573670119147937,
+      "grad_norm": 0.86328125,
+      "learning_rate": 3.7e-05,
+      "loss": 1.8113,
+      "step": 37
+    },
+    {
+      "epoch": 0.0032427012554800583,
+      "grad_norm": 0.84765625,
+      "learning_rate": 3.8e-05,
+      "loss": 1.8451,
+      "step": 38
+    },
+    {
+      "epoch": 0.0033280354990453233,
+      "grad_norm": 0.73828125,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 1.7651,
+      "step": 39
+    },
+    {
+      "epoch": 0.003413369742610588,
+      "grad_norm": 0.82421875,
+      "learning_rate": 4e-05,
+      "loss": 1.8078,
+      "step": 40
+    },
+    {
+      "epoch": 0.0034987039861758525,
+      "grad_norm": 0.74609375,
+      "learning_rate": 4.1e-05,
+      "loss": 1.6197,
+      "step": 41
+    },
+    {
+      "epoch": 0.003584038229741117,
+      "grad_norm": 0.90234375,
+      "learning_rate": 4.2e-05,
+      "loss": 1.7857,
+      "step": 42
+    },
+    {
+      "epoch": 0.003669372473306382,
+      "grad_norm": 0.828125,
+      "learning_rate": 4.3e-05,
+      "loss": 1.8162,
+      "step": 43
+    },
+    {
+      "epoch": 0.0037547067168716467,
+      "grad_norm": 0.7265625,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 1.8103,
+      "step": 44
+    },
+    {
+      "epoch": 0.0038400409604369113,
+      "grad_norm": 0.69921875,
+      "learning_rate": 4.5e-05,
+      "loss": 1.6104,
+      "step": 45
+    },
+    {
+      "epoch": 0.003925375204002176,
+      "grad_norm": 0.6875,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 1.6466,
+      "step": 46
+    },
+    {
+      "epoch": 0.004010709447567441,
+      "grad_norm": 0.6640625,
+      "learning_rate": 4.7e-05,
+      "loss": 1.5459,
+      "step": 47
+    },
+    {
+      "epoch": 0.0040960436911327055,
+      "grad_norm": 0.703125,
+      "learning_rate": 4.8e-05,
+      "loss": 1.7963,
+      "step": 48
+    },
+    {
+      "epoch": 0.00418137793469797,
+      "grad_norm": 0.796875,
+      "learning_rate": 4.9e-05,
+      "loss": 1.9184,
+      "step": 49
+    },
+    {
+      "epoch": 0.004266712178263235,
+      "grad_norm": 0.7109375,
+      "learning_rate": 5e-05,
+      "loss": 1.6674,
+      "step": 50
+    },
+    {
+      "epoch": 0.004352046421828499,
+      "grad_norm": 0.765625,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 1.7562,
+      "step": 51
+    },
+    {
+      "epoch": 0.004437380665393764,
+      "grad_norm": 0.76171875,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 1.6926,
+      "step": 52
+    },
+    {
+      "epoch": 0.004522714908959029,
+      "grad_norm": 0.8515625,
+      "learning_rate": 5.300000000000001e-05,
+      "loss": 1.833,
+      "step": 53
+    },
+    {
+      "epoch": 0.004608049152524294,
+      "grad_norm": 0.73046875,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 1.9062,
+      "step": 54
+    },
+    {
+      "epoch": 0.0046933833960895585,
+      "grad_norm": 0.68359375,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 1.6742,
+      "step": 55
+    },
+    {
+      "epoch": 0.004778717639654823,
+      "grad_norm": 0.67578125,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 1.8362,
+      "step": 56
+    },
+    {
+      "epoch": 0.004864051883220088,
+      "grad_norm": 0.69921875,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 1.7329,
+      "step": 57
+    },
+    {
+      "epoch": 0.004949386126785352,
+      "grad_norm": 0.6171875,
+      "learning_rate": 5.8e-05,
+      "loss": 1.8006,
+      "step": 58
+    },
+    {
+      "epoch": 0.005034720370350617,
+      "grad_norm": 0.734375,
+      "learning_rate": 5.9e-05,
+      "loss": 1.7329,
+      "step": 59
+    },
+    {
+      "epoch": 0.005120054613915881,
+      "grad_norm": 0.69140625,
+      "learning_rate": 6e-05,
+      "loss": 1.777,
+      "step": 60
+    },
+    {
+      "epoch": 0.005205388857481147,
+      "grad_norm": 0.671875,
+      "learning_rate": 6.1e-05,
+      "loss": 1.6312,
+      "step": 61
+    },
+    {
+      "epoch": 0.0052907231010464115,
+      "grad_norm": 0.59765625,
+      "learning_rate": 6.2e-05,
+      "loss": 1.506,
+      "step": 62
+    },
+    {
+      "epoch": 0.005376057344611676,
+      "grad_norm": 0.6796875,
+      "learning_rate": 6.3e-05,
+      "loss": 1.6359,
+      "step": 63
+    },
+    {
+      "epoch": 0.005461391588176941,
+      "grad_norm": 0.65234375,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 1.6188,
+      "step": 64
+    },
+    {
+      "epoch": 0.005546725831742205,
+      "grad_norm": 0.6328125,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 1.7215,
+      "step": 65
+    },
+    {
+      "epoch": 0.00563206007530747,
+      "grad_norm": 0.671875,
+      "learning_rate": 6.6e-05,
+      "loss": 1.868,
+      "step": 66
+    },
+    {
+      "epoch": 0.005717394318872734,
+      "grad_norm": 0.5859375,
+      "learning_rate": 6.7e-05,
+      "loss": 1.7063,
+      "step": 67
+    },
+    {
+      "epoch": 0.005802728562437999,
+      "grad_norm": 0.59375,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 1.6246,
+      "step": 68
+    },
+    {
+      "epoch": 0.0058880628060032645,
+      "grad_norm": 0.578125,
+      "learning_rate": 6.9e-05,
+      "loss": 1.8109,
+      "step": 69
+    },
+    {
+      "epoch": 0.005973397049568529,
+      "grad_norm": 0.6796875,
+      "learning_rate": 7e-05,
+      "loss": 1.9047,
+      "step": 70
+    },
+    {
+      "epoch": 0.006058731293133794,
+      "grad_norm": 0.70703125,
+      "learning_rate": 7.1e-05,
+      "loss": 1.6801,
+      "step": 71
+    },
+    {
+      "epoch": 0.006144065536699058,
+      "grad_norm": 0.66015625,
+      "learning_rate": 7.2e-05,
+      "loss": 1.7572,
+      "step": 72
+    },
+    {
+      "epoch": 0.006229399780264323,
+      "grad_norm": 0.61328125,
+      "learning_rate": 7.3e-05,
+      "loss": 1.7509,
+      "step": 73
+    },
+    {
+      "epoch": 0.006314734023829587,
+      "grad_norm": 0.60546875,
+      "learning_rate": 7.4e-05,
+      "loss": 1.8562,
+      "step": 74
+    },
+    {
+      "epoch": 0.006400068267394852,
+      "grad_norm": 0.5625,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.6753,
+      "step": 75
+    },
+    {
+      "epoch": 0.006485402510960117,
+      "grad_norm": 0.6015625,
+      "learning_rate": 7.6e-05,
+      "loss": 1.6632,
+      "step": 76
+    },
+    {
+      "epoch": 0.006570736754525381,
+      "grad_norm": 0.78515625,
+      "learning_rate": 7.7e-05,
+      "loss": 1.8732,
+      "step": 77
+    },
+    {
+      "epoch": 0.006656070998090647,
+      "grad_norm": 0.59765625,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 1.6412,
+      "step": 78
+    },
+    {
+      "epoch": 0.006741405241655911,
+      "grad_norm": 0.63671875,
+      "learning_rate": 7.900000000000001e-05,
+      "loss": 1.695,
+      "step": 79
+    },
+    {
+      "epoch": 0.006826739485221176,
+      "grad_norm": 0.58203125,
+      "learning_rate": 8e-05,
+      "loss": 1.6164,
+      "step": 80
+    },
+    {
+      "epoch": 0.00691207372878644,
+      "grad_norm": 0.60546875,
+      "learning_rate": 8.1e-05,
+      "loss": 1.8737,
+      "step": 81
+    },
+    {
+      "epoch": 0.006997407972351705,
+      "grad_norm": 0.5625,
+      "learning_rate": 8.2e-05,
+      "loss": 1.8113,
+      "step": 82
+    },
+    {
+      "epoch": 0.00708274221591697,
+      "grad_norm": 0.6328125,
+      "learning_rate": 8.3e-05,
+      "loss": 1.8269,
+      "step": 83
+    },
+    {
+      "epoch": 0.007168076459482234,
+      "grad_norm": 0.5390625,
+      "learning_rate": 8.4e-05,
+      "loss": 1.6431,
+      "step": 84
+    },
+    {
+      "epoch": 0.007253410703047499,
+      "grad_norm": 0.5390625,
+      "learning_rate": 8.5e-05,
+      "loss": 1.6118,
+      "step": 85
+    },
+    {
+      "epoch": 0.007338744946612764,
+      "grad_norm": 0.5,
+      "learning_rate": 8.6e-05,
+      "loss": 1.5627,
+      "step": 86
+    },
+    {
+      "epoch": 0.007424079190178029,
+      "grad_norm": 0.51953125,
+      "learning_rate": 8.7e-05,
+      "loss": 1.686,
+      "step": 87
+    },
+    {
+      "epoch": 0.007509413433743293,
+      "grad_norm": 0.50390625,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 1.6378,
+      "step": 88
+    },
+    {
+      "epoch": 0.007594747677308558,
+      "grad_norm": 0.48046875,
+      "learning_rate": 8.900000000000001e-05,
+      "loss": 1.529,
+      "step": 89
+    },
+    {
+      "epoch": 0.007680081920873823,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9e-05,
+      "loss": 1.6327,
+      "step": 90
+    },
+    {
+      "epoch": 0.007765416164439087,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.1e-05,
+      "loss": 1.7247,
+      "step": 91
+    },
+    {
+      "epoch": 0.007850750408004353,
+      "grad_norm": 0.5,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 1.6186,
+      "step": 92
+    },
+    {
+      "epoch": 0.007936084651569617,
+      "grad_norm": 0.546875,
+      "learning_rate": 9.300000000000001e-05,
+      "loss": 1.5325,
+      "step": 93
+    },
+    {
+      "epoch": 0.008021418895134882,
+      "grad_norm": 0.48828125,
+      "learning_rate": 9.4e-05,
+      "loss": 1.4677,
+      "step": 94
+    },
+    {
+      "epoch": 0.008106753138700146,
+      "grad_norm": 0.55078125,
+      "learning_rate": 9.5e-05,
+      "loss": 1.8127,
+      "step": 95
+    },
+    {
+      "epoch": 0.008192087382265411,
+      "grad_norm": 0.54296875,
+      "learning_rate": 9.6e-05,
+      "loss": 1.5892,
+      "step": 96
+    },
+    {
+      "epoch": 0.008277421625830676,
+      "grad_norm": 0.466796875,
+      "learning_rate": 9.7e-05,
+      "loss": 1.666,
+      "step": 97
+    },
+    {
+      "epoch": 0.00836275586939594,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.8e-05,
+      "loss": 1.5446,
+      "step": 98
+    },
+    {
+      "epoch": 0.008448090112961205,
+      "grad_norm": 0.57421875,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 1.8449,
+      "step": 99
+    },
+    {
+      "epoch": 0.00853342435652647,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0001,
+      "loss": 1.5976,
+      "step": 100
+    },
+    {
+      "epoch": 0.008618758600091734,
+      "grad_norm": 0.6640625,
+      "learning_rate": 9.999497487437187e-05,
+      "loss": 1.4564,
+      "step": 101
+    },
+    {
+      "epoch": 0.008704092843656999,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.998994974874373e-05,
+      "loss": 1.9022,
+      "step": 102
+    },
+    {
+      "epoch": 0.008789427087222263,
+      "grad_norm": 0.435546875,
+      "learning_rate": 9.998492462311558e-05,
+      "loss": 1.4713,
+      "step": 103
+    },
+    {
+      "epoch": 0.008874761330787528,
+      "grad_norm": 0.484375,
+      "learning_rate": 9.997989949748744e-05,
+      "loss": 1.507,
+      "step": 104
+    },
+    {
+      "epoch": 0.008960095574352792,
+      "grad_norm": 0.578125,
+      "learning_rate": 9.99748743718593e-05,
+      "loss": 1.5528,
+      "step": 105
+    },
+    {
+      "epoch": 0.009045429817918059,
+      "grad_norm": 0.4609375,
+      "learning_rate": 9.996984924623116e-05,
+      "loss": 1.4151,
+      "step": 106
+    },
+    {
+      "epoch": 0.009130764061483323,
+      "grad_norm": 0.4921875,
+      "learning_rate": 9.996482412060301e-05,
+      "loss": 1.7527,
+      "step": 107
+    },
+    {
+      "epoch": 0.009216098305048588,
+      "grad_norm": 0.486328125,
+      "learning_rate": 9.995979899497487e-05,
+      "loss": 1.622,
+      "step": 108
+    },
+    {
+      "epoch": 0.009301432548613852,
+      "grad_norm": 0.4765625,
+      "learning_rate": 9.995477386934674e-05,
+      "loss": 1.613,
+      "step": 109
+    },
+    {
+      "epoch": 0.009386766792179117,
+      "grad_norm": 0.54296875,
+      "learning_rate": 9.99497487437186e-05,
+      "loss": 1.7025,
+      "step": 110
+    },
+    {
+      "epoch": 0.009472101035744382,
+      "grad_norm": 0.4453125,
+      "learning_rate": 9.994472361809045e-05,
+      "loss": 1.5231,
+      "step": 111
+    },
+    {
+      "epoch": 0.009557435279309646,
+      "grad_norm": 0.482421875,
+      "learning_rate": 9.993969849246232e-05,
+      "loss": 1.4343,
+      "step": 112
+    },
+    {
+      "epoch": 0.00964276952287491,
+      "grad_norm": 0.427734375,
+      "learning_rate": 9.993467336683417e-05,
+      "loss": 1.5259,
+      "step": 113
+    },
+    {
+      "epoch": 0.009728103766440175,
+      "grad_norm": 0.486328125,
+      "learning_rate": 9.992964824120603e-05,
+      "loss": 1.7502,
+      "step": 114
+    },
+    {
+      "epoch": 0.00981343801000544,
+      "grad_norm": 0.462890625,
+      "learning_rate": 9.99246231155779e-05,
+      "loss": 1.5343,
+      "step": 115
+    },
+    {
+      "epoch": 0.009898772253570704,
+      "grad_norm": 0.55078125,
+      "learning_rate": 9.991959798994976e-05,
+      "loss": 1.7215,
+      "step": 116
+    },
+    {
+      "epoch": 0.009984106497135969,
+      "grad_norm": 0.4609375,
+      "learning_rate": 9.99145728643216e-05,
+      "loss": 1.4887,
+      "step": 117
+    },
+    {
+      "epoch": 0.010069440740701234,
+      "grad_norm": 0.443359375,
+      "learning_rate": 9.990954773869348e-05,
+      "loss": 1.6494,
+      "step": 118
+    },
+    {
+      "epoch": 0.010154774984266498,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.990452261306533e-05,
+      "loss": 1.5469,
+      "step": 119
+    },
+    {
+      "epoch": 0.010240109227831763,
+      "grad_norm": 0.47265625,
+      "learning_rate": 9.989949748743719e-05,
+      "loss": 1.5484,
+      "step": 120
+    },
+    {
+      "epoch": 0.010325443471397027,
+      "grad_norm": 0.41015625,
+      "learning_rate": 9.989447236180905e-05,
+      "loss": 1.5742,
+      "step": 121
+    },
+    {
+      "epoch": 0.010410777714962294,
+      "grad_norm": 0.4296875,
+      "learning_rate": 9.988944723618092e-05,
+      "loss": 1.4625,
+      "step": 122
+    },
+    {
+      "epoch": 0.010496111958527558,
+      "grad_norm": 0.482421875,
+      "learning_rate": 9.988442211055276e-05,
+      "loss": 1.5756,
+      "step": 123
+    },
+    {
+      "epoch": 0.010581446202092823,
+      "grad_norm": 0.39453125,
+      "learning_rate": 9.987939698492463e-05,
+      "loss": 1.4773,
+      "step": 124
+    },
+    {
+      "epoch": 0.010666780445658088,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.987437185929649e-05,
+      "loss": 1.5322,
+      "step": 125
+    },
+    {
+      "epoch": 0.010752114689223352,
+      "grad_norm": 0.53125,
+      "learning_rate": 9.986934673366835e-05,
+      "loss": 1.5734,
+      "step": 126
+    },
+    {
+      "epoch": 0.010837448932788617,
+      "grad_norm": 0.43359375,
+      "learning_rate": 9.98643216080402e-05,
+      "loss": 1.5173,
+      "step": 127
+    },
+    {
+      "epoch": 0.010922783176353881,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.985929648241207e-05,
+      "loss": 1.4155,
+      "step": 128
+    },
+    {
+      "epoch": 0.011008117419919146,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.985427135678392e-05,
+      "loss": 1.5875,
+      "step": 129
+    },
+    {
+      "epoch": 0.01109345166348441,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.984924623115578e-05,
+      "loss": 1.4325,
+      "step": 130
+    },
+    {
+      "epoch": 0.011178785907049675,
+      "grad_norm": 0.4453125,
+      "learning_rate": 9.984422110552765e-05,
+      "loss": 1.6385,
+      "step": 131
+    },
+    {
+      "epoch": 0.01126412015061494,
+      "grad_norm": 0.43359375,
+      "learning_rate": 9.983919597989951e-05,
+      "loss": 1.5616,
+      "step": 132
+    },
+    {
+      "epoch": 0.011349454394180204,
+      "grad_norm": 0.41796875,
+      "learning_rate": 9.983417085427136e-05,
+      "loss": 1.4178,
+      "step": 133
+    },
+    {
+      "epoch": 0.011434788637745469,
+      "grad_norm": 0.44921875,
+      "learning_rate": 9.982914572864322e-05,
+      "loss": 1.548,
+      "step": 134
+    },
+    {
+      "epoch": 0.011520122881310733,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.982412060301508e-05,
+      "loss": 1.5455,
+      "step": 135
+    },
+    {
+      "epoch": 0.011605457124875998,
+      "grad_norm": 0.46875,
+      "learning_rate": 9.981909547738694e-05,
+      "loss": 1.5639,
+      "step": 136
+    },
+    {
+      "epoch": 0.011690791368441263,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.98140703517588e-05,
+      "loss": 1.7033,
+      "step": 137
+    },
+    {
+      "epoch": 0.011776125612006529,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.980904522613065e-05,
+      "loss": 1.4245,
+      "step": 138
+    },
+    {
+      "epoch": 0.011861459855571794,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.980402010050252e-05,
+      "loss": 1.4474,
+      "step": 139
+    },
+    {
+      "epoch": 0.011946794099137058,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.979899497487438e-05,
+      "loss": 1.5824,
+      "step": 140
+    },
+    {
+      "epoch": 0.012032128342702323,
+      "grad_norm": 0.435546875,
+      "learning_rate": 9.979396984924624e-05,
+      "loss": 1.5278,
+      "step": 141
+    },
+    {
+      "epoch": 0.012117462586267587,
+      "grad_norm": 0.4296875,
+      "learning_rate": 9.978894472361809e-05,
+      "loss": 1.5589,
+      "step": 142
+    },
+    {
+      "epoch": 0.012202796829832852,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.978391959798995e-05,
+      "loss": 1.3584,
+      "step": 143
+    },
+    {
+      "epoch": 0.012288131073398116,
+      "grad_norm": 0.458984375,
+      "learning_rate": 9.977889447236181e-05,
+      "loss": 1.5057,
+      "step": 144
+    },
+    {
+      "epoch": 0.012373465316963381,
+      "grad_norm": 0.451171875,
+      "learning_rate": 9.977386934673367e-05,
+      "loss": 1.5435,
+      "step": 145
+    },
+    {
+      "epoch": 0.012458799560528646,
+      "grad_norm": 0.431640625,
+      "learning_rate": 9.976884422110552e-05,
+      "loss": 1.5242,
+      "step": 146
+    },
+    {
+      "epoch": 0.01254413380409391,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.97638190954774e-05,
+      "loss": 1.5619,
+      "step": 147
+    },
+    {
+      "epoch": 0.012629468047659175,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.975879396984925e-05,
+      "loss": 1.6439,
+      "step": 148
+    },
+    {
+      "epoch": 0.01271480229122444,
+      "grad_norm": 0.404296875,
+      "learning_rate": 9.975376884422111e-05,
+      "loss": 1.4136,
+      "step": 149
+    },
+    {
+      "epoch": 0.012800136534789704,
+      "grad_norm": 0.400390625,
+      "learning_rate": 9.974874371859297e-05,
+      "loss": 1.5351,
+      "step": 150
+    },
+    {
+      "epoch": 0.012885470778354969,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.974371859296483e-05,
+      "loss": 1.5089,
+      "step": 151
+    },
+    {
+      "epoch": 0.012970805021920233,
+      "grad_norm": 0.41796875,
+      "learning_rate": 9.973869346733668e-05,
+      "loss": 1.3977,
+      "step": 152
+    },
+    {
+      "epoch": 0.013056139265485498,
+      "grad_norm": 0.43359375,
+      "learning_rate": 9.973366834170856e-05,
+      "loss": 1.4765,
+      "step": 153
+    },
+    {
+      "epoch": 0.013141473509050762,
+      "grad_norm": 0.427734375,
+      "learning_rate": 9.97286432160804e-05,
+      "loss": 1.3851,
+      "step": 154
+    },
+    {
+      "epoch": 0.013226807752616029,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.972361809045227e-05,
+      "loss": 1.4794,
+      "step": 155
+    },
+    {
+      "epoch": 0.013312141996181293,
+      "grad_norm": 0.390625,
+      "learning_rate": 9.971859296482412e-05,
+      "loss": 1.661,
+      "step": 156
+    },
+    {
+      "epoch": 0.013397476239746558,
+      "grad_norm": 0.404296875,
+      "learning_rate": 9.971356783919599e-05,
+      "loss": 1.4593,
+      "step": 157
+    },
+    {
+      "epoch": 0.013482810483311822,
+      "grad_norm": 0.458984375,
+      "learning_rate": 9.970854271356784e-05,
+      "loss": 1.4863,
+      "step": 158
+    },
+    {
+      "epoch": 0.013568144726877087,
+      "grad_norm": 0.390625,
+      "learning_rate": 9.97035175879397e-05,
+      "loss": 1.5137,
+      "step": 159
+    },
+    {
+      "epoch": 0.013653478970442352,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.969849246231156e-05,
+      "loss": 1.4967,
+      "step": 160
+    },
+    {
+      "epoch": 0.013738813214007616,
+      "grad_norm": 0.400390625,
+      "learning_rate": 9.969346733668343e-05,
+      "loss": 1.5225,
+      "step": 161
+    },
+    {
+      "epoch": 0.01382414745757288,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.968844221105527e-05,
+      "loss": 1.4847,
+      "step": 162
+    },
+    {
+      "epoch": 0.013909481701138145,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.968341708542715e-05,
+      "loss": 1.4611,
+      "step": 163
+    },
+    {
+      "epoch": 0.01399481594470341,
+      "grad_norm": 0.388671875,
+      "learning_rate": 9.9678391959799e-05,
+      "loss": 1.6441,
+      "step": 164
+    },
+    {
+      "epoch": 0.014080150188268675,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.967336683417086e-05,
+      "loss": 1.5616,
+      "step": 165
+    },
+    {
+      "epoch": 0.01416548443183394,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.966834170854272e-05,
+      "loss": 1.4336,
+      "step": 166
+    },
+    {
+      "epoch": 0.014250818675399204,
+      "grad_norm": 0.3828125,
+      "learning_rate": 9.966331658291458e-05,
+      "loss": 1.3676,
+      "step": 167
+    },
+    {
+      "epoch": 0.014336152918964468,
+      "grad_norm": 0.416015625,
+      "learning_rate": 9.965829145728643e-05,
+      "loss": 1.6301,
+      "step": 168
+    },
+    {
+      "epoch": 0.014421487162529733,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.96532663316583e-05,
+      "loss": 1.5302,
+      "step": 169
+    },
+    {
+      "epoch": 0.014506821406094998,
+      "grad_norm": 0.400390625,
+      "learning_rate": 9.964824120603016e-05,
+      "loss": 1.516,
+      "step": 170
+    },
+    {
+      "epoch": 0.014592155649660264,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.964321608040202e-05,
+      "loss": 1.5338,
+      "step": 171
+    },
+    {
+      "epoch": 0.014677489893225528,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.963819095477387e-05,
+      "loss": 1.4734,
+      "step": 172
+    },
+    {
+      "epoch": 0.014762824136790793,
+      "grad_norm": 0.435546875,
+      "learning_rate": 9.963316582914573e-05,
+      "loss": 1.3936,
+      "step": 173
+    },
+    {
+      "epoch": 0.014848158380356058,
+      "grad_norm": 0.37890625,
+      "learning_rate": 9.962814070351759e-05,
+      "loss": 1.4652,
+      "step": 174
+    },
+    {
+      "epoch": 0.014933492623921322,
+      "grad_norm": 0.41796875,
+      "learning_rate": 9.962311557788945e-05,
+      "loss": 1.4478,
+      "step": 175
+    },
+    {
+      "epoch": 0.015018826867486587,
+      "grad_norm": 0.435546875,
+      "learning_rate": 9.961809045226132e-05,
+      "loss": 1.4698,
+      "step": 176
+    },
+    {
+      "epoch": 0.015104161111051851,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.961306532663316e-05,
+      "loss": 1.5142,
+      "step": 177
+    },
+    {
+      "epoch": 0.015189495354617116,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.960804020100503e-05,
+      "loss": 1.5024,
+      "step": 178
+    },
+    {
+      "epoch": 0.01527482959818238,
+      "grad_norm": 0.416015625,
+      "learning_rate": 9.960301507537689e-05,
+      "loss": 1.4897,
+      "step": 179
+    },
+    {
+      "epoch": 0.015360163841747645,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.959798994974875e-05,
+      "loss": 1.4291,
+      "step": 180
+    },
+    {
+      "epoch": 0.01544549808531291,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.95929648241206e-05,
+      "loss": 1.333,
+      "step": 181
+    },
+    {
+      "epoch": 0.015530832328878174,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.958793969849247e-05,
+      "loss": 1.5698,
+      "step": 182
+    },
+    {
+      "epoch": 0.015616166572443439,
+      "grad_norm": 0.4453125,
+      "learning_rate": 9.958291457286432e-05,
+      "loss": 1.7028,
+      "step": 183
+    },
+    {
+      "epoch": 0.015701500816008705,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.957788944723619e-05,
+      "loss": 1.4966,
+      "step": 184
+    },
+    {
+      "epoch": 0.01578683505957397,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.957286432160805e-05,
+      "loss": 1.4275,
+      "step": 185
+    },
+    {
+      "epoch": 0.015872169303139234,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.956783919597991e-05,
+      "loss": 1.3239,
+      "step": 186
+    },
+    {
+      "epoch": 0.0159575035467045,
+      "grad_norm": 0.4296875,
+      "learning_rate": 9.956281407035176e-05,
+      "loss": 1.5804,
+      "step": 187
+    },
+    {
+      "epoch": 0.016042837790269764,
+      "grad_norm": 0.39453125,
+      "learning_rate": 9.955778894472362e-05,
+      "loss": 1.5098,
+      "step": 188
+    },
+    {
+      "epoch": 0.016128172033835028,
+      "grad_norm": 0.37890625,
+      "learning_rate": 9.955276381909548e-05,
+      "loss": 1.5299,
+      "step": 189
+    },
+    {
+      "epoch": 0.016213506277400293,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.954773869346734e-05,
+      "loss": 1.4604,
+      "step": 190
+    },
+    {
+      "epoch": 0.016298840520965557,
+      "grad_norm": 0.431640625,
+      "learning_rate": 9.954271356783919e-05,
+      "loss": 1.4824,
+      "step": 191
+    },
+    {
+      "epoch": 0.016384174764530822,
+      "grad_norm": 0.44921875,
+      "learning_rate": 9.953768844221107e-05,
+      "loss": 1.5933,
+      "step": 192
+    },
+    {
+      "epoch": 0.016469509008096087,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.953266331658292e-05,
+      "loss": 1.6971,
+      "step": 193
+    },
+    {
+      "epoch": 0.01655484325166135,
+      "grad_norm": 0.423828125,
+      "learning_rate": 9.952763819095478e-05,
+      "loss": 1.7039,
+      "step": 194
+    },
+    {
+      "epoch": 0.016640177495226616,
+      "grad_norm": 0.4140625,
+      "learning_rate": 9.952261306532664e-05,
+      "loss": 1.5222,
+      "step": 195
+    },
+    {
+      "epoch": 0.01672551173879188,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.95175879396985e-05,
+      "loss": 1.5505,
+      "step": 196
+    },
+    {
+      "epoch": 0.016810845982357145,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.951256281407035e-05,
+      "loss": 1.35,
+      "step": 197
+    },
+    {
+      "epoch": 0.01689618022592241,
+      "grad_norm": 0.419921875,
+      "learning_rate": 9.950753768844223e-05,
+      "loss": 1.4943,
+      "step": 198
+    },
+    {
+      "epoch": 0.016981514469487674,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.950251256281408e-05,
+      "loss": 1.4103,
+      "step": 199
+    },
+    {
+      "epoch": 0.01706684871305294,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.949748743718594e-05,
+      "loss": 1.3057,
+      "step": 200
+    },
+    {
+      "epoch": 0.017152182956618203,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.94924623115578e-05,
+      "loss": 1.6187,
+      "step": 201
+    },
+    {
+      "epoch": 0.017237517200183468,
+      "grad_norm": 0.37890625,
+      "learning_rate": 9.948743718592966e-05,
+      "loss": 1.5999,
+      "step": 202
+    },
+    {
+      "epoch": 0.017322851443748732,
+      "grad_norm": 0.447265625,
+      "learning_rate": 9.948241206030151e-05,
+      "loss": 1.4204,
+      "step": 203
+    },
+    {
+      "epoch": 0.017408185687313997,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.947738693467337e-05,
+      "loss": 1.4472,
+      "step": 204
+    },
+    {
+      "epoch": 0.01749351993087926,
+      "grad_norm": 0.3828125,
+      "learning_rate": 9.947236180904523e-05,
+      "loss": 1.5425,
+      "step": 205
+    },
+    {
+      "epoch": 0.017578854174444526,
+      "grad_norm": 0.390625,
+      "learning_rate": 9.94673366834171e-05,
+      "loss": 1.4854,
+      "step": 206
+    },
+    {
+      "epoch": 0.01766418841800979,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.946231155778894e-05,
+      "loss": 1.4918,
+      "step": 207
+    },
+    {
+      "epoch": 0.017749522661575055,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.94572864321608e-05,
+      "loss": 1.4233,
+      "step": 208
+    },
+    {
+      "epoch": 0.01783485690514032,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.945226130653267e-05,
+      "loss": 1.3717,
+      "step": 209
+    },
+    {
+      "epoch": 0.017920191148705585,
+      "grad_norm": 0.36328125,
+      "learning_rate": 9.944723618090453e-05,
+      "loss": 1.4905,
+      "step": 210
+    },
+    {
+      "epoch": 0.018005525392270853,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.944221105527639e-05,
+      "loss": 1.3972,
+      "step": 211
+    },
+    {
+      "epoch": 0.018090859635836117,
+      "grad_norm": 0.408203125,
+      "learning_rate": 9.943718592964824e-05,
+      "loss": 1.4586,
+      "step": 212
+    },
+    {
+      "epoch": 0.018176193879401382,
+      "grad_norm": 0.36328125,
+      "learning_rate": 9.94321608040201e-05,
+      "loss": 1.5897,
+      "step": 213
+    },
+    {
+      "epoch": 0.018261528122966646,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.942713567839197e-05,
+      "loss": 1.5939,
+      "step": 214
+    },
+    {
+      "epoch": 0.01834686236653191,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.942211055276383e-05,
+      "loss": 1.5312,
+      "step": 215
+    },
+    {
+      "epoch": 0.018432196610097176,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.941708542713568e-05,
+      "loss": 1.4241,
+      "step": 216
+    },
+    {
+      "epoch": 0.01851753085366244,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.941206030150754e-05,
+      "loss": 1.3628,
+      "step": 217
+    },
+    {
+      "epoch": 0.018602865097227705,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.94070351758794e-05,
+      "loss": 1.4801,
+      "step": 218
+    },
+    {
+      "epoch": 0.01868819934079297,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.940201005025126e-05,
+      "loss": 1.5228,
+      "step": 219
+    },
+    {
+      "epoch": 0.018773533584358234,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.939698492462311e-05,
+      "loss": 1.4359,
+      "step": 220
+    },
+    {
+      "epoch": 0.0188588678279235,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.939195979899499e-05,
+      "loss": 1.4317,
+      "step": 221
+    },
+    {
+      "epoch": 0.018944202071488763,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.938693467336683e-05,
+      "loss": 1.5913,
+      "step": 222
+    },
+    {
+      "epoch": 0.019029536315054028,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.93819095477387e-05,
+      "loss": 1.505,
+      "step": 223
+    },
+    {
+      "epoch": 0.019114870558619292,
+      "grad_norm": 0.390625,
+      "learning_rate": 9.937688442211056e-05,
+      "loss": 1.6387,
+      "step": 224
+    },
+    {
+      "epoch": 0.019200204802184557,
+      "grad_norm": 0.416015625,
+      "learning_rate": 9.937185929648242e-05,
+      "loss": 1.5977,
+      "step": 225
+    },
+    {
+      "epoch": 0.01928553904574982,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.936683417085427e-05,
+      "loss": 1.4547,
+      "step": 226
+    },
+    {
+      "epoch": 0.019370873289315086,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.936180904522614e-05,
+      "loss": 1.4913,
+      "step": 227
+    },
+    {
+      "epoch": 0.01945620753288035,
+      "grad_norm": 0.4296875,
+      "learning_rate": 9.935678391959799e-05,
+      "loss": 1.5211,
+      "step": 228
+    },
+    {
+      "epoch": 0.019541541776445615,
+      "grad_norm": 0.515625,
+      "learning_rate": 9.935175879396985e-05,
+      "loss": 1.5645,
+      "step": 229
+    },
+    {
+      "epoch": 0.01962687602001088,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.934673366834172e-05,
+      "loss": 1.4796,
+      "step": 230
+    },
+    {
+      "epoch": 0.019712210263576144,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.934170854271358e-05,
+      "loss": 1.4586,
+      "step": 231
+    },
+    {
+      "epoch": 0.01979754450714141,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.933668341708543e-05,
+      "loss": 1.5003,
+      "step": 232
+    },
+    {
+      "epoch": 0.019882878750706674,
+      "grad_norm": 0.390625,
+      "learning_rate": 9.933165829145729e-05,
+      "loss": 1.5719,
+      "step": 233
+    },
+    {
+      "epoch": 0.019968212994271938,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.932663316582915e-05,
+      "loss": 1.3316,
+      "step": 234
+    },
+    {
+      "epoch": 0.020053547237837203,
+      "grad_norm": 0.388671875,
+      "learning_rate": 9.932160804020101e-05,
+      "loss": 1.5506,
+      "step": 235
+    },
+    {
+      "epoch": 0.020138881481402467,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.931658291457286e-05,
+      "loss": 1.4023,
+      "step": 236
+    },
+    {
+      "epoch": 0.020224215724967732,
+      "grad_norm": 0.3828125,
+      "learning_rate": 9.931155778894474e-05,
+      "loss": 1.5384,
+      "step": 237
+    },
+    {
+      "epoch": 0.020309549968532997,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.930653266331659e-05,
+      "loss": 1.332,
+      "step": 238
+    },
+    {
+      "epoch": 0.02039488421209826,
+      "grad_norm": 0.39453125,
+      "learning_rate": 9.930150753768845e-05,
+      "loss": 1.4412,
+      "step": 239
+    },
+    {
+      "epoch": 0.020480218455663526,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.929648241206031e-05,
+      "loss": 1.4309,
+      "step": 240
+    },
+    {
+      "epoch": 0.02056555269922879,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.929145728643217e-05,
+      "loss": 1.4405,
+      "step": 241
+    },
+    {
+      "epoch": 0.020650886942794055,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.928643216080402e-05,
+      "loss": 1.3873,
+      "step": 242
+    },
+    {
+      "epoch": 0.02073622118635932,
+      "grad_norm": 0.37890625,
+      "learning_rate": 9.928140703517588e-05,
+      "loss": 1.6588,
+      "step": 243
+    },
+    {
+      "epoch": 0.020821555429924588,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.927638190954774e-05,
+      "loss": 1.4384,
+      "step": 244
+    },
+    {
+      "epoch": 0.020906889673489852,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.927135678391961e-05,
+      "loss": 1.4542,
+      "step": 245
+    },
+    {
+      "epoch": 0.020992223917055117,
+      "grad_norm": 0.36328125,
+      "learning_rate": 9.926633165829147e-05,
+      "loss": 1.4885,
+      "step": 246
+    },
+    {
+      "epoch": 0.02107755816062038,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.926130653266332e-05,
+      "loss": 1.5343,
+      "step": 247
+    },
+    {
+      "epoch": 0.021162892404185646,
+      "grad_norm": 0.38671875,
+      "learning_rate": 9.925628140703518e-05,
+      "loss": 1.4286,
+      "step": 248
+    },
+    {
+      "epoch": 0.02124822664775091,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.925125628140703e-05,
+      "loss": 1.5849,
+      "step": 249
+    },
+    {
+      "epoch": 0.021333560891316175,
+      "grad_norm": 0.408203125,
+      "learning_rate": 9.92462311557789e-05,
+      "loss": 1.3079,
+      "step": 250
+    },
+    {
+      "epoch": 0.02141889513488144,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.924120603015075e-05,
+      "loss": 1.6098,
+      "step": 251
+    },
+    {
+      "epoch": 0.021504229378446704,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.923618090452261e-05,
+      "loss": 1.5455,
+      "step": 252
+    },
+    {
+      "epoch": 0.02158956362201197,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.923115577889448e-05,
+      "loss": 1.5326,
+      "step": 253
+    },
+    {
+      "epoch": 0.021674897865577233,
+      "grad_norm": 0.37890625,
+      "learning_rate": 9.922613065326634e-05,
+      "loss": 1.4347,
+      "step": 254
+    },
+    {
+      "epoch": 0.021760232109142498,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.922110552763819e-05,
+      "loss": 1.554,
+      "step": 255
+    },
+    {
+      "epoch": 0.021845566352707763,
+      "grad_norm": 0.39453125,
+      "learning_rate": 9.921608040201006e-05,
+      "loss": 1.4348,
+      "step": 256
+    },
+    {
+      "epoch": 0.021930900596273027,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.921105527638191e-05,
+      "loss": 1.3949,
+      "step": 257
+    },
+    {
+      "epoch": 0.022016234839838292,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.920603015075377e-05,
+      "loss": 1.3103,
+      "step": 258
+    },
+    {
+      "epoch": 0.022101569083403556,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.920100502512563e-05,
+      "loss": 1.3763,
+      "step": 259
+    },
+    {
+      "epoch": 0.02218690332696882,
+      "grad_norm": 0.33984375,
+      "learning_rate": 9.91959798994975e-05,
+      "loss": 1.3494,
+      "step": 260
+    },
+    {
+      "epoch": 0.022272237570534086,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.919095477386935e-05,
+      "loss": 1.3265,
+      "step": 261
+    },
+    {
+      "epoch": 0.02235757181409935,
+      "grad_norm": 0.30078125,
+      "learning_rate": 9.918592964824122e-05,
+      "loss": 1.2197,
+      "step": 262
+    },
+    {
+      "epoch": 0.022442906057664615,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.918090452261307e-05,
+      "loss": 1.4191,
+      "step": 263
+    },
+    {
+      "epoch": 0.02252824030122988,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.917587939698493e-05,
+      "loss": 1.3404,
+      "step": 264
+    },
+    {
+      "epoch": 0.022613574544795144,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.917085427135678e-05,
+      "loss": 1.3232,
+      "step": 265
+    },
+    {
+      "epoch": 0.02269890878836041,
+      "grad_norm": 0.37890625,
+      "learning_rate": 9.916582914572866e-05,
+      "loss": 1.3534,
+      "step": 266
+    },
+    {
+      "epoch": 0.022784243031925673,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.91608040201005e-05,
+      "loss": 1.396,
+      "step": 267
+    },
+    {
+      "epoch": 0.022869577275490938,
+      "grad_norm": 0.41015625,
+      "learning_rate": 9.915577889447237e-05,
+      "loss": 1.5844,
+      "step": 268
+    },
+    {
+      "epoch": 0.022954911519056202,
+      "grad_norm": 0.40625,
+      "learning_rate": 9.915075376884423e-05,
+      "loss": 1.4995,
+      "step": 269
+    },
+    {
+      "epoch": 0.023040245762621467,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.914572864321609e-05,
+      "loss": 1.4104,
+      "step": 270
+    },
+    {
+      "epoch": 0.02312558000618673,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.914070351758794e-05,
+      "loss": 1.4622,
+      "step": 271
+    },
+    {
+      "epoch": 0.023210914249751996,
+      "grad_norm": 0.361328125,
+      "learning_rate": 9.913567839195981e-05,
+      "loss": 1.3194,
+      "step": 272
+    },
+    {
+      "epoch": 0.02329624849331726,
+      "grad_norm": 0.37890625,
+      "learning_rate": 9.913065326633166e-05,
+      "loss": 1.4319,
+      "step": 273
+    },
+    {
+      "epoch": 0.023381582736882525,
+      "grad_norm": 0.376953125,
+      "learning_rate": 9.912562814070352e-05,
+      "loss": 1.4195,
+      "step": 274
+    },
+    {
+      "epoch": 0.02346691698044779,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.912060301507539e-05,
+      "loss": 1.3806,
+      "step": 275
+    },
+    {
+      "epoch": 0.023552251224013058,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.911557788944725e-05,
+      "loss": 1.4023,
+      "step": 276
+    },
+    {
+      "epoch": 0.023637585467578322,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.91105527638191e-05,
+      "loss": 1.4861,
+      "step": 277
+    },
+    {
+      "epoch": 0.023722919711143587,
+      "grad_norm": 0.375,
+      "learning_rate": 9.910552763819096e-05,
+      "loss": 1.3139,
+      "step": 278
+    },
+    {
+      "epoch": 0.02380825395470885,
+      "grad_norm": 0.375,
+      "learning_rate": 9.910050251256282e-05,
+      "loss": 1.4453,
+      "step": 279
+    },
+    {
+      "epoch": 0.023893588198274116,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.909547738693468e-05,
+      "loss": 1.5874,
+      "step": 280
+    },
+    {
+      "epoch": 0.02397892244183938,
+      "grad_norm": 0.400390625,
+      "learning_rate": 9.909045226130653e-05,
+      "loss": 1.587,
+      "step": 281
+    },
+    {
+      "epoch": 0.024064256685404645,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.90854271356784e-05,
+      "loss": 1.5391,
+      "step": 282
+    },
+    {
+      "epoch": 0.02414959092896991,
+      "grad_norm": 0.3203125,
+      "learning_rate": 9.908040201005026e-05,
+      "loss": 1.39,
+      "step": 283
+    },
+    {
+      "epoch": 0.024234925172535175,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.90753768844221e-05,
+      "loss": 1.489,
+      "step": 284
+    },
+    {
+      "epoch": 0.02432025941610044,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.907035175879398e-05,
+      "loss": 1.3527,
+      "step": 285
+    },
+    {
+      "epoch": 0.024405593659665704,
+      "grad_norm": 0.423828125,
+      "learning_rate": 9.906532663316583e-05,
+      "loss": 1.3988,
+      "step": 286
+    },
+    {
+      "epoch": 0.02449092790323097,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.906030150753769e-05,
+      "loss": 1.5052,
+      "step": 287
+    },
+    {
+      "epoch": 0.024576262146796233,
+      "grad_norm": 0.361328125,
+      "learning_rate": 9.905527638190955e-05,
+      "loss": 1.4864,
+      "step": 288
+    },
+    {
+      "epoch": 0.024661596390361498,
+      "grad_norm": 0.36328125,
+      "learning_rate": 9.905025125628141e-05,
+      "loss": 1.3851,
+      "step": 289
+    },
+    {
+      "epoch": 0.024746930633926762,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.904522613065326e-05,
+      "loss": 1.4283,
+      "step": 290
+    },
+    {
+      "epoch": 0.024832264877492027,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.904020100502514e-05,
+      "loss": 1.3651,
+      "step": 291
+    },
+    {
+      "epoch": 0.02491759912105729,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.903517587939699e-05,
+      "loss": 1.4531,
+      "step": 292
+    },
+    {
+      "epoch": 0.025002933364622556,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.903015075376885e-05,
+      "loss": 1.4701,
+      "step": 293
+    },
+    {
+      "epoch": 0.02508826760818782,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.902512562814071e-05,
+      "loss": 1.4249,
+      "step": 294
+    },
+    {
+      "epoch": 0.025173601851753085,
+      "grad_norm": 0.310546875,
+      "learning_rate": 9.902010050251257e-05,
+      "loss": 1.2651,
+      "step": 295
+    },
+    {
+      "epoch": 0.02525893609531835,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.901507537688442e-05,
+      "loss": 1.404,
+      "step": 296
+    },
+    {
+      "epoch": 0.025344270338883614,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.901005025125628e-05,
+      "loss": 1.5218,
+      "step": 297
+    },
+    {
+      "epoch": 0.02542960458244888,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.900502512562815e-05,
+      "loss": 1.2835,
+      "step": 298
+    },
+    {
+      "epoch": 0.025514938826014143,
+      "grad_norm": 0.33984375,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 1.3895,
+      "step": 299
+    },
+    {
+      "epoch": 0.025600273069579408,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.899497487437186e-05,
+      "loss": 1.5718,
+      "step": 300
+    },
+    {
+      "epoch": 0.025685607313144673,
+      "grad_norm": 0.427734375,
+      "learning_rate": 9.898994974874373e-05,
+      "loss": 1.6433,
+      "step": 301
+    },
+    {
+      "epoch": 0.025770941556709937,
+      "grad_norm": 0.400390625,
+      "learning_rate": 9.898492462311558e-05,
+      "loss": 1.465,
+      "step": 302
+    },
+    {
+      "epoch": 0.025856275800275202,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.897989949748744e-05,
+      "loss": 1.5619,
+      "step": 303
+    },
+    {
+      "epoch": 0.025941610043840466,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.89748743718593e-05,
+      "loss": 1.367,
+      "step": 304
+    },
+    {
+      "epoch": 0.02602694428740573,
+      "grad_norm": 0.423828125,
+      "learning_rate": 9.896984924623117e-05,
+      "loss": 1.3791,
+      "step": 305
+    },
+    {
+      "epoch": 0.026112278530970996,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.896482412060301e-05,
+      "loss": 1.9469,
+      "step": 306
+    },
+    {
+      "epoch": 0.02619761277453626,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.895979899497489e-05,
+      "loss": 1.3326,
+      "step": 307
+    },
+    {
+      "epoch": 0.026282947018101525,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.895477386934674e-05,
+      "loss": 1.455,
+      "step": 308
+    },
+    {
+      "epoch": 0.026368281261666793,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.89497487437186e-05,
+      "loss": 1.3405,
+      "step": 309
+    },
+    {
+      "epoch": 0.026453615505232057,
+      "grad_norm": 0.375,
+      "learning_rate": 9.894472361809046e-05,
+      "loss": 1.4276,
+      "step": 310
+    },
+    {
+      "epoch": 0.026538949748797322,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.893969849246232e-05,
+      "loss": 1.3773,
+      "step": 311
+    },
+    {
+      "epoch": 0.026624283992362587,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.893467336683417e-05,
+      "loss": 1.4056,
+      "step": 312
+    },
+    {
+      "epoch": 0.02670961823592785,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.892964824120604e-05,
+      "loss": 1.2113,
+      "step": 313
+    },
+    {
+      "epoch": 0.026794952479493116,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.89246231155779e-05,
+      "loss": 1.44,
+      "step": 314
+    },
+    {
+      "epoch": 0.02688028672305838,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.891959798994975e-05,
+      "loss": 1.3373,
+      "step": 315
+    },
+    {
+      "epoch": 0.026965620966623645,
+      "grad_norm": 0.4453125,
+      "learning_rate": 9.891457286432161e-05,
+      "loss": 1.3961,
+      "step": 316
+    },
+    {
+      "epoch": 0.02705095521018891,
+      "grad_norm": 0.388671875,
+      "learning_rate": 9.890954773869347e-05,
+      "loss": 1.2976,
+      "step": 317
+    },
+    {
+      "epoch": 0.027136289453754174,
+      "grad_norm": 0.431640625,
+      "learning_rate": 9.890452261306533e-05,
+      "loss": 1.5118,
+      "step": 318
+    },
+    {
+      "epoch": 0.02722162369731944,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.889949748743718e-05,
+      "loss": 1.4033,
+      "step": 319
+    },
+    {
+      "epoch": 0.027306957940884703,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.889447236180906e-05,
+      "loss": 1.3548,
+      "step": 320
+    },
+    {
+      "epoch": 0.027392292184449968,
+      "grad_norm": 0.328125,
+      "learning_rate": 9.88894472361809e-05,
+      "loss": 1.4382,
+      "step": 321
+    },
+    {
+      "epoch": 0.027477626428015232,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.888442211055277e-05,
+      "loss": 1.3789,
+      "step": 322
+    },
+    {
+      "epoch": 0.027562960671580497,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.887939698492463e-05,
+      "loss": 1.417,
+      "step": 323
+    },
+    {
+      "epoch": 0.02764829491514576,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.887437185929649e-05,
+      "loss": 1.67,
+      "step": 324
+    },
+    {
+      "epoch": 0.027733629158711026,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.886934673366834e-05,
+      "loss": 1.5434,
+      "step": 325
+    },
+    {
+      "epoch": 0.02781896340227629,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.886432160804021e-05,
+      "loss": 1.5579,
+      "step": 326
+    },
+    {
+      "epoch": 0.027904297645841555,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.885929648241206e-05,
+      "loss": 1.5034,
+      "step": 327
+    },
+    {
+      "epoch": 0.02798963188940682,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.885427135678393e-05,
+      "loss": 1.5051,
+      "step": 328
+    },
+    {
+      "epoch": 0.028074966132972085,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.884924623115577e-05,
+      "loss": 1.4347,
+      "step": 329
+    },
+    {
+      "epoch": 0.02816030037653735,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.884422110552765e-05,
+      "loss": 1.2852,
+      "step": 330
+    },
+    {
+      "epoch": 0.028245634620102614,
+      "grad_norm": 0.39453125,
+      "learning_rate": 9.88391959798995e-05,
+      "loss": 1.5375,
+      "step": 331
+    },
+    {
+      "epoch": 0.02833096886366788,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.883417085427136e-05,
+      "loss": 1.4506,
+      "step": 332
+    },
+    {
+      "epoch": 0.028416303107233143,
+      "grad_norm": 0.361328125,
+      "learning_rate": 9.882914572864322e-05,
+      "loss": 1.4868,
+      "step": 333
+    },
+    {
+      "epoch": 0.028501637350798407,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.882412060301508e-05,
+      "loss": 1.5473,
+      "step": 334
+    },
+    {
+      "epoch": 0.028586971594363672,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.881909547738693e-05,
+      "loss": 1.3555,
+      "step": 335
+    },
+    {
+      "epoch": 0.028672305837928937,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.881407035175881e-05,
+      "loss": 1.4413,
+      "step": 336
+    },
+    {
+      "epoch": 0.0287576400814942,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.880904522613066e-05,
+      "loss": 1.419,
+      "step": 337
+    },
+    {
+      "epoch": 0.028842974325059466,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.880402010050252e-05,
+      "loss": 1.4604,
+      "step": 338
+    },
+    {
+      "epoch": 0.02892830856862473,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.879899497487438e-05,
+      "loss": 1.4172,
+      "step": 339
+    },
+    {
+      "epoch": 0.029013642812189995,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.879396984924624e-05,
+      "loss": 1.4164,
+      "step": 340
+    },
+    {
+      "epoch": 0.029098977055755263,
+      "grad_norm": 0.50390625,
+      "learning_rate": 9.878894472361809e-05,
+      "loss": 1.3261,
+      "step": 341
+    },
+    {
+      "epoch": 0.029184311299320528,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.878391959798995e-05,
+      "loss": 1.4579,
+      "step": 342
+    },
+    {
+      "epoch": 0.029269645542885792,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.877889447236182e-05,
+      "loss": 1.4607,
+      "step": 343
+    },
+    {
+      "epoch": 0.029354979786451057,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.877386934673368e-05,
+      "loss": 1.4725,
+      "step": 344
+    },
+    {
+      "epoch": 0.02944031403001632,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.876884422110553e-05,
+      "loss": 1.2707,
+      "step": 345
+    },
+    {
+      "epoch": 0.029525648273581586,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.87638190954774e-05,
+      "loss": 1.2084,
+      "step": 346
+    },
+    {
+      "epoch": 0.02961098251714685,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.875879396984925e-05,
+      "loss": 1.5012,
+      "step": 347
+    },
+    {
+      "epoch": 0.029696316760712115,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.875376884422111e-05,
+      "loss": 1.4598,
+      "step": 348
+    },
+    {
+      "epoch": 0.02978165100427738,
+      "grad_norm": 0.36328125,
+      "learning_rate": 9.874874371859297e-05,
+      "loss": 1.5267,
+      "step": 349
+    },
+    {
+      "epoch": 0.029866985247842644,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.874371859296482e-05,
+      "loss": 1.4719,
+      "step": 350
+    },
+    {
+      "epoch": 0.02995231949140791,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.873869346733668e-05,
+      "loss": 1.4309,
+      "step": 351
+    },
+    {
+      "epoch": 0.030037653734973174,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.873366834170855e-05,
+      "loss": 1.4229,
+      "step": 352
+    },
+    {
+      "epoch": 0.030122987978538438,
+      "grad_norm": 0.361328125,
+      "learning_rate": 9.872864321608041e-05,
+      "loss": 1.321,
+      "step": 353
+    },
+    {
+      "epoch": 0.030208322222103703,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.872361809045226e-05,
+      "loss": 1.5606,
+      "step": 354
+    },
+    {
+      "epoch": 0.030293656465668967,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.871859296482413e-05,
+      "loss": 1.418,
+      "step": 355
+    },
+    {
+      "epoch": 0.030378990709234232,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.871356783919598e-05,
+      "loss": 1.3772,
+      "step": 356
+    },
+    {
+      "epoch": 0.030464324952799497,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.870854271356784e-05,
+      "loss": 1.483,
+      "step": 357
+    },
+    {
+      "epoch": 0.03054965919636476,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.870351758793969e-05,
+      "loss": 1.4651,
+      "step": 358
+    },
+    {
+      "epoch": 0.030634993439930026,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.869849246231157e-05,
+      "loss": 1.3877,
+      "step": 359
+    },
+    {
+      "epoch": 0.03072032768349529,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.869346733668342e-05,
+      "loss": 1.5275,
+      "step": 360
+    },
+    {
+      "epoch": 0.030805661927060555,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.868844221105528e-05,
+      "loss": 1.462,
+      "step": 361
+    },
+    {
+      "epoch": 0.03089099617062582,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.868341708542714e-05,
+      "loss": 1.5315,
+      "step": 362
+    },
+    {
+      "epoch": 0.030976330414191084,
+      "grad_norm": 0.375,
+      "learning_rate": 9.8678391959799e-05,
+      "loss": 1.44,
+      "step": 363
+    },
+    {
+      "epoch": 0.03106166465775635,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.867336683417085e-05,
+      "loss": 1.429,
+      "step": 364
+    },
+    {
+      "epoch": 0.031146998901321613,
+      "grad_norm": 0.390625,
+      "learning_rate": 9.866834170854273e-05,
+      "loss": 1.4823,
+      "step": 365
+    },
+    {
+      "epoch": 0.031232333144886878,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.866331658291457e-05,
+      "loss": 1.596,
+      "step": 366
+    },
+    {
+      "epoch": 0.031317667388452146,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.865829145728644e-05,
+      "loss": 1.482,
+      "step": 367
+    },
+    {
+      "epoch": 0.03140300163201741,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.86532663316583e-05,
+      "loss": 1.4981,
+      "step": 368
+    },
+    {
+      "epoch": 0.031488335875582675,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.864824120603016e-05,
+      "loss": 1.3063,
+      "step": 369
+    },
+    {
+      "epoch": 0.03157367011914794,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.864321608040201e-05,
+      "loss": 1.267,
+      "step": 370
+    },
+    {
+      "epoch": 0.031659004362713204,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.863819095477388e-05,
+      "loss": 1.2516,
+      "step": 371
+    },
+    {
+      "epoch": 0.03174433860627847,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.863316582914573e-05,
+      "loss": 1.3485,
+      "step": 372
+    },
+    {
+      "epoch": 0.03182967284984373,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.86281407035176e-05,
+      "loss": 1.4993,
+      "step": 373
+    },
+    {
+      "epoch": 0.031915007093409,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.862311557788944e-05,
+      "loss": 1.4814,
+      "step": 374
+    },
+    {
+      "epoch": 0.03200034133697426,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.861809045226132e-05,
+      "loss": 1.5224,
+      "step": 375
+    },
+    {
+      "epoch": 0.03208567558053953,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.861306532663317e-05,
+      "loss": 1.4168,
+      "step": 376
+    },
+    {
+      "epoch": 0.03217100982410479,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.860804020100503e-05,
+      "loss": 1.2661,
+      "step": 377
+    },
+    {
+      "epoch": 0.032256344067670056,
+      "grad_norm": 0.328125,
+      "learning_rate": 9.860301507537689e-05,
+      "loss": 1.3328,
+      "step": 378
+    },
+    {
+      "epoch": 0.03234167831123532,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.859798994974875e-05,
+      "loss": 1.2235,
+      "step": 379
+    },
+    {
+      "epoch": 0.032427012554800586,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.85929648241206e-05,
+      "loss": 1.6012,
+      "step": 380
+    },
+    {
+      "epoch": 0.03251234679836585,
+      "grad_norm": 0.361328125,
+      "learning_rate": 9.858793969849246e-05,
+      "loss": 1.4359,
+      "step": 381
+    },
+    {
+      "epoch": 0.032597681041931115,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.858291457286433e-05,
+      "loss": 1.4301,
+      "step": 382
+    },
+    {
+      "epoch": 0.03268301528549638,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.857788944723619e-05,
+      "loss": 1.3836,
+      "step": 383
+    },
+    {
+      "epoch": 0.032768349529061644,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.857286432160805e-05,
+      "loss": 1.3022,
+      "step": 384
+    },
+    {
+      "epoch": 0.03285368377262691,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.85678391959799e-05,
+      "loss": 1.4875,
+      "step": 385
+    },
+    {
+      "epoch": 0.03293901801619217,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.856281407035176e-05,
+      "loss": 1.3937,
+      "step": 386
+    },
+    {
+      "epoch": 0.03302435225975744,
+      "grad_norm": 0.43359375,
+      "learning_rate": 9.855778894472362e-05,
+      "loss": 1.4887,
+      "step": 387
+    },
+    {
+      "epoch": 0.0331096865033227,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.855276381909548e-05,
+      "loss": 1.3579,
+      "step": 388
+    },
+    {
+      "epoch": 0.03319502074688797,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.854773869346733e-05,
+      "loss": 1.3316,
+      "step": 389
+    },
+    {
+      "epoch": 0.03328035499045323,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.85427135678392e-05,
+      "loss": 1.3678,
+      "step": 390
+    },
+    {
+      "epoch": 0.033365689234018496,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.853768844221106e-05,
+      "loss": 1.4474,
+      "step": 391
+    },
+    {
+      "epoch": 0.03345102347758376,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.853266331658292e-05,
+      "loss": 1.388,
+      "step": 392
+    },
+    {
+      "epoch": 0.033536357721149025,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.852763819095477e-05,
+      "loss": 1.2983,
+      "step": 393
+    },
+    {
+      "epoch": 0.03362169196471429,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.852261306532664e-05,
+      "loss": 1.5462,
+      "step": 394
+    },
+    {
+      "epoch": 0.033707026208279554,
+      "grad_norm": 0.310546875,
+      "learning_rate": 9.851758793969849e-05,
+      "loss": 1.3843,
+      "step": 395
+    },
+    {
+      "epoch": 0.03379236045184482,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.851256281407035e-05,
+      "loss": 1.5755,
+      "step": 396
+    },
+    {
+      "epoch": 0.033877694695410084,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.850753768844222e-05,
+      "loss": 1.3772,
+      "step": 397
+    },
+    {
+      "epoch": 0.03396302893897535,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.850251256281408e-05,
+      "loss": 1.3809,
+      "step": 398
+    },
+    {
+      "epoch": 0.03404836318254061,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.849748743718593e-05,
+      "loss": 1.4123,
+      "step": 399
+    },
+    {
+      "epoch": 0.03413369742610588,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.84924623115578e-05,
+      "loss": 1.5279,
+      "step": 400
+    },
+    {
+      "epoch": 0.03421903166967114,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.848743718592965e-05,
+      "loss": 1.5218,
+      "step": 401
+    },
+    {
+      "epoch": 0.034304365913236406,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.848241206030151e-05,
+      "loss": 1.5075,
+      "step": 402
+    },
+    {
+      "epoch": 0.03438970015680167,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.847738693467337e-05,
+      "loss": 1.3305,
+      "step": 403
+    },
+    {
+      "epoch": 0.034475034400366936,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.847236180904524e-05,
+      "loss": 1.3203,
+      "step": 404
+    },
+    {
+      "epoch": 0.0345603686439322,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.846733668341709e-05,
+      "loss": 1.3874,
+      "step": 405
+    },
+    {
+      "epoch": 0.034645702887497465,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.846231155778895e-05,
+      "loss": 1.4767,
+      "step": 406
+    },
+    {
+      "epoch": 0.03473103713106273,
+      "grad_norm": 0.3203125,
+      "learning_rate": 9.845728643216081e-05,
+      "loss": 1.3661,
+      "step": 407
+    },
+    {
+      "epoch": 0.034816371374627994,
+      "grad_norm": 0.400390625,
+      "learning_rate": 9.845226130653267e-05,
+      "loss": 1.3396,
+      "step": 408
+    },
+    {
+      "epoch": 0.03490170561819326,
+      "grad_norm": 0.33984375,
+      "learning_rate": 9.844723618090452e-05,
+      "loss": 1.4862,
+      "step": 409
+    },
+    {
+      "epoch": 0.03498703986175852,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.84422110552764e-05,
+      "loss": 1.4893,
+      "step": 410
+    },
+    {
+      "epoch": 0.03507237410532379,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.843718592964824e-05,
+      "loss": 1.4895,
+      "step": 411
+    },
+    {
+      "epoch": 0.03515770834888905,
+      "grad_norm": 0.328125,
+      "learning_rate": 9.84321608040201e-05,
+      "loss": 1.4898,
+      "step": 412
+    },
+    {
+      "epoch": 0.03524304259245432,
+      "grad_norm": 0.3203125,
+      "learning_rate": 9.842713567839197e-05,
+      "loss": 1.2919,
+      "step": 413
+    },
+    {
+      "epoch": 0.03532837683601958,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.842211055276383e-05,
+      "loss": 1.4775,
+      "step": 414
+    },
+    {
+      "epoch": 0.035413711079584846,
+      "grad_norm": 0.298828125,
+      "learning_rate": 9.841708542713568e-05,
+      "loss": 1.3153,
+      "step": 415
+    },
+    {
+      "epoch": 0.03549904532315011,
+      "grad_norm": 0.294921875,
+      "learning_rate": 9.841206030150754e-05,
+      "loss": 1.2187,
+      "step": 416
+    },
+    {
+      "epoch": 0.035584379566715375,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.84070351758794e-05,
+      "loss": 1.3458,
+      "step": 417
+    },
+    {
+      "epoch": 0.03566971381028064,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.840201005025126e-05,
+      "loss": 1.3554,
+      "step": 418
+    },
+    {
+      "epoch": 0.035755048053845905,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.839698492462313e-05,
+      "loss": 1.3247,
+      "step": 419
+    },
+    {
+      "epoch": 0.03584038229741117,
+      "grad_norm": 0.33984375,
+      "learning_rate": 9.839195979899497e-05,
+      "loss": 1.4921,
+      "step": 420
+    },
+    {
+      "epoch": 0.035925716540976434,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.838693467336684e-05,
+      "loss": 1.3241,
+      "step": 421
+    },
+    {
+      "epoch": 0.036011050784541705,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.83819095477387e-05,
+      "loss": 1.456,
+      "step": 422
+    },
+    {
+      "epoch": 0.03609638502810697,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.837688442211056e-05,
+      "loss": 1.2913,
+      "step": 423
+    },
+    {
+      "epoch": 0.036181719271672234,
+      "grad_norm": 0.4609375,
+      "learning_rate": 9.837185929648241e-05,
+      "loss": 1.5285,
+      "step": 424
+    },
+    {
+      "epoch": 0.0362670535152375,
+      "grad_norm": 0.33984375,
+      "learning_rate": 9.836683417085427e-05,
+      "loss": 1.4796,
+      "step": 425
+    },
+    {
+      "epoch": 0.036352387758802764,
+      "grad_norm": 0.3203125,
+      "learning_rate": 9.836180904522613e-05,
+      "loss": 1.3067,
+      "step": 426
+    },
+    {
+      "epoch": 0.03643772200236803,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.8356783919598e-05,
+      "loss": 1.5642,
+      "step": 427
+    },
+    {
+      "epoch": 0.03652305624593329,
+      "grad_norm": 0.328125,
+      "learning_rate": 9.835175879396984e-05,
+      "loss": 1.4515,
+      "step": 428
+    },
+    {
+      "epoch": 0.03660839048949856,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.834673366834172e-05,
+      "loss": 1.3682,
+      "step": 429
+    },
+    {
+      "epoch": 0.03669372473306382,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.834170854271357e-05,
+      "loss": 1.4455,
+      "step": 430
+    },
+    {
+      "epoch": 0.036779058976629087,
+      "grad_norm": 0.302734375,
+      "learning_rate": 9.833668341708543e-05,
+      "loss": 1.4017,
+      "step": 431
+    },
+    {
+      "epoch": 0.03686439322019435,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.833165829145729e-05,
+      "loss": 1.3697,
+      "step": 432
+    },
+    {
+      "epoch": 0.036949727463759616,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.832663316582915e-05,
+      "loss": 1.2995,
+      "step": 433
+    },
+    {
+      "epoch": 0.03703506170732488,
+      "grad_norm": 0.365234375,
+      "learning_rate": 9.8321608040201e-05,
+      "loss": 1.4741,
+      "step": 434
+    },
+    {
+      "epoch": 0.037120395950890145,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.831658291457288e-05,
+      "loss": 1.3424,
+      "step": 435
+    },
+    {
+      "epoch": 0.03720573019445541,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.831155778894473e-05,
+      "loss": 1.3283,
+      "step": 436
+    },
+    {
+      "epoch": 0.037291064438020674,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.830653266331659e-05,
+      "loss": 1.4287,
+      "step": 437
+    },
+    {
+      "epoch": 0.03737639868158594,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.830150753768844e-05,
+      "loss": 1.5556,
+      "step": 438
+    },
+    {
+      "epoch": 0.0374617329251512,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.829648241206031e-05,
+      "loss": 1.398,
+      "step": 439
+    },
+    {
+      "epoch": 0.03754706716871647,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.829145728643216e-05,
+      "loss": 1.4242,
+      "step": 440
+    },
+    {
+      "epoch": 0.03763240141228173,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.828643216080402e-05,
+      "loss": 1.3893,
+      "step": 441
+    },
+    {
+      "epoch": 0.037717735655847,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.828140703517589e-05,
+      "loss": 1.5674,
+      "step": 442
+    },
+    {
+      "epoch": 0.03780306989941226,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.827638190954775e-05,
+      "loss": 1.3771,
+      "step": 443
+    },
+    {
+      "epoch": 0.037888404142977526,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.82713567839196e-05,
+      "loss": 1.4542,
+      "step": 444
+    },
+    {
+      "epoch": 0.03797373838654279,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.826633165829147e-05,
+      "loss": 1.446,
+      "step": 445
+    },
+    {
+      "epoch": 0.038059072630108055,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.826130653266332e-05,
+      "loss": 1.3855,
+      "step": 446
+    },
+    {
+      "epoch": 0.03814440687367332,
+      "grad_norm": 0.33984375,
+      "learning_rate": 9.825628140703518e-05,
+      "loss": 1.4553,
+      "step": 447
+    },
+    {
+      "epoch": 0.038229741117238585,
+      "grad_norm": 0.416015625,
+      "learning_rate": 9.825125628140704e-05,
+      "loss": 1.4399,
+      "step": 448
+    },
+    {
+      "epoch": 0.03831507536080385,
+      "grad_norm": 0.328125,
+      "learning_rate": 9.82462311557789e-05,
+      "loss": 1.4373,
+      "step": 449
+    },
+    {
+      "epoch": 0.038400409604369114,
+      "grad_norm": 0.484375,
+      "learning_rate": 9.824120603015075e-05,
+      "loss": 1.5411,
+      "step": 450
+    },
+    {
+      "epoch": 0.03848574384793438,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.823618090452262e-05,
+      "loss": 1.3595,
+      "step": 451
+    },
+    {
+      "epoch": 0.03857107809149964,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.823115577889448e-05,
+      "loss": 1.2787,
+      "step": 452
+    },
+    {
+      "epoch": 0.03865641233506491,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.822613065326634e-05,
+      "loss": 1.3211,
+      "step": 453
+    },
+    {
+      "epoch": 0.03874174657863017,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.822110552763819e-05,
+      "loss": 1.2747,
+      "step": 454
+    },
+    {
+      "epoch": 0.03882708082219544,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.821608040201005e-05,
+      "loss": 1.3369,
+      "step": 455
+    },
+    {
+      "epoch": 0.0389124150657607,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.821105527638191e-05,
+      "loss": 1.4238,
+      "step": 456
+    },
+    {
+      "epoch": 0.038997749309325966,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.820603015075378e-05,
+      "loss": 1.4062,
+      "step": 457
+    },
+    {
+      "epoch": 0.03908308355289123,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.820100502512564e-05,
+      "loss": 1.5235,
+      "step": 458
+    },
+    {
+      "epoch": 0.039168417796456495,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.819597989949749e-05,
+      "loss": 1.4112,
+      "step": 459
+    },
+    {
+      "epoch": 0.03925375204002176,
+      "grad_norm": 0.388671875,
+      "learning_rate": 9.819095477386935e-05,
+      "loss": 1.3974,
+      "step": 460
+    },
+    {
+      "epoch": 0.039339086283587024,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.818592964824121e-05,
+      "loss": 1.566,
+      "step": 461
+    },
+    {
+      "epoch": 0.03942442052715229,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.818090452261307e-05,
+      "loss": 1.4603,
+      "step": 462
+    },
+    {
+      "epoch": 0.03950975477071755,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.817587939698492e-05,
+      "loss": 1.467,
+      "step": 463
+    },
+    {
+      "epoch": 0.03959508901428282,
+      "grad_norm": 0.2578125,
+      "learning_rate": 9.81708542713568e-05,
+      "loss": 2.2431,
+      "step": 464
+    },
+    {
+      "epoch": 0.03968042325784808,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.816582914572864e-05,
+      "loss": 1.3986,
+      "step": 465
+    },
+    {
+      "epoch": 0.03976575750141335,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.81608040201005e-05,
+      "loss": 1.3912,
+      "step": 466
+    },
+    {
+      "epoch": 0.03985109174497861,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.815577889447236e-05,
+      "loss": 1.3503,
+      "step": 467
+    },
+    {
+      "epoch": 0.039936425988543876,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.815075376884423e-05,
+      "loss": 1.2785,
+      "step": 468
+    },
+    {
+      "epoch": 0.04002176023210914,
+      "grad_norm": 0.29296875,
+      "learning_rate": 9.814572864321608e-05,
+      "loss": 1.4368,
+      "step": 469
+    },
+    {
+      "epoch": 0.040107094475674405,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.814070351758794e-05,
+      "loss": 1.248,
+      "step": 470
+    },
+    {
+      "epoch": 0.04019242871923967,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.81356783919598e-05,
+      "loss": 1.2747,
+      "step": 471
+    },
+    {
+      "epoch": 0.040277762962804935,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.813065326633167e-05,
+      "loss": 1.1944,
+      "step": 472
+    },
+    {
+      "epoch": 0.0403630972063702,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.812562814070351e-05,
+      "loss": 1.3729,
+      "step": 473
+    },
+    {
+      "epoch": 0.040448431449935464,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.812060301507539e-05,
+      "loss": 1.4525,
+      "step": 474
+    },
+    {
+      "epoch": 0.04053376569350073,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.811557788944724e-05,
+      "loss": 1.5407,
+      "step": 475
+    },
+    {
+      "epoch": 0.04061909993706599,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.81105527638191e-05,
+      "loss": 1.4424,
+      "step": 476
+    },
+    {
+      "epoch": 0.04070443418063126,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.810552763819096e-05,
+      "loss": 1.3422,
+      "step": 477
+    },
+    {
+      "epoch": 0.04078976842419652,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.810050251256282e-05,
+      "loss": 1.4135,
+      "step": 478
+    },
+    {
+      "epoch": 0.04087510266776179,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.809547738693467e-05,
+      "loss": 1.343,
+      "step": 479
+    },
+    {
+      "epoch": 0.04096043691132705,
+      "grad_norm": 0.302734375,
+      "learning_rate": 9.809045226130655e-05,
+      "loss": 1.5094,
+      "step": 480
+    },
+    {
+      "epoch": 0.041045771154892316,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.80854271356784e-05,
+      "loss": 1.185,
+      "step": 481
+    },
+    {
+      "epoch": 0.04113110539845758,
+      "grad_norm": 0.419921875,
+      "learning_rate": 9.808040201005026e-05,
+      "loss": 1.4119,
+      "step": 482
+    },
+    {
+      "epoch": 0.041216439642022845,
+      "grad_norm": 0.375,
+      "learning_rate": 9.807537688442211e-05,
+      "loss": 1.2592,
+      "step": 483
+    },
+    {
+      "epoch": 0.04130177388558811,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.807035175879398e-05,
+      "loss": 1.3392,
+      "step": 484
+    },
+    {
+      "epoch": 0.041387108129153374,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.806532663316583e-05,
+      "loss": 1.5251,
+      "step": 485
+    },
+    {
+      "epoch": 0.04147244237271864,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.806030150753769e-05,
+      "loss": 1.5757,
+      "step": 486
+    },
+    {
+      "epoch": 0.04155777661628391,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.805527638190956e-05,
+      "loss": 1.3507,
+      "step": 487
+    },
+    {
+      "epoch": 0.041643110859849175,
+      "grad_norm": 0.431640625,
+      "learning_rate": 9.805025125628142e-05,
+      "loss": 1.3745,
+      "step": 488
+    },
+    {
+      "epoch": 0.04172844510341444,
+      "grad_norm": 0.310546875,
+      "learning_rate": 9.804522613065327e-05,
+      "loss": 1.3922,
+      "step": 489
+    },
+    {
+      "epoch": 0.041813779346979704,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.804020100502513e-05,
+      "loss": 1.3142,
+      "step": 490
+    },
+    {
+      "epoch": 0.04189911359054497,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.803517587939699e-05,
+      "loss": 1.2545,
+      "step": 491
+    },
+    {
+      "epoch": 0.04198444783411023,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.803015075376885e-05,
+      "loss": 1.3803,
+      "step": 492
+    },
+    {
+      "epoch": 0.0420697820776755,
+      "grad_norm": 0.41015625,
+      "learning_rate": 9.802512562814071e-05,
+      "loss": 1.4888,
+      "step": 493
+    },
+    {
+      "epoch": 0.04215511632124076,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.802010050251256e-05,
+      "loss": 1.5492,
+      "step": 494
+    },
+    {
+      "epoch": 0.04224045056480603,
+      "grad_norm": 0.310546875,
+      "learning_rate": 9.801507537688442e-05,
+      "loss": 1.4367,
+      "step": 495
+    },
+    {
+      "epoch": 0.04232578480837129,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.801005025125629e-05,
+      "loss": 1.3596,
+      "step": 496
+    },
+    {
+      "epoch": 0.042411119051936556,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.800502512562815e-05,
+      "loss": 1.5326,
+      "step": 497
+    },
+    {
+      "epoch": 0.04249645329550182,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.8e-05,
+      "loss": 1.3891,
+      "step": 498
+    },
+    {
+      "epoch": 0.042581787539067086,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.799497487437186e-05,
+      "loss": 1.4772,
+      "step": 499
+    },
+    {
+      "epoch": 0.04266712178263235,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.798994974874372e-05,
+      "loss": 1.4528,
+      "step": 500
+    },
+    {
+      "epoch": 0.042752456026197615,
+      "grad_norm": 0.365234375,
+      "learning_rate": 9.798492462311558e-05,
+      "loss": 1.421,
+      "step": 501
+    },
+    {
+      "epoch": 0.04283779026976288,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.797989949748743e-05,
+      "loss": 1.4449,
+      "step": 502
+    },
+    {
+      "epoch": 0.042923124513328144,
+      "grad_norm": 0.36328125,
+      "learning_rate": 9.797487437185931e-05,
+      "loss": 1.4574,
+      "step": 503
+    },
+    {
+      "epoch": 0.04300845875689341,
+      "grad_norm": 0.466796875,
+      "learning_rate": 9.796984924623116e-05,
+      "loss": 1.6706,
+      "step": 504
+    },
+    {
+      "epoch": 0.04309379300045867,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.796482412060302e-05,
+      "loss": 1.3832,
+      "step": 505
+    },
+    {
+      "epoch": 0.04317912724402394,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.795979899497488e-05,
+      "loss": 1.4373,
+      "step": 506
+    },
+    {
+      "epoch": 0.0432644614875892,
+      "grad_norm": 0.376953125,
+      "learning_rate": 9.795477386934674e-05,
+      "loss": 1.5441,
+      "step": 507
+    },
+    {
+      "epoch": 0.04334979573115447,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.794974874371859e-05,
+      "loss": 1.4249,
+      "step": 508
+    },
+    {
+      "epoch": 0.04343512997471973,
+      "grad_norm": 0.36328125,
+      "learning_rate": 9.794472361809047e-05,
+      "loss": 1.38,
+      "step": 509
+    },
+    {
+      "epoch": 0.043520464218284996,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.793969849246231e-05,
+      "loss": 1.3048,
+      "step": 510
+    },
+    {
+      "epoch": 0.04360579846185026,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.793467336683418e-05,
+      "loss": 1.4716,
+      "step": 511
+    },
+    {
+      "epoch": 0.043691132705415525,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.792964824120604e-05,
+      "loss": 1.3129,
+      "step": 512
+    },
+    {
+      "epoch": 0.04377646694898079,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.79246231155779e-05,
+      "loss": 1.4984,
+      "step": 513
+    },
+    {
+      "epoch": 0.043861801192546054,
+      "grad_norm": 0.451171875,
+      "learning_rate": 9.791959798994975e-05,
+      "loss": 1.3882,
+      "step": 514
+    },
+    {
+      "epoch": 0.04394713543611132,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.791457286432161e-05,
+      "loss": 1.4305,
+      "step": 515
+    },
+    {
+      "epoch": 0.044032469679676584,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.790954773869347e-05,
+      "loss": 1.2544,
+      "step": 516
+    },
+    {
+      "epoch": 0.04411780392324185,
+      "grad_norm": 0.291015625,
+      "learning_rate": 9.790452261306533e-05,
+      "loss": 1.4185,
+      "step": 517
+    },
+    {
+      "epoch": 0.04420313816680711,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.789949748743718e-05,
+      "loss": 1.355,
+      "step": 518
+    },
+    {
+      "epoch": 0.04428847241037238,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.789447236180906e-05,
+      "loss": 1.5492,
+      "step": 519
+    },
+    {
+      "epoch": 0.04437380665393764,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.788944723618091e-05,
+      "loss": 1.3477,
+      "step": 520
+    },
+    {
+      "epoch": 0.044459140897502906,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.788442211055277e-05,
+      "loss": 1.3968,
+      "step": 521
+    },
+    {
+      "epoch": 0.04454447514106817,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.787939698492463e-05,
+      "loss": 1.2862,
+      "step": 522
+    },
+    {
+      "epoch": 0.044629809384633436,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.78743718592965e-05,
+      "loss": 1.2657,
+      "step": 523
+    },
+    {
+      "epoch": 0.0447151436281987,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.786934673366834e-05,
+      "loss": 1.3799,
+      "step": 524
+    },
+    {
+      "epoch": 0.044800477871763965,
+      "grad_norm": 0.287109375,
+      "learning_rate": 9.78643216080402e-05,
+      "loss": 1.1775,
+      "step": 525
+    },
+    {
+      "epoch": 0.04488581211532923,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.785929648241207e-05,
+      "loss": 1.3675,
+      "step": 526
+    },
+    {
+      "epoch": 0.044971146358894494,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.785427135678393e-05,
+      "loss": 1.3998,
+      "step": 527
+    },
+    {
+      "epoch": 0.04505648060245976,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.784924623115579e-05,
+      "loss": 1.3718,
+      "step": 528
+    },
+    {
+      "epoch": 0.04514181484602502,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.784422110552764e-05,
+      "loss": 1.4417,
+      "step": 529
+    },
+    {
+      "epoch": 0.04522714908959029,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.78391959798995e-05,
+      "loss": 1.3526,
+      "step": 530
+    },
+    {
+      "epoch": 0.04531248333315555,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.783417085427136e-05,
+      "loss": 1.3731,
+      "step": 531
+    },
+    {
+      "epoch": 0.04539781757672082,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.782914572864322e-05,
+      "loss": 1.3264,
+      "step": 532
+    },
+    {
+      "epoch": 0.04548315182028608,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.782412060301507e-05,
+      "loss": 1.3301,
+      "step": 533
+    },
+    {
+      "epoch": 0.045568486063851346,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.781909547738694e-05,
+      "loss": 1.43,
+      "step": 534
+    },
+    {
+      "epoch": 0.04565382030741661,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.78140703517588e-05,
+      "loss": 1.3434,
+      "step": 535
+    },
+    {
+      "epoch": 0.045739154550981875,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.780904522613066e-05,
+      "loss": 1.3274,
+      "step": 536
+    },
+    {
+      "epoch": 0.04582448879454714,
+      "grad_norm": 0.296875,
+      "learning_rate": 9.780402010050251e-05,
+      "loss": 1.3934,
+      "step": 537
+    },
+    {
+      "epoch": 0.045909823038112405,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.779899497487438e-05,
+      "loss": 1.375,
+      "step": 538
+    },
+    {
+      "epoch": 0.04599515728167767,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.779396984924623e-05,
+      "loss": 1.4701,
+      "step": 539
+    },
+    {
+      "epoch": 0.046080491525242934,
+      "grad_norm": 0.296875,
+      "learning_rate": 9.77889447236181e-05,
+      "loss": 1.3611,
+      "step": 540
+    },
+    {
+      "epoch": 0.0461658257688082,
+      "grad_norm": 0.287109375,
+      "learning_rate": 9.778391959798996e-05,
+      "loss": 1.3605,
+      "step": 541
+    },
+    {
+      "epoch": 0.04625116001237346,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.777889447236182e-05,
+      "loss": 1.3663,
+      "step": 542
+    },
+    {
+      "epoch": 0.04633649425593873,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.777386934673367e-05,
+      "loss": 1.3364,
+      "step": 543
+    },
+    {
+      "epoch": 0.04642182849950399,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.776884422110554e-05,
+      "loss": 1.3242,
+      "step": 544
+    },
+    {
+      "epoch": 0.04650716274306926,
+      "grad_norm": 0.36328125,
+      "learning_rate": 9.776381909547739e-05,
+      "loss": 1.377,
+      "step": 545
+    },
+    {
+      "epoch": 0.04659249698663452,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.775879396984925e-05,
+      "loss": 1.4118,
+      "step": 546
+    },
+    {
+      "epoch": 0.046677831230199786,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.77537688442211e-05,
+      "loss": 1.2553,
+      "step": 547
+    },
+    {
+      "epoch": 0.04676316547376505,
+      "grad_norm": 0.3203125,
+      "learning_rate": 9.774874371859298e-05,
+      "loss": 1.3402,
+      "step": 548
+    },
+    {
+      "epoch": 0.046848499717330315,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.774371859296483e-05,
+      "loss": 1.3096,
+      "step": 549
+    },
+    {
+      "epoch": 0.04693383396089558,
+      "grad_norm": 0.361328125,
+      "learning_rate": 9.773869346733669e-05,
+      "loss": 1.4166,
+      "step": 550
+    },
+    {
+      "epoch": 0.047019168204460844,
+      "grad_norm": 0.28125,
+      "learning_rate": 9.773366834170855e-05,
+      "loss": 1.385,
+      "step": 551
+    },
+    {
+      "epoch": 0.047104502448026116,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.772864321608041e-05,
+      "loss": 1.317,
+      "step": 552
+    },
+    {
+      "epoch": 0.04718983669159138,
+      "grad_norm": 0.310546875,
+      "learning_rate": 9.772361809045226e-05,
+      "loss": 1.415,
+      "step": 553
+    },
+    {
+      "epoch": 0.047275170935156645,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.771859296482414e-05,
+      "loss": 1.3261,
+      "step": 554
+    },
+    {
+      "epoch": 0.04736050517872191,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.771356783919598e-05,
+      "loss": 1.4613,
+      "step": 555
+    },
+    {
+      "epoch": 0.047445839422287174,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.770854271356785e-05,
+      "loss": 1.3791,
+      "step": 556
+    },
+    {
+      "epoch": 0.04753117366585244,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.770351758793971e-05,
+      "loss": 1.3732,
+      "step": 557
+    },
+    {
+      "epoch": 0.0476165079094177,
+      "grad_norm": 0.376953125,
+      "learning_rate": 9.769849246231157e-05,
+      "loss": 1.4468,
+      "step": 558
+    },
+    {
+      "epoch": 0.04770184215298297,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.769346733668342e-05,
+      "loss": 1.183,
+      "step": 559
+    },
+    {
+      "epoch": 0.04778717639654823,
+      "grad_norm": 0.302734375,
+      "learning_rate": 9.768844221105528e-05,
+      "loss": 1.3469,
+      "step": 560
+    },
+    {
+      "epoch": 0.0478725106401135,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.768341708542714e-05,
+      "loss": 1.4412,
+      "step": 561
+    },
+    {
+      "epoch": 0.04795784488367876,
+      "grad_norm": 0.328125,
+      "learning_rate": 9.7678391959799e-05,
+      "loss": 1.4291,
+      "step": 562
+    },
+    {
+      "epoch": 0.048043179127244026,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.767336683417085e-05,
+      "loss": 1.2476,
+      "step": 563
+    },
+    {
+      "epoch": 0.04812851337080929,
+      "grad_norm": 0.328125,
+      "learning_rate": 9.766834170854271e-05,
+      "loss": 1.5017,
+      "step": 564
+    },
+    {
+      "epoch": 0.048213847614374555,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.766331658291458e-05,
+      "loss": 1.3366,
+      "step": 565
+    },
+    {
+      "epoch": 0.04829918185793982,
+      "grad_norm": 0.427734375,
+      "learning_rate": 9.765829145728644e-05,
+      "loss": 1.7742,
+      "step": 566
+    },
+    {
+      "epoch": 0.048384516101505085,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.76532663316583e-05,
+      "loss": 1.4306,
+      "step": 567
+    },
+    {
+      "epoch": 0.04846985034507035,
+      "grad_norm": 0.298828125,
+      "learning_rate": 9.764824120603015e-05,
+      "loss": 1.3025,
+      "step": 568
+    },
+    {
+      "epoch": 0.048555184588635614,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.764321608040201e-05,
+      "loss": 1.4327,
+      "step": 569
+    },
+    {
+      "epoch": 0.04864051883220088,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.763819095477387e-05,
+      "loss": 1.4312,
+      "step": 570
+    },
+    {
+      "epoch": 0.04872585307576614,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.763316582914574e-05,
+      "loss": 1.4382,
+      "step": 571
+    },
+    {
+      "epoch": 0.04881118731933141,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.762814070351758e-05,
+      "loss": 1.2527,
+      "step": 572
+    },
+    {
+      "epoch": 0.04889652156289667,
+      "grad_norm": 0.36328125,
+      "learning_rate": 9.762311557788946e-05,
+      "loss": 1.4433,
+      "step": 573
+    },
+    {
+      "epoch": 0.04898185580646194,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.761809045226131e-05,
+      "loss": 1.4039,
+      "step": 574
+    },
+    {
+      "epoch": 0.0490671900500272,
+      "grad_norm": 0.388671875,
+      "learning_rate": 9.761306532663317e-05,
+      "loss": 1.3426,
+      "step": 575
+    },
+    {
+      "epoch": 0.049152524293592466,
+      "grad_norm": 0.302734375,
+      "learning_rate": 9.760804020100502e-05,
+      "loss": 1.2894,
+      "step": 576
+    },
+    {
+      "epoch": 0.04923785853715773,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.76030150753769e-05,
+      "loss": 1.4818,
+      "step": 577
+    },
+    {
+      "epoch": 0.049323192780722995,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.759798994974874e-05,
+      "loss": 1.6552,
+      "step": 578
+    },
+    {
+      "epoch": 0.04940852702428826,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.75929648241206e-05,
+      "loss": 1.3383,
+      "step": 579
+    },
+    {
+      "epoch": 0.049493861267853524,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.758793969849247e-05,
+      "loss": 1.4319,
+      "step": 580
+    },
+    {
+      "epoch": 0.04957919551141879,
+      "grad_norm": 0.328125,
+      "learning_rate": 9.758291457286433e-05,
+      "loss": 1.3755,
+      "step": 581
+    },
+    {
+      "epoch": 0.04966452975498405,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.757788944723618e-05,
+      "loss": 1.2544,
+      "step": 582
+    },
+    {
+      "epoch": 0.04974986399854932,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.757286432160805e-05,
+      "loss": 1.3336,
+      "step": 583
+    },
+    {
+      "epoch": 0.04983519824211458,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.75678391959799e-05,
+      "loss": 1.5186,
+      "step": 584
+    },
+    {
+      "epoch": 0.04992053248567985,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.756281407035176e-05,
+      "loss": 1.3453,
+      "step": 585
+    },
+    {
+      "epoch": 0.05000586672924511,
+      "grad_norm": 0.271484375,
+      "learning_rate": 9.755778894472363e-05,
+      "loss": 1.5808,
+      "step": 586
+    },
+    {
+      "epoch": 0.050091200972810376,
+      "grad_norm": 0.310546875,
+      "learning_rate": 9.755276381909549e-05,
+      "loss": 1.3914,
+      "step": 587
+    },
+    {
+      "epoch": 0.05017653521637564,
+      "grad_norm": 0.30078125,
+      "learning_rate": 9.754773869346734e-05,
+      "loss": 1.2982,
+      "step": 588
+    },
+    {
+      "epoch": 0.050261869459940905,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.754271356783921e-05,
+      "loss": 1.4805,
+      "step": 589
+    },
+    {
+      "epoch": 0.05034720370350617,
+      "grad_norm": 0.390625,
+      "learning_rate": 9.753768844221106e-05,
+      "loss": 1.3696,
+      "step": 590
+    },
+    {
+      "epoch": 0.050432537947071435,
+      "grad_norm": 0.310546875,
+      "learning_rate": 9.753266331658292e-05,
+      "loss": 1.428,
+      "step": 591
+    },
+    {
+      "epoch": 0.0505178721906367,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.752763819095477e-05,
+      "loss": 1.3564,
+      "step": 592
+    },
+    {
+      "epoch": 0.050603206434201964,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.752261306532665e-05,
+      "loss": 1.1273,
+      "step": 593
+    },
+    {
+      "epoch": 0.05068854067776723,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.75175879396985e-05,
+      "loss": 1.2114,
+      "step": 594
+    },
+    {
+      "epoch": 0.05077387492133249,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.751256281407036e-05,
+      "loss": 1.4617,
+      "step": 595
+    },
+    {
+      "epoch": 0.05085920916489776,
+      "grad_norm": 0.28515625,
+      "learning_rate": 9.750753768844222e-05,
+      "loss": 1.3907,
+      "step": 596
+    },
+    {
+      "epoch": 0.05094454340846302,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.750251256281408e-05,
+      "loss": 1.522,
+      "step": 597
+    },
+    {
+      "epoch": 0.05102987765202829,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.749748743718593e-05,
+      "loss": 1.3178,
+      "step": 598
+    },
+    {
+      "epoch": 0.05111521189559355,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.749246231155779e-05,
+      "loss": 1.4773,
+      "step": 599
+    },
+    {
+      "epoch": 0.051200546139158816,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.748743718592965e-05,
+      "loss": 1.2632,
+      "step": 600
+    },
+    {
+      "epoch": 0.05128588038272408,
+      "grad_norm": 0.33984375,
+      "learning_rate": 9.748241206030152e-05,
+      "loss": 1.4679,
+      "step": 601
+    },
+    {
+      "epoch": 0.051371214626289345,
+      "grad_norm": 0.328125,
+      "learning_rate": 9.747738693467338e-05,
+      "loss": 1.4967,
+      "step": 602
+    },
+    {
+      "epoch": 0.05145654886985461,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.747236180904523e-05,
+      "loss": 1.316,
+      "step": 603
+    },
+    {
+      "epoch": 0.051541883113419874,
+      "grad_norm": 0.294921875,
+      "learning_rate": 9.746733668341709e-05,
+      "loss": 1.3784,
+      "step": 604
+    },
+    {
+      "epoch": 0.05162721735698514,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.746231155778895e-05,
+      "loss": 1.4194,
+      "step": 605
+    },
+    {
+      "epoch": 0.051712551600550404,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.745728643216081e-05,
+      "loss": 1.4359,
+      "step": 606
+    },
+    {
+      "epoch": 0.05179788584411567,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.745226130653266e-05,
+      "loss": 1.3742,
+      "step": 607
+    },
+    {
+      "epoch": 0.05188322008768093,
+      "grad_norm": 0.310546875,
+      "learning_rate": 9.744723618090452e-05,
+      "loss": 1.3153,
+      "step": 608
+    },
+    {
+      "epoch": 0.0519685543312462,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.744221105527638e-05,
+      "loss": 1.4075,
+      "step": 609
+    },
+    {
+      "epoch": 0.05205388857481146,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.743718592964825e-05,
+      "loss": 1.3565,
+      "step": 610
+    },
+    {
+      "epoch": 0.052139222818376726,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.74321608040201e-05,
+      "loss": 1.5243,
+      "step": 611
+    },
+    {
+      "epoch": 0.05222455706194199,
+      "grad_norm": 0.298828125,
+      "learning_rate": 9.742713567839197e-05,
+      "loss": 1.3937,
+      "step": 612
+    },
+    {
+      "epoch": 0.052309891305507256,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.742211055276382e-05,
+      "loss": 1.3352,
+      "step": 613
+    },
+    {
+      "epoch": 0.05239522554907252,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.741708542713568e-05,
+      "loss": 1.3526,
+      "step": 614
+    },
+    {
+      "epoch": 0.052480559792637785,
+      "grad_norm": 0.29296875,
+      "learning_rate": 9.741206030150754e-05,
+      "loss": 1.3136,
+      "step": 615
+    },
+    {
+      "epoch": 0.05256589403620305,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.74070351758794e-05,
+      "loss": 1.3903,
+      "step": 616
+    },
+    {
+      "epoch": 0.05265122827976832,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.740201005025125e-05,
+      "loss": 1.388,
+      "step": 617
+    },
+    {
+      "epoch": 0.052736562523333586,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.739698492462313e-05,
+      "loss": 1.3319,
+      "step": 618
+    },
+    {
+      "epoch": 0.05282189676689885,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.739195979899498e-05,
+      "loss": 1.4627,
+      "step": 619
+    },
+    {
+      "epoch": 0.052907231010464115,
+      "grad_norm": 0.30078125,
+      "learning_rate": 9.738693467336684e-05,
+      "loss": 1.3808,
+      "step": 620
+    },
+    {
+      "epoch": 0.05299256525402938,
+      "grad_norm": 0.38671875,
+      "learning_rate": 9.73819095477387e-05,
+      "loss": 1.3673,
+      "step": 621
+    },
+    {
+      "epoch": 0.053077899497594644,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.737688442211056e-05,
+      "loss": 1.3133,
+      "step": 622
+    },
+    {
+      "epoch": 0.05316323374115991,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.737185929648241e-05,
+      "loss": 1.5619,
+      "step": 623
+    },
+    {
+      "epoch": 0.05324856798472517,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.736683417085427e-05,
+      "loss": 1.3774,
+      "step": 624
+    },
+    {
+      "epoch": 0.05333390222829044,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.736180904522614e-05,
+      "loss": 1.2794,
+      "step": 625
+    },
+    {
+      "epoch": 0.0534192364718557,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.7356783919598e-05,
+      "loss": 1.2726,
+      "step": 626
+    },
+    {
+      "epoch": 0.05350457071542097,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.735175879396985e-05,
+      "loss": 1.3671,
+      "step": 627
+    },
+    {
+      "epoch": 0.05358990495898623,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.734673366834172e-05,
+      "loss": 1.5113,
+      "step": 628
+    },
+    {
+      "epoch": 0.053675239202551496,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.734170854271357e-05,
+      "loss": 1.3327,
+      "step": 629
+    },
+    {
+      "epoch": 0.05376057344611676,
+      "grad_norm": 0.28125,
+      "learning_rate": 9.733668341708543e-05,
+      "loss": 1.3168,
+      "step": 630
+    },
+    {
+      "epoch": 0.053845907689682025,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.73316582914573e-05,
+      "loss": 1.4929,
+      "step": 631
+    },
+    {
+      "epoch": 0.05393124193324729,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.732663316582916e-05,
+      "loss": 1.1441,
+      "step": 632
+    },
+    {
+      "epoch": 0.054016576176812554,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.7321608040201e-05,
+      "loss": 1.4573,
+      "step": 633
+    },
+    {
+      "epoch": 0.05410191042037782,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.731658291457287e-05,
+      "loss": 1.5108,
+      "step": 634
+    },
+    {
+      "epoch": 0.054187244663943084,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.731155778894473e-05,
+      "loss": 1.3039,
+      "step": 635
+    },
+    {
+      "epoch": 0.05427257890750835,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.730653266331659e-05,
+      "loss": 1.2709,
+      "step": 636
+    },
+    {
+      "epoch": 0.05435791315107361,
+      "grad_norm": 0.361328125,
+      "learning_rate": 9.730150753768845e-05,
+      "loss": 1.3005,
+      "step": 637
+    },
+    {
+      "epoch": 0.05444324739463888,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.72964824120603e-05,
+      "loss": 1.2627,
+      "step": 638
+    },
+    {
+      "epoch": 0.05452858163820414,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.729145728643216e-05,
+      "loss": 1.4493,
+      "step": 639
+    },
+    {
+      "epoch": 0.054613915881769406,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.728643216080403e-05,
+      "loss": 1.2773,
+      "step": 640
+    },
+    {
+      "epoch": 0.05469925012533467,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.728140703517589e-05,
+      "loss": 1.2734,
+      "step": 641
+    },
+    {
+      "epoch": 0.054784584368899936,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.727638190954774e-05,
+      "loss": 1.3252,
+      "step": 642
+    },
+    {
+      "epoch": 0.0548699186124652,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.72713567839196e-05,
+      "loss": 1.5284,
+      "step": 643
+    },
+    {
+      "epoch": 0.054955252856030465,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.726633165829146e-05,
+      "loss": 1.2823,
+      "step": 644
+    },
+    {
+      "epoch": 0.05504058709959573,
+      "grad_norm": 0.41015625,
+      "learning_rate": 9.726130653266332e-05,
+      "loss": 1.5189,
+      "step": 645
+    },
+    {
+      "epoch": 0.055125921343160994,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.725628140703517e-05,
+      "loss": 1.4556,
+      "step": 646
+    },
+    {
+      "epoch": 0.05521125558672626,
+      "grad_norm": 0.375,
+      "learning_rate": 9.725125628140705e-05,
+      "loss": 1.2523,
+      "step": 647
+    },
+    {
+      "epoch": 0.05529658983029152,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.72462311557789e-05,
+      "loss": 1.2908,
+      "step": 648
+    },
+    {
+      "epoch": 0.05538192407385679,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.724120603015076e-05,
+      "loss": 1.4425,
+      "step": 649
+    },
+    {
+      "epoch": 0.05546725831742205,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.723618090452262e-05,
+      "loss": 1.4079,
+      "step": 650
+    },
+    {
+      "epoch": 0.05555259256098732,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.723115577889448e-05,
+      "loss": 1.3829,
+      "step": 651
+    },
+    {
+      "epoch": 0.05563792680455258,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.722613065326633e-05,
+      "loss": 1.427,
+      "step": 652
+    },
+    {
+      "epoch": 0.055723261048117846,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.72211055276382e-05,
+      "loss": 1.4874,
+      "step": 653
+    },
+    {
+      "epoch": 0.05580859529168311,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.721608040201005e-05,
+      "loss": 1.31,
+      "step": 654
+    },
+    {
+      "epoch": 0.055893929535248375,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.721105527638192e-05,
+      "loss": 1.3139,
+      "step": 655
+    },
+    {
+      "epoch": 0.05597926377881364,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.720603015075376e-05,
+      "loss": 1.2544,
+      "step": 656
+    },
+    {
+      "epoch": 0.056064598022378904,
+      "grad_norm": 0.375,
+      "learning_rate": 9.720100502512564e-05,
+      "loss": 1.3986,
+      "step": 657
+    },
+    {
+      "epoch": 0.05614993226594417,
+      "grad_norm": 0.38671875,
+      "learning_rate": 9.719597989949749e-05,
+      "loss": 1.4181,
+      "step": 658
+    },
+    {
+      "epoch": 0.056235266509509434,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.719095477386935e-05,
+      "loss": 2.0389,
+      "step": 659
+    },
+    {
+      "epoch": 0.0563206007530747,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.718592964824121e-05,
+      "loss": 1.5742,
+      "step": 660
+    },
+    {
+      "epoch": 0.05640593499663996,
+      "grad_norm": 0.294921875,
+      "learning_rate": 9.718090452261307e-05,
+      "loss": 1.3886,
+      "step": 661
+    },
+    {
+      "epoch": 0.05649126924020523,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.717587939698492e-05,
+      "loss": 1.3186,
+      "step": 662
+    },
+    {
+      "epoch": 0.05657660348377049,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.71708542713568e-05,
+      "loss": 1.4196,
+      "step": 663
+    },
+    {
+      "epoch": 0.05666193772733576,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.716582914572865e-05,
+      "loss": 1.2109,
+      "step": 664
+    },
+    {
+      "epoch": 0.05674727197090102,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.716080402010051e-05,
+      "loss": 1.3911,
+      "step": 665
+    },
+    {
+      "epoch": 0.056832606214466286,
+      "grad_norm": 0.29296875,
+      "learning_rate": 9.715577889447237e-05,
+      "loss": 1.3251,
+      "step": 666
+    },
+    {
+      "epoch": 0.05691794045803155,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.715075376884423e-05,
+      "loss": 1.4653,
+      "step": 667
+    },
+    {
+      "epoch": 0.057003274701596815,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.714572864321608e-05,
+      "loss": 1.369,
+      "step": 668
+    },
+    {
+      "epoch": 0.05708860894516208,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.714070351758794e-05,
+      "loss": 1.4326,
+      "step": 669
+    },
+    {
+      "epoch": 0.057173943188727344,
+      "grad_norm": 0.41015625,
+      "learning_rate": 9.71356783919598e-05,
+      "loss": 1.3603,
+      "step": 670
+    },
+    {
+      "epoch": 0.05725927743229261,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.713065326633167e-05,
+      "loss": 1.4364,
+      "step": 671
+    },
+    {
+      "epoch": 0.05734461167585787,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.712562814070352e-05,
+      "loss": 2.7462,
+      "step": 672
+    },
+    {
+      "epoch": 0.05742994591942314,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.712060301507538e-05,
+      "loss": 1.3715,
+      "step": 673
+    },
+    {
+      "epoch": 0.0575152801629884,
+      "grad_norm": 0.2890625,
+      "learning_rate": 9.711557788944724e-05,
+      "loss": 1.2796,
+      "step": 674
+    },
+    {
+      "epoch": 0.05760061440655367,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.71105527638191e-05,
+      "loss": 1.5047,
+      "step": 675
+    },
+    {
+      "epoch": 0.05768594865011893,
+      "grad_norm": 0.328125,
+      "learning_rate": 9.710552763819096e-05,
+      "loss": 1.2029,
+      "step": 676
+    },
+    {
+      "epoch": 0.057771282893684196,
+      "grad_norm": 0.462890625,
+      "learning_rate": 9.710050251256281e-05,
+      "loss": 1.5285,
+      "step": 677
+    },
+    {
+      "epoch": 0.05785661713724946,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.709547738693468e-05,
+      "loss": 1.4128,
+      "step": 678
+    },
+    {
+      "epoch": 0.057941951380814725,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.709045226130654e-05,
+      "loss": 1.4114,
+      "step": 679
+    },
+    {
+      "epoch": 0.05802728562437999,
+      "grad_norm": 0.28515625,
+      "learning_rate": 9.70854271356784e-05,
+      "loss": 1.3152,
+      "step": 680
+    },
+    {
+      "epoch": 0.058112619867945255,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.708040201005025e-05,
+      "loss": 1.3538,
+      "step": 681
+    },
+    {
+      "epoch": 0.058197954111510526,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.707537688442212e-05,
+      "loss": 1.3003,
+      "step": 682
+    },
+    {
+      "epoch": 0.05828328835507579,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.707035175879397e-05,
+      "loss": 1.3926,
+      "step": 683
+    },
+    {
+      "epoch": 0.058368622598641055,
+      "grad_norm": 0.310546875,
+      "learning_rate": 9.706532663316583e-05,
+      "loss": 1.3557,
+      "step": 684
+    },
+    {
+      "epoch": 0.05845395684220632,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.706030150753768e-05,
+      "loss": 1.2721,
+      "step": 685
+    },
+    {
+      "epoch": 0.058539291085771585,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.705527638190956e-05,
+      "loss": 1.3897,
+      "step": 686
+    },
+    {
+      "epoch": 0.05862462532933685,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.70502512562814e-05,
+      "loss": 1.286,
+      "step": 687
+    },
+    {
+      "epoch": 0.058709959572902114,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.704522613065327e-05,
+      "loss": 1.4595,
+      "step": 688
+    },
+    {
+      "epoch": 0.05879529381646738,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.704020100502513e-05,
+      "loss": 1.435,
+      "step": 689
+    },
+    {
+      "epoch": 0.05888062806003264,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.703517587939699e-05,
+      "loss": 1.3686,
+      "step": 690
+    },
+    {
+      "epoch": 0.05896596230359791,
+      "grad_norm": 0.33984375,
+      "learning_rate": 9.703015075376884e-05,
+      "loss": 1.3749,
+      "step": 691
+    },
+    {
+      "epoch": 0.05905129654716317,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.702512562814072e-05,
+      "loss": 1.1963,
+      "step": 692
+    },
+    {
+      "epoch": 0.05913663079072844,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.702010050251257e-05,
+      "loss": 1.3632,
+      "step": 693
+    },
+    {
+      "epoch": 0.0592219650342937,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.701507537688443e-05,
+      "loss": 1.5149,
+      "step": 694
+    },
+    {
+      "epoch": 0.059307299277858966,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.701005025125629e-05,
+      "loss": 1.3232,
+      "step": 695
+    },
+    {
+      "epoch": 0.05939263352142423,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.700502512562815e-05,
+      "loss": 1.2522,
+      "step": 696
+    },
+    {
+      "epoch": 0.059477967764989495,
+      "grad_norm": 0.28515625,
+      "learning_rate": 9.7e-05,
+      "loss": 1.4551,
+      "step": 697
+    },
+    {
+      "epoch": 0.05956330200855476,
+      "grad_norm": 0.310546875,
+      "learning_rate": 9.699497487437188e-05,
+      "loss": 1.0857,
+      "step": 698
+    },
+    {
+      "epoch": 0.059648636252120024,
+      "grad_norm": 0.28125,
+      "learning_rate": 9.698994974874372e-05,
+      "loss": 1.309,
+      "step": 699
+    },
+    {
+      "epoch": 0.05973397049568529,
+      "grad_norm": 0.369140625,
+      "learning_rate": 9.698492462311559e-05,
+      "loss": 1.4626,
+      "step": 700
+    },
+    {
+      "epoch": 0.05981930473925055,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.697989949748743e-05,
+      "loss": 1.4734,
+      "step": 701
+    },
+    {
+      "epoch": 0.05990463898281582,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.697487437185931e-05,
+      "loss": 1.3883,
+      "step": 702
+    },
+    {
+      "epoch": 0.05998997322638108,
+      "grad_norm": 0.310546875,
+      "learning_rate": 9.696984924623116e-05,
+      "loss": 1.4671,
+      "step": 703
+    },
+    {
+      "epoch": 0.06007530746994635,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.696482412060302e-05,
+      "loss": 1.3927,
+      "step": 704
+    },
+    {
+      "epoch": 0.06016064171351161,
+      "grad_norm": 0.296875,
+      "learning_rate": 9.695979899497488e-05,
+      "loss": 1.2563,
+      "step": 705
+    },
+    {
+      "epoch": 0.060245975957076876,
+      "grad_norm": 0.30078125,
+      "learning_rate": 9.695477386934674e-05,
+      "loss": 1.2925,
+      "step": 706
+    },
+    {
+      "epoch": 0.06033131020064214,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.694974874371859e-05,
+      "loss": 1.3398,
+      "step": 707
+    },
+    {
+      "epoch": 0.060416644444207405,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.694472361809045e-05,
+      "loss": 1.3857,
+      "step": 708
+    },
+    {
+      "epoch": 0.06050197868777267,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.693969849246232e-05,
+      "loss": 1.4813,
+      "step": 709
+    },
+    {
+      "epoch": 0.060587312931337935,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.693467336683418e-05,
+      "loss": 1.4029,
+      "step": 710
+    },
+    {
+      "epoch": 0.0606726471749032,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.692964824120604e-05,
+      "loss": 1.46,
+      "step": 711
+    },
+    {
+      "epoch": 0.060757981418468464,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.692462311557789e-05,
+      "loss": 1.3441,
+      "step": 712
+    },
+    {
+      "epoch": 0.06084331566203373,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.691959798994975e-05,
+      "loss": 1.4921,
+      "step": 713
+    },
+    {
+      "epoch": 0.06092864990559899,
+      "grad_norm": 0.30078125,
+      "learning_rate": 9.691457286432161e-05,
+      "loss": 1.2976,
+      "step": 714
+    },
+    {
+      "epoch": 0.06101398414916426,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.690954773869348e-05,
+      "loss": 1.3878,
+      "step": 715
+    },
+    {
+      "epoch": 0.06109931839272952,
+      "grad_norm": 0.365234375,
+      "learning_rate": 9.690452261306532e-05,
+      "loss": 1.4598,
+      "step": 716
+    },
+    {
+      "epoch": 0.06118465263629479,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.689949748743719e-05,
+      "loss": 1.3242,
+      "step": 717
+    },
+    {
+      "epoch": 0.06126998687986005,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.689447236180905e-05,
+      "loss": 1.418,
+      "step": 718
+    },
+    {
+      "epoch": 0.061355321123425316,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.688944723618091e-05,
+      "loss": 1.5938,
+      "step": 719
+    },
+    {
+      "epoch": 0.06144065536699058,
+      "grad_norm": 0.2734375,
+      "learning_rate": 9.688442211055276e-05,
+      "loss": 1.1907,
+      "step": 720
+    },
+    {
+      "epoch": 0.061525989610555845,
+      "grad_norm": 0.294921875,
+      "learning_rate": 9.687939698492463e-05,
+      "loss": 1.2858,
+      "step": 721
+    },
+    {
+      "epoch": 0.06161132385412111,
+      "grad_norm": 0.298828125,
+      "learning_rate": 9.687437185929648e-05,
+      "loss": 1.2152,
+      "step": 722
+    },
+    {
+      "epoch": 0.061696658097686374,
+      "grad_norm": 0.302734375,
+      "learning_rate": 9.686934673366834e-05,
+      "loss": 1.3035,
+      "step": 723
+    },
+    {
+      "epoch": 0.06178199234125164,
+      "grad_norm": 0.29296875,
+      "learning_rate": 9.686432160804021e-05,
+      "loss": 1.3546,
+      "step": 724
+    },
+    {
+      "epoch": 0.061867326584816904,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.685929648241207e-05,
+      "loss": 1.6224,
+      "step": 725
+    },
+    {
+      "epoch": 0.06195266082838217,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.685427135678392e-05,
+      "loss": 1.1704,
+      "step": 726
+    },
+    {
+      "epoch": 0.06203799507194743,
+      "grad_norm": 0.296875,
+      "learning_rate": 9.684924623115579e-05,
+      "loss": 1.2085,
+      "step": 727
+    },
+    {
+      "epoch": 0.0621233293155127,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.684422110552764e-05,
+      "loss": 1.2988,
+      "step": 728
+    },
+    {
+      "epoch": 0.06220866355907796,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.68391959798995e-05,
+      "loss": 1.2337,
+      "step": 729
+    },
+    {
+      "epoch": 0.062293997802643226,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.683417085427137e-05,
+      "loss": 1.2785,
+      "step": 730
+    },
+    {
+      "epoch": 0.06237933204620849,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.682914572864323e-05,
+      "loss": 1.4845,
+      "step": 731
+    },
+    {
+      "epoch": 0.062464666289773756,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.682412060301508e-05,
+      "loss": 1.2532,
+      "step": 732
+    },
+    {
+      "epoch": 0.06255000053333902,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.681909547738694e-05,
+      "loss": 1.4766,
+      "step": 733
+    },
+    {
+      "epoch": 0.06263533477690429,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.68140703517588e-05,
+      "loss": 1.2843,
+      "step": 734
+    },
+    {
+      "epoch": 0.06272066902046955,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.680904522613066e-05,
+      "loss": 1.5138,
+      "step": 735
+    },
+    {
+      "epoch": 0.06280600326403482,
+      "grad_norm": 0.328125,
+      "learning_rate": 9.680402010050251e-05,
+      "loss": 1.4527,
+      "step": 736
+    },
+    {
+      "epoch": 0.06289133750760008,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.679899497487439e-05,
+      "loss": 1.3572,
+      "step": 737
+    },
+    {
+      "epoch": 0.06297667175116535,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.679396984924623e-05,
+      "loss": 1.283,
+      "step": 738
+    },
+    {
+      "epoch": 0.06306200599473061,
+      "grad_norm": 0.3203125,
+      "learning_rate": 9.67889447236181e-05,
+      "loss": 1.4313,
+      "step": 739
+    },
+    {
+      "epoch": 0.06314734023829588,
+      "grad_norm": 0.38671875,
+      "learning_rate": 9.678391959798996e-05,
+      "loss": 1.504,
+      "step": 740
+    },
+    {
+      "epoch": 0.06323267448186114,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.677889447236182e-05,
+      "loss": 1.2715,
+      "step": 741
+    },
+    {
+      "epoch": 0.06331800872542641,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.677386934673367e-05,
+      "loss": 1.2285,
+      "step": 742
+    },
+    {
+      "epoch": 0.06340334296899167,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.676884422110553e-05,
+      "loss": 1.3994,
+      "step": 743
+    },
+    {
+      "epoch": 0.06348867721255694,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.67638190954774e-05,
+      "loss": 1.5629,
+      "step": 744
+    },
+    {
+      "epoch": 0.0635740114561222,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.675879396984924e-05,
+      "loss": 1.424,
+      "step": 745
+    },
+    {
+      "epoch": 0.06365934569968747,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.675376884422112e-05,
+      "loss": 1.2753,
+      "step": 746
+    },
+    {
+      "epoch": 0.06374467994325272,
+      "grad_norm": 0.36328125,
+      "learning_rate": 9.674874371859297e-05,
+      "loss": 1.3822,
+      "step": 747
+    },
+    {
+      "epoch": 0.063830014186818,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.674371859296483e-05,
+      "loss": 1.3389,
+      "step": 748
+    },
+    {
+      "epoch": 0.06391534843038325,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.673869346733668e-05,
+      "loss": 1.3684,
+      "step": 749
+    },
+    {
+      "epoch": 0.06400068267394853,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.673366834170855e-05,
+      "loss": 1.3797,
+      "step": 750
+    },
+    {
+      "epoch": 0.06408601691751378,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.67286432160804e-05,
+      "loss": 1.2652,
+      "step": 751
+    },
+    {
+      "epoch": 0.06417135116107905,
+      "grad_norm": 0.373046875,
+      "learning_rate": 9.672361809045226e-05,
+      "loss": 1.5532,
+      "step": 752
+    },
+    {
+      "epoch": 0.06425668540464431,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.671859296482412e-05,
+      "loss": 1.351,
+      "step": 753
+    },
+    {
+      "epoch": 0.06434201964820958,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.671356783919599e-05,
+      "loss": 1.4864,
+      "step": 754
+    },
+    {
+      "epoch": 0.06442735389177484,
+      "grad_norm": 0.3671875,
+      "learning_rate": 9.670854271356784e-05,
+      "loss": 1.3885,
+      "step": 755
+    },
+    {
+      "epoch": 0.06451268813534011,
+      "grad_norm": 0.3203125,
+      "learning_rate": 9.670351758793971e-05,
+      "loss": 1.3694,
+      "step": 756
+    },
+    {
+      "epoch": 0.06459802237890537,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.669849246231156e-05,
+      "loss": 1.3275,
+      "step": 757
+    },
+    {
+      "epoch": 0.06468335662247064,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.669346733668342e-05,
+      "loss": 1.3146,
+      "step": 758
+    },
+    {
+      "epoch": 0.0647686908660359,
+      "grad_norm": 0.4921875,
+      "learning_rate": 9.668844221105528e-05,
+      "loss": 1.4108,
+      "step": 759
+    },
+    {
+      "epoch": 0.06485402510960117,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.668341708542715e-05,
+      "loss": 1.3766,
+      "step": 760
+    },
+    {
+      "epoch": 0.06493935935316643,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.6678391959799e-05,
+      "loss": 1.3579,
+      "step": 761
+    },
+    {
+      "epoch": 0.0650246935967317,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.667336683417087e-05,
+      "loss": 1.358,
+      "step": 762
+    },
+    {
+      "epoch": 0.06511002784029696,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.666834170854272e-05,
+      "loss": 1.3018,
+      "step": 763
+    },
+    {
+      "epoch": 0.06519536208386223,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.666331658291458e-05,
+      "loss": 1.4102,
+      "step": 764
+    },
+    {
+      "epoch": 0.06528069632742749,
+      "grad_norm": 0.30078125,
+      "learning_rate": 9.665829145728643e-05,
+      "loss": 1.2921,
+      "step": 765
+    },
+    {
+      "epoch": 0.06536603057099276,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.66532663316583e-05,
+      "loss": 1.3826,
+      "step": 766
+    },
+    {
+      "epoch": 0.06545136481455802,
+      "grad_norm": 0.3203125,
+      "learning_rate": 9.664824120603015e-05,
+      "loss": 1.2876,
+      "step": 767
+    },
+    {
+      "epoch": 0.06553669905812329,
+      "grad_norm": 0.33984375,
+      "learning_rate": 9.664321608040201e-05,
+      "loss": 1.42,
+      "step": 768
+    },
+    {
+      "epoch": 0.06562203330168855,
+      "grad_norm": 0.298828125,
+      "learning_rate": 9.663819095477388e-05,
+      "loss": 1.342,
+      "step": 769
+    },
+    {
+      "epoch": 0.06570736754525382,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.663316582914574e-05,
+      "loss": 1.3334,
+      "step": 770
+    },
+    {
+      "epoch": 0.06579270178881907,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.662814070351759e-05,
+      "loss": 1.3406,
+      "step": 771
+    },
+    {
+      "epoch": 0.06587803603238435,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.662311557788946e-05,
+      "loss": 1.4716,
+      "step": 772
+    },
+    {
+      "epoch": 0.0659633702759496,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.661809045226131e-05,
+      "loss": 1.3348,
+      "step": 773
+    },
+    {
+      "epoch": 0.06604870451951488,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.661306532663317e-05,
+      "loss": 1.3903,
+      "step": 774
+    },
+    {
+      "epoch": 0.06613403876308013,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.660804020100504e-05,
+      "loss": 1.3265,
+      "step": 775
+    },
+    {
+      "epoch": 0.0662193730066454,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.66030150753769e-05,
+      "loss": 1.2455,
+      "step": 776
+    },
+    {
+      "epoch": 0.06630470725021066,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.659798994974875e-05,
+      "loss": 1.3662,
+      "step": 777
+    },
+    {
+      "epoch": 0.06639004149377593,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.659296482412061e-05,
+      "loss": 1.4612,
+      "step": 778
+    },
+    {
+      "epoch": 0.0664753757373412,
+      "grad_norm": 0.298828125,
+      "learning_rate": 9.658793969849247e-05,
+      "loss": 1.3211,
+      "step": 779
+    },
+    {
+      "epoch": 0.06656070998090646,
+      "grad_norm": 0.310546875,
+      "learning_rate": 9.658291457286432e-05,
+      "loss": 1.3174,
+      "step": 780
+    },
+    {
+      "epoch": 0.06664604422447173,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.657788944723618e-05,
+      "loss": 1.387,
+      "step": 781
+    },
+    {
+      "epoch": 0.06673137846803699,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.657286432160804e-05,
+      "loss": 1.3186,
+      "step": 782
+    },
+    {
+      "epoch": 0.06681671271160226,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.65678391959799e-05,
+      "loss": 1.449,
+      "step": 783
+    },
+    {
+      "epoch": 0.06690204695516752,
+      "grad_norm": 0.3828125,
+      "learning_rate": 9.656281407035175e-05,
+      "loss": 1.5026,
+      "step": 784
+    },
+    {
+      "epoch": 0.06698738119873279,
+      "grad_norm": 0.28125,
+      "learning_rate": 9.655778894472363e-05,
+      "loss": 1.2671,
+      "step": 785
+    },
+    {
+      "epoch": 0.06707271544229805,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.655276381909548e-05,
+      "loss": 1.3618,
+      "step": 786
+    },
+    {
+      "epoch": 0.06715804968586332,
+      "grad_norm": 0.271484375,
+      "learning_rate": 9.654773869346734e-05,
+      "loss": 1.1871,
+      "step": 787
+    },
+    {
+      "epoch": 0.06724338392942858,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.65427135678392e-05,
+      "loss": 1.2311,
+      "step": 788
+    },
+    {
+      "epoch": 0.06732871817299385,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.653768844221106e-05,
+      "loss": 1.3273,
+      "step": 789
+    },
+    {
+      "epoch": 0.06741405241655911,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.653266331658291e-05,
+      "loss": 1.3785,
+      "step": 790
+    },
+    {
+      "epoch": 0.06749938666012438,
+      "grad_norm": 0.29296875,
+      "learning_rate": 9.652763819095479e-05,
+      "loss": 1.208,
+      "step": 791
+    },
+    {
+      "epoch": 0.06758472090368964,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.652261306532664e-05,
+      "loss": 1.2334,
+      "step": 792
+    },
+    {
+      "epoch": 0.06767005514725491,
+      "grad_norm": 0.302734375,
+      "learning_rate": 9.65175879396985e-05,
+      "loss": 1.3317,
+      "step": 793
+    },
+    {
+      "epoch": 0.06775538939082017,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.651256281407036e-05,
+      "loss": 1.0677,
+      "step": 794
+    },
+    {
+      "epoch": 0.06784072363438544,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.650753768844222e-05,
+      "loss": 1.3388,
+      "step": 795
+    },
+    {
+      "epoch": 0.0679260578779507,
+      "grad_norm": 0.328125,
+      "learning_rate": 9.650251256281407e-05,
+      "loss": 1.2782,
+      "step": 796
+    },
+    {
+      "epoch": 0.06801139212151597,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.649748743718593e-05,
+      "loss": 1.3381,
+      "step": 797
+    },
+    {
+      "epoch": 0.06809672636508123,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.64924623115578e-05,
+      "loss": 1.3785,
+      "step": 798
+    },
+    {
+      "epoch": 0.0681820606086465,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.648743718592966e-05,
+      "loss": 1.2829,
+      "step": 799
+    },
+    {
+      "epoch": 0.06826739485221175,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.64824120603015e-05,
+      "loss": 1.3086,
+      "step": 800
+    },
+    {
+      "epoch": 0.06835272909577703,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.647738693467338e-05,
+      "loss": 1.5021,
+      "step": 801
+    },
+    {
+      "epoch": 0.06843806333934228,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.647236180904523e-05,
+      "loss": 1.2855,
+      "step": 802
+    },
+    {
+      "epoch": 0.06852339758290756,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.646733668341709e-05,
+      "loss": 1.4065,
+      "step": 803
+    },
+    {
+      "epoch": 0.06860873182647281,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.646231155778895e-05,
+      "loss": 1.3367,
+      "step": 804
+    },
+    {
+      "epoch": 0.06869406607003808,
+      "grad_norm": 0.310546875,
+      "learning_rate": 9.645728643216081e-05,
+      "loss": 1.3224,
+      "step": 805
+    },
+    {
+      "epoch": 0.06877940031360334,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.645226130653266e-05,
+      "loss": 1.5007,
+      "step": 806
+    },
+    {
+      "epoch": 0.06886473455716861,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.644723618090454e-05,
+      "loss": 1.2908,
+      "step": 807
+    },
+    {
+      "epoch": 0.06895006880073387,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.644221105527639e-05,
+      "loss": 1.3805,
+      "step": 808
+    },
+    {
+      "epoch": 0.06903540304429914,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.643718592964825e-05,
+      "loss": 1.4245,
+      "step": 809
+    },
+    {
+      "epoch": 0.0691207372878644,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.64321608040201e-05,
+      "loss": 1.4083,
+      "step": 810
+    },
+    {
+      "epoch": 0.06920607153142967,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.642713567839196e-05,
+      "loss": 1.4118,
+      "step": 811
+    },
+    {
+      "epoch": 0.06929140577499493,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.642211055276382e-05,
+      "loss": 1.4157,
+      "step": 812
+    },
+    {
+      "epoch": 0.0693767400185602,
+      "grad_norm": 0.291015625,
+      "learning_rate": 9.641708542713568e-05,
+      "loss": 1.2907,
+      "step": 813
+    },
+    {
+      "epoch": 0.06946207426212546,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.641206030150755e-05,
+      "loss": 1.3037,
+      "step": 814
+    },
+    {
+      "epoch": 0.06954740850569073,
+      "grad_norm": 0.375,
+      "learning_rate": 9.64070351758794e-05,
+      "loss": 1.569,
+      "step": 815
+    },
+    {
+      "epoch": 0.06963274274925599,
+      "grad_norm": 0.294921875,
+      "learning_rate": 9.640201005025126e-05,
+      "loss": 1.3873,
+      "step": 816
+    },
+    {
+      "epoch": 0.06971807699282126,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.639698492462312e-05,
+      "loss": 1.3566,
+      "step": 817
+    },
+    {
+      "epoch": 0.06980341123638652,
+      "grad_norm": 0.3203125,
+      "learning_rate": 9.639195979899498e-05,
+      "loss": 1.2744,
+      "step": 818
+    },
+    {
+      "epoch": 0.06988874547995179,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.638693467336683e-05,
+      "loss": 1.4438,
+      "step": 819
+    },
+    {
+      "epoch": 0.06997407972351705,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.63819095477387e-05,
+      "loss": 1.4021,
+      "step": 820
+    },
+    {
+      "epoch": 0.07005941396708232,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.637688442211055e-05,
+      "loss": 1.2743,
+      "step": 821
+    },
+    {
+      "epoch": 0.07014474821064758,
+      "grad_norm": 0.294921875,
+      "learning_rate": 9.637185929648242e-05,
+      "loss": 1.2944,
+      "step": 822
+    },
+    {
+      "epoch": 0.07023008245421285,
+      "grad_norm": 0.30078125,
+      "learning_rate": 9.636683417085428e-05,
+      "loss": 1.2728,
+      "step": 823
+    },
+    {
+      "epoch": 0.0703154166977781,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.636180904522614e-05,
+      "loss": 1.3109,
+      "step": 824
+    },
+    {
+      "epoch": 0.07040075094134338,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.635678391959799e-05,
+      "loss": 1.3518,
+      "step": 825
+    },
+    {
+      "epoch": 0.07048608518490863,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.635175879396985e-05,
+      "loss": 1.3864,
+      "step": 826
+    },
+    {
+      "epoch": 0.0705714194284739,
+      "grad_norm": 0.28125,
+      "learning_rate": 9.634673366834171e-05,
+      "loss": 1.3201,
+      "step": 827
+    },
+    {
+      "epoch": 0.07065675367203916,
+      "grad_norm": 0.28515625,
+      "learning_rate": 9.634170854271357e-05,
+      "loss": 1.3849,
+      "step": 828
+    },
+    {
+      "epoch": 0.07074208791560443,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.633668341708542e-05,
+      "loss": 1.4096,
+      "step": 829
+    },
+    {
+      "epoch": 0.07082742215916969,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.63316582914573e-05,
+      "loss": 1.3157,
+      "step": 830
+    },
+    {
+      "epoch": 0.07091275640273496,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.632663316582915e-05,
+      "loss": 1.1486,
+      "step": 831
+    },
+    {
+      "epoch": 0.07099809064630022,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.632160804020101e-05,
+      "loss": 1.2551,
+      "step": 832
+    },
+    {
+      "epoch": 0.07108342488986549,
+      "grad_norm": 0.296875,
+      "learning_rate": 9.631658291457287e-05,
+      "loss": 1.3846,
+      "step": 833
+    },
+    {
+      "epoch": 0.07116875913343075,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.631155778894473e-05,
+      "loss": 1.4594,
+      "step": 834
+    },
+    {
+      "epoch": 0.07125409337699602,
+      "grad_norm": 0.302734375,
+      "learning_rate": 9.630653266331658e-05,
+      "loss": 1.2315,
+      "step": 835
+    },
+    {
+      "epoch": 0.07133942762056128,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.630150753768846e-05,
+      "loss": 1.3929,
+      "step": 836
+    },
+    {
+      "epoch": 0.07142476186412655,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.62964824120603e-05,
+      "loss": 1.4372,
+      "step": 837
+    },
+    {
+      "epoch": 0.07151009610769181,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.629145728643217e-05,
+      "loss": 1.5075,
+      "step": 838
+    },
+    {
+      "epoch": 0.07159543035125708,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.628643216080403e-05,
+      "loss": 1.3296,
+      "step": 839
+    },
+    {
+      "epoch": 0.07168076459482234,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.628140703517589e-05,
+      "loss": 1.302,
+      "step": 840
+    },
+    {
+      "epoch": 0.07176609883838761,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.627638190954774e-05,
+      "loss": 1.5026,
+      "step": 841
+    },
+    {
+      "epoch": 0.07185143308195287,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.62713567839196e-05,
+      "loss": 1.3536,
+      "step": 842
+    },
+    {
+      "epoch": 0.07193676732551814,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.626633165829146e-05,
+      "loss": 1.4149,
+      "step": 843
+    },
+    {
+      "epoch": 0.07202210156908341,
+      "grad_norm": 0.29296875,
+      "learning_rate": 9.626130653266333e-05,
+      "loss": 1.3225,
+      "step": 844
+    },
+    {
+      "epoch": 0.07210743581264867,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.625628140703517e-05,
+      "loss": 1.3959,
+      "step": 845
+    },
+    {
+      "epoch": 0.07219277005621394,
+      "grad_norm": 0.3203125,
+      "learning_rate": 9.625125628140704e-05,
+      "loss": 1.2814,
+      "step": 846
+    },
+    {
+      "epoch": 0.0722781042997792,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.62462311557789e-05,
+      "loss": 1.2217,
+      "step": 847
+    },
+    {
+      "epoch": 0.07236343854334447,
+      "grad_norm": 0.287109375,
+      "learning_rate": 9.624120603015076e-05,
+      "loss": 1.4332,
+      "step": 848
+    },
+    {
+      "epoch": 0.07244877278690973,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.623618090452262e-05,
+      "loss": 1.3367,
+      "step": 849
+    },
+    {
+      "epoch": 0.072534107030475,
+      "grad_norm": 0.30078125,
+      "learning_rate": 9.623115577889447e-05,
+      "loss": 1.3362,
+      "step": 850
+    },
+    {
+      "epoch": 0.07261944127404026,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.622613065326633e-05,
+      "loss": 1.4656,
+      "step": 851
+    },
+    {
+      "epoch": 0.07270477551760553,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.62211055276382e-05,
+      "loss": 1.5471,
+      "step": 852
+    },
+    {
+      "epoch": 0.07279010976117078,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.621608040201006e-05,
+      "loss": 1.3255,
+      "step": 853
+    },
+    {
+      "epoch": 0.07287544400473606,
+      "grad_norm": 0.296875,
+      "learning_rate": 9.62110552763819e-05,
+      "loss": 1.2592,
+      "step": 854
+    },
+    {
+      "epoch": 0.07296077824830131,
+      "grad_norm": 0.296875,
+      "learning_rate": 9.620603015075378e-05,
+      "loss": 1.2034,
+      "step": 855
+    },
+    {
+      "epoch": 0.07304611249186659,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.620100502512563e-05,
+      "loss": 1.4065,
+      "step": 856
+    },
+    {
+      "epoch": 0.07313144673543184,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.619597989949749e-05,
+      "loss": 1.3095,
+      "step": 857
+    },
+    {
+      "epoch": 0.07321678097899711,
+      "grad_norm": 0.400390625,
+      "learning_rate": 9.619095477386934e-05,
+      "loss": 1.4808,
+      "step": 858
+    },
+    {
+      "epoch": 0.07330211522256237,
+      "grad_norm": 0.302734375,
+      "learning_rate": 9.618592964824122e-05,
+      "loss": 1.3797,
+      "step": 859
+    },
+    {
+      "epoch": 0.07338744946612764,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.618090452261306e-05,
+      "loss": 1.4729,
+      "step": 860
+    },
+    {
+      "epoch": 0.0734727837096929,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.617587939698493e-05,
+      "loss": 1.401,
+      "step": 861
+    },
+    {
+      "epoch": 0.07355811795325817,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.617085427135679e-05,
+      "loss": 1.2793,
+      "step": 862
+    },
+    {
+      "epoch": 0.07364345219682343,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.616582914572865e-05,
+      "loss": 1.3485,
+      "step": 863
+    },
+    {
+      "epoch": 0.0737287864403887,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.61608040201005e-05,
+      "loss": 1.4182,
+      "step": 864
+    },
+    {
+      "epoch": 0.07381412068395396,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.615577889447237e-05,
+      "loss": 1.3818,
+      "step": 865
+    },
+    {
+      "epoch": 0.07389945492751923,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.615075376884422e-05,
+      "loss": 1.3538,
+      "step": 866
+    },
+    {
+      "epoch": 0.07398478917108449,
+      "grad_norm": 0.361328125,
+      "learning_rate": 9.614572864321608e-05,
+      "loss": 1.3324,
+      "step": 867
+    },
+    {
+      "epoch": 0.07407012341464976,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.614070351758795e-05,
+      "loss": 1.3011,
+      "step": 868
+    },
+    {
+      "epoch": 0.07415545765821502,
+      "grad_norm": 0.29296875,
+      "learning_rate": 9.613567839195981e-05,
+      "loss": 1.4267,
+      "step": 869
+    },
+    {
+      "epoch": 0.07424079190178029,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.613065326633166e-05,
+      "loss": 1.4434,
+      "step": 870
+    },
+    {
+      "epoch": 0.07432612614534555,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.612562814070353e-05,
+      "loss": 1.3762,
+      "step": 871
+    },
+    {
+      "epoch": 0.07441146038891082,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.612060301507538e-05,
+      "loss": 1.2908,
+      "step": 872
+    },
+    {
+      "epoch": 0.07449679463247608,
+      "grad_norm": 0.294921875,
+      "learning_rate": 9.611557788944724e-05,
+      "loss": 1.1824,
+      "step": 873
+    },
+    {
+      "epoch": 0.07458212887604135,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.611055276381909e-05,
+      "loss": 1.4014,
+      "step": 874
+    },
+    {
+      "epoch": 0.0746674631196066,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.610552763819097e-05,
+      "loss": 1.3461,
+      "step": 875
+    },
+    {
+      "epoch": 0.07475279736317188,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.610050251256282e-05,
+      "loss": 1.4505,
+      "step": 876
+    },
+    {
+      "epoch": 0.07483813160673713,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.609547738693468e-05,
+      "loss": 1.3212,
+      "step": 877
+    },
+    {
+      "epoch": 0.0749234658503024,
+      "grad_norm": 0.2890625,
+      "learning_rate": 9.609045226130654e-05,
+      "loss": 1.2537,
+      "step": 878
+    },
+    {
+      "epoch": 0.07500880009386766,
+      "grad_norm": 0.296875,
+      "learning_rate": 9.60854271356784e-05,
+      "loss": 1.3349,
+      "step": 879
+    },
+    {
+      "epoch": 0.07509413433743294,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.608040201005025e-05,
+      "loss": 1.2169,
+      "step": 880
+    },
+    {
+      "epoch": 0.0751794685809982,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.607537688442211e-05,
+      "loss": 1.2766,
+      "step": 881
+    },
+    {
+      "epoch": 0.07526480282456346,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.607035175879397e-05,
+      "loss": 1.4459,
+      "step": 882
+    },
+    {
+      "epoch": 0.07535013706812872,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.606532663316584e-05,
+      "loss": 1.2888,
+      "step": 883
+    },
+    {
+      "epoch": 0.075435471311694,
+      "grad_norm": 0.27734375,
+      "learning_rate": 9.60603015075377e-05,
+      "loss": 1.3252,
+      "step": 884
+    },
+    {
+      "epoch": 0.07552080555525925,
+      "grad_norm": 0.33984375,
+      "learning_rate": 9.605527638190955e-05,
+      "loss": 1.417,
+      "step": 885
+    },
+    {
+      "epoch": 0.07560613979882452,
+      "grad_norm": 0.3203125,
+      "learning_rate": 9.605025125628141e-05,
+      "loss": 1.256,
+      "step": 886
+    },
+    {
+      "epoch": 0.07569147404238978,
+      "grad_norm": 0.310546875,
+      "learning_rate": 9.604522613065327e-05,
+      "loss": 1.7487,
+      "step": 887
+    },
+    {
+      "epoch": 0.07577680828595505,
+      "grad_norm": 0.291015625,
+      "learning_rate": 9.604020100502513e-05,
+      "loss": 1.3971,
+      "step": 888
+    },
+    {
+      "epoch": 0.07586214252952031,
+      "grad_norm": 0.5859375,
+      "learning_rate": 9.603517587939698e-05,
+      "loss": 1.5179,
+      "step": 889
+    },
+    {
+      "epoch": 0.07594747677308558,
+      "grad_norm": 0.310546875,
+      "learning_rate": 9.603015075376884e-05,
+      "loss": 1.3618,
+      "step": 890
+    },
+    {
+      "epoch": 0.07603281101665084,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.60251256281407e-05,
+      "loss": 1.4516,
+      "step": 891
+    },
+    {
+      "epoch": 0.07611814526021611,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.602010050251257e-05,
+      "loss": 1.3767,
+      "step": 892
+    },
+    {
+      "epoch": 0.07620347950378137,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.601507537688442e-05,
+      "loss": 1.3326,
+      "step": 893
+    },
+    {
+      "epoch": 0.07628881374734664,
+      "grad_norm": 0.3203125,
+      "learning_rate": 9.601005025125629e-05,
+      "loss": 1.3592,
+      "step": 894
+    },
+    {
+      "epoch": 0.0763741479909119,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.600502512562814e-05,
+      "loss": 1.4043,
+      "step": 895
+    },
+    {
+      "epoch": 0.07645948223447717,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.6e-05,
+      "loss": 1.4969,
+      "step": 896
+    },
+    {
+      "epoch": 0.07654481647804243,
+      "grad_norm": 0.302734375,
+      "learning_rate": 9.599497487437186e-05,
+      "loss": 1.4438,
+      "step": 897
+    },
+    {
+      "epoch": 0.0766301507216077,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.598994974874373e-05,
+      "loss": 1.3689,
+      "step": 898
+    },
+    {
+      "epoch": 0.07671548496517296,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.598492462311557e-05,
+      "loss": 1.3063,
+      "step": 899
+    },
+    {
+      "epoch": 0.07680081920873823,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.597989949748745e-05,
+      "loss": 1.3885,
+      "step": 900
+    },
+    {
+      "epoch": 0.07688615345230349,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.59748743718593e-05,
+      "loss": 1.2972,
+      "step": 901
+    },
+    {
+      "epoch": 0.07697148769586876,
+      "grad_norm": 0.279296875,
+      "learning_rate": 9.596984924623116e-05,
+      "loss": 1.3185,
+      "step": 902
+    },
+    {
+      "epoch": 0.07705682193943401,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.596482412060302e-05,
+      "loss": 1.6207,
+      "step": 903
+    },
+    {
+      "epoch": 0.07714215618299929,
+      "grad_norm": 0.30078125,
+      "learning_rate": 9.595979899497489e-05,
+      "loss": 1.3045,
+      "step": 904
+    },
+    {
+      "epoch": 0.07722749042656454,
+      "grad_norm": 0.361328125,
+      "learning_rate": 9.595477386934673e-05,
+      "loss": 1.5434,
+      "step": 905
+    },
+    {
+      "epoch": 0.07731282467012981,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.59497487437186e-05,
+      "loss": 1.2041,
+      "step": 906
+    },
+    {
+      "epoch": 0.07739815891369507,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.594472361809046e-05,
+      "loss": 1.516,
+      "step": 907
+    },
+    {
+      "epoch": 0.07748349315726034,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.593969849246232e-05,
+      "loss": 1.1499,
+      "step": 908
+    },
+    {
+      "epoch": 0.07756882740082562,
+      "grad_norm": 0.345703125,
+      "learning_rate": 9.593467336683417e-05,
+      "loss": 1.5208,
+      "step": 909
+    },
+    {
+      "epoch": 0.07765416164439087,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.592964824120604e-05,
+      "loss": 1.44,
+      "step": 910
+    },
+    {
+      "epoch": 0.07773949588795614,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.592462311557789e-05,
+      "loss": 1.3304,
+      "step": 911
+    },
+    {
+      "epoch": 0.0778248301315214,
+      "grad_norm": 0.353515625,
+      "learning_rate": 9.591959798994975e-05,
+      "loss": 1.3398,
+      "step": 912
+    },
+    {
+      "epoch": 0.07791016437508667,
+      "grad_norm": 0.3125,
+      "learning_rate": 9.591457286432162e-05,
+      "loss": 1.3406,
+      "step": 913
+    },
+    {
+      "epoch": 0.07799549861865193,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.590954773869348e-05,
+      "loss": 1.3676,
+      "step": 914
+    },
+    {
+      "epoch": 0.0780808328622172,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.590452261306533e-05,
+      "loss": 1.2863,
+      "step": 915
+    },
+    {
+      "epoch": 0.07816616710578246,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.589949748743719e-05,
+      "loss": 1.4582,
+      "step": 916
+    },
+    {
+      "epoch": 0.07825150134934773,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.589447236180905e-05,
+      "loss": 1.3654,
+      "step": 917
+    },
+    {
+      "epoch": 0.07833683559291299,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.588944723618091e-05,
+      "loss": 1.3158,
+      "step": 918
+    },
+    {
+      "epoch": 0.07842216983647826,
+      "grad_norm": 0.294921875,
+      "learning_rate": 9.588442211055276e-05,
+      "loss": 1.3044,
+      "step": 919
+    },
+    {
+      "epoch": 0.07850750408004352,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.587939698492462e-05,
+      "loss": 1.3499,
+      "step": 920
+    },
+    {
+      "epoch": 0.07859283832360879,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.587437185929649e-05,
+      "loss": 1.3094,
+      "step": 921
+    },
+    {
+      "epoch": 0.07867817256717405,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.586934673366835e-05,
+      "loss": 1.5208,
+      "step": 922
+    },
+    {
+      "epoch": 0.07876350681073932,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.586432160804021e-05,
+      "loss": 1.3929,
+      "step": 923
+    },
+    {
+      "epoch": 0.07884884105430458,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.585929648241206e-05,
+      "loss": 1.1968,
+      "step": 924
+    },
+    {
+      "epoch": 0.07893417529786985,
+      "grad_norm": 0.29296875,
+      "learning_rate": 9.585427135678392e-05,
+      "loss": 1.3747,
+      "step": 925
+    },
+    {
+      "epoch": 0.0790195095414351,
+      "grad_norm": 0.298828125,
+      "learning_rate": 9.584924623115578e-05,
+      "loss": 1.2747,
+      "step": 926
+    },
+    {
+      "epoch": 0.07910484378500038,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.584422110552764e-05,
+      "loss": 1.3243,
+      "step": 927
+    },
+    {
+      "epoch": 0.07919017802856564,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.583919597989949e-05,
+      "loss": 1.4356,
+      "step": 928
+    },
+    {
+      "epoch": 0.07927551227213091,
+      "grad_norm": 0.287109375,
+      "learning_rate": 9.583417085427137e-05,
+      "loss": 1.3907,
+      "step": 929
+    },
+    {
+      "epoch": 0.07936084651569617,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.582914572864322e-05,
+      "loss": 1.3833,
+      "step": 930
+    },
+    {
+      "epoch": 0.07944618075926144,
+      "grad_norm": 0.2890625,
+      "learning_rate": 9.582412060301508e-05,
+      "loss": 1.258,
+      "step": 931
+    },
+    {
+      "epoch": 0.0795315150028267,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.581909547738694e-05,
+      "loss": 1.248,
+      "step": 932
+    },
+    {
+      "epoch": 0.07961684924639197,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.58140703517588e-05,
+      "loss": 1.3453,
+      "step": 933
+    },
+    {
+      "epoch": 0.07970218348995722,
+      "grad_norm": 0.30078125,
+      "learning_rate": 9.580904522613065e-05,
+      "loss": 1.3833,
+      "step": 934
+    },
+    {
+      "epoch": 0.0797875177335225,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.580402010050251e-05,
+      "loss": 1.3944,
+      "step": 935
+    },
+    {
+      "epoch": 0.07987285197708775,
+      "grad_norm": 0.30078125,
+      "learning_rate": 9.579899497487438e-05,
+      "loss": 1.2375,
+      "step": 936
+    },
+    {
+      "epoch": 0.07995818622065302,
+      "grad_norm": 0.3359375,
+      "learning_rate": 9.579396984924624e-05,
+      "loss": 1.4489,
+      "step": 937
+    },
+    {
+      "epoch": 0.08004352046421828,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.578894472361809e-05,
+      "loss": 1.6765,
+      "step": 938
+    },
+    {
+      "epoch": 0.08012885470778355,
+      "grad_norm": 0.298828125,
+      "learning_rate": 9.578391959798996e-05,
+      "loss": 1.3016,
+      "step": 939
+    },
+    {
+      "epoch": 0.08021418895134881,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.577889447236181e-05,
+      "loss": 1.4102,
+      "step": 940
+    },
+    {
+      "epoch": 0.08029952319491408,
+      "grad_norm": 0.376953125,
+      "learning_rate": 9.577386934673367e-05,
+      "loss": 1.5067,
+      "step": 941
+    },
+    {
+      "epoch": 0.08038485743847934,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.576884422110553e-05,
+      "loss": 1.4139,
+      "step": 942
+    },
+    {
+      "epoch": 0.08047019168204461,
+      "grad_norm": 0.2734375,
+      "learning_rate": 9.57638190954774e-05,
+      "loss": 1.4305,
+      "step": 943
+    },
+    {
+      "epoch": 0.08055552592560987,
+      "grad_norm": 0.3203125,
+      "learning_rate": 9.575879396984924e-05,
+      "loss": 1.2919,
+      "step": 944
+    },
+    {
+      "epoch": 0.08064086016917514,
+      "grad_norm": 0.28515625,
+      "learning_rate": 9.575376884422112e-05,
+      "loss": 1.2507,
+      "step": 945
+    },
+    {
+      "epoch": 0.0807261944127404,
+      "grad_norm": 0.279296875,
+      "learning_rate": 9.574874371859297e-05,
+      "loss": 1.2059,
+      "step": 946
+    },
+    {
+      "epoch": 0.08081152865630567,
+      "grad_norm": 0.30859375,
+      "learning_rate": 9.574371859296483e-05,
+      "loss": 1.2353,
+      "step": 947
+    },
+    {
+      "epoch": 0.08089686289987093,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.573869346733669e-05,
+      "loss": 1.4021,
+      "step": 948
+    },
+    {
+      "epoch": 0.0809821971434362,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.573366834170855e-05,
+      "loss": 1.3864,
+      "step": 949
+    },
+    {
+      "epoch": 0.08106753138700146,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.57286432160804e-05,
+      "loss": 1.5179,
+      "step": 950
+    },
+    {
+      "epoch": 0.08115286563056673,
+      "grad_norm": 0.294921875,
+      "learning_rate": 9.572361809045227e-05,
+      "loss": 1.3711,
+      "step": 951
+    },
+    {
+      "epoch": 0.08123819987413199,
+      "grad_norm": 0.3203125,
+      "learning_rate": 9.571859296482413e-05,
+      "loss": 1.2756,
+      "step": 952
+    },
+    {
+      "epoch": 0.08132353411769726,
+      "grad_norm": 0.37109375,
+      "learning_rate": 9.571356783919599e-05,
+      "loss": 1.2157,
+      "step": 953
+    },
+    {
+      "epoch": 0.08140886836126252,
+      "grad_norm": 0.298828125,
+      "learning_rate": 9.570854271356784e-05,
+      "loss": 1.3342,
+      "step": 954
+    },
+    {
+      "epoch": 0.08149420260482779,
+      "grad_norm": 0.296875,
+      "learning_rate": 9.57035175879397e-05,
+      "loss": 1.2293,
+      "step": 955
+    },
+    {
+      "epoch": 0.08157953684839304,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.569849246231156e-05,
+      "loss": 1.4536,
+      "step": 956
+    },
+    {
+      "epoch": 0.08166487109195832,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.569346733668342e-05,
+      "loss": 1.2433,
+      "step": 957
+    },
+    {
+      "epoch": 0.08175020533552357,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.568844221105529e-05,
+      "loss": 1.278,
+      "step": 958
+    },
+    {
+      "epoch": 0.08183553957908885,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.568341708542713e-05,
+      "loss": 1.3502,
+      "step": 959
+    },
+    {
+      "epoch": 0.0819208738226541,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.5678391959799e-05,
+      "loss": 1.4049,
+      "step": 960
+    },
+    {
+      "epoch": 0.08200620806621937,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.567336683417086e-05,
+      "loss": 1.3969,
+      "step": 961
+    },
+    {
+      "epoch": 0.08209154230978463,
+      "grad_norm": 0.337890625,
+      "learning_rate": 9.566834170854272e-05,
+      "loss": 1.307,
+      "step": 962
+    },
+    {
+      "epoch": 0.0821768765533499,
+      "grad_norm": 0.29296875,
+      "learning_rate": 9.566331658291457e-05,
+      "loss": 1.2148,
+      "step": 963
+    },
+    {
+      "epoch": 0.08226221079691516,
+      "grad_norm": 0.294921875,
+      "learning_rate": 9.565829145728644e-05,
+      "loss": 1.3293,
+      "step": 964
+    },
+    {
+      "epoch": 0.08234754504048043,
+      "grad_norm": 0.365234375,
+      "learning_rate": 9.565326633165829e-05,
+      "loss": 1.3477,
+      "step": 965
+    },
+    {
+      "epoch": 0.08243287928404569,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.564824120603016e-05,
+      "loss": 1.412,
+      "step": 966
+    },
+    {
+      "epoch": 0.08251821352761096,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.5643216080402e-05,
+      "loss": 1.3719,
+      "step": 967
+    },
+    {
+      "epoch": 0.08260354777117622,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.563819095477388e-05,
+      "loss": 1.3444,
+      "step": 968
+    },
+    {
+      "epoch": 0.08268888201474149,
+      "grad_norm": 0.330078125,
+      "learning_rate": 9.563316582914573e-05,
+      "loss": 1.4448,
+      "step": 969
+    },
+    {
+      "epoch": 0.08277421625830675,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.562814070351759e-05,
+      "loss": 1.3273,
+      "step": 970
+    },
+    {
+      "epoch": 0.08285955050187202,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.562311557788945e-05,
+      "loss": 1.2452,
+      "step": 971
+    },
+    {
+      "epoch": 0.08294488474543728,
+      "grad_norm": 0.322265625,
+      "learning_rate": 9.561809045226131e-05,
+      "loss": 1.3695,
+      "step": 972
+    },
+    {
+      "epoch": 0.08303021898900255,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.561306532663316e-05,
+      "loss": 1.3225,
+      "step": 973
+    },
+    {
+      "epoch": 0.08311555323256782,
+      "grad_norm": 0.28125,
+      "learning_rate": 9.560804020100504e-05,
+      "loss": 1.2685,
+      "step": 974
+    },
+    {
+      "epoch": 0.08320088747613308,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.560301507537689e-05,
+      "loss": 1.2856,
+      "step": 975
+    },
+    {
+      "epoch": 0.08328622171969835,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.559798994974875e-05,
+      "loss": 1.3447,
+      "step": 976
+    },
+    {
+      "epoch": 0.08337155596326361,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.559296482412061e-05,
+      "loss": 1.3304,
+      "step": 977
+    },
+    {
+      "epoch": 0.08345689020682888,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.558793969849247e-05,
+      "loss": 1.3872,
+      "step": 978
+    },
+    {
+      "epoch": 0.08354222445039414,
+      "grad_norm": 0.3046875,
+      "learning_rate": 9.558291457286432e-05,
+      "loss": 1.1908,
+      "step": 979
+    },
+    {
+      "epoch": 0.08362755869395941,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.55778894472362e-05,
+      "loss": 1.5538,
+      "step": 980
+    },
+    {
+      "epoch": 0.08371289293752467,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.557286432160804e-05,
+      "loss": 1.3397,
+      "step": 981
+    },
+    {
+      "epoch": 0.08379822718108994,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.556783919597991e-05,
+      "loss": 1.2492,
+      "step": 982
+    },
+    {
+      "epoch": 0.0838835614246552,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.556281407035176e-05,
+      "loss": 1.3243,
+      "step": 983
+    },
+    {
+      "epoch": 0.08396889566822047,
+      "grad_norm": 0.33203125,
+      "learning_rate": 9.555778894472363e-05,
+      "loss": 1.4901,
+      "step": 984
+    },
+    {
+      "epoch": 0.08405422991178572,
+      "grad_norm": 0.294921875,
+      "learning_rate": 9.555276381909548e-05,
+      "loss": 1.2068,
+      "step": 985
+    },
+    {
+      "epoch": 0.084139564155351,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.554773869346734e-05,
+      "loss": 1.449,
+      "step": 986
+    },
+    {
+      "epoch": 0.08422489839891625,
+      "grad_norm": 0.3515625,
+      "learning_rate": 9.55427135678392e-05,
+      "loss": 1.2157,
+      "step": 987
+    },
+    {
+      "epoch": 0.08431023264248153,
+      "grad_norm": 0.294921875,
+      "learning_rate": 9.553768844221107e-05,
+      "loss": 1.3926,
+      "step": 988
+    },
+    {
+      "epoch": 0.08439556688604678,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.553266331658291e-05,
+      "loss": 1.3208,
+      "step": 989
+    },
+    {
+      "epoch": 0.08448090112961205,
+      "grad_norm": 0.357421875,
+      "learning_rate": 9.552763819095478e-05,
+      "loss": 1.7452,
+      "step": 990
+    },
+    {
+      "epoch": 0.08456623537317731,
+      "grad_norm": 0.298828125,
+      "learning_rate": 9.552261306532664e-05,
+      "loss": 1.3488,
+      "step": 991
+    },
+    {
+      "epoch": 0.08465156961674258,
+      "grad_norm": 0.314453125,
+      "learning_rate": 9.55175879396985e-05,
+      "loss": 1.4598,
+      "step": 992
+    },
+    {
+      "epoch": 0.08473690386030784,
+      "grad_norm": 0.390625,
+      "learning_rate": 9.551256281407036e-05,
+      "loss": 1.2757,
+      "step": 993
+    },
+    {
+      "epoch": 0.08482223810387311,
+      "grad_norm": 0.34375,
+      "learning_rate": 9.550753768844221e-05,
+      "loss": 1.3147,
+      "step": 994
+    },
+    {
+      "epoch": 0.08490757234743837,
+      "grad_norm": 0.291015625,
+      "learning_rate": 9.550251256281407e-05,
+      "loss": 1.3024,
+      "step": 995
+    },
+    {
+      "epoch": 0.08499290659100364,
+      "grad_norm": 0.365234375,
+      "learning_rate": 9.549748743718593e-05,
+      "loss": 1.3254,
+      "step": 996
+    },
+    {
+      "epoch": 0.0850782408345689,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.54924623115578e-05,
+      "loss": 1.4054,
+      "step": 997
+    },
+    {
+      "epoch": 0.08516357507813417,
+      "grad_norm": 0.31640625,
+      "learning_rate": 9.548743718592965e-05,
+      "loss": 1.407,
+      "step": 998
+    },
+    {
+      "epoch": 0.08524890932169943,
+      "grad_norm": 0.302734375,
+      "learning_rate": 9.548241206030151e-05,
+      "loss": 1.2184,
+      "step": 999
+    },
+    {
+      "epoch": 0.0853342435652647,
+      "grad_norm": 0.34765625,
+      "learning_rate": 9.547738693467337e-05,
+      "loss": 1.4606,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 20000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.58314499784704e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}